One has to acknowledge that different types of targets (e.g. an extracellular enzyme vs. an intracellular protein vs. a gene therapy target) and different disease contexts demand different playbooks. We connect all the prior analyses to practical plans, ensuring that our strategy to drug the target and prove its value is as sound as the biology itself. I have listed the key considerations below for founders with identified targets looking to initiate the development phase.

1. Timeline to conviction

The central insight here is that target biology sets a floor on how fast you can reach human proof-of-concept, and everything else (trial design, financing, endpoints) builds from that constraint.

Fast-readout symptomatic targets (hours to weeks)

Vertex Pharmaceuticals / Suzetrigine (NaV1.8 inhibitor, acute pain) → clean example of genetically validated target enabling rapid conviction. NaV1.8 loss-of-function mutations cause pain insensitivity in humans. Vertex designed Phase 2 around post-surgical pain models (abdominoplasty, bunionectomy) where the primary endpoint - pain reduction at 48 hours - gives a binary signal in under a week per patient. Published in NEJM August 2023, positive Phase 3 readout January 2024, FDA approval January 2025 as Journavx. From Phase 2 publication to approval in roughly two years, making it one of the fastest first-in-class timelines in recent memory.

Disease-modifying targets requiring years of patience

Denali Therapeutics / BIIB122 (LRRK2 inhibitor, Parkinson’s disease). The Phase 2b LUMA study enrolled ~640 early-PD patients; clinical readout is expected in 2026 (roughly 6+ years after Phase 1 initiation). Denali’s strategy for surviving this timeline: (1) a $560M upfront + $465M equity Biogen partnership to share Phase 2b costs, (2) CSF and urinary LRRK2 phosphorylation biomarkers to demonstrate intermediate target engagement, (3) a diversified pipeline with nearer-term assets (tividenofusp alfa BLA submitted 2025) to provide interim milestones, and (4) $1.28B cash runway extending into 2028 plus a $275M Royalty Pharma deal. The $422.8M net loss in FY2024 shows the capital intensity of disease-modification programs.

Novartis / Pelacarsen (antisense, Lp(a) lowering for CV events). This is the extreme end: an event-driven cardiovascular outcomes trial where the timeline is partly beyond management’s control, it depends on MACE events accumulating in the study population. Results originally expected 2025, now pushed to 2026. Only a large-pharma balance sheet can absorb this kind of timeline uncertainty.

GLP-1 agonists in Parkinson’s (lixisenatide). Phase 2 results in NEJM (April 2024) showed lixisenatide slowed motor disability progression at 12 months. But confirming disease modification in neurodegeneration (distinguishing symptomatic from disease-modifying effects) requires years of follow-up. Multiple parallel GLP-1 PD trials (exenatide, NLY01, PT320) are now running, pushing the field toward biomarker-intermediate strategies using alpha-synuclein seeding assays and NfL levels.

Clever trial designs compress the timeline

Spyre Therapeutics - 9 proof-of-concept readouts in 2 years. Spyre is running two innovative designs simultaneously: SKYLINE-UC, a Phase 2 platform trial in ulcerative colitis testing three next-gen monotherapies (anti-α4β7, anti-TL1A, anti-IL-23) AND their combinations under a single protocol; and SKYWAY-RD, a basket trial evaluating anti-TL1A across three rheumatic diseases (RA, PsA, axSpA). With $757M in cash and runway into H2 2028, Spyre’s entire corporate strategy is organized around maximizing POC readouts per dollar.

Abivax / Obefazimod (UC, 8-week induction). Even though UC is a chronic inflammatory condition, Abivax designed an 8-week induction endpoint (Modified Mayo Score) that delivered Phase 2b proof-of-concept in months rather than years. Long-term extension then built progressive conviction (52.5% clinical remission at 96 weeks). Phase 3 ABTECT-1 and ABTECT-2 hit primary endpoints in July 2025; stock surged >500%. NDA submission planned H2 2026.

Regulatory pathway stacking as strategy

Breakthrough Therapy Designation has been granted to 634 of 1,622 requests (39%), and it reduces clinical development time by roughly 30%. Parexel analysis found that 52% of orphan drugs received BTD, and 60% of approved orphan BTD drugs stacked 2+ expedited programs. The most aggressive example: Krazati (adagrasib) achieved a 4.1-year development timeline (versus the 10.1-year average) by stacking accelerated approval + BTD + Fast Track + Priority Review.

Recent notable BTDs: Lilly’s sofetabart mipitecan (FRα ADC, 50% ORR across all FRα expression levels, January 2026); Cogent’s bezuclastinib (systemic mastocytosis, October 2025); Sanofi’s tolebrutinib (non-relapsing secondary progressive MS, December 2024).

2. Biomarker plan

The FDA-NIH BEST (Biomarkers, EndpointS, and other Tools) glossary defines a biomarker as “a defined characteristic that is measured as an indicator of normal biological processes, pathogenic processes, or responses to an exposure or intervention, including therapeutic interventions.” Molecular, histologic, radiographic, or physiologic characteristics are types of biomarkers. Important: biomarker is NOT an assessment of how a patient feels, functions, or survives. This distinction separates biomarkers from clinical outcomes, and confusing the two has destroyed billions in value.

The FDA recognizes seven biomarker categories: susceptibility/risk, diagnostic, monitoring, prognostic, predictive, pharmacodynamic/response, and safety. For drug development strategy, founders need to think about biomarkers in a chain (from molecule to patient) where each link answers a different question.

Target engagement biomarkers that informed dose selection

Eli Lilly / Donanemab, the “treat-to-target” paradigm. Donanemab used amyloid PET as a treatment-stopping biomarker: when patients achieved amyloid-negative status (<24.1 centiloids), dosing was stopped. 67.8% of patients reached this threshold by 76 weeks. In this case, a target engagement biomarker dictates not just dose but treatment duration.

Mechanism biomarkers versus disease biomarkers: the critical distinction

The anti-amyloid Alzheimer’s programs provide a clear illustration of this hierarchy:

The critical lesson here: lecanemab produced large amyloid reductions (target engagement) AND CSF p-tau181 reductions (mechanism), but volumetric MRI paradoxically showed greater brain volume loss in the treatment group. Mechanism confirmed but disease biomarker gave mixed signals.

Alector / Latozinemab (anti-sortilin, FTD). The Phase 2 INFRONT-3 trial (2025) showed a “significant effect” on progranulin levels, the mechanism biomarker moved in the right direction. But MRI, fluid biomarkers of neurodegeneration, and clinical decline showed no apparent effect. Alector laid off ~50% of workforce and abandoned the program entirely. The drug clearly hit its target but failed to modify the disease. Mechanism engagement ≠ disease modification.

Surrogate endpoints enabling accelerated approval (2023-2026)

Biomarker absence shapes entire development strategies

MASH: the $100B endpoint problem. For decades, the only gold standard for MASH was liver biopsy, which is invasive, sampling only 1/50,000th of the liver, with inter-reader kappa scores of just 0.33-0.61. Screening failure rates are astronomical (~13% of biopsied patients don’t confirm MASH). This forced Phase 3 trials like Madrigal’s MAESTRO-NASH to require serial biopsies (expensive, slow, and limiting enrollment). The paradigm shift came in September 2025 when FDA accepted FibroScan/VCTE (vibration-controlled transient elastography) as a “reasonably likely surrogate endpoint”. It was the first non-invasive test accepted for MASH. Lilly immediately announced plans for Phase 3 trials of tirzepatide using purely non-invasive assessment. Additionally, FDA qualified AIM-MASH, the first AI tool for MASH histological scoring (Nature Medicine 2024), to reduce biopsy variability.

Example where biomarker strategy was inadequate

CETP inhibitors: the $10B+ surrogate biomarker trap. Torcetrapib (Pfizer), dalcetrapib (Roche), and evacetrapib (Lilly) all successfully raised HDL cholesterol—the surrogate that was “reasonably likely to predict clinical benefit.” Torcetrapib increased death risk. Dalcetrapib showed zero efficacy. Evacetrapib was stopped for futility. The surrogate moved perfectly; the clinical outcome never followed. This remains the canonical warning against anchoring development strategy on an inadequately validated surrogate.

3. Model selection

Strategic model choices that de-risked programs

Emulate / Liver-Chip (organ-on-chip for DILI prediction). Validated across 870 chips, Emulate’s Liver-Chip detects drug-induced liver injury with 87% sensitivity. Economic modeling estimates broad adoption could reduce small-molecule liver toxicity failures by up to 87% and generate $3 billion/year in R&D productivity gains. Now used by 21 of the top 25 pharma companies and submitted to FDA’s ISTAND program.

Biocytogen (drug-target humanized mice for IND toxicology). Rather than using non-human primates, Biocytogen developed mice with specific human gene replacements. Over 20 drug candidates supported by these models have received IND approvals from FDA, NMPA, and TGA. This aligns with FDA’s April 2025 announcement to reduce NHP reliance.

Vivodyne (robotic 3D organ tissues + AI). Founded by organ-chip pioneer Dan Huh’s group, Vivodyne’s robotic Data Engine tests 10,000+ independent human tissues per run. Secured $38M seed (2024) + $40M Series A (May 2025, Khosla Ventures lead). Already partnered with a majority of top-10 pharma companies. Their deliberate bet on throughput (100,000+ tissue tests in 2 weeks) addresses the scalability gap that historically limited organoid adoption.

Recent failures attributable to wrong model selection

Takeda / TAK-875 (fasiglifam, GPR40 agonist, T2D). Phase 3 terminated due to severe DILI despite clean preclinical animal safety profiles. The reactive acyl glucuronide metabolite caused covalent binding above the DILI risk threshold and inhibited bile acid transporters, the mechanisms standard rat and NHP assays failed to capture. Retrospective testing on Emulate’s Liver-Chip detected the toxicity signal.

Merck KGaA / Evobrutinib (BTK inhibitor, MS). Phase 3 halted for liver injury even though preclinical safety profile was clean. A 2025 Nature Communications ToxPredictor model trained on human hepatocyte RNA-seq correctly classified evobrutinib as “high risk” with low margin of safety. Standard preclinical species missed it.

Alzheimer’s disease mouse models. Over 155 terminated AD clinical trials as of 2022. A 2024 eLife paper documented that transgenic AD mice (5XFAD, 3xTg-AD, APP/PS1) express familial AD mutations found in <1% of all AD cases. Past amyloid vaccines worked in mice but caused brain swelling in humans. Second-generation humanized App models and iPSC-derived cerebral organoids are now emerging as alternatives.

The model decision framework

The most sophisticated programs now build tiered decision trees: (1) 2D cultures for initial high-throughput screening, (2) organoids/organ-chips for human-specific efficacy and toxicity questions, (3) humanized mice for in vivo target engagement and systemic PK/PD, (4) PDX models for tumor heterogeneity, (5) standard animals for regulatory-required systemic toxicity.

The regulatory landscape supports this: FDA Modernization Act 2.0 (2022) removed mandatory animal testing. The FDA’s April 2025 roadmap aims to make animal testing “the exception rather than the norm” within 3-5 years, starting with monoclonal antibodies. The PDX model market is valued at ~$540M in 2025, projected to reach $1B by 2030.

4. Modality constraints

Target location as the primary modality determinant

Extracellular targets → antibodies. AbbVie acquired Aliada Therapeutics for $1.4 billion (October 2024) for its anti-pyroglutamate amyloid-β antibody with BBB-crossing technology (MODEL™ platform using transferrin receptor and CD98 binders). AbbVie had already killed its undifferentiated internal anti-amyloid antibody ABBV-916. The mAb clinical pipeline is the largest across all modalities, with 18% growth in the past year.

Intracellular targets → alternative modalities. Tecelra (afamitresgene autoleucel), approved 2024 for synovial sarcoma, is a TCR-T therapy targeting intracellular MAGE-A4 presented on cell surfaces via HLA. The target is inside the cancer cell and only accessible via MHC presentation, antibodies cannot reach it. This is also why targeted protein degraders (PROTACs, molecular glues) are gaining traction: ~85% of the human proteome remains “undruggable” by conventional small molecules, and these targets are overwhelmingly intracellular.

Gene/RNA targets → editing and silencing.

• Verve Therapeutics / VERVE-102 (PCSK9 base editing): Single IV infusion achieved 53% LDL-C reduction and 60% PCSK9 reduction at highest dose. Eli Lilly acquired Verve in June 2025. The rationale: PCSK9 mAbs have ~50% discontinuation rates; inclisiran (siRNA) has ~20% first-year discontinuation. One-time permanent gene inactivation addresses the adherence crisis.

• CRISPR Therapeutics / CTX310 (ANGPTL3 gene editing): Single IV dose achieved -73% ANGPTL3, -55% triglycerides, -49% LDL (published NEJM November 2025). Compared to Regeneron’s evinacumab (anti-ANGPTL3 mAb requiring monthly IV infusions), gene editing offers a one-time cure.

• Intellia / NTLA-2001 (TTR gene editing): >90% TTR knockdown sustained over 2+ years, now in two Phase 3 trials.

Delivery as the rate-limiting step (and value driver)

Three data points frame the delivery challenge: more than 98% of small molecules and nearly 100% of large molecules fail to cross the blood-brain barrier at therapeutic concentrations. All six FDA-approved siRNA therapies target liver-expressed genes via GalNAc conjugation → Valuable billion-dollar assets in cracking extrahepatic delivery.

The deals confirm this thesis:

• Novartis / Avidity Biosciences ($12 billion acquisition, October 2025): Avidity’s antibody-oligonucleotide conjugates (AOCs) fuse anti-TfR1 antibodies with oligonucleotides to deliver RNA therapeutics directly to skeletal and cardiac muscle. Lead program del-zota achieved ~25% normal dystrophin restoration in DMD (versus ~1% for eteplirsen’s naked PMO). Same payload, radically different delivery, 25x the efficacy.

• Eli Lilly / Sangamo ($1.4 billion milestone deal): Licensed Sangamo’s STAC-BBB capsid (a proprietary AAV that crosses the blood-brain barrier) for CNS gene therapies.

Cost and timeline hierarchy across modalities

[Source: 1, 2, 3, 4, 5, 6, 7]

IP forces modality diversification

The crowded PCSK9 antibody patent landscape (Amgen/Regeneron composition patents) likely incentivized Verve to pursue base editing and others to develop oral small molecules. ADC patent filings have more than doubled since 2018, reaching ~500 new patent families per year with 3,761 families analyzed. Biologics enjoy 12.5 years of market exclusivity versus 5 years for small molecules. It’s a powerful incentive to choose biologic modalities when the target biology permits it.

5. Clinical endpoint strategy

A clinical endpoint is a precisely defined variable intended to reflect an outcome of interest that is statistically analyzed to address a specific research question. The definition typically specifies the type of assessment made, the timing of those assessments, the tools used, and how multiple assessments within an individual are combined. Endpoints are not inherent properties of diseases, they are constructed measures that the FDA and your trial are designed around.

The FDA distinguishes three families of endpoints:

Primary endpoints are the basis for determining whether the trial meets its objective. They are the centerpiece of the regulatory submission and the trial must be statistically powered to detect a treatment effect on them.

Secondary endpoints provide supportive evidence about the drug’s effect on the primary endpoint or demonstrate additional clinical effects. They are analyzed only after the primary endpoint achieves statistical significance (per FDA’s hierarchical testing framework), and they can support additional labeling claims. But secondary endpoint results cannot save a missed primary endpoint.

Exploratory endpoints generate hypotheses for future trials. They have no regulatory weight but can inform dose selection, identify responsive subpopulations, or reveal unexpected effects.

Within these families, endpoints fall into two fundamental categories. Clinical outcome assessments directly measure how a patient feels, functions, or survives, such as pain scores, cognitive scales, overall survival, or physical function tests. These measure the thing we actually care about. Surrogate endpoints are biomarker-based substitutes that predict clinical benefit without directly measuring it, like tumor shrinkage predicting longer survival, or HbA1c reduction predicting fewer diabetic complications.

This distinction matters for regulatory strategy because it determines your approval pathway.

Co-primary and composite endpoints

When a disease affects patients through multiple dimensions, the FDA may require effects on more than one endpoint. Migraine trials use co-primary endpoints: pain freedom at 2 hours AND relief of the most bothersome symptom. Both must be met. This raises the statistical bar (multiplicity adjustments are required) but ensures the drug addresses the full clinical picture.

Composite endpoints combine multiple events into a single measure. MACE (major adverse cardiovascular events: cardiovascular death, non-fatal MI, non-fatal stroke) is the standard composite for cardiovascular outcome trials. The advantage is statistical efficiency: events accumulate faster when multiple outcomes count. The risk is that the composite can be driven by a less important component. A drug might reduce hospitalizations but not death, yet “meet” the composite endpoint.

Patient stratification that improved success rates

Donanemab‘s TRAILBLAZER-ALZ 2 stratified patients by tau burden on PET imaging. In the low/medium tau population, donanemab showed 35% slowing of CDR-SB decline versus placebo, a much stronger signal than in the overall population. Tissue-agnostic approvals (pembrolizumab for MSI-H/dMMR tumors; larotrectinib for NTRK fusions) represent the ultimate biomarker-enrichment strategy: single-arm basket trials enriched entirely by molecular biomarker, achieving high enough ORRs that randomized controls became unnecessary.

Biomarker-driven trials achieve 15.9% LOA, double the overall 7.9% average. A Boehringer Ingelheim-supported 2023 preprint demonstrated that AI-based stratification into “slow” versus “fast” Alzheimer’s progressors could make enrichment trials >13% cheaper with improved success probability.

The comparator problem

Your endpoint strategy must account for the standard of care at the time your trial reads out (not when it was designed). Pfizer’s KEYLYNK-010 in mCRPC randomized patients against abiraterone/enzalutamide, but the CARD trial established cabazitaxel as the new standard during enrollment. The STAMPEDE trial solved this adaptively: when docetaxel proved better, it became the new control arm for subsequent comparisons → continuous relevance.

6. The numbers that tie it all together

Success rates: the denominator matters most

Overall probability of success from Phase 1 to approval is 7.9% (BIO, 2011-2020) and trending downward to 6.7% (Citeline, 2014-2023). Phase 2 remains the biggest bottleneck at 28-29% success. But these averages hide massive variation:

• Hematology: 23.9% LOA (highest)

• Rare disease gene therapy: 18.5% LOA

• CAR-T: 17.3% LOA

• Rare disease overall: 17.0% LOA

• Biomarker-selected trials: 15.9% LOA (2x average)

• Oncology: 5.3% LOA

• Chronic high-prevalence disease: 5.9% LOA

Cost benchmarks for founders

Average drug development cost reached $2.23 billion per asset in 2024 (Deloitte), up from $2.12B in 2023. Mean out-of-pocket cost per successful drug is $172.7 million (JAMA Network Open, 2024), rising to $515.8M when failure costs are included, and $879.3M when capital costs are added. Phase-level averages: Phase 1 ~$5.3M, Phase 2 ~$18.5M, Phase 3 ~$52.8M. But these mask modality-specific realities: rare disease gene therapy can go from discovery to NDA for under $25M (Elpida Therapeutics), while large-indication cardiovascular outcomes trials consume hundreds of millions.

The AI prediction frontier

A scoping review identified 142 studies (2013-2024) applying AI to clinical trial risk prediction, achieving up to 85% accuracy in forecasting outcomes and AUROC up to 96% for specific predictions. AI-accelerated trial design may reduce timelines by 30-50% and costs by up to 40%. The top predictive factors: disease indication, target, modality, drug novelty, and sponsor type.

Connecting the pieces: target typology as the organizing principle

The five decisions above are not independent. They form an interlocking system where the target typology determines the natural constraints and the founder’s job is to build a program architecture that is internally consistent.

A genetically validated extracellular target (Tier 2 evidence, clear LOF phenotype, accessible to mAbs) with a symptomatic endpoint is the lowest-risk, fastest-to-conviction profile. You can move quickly through validation, pick an established modality, use target engagement biomarkers to set dose, and power a pivotal trial on a clinical endpoint with weeks-to-months readout. Your primary risk is competitive as someone else is probably working on the same target. Your capital efficiency should be high.

A perturbation-derived intracellular target (Tier 3-4 evidence, requires novel modality, disease-modifying intent in a progressive condition) is the highest-risk, highest-reward profile. You need extensive orthogonal validation before IND, must solve a delivery problem, may lack intermediate biomarkers, and face a years-long timeline to clinical conviction. Your primary risk is biological, you might have the wrong target. Your capital requirements will be an order of magnitude higher.

Most programs fall somewhere between these extremes, and the strategic art is in honestly assessing where your target sits and building a plan that neither overreaches (spending as if you have Tier 2 evidence when you have Tier 4) nor under-invests (running a minimal preclinical package when your biology demands more).

This series has tried to lay out the logic of that structure, from how targets are identified and validated, to how development strategy follows from target biology. The gap between a scientific insight about a target and an approved medicine is technical, strategic, financial, and organizational. The founders who close that gap most efficiently are the ones who understand that every decision downstream (the biomarker, the model, the modality, the endpoint, the financing) is already constrained by the biology they chose at the start.

Build the plan that matches your biology. Then execute it faster and better than anyone else.

Acknowledgements

And that is the end of the “Betting on Biology” series! Appreciate everyone who has read and commented, as always suggestions of topics are welcome. A big thank you to Satvik Dasariraju for the inspiration, thoughtful comments, and prompt answers to my many questions; to Alex Colville for invaluable writing guidance throughout the process; and to every age1 crew for helpful pointers. Special thanks to Aerska, Loyal, and the amazing founders for the inspiration.

Cheers, see you next piece/series!

So what’s next?

First, a case study, then a call for proposals.

Aerska’s approach and pace of progress underscore our thesis-driven approach to early-stage biotech investing.

We look for enabling technologies that propel therapeutic asset concepts to have the potential to go on to become civilization-changing medicines.

In the case of neurodegenerative disease, we identified four core challenges:

1. Crossing the blood-brain barrier

2. The patients most likely to benefit from therapeutics (prodromal phase, early disease course, etc.) are the ones least likely to go through invasive procedures (stereotactic injections and spinal taps are common). Systemically-administered approaches have been intractable until extremely recently

3. Intracellular proteinopathies (ubiquitous in neurodegenerative disease) are not accessible to antibodies, whether or not they are brain-shuttled

4. Target overcrowding

That’s why brain-shuttled RNAi, aimed at early intervention in high-risk cohorts, has the potential to change how we think about neurodegeneration, one of the greatest challenges of our time.

We are laser-focused on the following:

Please reach out if any of the following topics resonate and you have a thesis on how to surmount these challenges.

A lot of the time, founders teach us the best spaces to build in, or invent new spaces altogether! While our focus and interests go beyond the five areas listed below, we have a hunch these five will be especially important for the 2030s.

Osteoarthritis (and other approaches for joint/tendon degeneration)

• OA is a leading cause of disability and will affect 1B individuals by 2050

• Approaches that only improve cartilage without affecting pain and function will have intractable clinical development paths given WOMAC is the primary endpoint (e.g., FGF18); approaches that solely address pain will not address the underlying cause of the disease and may actually accelerate deterioration (e.g., NGF inhibitors).

• Cartilage-anchoring protein (e.g., Sanofi’s SAR446959) and siRNA approaches are new to the scene in OA

Chronic kidney disease

• The definition of CKD is eGFR < 60 or urine albumin/creatine >= 30. eGFR drops by 1 point per year after age 40, meaning pretty much everyone above a certain age (~70) has CKD (symptomatic, secondary to hypertension or diabetes, or asymptomatic). eGFR and uACR are “integrators” of cumulative metabolic aging.

• We never regrow nephrons. Current SOC (standard of care) only affects blood pressure (fluid balance) or blood glucose.

• Unlike HbA1c or LDL-C, there is no such thing as “sufficiently controlled” GFR decline

• SGLT2 inhibitors are multi-billion dollar drugs but have barely penetrated the asymptomatic CKD market.

Delaying menopause

• See deep dives from our team here (2024) and here (2026)!

Prevention of vision loss

• MR/GWAS approaches are uncovering many promising targets for wet AMD, dry AMD, glaucoma, etc. Selecting the right modality with tissue specificity is key.

• There are no approved drugs for intermediate AMD, despite it being the optimal phase to intervene in (hard to rescue vision loss when it is pretty much 100% gone).

• Targets that affect AMD progression may be different than those that affect “the path to AMD” (prevention; subclinical to intermediate AMD)

• Pharmacological SOC for presbyopia are eye drops that induce pupil contraction (no modulation of underlying biology: loss of flexibility in the eye’s crystalline lens)

Age-related cardiorespiratory deficit

• Cardiomyocytes are post-mitotic and do not regenerate in adult humans

• Modalities that enable access to cardiomyocyte-specific and intracellular targets are now here (e.g., siRNA conjugated to TfR1 ligand or antibody) and in development for rare conditions (e.g., PLN cardiomyopathy)

• MR/GWAS with cardiomyocyte-specific eQTLs has revealed some interesting causal targets that are interesting for common chronic heart diseases but also for modifier effects in rare/monogenic heart diseases (e.g., dilated cardiomyopathy subtypes)

• An extremely common consequence of aging with functional, physical, and socioemotional effects is being “short of breath” (age-related decrease in VO₂ max, endurance, fatigue resistance, etc.) We now have MR/GWAS hits suggestive of an effect in preventing CAD but also improving markers of cardiorespiratory function (e.g., FEV1, bronchiectasis)

Identifying a plausible target is only the beginning. Target validation is the systematic process of confirming that modulating a specific molecular target will produce a therapeutic effect in human patients before committing to expensive clinical development. It is a structured body of evidence addressing a chain of linked questions: Is this target causally involved in the disease, or correlated with it? If causal, in which direction should it be modulated (inhibited or activated)? In which tissue must the drug act? In which patients, at which disease stage? And, importantly: what experiment could disprove all of this most efficiently?

Echoing the previous parts of the series: focusing on a wrong target is EXPENSIVE. 40%-50% of clinical drug development failures are due to a lack of clinical efficacy. Phase III trials, where most of this value is destroyed, have a median pivotal cost of $19M with large outcome trials routinely costing $100-350M. Including the cost of all prior failed programs, the fully-loaded cost per approved drug now ranges from $1.5 billion to $2.8 billion. Beyond financial loss, more importantly, a Phase III failure burns a decade of scientific effort and leaves patients without the therapy they were promised.

The data on what reduces this failure rate is unambiguous. Nelson et al.’s landmark 2015 Nature Genetics analysis showed that drug programs targeting proteins with human genetic evidence of disease association were twice as likely to gain approval. A 2024 follow-up in Nature that leverages a decade of additional GWAS data and pipeline outcomes revised that estimate upward: genetically supported targets are now 2.6× more likely to succeed, and the advantage increases with confidence in the causal gene. Alnylam Pharmaceuticals, which has made genetic target validation a foundational R&D principle, reports a clinical trial success rate 6× the industry average. And when AstraZeneca implemented its “5R framework“ (right target, right tissue, right safety, right patient, right commercial potential) its success rate from candidate nomination to Phase III completion rose from 4% to 19% in five years, and has since climbed to 23%, nearly six times the pre-reform baseline and well above the industry average.

This post lays out a framework for evaluating target validation when building a drug program. We cover directionality testing, context dependence, the hierarchy of evidence, and what we call “killer experiments”: the cheapest insurance money can buy.
Directionality testing: the billion dollar coin flip

A fundamental early question in target validation is the direction of intervention - should we activate (agonize) or inhibit the target? This decision must align with the disease biology. If the disease state already suppresses the target’s activity or levels, further inhibition may yield little benefit. For example, trials of myostatin inhibitors in Duchenne muscular dystrophy (DMD) failed partly because DMD patients already have very low myostatin levels. In such a context, trying to inhibit an already diminished factor is a losing battle. Conversely, if considering an inhibitory drug for a condition like obesity, one would prefer the target to be overactive or elevated in obesity – inhibiting something that is already low or normal in obese patients is unlikely to help. The target’s expression across age and disease states provides clues: ideally, an obesity target intended for blockade should be increased in obese individuals (and in contrast a target to agonize might be abnormally low or underactive in disease).

The GIP receptor directionality paradox

Glucose-dependent insulinotropic polypeptide (GIP) exemplifies how tricky directionality can be. Human genetics indicate that reducing GIP signaling might be beneficial: loss-of-function variants in the GIP receptor (GIPR) are associated with lower BMI and protection against obesity. This suggests that antagonizing GIPR could have anti-obesity effects, and indeed GIPR antagonists are being explored for weight loss, with Amgen’s MariTide achieving approximately 20% weight loss at 52 weeks.

Paradoxically, agonizing the GIP receptor has also shown promise when combined with GLP-1 agonism. The dual GIP/GLP-1 agonist tirzepatide produces approximately 21% weight loss in the SURMOUNT trials. How could activating a receptor that genetics say is better off silent lead to weight reduction?

The resolution involves three mechanisms. Firstly, GIPR agonism and antagonism reduce food intake through different neuronal populations. Agonism acts on GABAergic neurons to suppress appetite directly, while antagonism acts on non-GABAergic neurons to potentiate GLP-1 receptor signaling. The practical implication is that antagonism requires concurrent GLP-1R activation to work, while agonism does not. Secondly, chronic agonism can lead to functional antagonism through receptor desensitization. Continuous stimulation of GIPR causes the receptor to internalize or downregulate, “burning out” the receptor’s activity. Thirdly, this effect seems more pronounced in humans than in rodents. Thus, a long-acting agonist can paradoxically achieve the same outcome as an antagonist by inducing ligand-induced tolerance.

The paradox and the mechanistic insights has spawned over $300 million in startup financing in the past fourteen months:

Antag Therapeutics (Copenhagen, €80M Series A, Versant/Novo Holdings) launched with AT-7687, a pure GIPR antagonist peptide co-developed by GLP-1 co-discoverer Jens Juul Holst. Phase 1 data reported in January 2026 showed enhanced weight loss when combined with the amylin analog cagrilintide — superior to either monotherapy. Preclinical data showed subnanomolar antagonistic potency (pKB 9.5) with a 27-hour half-life.

Helicore Biopharma (San Francisco, $65M Series A, Versant/OrbiMed) takes a different approach: its lead asset HCR-188 is a monoclonal antibody that binds circulating GIP ligand rather than blocking the receptor directly. This could neutralize GIP signaling in both periphery and CNS. Phase 1 is underway.

Alveus Therapeutics (Philadelphia, $160M Series A, New Rhein/Sanofi Capital) is advancing ALV-100, a bifunctional GIPR antagonist/GLP-1R agonist fusion. Phase 1b dosed its first patient in January 2026.

This shows the importance of testing both directions: a “directionality test” (e.g. comparing agonist vs antagonist molecules or genetic upregulation vs knockout) is critical to reveal which way of nudging the pathway yields benefit. It’s not always intuitive, human biology may respond differently than mouse models or simplistic assumptions. Rigorous experimentation is needed to confirm the correct strategy (for instance, testing whether sustained GIPR activation actually decreases signaling over time, and whether blocking GIPR outright has similar or distinct effects on metabolism). Such tests prevent pursuing the wrong approach on a valid target. In summary, getting the direction right can make or break a target: we must ask whether to floor the gas or hit the brakes, and the answer must be grounded in empirical evidence about the disease context.

Context Matters

Directionality is closely tied to when and where the target acts. Some targets produce opposite effects in different tissues or disease stages, which means that even a correctly identified target with the right directionality can fail if the biological context is wrong. We evaluate context across three dimensions: disease-state specificity, tissue of action, and cross-species conservation.

Disease-state specificity

GDF15, a hormone that suppresses appetite by acting on brainstem receptors (GFRAL), illustrates the problem. Pharmacologically raising GDF15 causes weight loss in preclinical models by reducing appetite. But in chronically obese individuals, GDF15 is often already elevated with no apparent effect on appetite or weight. Obese humans and rodents have significantly higher circulating GDF15 levels than lean controls, yet their food intake does not decrease accordingly, implying a form of resistance or desensitization. One hypothesis is that high GDF15 in obesity upregulates a negative regulator (the metalloprotease MMP14) that cleaves the GFRAL receptor, dampening downstream signaling. A “miracle” anorectic hormone in a short-term mouse test may fizzle in the chronic human condition due to feedback mechanisms.

The lesson: evaluating a target requires checking disease-specific context. Is the target pathway intact and actionable in the relevant tissue? Does the patient’s disease stage or concurrent treatments change the target’s baseline activity? For DMD, chronic corticosteroids and muscle degeneration changed myostatin’s baseline. For GDF15, long-term obesity changed receptor responsiveness. We must ensure the biology we aim to manipulate is not already maximally compensated by the body’s own feedback responses.

Tissue-of-action

Tissue-of-action determines everything downstream. This is the validation principle that shaped Aerska. Dozens of CNS programs have failed not because the target biology was wrong, but because the drug never reached the brain in sufficient concentrations. The blood-brain barrier excludes over 98% of small molecules and virtually all large biologics. An siRNA that silences target mRNA with exquisite specificity is worthless if it cannot cross that barrier.

Aerska’s brain shuttle conjugates siRNA payloads to transferrin receptor-targeting antibodies, exploiting the brain’s iron transport machinery to achieve uniform deep brain distribution. The competitive landscape validates this approach: AbbVie acquired brain shuttle company Aliada Therapeutics for $1.4 billion, and Regeneron/Alnylam signed a roughly $1 billion deal for CNS siRNA delivery. The tissue-of-action problem in CNS is being solved, which means genetic targets that were previously undruggable are becoming addressable.

BioAge Labs’ brain-penetrant NLRP3 inhibitor BGE-102 provides a complementary example. Phase 1 data showed 86% median reduction in hsCRP, substantially outperforming peripheral-only NLRP3 inhibitors. This suggests that central inflammation drives systemic effects more potently than peripheral inflammation alone. The tissue-of-action experiment asks “does the drug get there?” as well as “does getting there matter for the phenotype?”, and the answer to the second question tend to generate more value.

Cross-species conservation

A powerful and underappreciated form of context validation is cross-species conservation. If the same molecular signatures of a disease appear across species with shared environments and evolutionary history, the probability of causal relevance increases substantially. This is the core scientific insight behind Loyal’s approach.

Canine aging recapitulates human aging hallmarks with specific enrichments for telomere attrition (5.0×), genomic instability (2.5×), and loss of proteostasis (1.9×). Dogs and humans have cohabitated for ~23,000 years, share environmental exposures (diet, pollutants, stress), develop the same age-related diseases (cancer, cognitive decline, arthritis, cardiac disease), and show overlapping genetic determinants of lifespan: IGF1 pathway variants influence size and longevity in both species. A recent comprehensive review of canine immunosenescence confirms that dogs display the same age-related immune decline as humans: decreased naïve T-cells, inverted CD4:CD8 ratios, thymic involution, and elevated inflammatory cytokines (IL-1β, IL-6, TNF-α). Importantly, canine caloric restriction studies have already shown improved CD4:CD8 ratios and delayed thymic atrophy which directly mirror the mechanism LOY-002 is designed to exploit.

The limitations are real and worth stating clearly. Proteomic overlap between dog and human aging was weaker than transcriptomic overlap, likely due to species bias in the SomaScan assay (optimized for human proteins). The study was cross-sectional (different dogs at different ages, not the same dogs tracked longitudinally), single-breed (beagles), and relatively small (n=40). But as translational models go, the dog sits in a unique position: long-lived enough to develop genuine age-related pathology, genetically diverse enough to model human variation, and sharing the same environment as the humans it models.

Hierarchy of Evidence

Not all validation is created equal. In Part I, we described the hierarchy of target identification: human perturbational biology > human genetics > in vivo models > in vitro assays. A parallel hierarchy applies to validation evidence:

Tier 1: Orthogonal convergence in humans. Multiple independent methods of modulating the target in human systems produce the same outcome. This is the gold standard because it controls for off-target effects, model artifacts, and species differences simultaneously.

Tier 2: Human genetic validation with dose-response. Loss-of-function and gain-of-function variants in the target gene produce graded, directionally consistent phenotypes in human populations.

Tier 3: Cross-species conservation with mechanistic clarity. The same molecular signatures appear across species with shared evolutionary and environmental history, and the mechanistic pathway is well characterized.

Tier 4: Preclinical pharmacology alone. The target has been validated only in cell-based assays or animal models without human genetic support or cross-species mechanistic conservation.

Every step up this hierarchy substantially de-risks the program. Moving from Tier 4 to Tier 2 by incorporating human genetic evidence early is one of the highest-ROI activities in preclinical drug development. The hierarchy also matters for speed of iteration: Tier 1 and Tier 2 programs can fail fast on delivery or safety without wasting years debating target biology. Tier 4 programs often do not know they have the wrong target until Phase 2, by which point millions have been spent.

Three axis of target evaluation

When evaluating a prospective drug target, it’s useful to score it on three critical axes: causality, context, and controllability.

• Causality: does perturbing the target change disease (in humans ideally)?

• Context: which cell type, state, timepoint, comorbidity, ancestry?

• Controllability: can we modulate it to the required degree, safely, in the right tissue?

In practice, these three axes are interdependent. The most compelling targets check all boxes: clear causality, relevant context, and feasible controllability. If one axis is lacking, the program is riskier. For example, a target might have great human genetic causality, but if it’s an undruggable protein, that’s a dead end (unless new tech emerges). Or a highly druggable target might fail if it turns out to be involved in too many normal functions (context/safety issue). By systematically assessing causality, context, and controllability, we ensure a holistic evaluation of target viability. A successful drug target must have a strong reason to believe it drives disease, act in a way we can intervene in the patient, and be amenable to pharmacological control. These principles guide us in choosing targets that are not only scientifically interesting but also actionable and translatable to therapy.

On Killer Experiments

Once a target looks promising on paper across causality, context, and controllability, the next step is a series of “killer experiments.” These are critical tests (often done in early research) explicitly designed to kill the project if the target isn’t truly valid. Instead of seeking only confirmatory evidence, we actively stress-test the target hypothesis. By setting up rigorous experiments that a false target would fail, we can save time and money by abandoning losers early.

We evaluate targets across nine categories of killer experiment. Each one tests a specific assumption that, if wrong, would invalidate the therapeutic program. Not every target requires all nine, but every target requires honest assessment of which experiments are most informative for its specific risk profile.

Directionality Test (Agonize vs. Inhibit)

Does modulating the target in your intended direction produce the expected phenotypic change in disease-relevant systems?

The GIPR paradox is instructive here. Test the effect of both activating and inhibiting the target. This determines which direction of modulation is therapeutic.

For aging targets specifically, directionality is complicated by the fact that many pathways have context-dependent effects (e.g., mTOR activation is beneficial during development and wound healing but detrimental during aging). The directionality test must specify the disease context

Domain-Specific Modulation

Does the effect depend on modulating a specific domain, isoform, or conformation of the target?

Probe different functional domains of the target with precise tools. Some proteins have multiple activities or domains (enzymatic site, scaffolding region, etc.). Using selective molecules or mutations that affect only one domain can reveal what aspect of the target is relevant to disease.

Scholar Rock’s apitegromab exemplifies this. By targeting only latent myostatin (avoiding GDF11 and activin A cross-reactivity), it met Phase 3 primary endpoints in SMA where broad myostatin inhibitors failed in DMD. In the Phase 2 EMBRAZE trial, apitegromab combined with tirzepatide preserved 54.9% of lean mass in 24 weeks during weight loss. The same target protein, but domain specificity determined success versus failure. This experiment is particularly relevant for complex targets with multiple binding partners or conformational states.

Tissue-of-Action Test

Does the target reside in the tissue your drug can reach, and is that tissue the site of disease pathology?

Confirm the target’s key site of action.

This is the experiment Aerska’s platform is designed to answer. BioAge Labs’ brain-penetrant NLRP3 inhibitor BGE-102 provides another example: Phase 1 data showed 86% median reduction in hsCRP which substantially outperformed peripheral-only NLRP3 inhibitors. This result suggests that central inflammation drives systemic effects more potently than peripheral inflammation alone. The tissue-of-action experiment asks “does the drug get there?” as well as “does getting there matter for the phenotype?” → the answer to which can generate more value!

Dose-Response & Reversibility

Does the phenotypic effect scale with the degree of target modulation, and is it reversible when modulation stops?

Demonstrate that the phenotype follows target modulation in a dose-dependent manner and can be reversed. A strong target should show a graded response: e.g. 50% inhibition yields partial improvement, full inhibition yields max improvement. This dose–response relationship supports causality (it’s a hallmark of true cause-effect). Additionally, if turning the target off produces a phenotype, then turning it back on (or letting it recover) should reverse the phenotype.

Orthogonal Perturbation

Do independent methods of modulating the same target produce convergent outcomes?

Use multiple independent methods to modulate the target and see if they converge on the same outcome. Check out the previous 3 blogs where we went deep in this for the typology.

Target Engagement → Phenotype Link

(”PK/PD chain”)

Can you trace a quantitative chain from drug exposure to target engagement to downstream biomarker to clinical endpoint?

Show that the degree of target engagement correlates with the degree of phenotypic effect. It tells us the minimum level of target modulation required for benefit.

Ibrutinib in CLL: covalent binding to BTK Cys-481 → >95% target occupancy at clinical doses → BCR signaling suppression (ERK, NF-κB) → lymphocytosis from nodal redistribution → 71% overall response rate. Resistance mutations at the covalent binding site (C481S) prove the drug works exclusively through this mechanism. Higher trough BTK occupancy correlates with improved progression-free survival across multiple BTK inhibitors (ibrutinib, zanubrutinib, acalabrutinib). This level of mechanistic completeness enables rational dose selection, predicts resistance mechanisms, and provides regulatory confidence.

Early Safety Liabilities Scan

Does the target have essential functions in healthy tissue that would create on-mechanism toxicity?

Assess potential safety issues before investing too far. Even in early validation, we can do “safety scans” in silico and in vitro: check if the target gene is expressed in vital tissues, examine human mutations, pathway analysis etc.

The activin pathway illustrates this. ACE-031’s broad activin receptor trapping caused epistaxis and telangiectasia from BMP9/10 inhibition in DMD trials: on-mechanism toxicity from hitting receptors in healthy vascular endothelium. The same vascular biology became sotatercept’s therapeutic mechanism in PAH only after the disease context was matched to the on-target pharmacology.

Disease State vs. Healthy Differences

*(Expression with Age/Disease)*

Is the target differentially expressed or active in the disease state versus healthy tissue?

Examine how the target (or its ligand) changes with age and disease progression.

For Pfizer’s ponsegromab study in cachexia, they enrolled only patients with GDF15 ≥1,500 pg/mL to ensure the disease was mechanistically driven by the target. In Loyal’s canine aging work, the omic data provides exactly this kind of disease-state characterization. It shows which pathways are differentially active in aged versus young dogs at the molecular level before designing the intervention.

Ligand vs. Receptor Dynamics

In a ligand-receptor system, is the bottleneck the ligand, the receptor, or the signaling machinery?

Test whether targeting the ligand or the receptor yields the desired effect, and whether acute vs. sustained stimulation differ. It tests how the system responds to continuous vs. pulsatile signaling and whether we should intervene at the level of the signal or the sensor.

When is enough validation enough?

A framework this comprehensive raises a natural question: how much validation is sufficient before proceeding to clinical development? The answer depends on the tier of evidence and the cost of being wrong.

For Tier 1 and Tier 2 targets with strong human genetic causality, the primary risks are delivery and safety, not target biology. These programs can afford to move faster on validation because the core hypothesis is supported by the strongest available evidence. The killer experiments that matter most are tissue-of-action (can the drug get there?), target engagement (does it hit the target sufficiently?), and early safety scans (will on-mechanism toxicity limit the therapeutic window?). Spending two additional years validating whether the target is causal is wasteful when human genetics has already answered that question.

For Tier 3 and Tier 4 targets, the calculus reverses. Here, the risk of having the wrong target entirely is significant, and the killer experiments around orthogonal perturbation, disease-state differences, and dose-response carry the most value. The investment in thorough validation before entering clinical development is the highest-ROI expenditure available, because the alternative is discovering the target was wrong in a $200 million Phase 2 trial.

The general principle: validate enough to know where the remaining risk sits, then design the clinical program to address that risk efficiently. A program with Tier 2 evidence and a clear tissue-of-action plan is ready for IND-enabling studies. A program with only Tier 4 evidence and untested directionality is not. The killer experiments table is not a checklist to complete exhaustively, it is more like a tool for identifying which assumptions carry the most risk for a given target and testing those assumptions first.

This is the fourth piece in the “Betting on Biology” series. In our last piece of the series next week, we’ll analyse the strategic decisions coming out of target identification and validation. Essentially, what should we do with this information? What factors should we consider and how are they influenced by the target discovery typology? See you soon.

Acknowledgements

A big thank you to Satvik Dasariraju for the inspiration, thoughtful comments, and prompt answers to my many questions; to Alex Colville for invaluable writing guidance throughout the process; and to every age1 crew for helpful pointers. Special thanks to Aerska, Loyal, and the amazing founders for the inspiration. Cheers!

This speed matters commercially, but not because obesity represents a “post-GLP-1 market” as GLP-1 receptor agonists currently dominate obesity pharmacotherapy. Rather, the market now demands targets that address three specific unmet needs GLP-1s cannot solve: (1) the 9-27% non-responder rate, (2) the 25-40% lean mass loss that accompanies fat reduction that is particularly problematic in older adults where sarcopenia compounds frailty risk, and (3) the 37-81% one-year discontinuation rate driven by cost, insurance type, comorbidities, and absence of type 2 diabetes etc. These gaps create commercial space for mechanistically distinct targets, particularly those addressing adipocyte biology, metabolic memory, or muscle-sparing pathways.

The platform choice reveals a testable hypothesis about chronic, age-related disease target identification: that physiological context is a NEED. Traditional 2D adipocyte cultures fail to recapitulate the insulin resistance phenotype central to metabolic disease; co-culture with macrophages in 3D systems is required to observe reduced glucose uptake and impaired GLUT4 translocation characteristic of inflamed adipose tissue. More fundamentally, in vitro systems, even sophisticated vascularized 3D adipose tissue models cannot fully replicate the in vivo environment as the constitutive elements (hormones, cytokines) exist at fixed concentrations rather than the dynamic, responsive levels found in living tissue.

We need to think about whether this gap between simplified systems and living biology actually matters for target identification. For some diseases, it often does not. Cancer cell survival screens, ion channel pharmacology, and receptor-ligand binding assays work in isolated systems because the relevant biology operates at the molecular level. The target either blocks a specific enzymatic reaction or it does not, the complex tissue environment in these circumstances adds noise instead of signal.

Age-related metabolic diseases may occupy a different territory. Visceral adipose tissue secretes leptin that regulates hypothalamic appetite circuits, releases free fatty acids that drive hepatic insulin resistance, recruits macrophages that sustain inflammatory signaling through TNF-α and IL-6, and responds to sympathetic innervation controlling lipolysis. An adipocyte on plastic loses mechanical tension from extracellular matrix, paracrine signals from neighboring pre-adipocytes and immune cells, endocrine inputs from distant organs, and vascular perfusion delivering nutrients and immune factors. Whether these features are signal (revealing therapeutic targets invisible in reductionist systems) or noise (obscuring molecular mechanisms) determines which experimental platform can distinguish causal drivers from the vastly larger set of genes that correlate with disease readouts in simplified models but prove irrelevant in human pathophysiology.

Three strategies have emerged attempting to preserve physiological context:

• In vivo perturbation at scale

• Human primary tissue perturbation platforms

• Comparative and evolutionary biology

What follows is a systematic analysis of these three approaches: the evidence supporting each, the failure modes that undermine them, and the criteria for determining when physiological context genuinely improves target identification versus when it adds cost and complexity without increasing the probability of finding causal drivers.

In Vivo Perturbation at Scale

As we defined perturbation at scale in our last piece, this method highlights the in vivo element. In practice, this means delivering a library of perturbations (e.g. CRISPR guides, shRNAs or cDNAs) into a disease-relevant model, then measuring each cell’s “barcode” and phenotype (often by single-cell RNA sequencing) to link interventions to outcomes.

Gordian Biotechnology has pioneered this “mosaic” tissue approach to test hundreds of gene therapies simultaneously within a single animal. The platform’s success really comes down to three things:

• Pool delivery: Gordian injects hundreds of AAV vectors (each targeting a specific gene) at low dose into the tissue of an aged or diseased animal. The low multiplicity of infection ensures that most cells pick up only one perturbation, creating a true mosaic of treated and untreated cells.

• Tissue specificity: Each viral construct uses cell-type-specific promoters or serotypes so that only the desired cells are edited.

• Barcode tracking: Each vector carries a DNA or RNA barcode unique to its perturbation. After treatment, the tissue is dissociated and subjected to single-cell (or single-nucleus) RNA sequencing. In the sequencing data, the barcode reveals which perturbation each cell received, while the transcriptome reports that cell’s response.

This high-resolution readout offers two distinct benefits:

1. Cellular resolution: It distinguishes effects across subpopulations (e.g., separating mature adipocytes from stem cells or infiltrating macrophages) within the same tissue.

2. Rich phenotypic mapping: scRNA-seq captures shifts in thousands of genes instead of simple survival. This allows for deep analysis of pathway-level readouts, such as inflammation, insulin signaling, and fibrotic remodeling.

Gordian applied this platform to identify targets in obese, aged mice on a long-term high-fat (”GAN”) diet. By targeting hundreds of genes in visceral fat, they identified interventions that shifted the aged transcriptome toward a “rejuvenated” state. Knocking down certain genes induced a ‘beiging’ of fat cells and improved insulin signaling, and other targets reduced lipid synthesis or inflammation in the tissue.

Limitations of the approach:

• Interpreting “rejuvenation”

  • As delineated in Part II of the series, signatures of rejuvenation need to be analysed more critically. A “more youthful” transcriptome might be defined by comparison to young healthy tissue (lower levels of inflammatory and senescence genes, etc.). However, the signatures risk failing to improve organismal health or lifespan when tested long-term. A perturbation might down-regulate certain age markers simply by activating stress responses or by causing cell turnover, without truly restoring function.

• Delivery

  • In vivo screens rely on viral vectors or nanoparticles with limited tropism (e.g. poor blood-brain barrier crossing, liver accumulation, exclusion from large or fibrotic tissues), and far fewer cells can be transduced or recovered than in vitro, limiting library size and statistical power.

  • The host immune system can neutralize vectors or eliminate perturbed cells via pre-existing anti-AAV antibodies or immune responses to Cas9, reporters, or newly introduced proteins, causing loss of signal and unintended inflammation.

• Cellular heterogeneity

  • Transduction is invariably uneven, some cell types or regions get many perturbations while others get none. After treatment, tissue dissociation and single-cell capture introduce bias: certain fragile or large cells (e.g. aged neurons, adipocytes) may die or fail to encapsulate, so the sequenced population is not a random sample of the in vivo pool.

  • Likewise, droplet-based scRNA-seq only captures a fraction of each cell’s mRNAs (including the perturbation barcode), so many perturbed cells may yield no readable guide sequence. Separately, even if a guide RNA is detected, the genome editing it induces may be incomplete: cells may have 0, 1 or 2 alleles cut, and there is no easy single-cell readout of editing efficiency.

• Model-dependent limitations

  • Mice differ from humans in immunology, metabolism, gene lifespans, etc. In a notable example, all human myeloid leukemia-specific targets would be missed by mouse screens simply because the relevant gene is not expressed in mice (Consider: humanized systems?)

How to stress-test this approach?

1. Dose-responsiveness: A true causal target generally shows graded effects with perturbation strength (e.g. varying AAV dose or guide efficiency).

2. Orthogonal assays: Said it in other parts of the series, just can not emphasize enough. For example, if a CRISPR knockout yields a beneficial transcriptional shift, demonstrate that restoring the gene’s expression (genetic rescue) or using an independent modality (e.g. RNAi or a small-molecule ligand) produces the opposite effect. Such paired perturb-and-rescue helps confirm on-target action. To use the Gordian example, they followed up promising hits by administering them (as single-gene gene therapies) at full dose and looking for actual phenotypic improvements (e.g. better glucose tolerance, reduced fibrosis, improved heart function). Only if an intervention not only reprograms the gene expression profile and yields measurable functional gains can one be confident it represents a genuine rejuvenation. We need to emphasize functional endpoints.

3. Cross-validate in different models: Validating hits in human-derived organoids or primary tissues is particularly essential to confirm that the regulatory networks are evolutionarily conserved. Ultimately, triangulating results across diverse genetic backgrounds helps determine if a hit is a universal driver or merely dependent on a specific genetic modifier landscape, which is crucial for predicting success in a heterogeneous patient population.

Ex-vivo Human Primary Tissue Perturbation Platform

Human primary tissue perturbation platforms use real human tissues (e.g. donated organs, biopsies, or primary cell cultures) as the testing ground for potential treatments.

*Aim: Identify targets that have immediate human relevance

Fresh human tissue, often 100 to 500 µm thick, is cultured under oxygenated conditions to maintain its 3D structure, native extracellular matrix, and all the types of cells it contains. Ochre Bio, a startup focused on liver disease drug discovery, has created a platform based on this concept. It generates large-scale perturbation profiles in human livers and hepatocytes, including genome-wide gene knockdown in primary hepatocytes and single-cell profiling of perfused donor livers, to facilitate target discovery. These platforms fall between simple in vitro tests and animal studies. They maintain human-specific multicellular complexity while allowing for experimental control. Many liver and lung diseases are highly species-specific, so these models fill an important gap in perturbational biology by providing validation in a human context. This strategy already produced two rich datasets with GSK: one mapping gene knockdowns to expression changes in human hepatocytes, and another profiling the cell-type-specific gene expression of perfused diseased livers.

Limitations of the approach

• Ethical and regulatory constraints (especially for solid organs)

  • Access to human organs is limited to surgical waste, transplant discards, or end-stage diseased tissue, because sampling healthy organs or performing repeated biopsies for research is ethically impermissible. Consequently, most platforms operate on late-stage, highly perturbed biology rather than early causal disease states, introducing systematic sampling bias.

  • Informed-consent, privacy, and biosafety regulations restrict permissible perturbations, longitudinal studies, and data sharing (e.g. limits on genome-wide editing, chronic culture, or clinical linkage). These constraints truncate experimental time horizons and narrow the searchable biological space independently of technical feasibility.

• Limited viability and throughput

  • Slices typically remain viable for only a few days (often 2-5 days) before hypoxia and nutrient gradients degrade function. Longer assays are infeasible. This contrasts with permanent cell lines or organoids that can be passaged indefinitely. Likewise, because each experiment requires fresh human tissue, throughput is low: only tens of slices can be prepared per sample, and scaling to hundreds of perturbations is challenging.

• Tissue access

  • Reliable access to fresh human tissue can be a bottleneck (need donor organ or surgical samples). Samples come from individual patients, so biological variability is high. Intrinsic heterogeneity (cell composition, fibrosis grade, tumor content) means inter-slice and inter-donor reproducibility is limited. For example, the same tumor sliced from different areas or donors can yield different responses, making trends harder to detect. This “real-world” variability must be controlled by using multiple donors and replicates.

• Delivery constraints

  • Genetic perturbation in thick tissue is nontrivial. Transducing all cells with CRISPR or shRNA vectors may be inefficient or uneven, especially in the core of a slice. Similarly, large biologics or nanoparticles may penetrate poorly. Slice size (few hundred µm) is a tradeoff: thick enough to preserve cells, but thin enough for diffusion. In practice, diffusion limits mean that drug delivery and genetic editing efficacy can vary across the slice.

• Experimental complexity

  • Because slices are complex, data analysis is more challenging. Background cell heterogeneity, matrix autofluorescence, and variable oxygenation all add noise. Multimodal assays (scRNA-seq, imaging, ELISA, etc.) must be carefully controlled. Moreover, standardization is difficult: there are few consensus protocols for slice culture and analysis, leading to batch-to-batch variability.

All these factors limit reproducibility compared to simpler systems.

How to stress-test this approach?

1. Benchmark against known human data: Wherever possible, compare slice responses to existing in vivo or clinical data. For example, test a perturbation with a known human outcome: if an antifibrotic drug ameliorates markers in slices similarly to patient or animal data, that builds confidence. In a liver PCLS study, the EGFR inhibitor erlotinib produced the same antifibrotic gene-expression changes in slices as observed in prior cirrhosis models. Likewise, demonstrating a slice’s sensitivity to a standard therapy (e.g. the mTOR inhibitor rapamycin reduced PDAC slice proliferation) shows the model can recapitulate expected drug effects.

2. Replicate across donors and timepoints: True effects should persist across tissue from multiple patients and across independent cultures. Time-course analysis helps ensure the effect is not a transient stress artifact. Lack of signal reproducibility is a red flag for a technical artifact.

3. Rescue assays: Whenever possible, perform complementary “rescue” experiments. For instance, if knocking out Gene X in slices induces pathology, reintroducing a wild-type copy or adding a pathway agonist should reverse the effect. Conversely, applying an independent perturbation (e.g. a small molecule versus genetic KO) targeting the same pathway provides convergence. Orthogonal readouts are also critical: verify gene expression changes with protein-level assays or histology. In practice, slice studies measure both molecular and functional endpoints. For example, drug-treated liver slices might show downregulation of proliferation genes by RNA-seq and decreased Ki67 staining by IHC, or altered metabolite levels by mass spectrometry. Such multi-layer validation (gene, protein, phenotype) guards against false positives.

4. Compare to clinical endpoints: When possible, link ex vivo findings to patient data. For example, if a slice perturbation upregulates a fibrosis gene signature, check if that signature predicts fibrosis severity in patient cohorts or is known from GWAS. Aligning slice biomarker changes with human pathology strengthens causal claims. This might involve correlating slice-derived profiles with public datasets or biobank samples.

Comparative/Evolutionary Biology

Comparative and evolutionary biology-driven discovery mines the traits of long-lived or disease-resistant species (and extraordinary humans) to find mechanisms that could be turned into therapies by examining comparative gene expression, regulatory regions, protein modifications, evolutionary conservation, stress-response pathways etc.

The premise is that nature has already solved many problems: some animals naturally don’t get diseases that afflict humans, or survive conditions (freezing, low oxygen, radiation) that would normally kill human cells. The underlying genes, proteins, or pathways responsible can be identified and then mimicked or activated in humans.

• Cancer resistance: An elephant has roughly 100-fold more cells than a human and lives as long, yet its lifetime cancer mortality is only ~5% (versus up to ~20-25% in humans). The discrepancy is known as Peto’s paradox. A major reason appears to be an evolutionary expansion of the tumor-suppressor gene TP53. Whereas humans have a single TP53 gene, elephants carry at least 20 copies (including 19 retrotransposed duplicates) that are actually expressed. Extra TP53 means extra dosing of p53 protein guarding the cell and elephant cells undergo apoptosis in the face of DNA damage at roughly double the rate of human cells, before cancer develops. Now the translational step is to harness this mechanism for humans. Peel Therapeutics, for example, is developing therapies around elephant p53 (“EP53”) to recapitulate this hyper-apoptotic, cancer-suppressing phenotype in human tissues.

• Stress tolerance: Cross-species genomics is yielding therapeutic insights as well. Fauna Bio, for example, studies hibernating mammals to find factors that could protect human organs from disease, and has established strategic collaboration with Eli Lilly for its platform. Hibernators like the 13-lined ground squirrel endure extreme metabolic suppression and even near-freezing body temperatures for long periods, entering states akin to severe ischemia, but their organs (heart, brain, kidneys) exit torpor undamaged once they rewarm. By comparing gene expression in such animals, Fauna’s team identified one gene, code-named Faun269g, that stood out as a driver of multi-organ stress tolerance. By studying how a hibernating species naturally avoids heart failure under harsh conditions, researchers uncovered a promising drug target (and a following lead compound) now moving through preclinical development.

• Musculoskeletal health: Many adaptations identified through evolution already have built-in safety and efficacy. Hibernating grizzly bears provide a robust proof of concept for osteoporosis treatment by maintaining structural bone integrity through a biological mechanism that prevents the disuse-atrophy seen in other mammals. They bypass the “use it or lose it” paradigm by reducing cortical bone turnover activation frequency by 75% while maintaining a strict homeostatic balance between formation and resorption. This balanced turnover, likely mediated by PTH signaling that prevents osteocyte apoptosis and suppresses osteoclastogenesis, validated sclerostin as a high-confidence target for therapies like Romosozumab.

Comparative screens can also flag mechanisms that never vary in our species (e.g. an embryonically lethal tweak in humans). A comparative genomics analysis of 57 mammals, for instance, linked longevity to enhanced protein stability and demonstrated that such comparative hits reveal human lifespan biology. Additionally, if independent long-lived lineages share a genomic feature, it’s likely fundamental. For example, the insulin/IGF-1 pathway is repeatedly implicated in lifespan regulation from worms to whales.

Limitations of the approach

• Phylogenetic gaps

  • Animals vary enormously in lifespan and life history. Comparing organisms whose lifespans differ by orders of magnitude (a 2-day mayfly versus a 200-year whale) adds statistical and conceptual complexity. It can be hard to align aging metrics or isolate the right comparisons, so findings may reflect confounders (body size, metabolic rate, etc.) rather than causal mechanisms.

• Model tractability

  • Many species with exceptional biology are difficult to study. Large or long-lived animals (whales, elephants) pose obvious husbandry challenges, while even small exotic models (naked mole-rats, bats) have specialized needs. There is a difficulty in finding animal models that both emulate human aging processes and are practical for the lab. If a candidate adaptation can’t be experimentally tested (for example, because no lab animal shares it), its value remains speculative.

• Translation gap

  • Not all evolutionary solutions map to humans. Many longevity-enhancing variants in other mammals have no human equivalent, or were fixed (and invisible) in our lineage. One comparative-genomics study found little overlap between genes identified in long-lived mammals and human GWAS signals for lifespan. In other words, a “hit” from an animal screen might not be druggable in people or might simply reflect evolutionary chance.

• Context specificity

  • Adaptations often serve species-specific niches. For example, a gene that protects a burrowing rodent from low oxygen has no obvious parallel in human biology. Comparative hits must be evaluated in context: some mechanisms may be irrelevant outside the original evolutionary setting. In short, this approach works best when it focuses on broad, conserved processes (e.g. DNA repair, metabolism) and is less useful for pathologies rooted in uniquely human traits.

How to stress-test this approach?

1. Phylogenetic replication: Demand that a candidate emerges independently in multiple lineages. Rigorous screens can compare the top and bottom deciles of species by lifespan or stress-resistance, then require that each candidate feature is fixed in all “long-lived” groups and absent (or different) in “short-lived” ones. Permutation or phylogenetic ANOVA tests are then applied to confirm significance, ensuring the signal isn’t due to shared ancestry or chance.

2. Human-data alignment: Check whether the candidate gene or pathway shows any relation to human aging or disease. If possible, examine human genetic or clinical data: do variants in the target correlate with healthspan? Is the gene expressed or dysregulated in patient tissues? This cross-check guards against chasing species-specific quirks.

3. Functional validation: Test the target in relevant living models. For a gene identified in elephants or hibernators, one might use CRISPR or drugs in human cells or mice to mimic the predicted effect. For example, overexpress the long-lived species’ version of the gene (or knock down its human ortholog) and measure outcomes: cell stress resistance, metabolic function, or neuronal survival. Success in multiple assays increases confidence that the target truly drives the desired effect.

4. Diverse phenotypic assays: Use *orthogonal readouts. If a target is supposed to improve metabolism, verify effects on both molecular markers (insulin sensitivity, mitochondrial function) and organismal health (weight, endurance) in animals. For neurodegenerative targets, check both biochemical markers (protein aggregation, synaptic function) and behavior (cognitive tests). Converging evidence from different angles is the final proof that a comparative hit is biologically meaningful.

*You might have noticed this has come up a lot. It is fundamental and logical to validate typologies orthogonally and it highlights and acknowledges the importance of context.

This is the third piece in the “Betting on Biology” series. At this point, we have talked about the major target identification approaches. In the next piece, we will think about the important questions to ask in deciding the targets: How should we test directionality? How do we prioritize evidence? What are the killer tests to consider? See you soon.

Acknowledgements

A big thank you to Satvik Dasariraju for the inspiration, thoughtful comments, and prompt answers to my many questions; to Alex Colville for invaluable writing guidance throughout the process; and to every age1 crew for helpful pointers. Special thanks to Gordian, Fauna, and the amazing founders for the inspiration. Cheers!

In biology, a perturbation is any intentional change made to a cell, like turning a gene off, overexpressing a protein, or adding a drug. Perturbation at scale is the transition from doing this to one gene at a time to doing it to every gene in the genome simultaneously. By using tools such as CRISPR to knock down thousands of genes across millions of individual cells and then sequencing the results, we create a high-resolution data point for every possible intervention. We can test every possibility at once, quickly assess causal hypotheses and rule out what doesn’t work. A useful perturbation dataset should be evaluated via:

1. Breadth: Genome-wide or near-genome-wide perturbation coverage

2. Depth: High-content readouts (single-cell transcriptomics, epigenomics, spatial or functional phenotypes)

3. Context: Perturbations resolved across cell types, states, and increasingly, combinatorial or “combo” modalities

In early January, Gordian Biotechnology announced a research collaboration with Pfizer to apply its large-scale in vivo mosaic screening platform toward obesity target discovery, directly interrogating hundreds of gene perturbations within living adipose tissue. On January 20, 2026, Parse Biosciences and Graph Therapeutics announced a partnership to construct a large-scale functional immune perturbation atlas, combining pooled genetic perturbations with single-cell readouts across immune states. Just days earlier, the Norman lab reported the largest exhaustive genetic interaction map ever built in human cells: 665,856 pairwise perturbations across 46 million clonal lineages, and critically, the first map at this scale using measurable traits other than whether a cell lives or grows. Adding to this momentum, Illumina introduced the “Billion Cell Atlas“ on January 13, a genome-wide dataset designed to capture how one billion individual cells respond to CRISPR-driven genetic changes in 200+ cell lines. These news and datasets are transformative because they shift the focus from cell survival to cell state, a cell’s functional identity. After getting single-cell sequencing of a perturbed cell, you gain two major useful capabilities:

1. Navigation: We can treat cell identity as a position on a map. This allows us to identify exactly which interventions shift a cell from a “diseased” toward a “healthy” reference state, or towards a different identity, assuming that a priori healthy vs. old cells from primary samples have been collected.

2. Prediction: Because these maps capture millions of interactions simultaneously, they serve as a predictive model of the cell’s internal logic in the long run. This “searchable index” holds complex biological information that is far too non-linear and emergent to be derived using traditional math or differential equations.

The idea itself is not new. The 1979-1980 Heidelberg screens in Drosophila pioneered systematic mutagenesis, then in the 1980s, High-throughput screening (HTS) industrialized chemical perturbation in pharma, scaled up from 800 compounds per week in 1986 to 7,200 per week by 1989 in Pfizer. Genome-wide RNAi libraries followed in the 2000s, and CRISPR-Cas9 screens in the 2010s expanded the search space further, yielding bona fide clinical targets. Research utilizing the GeCKO (Genome-scale CRISPR-Cas9 Knockout) library successfully identified CUL3 (Cullin 3) as a high-ranking candidate gene whose loss gives resistance to the BRAF inhibitor vemurafenib in melanoma models. By perturbing gene function at the DNA level, researchers demonstrated that the complete loss of CUL3 function significantly alters cellular responses to targeted therapies. These findings illuminate the importance of the neddylation pathway, which is required for the activation of Cullin proteins like CUL3, and provides a genetic rationale for investigating therapeutic inhibitors like pevonedistat that target this specific regulatory mechanism. In parallel, the Connectivity Map (Broad Institute, 2006) demonstrated that cell states themselves could be profiled and indexed at scale.

What is different now is not just throughput, but dimensionality. We have crossed a phase boundary: from perturbation as a sparse experiment to perturbation as a high-resolution map of causal biology. This shift reframes the core bottleneck in drug discovery. The problem is no longer a shortage of targets, rather a shortage of high-dimensional understanding of those targets. What does this gene do in a human cell? In a diseased tissue? Under aging, stress, or immune activation? Will modulating it repair the system, or trigger compensatory failure?

Modern automated platforms can now identify potential drug targets 10,000 times faster than traditional methods. Additionally, the “evidence base” for target-disease associations has exploded. Since 2015, the accumulation of supportive biomedical data driven by GWAS, CRISPR screens, and single-cell “omics” has reached a scale where nearly 30 million pieces of evidence now exist in open platforms. Despite this massive influx of candidates, the rate of FDA approvals has not followed suit, with approvals for New Molecular Entities largely fluctuated between 35 and 55 annually for the past decade.

[*Conceptual graphic]

What remains scarce, and increasingly decisive, is the ability to interpret perturbations, extract causal structure from high-dimensional data, and select the right lead from an overwhelming search space.

This essay argues that perturbation at scale is not just a faster way to find targets. It is a fundamental reorganization of how we learn biology, and a necessary response to the widening gap between experimental capacity and causal understanding.

1. Transcriptomic state-shift screens

Transcriptomic state-shift screening uses large-scale gene expression measurements to identify perturbations that move cells from an undesired biological state toward a predefined target state. The core idea is to treat cell identity as a position in a transcriptomic state space: a low-dimensional manifold learned from gene expression data, such as an axis spanning “aged” to “youthful” or diseased to healthy cellular programs.

In practice, researchers first construct a reference manifold of cellular states using single-cell RNA sequencing (scRNA-seq). Because human aging and disease are driven by diverse epigenetic codes across various genetic backgrounds, this requires collecting primary tissue samples from a broad range of healthy, diseased, and aged individuals. It is necessary to capture real biological variation rather than literature-based assumptions from immortalized cell lines, and ensure the resulting models are grounded in human biology. To maintain a clear signal, the cell type must be held constant across these samples. Thousands of candidate perturbations, such as pooled transcription-factor overexpression libraries, CRISPRa/CRISPRi guides, or small molecules, are then introduced in parallel. However, these exhaustive screens are typically not performed on the original primary donor samples, they are carried out instead in more scalable, ‘accessible’ systems like a curated set of disease-relevant cell lines. Primary cells are generally too scarce, fragile, and short-lived to survive the massive throughput required to test millions of combinations. Each cell’s transcriptome and perturbation identity are jointly captured, typically via pooled single-cell readouts.

By comparing the perturbed cells to our healthy reference, we can see exactly which interventions push a diseased cell back towards a healthy state. The effect of a perturbation is quantified using metrics such as transcriptomic distance, state displacement, or a model-derived aging score.

For example, NewLimit’s pooled reprogramming screens evaluate whether specific transcription-factor combinations shift aged cells toward a youthful transcriptomic profile. Hits are perturbations that reproducibly and significantly reduce inferred transcriptomic age or induce the target gene expression.

Transcriptomic state-shift screens provide an information-dense readout of perturbation effects. Single-cell resolution resolves heterogeneous cellular responses and avoids false negatives caused by bulk averaging, while enabling thousands of perturbations to be assayed in a single pooled experiment, with each cell effectively serving as an independent test.

As a result, these screens scale efficiently to very large perturbation spaces. For example, NewLimit reports pooled evaluation of >4,000 transcription-factor combinations within a single experimental framework. The resulting high-dimensional expression profiles can be quantitatively analyzed using statistical or machine-learning models to rank perturbations by the magnitude and consistency of their induced state shift.

Crucially, transcriptome-based hits can uncover previously unappreciated biology. Rather than relying on proxy phenotypes or viability effects, the approach directly measures whether perturbing a given target drives cells toward a particular molecular state, establishing a causal link between target modulation and the desired transcriptomic outcome. This strategy allows researchers to go beyond the inherent limitations of human genetics, which is fundamentally filtered by natural selection. Traditional genomic studies are powerful, but they are restricted to variants that have successfully persisted through evolution. Many of the most powerful biological levers remain undiscovered because they are tied to evolutionary constraints or developmental necessities. For example, a mutation that could theoretically reverse an adult disease state might never be observed in human populations because it would be embryonic lethal during early development. Furthermore, evolution does not “trace” every possible genetic combination, which leaves vast regions of the biological landscape unexplored.

Limitations of the approach

• Transcriptome ≠ function

  • Changes in gene expression profiles do not always translate to improved cellular or organismal function. An intervention may make a cell’s transcriptomic signature resemble a desired state (e.g. a “younger” or less diseased profile) without fixing underlying functional deficits. In aging research, many treatments targeting single molecular hallmarks altered gene expression but failed to increase lifespan or restore health in vivo.

  • Similarly, in immune modulation, simply reversing exhaustion-associated transcripts in T cells is not guaranteed to restore their antitumor function. It remains challenging to assess whether altered exhaustion markers truly correlate with durable functional recovery.

• Undesired coupling of state changes

  • The desired state shift can be coupled to unwanted cellular changes. A prominent example is partial cellular reprogramming to reverse aging: while short-term OSKM factor expression can reset epigenetic age and gene expression, it also pushes cells toward a pluripotent-like state. This dedifferentiation erodes cell identity and can lead to uncontrolled proliferation.

• In vitro hits vs. in vivo efficacy

  • Transcriptome-based screens often identify perturbations that look promising in cultured cells but falter in living organisms. One reason is that complex physiological contexts are not captured in vitro, cell-line gene expression responses may not replicate in tissues or patients. Notably, despite the excitement around reprogramming factors and chemical cocktails that reverse cellular age in vitro, only a single peer-reviewed study from Rejuvenate Bio thus far has shown a significant lifespan extension in normal adult animals using these methods.

• Logistical constraints

  • High-resolution mapping currently relies on specific physical requirements. First, we need high-quality “reference” cells; while we can biopsy living tissue, collecting from autopsies is difficult because cell states begin to degrade immediately after death. Second, once we identify a gene that can shift a cell state, we must be able to actually reach that cell in a patient. This limits the current impact of these screens to cell types where we already have reliable ways to deliver a drug, like those in the blood, liver, or eye.

How to stress-test this approach?

Transcriptomic hits require rigorous downstream validation to confirm that molecular changes translate into genuine functional benefits. Key tests include:

(1) Functional Assays: Does perturbation actually make old or diseased cells act young or healthy? For example, a hit from a gene expression screen should be tested in a functional readout like regenerative capacity, stress resistance, or a disease-specific behavior. NewLimit followed this principle by taking a top TF hit into animal models - formulating it as an LNP-mRNA therapy and showing it restored liver regeneration in old mice. Such an in vivo rescue of function is the ultimate proof that the transcriptomic shift was meaningful.

(2) Orthogonal Markers: Check other layers of biology, e.g., does the intervention also reduce epigenetic age or improve protein-level biomarkers? Consistency across orthogonal “youth” metrics builds confidence that the target has true causal power.

(3) Specificity Checks: Ensure the perturbation isn’t causing broad stress or proliferation that masquerades as a rejuvenation signature. This can be done by measuring whether cell-type identity markers remain intact (no unwanted lineage changes), and by looking for adverse transcriptional programs (p53 activation, inflammation) in the single-cell data.

(4) Time-course durability: A durable effect is more likely to be therapeutically relevant so evaluate if the desired expression changes persist after the perturbation is withdrawn.

2. Phenotypic screens

Phenotypic screening interrogates biological targets indirectly by measuring high-level cellular outcomes (phenotypes) rather than directly assaying a specific molecular species (e.g., a protein’s enzymatic activity or a transcript’s abundance). A phenotype here refers to observable cellular properties such as morphology, organelle organization, growth, survival, or reporter activity.

In practice, phenotypic screening often relies on automated, high-content assays that can capture many features simultaneously across thousands to millions of perturbations. A canonical example is the platform developed by Recursion, which makes extensive use of the Cell Painting assay. In this approach, cells are exposed to perturbagens, such as small molecules, CRISPR-based gene knockouts, or RNA interference, then stained with a standardized set of fluorescent dyes that label major cellular compartments. Cells are imaged at scale using automated microscopy.

Computer vision and machine-learning models extract hundreds to thousands of quantitative features per cell, describing aspects of size, shape, texture, intensity, and spatial organization. The result is a high-dimensional phenotypic signature for each perturbation.

These high-dimensional phenotypic profiles are typically embedded into a lower-dimensional latent space using dimensionality-reduction or representation-learning methods. Perturbations that induce similar cellular changes cluster together in this space. At Recursion, this strategy has been applied to build a large-scale cellular “data atlas” comprising hundreds of millions of images across dozens of human cell types, spanning both genetic perturbations (e.g., gene knockouts) and chemical treatments.

Mining such an atlas enables mechanistic inference by similarity. For example, if an uncharacterized compound induces a phenotype that clusters with the knockout of a gene from a known pathway, this suggests that the compound may act on that pathway or on a molecular target upstream or downstream of it. Importantly, this inference is probabilistic: phenotypic similarity constrains the space of plausible mechanisms but does not, on its own, prove direct target engagement.

Not all phenotypic screens rely on rich imaging. Some focus on more targeted functional readouts, such as cell viability, proliferation rate, or reporter-gene expression under defined conditions (e.g., metabolic stress, drug resistance, or inflammatory signaling). These assays still qualify as phenotypic because they measure an integrated cellular response rather than a single molecular interaction.

A common strategy for deconvolving phenotypic hits (i.e., linking an observed phenotype to an underlying mechanism) is signature mapping. In this approach, the phenotype of a hit (whether an image-based fingerprint or a gene-expression profile) is compared against a reference library of signatures derived from perturbations with known targets or mechanisms. The original Connectivity Map applied this concept to transcriptional profiles; analogous logic is now widely used with imaging-based phenotypic data.

Another powerful deconvolution method is resistance profiling. When a small molecule produces a strong phenotypic effect but its target is unknown, researchers can perform a secondary genetic screen (often using a genome-wide CRISPR knockout library) to identify genes whose loss suppresses or abrogates the compound’s effect. Genes that confer resistance are frequently directly explained by, or functionally connected to, the compound’s mechanism of action.

Conceptually, phenotypic screening asks: does any perturbation make cells look or behave more normally under a given condition? Mechanistic insight is then obtained not from the primary screen itself, but from downstream pattern recognition, comparative signature analysis, and targeted secondary experiments.

Phenotypic screening is most effective when the biological objective is already proven, but the chemical path to reach it is unknown. The development of risdiplam (for Spinal Muscular Atrophy) follows this logic. Before risdiplam, nusinersen (an ASO) and zolgensma (a gene therapy) had proved the “objective function.” They showed that increasing SMN protein levels in patients significantly improved motor function. While successful, they required complex delivery: nusinersen requires lifelong, repeated injections into the spinal fluid (intrathecal), and zolgensma is a one-time viral gene replacement delivered via IV. The industry wanted a small molecule (a pill) that could be taken at home and distributed systemically. The problem was, The SMN protein itself lacks a traditional “pocket” for a small molecule to bind to, making it a difficult target for conventional drug design. Instead of trying to engineer a stabilizer for an “undruggable” protein, researchers used a phenotypic screen to find a small molecule that could achieve the same outcome as nusinersen: correcting SMN2 splicing. They set the objective to match the success of the previous therapies and let the screen find the lever.

High-content imaging in particular captures an integrative picture of cell state: subtle changes in organelle shape, texture, or cell size can reflect pathway perturbations that genomic assays might miss. Modern machine learning (ML) allows extraction of these subtle phenotypes and clustering of related perturbations far beyond human visual ability. Another strength is scalability: platforms like Recursion’s can test millions of perturbation-condition combinations and profile each with thousands of features. This yields an extensive phenotypic map that can be mined repeatedly (a “searchable” index of biology). Also, phenotypic hits are directly linked to functional outcomes by definition, e.g., a compound that restores a diseased cell morphology or survival has already passed a kind of functional test in vitro. This tends to prioritize biologically relevant effects.

When combined with reference perturbations, phenotypic assays can also suggest mechanisms: Recursion demonstrated that morphologically similar knockouts often belong to the same pathway or complex (e.g., knocking out JAK1, STAT3, and other IL-6 signaling components all produced closely related phenotypic signatures). Thus, phenotypic screening can both identify and begin to mechanistically annotate novel targets.

Limitations of the approach

• Ambiguity

  • Visual or functional phenotypes are often not specific. Many mechanistically distinct treatments cause overlapping phenotypic changes. For example, any severe stress or toxic insult may make cells round up or detach, a phenotype that doesn’t pinpoint a unique target. Such “mechanistic ambiguity” means the same cellular outcome (e.g. apoptosis, cell arrest) could result from very different molecular triggers.

  • Large-scale cell imaging often catches false signals from dust, lint, or blurry focus. Some chemicals also glow or dim on their own, making them look like a “hit” when they aren’t. Also, dead cells can trick the computer because of their morphological changes.

• Cell type/model specificity

  • A phenotype observed in a particular cell line (often aneuploid cancer cells or immortalized lines) may not translate to primary cells or in vivo tissue.

  • ML models in cell testing can be tricked by hidden patterns, like which batch a sample came from or where it sat on a tray. If these aren’t managed, the computer learns to group samples by these accidental setup differences instead of their actual biological changes.

  • Phenotypic profiles can cluster compounds with similar action, they do not automatically reveal the molecular target. This is the flip side of not starting from a hypothesis.

• “Common pathways”

  • Phenotypic screening can yield hits acting on known pathways like generic stress response that might alleviate multiple cellular problems transiently without truly addressing disease drivers.

  • Certain mechanisms recur across phenotypic screens because they universally perturb cells. For example, HDAC inhibitors, bromodomain (BRD4) inhibitors, mTOR pathway inhibitors, tubulin disruptors, and mitochondrial poisons are famous for appearing as hits in many phenotypic assays. Modulating these targets produces strong phenotypes (changes in gene expression, cell cycle arrest, cell death, etc.) in a wide range of contexts.

How to stress-test this approach?

(1) Reproducibility and specificity: Re-test the hit in the same assay across independent experiments and controls. Does the phenotype consistently reproduce? Does the intervention only correct the intended phenotype or does it also cause other aberrations? For example, if a small molecule normalizes cell morphology, one would verify it doesn’t also, say, trigger massive cell death or stress pathways as side-effects (using additional stains or assays).

(2) Mechanism deconvolution: Apply signature mapping or resistance profiling to nail down the target. If the hit is a compound, perform a genome-wide CRISPR resistance screen: if cells lacking protein X become immune to the drug’s effect, that strongly implicates X as the target or pathway node. Conversely, if no clear resistance emerges or if multiple unrelated genes confer resistance, the mechanism may be off-target or poly-target. Another approach is proteomic pull-down (for compounds) to see what the molecule binds. For an image-based signature, check if it clusters with any known reference perturbation in the atlas – that can generate a hypothesis (“this looks like a TNF-alpha inhibition signature”) which can be tested by directly measuring pathway readouts or using known inhibitors.

(3) Orthogonal phenotypes: Validate the hit in a different assay that measures the core desired function. For instance, if the primary screen was based on morphology, test whether the hit also improves a relevant functional metric (e.g., contractility in a cardiac fibrosis model, or survival in a toxicity model). Recursion’s philosophy of “patient connectivity” is relevant here – they seek evidence that perturbing the target maps onto real disease biology. This could mean testing the hit in patient-derived cells or organoids to see if it reproduces the beneficial effect.

(4) In Vivo efficacy: Ultimately, a “killer” validation is demonstrating the hit’s effect in an animal model. If a compound arises from a phenotypic screen, one would administer it in a disease model and look for phenotypic improvement (tumor shrinkage, improved tissue function, etc.). Phenotypic hits can be less validated on mechanism, but more directly validated on outcome: if it works in vivo, that validates the approach, even if the target is unknown – though target ID will still be needed for optimization.

(5) Counter-screening for off-targets: Ensure the hit is not acting through known undesirable mechanisms. For example, one might counter-screen the compound in a panel of assays for common toxic liabilities (hERG channel, etc.) or check if the phenotypic change is simply due to cell cycle arrest or apoptosis (common artifacts). In summary, the path from a phenotypic screen to a real target demands connecting the phenotypic outcome back to a molecular cause and confirming that cause actually drives disease reversal in vivo. Only through such multi-pronged validation can one be confident that a phenotypic screening hit represents a viable, novel target (and not a dead-end artifact).

This is the second piece in the “Betting on Biology” series. Next week, we explore two distinct strategies for anchoring drug discovery in real-world biology: first, using human primary tissue perturbation platforms to test how interventions behave in the complex, 3D environment of fresh patient samples directly. We will go into Gordian’s in vivo pooled screening as a case study to highlight the importance of physiological context. Second, leveraging comparative biology to identify novel targets by studying how different species have evolved unique ways to resist disease. See you soon.

Acknowledgements

A big thank you to Satvik Dasariraju for the inspiration, thoughtful comments, and prompt answers to my many questions; to Alex Colville for invaluable writing guidance throughout the process; to every age1 crew for helpful pointers; and to many amazing builders whose innovation and passion shaped the thesis of this piece. Cheers!

The industry’s fallback strategy today often is: take a validated target (with human or clinical evidence) and either improve convenience of the same modality or apply a new modality. That approach can yield quick gains but also crowding. For example, the PCSK9 gene (a cholesterol-regulating protease) was genetically implicated in cardiovascular risk: gain-of-function variants cause high LDL, while natural loss-of-function mutations give protection. Once PCSK9 was proven causal, dozens of companies rushed in: from first-generation monoclonal antibodies (evolocumab/alirocumab) to RNAi drugs, small molecules, vaccines, and now CRISPR-based gene editing. In one recent deal, Eli Lilly paid up to $1.3 billion for Verve Therapeutics’ one-time PCSK9 gene-editing program. This “validated target + improved/new modality” pattern yields value (the first PCSK9 drug lowered heart events), but it also means intense competition: the median number of approved drugs per target is only ~2, suggesting the lead drug captures most of the returns. Meanwhile, chasing untested new targets is intuitively the mirror risk. We see a widening pipeline gap in new targets, with novel target entry down threefold over the past decade despite unprecedented pipeline size and capital availability.

(Source: LEK report on novel targets)

We need better tools to navigate this landscape. By the end of the series, we want to create a clear and critical framework for identifying targets so that

• We can distinguish real target information from model artifacts.

• It provides a useful classification of target-origin stories along with a set of killer tests.

• We understand better what are the compelling ways to discover molecular targets for age-related diseases today.

Definition and Scope of “Target”

We use “target” broadly. Consider these classes:

Single gene/protein

Proteases (PCSK9), kinases (mTOR), receptors (IL-6), enzymes (BACE1) targeted directly by small molecules/biologics

Pathway

NLRP3 inflammasome pathway etc.

Cellular state

Pathogenic cells (by Arda Tx), exhausted T-cells etc.

Circuit

Neural circuits (e.g. the basal ganglia loop), immune circuits (e.g. signals in tumor immunity) etc. → multi-target interventions / system-level perturbations.

Combo

Combine different classes of targets together.

*Less than 0.1% of all unique human target-indication pairings have advanced beyond preliminary testing stages in drug discovery. Vast swaths of biology are never tried because we can only prioritize a few targets at a time.

Typology of Target Identification

1. Pathway biology/mechanism-first targets

This classic strategy begins with a well-characterized physiological node (e.g. a hormone or pathway) and tests perturbing it in vivo. Its strength is obvious mechanistic coherence: if a gene or protein is known to regulate a key process (glucose, inflammation, etc.), targeting it feels logical. Decades of academic research reveals amylin as a pancreatic hormone co-secreted with insulin that regulates glucose homeostasis by inhibiting gastric emptying, inhibiting the release of the counter‐regulatory hormone glucagon and inducing meal‐ending satiety. The development of petrelintide leveraged its known mechanism.

Limitations of the approach

• Correlation vs Causality: Signal ≠ cause

  • Novo Nordisk’s EVOKE/EVOKE+ trial results (November 2025) illustrate the risk associated with selecting targets through correlational pathway biology. GLP-1 receptor agonists appeared promising based on three evidence streams: 37% lower dementia incidence in diabetic patients receiving GLP-1 drugs versus other antidiabetics, preclinical neuroprotective effects, and mechanistic links between insulin resistance and neurodegeneration. Yet oral semaglutide failed to slow cognitive decline despite engaging AD-related biomarkers. This disconnect likely reflects that the observational correlation may stem from selection bias - healthier patients choosing GLP-1 therapy - rather than neuroprotection, and metabolic dysfunction may parallel rather than drive Alzheimer’s pathogenesis. Pathway intersection does not equal causal involvement. However, interpretation requires caution as more results are revealed, and GLP-1 agonists may address contributory rather than primary pathology, potentially adding value in multifactorial treatment strategies even if insufficient alone.

• Compensation

  • Compensatory mechanisms refer to adaptive responses within biological networks in which redundant or parallel pathways substitute for the inhibited target, preserving the disease phenotype. These mechanisms are especially common in signaling pathways that are highly interconnected, evolutionarily conserved, and involved in stress or inflammatory responses. In complex diseases, compensation can occur at the level of parallel kinases, such as the c-Jun N-terminal kinase (JNK) or extracellular signal-regulated kinase (ERK) pathways substituting for inhibited p38; feedback loops increasing upstream signaling; and context-specific rewiring not captured in simplified models. The p38 MAPK pathway prompted inhibiting p38 as a treatment for arthritis, COPD, and autoimmune disease, with positive in vitro results. In vivo, parallel cascades compensate.

  • When compensatory mechanisms are likely, founders should prioritize upstream or less redundant targets, validate biology early in human systems, and design programs that anticipate network adaptation rather than single-node inhibition. In practice, this means using combination or polypharmacology strategies, stratifying by disease context and stage, and measuring beyond target engagement as an early go/no-go signal.

• Wrong disease context

  • Targets validated in rodents or cell lines can fail in humans (“Translational Drift”), since models reproduce only fragments of disease biology. Conservation of genes is an inevitable consideration, in specific cases non-rodent animal models or “humanised” mouse models need to be employed.

How to stress-test this typology?

We need to aggressively falsify them. First, test the target across multiple biological contexts: use different models (e.g. human cells or organoids as well as animals) and even non-disease settings to see if the intervention still affects the intended pathway. Show a clear dose-response between target perturbation and disease-relevant outcomes (e.g. pathology burden or functional improvement), not just surrogate markers. For example, rather than only measuring biomarker levels (like amyloid reduction), test cognition or survival. Second, employ genetic knockouts or CRISPR in human-derived cells/organs to confirm that loss of the target recapitulates the protective effect. A true causal node should produce a phenotype whenever it is perturbed. Third, if available, leverage human evidence: for instance, examine whether rare human knockouts or loss-of-function variants in the target gene have the predicted effect. This leads to our discussion of the second typology below on genetics.

*Push the target to fail by varying context, modality, and readout; only a driver that survives all falsification attempts is worthy of pursuit.

2. Human Genetic Causality

Human genetic evidence can identify causal targets in three broad ways:

1. Obvious traits / large-effect loss of function (LOF)

Some traits have rare “smoking-gun” mutations that cleanly shift biology. The poster child is PCSK9, similarly ANGPTL3 LOF carriers have much lower triglycerides/LDL (-27% and -9%) and lower atherosclerosis risk (-41%). The data supported the logic behind ANGPTL3 antibody evinacumab.

There are several limitations to this approach:

• The biological context differs. Germline LOF represents lifelong, developmental perturbation across all tissues, whereas pharmacological inhibition occurs post-development in adults. A protein essential for early organ development may be safely targetable in mature tissues, or conversely, adaptive compensation during development may hide toxicities that appear with acute inhibition.

• LOF genetics primarily inform disease prevention rather than progression or reversal. For continuous pathologies like atherosclerosis, where disease accumulates gradually, prevention and progression targets largely overlap. However, for diseases with distinct pre-disease and post-disease states, such as osteoarthritis, where cartilage deterioration crosses thresholds of chondrocyte loss and chronic inflammation, genetic protection from cartilage damage may not predict whether inhibiting the same target saves already-degraded joints.

• Compared to obvious large-effect targets in European-ancestry populations already extensively mined, African, East Asian, Latin American, and other populations harbor distinct genetic architectures and may yet reveal high-value monogenic targets invisible in current datasets, particularly for diseases with variable prevalence or presentation across populations.

1. Genome-wide association studies (GWAS) / Mendelian randomization (MR) / Quantitative trait locus (QTL) methods

Genome-wide association studies first identify SNPs associated with disease, but establishing causality requires functional follow-up. QTL mapping links these variants to molecular traits (e.g. identifying SNPs as eQTLs or pQTLs affecting gene expression or protein levels). MR uses the genetic variant as an “exposure” to test if changing it causally impacts disease risk. Because alleles are randomly inherited, MR approximates a randomized trial and minimizes confounding. Bidirectional MR explicitly tests each direction (exposure→disease and disease→exposure) to validate causal direction. For drug-target validation, a cis‐MR approach uses SNPs near a candidate gene (e.g. affecting its protein) as instruments to directly test if that protein affects disease. GWAS/QTL mapping and MR analyses together build a causal chain from genotype to phenotype to disease.

This approach can inspire novel targets (e.g. bidirectional MR has highlighted INHBC in metabolic disease or FGF21 in kidney disease). Human genetic studies place INHBE (encoding the liver-secreted hepatokine Activin E) in the causal chain from GWAS to therapeutic target. Rare predicted loss-of-function (pLOF) variants in INHBE robustly associate with lower waist-to-hip ratio and less visceral fat. In vitro, an INHBE pLOF mutation causes ~90% reduction in secreted Activin E, effectively acting as a protein-level QTL and implying that attenuating INHBE signaling is beneficial. Carriers of INHBE pLOF have a favorable metabolic profile (lower triglycerides, higher HDL, lower glucose) and significantly less visceral adiposity than non-carriers, consistent with MR predictions of lower diabetes risk. This human genetics evidence effectively de-risked INHBE as an obesity target. Recently, in early clinical tests, INHBE silencing recapitulates the predicted effects: Arrowhead’s ARO-INHBE (with tirzepatide) achieved ~23% visceral fat reduction (versus ~7% for tirzepatide alone) and also raised lean mass ~3.6%, while Wave’s WVE-007 (single 240 mg dose) cut visceral fat ~9.4% and increased lean mass ~3.2%. These results closely align with the GWAS→QTL→MR→clinical framework for INHBE target validation.

The main failure modes here come from the limitations of genetic inference. Weak instruments (variants with tiny effect on the target) make estimates fragile or biased. Horizontal pleiotropy (a variant influencing multiple pathways) can be mistaken as causality and inflate false positives. Tissue-specificity also needs care: using a blood eQTL to infer a brain disease target can mislead. Effect sizes from discovery MR studies are often overestimated (the “winner’s curse”) and may shrink on replication.

2. “Intelligent traits”

Some methods refine traits to isolate mechanisms (e.g. waist-to-hip ratio adjusted for BMI to capture fat distribution independently of obesity, or defining “remnant cholesterol” to study triglyceride metabolism). In practice, an “intelligent” trait may generate associations that reflect statistical nuances rather than biology. To stress-test such leads, one should examine them in multiple populations and ensure they replicate outside the original cohort. Ideally, the engineered phenotype should map to the same gene or pathway in independent analyses and have a clear mechanistic rationale; otherwise, it may simply reflect model artifacts.

How to stress-test this typology?

To de-risk MR/QTL findings, one should assemble multiple independent genetic instruments (an allelic series) for the same target: if several distinct SNPs that each alter the gene/protein in graded ways all show consistent disease effects, confidence increases. One should also perform colocalization or multivariable MR with tissue-specific datasets (e.g. GTEx) to ensure the genetic signal aligns with the disease-relevant tissue to mitigate the “wrong tissue” risk. Wherever possible, orthogonal perturbation helps. Knock down or pharmacologically inhibit the candidate target in a human cell or animal model and check if the phenotype matches the genetic prediction. Finally, replicate the MR finding in larger independent cohorts to ensure the effect is not a statistical artifact. These steps help distinguish true causal targets from others.

3. Human Perturbational Biology (Non-genetic)

While the last typology establishes whether a pathway is causally involved in disease, some of the most actionable biological insights in medicine have come from observing unexpected phenotypes following human intervention.This strategy uses natural and clinical “experiments” in humans (drug treatments, infections, surgeries, etc.) to infer targets. Famously, sildenafil (Viagra) was repurposed from treating angina to erectile dysfunction. Likewise, large observational studies have linked routine shingles (herpes zoster) vaccination to lower dementia risk. For example, a Welsh regression‑discontinuity analysis found that live zoster vaccination cut new dementia diagnoses by about 3.5 percentage points over 7 years (∼20% relative reduction). Additionally, from the CANTOS trial (an IL‑1β antibody for atherosclerosis), canakinumab reduced recurrent cardiovascular events as well as dramatically lowered lung cancer and total cancer mortality. These findings suggest that perturbing IL‑1β (an innate‐immune cytokine) conferred unexpected protection against smoking‑related cancer.

Limitations of the approach

• Search space and retrospective constraints

  • The typology is largely retrospective; we are restricted to studying biological pathways that nature or medicine has already “intervened” upon. This creates a “streetlight effect” where we only look where the light is brightest. Vast areas of the genome or non-direct effects without clear genetic associations remain invisible to this typology, and because we rely on existing perturbations, identifying “first-in-class” targets that don’t have a natural human proxy is especially challenging.

• Selection bias

  • Vaccinated, treated, or trial-enrolled populations differ systematically from controls (health-seeking behavior, comorbidities, access to care). Apparent benefits may partly reflect who gets treated rather than what was perturbed.

  • Effects may be specific to the population, disease stage, or exposure context.

• Causal confounding

  • Determining the direction of a relationship is a primary hurdle when using EHR-linked frameworks or MR. The underlying disease (indication) often carries its own inherent risks that correlate with the outcome being measured. A drug may appear to cause an adverse event, but that event may actually be a symptom of the very disease the drug was meant to treat.

• Mechanistic ambiguity and pleiotropy

  • Difficult to attribute observed phenotypes to a specific target rather than off-target or downstream immune effects.

How to stress-test this pathology?

Orthogonal validation is important. If a candidate pathway emerges from a drug study, test whether other drugs or interventions expected to act on the same pathway produce consistent molecular changes. For example, if two unrelated anti-diabetic agents both reduce a particular inflammatory marker, that strengthens the link. Otherwise, if a signature is unique to a single drug, it may be an off-target effect. Within individuals, longitudinal designs (multiple timepoints) help separate direct drug effects from temporal noise. Finally, use orthogonal readouts (biochemical assays, imaging, functional tests) rather than relying solely on high-throughput omics. A purported target should correlate with improvements in hard endpoints across perturbations, not just with a surrogate biomarker. To increase the efficiency of screening, we see recent efforts in perturbing biology deliberately in scale.

This is the first of a five-part series. Next week, we’ll publish an in-depth analysis of perturbation-based target discovery, dissecting how to separate signals from noise when scale meets biology. Stay tuned!

Acknowledgements

A big thank you to Satvik Dasariraju for the inspiration, thoughtful comments, and prompt answers to my many questions; to Alex Colville for invaluable writing guidance throughout the process; and to all dedicated age1 crew for helpful pointers. Cheers!

To better understand the drivers of biotech company closures, we conducted what we believe is the first systematic review of all publicly disclosed biotech shutdowns (sourced from Fierce Biotech and Biospace) between January 2023 and August 2025. We documented the stated reasons for failure in 64 companies and categorized each into one or more of six major clusters:

1. Capital Crunch: Companies in this category either exhausted cash runway too quickly (Excessive Burn Rate), were unable to secure additional financing to continue operations (Lack of Investor Confidence), or were caught in an extreme downturn (Macroeconomic Landscape).

2. Clinical Failure: In these cases, the company’s lead program(s) failed to meet primary endpoints or generated unacceptable safety signals in preclinical or clinical trials.

3. Strategic Misalignment: This category encompasses companies whose flawed or poorly timed strategic choices left them vulnerable. This includes over-reliance on a single asset or partner, costly failed mergers, overextension into too many programs, or the pursuit of markets misaligned with the company’s capabilities or resources.

4. Commercial Shortfall: Companies whose clinical-stage or approved asset failed to achieve market traction due to weak sales, crowded competitive pressure, insufficient differentiation, pricing/reimbursement barriers, or an overestimated market opportunity fall into this category.

5. Regulatory Setback: This category includes companies that faced adverse regulatory actions such as a clinical hold, complete response letter, refuse-to-file decision, or an unexpected data requirement that derailed the company’s path to market.

6. Operational Breakdown: This rarer category includes companies whose internal execution failures such as mismanaged trials, governance lapses, founding team indecision, or platform immaturity, accelerated failure. We predict this category is underrepresented in our analysis, skewing toward pre-seed/seed biotechs whose data was unavailable.

See the raw data here.

While clinical trial failures accounted for approximately one-third of public explanations for recent biotech closures, capital crunch remained the most significant factor, underscoring that financing risk, not science alone, is the dominant existential threat. A subgroup analysis revealed that of companies for whom capital crunch was a primary cause of failure, lack of investor confidence accounted for nearly three-fourths (72.5%) of cases, whereas excessive burn rate (19.6%) and macroeconomic landscape (7.8%) made up the minority. However, these three players are often interrelated, and thus cannot be considered in isolation—rapid cash burn can erode investor confidence, just as a poor funding climate can magnify the effects of either; thus, this subjective analysis cannot be taken for a definitive causal hierarchy of risk, but rather as a directional snapshot of publicly visible patterns.

Lastly, strategic errors, regulatory barriers, commercial blunders, and operational breakdowns together accounted for over 20% of publicly stated explanations for recent biotech failures. It is important to note, however, that the analysis above is limited to publicly disclosed information, and without access to internal decision-making processes, private financial data, and unpublished trial results, it is difficult to draw definitive conclusions about any given company’s demise. age1 looks forward to when a more rigorous, objective, and data-complete analysis can be conducted across the biotech sector to better inform founders, investors, and the public about the drivers of survival and failure in this industry.

In this series, we highlighted several unnerving stats about building in biotech. But don’t let us scare you away! By learning from the blunders above, and recognizing key patterns of failure, biotech founders have the opportunity to develop adaptable strategies to substantially increase survival odds.

age1’s Biotech Operator Checklist: Avoiding Common Pitfalls

I. Capital & Runway Planning

1. Have we explored all potential funding avenues (CPRIT, STTR, SBIR, ARPA-H, and other non-dilutive sources) alongside traditional investor capital?

2. When raising money, are we accounting for the expected, the unexpected, and planning long-term? On top of clinical trial costs, capital is necessary to fund CMC, commercial planning, and recovery from any obstacles along the way.

3. Can we sustain operations for ≥12 months without any additional cash infusion?

4. Does >30% of our future runway rely on payments from a single partner or counterparty?

5. Are we timing fundraising to valuation peaks around pivotal milestones, rather than waiting until runway is critically low?

II. Scientific & Clinical Rigor

6. Have third-party labs reproduced our core mechanistic readouts with blinded protocols? If not, is there sufficient genetic, epidemiological, preclinical, and mechanistic evidence to justify capital influx?

7. Have we thoroughly examined all safety and efficacy data to date on analogous drugs, reviewed historical literature, and conferred with independent KOLs to uncover blind spots?

8. Does our chosen indication clearly align with mechanism and manageable clinical endpoints? Are market size, payer burden, and mechanistic rationale aligned?

9. Are trial endpoints rigorously validated, clinically relevant, and powered adequately for symptom benefit—not just biomarkers?

III. Market & Portfolio Strategy

10. Is our pipeline breadth manageable for our capital and resources? While a multiasset portfolio is critical to risk management and financial success, spreading too thin can leave a company underfunded, understaffed, and incapable of reaching an inflection point.

11. Have we begun payor conversations ≥6 months pre-approval, with early onboarding of access teams? A 2022 McKinsey survey of first-time launchers revealed that successful companies onboarded a market access function team 4-6 months earlier than their less successful peers, 75% of which admitted that being late in the payor game was a costly choice.

IV. Operational Resilience

12. Has leadership conferred with enough third-party KOLs and mentors (clinicians, statisticians, CMC, ex-FDA) to identify blind spots and surface operational, regulatory, or clinical risks before committing significant capital or advancing into costly trials?

13. Is there a qualified backup for every critical supplier? As EY notes, today’s biotech manufacturing supply chain landscape remains uncertain, making this checklist item increasingly pertinent to today’s founders.

V. Risk & Crisis Management

14. Do we have a crisis-communications plan for bad data or safety events? Have we predefined quantitative criteria that would warrant shuttering trials before draining capital on a sinking ship?

Sources we liked:

Painful Truth: Successful Failure Of A Biotech Startup – LifeSciVC

The Intelligent Entrepreneur: Real and Illusory Margins of Safety in Company Building – LifeSciVC

Small but mighty: Priming biotech first-time launchers to compete with established players – McKinsey

Biotech Startups And The Hard Truth Of Innovation – Forbes

Good Science Doesn’t Guarantee Success: Why Some Biotech Startups Fail – BioSpace

So you want to start a biotech company – Sadhana Chitale, Colm Lawler, and Arthur Klausner

The Entrepreneur's Guide to a Biotech Startup, 4th Edition – Peter Kolchinsky

EY Biotech Beyond Borders Report 2025 – EY

We’d love to hear from our audience here as well: what questions would you recommend founders ask themselves before taking the next step?

In this article, age1 profiles (autopsies?) eight selected biotechs as case studies in the scientific, managerial, and operational missteps that can sink an early-stage company. We selected companies that garnered considerable attention in the biotech community at the time of their failure, and whose failures could be attributed to specific reasons that we believe, if detailed here, could offer valuable lessons to aspiring founders hoping to beat the odds.

It's important to note that this article only focuses on higher-profile biotech failures—companies whose rise and fall left enough of a paper trail for post-mortem analysis. In the last decade, there have likely been thousands of biotech shutdowns that simply did not progress far enough to attract capital. These seed-stage failures can arise for several reasons: inexperienced management (repeat entrepreneurs have consistently been found to be more successful the second or third time around), an inability to secure funding for R&D costs, and intense crowd competition. Many founders face their greatest hurdle at the "Series A Cliff," the gap between seed-stage enthusiasm and the rigorous clinical, commercial, and operational milestones required to attract institutional capital.

1. Zafgen

In 2015, metabolic disorder biotech Zafgen was one to watch. That January, the company's lead drug, beloranib, a MetAP2 inhibitor, showed a positive hypothalamic injury-associated obesity (HIAO) Phase 2 signal, and in May, the company announced a similar positive Phase 2 for beloranib in Prader-Willi Syndrome. A 12-week, double‑blind study in 17 adults met its primary endpoint (p < 0.01), with weight loss of up to ~11 kg from baseline; the company had progressed smoothly through the critical Phase 2 mark. The drug was supposedly advancing smoothly in Phase 3 trials when, in early October of 2015, the company abruptly canceled its appearance at an investor meeting, and didn't hold analyst or investor calls for days. The stock plunged nearly 60%.

Only days later, Zafgen announced that there had been a patient death in the Phase III trial, but refused to disclose details around the tragic outcome publicly. Months later, Zafgen acknowledged that yet another patient died of the same cause: pulmonary embolism. Two others experienced non-fatal deep vein thrombosis.

That December, the FDA escalated from partial to full hold on beloranib, and the following summer, the company was forced to reduce its workforce by 34%. President Patrick Loustau and CCO Alicia Secor also left the company. In July 2016, the company refocused on yet another MetAP2 inhibitor (ZGN‑1061). Despite Zafgen's claims of a safer new formulation, the FDA halted development of the drug in November 2018 based on the cardiovascular safety concerns identified in its predecessor's outcomes. Although Zafgen reached a July 2019 agreement with the FDA on an in vivo animal study design and protocol to establish relevant safety margins for ZGN-1061, preliminary results did not raise confidence in the Zafgen team, and on December 17th, 2019, Zafgen merged with Chondrial Therapeutics, having ended the year in a deficit of $396.4 million and no approved products. Zafgen now operates as Larimar Therapeutics.

What went wrong?

Three fundamental missteps stand out: ignorance of a predictable safety signal, opacity that destroyed investor confidence, and target class tunnel vision even after two deaths and an FDA clinical hold. Anti-angiogenic agents like beloranib had long been linked to thrombotic events; after the first patient death, however, the Zafgen team noted that "at the time, it was unclear if the death should be attributed to beloranib treatment," and the thrombosis-free patients were free to continue taking the drug in an open-label extension. Only after the second patient death did dosing stop; the Zafgen team later conducted mechanistic research suggesting "beloranib slows endothelial cell proliferation, which influences pro- and anticoagulant factors on the cell surface… prolonged exposure (≥ 12 h) from the suspension formulation is hypothesized to contribute to the imbalance of thrombotic events."

The company's secrecy also did not help its case. Zafgen's decision to cancel the investor roadshow and maintain what Fierce Biotech called an "awkward, stony silence" as safety rumors swirled around the beloranib program created an information vacuum that was inevitably going to be filled by speculation. Investors learned of the first patient death only after the stock had begun to slide, which only compounded the trust deficit when a second death was later disclosed. And the company's platform tunnel vision in the years following did not help their cause; rather than diversify, management poured scarce capital into a backup MetAP2 inhibitor (ZGN-1061), which investors and regulators could not credibly separate from beloranib's clot-risk baggage.

"Without such basic disclosures from the company [such as drug metabolization, cellular uptake, and clearance data], and the preclinical work showing that ZGN-1061 may induce pro-clotting factors… we have viewed the aforementioned variables as significant risks," commented Leerink Partners analyst Joseph Schwartz.

A decade after Zafgen's fall, Sarepta Therapeutics followed an eerily similar path. Fast forward a few years: Elevidys, Sarepta's AAV-based Duchenne gene therapy, won accelerated approval in 2023 and a label expansion in 2024 despite lingering FDA and critics' doubts. After two liver-failure deaths linked to Elevidys's viral delivery method, the company rebuffed regulators' demands to stop delivering product, then issued a press release about layoffs that omitted news of a third fatality, prompting Wall Street outrage when the omission surfaced days later and caused the stock to plunge 36%. In both cases, the opacity spawned shareholder litigation. A securities-fraud class-action suit alleges Sarepta made "false and/or misleading statements," failing to disclose that Elevidys posed "significant safety risks," that trial protocols were unable to catch severe side effects, and that the undisclosed morbidity would trigger trial halts, regulatory scrutiny, and jeopardise expanded approvals. While Sarepta remains in business for the time being, the twin stories underscore the importance of conservative clinical risk-benefit analysis and close monitoring of safety signals, especially of troublesome outcomes grounded in mechanistic rationale. Platforms and assets can be re-engineered; credibility cannot. Once forfeited, future developments and data are discounted, and capital moves on to the next racehorse.

2. Goldfinch Bio

When Third Rock Ventures unveiled Goldfinch Bio, a kidney disease precision medicine company, the biotech was set to be the first of its kind in the world of kidney disease therapy. Goldfinch launched in 2016 with $55 million, money that seeded development of the Kidney Genome Atlas (KGA) and an ion channel program aimed at CKD. Its lead candidate, GFB-887, targeted the TRPC5-Rac1 pathway to protect podocytes, the filtration linings of the kidney. In May 2019, Goldfinch also entered into a multi-billion-dollar milestone collaboration with Gilead Sciences. Five months later, in October 2019, Goldfinch acquired its second asset (CB1 inverse agonist GFB‑024) from Takeda. In 2020, Goldfinch raised $100 million in Series B, bringing its total capital raised to over $200 million in its first four years.

The shutdown came just under a year after the company released data from a Phase 2 showing that GFB-887's ability to reduce proteinuria compared to placebo was significant in patients with focal segmental glomerulosclerosis, a rare kidney disease marked by blood vessel scarring in the glomerulus; however, the drug showed no effect in the initial cohort of 44 diabetic nephropathy (DN) patients.

Ultimately, DN was a disease that affected up to 40% of diabetics, and FSGS was an orphan indication (<40,000 U.S. patients) unlikely to justify a multibillion-dollar valuation on its own. The study was terminated in November 2022 due to "business reasons." At the same time, the company was already reducing its staff. The company released positive Phase 1 data on GFB-024 in late 2022, but the preliminary success was ultimately not enough to reassure investors after the lead candidate produced lukewarm results. Operations dwindled, and by January 27th, Goldfinch had officially closed.

What went wrong?

According to The Wall Street Journal, Goldfinch was attempting to secure additional funding before closure, but the previous Phase 2 failure posed too great a risk to investors. "Unfortunately, we had funding challenges, just like I think the rest of the environment, particularly private companies, in the current macro environment,” said CEO Tony Johnson. This explanation certainly played a part in the failure; the economic shutdown caused by COVID-19 heavily affected the biotech industry, and in a downturn, investors would move reasonably toward assets with quick de‑risking milestones; kidney precision medicine had neither in 2022.

However, another lesson is that indication selection matters as much as the mechanism. While the drug was safe and efficacious in both indications, FSGS, given its name, is a disease of the glomerular system; DN is a kidney-wide disorder affecting both glomeruli and tubules. Goldfinch Bio started out as a precision medicine company aimed at genetically defined, rare podocytopathies; pivoting to diabetic nephropathy, a highly prevalent, genetically heterogeneous disorder that would demand large, costly outcome trials, pulled the company far outside its original niche.

For $15 million upfront, and up to $520 million in milestones, Goldfinch was able to appease creditors somewhat by selling its TRPC4/5 channel candidates to Karuna Therapeutics, a biotech aimed at curing mood disorders. Ironically, however, Goldfinch collapsed just as the renal medicine market began heating up. Novartis alone has poured more than $5 billion into the space—in June 2023, the pharma bought out Chinook for $3.2 billion in IgA-nephropathy assets, followed by a $1.7B acquisition of Regulus Therapeutics two years later. AbbVie and Vertex Pharmaceuticals are both advancing frontier clinical-stage assets for ADPKD, and venture money has kept pace: early-stage renal biotechs like Renasant and Judo Bio are entering the arena with tens of millions in seed funding.

3. SQZ Biotechnologies

Armon Sharei was still an MIT graduate student when he discovered the technology behind SQZ. By pushing—or rather, "squeezing" cells (hence the name "SQZ") through microcapillary chips, a cell membrane could temporarily open enough such that proteins and reagents could enter without viral vectors or electroporation. Sharei spun out the company in 2013, branding the platform Cell Squeeze™, and was lucky enough to strike an early success most founders could only dream of. In December 2015, Roche offered SQZ a $1 billion option‑and‑license partnership deal. With Roche's stamp of approval, SQZ was able to raise substantial venture funding, culminating in a 2020 $65 million Series D and an October IPO via SPAC that same year.

At the time of the transaction, SQZ fielded three distinct therapies: APCs, AACs, and TACs, targeting infectious disease, cancer, and autoimmunity. The company acknowledged in IPO filings that Roche's milestone payouts were heavily dependent on the success of its first two clinical programs, SQZ-PBMC-HPV (Phase 1) and SQZ-AAC-HPV (preclinical), both aimed at HPV-positive tumors. Yet the company continued to stretch its platform further, announcing plans to expand into celiac and Parkinson's disease in the near future. Public updates stayed upbeat, but the company struggled behind the scenes as clinical progress faltered. By the end of 2022, the company finally announced a "strategic reprioritization" to focus solely on its lead HPV-targeting candidates and pause all other projects. Sharei and his CFO had left the company, the workforce was cut by 60%, and SQZ reported an 80% loss for the year. But the company suffered its biggest blow in 2023, when Roche declined to exercise its option on SQZ's HPV‑positive solid‑tumor program, and overnight, the billion‑dollar milestone stream that investors were waiting on disappeared.

The board was forced to gut 80% of the remaining workforce, leaving barely a dozen employees to keep trials alive. According to FierceBiotech, SQZ reportedly spent the remainder of 2023 searching for an alternative partner, but none came forward despite positive clinical results: eight of the twenty patients in SQZ's APC platform trial reported cancer stabilization. Bankers contacted 52 potential buyers; five signed NDAs, and three offered term sheets. Ultimately, Stemcell Technologies bought SQZ's IP on December 19th, 2023, for $11.8 million, and in March 2024, shareholders voted to liquidate. Undeterred, Sharei quickly regrouped, launching a new cell-engineering technology venture called Portal Biotechnologies in late 2023, which has thus far raised over $22 million in funding.

What went wrong?

As much as SQZ touted the $1 billion number from Roche, only $94 million of Roche's money had been received up to mid-2020; the vast majority of capital was contingent on long-term success (clinical, regulatory, sales milestones). And like many biotech startups at IPO, SQZ's prospectus admitted "we have incurred significant losses since inception and expect to incur significant additional losses for the foreseeable future, and we have no products that have generated any commercial revenue and we may never achieve or maintain profitability." This isn't unusual language for a pre-commercial biotech—SQZ was entirely reliant on investor funding and partnership payments to continue operating. However, SQZ's broad pipeline suggested dangerous dilution for a pre-revenue company, spreading limited resources across too many projects. An IPO framed around one or two assets would have set more modest expectations for investors. And amid the IPO frenzy, SQZ had still not settled on a defined company vision: platform or therapeutic.

Even in the best-case scenario (Roche staying on), the "alternate licensing" partnership structure meant SQZ would share rights and revenues for any success. Roche and SQZ would split commercial rights on alternating programs in oncology, and Roche held options on additional antigen targets. As SQZ's prospectus stated, success under the Roche collaboration "may be obtained at the expense of or to the detriment of our other wholly owned product candidates." While the Roche deal was an incredible early opportunity, it also implicitly capped SQZ's future profitability. Moreover, since that 2015 deal, Roche remained SQZ's lifeline. After noting the IPO cash would last only to 2022, the prospectus document states, "other than potential payments, if any, from our collaboration agreement with Roche for SQZ-PBMC-HPV, we do not yet have any committed source of funding for the completion of clinical development or commercialization of these product candidates."

Ultimately, the SQZ story demonstrates three critical lessons: contingent deals aren't cash equivalents, lead asset strength will always dominate preclinical pipeline breadth, and single-partner dependency can become an existential liability if diversification and fallback capital aren't in place.

4. Walking Fish Therapeutics

According to CEO Rusty Williams, M.D., Ph.D., Walking Fish Therapeutics, a B-cell therapy company with a three-asset portfolio, was two months away from requesting FDA approval to enter the clinic when an unnamed top investor pulled out at the last minute. The investor cited concerns over a long-term manufacturing lease; Walking Fish evidently lacked runway. The company had already begun plans to shutter when yet another insider followed the first investor out. Thirty-five employees were laid off, and Williams quite promptly moved on to his next venture at Ten30 Bio.

The May 2024 shutdown came as a public surprise. Walking Fish had closed a whopping $50M Series A (later increased to $73M) in early 2022, and had secured two ARPA-H grants just months before the shutdown. Walking Fish's lead candidate, WFX-001, aimed to cure Fabry Disease, a rare lysosomal storage condition. The company had plans to expand into other indications, envisioning a fully-fledged platform for B-cell-based therapies to restore malfunctioning protein and enzymatic activity, as well as reconstitute an innate antibody response in cancer therapy.

What went wrong?

While Williams did not disclose the funding split of the many investors in his Series B, the immediate collapse of the funding round after a single late‑stage lead investor withdrew suggests that Walking Fish had undertaken significant fixed costs, such as R&D and manufacturing expenses, based on anticipated, rather than secured, capital. This lack of adequate contingency financing meant they lacked the time to secure an alternative investor before running out of cash. As Williams admitted, "the cost of cell therapy manufacturing and associated clinical trial costs were difficult to overcome." "I think that B-cell therapy is likely to have its day; I think that what we were doing would have worked," said Williams in an interview with Fierce Biotech. "But it was just too expensive to get the phase 1 done in that financing environment."

5. Allakos

Founded in 2012, Allakos's lead candidate was AK002 (Lirentelimab), a monoclonal antibody targeting Siglec-8 (a receptor found on eosinophils and mast cells) to treat severe food allergies and inflammatory GI diseases. Altogether, Allakos's strategy was simple: to advance its lead antibody through clinical trials in multiple indications, demonstrate efficacy, and go public to fund Phase 3. The biological rationale was clear, as elevated eosinophils are known to play a role in disorders like eosinophilic esophagitis and gastritis. Novo Ventures led the company's December 2012 $32 million Series A investing with Roche and others. Follow-on early rounds added another ~$32 million, and a large $100 million Series B in early 2017 marked the last chunk of funding. By mid-2018, Allakos's lead program was progressing through a Phase 2 trial, and the company raised a $128 million IPO that summer, with the plan to put aside $94 million to advance AK002.

In mid-2019, the company announced positive Phase 2 results in patients with eosinophilic gastritis and duodenitis, sending Allakos stock soaring. By late 2020 into 2021, Allakos's market cap swelled to the multi-billion-dollar range in anticipation of Phase 3 results. The trouble came in the fall of 2021, when two pivotal studies (Phase 3 ENIGMA-2 for eosinophilic gastritis or duodenitis, and Phase 2/3 KRYPTOS in patients with eosinophilic esophagitis alone) missed their co-primary endpoints despite successfully lowering eosinophil counts. That December, Allakos shares fell nearly 90%, wiping out approximately $3.5 billion in market value, and the company never recovered.

What went wrong?

While eosinophil depletion offered an intuitively appealing biomarker, ENIGMA-1's Phase 2 reported only modest symptom improvements insufficiently robust to justify a Phase 3 investment. In Phase 2, treatment cut mean total symptom score (TSS) by a relative change of -48% vs -22% on placebo, yet when ENIGMA-2 used a more rigorous absolute change endpoint and a 6-item TSS (dropping two low-severity items from the 8-item diary), no clinically meaningful benefit emerged. Post-hoc analysis also suggests Phase 3 enrolled a broader, lower-severity population than the 17 legacy ENIGMA-1 sites, greatly increasing heterogeneity. Even ENIGMA-1's headline 63% "treatment response" rate (vs. 5% for placebo) was inflated by design, requiring both a >30% TSS reduction and a >75% histologic eosinophil reduction, anchoring the outcome to the very biomarker under evaluation.

Lastly, while a -22% symptom score reduction in ENIGMA-1's placebo arm might seem notable, such a drop was actually uncharacteristically low for this indication, where mean placebo improvement often approaches ~40%, a number not far off from lirentelimab's -48% figure. In ENIGMA-2, where histology and symptoms were treated as separate co-primary endpoints, placebo improved symptoms by ~45-50% (consistent with other eosinophilic gastrointestinal disease trials), while Lirentelimab again failed to produce significant symptom benefit despite achieving an 80-90% eosinophil reduction versus placebo.

Only in 2023 did Kliewer et al. find that eosinophilic oesophagitis pathology is largely independent of excessive eosinophil production, which should have suggested to Allakos that this would be true for other indications. However, management pivoted to AK006 (an anti‑Siglec‑6 that would inhibit mast cell activity to reduce eosinophils indirectly) and continued developing Lirentelimab in atopic dermatitis (ATLAS) and chronic spontaneous urticaria (MAVERICK). Both failed in early 2024:

• ATLAS: 23% of Lirentelimab vs. 18% of placebo patients achieved a 75% or greater improvement in the Eczema Area and Severity Index.

• MAVERICK: Urticaria Activity Score dropped 27% in treated patients vs. 26% in placebo.

Allakos's downfall stemmed from a failure to interrogate core mechanistic assumptions, adapt clinical endpoints and trial design to prioritize patient-relevant outcomes, and recalibrate its strategy as the biological picture evolved, recognizing that while its products excelled at reducing eosinophils, functional outcomes faltered. Even after new evidence emerged, the company continued to hinge its pipeline on Siglec‑targeted antibodies aimed at modulating these cells indirectly.

Clinical trial misalignment was only exacerbated by operational excesses. Allakos aggressively scaled operations, incurring net losses of $246.1 million in the first half of 2022 alone, nearly double those of the previous year. By mid-2022, the deficit stood at $859 million. Allakos was forced into a rapid restructuring that February (termed the "Reorganization Plan"), terminating its substantial manufacturing commitments with Lonza AG and cutting ~35% of its workforce. While these actions slightly extended runway, they came too late to alter the company's trajectory. When alternative asset AK006 (a Siglec-6 activator) failed this January 2025 in a chronic spontaneous urticaria (CSU) trial, Allakos shrank to 15 employees and merged with Concentra Biosciences in April.

6. Sirtris Pharmaceuticals

In April 2008, GlaxoSmithKline (GSK) agreed to acquire David Sinclair's spinout, Sirtris Pharmaceuticals, for ~$720M cash to access the company's lead sirtuin asset (SRT501) for metabolic and degenerative diseases. Notably, GSK paid this price at an 84% premium—that is, at a price 84% greater than the company's pre-acquisition valuation. Critics commented that the GSK acquisition was naïve, as the company had only a single compound in trials (SRT2104 in Phase 1b for ulcerative colitis), and there was insufficient evidence for human efficacy.

The mechanistic premise became even more controversial after Pfizer and other groups (Amgen, University of Washington) increasingly failed to replicate Sincair's data. In May 2010, GSK suspended a Phase 2a trial of high-dose micronised SRT501 in advanced multiple myeloma after several patients developed nephropathies; as GSK told FierceBiotech, the compound "may only offer minimal efficacy while having a potential to indirectly exacerbate a renal complication common in this patient population." Finally, in late November 2010, GSK ceased all programs related to SRT501, but stated that they remained interested in developing SRT2104 and SRT2379, biosimilars to SRT501 with "more favorable properties." By 2013, GSK shut down Sirtris's Cambridge lab, and the company was effectively dissolved into GSK's R&D.

Unfortunately, a 2014 trial found that a 28 day course of SRT2104 had no protective effect in type II diabetes patients, nor ulcerative colitis in a 2016 trial, and SRT2379 was unable to modulate the inflammatory response of healthy male subjects after exposure to lipopolysaccharide (however, SRT2104 did prove successful in stimulating anti-inflammatory and anticoagulant responses in a later trial). In any case, no Sirtris compounds advanced quickly enough within GSK, and the pharma company seemingly let go of the sirtuin platform after 2015. While many other groups in the years following have continued exploring new indications for compounds in the SIRT-1 activator class, a sirtuin activator has yet to be approved and deployed into the market more than 25 years after Sincair's initial yeast aging discovery.

What went wrong?

The Sirtis story represents both a lesson to founders and investors. Paying an 84% premium for a platform with unvalidated data put GSK in a problematic position when third‑party labs contradicted Sirtris's SIRT1 activation data; ultimately, GSK should have reproduced the company's data independently before writing a nine‑figure check. On the other hand, Sirtris should have controlled for assay artifacts. Pfizer's study found that Sinclair's activator compounds bound to the SIRT1-fluorescent peptide complex, but when they removed the fluorogenic group, calorimetry confirmed that the compound only bound to the artificial peptide-enzyme complex and not SIRT1. GSK may have been able to get ahead of the criticism by imposing a more stringent post‑merger governance model; however, the pharma seemingly granted Sirtris full autonomy in Cambridge.

7. Athersys

Athersys spent more than two decades betting on MultiStem, which the company described as an "off-the-shelf" stem cell therapy designed to promote tissue repair and healing. Since 2011, the company had financed its moonshot through funding from Aspire Capital. In May 2022, Aspire agreed to buy up to $100 million of shares in the company over 24 months. Only a few weeks later, however, the company faced an overseas disappointment: a critical Phase 2/3 ischemic stroke trial sponsored by Healios, Athersys' Japanese partner, failed to meet its primary endpoint. The finding triggered a 60% share price collapse, prompting leadership to undertake a massive 70% reduction in staff mid-2022, gutting the company's R&D team with the intention of regaining investor trust by cutting costs. However, Aspire Capital Fund had only purchased shares with a caveat that it could cut and run under a range of circumstances, including the departure of any Athersys executives for any reason. Athersys's $100 million committed equity financing line was cut that July.

Hoping to salvage the program, the company persuaded the FDA in early 2023 to tweak its MASTERS‑2 design, shifting the primary outcome from a 90‑day functional score to a 365‑day measure, aiming to replicate encouraging long‑term trends seen in prior Japan stroke trials. However, after gaining FDA permission, Athersys reported in October 2023 that the results of an interim analysis proved the trial was still underpowered; 300 patients weren't enough to show efficacy, and Athersys could not afford to expand. At this point, Athersys's stock was a penny stock (quite literally trading at ~$0.14).

"Since announcing the results of the blinded interim analysis of MASTERS-2… we've been actively engaged in exploring strategic options with interested parties to determine the best path forward for MultiStem and Athersys," said Daniel Camardo, CEO. But options were slim. No partner stepped forward, and on January 5th, 2024, the company filed for Chapter 11 bankruptcy, selling its remaining assets to Healios.

What went wrong?

Athersys had been an established company for fifteen years, with over $500 million invested, and zero backup pipeline. MultiStem was Athersys's bread and butter, and their therapy of choice for every indication they pursued (stroke, trauma, ARDS). The company either misjudged Aspire Capital's commitment or overlooked crucial deal clauses, as layoffs executed without regard for financing extinguished $100 million in funding. And lastly, launching a pivotal trial in the same indication where a previous attempt had failed, based solely on aggressive statistical assumptions, boxed the company into a precarious corner.

8. Argos Therapeutics

Argos Therapeutics was a Durham-based biotech that spent over two decades developing Arcelis, an immunotherapy platform that aimed to leverage a patient's own dendritic cells and tumor RNA to create a metastatic kidney cancer (mRCC) vaccine (named rocapuldencel-T, formerly AGS-003). After success in Phases 1 and 2, Argos moved the therapy into a pivotal 2012 Phase 3 (ADAPT), combining rocapuldencel-T with the standard-of-care drug sunitinib in 462 patients. The study finalized enrollment in July 2015 and commenced treatment. However, in February 2017, an independent data monitoring committee (IDMC) analyzed interim results and recommended the trial be stopped for futility, as the vaccine was "unlikely to demonstrate a significant improvement in overall survival."

Nonetheless, Argos persisted. But 2018 data remained negative (in fact, the data were somewhat disappointing—median overall survival was slightly longer in the sunitinib arm than in the combination arm (28.2 months vs. 31.2 months). Finally, the company shut down ADAPT. NASDAQ delisted the stock in spring 2018 after trading below the minimum threshold, and Argos was forced to lay off another 20 people that year. In November 2018, Argos filed for bankruptcy. Argos had been simultaneously running HIV trials with AGS-004, but a Phase II study in acutely treated HIV patients saw no evidence that the vaccine enabled stopping ART, and a UNC/Argos pilot trial combining AGS‑004 with vorinostat also failed to induce greater HIV-specific immune responses. In a 2019 court-supervised asset auction, Argos's remaining technology was sold to a consortium of two Korean biotechs that included its former partner Genexine.

What went wrong?

Instead of heeding third-party evaluations, Argos sought to buy time, proposing a protocol amendment to the FDA to continue the study with some changes, and in April 2017, the FDA acquiesced. Instead of taking the failure in stride, Argos's decision to move regulatory goalposts instead of rethinking its pipeline locked stakeholders into a dead program and consumed capital that could have funded a more promising candidate or pipeline-rescuing acquisition.

Takeaways

The grim baseline is well known. Only one in every 5,000 compounds (0.02%) discovered and tested preclinically get approved; of the drugs started in clinical trials in human participants, only 10% would secure FDA approval. In a later analysis of the period between 2014 and 2023, Citeline estimated the average likelihood of approval (LOA) for a new Phase I drug to be 6.7%, an all-time low. According to Citeline, the leading cause of failure remains the "Phase II hurdle," with a 28% completion rate compared to Phase I (47%) and Phase III (55%).

Yet molecules are only half the story; many companies fail for reasons that have little to do with science. For some fledgling companies, lack of access to supply chains, industrial-scale CMC expertise, and regulatory teams sink the startup rather than drug efficacy; it is mainly for this reason that securing a corporate partner can be the decisive bridge to carry a program through trials. In a study in Nature Biotechnology, analysts from two Dutch venture firms mined GlobalData's Pharma database for every large‑pharma/biotech startup deal made between 2004 and 2019. They then asked whether prior pharma ties predicted startup "success," defined as going public, being acquired, or securing at least one drug approval during that window. Startups with a large‑pharma investor to guide operations nearly doubled their median odds of success (37% vs 18%) and achieved bigger outcomes—market cap rose from $138M to $332M, and median acquisition value from $136M to $377M.

According to Fierce Biotech, venture failures peaked in 2023 with 27 closures, which dropped to 22 in 2024; sources have yet to compile data on how well biotechs fared in 2025, although age1's analysis tallied 15 as of August 14th. However, the same reasons for failure arise year after year. In 2022, BioSpace interviewed several CEOs and VCs on this topic: many mentioned overlapping themes: poor capital management, inadequate flexibility, and miscommunication across hiring, development, management, and more. It's worth stating clearly that most biotechs do not fail because their founders are inept or their science is faulty. In fact, the opposite is true—the companies profiled in this piece were founded by credible, well-intentioned teams and built around plausible mechanisms. Still, building in biotech, for the reasons outlined above and more, is brutally hard. More specifically, failure is far and beyond the statistical default. Some of these companies did not fail from one single decision, but a series of missed pivots, unhedged assumptions, and cracks that gradually widened over time, whose consequences were only visible in hindsight.

Ultimately, our analysis of biotech failures reveals that clinical readouts are rarely the only cause of shutdown. While disappointing clinical trial outcomes are the most ubiquitous catalysts for a biotech's downfall, what separates those who can bounce back from those who can't is behind-the-scenes operations. Importantly, operational robustness can sometimes overcome scientific setbacks (e.g. Exelixis's strategic reprioritization post-COMET-1 trial failure). Conversely, operational blunders can sink even scientifically promising ventures (e.g. Dendreon's miscalculation of demand, COGS, and CMC for prostate cancer treatment, Provenge).

One factor that separates survivors from casualties is whether management aligns trial endpoints with true patient benefit. Allakos drove eosinophil counts down, yet ignored tepid symptom data; its pivotal ENIGMA‑2 study missed both co‑primary clinical endpoints and imploded a multi‑billion‑dollar valuation. While biomarker wins are necessary, they are never sufficient. Clinical rigidity (likely a byproduct of belief in the sunk-cost fallacy) also remains a major driver of biotech failures: Argos pressed on with their renal‑cell vaccine after a data‑monitoring committee declared futility. An extra year of spending only confirmed the verdict and left little capital for a pivot. Those with a long-term vision, including plans for expanding into new indications or developing additional assets beyond their lead candidate, will consistently outperform startups focused solely on short-term goals.

Balance sheet architecture is another tripwire. Walking Fish relied on a single late‑stage investor and spent against money that was not yet in its pockets; when that backer walked, the Fish couldn't: fixed costs outpaced runway, and the company closed despite a recent $73 million raise. Athersys, by contrast, conducted a mass layoff to cut costs after a stroke trial failed, breaching the fine print on a $100 million equity financial agreement. Contingency capital is an inevitable and essential source of funds, but can never be relied upon to the extent that a loss of such funds would bring down the company.

Overdependence on single partnerships also remains a source of caution. While SQZ publicized a "$1 billion" Roche alliance, only $94 million was ever wired. When Roche declined its option, it led to liquidation. Investor trust is equally essential. Zafgen's decision to withdraw from an investor conference after a trial death and stay silent for days was significantly damaging; the stock had already lost half its value by the time management confirmed the fatal thrombotic events. Even if transparency can't revive bad data, opacity compounds the damage.

Another recurring error is letting valuation outpace validation. GSK paid an 84% premium to acquire Sirtris on the strength of unreplicated assays; subsequent independent work showed the lead compound bound to an artificial fluorescent substrate, not to SIRT1 itself. By the time Sinclair could publish a rebuttal, it was too late—the program was abandoned within five years. Ultimately, a late‑stage independent diligence repeat can turn out to be cheaper than a nine‑figure investment.

Across these cases emerge consistent patterns: financial mismanagement, communication failure, over-reliance on single partners, clinical rigidity, and more. While addressing these factors can't guarantee success, they do significantly influence whether a missed primary endpoint becomes a temporary setback or an obituary. Stay tuned for Part II: next week, age1 will publish a systematic review of 64 publicly disclosed biotech shutdowns (2023–2025), organized into six root-cause clusters, alongside an operator checklist we believe every founder should use as they build.

Acknowledgements

Thank you to Alex Colville for providing guidance wherever necessary, and thank you to Lily Clayton for thoughtful graphic design! Lastly, thank you Bruce Booth at Atlas Ventures for inspiring this piece, and to Michael Reisman for his work designing and generating the data for Figure 2 (“From Phase 1 to Approval: Who Actually Makes It”).

As an early-stage investor, age1 is dedicated to backing individuals—founders with cutting-edge, visionary projects and ideas that could add healthy years and decades to human life. Yet startup audacity and corporate scale are not opposites so much as complements: nimble teams can scout novel mechanisms, while pharma excels at industrializing proven science. Growing corporate interest in the aging field inspired age1 to compile this leaderboard: a map of who inside pharma is leading in longevity, and in what ways.

Why is this list important?

1. Pragmatically, most early-stage companies would be extremely fortunate to get to Phase I/II trials backed only by VC funding—Series A/B funding may be just enough to run a Phase I and launch a proof-of-concept Phase II before the next round, but to run a Phase III trial can cost hundreds of millions and take multiple years. Unless the company is prepared to IPO with less clinical data, their best bet may be to partner with or be acquired by pharma. Founders who understand which companies have already staked claims in metabolic disease, neurodegeneration, fibrosis, or immune rejuvenation can craft a story that fits those partners’ strategic gap.

2. We hope that this report card also serves as a call to action for pharma. Facing unprecedented patent cliffs, top pharma companies cannot afford to miss out on one of the very few therapeutic areas with a market size significantly larger than the obesity market. Moreover, large-scale pharma companies have the unique resources and scale to effect transformative change in aging therapeutics through several strategic levers: establishing dedicated internal units focused explicitly on longevity and age-related disease, forging partnerships with innovative startups developing relevant platforms and therapies, directing their venture arms toward promising investments in the aging space, and more. Thus, this report card ideally offers pharmaceutical leaders a snapshot of where bold investment can still capture first-mover advantage and where they sit compared to their peers.

If we have missed any important data or insights that would enrich this article, we invite our audience to share with us; we are always open to feedback!

Without further ado, welcome to age1’s 2025 Pharma Report Card.

Table 1: Grading Criteria & Quantitative Stats

Note: All corporate partnership deal amounts include upfront payment and milestones (with the exception of Gero x Roche, for which upfront sum was undisclosed). Investment data was sourced from company press releases since 2020, as well as platforms such as Pitchbook, Crunchbase, and Tracxn, and validated by search. For the purpose of this report, "aging-thesis driven" refers to companies that explicitly publicly identify as focused on aging and longevity.

Merck & Co. (C-)

Merck is widely known for its work in oncology and vaccines, as well as its recent successes in treating cardiovascular conditions. The company’s greatest development in the age-related disease space is its development of enlicitide, the first oral macrocyclic PCSK9 inhibitor if approved. This June, Merck announced positive results from the first two of three Phase III trials of the drug, making it highly likely that the drug will soon advance to FDA approval. Alongside enlicitide, Merck is also running a Phase 2a clinical study of frespaciguat, an inhaled soluble guanylate cyclase stimulator for pulmonary hypertension. The company’s leadership in cardiovascular disease extends to its portfolio: in 2021, Merck completed a $11.5 billion acquisition of Acceleron Pharma, securing Winrevair—a first-in-class TGF-β/activin signaling modulator to treat pulmonary arterial hypertension granted FDA approval in 2024. And just this March, Merck entered a licensing agreement for HRS-5346, an oral small molecule Lp(a) inhibitor currently being evaluated in a Phase 2 clinical trial in China.

Merck has generally steered clear of any explicit longevity initiatives, but its venture arm has made modest investments in neurodegeneration. Merck Ventures was also a seed investor in Asceneuron, a spinoff company resulting from EMD Serono’s (a division of Merck) Entrepreneur Partnership Program building enzyme O-GlcNAcase (OGA) inhibitors that prevent the aggregation of tau proteins. Blocking O-GlcNAcylation has been shown to decrease amyloid plaque and improve cognitive function in aged mouse models. Unfortunately, Asceneuron unexpectedly halted its own Phase II trial of lead Alzheimer’s disease (AD) candidate ASN51 this March, but remains hopeful about its other clinical-stage tau-targeting therapies such as ASN90, currently in trials for the treatment of progressive supranuclear palsy. Nonetheless, Merck’s collaboration with Neuphoria on MK-1167, an alpha-7 nicotinic acetylcholine receptor allosteric modulator, remains promising, as the drug is currently in Phase II trials. In 2019 and 2023, the company acquired Calporta and Caraway (formerly Rheostat Therapeutics), respectively, both of which are developing drugs that target TRPML1, a lysosomal ion channel critical for cellular waste clearance. Impaired lysosomal function is increasingly recognized as a central driver of neurodegenerative diseases, including AD and PD, due to its role in clearing misfolded proteins and damaged organelles. However, the payout on these investments, exceeding $500 million each, remains unclear, as Merck has not shared any status updates on the development of its TRPML1 agonist preclinical program.

Merck’s strategy of pouring capital into derisked targets (PCSK9, TGF‑β/activin, sGC) and developing novel formulations (oral, inhaled) has resulted in well-executed, faster time‑to‑revenue drugs in cardiometabolic disease, but the company remains hesitant to bet on aging biology, either internally or externally. Venture investments in neurodegeneration remain aging-adjacent, and, like many of its peers, Merck has not publicized efforts to enter partnerships or collaborations with aging-thesis driven companies.

Sanofi (C-)

Sanofi, traditionally a leader in immunology and neuroinflammation, has taken a somewhat unconventional path into longevity R&D—while the French pharma does not have an aging unit like AstraZeneca or Eli Lilly, it has engaged in trials and investments that demonstrate a recognition that diseases of aging present huge opportunities, particularly in neurodegeneration and inflammation, two fields for which the company is arguably most well-known. Most recent is its May 2025 $470 million buyout of Vigil Neuroscience to acquire VG-3927, a small molecule agonist of the TREM2 protein expected to enhance microglial function in AD. In 2024, Sanofi quietly announced a $27 million investment in Ventyx, which is developing VTX3232, an NLRP3 inhibitor that can cross the blood-brain barrier. Chronic NLRP3 inflammasome activation has been shown to accelerate age-related diseases and conditions such as AD, Parkinson’s Disease (PD), metabolic disease, and even frailty. VTX3232 has already demonstrated success in safety and target engagement trials, with the results of subsequent studies expected later this year. In 2023, Sanofi Ventures also joined Novo Holdings in a $55 million Series B for NodThera, which is developing a similar small-molecule NLRP3 inflammasome inhibitor, highlighting the company’s clear focus on anti-inflammatory therapeutics—notably, 44 of its 86 clinical-stage projects are immunology and inflammation, comprising more than half of its pipeline.

Beyond investments, Sanofi’s clinical pipeline also includes some standouts: Frexalimab, a CD40-CD40L pathway inhibitor, is currently in Phase III trials for the treatment of Nonrelapsing Secondary Progressive Multiple Sclerosis (nrSPMS). Like AbbVie, Sanofi is also developing a gene therapy for wet age-related macular disease (SAR402663), albeit only in Phase I/II. The company is also running Phase I trials of a first-in-class trispecific nanobody for knee osteoarthritis (SAR446959), simultaneously inhibiting key cartilage-degrading enzymes (MMP13 and ADAMTS5) and anchoring the nanobody to the affected cartilage. Sanofi also boasts a promising vaccine pipeline, including three mRNA vaccines for RSV aimed at older adults. In 2022, Sanofi made a blockbuster $1.2 billion deal with Insilico Medicine to advance candidates for up to six new targets using the company’s Pharma.AI platform.

Ultimately, however, Sanofi sits at a C‑ because the company’s work in aging is currently a patchwork of late‑stage, disease‑centric projects (TREM2 for AD, NLRP3 for PD, trispecific nanobodies for osteoarthritis) rather than a cohesive longevity strategy. Sanofi’s alliance with an aging-thesis driven company like Insilico is promising, but it remains unclear whether Sanofi will use the company’s platform to develop assets in age-related disease, and the company has yet to establish an in-house aging group or publicly promote investment in the targeting of aging as a primary condition.

Pfizer (C)

Long known for its strengths in oncology, immunology, and vaccine development, Pfizer has slowly started to position itself as a participant in the longevity space. While Pfizer lacks a specialized aging division, the company has made its commitment to aging research explicit through investments and public campaigns. In 2022, Pfizer Ventures made a $500,000 investment in VitaDAO, a community collective organization that funds longevity researchers. One year later, Pfizer entered a research collaboration with Gero to discover drug targets for fibrotic diseases.

The company’s venture arm has also made four significant investments in neurodegenerative disease therapeutics. In 2017, Pfizer joined AbbVie in backing Aquinnah’s RNA-binding protein approach for ALS/AD as mentioned earlier, and, in 2019, co-led funding in lysosomal dysfunction biotech Arkuda Therapeutics (notably, Eli Lilly joined Arkuda in the following Series B round, making the biotechnology company one to watch). In 2020, Pfizer Ventures invested $15 million into Mission Therapeutics (joining Roche), a biotech developing selective deubiquitinase inhibitors to enhance mitophagy, and just this March, Pfizer contributed to a $31 million seed round for TRIMTECH Therapeutics (along with Eli Lilly), leveraging E3 ubiquitin ligase TRIM21 to degrade protein aggregates responsible for neuroinflammation. One of Pfizer’s most intriguing pipeline candidates is Ponsegromab, a GDF15 monoclonal antibody, which has shown positive Phase II results in cancer cachexia; GDF15 is one of the most consistently upregulated proteins with age. The company’s pipeline also includes drugs such as Dekavil, a precision IL-10-conjugated antibody that exploits fibronectin overexpression in inflamed cartilage; PF-07868489, a recombinant human IgG1 that targets bone morphogenetic protein (BMP9) for pulmonary arterial hypertension; and ervogastat, a first-in-class DGAT2 inhibitor for MASH.

Pfizer has also been highly vocal about advancing aging research, publishing content about longevity-associated genes, and launching its 2012 #GetOld campaign, which encouraged Americans to embrace aging in a world where they envision its therapeutics adding multiple healthy years. There are also rumors that Pfizer hosted an internal R&D retreat several years ago with a longevity theme. However, Pfizer seemingly remains longevity‑curious from the sidelines. Its publicity on healthy aging and small check investments into platforms like VitaDAO and Gero are encouraging, yet its own R&D remains focused on core oncology, immunology, and vaccines, and the company has not signaled any intention to create a dedicated aging unit.

Regeneron Pharmaceuticals (C)

As one of few big‑pharma players that operates an in-house industrial‑scale human genetics engine, Regeneron has the unique capacity to translate protective genetic variants into drugs that target aging and age‑related disease. Although Regeneron’s $256 million bid to acquire 23andMe’s bankruptcy‑auctioned genetic database ultimately fell through, its in‑house Regeneron Genetics Center (RGC) has already sequenced more than 2.7 million de‑identified exomes as of 2025, making the company a genomics powerhouse regardless of the deal’s end result. And indeed, Regeneron has shown the ability to leverage its genetic database for pharmaceutical innovation before. Its development of evinacumab was developed as a result of genetic analysis pointing to ANGPTL-3 LoF mutations as protective against cardiovascular disease. And while not discovered by Regeneron themselves, the company leveraged UT Southwestern’s 2006 finding that PCSK9 LoF variants confer lifelong low LDL-C and reduced coronary risk to develop Praluent, the first FDA-approved PCSK9 inhibitor. Regeneron has similarly leveraged genetic insights into targets for NAFLD (CIDEB, HSD17B13), heart failure and hypertension (NPR1), dyslipidemia (APOC3), and chronic obstructive pulmonary disease (COPD) (IL-4, IL-13). On the investment side, Regeneron completed two notable hundred-million-dollar partnerships in recent years:

• Alnylam Pharmaceuticals (2019), a leader in RNA interference therapeutics currently running Phase II trials on mivelsiran, an amyloid precursor protein (APP) siRNA,

• Intellia Therapeutics (2023), a CRISPR/Cas9 platform initially acquired for its potential in hemophilia treatment, but now the driving technology behind NTLA-2001, a gene-editing therapy in Phase III trials for transthyretin amyloidosis (ATTR-CM), a cardiomyopathy that rises steeply after age 60.

Regeneron has made great strides in aging under Vice President Dr. David Glass, head of the corporation’s unit studying aging and age-related disorders. Glass’s team has published extensively on the role of ActRII ligands in muscle wasting (the company is currently researching the safety and efficacy of Garetosmab, an Activin A-blocking antibody), laying the groundwork for potential interventions aimed at preserving muscle function in aging populations. However, Regeneron remains seemingly reluctant to take bigger action in the field—for example, we have yet to see an aging-focused company in its public portfolio via the newly launched Regeneron Ventures. Ultimately, Regeneron is doing the kind of work (large-scale multi-omics, gene editing, etc) that could uncover new territory in aging biology, but it’s applying it mostly to conventional disease categories for now, earning it a middling grade.

Roche (C)

Roche (and its subsidiaries in Genentech and Chugai) has been uniquely successful in the diagnostics space, which few pharma companies have entered. In 2024, Roche received FDA Breakthrough Device designation for its pTau217 blood test, co-developed with Eli Lilly. While not a cure, diagnosis before symptomology could allow patients to, at the very least, enter clinical trials while brain tissue is still salvageable. On the investment front, the Roche Venture Fund committed $290 million in 2022 to Freenome, an SF-based company developing machine learning-enabled multiomics blood tests to detect colorectal and lung cancers at their earliest stages. And in 2023, Roche partnered with Alnylam Pharmaceuticals to commercialize zilebesiran, an RNAi therapeutic reducing liver angiotensinogen mRNA levels to lower blood pressure. Roche also recently acquired Carmot Therapeutics, an obesity therapeutics company, for $2.7 billion. Carmot stands out in a crowded market for its “biased agonism” strategy which aims to improve efficacy and tolerability, and Roche is reportedly considering employing the Berkeley-based company’s lead candidate, CT-388, in combination with Roche’s own anti-myostatin antibody RO7204239. Lastly, this July, Roche’s Japanese subsidiary Chugai Pharmaceutical entered a deal with Gero (the aforementioned Pfizer partner), an AI target discovery platform that screens longitudinal data. The agreement includes an upfront payment of $250 million, with potential earnings reaching $1 billion through royalties.

Notable projects in Roche’s pipeline also include, but are not limited to:

• Nivegacetor, a Phase II oral gamma-secretase modulator licensed from Merck that lowers amyloidogenic Aβ42 and Aβ40 while increasing shorter, less aggregation-prone Aβ37/Aβ38,

• RO7269162, a y-secretase modulator meant to reduce amyloid formation in a 90-week Phase II trial for those at-risk or in the prodromal stages of AD,

• The appropriately named GYM-329, an anti-myostatin antibody marketed as a muscle preservation companion in obesity treatment, and

• Trontinemab, an amyloid-clearing bispecific monoclonal antibody employing a similar transferrin-receptor BBB shuttling technology to Aliada Therapeutics, heading to Phase III. RO7121932, an anti-CD20 which uses the same Brainshuttle^TM as Trotinemab, is currently in Phase I trials for MS.

Already on the market is Roche’s intravitreal injection Vabysmo, a one-time treatment dual VEGF and Ang-2 inhibitor for age-related macular disease and diabetic macular edema, and the first bispecific antibody for any age-related disease beyond cancer.

In 2022, Genentech published a content piece entitled “Turning Back the Clock,” in which the company describes a collaboration with the Salk Institute testing the long-term safety of inducing Yamanaka factors in healthy mice. Near the end of the article, Genentech made a surprising statement: “Our scientists are now exploring whether partial reprogramming can be used to tackle age-related disease.” Since then, members of Genentech’s Regenerative Medicine team such as Drs. Alessandro Ori, Kristen C. Browder, Henri Jasper, and more have put out several publications regarding potential applications of stem cells and modulating Yamanaka factors in vivo, but, expectedly, there hasn’t been any publicity on Genentech or Roche seriously pursuing epigenetic reprogramming as a therapeutic strategy in humans.

Despite the company’s successes in trials and diagnostics, Roche has no declared programs in aging pathology prevention. Nonetheless, Chugai’s recent partnership with Gero to identify antibody targets in age‑related disease is highly encouraging, and the company’s proprietary Brainshuttle technology has high potential to be redeployed across CNS aging indications; taken together, Roche sits comfortably with a respectable C grade.

Johnson & Johnson (C+)

While Johnson & Johnson’s (J&J) may be better known for its leadership in oncology and immunology, the company’s greatest strides in the aging space, like those of its peers Roche and Sanofi, have been largely contained to the arena of neurodegeneration. J&J is the only major pharmaceutical company with two tau-clearing drugs running in parallel: monoclonal antibody posdinemab (currently in Phase 2b, labeled “AuTonomy”) and vaccine JNJ-2056 (in a Phase II trial called “ReTain”), which have both been granted FDA Fast-Track status. J&J stands out for its novel strategies; Posdinemab attacks the phosphorylated-tau mid-domain, an epitope understudied compared to N/ C terminal tau antibodies, and JNJ-2056 is the first immunotherapy approach targeting tau in preclinical AD. J&J is also holding its own in the race to commercialize a gene therapy for age-related macular degeneration with the development of JNJ-1887 (Phase II), designed to increase expression of soluble CD59, an immunoregulatory glycoprotein, to protect retinal cells from geographic atrophy.

J&J is also engaging with aging biology via its venture arm, Johnson & Johnson Innovation – JJDC Inc. Much of JJDC’s capital prioritizes oncology—one of its most notable investments in this field was a 2014 seed investment in Inivata, a UK-based liquid biopsy company that developed a ctDNA test for early lung cancer detection (acquired by NeoGenomics in 2021). But recent investments signal a growing interest in aging-relevant pathways. Ten years ago, JJDC co-led a $33M Series B for Navitor Pharmaceuticals, a platform developing mTORC1 pathway modulators; in 2022, Jannsenn acquired Navitor spinout Anakuria Therapeutics, which is advancing rapamycin analog AT-20494 in ADPKD and other indications. In 2024, JJDC led an oversubscribed Series D for ONL Therapeutics, a retinal degeneration company developing small molecule Fas inhibitor ONL1204 as a therapy for dry age-related macular degeneration. And this January, J&J finally exercised its option to acquire Arkuda's portfolio of lysosomal function enhancers for the treatment of frontotemporal dementia.

J&J’s public visibility in the aging space is also notable: in 2023, the company launched its annual Japan Smart Healthy Aging QuickFire Challenge, which invites Japan-based founders to submit promising science or technologies with the potential to increase healthy life expectancy. J&J committed $300,000 in grant funding, lab access, and mentorship to four startups. While J&J’s commitment to aging in the Asia Pacific region had long evaded the U.S., the New Jersey-based corporation has recently begun employing scientists to research aging and metabolism in its home country, a positive indication of what is to come. Although J&J’s overall footprint in the aging space is narrower than its peers with broader metabolic or cardiovascular pushes, the company’s leadership in neuro & ophthalmology, as well as exciting acquisitions like that of Anakuria, place the company on an encouraging upward trajectory and secure its upper‑tier C grade.

AstraZeneca (C+)

AstraZeneca (AZ), historically concentrated in oncology and Cardiovascular, Renal, and Metabolism (CVRM), has made some of its boldest moves in the longevity space over just the past two years. In 2024, AZ announced a government-backed partnership with the University of Cambridge to establish a new functional genomics laboratory at the Milner Therapeutics Institute (MTI) with the stated goal of “improv[ing] health, ageing, and wellbeing.” Last May, AZ invested in SixPeaks Bio, which is developing dual-specific activin type IIA and B receptor antibodies and marketing itself as the muscle-preserving companion to GLP-1 RAs. AZ committed $80 million in exchange for an option to buy out the company at the time of IND application submission. Earlier transactions include AZ’s $1.8 billion acquisition of CinCor Pharma in 2023, bringing in aldosterone-synthase inhibitor baxdrostat, now in Phase III alongside dapagliflozin, and a 2021 exclusive licence of Dogma Therapeutics’ first-in-class oral PCSK9-inhibitor program (advanced internally as AZD0780, Phase II). Thus, the British-Swedish multinational pharmaceutical giant’s promising catalog of cardiovascular disease drugs currently in trials includes:

• AZD1705, a Phase I ANGPTL3 inhibitor for LDL-C management for the prevention of atherosclerotic cardiovascular disease,

• AZD4144, a Phase 1b NLPR3 inhibitor in trials for patients with atherosclerosis and chronic kidney disease,

• Baxdrostat + Dapagliflozin (Phase III), a combination of an aldosterone synthase inhibitor and SGLT2 inhibitor, respectively, for the prevention of heart failure, and

• AZD0780, an oral PCSK9 inhibitor in Phase II with a key advantage over Merck’s enlicitide: the drug would not require fasting before or after taking the medication (enlicitide requires an 8-hour fast before and a 30-minute fast after each dose).

AZ is also investigating the link between cardiovascular disease and COPD prevalence, as well as advancing a Phase III monoclonal antibody, Tezepelumab, for the indication, which targets thymic stromal lymphopoietin (TSLP).

The company’s recent hiring posts suggest an internal shift toward prioritizing aging research, particularly in the context of fighting cancer. As AZ principal scientist Dr. Cindy Guidi wrote in a recent recruitment message, “Why aging, you might ask? Over 60% of cancer cases are diagnosed in those aged 65 and above, highlighting that cancer is fundamentally a disease of aging. However, most preclinical models focus on younger subjects. Our goal is to target the aging pathways to develop preventative and more effective treatments for a broader patient demographic.” This kind of language around aging research from a pharma company is promising, but without broader visibility and aging-centric investments, AZ still has a long way to go before earning a higher grade.

AbbVie (C+)

Best recognized for its dominance in oncology, immunology, neuroscience, and medical aesthetics, AbbVie’s involvement in aging research has been marked by sizable investments with many yet-unrealized returns. The company’s most public venture into the aging space was its partnership with Alphabet to co-fund Calico, a biotechnology company aimed at understanding the underlying biology controlling aging. AbbVie partnered with Calico in 2014 and renewed the investment in 2018 and 2021 with an additional $500 million each. However, Calico has been shrouded in controversy over the decade since its founding—even its creator, Bill Maris, who left Google in 2016, expressed disillusionment with the company’s lack of vision, public transparency, and clinical progress since his departure. Most recently, Calico’s latest ALS drug, Fosigotifator, failed to outperform placebo in Phase II/III trials. The company’s current clinical trials focus on diseases such as Autosomal Dominant Polycystic Kidney Disease (ADPKD) and inherited Vanishing White Matter Disease (VWMD)—worthy causes, but not necessarily linked to aging hallmark biology, raising questions about Calico’s mission alignment.

However, Calico has quietly regained momentum this year with the addition of a patented anti-pregnancy-associated plasma protein A (PAPP-A) antibody to its pipeline. PAPP-A is a protease regulating insulin-like growth factor 1 (IGF-1) bioavailability (the IGF-1 signaling pathway was first implicated in longevity by Cynthia Kenyon, age1 cofounder Laura Deming’s former mentor, who found out that inhibiting insulin/IGF-1 signaling slowed down aging in C. elegans). PAPP-A inhibition has been shown to extend lifespan in mice, and 2020 data by Calico scientists suggest it does so by depleting IGF-responsive mesenchymal progenitors across tissues, thereby reducing fibrosis and other hallmarks of aging. If successful, the antibody could be one of the first drugs on the market to pharmacologically modulate an aging hallmark upstream of disease.

AbbVie has joined smaller bets in the neurodegeneration space, co-investing with Pfizer in Aquinnah Pharmaceuticals, an innovative Cambridge-based company pursuing novel therapies for ALS and AD utilizing RNA-binding proteins, as well as investing in Aliada Therapeutics, which is targeting amyloid via a novel, patented high-precision CNS drug delivery platform—bispecific antibodies dually bind to 3pE-Aβ as well as BBB transport receptor TfR1. AbbVie has also advanced several programs within its PD pipeline, finally gaining FDA approval for Vyalev, a subcutaneous infusion therapy increasing CNS dopamine availability, after multiple resubmissions and a protracted regulatory process. AbbVie has also continued to develop the once-daily oral selective D1/D5 dopamine partial agonist tavapadon in clinical trials. Nonetheless, both therapies aim to address symptoms rather than reverse disease pathology. Slightly more innovative is ABBV-1088, a PINK1 activator (acquired through AbbVie’s buyout of Mitokinin) in Phase I trials designed to address mitochondrial dysfunction in PD. Lastly, AbbVie’s Surabgene Lomparvovec (ABBV-RGX-314), a potential one-time gene therapy targeting VEGF to treat wet age-related macular disease, is currently in Phase III. AbbVie Ventures has invested in multiple companies focused on PD disease treatments, such as Endlyz Therapeutics, Portera Therapeutics, Capsida Biotherapeutics, and Nitrase Therapeutics.

AbbVie also made a strategic 2024 $15 million investment in Oisín Biotechnologies, a Seattle-based company developing genetic therapies to combat frailty by inducing adipocyte-specific apoptosis and delivering DNA follistatin-encoding DNA, antagonizing the myostatin pathway. Oisín has already seen success in preclinical trials, and there is no doubt the company’s value will increase as GLP-1 agonists increase in popularity, creating demand for companion therapies that prevent muscle wasting and improve strength in aging and post-obese populations.

AbbVie, like many of its peers, has yet to develop any in-house geroscience pipeline, nor has it made many public statements about a commitment to aging biology outside of Calico. AbbVie gets credit for recognizing longevity as a field worth funding, but until Calico can translate its strength in basic science to a marketable drug, or internal R&D dedicates a team to aging research, it remains deserving of its C+ grade.

Novo Nordisk (B-)

Novo Nordisk is best known today for its dominance in cardiometabolic disease therapies alongside rival Eli Lilly; however, less publicized is the company’s aim to leverage its GLP-1 receptor agonist franchise into treatments for neuroinflammatory and cardiovascular disease.

Novo is currently funding two Phase III trials (evoke and evoke+) to evaluate the effectiveness of GLP-1RAs against early AD (the two trials reportedly have identical designs, with one difference: evoke+ allows participants with evidence of small vessel pathology (CSVD) on baseline imaging; evoke does not). The trials follow a post-hoc analysis in 2022 showing that patients with type II diabetes treated with GLP-1RAs had a statistically significant 53% lower risk of all-cause dementia diagnosis versus patients receiving placebo. The evoke/evoke+ trials are estimated to be completed in October 2026. If semaglutide can prove to be effective in even a modest slowing of Alzheimer’s, it would be among the boldest successful efforts to repurpose a metabolic drug for brain aging. Novo’s SELECT trial in 17,604 overweight patients without diabetes showed semaglutide cut all major cardiovascular events by 20% and significantly improved mortality outcomes—treatment-group participants had a 4.3% rate of all-cause death after a 3.5-year follow-up compared to 5.2% in placebo. Lastly, Novo Nordisk is also investigating the potential of amylin analog cagrilintide to preserve bone mass in postmenopausal women with obesity using semaglutide.

In addition to GLP-1 RAs, Novo is also pursuing therapeutics for many common targets, including an ANGPTL3 monoclonal antibody and an NLRP3 inhibitor for cardiovascular disease. However, the pipeline projects that stand out are those with unconventional modalities:

• STEM‑PD, a Phase I (now Phase II ready) study in eight participants determining the efficacy of intraputamenal transplantation of hESC-derived dopaminergic cells for PD, currently under development by Novo Nordisk’s academic partners.

• A T1D DNA vaccine (Phase I) meant to deliver a plasmid encoding pre- and pro-insulin to preserve beta cell function.

• CDR132L (Phase II), an antisense oligonucleotide targeting microRNA molecule miR-132 for the treatment of heart failure (the product of a 2024 billion dollar acquisition of Cardior Therapeutics).

• FUSE (Phase II), a once-monthly peripheral-focused high-frequency ultrasound (PFUS) treatment for diabetes and obesity meant to regulate glucose metabolism via CNS stimulation (the result of a collaboration with Chicago-based GE Healthcare).

• Zalfmerin (Phase II), A long-acting FGF21 analog for the once-weekly treatment of MASH (being developed in collaboration with Omega Therapeutics and Cellularity). FGF21 has long been regarded as a “pro-longevity” hormonal protein—if advanced through Phase III trials, it could become the first FGF21 analog approved by the FDA.

Beyond its ambitious internal efforts to target age-related disease, the Novo ecosystem has also engaged in multiple external initiatives. In recent years, Novo Holdings (Novo Nordisk’s largest shareholder) made several significant early-stage investments in age-related disease prevention, including, but not limited to:

1. A joint $15 million seed financing round for Booster Therapeutics, a 20S proteasome activation platform aiming to restore the body’s ability to clear AD and PD-associated proteins.

2. A $100 M Series C leading investment in Asceneuron, the prior-mentioned EMD Serono spinout targeting tauopathies backed by Merck and joined by JJDC in a 2015 Series A.

3. Participation in a $23 million Series A for Tribune Therapeutics (co-founded in 2020 by Novo Holdings’s Seed Investments team with HealthCap Ventures), a biotech advancing lead candidate and CCN family inhibitor TRX-44 into trials for fibrotic conditions.

4. Participation in a €80 million Series A for Antag Therapeutics to support development of AT-7687, a once-weekly subcutaneous GIPR antagonist. Importantly, AT-7687 improved glycemic control and lipid profiles independent of weight changes in non-human primates, and achieved weight loss sans GI side effects.

5. Participation in a 2024 oversubscribed $95 million financing for Lexeo Therapeutics ($1.12 million worth in equity), an NYC-based biotech developing genetic medicine candidates for CVD and APOE4-associated AD.

The Novo Nordisk Foundation is also currently funding grants for PhDs and postdocs studying metabolic aging. In 2023, Novo Nordisk also joined the Hevolution Foundation’s Breakthrough Innovation Alliance, which funds early-stage aging research ideas—Novo sits on the committee that selects projects to receive incubator grants. While Novo Nordisk’s focus is still on specific age-related diseases rather than aging itself, the Danish multinational pharmaceutical company has demonstrated a clear commitment to leveraging metabolic research into broader age-related disease applications and is actively engaging with the early-stage longevity ecosystem through partnerships and investments, earning them a B-.

Novartis (B)

Novartis has shown interest in the longevity space for nearly a decade now. Even in 2014, the company conducted clinical trials of cancer drug and mTOR inhibitor everolimus, exploring its effects on improving the older adults’ immunity following the success of famed mTOR inhibitor rapamycin in extending mouse lifespan. While the trial attracted public interest, it was short-lived—the drug’s patents were nearing expiration before the trial’s end, and Novartis offloaded its aging-focused mTORC1 inhibition program to resTORbio in exchange for an equity stake. resTORbio IPOed in September 2018 and focused on developing RTB101, a selective TORC1 inhibitor, but shortly after the lead candidate failed in 2019 Phase III trials for symptomatic respiratory illness prevention, resTORbio completed a reverse merger with Adicet Bio, an allogeneic CAR-T cell therapy company.

Nonetheless, Novartis has continued to invest in longevity and has made several major moves in the space since then. In 2022, Novartis granted Cambrian Biopharma the license to develop next-generation selective mTOR inhibitors via subsidiary Tornado Therapeutics, a deal that allows Cambrian to advance these rapalogs in aging research while Novartis earns royalties. And a year earlier, Novartis acquired Gyroscope Therapeutics, a gene therapy company developing a one-time treatment for advanced-stage dry age-related macular degeneration. This June, Novartis struck a four-year partnership deal with ProFound Therapeutics, a protein detection platform company, committing $25 million upfront in addition to $750 million in milestones per identified cardiovascular disease target.

In 2023, however, Novartis stepped away from its longtime strategy of acquisitions and IP contributions for a financial stake rather than initiating internal efforts, and launched an in-house unit for Diseases of Aging and Regenerative Medicine (DARe). According to Dr. Michaela Kneissel, who joined Novartis nearly thirty years ago and is now the global head of DARe, the goal of the center is to “understand the biological drivers of aging to develop novel treatments for diseases related to aging.” The unit appears to be currently focused on exercise mimetics, a newly emergent field in aging biology that no other pharma has yet publicly pursued. At the 11th Aging Research and Drug Discovery (ARDD) conference last year, DARe representative Dr. Mara Fornaro shared several ambitious projects in the center’s pipeline, including but not limited to:

• LNA043, an FDA fast-tracked ANGPTL3 agonist that stimulates cartilage tissue regeneration in osteoarthritis (a promising project, but discontinued after a poor Phase II readout),

• Axatilimab, a colony-stimulating factor 1 receptor (CSF1R) inhibitor targeting inflammatory microglia to reverse age-associated neuromuscular changes,

• cGAS/STING (cyclic GMP-AMP Synthase, Stimulator of Interferon Genes) pathway inhibition (the small molecules for which were acquired for $90 million last year following a 2019 option and collaboration agreement with parent company IFM Therapeutics). cGAS/STING pathway upregulation drives excessive interferon/cytokine signaling linked to several age-related diseases.

Novartis’s DARe group is also recruiting researchers into its Basel-based Postdoctoral Fellowship Program, which aims to discover new methods for maturing patient-derived iPSC neurons to full adulthood and then accelerating their aging to model neurodegenerative pathologies for target discovery.

Novartis’s pipeline includes two neurodegeneration trials to watch: a Phase I tolerability study of NIO752, an intrathecally administered tau-targeting antisense oligonucleotide, for the treatment of early AD, and VHB937, a macrophage-specific TREM2-activating monoclonal antibody in Phase II for the treatment of ALS. Like Sanofi, Novartis appears highly focused on targeting inflammation as a root driver of aging—in addition to the above, DARe is also developing DFV‑89, an oral NLRP3 inflammasome inhibitor, and RHH‑646, a novel stimulator of cartilage tissue regeneration in knee osteoarthritis. The small molecule’s mechanism of action has not yet been made public, but it is a significant milestone in pharma’s growing acceptance of regenerative medicine in diseases of degeneration. Having discontinued the drug in a Phase II trial for IPF last year, Novartis is now progressing LTP001, an E3 ubiquitin ligase SMURF1 inhibitor, through Phase I/II for pulmonary arterial hypertension. And in 2021, Novartis's inclisiran became the FDA's fourth approved siRNA-based drug, and the first-ever siRNA approved for LDL-C reduction. The drug, developed under a license and collaboration agreement with the aforementioned Alnylam Pharmaceuticals, was revolutionary for its twice-a-year dosing regimen. Inclisiran is currently in trials for the primary and secondary prevention of cardiovascular events in high-risk patients. Novartis has also launched clinical trials of pelacarsen (TQJ230), an antisense inhibitor of apolipoprotein (a) synthesis, designed to reduce LDL‑C, lower lipoprotein (a) levels, and ultimately cut major adverse cardiovascular events (MACE).

In 2024, Novartis entered a collaboration with BioAge Labs, involving a $20 million upfront payment and potential future earnings of up to $530 million. The partnership, as stated in BioAge's press release, aims to combine "BioAge's extensive proprietary human longevity datasets and Novartis expertise in exercise biology." Paired with Novartis's early mTOR work, its dedicated DARe unit, and prevention‑focused cardiovascular trials, Novartis shows a consistent willingness to engage with fundamental aging mechanisms as well as put risk-reduction and resilience first in clinical trials, a visionary strategy not many of its peers are willing to follow just yet.

Eli Lilly (B)

On the product front, 2023–2025 has been extraordinary for Lilly. The company gained FDA approval for Zepbound (tirzepatide), a GLP-1/GIP twincretin, in 2023; this June, the company announced positive safety and efficacy results for Orforglipron, the first oral small molecule GLP-1 to successfully complete a Phase III trial. ACHIEVE-1 is only the first of seven Phase III trials for the drug, which is also being investigated as a potential treatment for obstructive sleep apnea and hypertension in obese adults. As is Novo Nordisk with Ozempic, Lilly’s tirzepatide has shown success in cardiovascular health trials, boasting a 38% reduction in heart failure, and notably, tirzepatide demonstrated superior weight loss compared to semaglutide in head-to-head trials. Ongoing trials such as SURMOUNT-MMO (Phase III), a landmark morbidity and mortality study in obesity, are positioning tirzepatide not just as a weight loss drug, but as a platform for systemic risk factor compression. In addition to cardiometabolic disease, Lilly is currently running trials dedicated to understanding tirzepatide’s impact on osteoarthritis, psoriasis, and MASH, all diseases with age as a primary risk factor.

At the 2023 Aging Research and Drug Discovery (ARDD) conference, Lilly VP Dr. Benjamin Yaden strongly advocated for the treatment of aging-related processes as a strategy to prevent disease in a discussion panel with Novo Nordisk Senior Director Erik Vernet. The company puts its money where its mouth is—Lilly’s 2025 pipeline is firing on multiple cylinders (cardiometabolic health, neuroinflammation/degeneration, and gene therapy) to conquer age-related diseases. RNA-targeting treatments have also become a strong point for Lilly as the corporation advances projects such as:

• Microtubule-Associated Protein Tau (MAPT)-targeting RNAi therapy for tauopathies,

• siRNAs targeting triglyceride lipase PNPLA3 and cholesterol sensor SCAP, key regulators in MASH pathophysiology, and

• The ANGPTL3 siRNA solbinsiran. As of this April, at least nine distinct ANGPTL3-targeting therapies have entered trials since the first-in-class drug was approved in 2021.

The company’s corporate business development page also highlights “longevity-related indications” as an area of interest within its cardiometabolic health unit, as well as “sarcopenia,” mitochondrial health,” and “next generation oral injectable approaches on myocardium and heart function, e.g., mitochondrial function, inflammation, repair/regeneration,” exciting avenues for future research.

As was mentioned earlier, Eli Lilly is invested in lysosomal enhancement innovator Arkuda Therapeutics alongside Pfizer. Eli Lilly’s internal pipeline efforts in the neurodegeneration space have also proven successful, with Donanemab (Kisunla™) becoming the third amyloid-targeting pharmaceutical to receive FDA approval last July—Lilly is currently recruiting participants for a Donanemab trial in patients with preclinical AD. The company has also partnered with Juvena Therapeutics and acquired Versanis Bio in its effort to develop therapies that preserve muscle mass in the pursuit of weight loss. Versnais Bio bought bimagrumab, an activin type II receptor (ActRII), from Novartis in 2021 after an unsuccessful Phase II/III trial for sporadic inclusion body myositis. Years later, Lilly integrated bimagrumab into its own Phase II pipeline for obesity, signaling a clear understanding of the metabolic–musculoskeletal implications of its GLP-1 therapies. This June, Lilly announced the results of Phase II trials for a combination therapy of semaglutide and bimagrumab, reporting highly encouraging results: participants in the highest-dose cohort lost 22% of their body weight over 72 weeks, with an impressive 93% attributed to fat loss. Lilly is also advancing Phase I trials of ActRIIA antibody LAE102 in healthy postmenopausal women as part of a collaboration with Laekna Therapeutics. In December 2023, Lilly entered a multi-year partnership with FaunaBio to apply its Convergence AI platform to identify novel targets in obesity treatment.

This January, Lilly committed up to $780 million in a global licensing deal with Mediar Therapeutics, a Boston-based biotech developing MTX-463, an anti-WISP1 antibody for the treatment of idiopathic pulmonary fibrosis (IPF). WISP1 is a matrix protein significantly upregulated in age-related musculoskeletal diseases; as such, WISP1 holds great potential as a novel therapeutic target. Mediar is also developing an anti-fibrotic targeting matricellular protein SMOC2, a novel target for fibrosis diseases. In 2024, Lilly joined a16z and others to finance a $43 million Series A for Arda Therapeutics, a San Francisco-based biotech developing cell-depletion therapies for age-related chronic diseases.

Most recently, Eli Lilly acquired Verve Therapeutics, a Boston-based company producing genetic therapies for cardiovascular disease, for $1.3 billion. While Verve is still focused on disease treatment at the moment, company leadership does envision a future in which its gene editing technologies are utilized to prevent age-related phenotypes and diseases. Verve’s lead program (VERVE-102), currently in Phase 1b trials, is an in vivo gene editing therapy targeting PCSK9, promising a “one-and-done” treatment method. Verve, and thus Eli Lilly’s strategy, stands in juxtaposition to that of last-place finisher Merck, whose peptide-based approach to PCSK9 inhibition lacks the ambition and future-oriented vision of pioneers like Lilly, granting it one of the highest grades on our report card.

Takeaways

Taken together, age1’s report card reveals an industry that has crossed the line of intriguing but unactualized research to clinical development. Most major players we scored are investing millions, if not billions, into next-generation therapies for age-related diseases. Our top performers, however, are companies that have made aging biology an explicit focus, whether through public visibility efforts, pipeline standouts, partnerships, investments, acquisitions, internal platforms, or any combination of the above. Novo Nordisk has managed to parlay its GLP-1 franchise into a broader metabolic aging strategy. Novartis’s DARe group has already displayed impressive progress in its aging pipeline, and the BioAge Labs collaboration is certainly one to watch. In a pharma panel on preclinical development at the 2024 Aging Research and Drug Discovery (ARDD), Eli Lilly reportedly publicly announced a commitment to being a “longevity company,” a major tonal shift for the industry as a whole, and an exciting sign of what may be to come. It is also telling that representatives from our three top picks have been selected as speakers for ARDD 2025, highlighting their leadership in the field.

Large pharmaceutical companies have advantages in the longevity arena: they have vast capital to finance large-scale trials like Novo Nordisk’s SELECT that smaller biotechs could never afford. They also possess seasoned clinical development teams and established relationships with both payors and agencies like the FDA, accelerating complex approval processes—their track record of established market therapies gives them leverage to negotiate new drugs. Perhaps above all, pharma giants can acquire proprietary data on a scale unattainable to most startups. In the age of artificial intelligence, the ability to mine millions of genomes may be the decisive advantage for discovering novel targets. As aging has long been known to be highly polygenic and polyomic, companies like Regeneron and Novartis, with resources like the Regeneron Genomics Center and BioAge’s proprietary longitudinal data, respectively, will propel massive breakthroughs in the coming years.

It would be naïve not to spell out why many large drug makers still hesitate to run trials in aging and resilience. For one, a McKinsey analysis of the top 36 biopharma companies’ assets launched since 2000 found a significant reduction in the average time assets took to accrue 50% of lifetime sales. In fact, most drugs reach 80% of their lifetime sales just ~35 months post-launch. This means that in an era of rapid clinical development and competition, R&D must promise fast monetization, putting pressure on pharma to generate revenue by turning to later-stage assets showing preliminary clinical efficacy with a defined regulatory and commercial path, criteria where aging therapies fall short today. McKinsey also notes that these data are compounded by loss-of-exclusivity (LOE) pressures. Blockbuster patents for drugs like Keytruda, Eylea, and Stelara, each generating billions annually, will soon go generic. Add to this the fact that payor economics are currently misaligned with prevention—the average insurance turnover for a commercially covered patient switches every three years, dissuading payors from covering a therapy with benefits that might not materialize for decades. These bottlenecks explain industry caution, but they also highlight actionable opportunities for pharma. Companies prepared to invest in discovering surrogate endpoints, work to select cohorts with higher baseline risk, and implement multimorbidity endpoints to reduce sample size requirements, and engage in early-stage discussions with the FDA about trial design (e.g., running adaptive trials in aging to allow for fast iteration) may reduce risk by bringing aging drugs to meaningful readouts sooner. 

Because the FDA still treats aging as a biological process rather than a condition, there is no established regulatory pathway (or U.S. reimbursement precedent) for a drug whose primary claim is “slowing aging” in humans. Excitingly, however, the ICD-11 codes now list “ageing associated decline in intrinsic capacity,” That code is not yet recognized in U.S. billing systems, but could be adopted soon. However, until then, large companies continue to channel capital toward aging-adjacent indications such as osteoarthritis and HFpEF, where endpoints and price benchmarks already exist. Truly paradigm-shifting longevity advances such as epigenetic reprogramming, menopause delay, and regeneration of post-mitotic tissue are largely absent from pharma pipelines because such projects lack an obvious regulatory path. Startups are not immune to this constraint, but equity financing may let them chase bold biology first and pin a conventional “proxy disease” label on the asset later in development.

Because the regulatory environment remains nascent and ill-defined, pursuing aging does inherently require that founders and pharma take on more risk in pursuing targets in this space. But the payoff can be immense (see age1’s 2023 market analysis). When the first GLP-1 agonist therapies for weight loss (Novo Nordisk’s Saxenda) arrived in 2014, the commercial landscape for it did not yet exist. There was no HCPCS J‑code (today, GLP-1 receptor agonists fall under code C5015-A), and most insurers refused to cover the treatment. Nonetheless, as more efficacious alternatives began to emerge in the early 2020s, the commercial environment was forced to adapt. By 2024, 18 % of large employers covered GLP‑1s for obesity (up sharply from single-digit estimates pre-Wegovy), and 15 states now cover GLP-1s in their Medicaid program. Longevity is at a similar pre‑infrastructure frontier. Pharma companies, with their scale and commercial familiarity, are in a unique position to surface transformative, mechanism-validated interventions through internal build, minority equity stakes, and M&A. Therefore, we contend that pharma companies should begin deploying capital into high‑potential aging-thesis driven platforms now. With patent expiries looming, competition intensifying, and development timelines tightening, early movers have the opportunity to shape the clinical and commercial landscape for an entire class of drugs and advance the next inevitable frontier in therapeutics development. Prove efficacy, and reimbursement and access frameworks will adjust accordingly.

Appendix:

While this article aimed to be as comprehensive as possible, there were several pharma companies we could not profile (Bayer, BMS, GSK) for the sake of article length, as well as many notable recent investments, collaborations, partnerships, and acquisitions in the age-related disease space made by many of the pharma companies profiled and their respective venture arms that we were unable to profile exhaustively here. These include:

AstraZeneca: Cincor Pharma, Icosavax, Caelum Biosciences, Dogma Tx, Silence Therapeutics, Ionis Pharmaceuticals, CSPC Pharmaceutical Group, Redx Pharma

Eli Lilly: Solu Therapeutics, Therini Bio, Prevail Therapeutics, Ampersand Biomedicines, Disarm Therapeutics, Haya Therapeutics, Laverock Therapeutics, Nido Biosciences, Regor Therapeutics, Orso Bio

JJDC: ​​Cardiac Dimensions, AviadoBio, Nanochon, SpectraWAVE, CuraSen Therapeutics, Perceive Biotherapeutics, Shockwave Medical, Anakuria Therapeutics

Merck: TRIMTECH Therapeutics, EyeBio, Cerevance, Acceleron Pharma

Novartis: Anthos Therapeutics, AstronauTx, Borealis Biosciences, FundaMental Pharma, ONL Therapeutics, Pliant Therapeutics, Regulus Therapeutics

Novo Nordisk: Avalyn Pharma, AnaCardio, NorthSea Therapeutics, Muna Therapeutics, Tanai Therapeutics, Eikonizo Therapeutics, Gensaic, Aspect Biosystems, Dicerna Pharmaceuticals, Corvidia Therapeutics

Pfizer: ReViral, Mediar Tx, MindImmune, Voyager

Sanofi: Vicore Pharma, Resalis Therapeutics, AgonAb, Gyroscope, Therini Bio, Vicebio

A huge thanks to Satvik Dasariraju and Alex Colville, who pored over each draft, filled any gaps, and guided the direction of this piece. Thank you to Lily Clayton as well for the beautiful graphics! Lastly, special thanks to Lauren Nam for sharing her unique perspective on the obstacles faced by pharma.

Table 2: Strategic Identities of Profiled Pharma Companies

But as “aging” (and “longevity”) have become more mainstream, there’s a temptation to stretch the definition to things that don’t belong in the category. Aging is not a single process. It’s a complex interplay of multiple biological processes that, over time, become dysregulated. Understanding it as such is crucial for developing effective interventions. So it’s worth thinking about what makes something genuinely an “aging” biotech approach, and how to tell if you’re really tackling aging.

What’s The End Goal?

A successful aging treatment would be something that:

• prevents diseases of aging, ideally more than one;

• preserves a healthy function that normally declines with age (like fertility, immune function, cognitive function, resilience, or physical fitness); or

• reverses the course of at least one age-related disease.

This seems pretty obvious, but it has non-trivial implications.

If the goal is to prevent or delay age-related disease and loss of function, then we’re envisioning a treatment that people start taking when they’re healthy (no clinical disease). It would look like “Take this pill and reduce your risk of heart attacks, diabetes, cancer, and so on.”

That’s not unprecedented. People with high cholesterol take statins to reduce heart attack risk, and people taking GLP-1 receptor agonists for diabetes or obesity have a lower risk of all-cause mortality.

But it does present special challenges.

A drug that prevents age-related diseases needs to be safe enough for a healthy person to take. And, it needs to target a clinical endpoint that’s relevant to preventing age-related disease, but doesn’t take an unrealistically long time to test in trials.

Alternatively, if your goal is to reverse age-related disease, you’re often trying to intervene on seriously damaged tissue.

Tackling a neurodegenerative disorder? Once a patient is diagnosed with Alzheimer’s, for example, they’ve already lost millions of neurons. Permanently.

Likewise, some other diseases of aging, like chronic kidney disease and cirrhosis, involve fibrosis, in which healthy tissue is converted to scar tissue.

In other words, reversing the progression of degenerative, age-related disorders would require replacing healthy cells that have been lost, or transforming damaged tissue into healthy tissue – essentially, true regenerative medicine or cellular reprogramming.

What’s Not Aging?

Some disease indications are generally bad signs for an R&D program’s prospects for ever coming up with a bona fide aging therapeutic.

This is not to say that developing treatments for these diseases isn’t worthwhile. It’s just a different endeavor from aging.

Oncology

Cancer is so deadly that we’ll tolerate pretty severe side effects in a cancer drug that works at all. Most cancer drugs work by killing cancer cells (and, we hope, not too many healthy cells). So, few effective cancer treatments are safe enough for a healthy person to take. Oncology is generally the wrong disease indication if you’re looking for a geroprotector.

Progerias

Some genetic disorders cause a so-called “accelerated aging” syndrome, or “progeria,” that in some ways mimics the health problems of old age. However, these are usually single-gene disorders, and real aging is more complex; monogenic aging differs from polygenic aging. A progeria treatment might be life-changing for these rare-disease patients, but it’s unlikely to translate to the average elderly person.

Aging Biomarkers and Clocks

Some molecular biomarkers (such as molecules found in the blood) change in levels with age. Some of these age-related changes, such as alterations in DNA methylation, have even been interpreted as “aging clocks,” with the idea that they can indicate an individual’s “biological age” or “rate of aging.” A natural idea is to try to target aging biomarkers – if you change the level of the biomarker, perhaps you’ll have “rejuvenated” the organism!

There are a few wrong equivocations:

One that has become a central debate in the aging field is that “aging biomarkers” are derived from studies indicating correlation, not causation. If you make an “old” biomarker phenotype look “youthful,” have you reversed a process of decline and dysfunction? Or have you blocked a protective mechanism that’s more needed in old age? Without more information, we can’t tell.

Then, even correlational studies generally show an association between the aging clock and chronological age, not actually an individual’s risk of death or disease in the future! Correlations between aging clocks and mortality tend to be weak, and shrink in bigger samples and after correcting for cell type distributions. Many “aging biomarkers” and “aging clocks” are worse at predicting future mortality than simply asking patients to rate themselves on overall “health.” Manipulating clocks could actually be neutral or even harmful.

If a startup claims to target aging biomarkers or aging clocks, or to measure them as a diagnostic, without committing to generating additional evidence that they’re doing something causally meaningful, they’re probably not going to move the needle on aging.

What Is Aging?

Some disease indications and endpoints seem especially promising because they’re relevant to the majority of elderly people and a wide range of age-related diseases, but are themselves measurable and clinically significant in the short term.

Frailty/Sarcopenia

Loss of muscle mass and strength with age is an important risk factor for dangerous falls and general disease vulnerability. Simple physical-function tests such as gait speed or the Short Physical Performance Battery (SPPB) consistently outperform individual blood biomarkers, and often rival sophisticated multi-marker panels, in predicting long-term mortality. A treatment that makes older adults less frail will make a vast difference to quality of life and likely reduce the risk of multiple age-related diseases.

Inflammaging or Immune Dysfunction

Age-related immune dysfunction underlies multiple diseases of aging, from cancer to dementia. The aging immune system simultaneously causes more inflammation and is less effective at fighting off pathogens and cancers. Improving immune function tackles a root cause of age-related disease, but reducing the risk of (especially respiratory) infection is also immediately measurable and consequential, since influenza and COVID-19 are leading causes of death in the elderly.

Cardiometabolic Disease

Obesity, insulin resistance, and metabolic syndrome are risk factors for nearly every age-related disease, and there’s now strong precedent for treating them with drugs directly, with or without a diabetes diagnosis. Because metabolic dysfunction precedes cardiovascular, neurodegenerative, and fibrotic disease, drugs that target it could be viable “geroprotectors.” GLP-1 agonists already show substantial cardio-renal benefits in humans and, in preclinical work, reverse multi-omic aging signatures and modestly extend lifespan in fast-aging mice. So far, though, no study has lengthened lifespan in healthy mammals or humans.

Fibrotic Disorders (like IPF)

Fibrosis, like chronic inflammation, is a tissue-level process that contributes to multiple diseases of aging. Some specific diseases, like idiopathic pulmonary fibrosis, are almost purely disorders of excessive fibrosis, and a generally effective and safe anti-fibrotic drug may have a broader range of applications.

Prodromes

Some neurodegenerative diseases, like Parkinson’s, have a prodrome that can begin to show distinctive symptoms (like loss of sense of smell and certain sleep disorders) up to twenty years before Parkinson’s itself is diagnosed.

Treatments that target prodromes, with the aim of both treating immediate symptoms and preventing or delaying disease onset, have a better chance of striking at the root of these disorders.

Lifespan and Mortality

Humans live a long time, so a longitudinal study that looks at all-cause mortality is expensive and slow. But such studies do exist and can show mortality benefits from chronic use of some drugs. Also, veterinary aging biotechs like Loyal are now directly treating lifespan as an endpoint, and are entering the clinic to test whether their drug candidates can help dogs live longer. Directly measuring longevity and all-cause mortality isn’t easy, but those who do it can provide unquestionable evidence that they’re really tackling aging and age-related disease.

There are also methodological approaches that seem like good signs that a company is likely to make progress on aging and age-related disease.

How do we solve aging?

Realistic Aging Models

Aged organisms are different from “disease models” that may simply have a single experimentally or genetically induced problem. Approaches to aging therapeutics are more credible if they work on:

• Aged organisms (e.g., mice >20 months)

• Spontaneously occurring disease in model organisms

• Human-derived tissue samples from older patients

They are also more credible if they draw from evidence about aging in non-model organisms (i.e., not the usual lab mice, rats, flies, and worms).

“Delay”, “Replace”, “Restore”, and “Pause” Approaches

• Delay – postpone the onset of age-related decline (the most common strategy).

• Replace – swap aged or damaged tissue for youthful counterparts (cell therapies, 3D-printed organs, xenotransplants).

• Restore – reverse cellular aging (partial or full reprogramming), pending safety validation.

• Pause – induce biological stasis (torpor, hibernation, cryopreservation) to halt the clock, useful for organ preservation.

These approaches, as this blog has previously noted, tend to involve newer therapeutic modalities like cell and gene therapies, not just the more traditional small molecule drugs.

Imagine your task is to maximize the amount of time a boulder takes to roll down a hill.

What are your options? 

1. Apply some force of friction to slow down its acceleration downwards.

2. Push the boulder back uphill.

In the field of longevity, we are putting nearly all of our time, energy, and resources into option (2), via trying to reverse aging with rejuvenation techniques. Even worse, we almost always begin interventions once the boulder is already barreling at full speed, with every hallmark of aging manifesting itself in detrimental ways. Wouldn’t it be far more effective to act on the forces acting on the boulder at the top of the hill, and slow it down before it reaches high velocity?

TL;DR:

• The most effective way to slow down the boulder is by addressing genome instability, ideally as early as possible.

• There are clinically relevant areas for intervention where clinical trials could show promising results.

In a piece coming out in a few days (stay tuned!), I'll dig deeper into the science underlying these interventions and lay out a new framework for looking at the field of genome instability in the context of aging.

Table of Contents

• Introduction

• The Idea

  • What is Aging, Anyway?

  • Why Genome Instability?

  • Where are We Now?

• Charting the Course

  • Tier 1: Prevent DNA Damage

  • Tier 2: Repress Genome Destabilizers

  • Tier 3: Activate Genome Stabilizers

  • Tier 4: Beyond Small Molecules

• Places to Intervene

  • Prevent Cancer

  • Delay Menopause

  • Slow Neurocognitive Disorders

  • Stabilize the Brain Post-Trauma

  • Hamper Fibrotic Disorders

• Venturing into the Unknown

• Acknowledgements

• Appendix 1: Challenges

  • Models

  • Readouts

  • Clinical Trials

• Appendix 2: Opportunities

  • Decades of Cancer Research

  • Screening and Artificial Intelligence

• Appendix 3: Labs Currently Working on Genome Instability in the Context of Aging

• Appendix 4: Companies Addressing Genome Instability

• Appendix 5: Interventions Improving Genome Stability

The Idea

What is Aging, Anyway?

This is a contentious topic, with many differing opinions and viewpoints. Most published work on this tends to have a fairly lowest-common-denominator definition of aging that is agreeable to most but is also quite vague. Take this definition written by the Biomarkers of Aging Consortium:

The vagueness makes it challenging to concretely define whether or not an intervention “reverses aging” or not. Per the original analogy - is the intervention truly pushing the boulder back up the hill? Does Paxlovid reverse aging, as it is a treatment for COVID? COVID is, after all, a “consequence of life…. that leads to functional decline.” Do GLP-1 agonists reverse aging as they ameliorate age-related metabolic pathology (interesting debate on this here)? Does rapamycin reverse aging?

I’ll propose a different definition of aging. Let’s go back to the rolling boulder example from before, and imagine the boulder reaching the bottom of the hill as representing death.

The distance the boulder has traveled represents biological age - your risk of mortality and onset of age-associated diseases. This tracks nicely with the fact mortality increases exponentially with age (source):

Just as there is a constant force (gravity) forcing the boulder down the hill, so there is a constant force (the root cause(s) of aging) causing us to age. See the below graphic for clarity:

This is not an academic exercise - the implications of thinking about aging as a boulder rolling down a mountain are profound. Firstly, the ability to slow biological aging becomes exponentially more difficult later in life, so late-life interventions for individuals in their 80s are (literally) fighting an uphill battle. Secondly, claims of “reversing biological age” require a decrease in the risk of age-related diseases with time, which has yet to be demonstrated by any therapy in mice or humans. Thirdly, if the experiences of biological aging are exponential, then the rate  of aging should have a linear (non-exponential) relationship with time and the root cause(s) of aging should be constant.

Lastly, and most importantly - by far the easiest way to increase the amount of time for the boulder to roll down the hill is not by trying to push it directly, but rather slowing its progression down the hill. This means tackling the acceleration of aging - its root causes. Thus, I call for improving genome stability.

But why genome instability?

(for the curious reader: the above model for aging is somewhat simplified, but it is remarkably similar to ideals posited by a few authors in the past [1] [2] [3])

Why Genome Instability?

If the effects of aging (such as mortality) have an exponential relationship with time, then the forces accelerating it (the root cause(s) of aging) should be constant across the lifespan, thus affecting a linear degradation in immediate downstream targets. Thus, a good way of hunting for the root cause(s) of aging is to find aging-associated hallmarks that change linearly with time. Here’s the boulder graphic again for clarity:

Linear relationships with age are relatively hard to come by. Protein aggregates, at least in yeast, show an exponential increase with age. Also, accumulation of many senescence markers (those related to age-associated nondividing & likely-damaged cells) in mice appears to dramatically increase towards the end of a mouse's lifespan. But what about something related to genome instability - such as somatic mutation frequency (from this paper)?

These look remarkably like straight lines.  This is even more interesting when combined with the finding that average lifespan scales incredibly closely to mutation rate across species.

This effect isn’t limited to mutation rate, either. DNA methylation patterns in rats, dogs, and baboons (though probably not mice!) all shift with age in a linear fashion:

The evidence for the failure of genomic maintenance being a primary driver of aging is not limited to graphs with straight lines. When I initially entered the aging field, I was surprised to learn that every single human disease with a strong claim of being a form of accelerated aging has a mutation in a genome maintenance gene. Several years later, I have yet to find a satisfying reason for this to be the case without accepting DNA damage to be a significant cause of aging. Werner’s syndrome patients - those suffering from a mutation in a key cell replication and telomere maintenance gene - show multifold signs of aging, including graying hair, early onset diabetes, and cataracts. It’s worth noting that these patients do not demonstrate an accelerated form of every aging symptom (more on this in a soon-to-be-released piece), but they certainly do look much older:

If impairing genome stability shortens lifespan, does improving genome stability increase lifespan? Early data suggests this to be the case. A cohort of centenarians has been demonstrated to have a more effective variant of the critical epigenetic maintenance protein SIRT6. A genetic variant in FOXO3A, a gene encoding a protein involved in the DNA damage response (among many other functions), is one of the few that has been conclusively linked to human longevity. This also applies across the animal kingdom - DNA repair pathways are consistently more present in longer-lived species and quieter in those that die sooner.

Where are We Now?

The microenvironment of individuals picking up the mantle of fixing genome stability in the context of aging could best be described as a “niche.” The field is, more than anything, currently focused on understanding the mechanistic drivers relating genome stability to aging. Nevertheless, the first glimmers of an effort to intervene in the root cause of aging can be found, both in academia and industry. To contextualize these endeavors in increasing levels of difficulty, I have split them into four buckets  of approaches that improve genome stability by:

1. Preventing DNA damage and instability.

2. Inhibiting factors that prevent DNA repair and stabilization.

3. Activating processes that reinforce DNA repair or organization.

4. (Next-generation) Harnessing gene editing or capabilities from other animals to improve genomic maintenance.

Charting the Course

Tier 1: Prevent DNA Damage

The most elegant way to prevent genome instability is to stop damage from happening in the first place. Unfortunately, this is a challenging proposition. DNA damage can come from nearly anywhere, including the sun and reactive oxygen species (ROS) derived from normal metabolism. Equally unfortunately, UV light is required for production of vitamin D, and ROS are used as part of healthy signaling processes throughout different tissues in the human body. High intake of antioxidants, the most well-known compounds thought to prevent DNA damage, is not associated with lower all-cause mortality. Cells have evolved incredibly sophisticated mechanisms for managing potential DNA damaging agents in a manner that makes modulation difficult.

Opportunities for interventions are still possible, however, particularly if implemented in a way that works cooperatively with cellular machinery. Taurine supplementation shows improvement in lifespan in mice while decreasing multiple types of DNA damage. Vitamin D may be protective against UV-induced DNA damage via an interesting mechanism involving preservation of expression of long genes. Many other supplements and natural products have been suggested to have effects on DNA damage prevention. Unfortunately, the literature on them remains murky at best - the effect sizes tend to be small, the compounds themselves prone to false positive readouts, and the dosing required to see an effect is frequently extremely high. However, the promise here is high - If a compound was discovered that could reliably block a type of DNA damage without considerable downside effects, it would be the holy grail of the genome stability field. So far, however, the results have been largely disappointing.

Tier 2: Repress Genome Destabilizers

Repressing a genetic pathway is the bread-and-butter of the biopharmaceutical industry. The best way to develop a drug is to find a chemical that blocks a deleterious pathway. If that’s not possible, find a chemical that blocks an inhibitor of a beneficial pathway. Unfortunately, this approach finds limited traction in the genome instability world. There are, generally speaking, precious few reasons for a cell to inhibit its own gene maintenance pathways unless it is actively dividing or apoptosing (in which case, it is generally not a good idea to meddle). Still, there are a few possibilities here. 

Some intriguing possibilities for blocking repressors of genes that improve genome stability could be found in germline cells, as they are generally thought to be more effective at genomic maintenance. The Schumacher lab demonstrated that the inhibition of one such repressor, DREAM, led to an increase in DNA repair capacity. Unfortunately, as in many such cases, the DREAM complex also has key roles in cell cycle progression. Though this could prove problematic in dividing cells, such an inhibition in permanently non-dividing cells (such as neurons) could have far fewer drawbacks. There is an exciting possibility that other repressors such as DREAM remain undiscovered, possibly including those with a lesser degree of involvement in other critical cell processes. 

Transposons, segments of DNA that replicate themselves and insert themselves in problematic places in the genome, are a known destabilizer of genetic architecture. Given that they consist of approximately 20% of the human genome, cells are forced to expend significant energy to repress them. Nucleoside reverse transcriptase inhibitors (NRTIs), drugs first deployed against HIV, have been shown to decrease age-associated inflammation by specifically inhibiting a type of transposon. This has, unsurprisingly, caught the attention of commercial actors. Transposon Therapeutics recently released positive Phase II data in a study focusing on progressive supranuclear palsy. Very recently, Altos Labs (a secretive longevity company funded by Jeff Bezos) has also been reported to be working on an oligonucleotide approach to treat accelerated aging disorders by inhibiting transposons. These are exciting developments, though it is worth noting the connection between transposon activation and human aging is suboptimal.

Tier 3: Activate Genome Stabilizers

The most intuitive way to improve genome stability is via upregulating or activating the proteins playing a role in maintaining it. Unfortunately, this is not so simple. The genome is closely regulated, and there are contexts in which optimal maintenance is not ideal.  Persistent activation of repair responses also sends a signal to the cell that something is terribly wrong - and the cell can react by activating persistent inflammatory signaling or apoptosing. Even if a safe target were found, a small molecule activator for it may not exist. The search for such a molecule would be especially difficult as those that could be utilized in the context of genome maintenance will need to be both safe and specific.

This is not to say that a small molecule activator improving genome maintenance would be impossible to find. Fairly recently, an activator of an enzyme involved in repairing oxidative damage was found to function via an unconventional mechanism. This is particularly exciting, as the enzyme is responsible for directly repairing DNA damage as opposed to being a master regulator of a complicated network. A number of compounds, including resveratrol, have also controversially been named as sirtuin activators capable of extending lifespan. Some evidence, too, suggests that NAD+ precursors play a role in boosting DNA repair capabilities, though the mechanism is not clear. Also, farrerol could be an activator of oxidative stress resistance pathways. Overall, our current arsenal of genome stabilizers remains extremely small. This is not only due to the complexity of the underlying biochemistry, but is also the result of a lack of appreciation for the potential of such molecules.

Tier 4: Beyond Small Molecules

Small molecules are a fairly indirect method for improving genome stability or preventing damage. What if we could simply directly increase the expression of maintenance mechanisms in humans? Alternatively, what if we could pull mechanisms from other animals and incorporate them into ourselves? Numerous companies are taking on these approaches already.

Genflow Biosciences is rapidly seeking to transfer the centenarian SIRT6 variant into the rest of us, by first seeking to treat Werner’s Syndrome and NASH. Matter Bio is finding ambitious ways to reverse the mutations that have already occurred, in some way breaking the conventional wisdom that improving genome stability will only slow the rate of aging. Peel Therapeutics is seeking to deliver an elephant copy of a key “guardian of the genome” gene. 

Many of these approaches come with caveats:

• Most applications for improving genome instability are relevant in the context of preventing a disease rather than curing one. In the grand scheme, the hope is to deploy them across all healthy adults. This is not quite compatible with gene delivery approaches, which do not currently have a safety profile that makes them compatible with use in healthy adults. As a result, the use of these techniques (for now) may only be in patients with a devastating disease. 

• The durability and targeting of gene delivery approaches is still a work in progress, limiting administration to certain tissues or organs. Some areas of the body (such as the brain) where improved genome stability would be the most useful are also the most difficult to deliver genes to.

• The underlying molecular mechanisms and pathways of many genes involved in genomic maintenance are unclear, and therefore the targeted disease indications may be challenging to identify.

None of these technical challenges are insurmountable, and incredibly ambitious approaches will need to be piloted in order for improvements in human maximum lifespan to occur.

Each tier comes with its trade-offs. Preventing DNA damage seems like the most intuitive and easiest approach, and using gene therapies to up-regulate repair is the most challenging. Yet from an evolutionary point of view, it is likely that our bodies have already evolved advanced mechanisms to deal with endogenous sources of damage, and it is only through the more radical approaches that we can push our defenses past their limits. The best-of-both-worlds approach may be to do something specific and context-dependent. For example, many Tier 3 approaches (those increasing the expression of genome stabilizing genes) lead to permanent cycle arrest, but this could be used in a cell type that is already post-mitotic. Alternatively, a Tier 1 approach may prove to be beneficial in the context of ameliorating rarer types of damage that the cell has not evolved adequate defenses against.

Regardless of the intervention, it will need to be paired with a disease to correct (at least in the short-to-medium term) for FDA clinical trials. I have listed a few promising areas below.

Places to Intervene

Prevent Cancer

As cancer is the second leading cause of death in the United States and is linked very tightly to failure of DNA repair, preventing it is a promising avenue through which to tackle improvements in genome stability more generally. Cancer prevention has been done before - tamoxifen can help prevent breast cancer and aspirin could help lower the risk of colorectal cancer. Nicotinamide, a molecule familiar to those in the aging space, has been shown to decrease skin cancer incidence in individuals who had previously been diagnosed with melanoma. The idea of using a drug to prevent cancer is extremely appealing, particularly as the potential patient population immediately includes everyone. There are a few caveats and challenges, however:

1. Not every genome maintenance pathway relevant to cancer is necessarily tied to aging.

2. Clinical trials for cancer prevention tend to be difficult and expensive. Even in the elderly, the annual incidence of cancer is quite rare, meaning trials that seek to prevent cancer in the healthy adult population will need to enroll many thousands of participants.

3. If the drug is going to be prescribed to everyone, it has to be extremely safe. This is generally true for any longevity drug but becomes acutely important in the context of long-term administration to prevent a possible future case of cancer rather than treating an acute disease.

Nevertheless, the possibility of being able to tackle both aging and cancer in one fell swoop is incredibly enticing.

Delay Menopause

The age of menopause is closely tied to women’s overall health (more on this in a wonderful post authored by Carol Magalhaes here!). There is intriguing evidence suggesting that improving DNA repair could be a direct method by which to delay the onset of menopause by prolonging the longevity of oocytes. In Genome-Wide Association studies (GWAS) investigating genetic associations with the age of onset of menopause, DNA repair genes frequently emerge as top hits. Variants in DNA regions near EXO1, one of the genes that emerged as a top hit, have also been shown to be linked to cognitive aging in a Taiwanese population and associated with prolonged life expectancy in centenarians. 

The mechanism here is unclear - it is possible that DNA repair keeps oocytes alive for longer and thereby directly delays menopause by maintaining the ovarian reserve, or it could have other functions in maintaining other ovarian cells or those involved in neuroendocrine signaling. Regardless, this is a promising avenue through which to significantly improve women’s health.

Slow Neurocognitive Disorders

Though oxidative damage to DNA occurs widely throughout the body, the brain appears to be uniquely sensitive. The reasons for this are likely multifaceted but may be linked to the inability of neurons to divide and thus dilute out damage, or caused by the high metabolic load in the brain. Deficiencies in oxidative damage repair, detection, or prevention are linked to ALS, Ataxia-Telangiectasia,  and Cockayne syndrome. Each of these three diseases has a pronounced neurodegenerative component, suggesting a central role of oxidative damage in destabilizing genome instability in neurons. Less causally, oxidative stress has also been implicated in the pathogenesis of Alzheimer’s disease (even at very early stages!) and Parkinson’s disease. Depletion of BRCA1, a protein responsible for DNA repair, has been shown to occur in Alzheimer’s disease brains with possible downstream effects on memory and cognitive function. Similarly, XRCC1 variants have been coupled with an increased risk of Alzheimer’s in a Polish cohort. 

Diseases aside, cognitive capabilities are one of the first functions to decrease in humans with age. In the brain, DNA damage wrecks havoc long before we start to acutely notice its effects. If we can find ways to improve our resilience, ideally as early as possible, we can both slow the incidence of disease and maintain our ability to think.

Stabilize the Brain Post-Trauma

Given that the brain is sensitive to oxidative damage, it will not surprise you to learn that events such as a stroke or seizure cause enormous amounts of DNA damage. Curiously, there’s evidence to suggest that failure to resolve DNA damage could even be causal for epilepsy, as over-activation of DNA damage sensors can lead to seizures in mice. Rare mutations in a single-strand DNA repair gene in humans manifest primarily as seizures and neurocognitive impairment. Debate exists around whether the true cause of neuronal cell death in the case of strokes and seizures is the DNA damage itself or a hyper-activated response to it. If you knock out a key DNA damage detector (PARP1) in mouse brains, they become much more resistant to neuronal damage induced by a stroke. The mechanism for this effect is also unclear, but could either be linked to direct cell death mechanisms or through DNA damage sensors burning through key energy metabolites that could be more effectively spent on repairing other forms of damage. In any case, if repair could be made more efficient then the amount of energy consumed for DNA damage signaling could be reduced, leading to neuronal protection.

Hamper Fibrotic Disorders

In the public eye, telomeres have been long connected to aging. If you work in the longevity field and mention that you are working on aging, a very common response is “oh, so like telomeres and stuff?” Unfortunately, the link between telomeres and aging in humans has proven to be fairly elusive. The connection is strong, however, in some diseases including pulmonary fibrosis. The gene encoding TERC, a component of telomerase (protein that extends telomeres), leads to inheritable pulmonary fibrosis when mutated. In mice, over-expressing telomerase leads to lung regeneration in models of fibrosis. 

The connection between genome instability and fibrotic diseases extends beyond telomeres. As an example, over-expression of a gene that promotes the formation of a key DNA repair complex leads to disease improvement. In kidneys, over-expression of a protein responsible for the repair of oxidative DNA damage in the context of fibrosis led to an improvement in outcomes and, surprisingly, a positive effect on inflammation. Critically, many of these treatments were administered after the damage already occurred, suggesting that improving DNA repair could be used not solely in a preventative context.

Venturing into the Unknown

Saving the genome will not be an easy task. The number of drugs we currently have on hand to address it is extremely limited. Doing it wrong means driving up a pathological response that brings with it inflammation and cell death. Nevertheless, the promise is too great to ignore. There is definitive evidence for the central role genome instability plays in our aging process. By tackling it, we have the potential to address healthspan and lifespan in one swoop. Instead of hopelessly seeking to push the boulder of aging back up the hill once it has already reached a critical velocity in some tragic emulation of Sisyphus, we can instead directly affect the exponential process hurling us towards death. Slowing down the rate of aging by maintaining the genome even 10% better could bring each of us decades of life. It is up to us to take up this mantle and protect our genomes to live longer, healthier lives.

Acknowledgments

This piece would not have been possible without months of conversations, editing, revising, and thinking. Discussions with the following experts informed several critical components of this piece:

• Dr. Jan Vijg

• Dr. Laura Niedernhofer

• Dr. Alex Cagan

• Dr. Jan Hoeijmakers

• Dr. Richard Frock

• Dr. Ashby Morrison

• Dr. Matt Yousefzadeh

• Dr. Karim Mekhail

The age1 team (Alex Colville, Kat Kajderowicz, Carol Magalhaes, Maggie Li, and Alex Kesin) provided helpful edits, insights, and the fire needed to push this piece out into the world.

Conversations with genome stability in aging enthusiasts (Ruxandra Teslo, Zane Koch, Ronnie Cutler, Stacy Li, and many others) provided joy in writing this piece through inspiring creativity, brainstorming, and passion.

Lastly, a thank you to the Buck Institute genome stability fan club (Edward Anderton, Carlos Galicia, Sierra Lore, and Brendan Hughes). They are instrumental in creating a warm intellectual environment that combines a childlike sense of wonder with the zeal necessary to desire to make a difference in the world.

Appendix 1: Challenges

Models

The most common studies for DNA repair tend to expose cells to extremely high doses of a genotoxic agent (e.g. X-rays, UV,  IR, cisplatin, doxorubicin) to induce high levels of DNA damage. The ability of a cell to resist this damage is representative of their DNA repair capacity, and screens that improve a cell’s resistance are noted as candidates for improving genome stability. Unfortunately, it is unclear if this is relevant for aging, given that different genotoxic agents cause unique types of genomic instability. Cells have diverse mechanisms to deal with threats to their encoding information, and it is unclear which types of defenses are relevant to aging. Even the applicability of cell models of accelerated aging disorders in humans is a topic of ongoing discussion. Worse still, it is unclear to what extent the same genomic maintenance pathways that fail in mice, worms, or yeast are translatable to humans. Werner’s syndrome models in mice fail to recapitulate the same accelerated aging phenotype as seen in humans, while Hutchinson-Gilford Progeria Syndrome models appear much more similar. Animal and cell models will be essential for understanding genome instability in humans, but we will need to deploy innovative techniques to ensure translatability.

Studying longevity in humans is challenging. Interventional lifespan studies are nearly impossible, as is nearly any type of aging-related exploratory investigation. Lifespan studies in mice are better, but take years. Worms, flies, and other species are faster - but with distance to humans come hints that findings may not translate. The evidence that core longevity mechanisms are conserved is enticing, highlighted by primordial pathways such as IGF signaling and addressed by interventions such as caloric restriction, but by no means is it clear that this is true for the entire causational aging spectrum. The Intervention Testing Program, a standardized protocol for testing potential longevity interventions in mice, has been a graveyard for many enticing approaches first discovered in lower organisms. Worse still, we will likely never know the potential approaches that would work in humans yet fail in mice. 

The challenge of translating animal data into human clinical trials is only more difficult for work in genome stability. Effect sizes are unlikely to be massive immediately. Optimal bands of effectiveness will be narrow at first until we reach a greater holistic understanding of how the cell maintains its genetic information. To succeed, we will need far better models. Genome stability will need to be simulated in a translatable context, and perturbed in manners that can recapitulate the way a cell naturally responds to the aging process. A significant amount of theoretical work will need to be undertaken to decouple the phenomena of cancer from aging. Creating the framework that could lead to successful interventions in humans is the first step in turbocharging the field.

Readouts

In many ways, the genome stability field is still in its infancy with regards to capacity for scaling. Many labs have their own in-house assays for different DNA repair processes, and few of these have been cross-verified by other groups. The most widely-used assay, the comet assay (so called due to measuring a DNA smear that looks somewhat like the tail of a comet) was developed 40 years ago, is low-throughput, and only measures a small number of types of DNA damage. The gold standard approaches, including those based on mass spectrometry, tend to be very expensive. There are researchers currently focusing on single assays that can measure many different types of DNA repair simultaneously, and the field is increasingly working towards cross-validating different techniques. Nevertheless, there would be an enormous benefit from having dedicated, inexpensive, high-throughput, and trusted ways to measure different types of DNA repair and stability.

Clinical Trials

Fixing genome instability will solve, largely, a problem with the rate of aging. Optimistically, upregulation of relevant enzymes may fix DNA lesions that have accumulated with time, or alleviate pressure on systems that have not been able to keep up with damage occurrence. Nevertheless, as improving genome maintenance is not likely to be immediately rejuvenative, the benefits in the short-term will be less obvious than any possible therapy that reverses aging, even though the likelihood of success in the long-term may be greater. Additionally, effectiveness will be greater the earlier the intervention has begun - it is entirely possible that as a person ages, other downstream consequences take on a dominant role in age-related decline. This poses challenges for clinical trial readouts - even the XPRIZE Healthspan competition requires a gain in function, which is theoretically impossible if the rate of aging is slowed rather than reversed. The benefit of going after the slowing of aging anyway, of course, is that the probability of meaningfully increasing human lifespan by improving genome stability is much higher, even if the results are not seen immediately.

Appendix 2: Opportunities

Decades of Cancer Research

Ironically, prime candidates for improving genome stability may be found in lists of the genes mutated in chemotherapy-resistant cancers. Cancers, in some ways, become gain-of-function screens for genome stability in their own right when subjected to genotoxic agents. Some tumors improve their pathways to cut out DNA lesions to evade the effects of cisplatin. Cancers can “learn” to improve double-strand break repair to survive radiotherapy. 

It will take time and effort to discover which pathways are key to slowing genomic attrition in aging. Peto’s paradox (the observation that larger animals do not necessarily get more cancer) means that animals that grow larger need to solve the problem of getting more cancer, and not all of these solutions slow down the rate of aging. Cancer could, nevertheless, be a promising avenue to identify which genes are most necessary for the successful repair of defined types of DNA stress. These “limiting” genes can then be screened for their impact on lifespan and healthspan.

Screening and Artificial Intelligence

With hundreds of proteins carefully titered by the cell for a given context, the likelihood of discovering the key interventions in genome stability by chance alone is extremely unlikely. Finding a way to scale will be fundamental to accelerating the pace of discovery. But this is much easier said than done - in screening, the devil is always in the details. A survival screen in the presence of a genotoxic agent may be a good place to start (and has been done by several groups), but enrichment may be found for genes involved in genotoxic agent efflux or senescence rather than core DNA maintenance machinery. It is also difficult to know which genotoxic agent best mimics the effect of aging. 

Given good models and screening approaches, artificial intelligence may be an inflection point creator for the field. The recent discovery of a more effective SIRT6 variant in centenarians would almost certainly never have been found by conventional screening, no matter how well-resourced. Evolution was ultimately the generator of this improvement, with the improved variant ironically a product of failed DNA repair. Artificial intelligence, even at this early stage, is able to generate improved versions of currently existing human proteins. These approaches, especially if combined with a built-in computational understanding of the aging process in humans, could rapidly accelerate our progress in this domain.

Appendix 3: Labs Currently Working on Genome Instability in the Context of Aging

This is a table of a small subset of current labs that are tackling genome instability in the context of aging. This list will be updated & curated frequently

Appendix 4: Companies Addressing Genome Instability

This is a table of current companies that are tackling genome instability in the context of aging. It will be updated frequently

Appendix 5: Interventions Improving Genome Stability

This is a table of current small molecules that have evidence of improving genome stability and improving healthspan and/or lifespan.

The current decade is the critical decade. 

Over the last ten years, the field has cleared some critical obstacles for a regulatory pathway. In 2015, a group of leading academics in the field met with the FDA about the TAME (Targeting Aging with Metformin) Trial and indicated the FDA agreed with the plan. In 2023, Loyal received a historic FDA milestone of reasonable expectation of effectiveness for lifespan extension in large dogs. And recently, BioAge announced a phase 2 clinical trial with Eli Lilly on a drug combination (azelaprag & tirzepatide) for musculoskeletal protection alongside GLP-1s. 

Over the next ten years, we will see more therapeutics, tools, and technologies in the field move from lab to clinic than ever before. The longevity funnel framework (LFF) is how we think about translation:

To rigorously show that healthy lifespan extension in humans is possible, we still have significant strides to make in each “layer” of the funnel:

Mechanisms of aging

1. Development of a casual understanding of the mechanisms of aging, in relation to functional outcomes.

Interventions for aging

1. Design more interventions that target aging with creative, ambitious, and scientifically rigorous approaches (including small molecule approaches and beyond).

2. Improving the safety and efficacy of existing modalities and developing new ones suitable for long-term interventions that potentially require repeat dosage.

Measures and models of aging

1. Development of models of aging, especially naturally aged model organisms and non-model organisms.

2. Validation of methods to measure and quantify aging at molecular, cellular, tissue, and organism levels.

3. Consensus around aging endpoints for clinical studies and subsequently useful biomarkers.

4. Clear strategies and FDA regulatory pathways for cost-effective aging clinical trials in humans.

Mechanisms of aging

Connecting our understanding of aging at the molecular and cellular levels with functional outcomes.

As we age, disease incidence increases dramatically (healthcare costs also skyrocket)! 

A cumulative loss of function and accumulation of damage leads to aging, which corresponds to a drop in survival probability on the Kaplan-Meier curve. Healthspan and lifespan extension ideally mitigate this process. We do not want one without the other. Hence, the goal for more years in good health = healthy lifespan extension. 

Aging is much more than a change in x over time. It is multimodal, multifactorial, and highly variable across space and time. The rate of aging varies across organs and throughout a lifetime. There are also sex-related differences, such as in immune system aging. 

Approaching the question of “What drives aging?” using the hallmarks of aging is reductionist (credit is still due as it provides an overview of current areas of research!). The hallmarks do not capture how aging is a complex network of changes across space and time. As a result, we still have many gaps in understanding of how pathways implicated in aging drive outcomes at the organism level. 

The age-old question of genotype to phenotype remains open. Although difficult, we have not yet established a quantitative and causal understanding of the mechanisms of aging. Targeting the drivers of aging is the ultimate goal of designing a therapeutic, so understanding the causality of mechanisms is critical. Regardless of what “theory of aging” these mechanisms are derived from (ex: metabolic regulation of aging, information theory of aging), the physiological outcomes are what matter. These mechanisms should be connected to function, feeling, and survival for clinical relevance.

If you think about aging as some process that maps causal mechanisms to phenotypes, a therapeutic targeting of these mechanisms is unlikely one-to-one or one-to-all in its effect. There is no one, all-encompassing “magical” pathway that modulates all aging phenotypes. Modulating sets of mechanisms correspond to a target-based approach whereas modulating sets of aging phenotypes correspond to a phenotype-based approach. To avoid exploring an infinite solution space, we need to account for both to varying degrees.

Relationships between mechanisms form complex networks. The same network structure applies to aging phenotypes at cellular, tissue, or organism levels. These networks then interact via biological pathways to produce systems-level, functional outcomes (ex: some of these interactions as a map). With aging, the interactions within and between these networks also change over time. 

Besides reducing the solution space, this is relevant because approaching from a systems biology perspective might lead to therapeutic approaches not considered before. Modulating one system could be the “fix” for another system’s age-related disease (ex: regulating cardiovascular systems’ effects on neurodegenerative diseases via targeting neurovascular coupling). 

Then, from a complex systems perspective, we can consider how these systems self-regulate for homeostasis. Aging leads to a breakdown of feedback loops over time, as systems lose their ability to maintain balance. This is even more challenging to model but could be a way forward to discern drivers of aging versus factors that are simply involved via related pathways/etc. 

For more on the biology of aging, here are some resources: 

• Historical perspective: a review of progress to date from Campisi et. al 

• Hallmarks perspective: the 2023 hallmarks of aging from López-Otín et. al (2013 version)

• Overview: the original longevity FAQ by Laura Deming, an extended longevity FAQ by José Luis Ricón

• Textbooks: Handbook of the Biology of Aging (covers some systems-level aging) and Systems Medicine: Physiological Circuits and the Dynamics of Disease, by Uri Alon (for systems-level modeling with some math)

Overall, capturing the networked nature of changes in aging will help improve the translatability of what we study. For computational approaches, this nuance will guide more meaningful inference-based learning. For tool development, studying aging across the biological scale will help answer more causality-related questions. Finally, for therapeutic interventions, a casual understanding of aging means more efficacious therapeutics that target drivers of aging. We’ll have better indication and target selection, a clear set of phenotypes to screen for, and an improved ability to customize the modality based on the mechanism of action. However, an incomplete picture does not impede our ability to move forward: we can design interventions with what we rigorously understand so far. 

Interventions for aging

Broader consideration of possible interventions for aging.

Small molecule drugs alone, single or combinatorial, are not medicine’s endgame. Exploring more alternative approaches is essential for addressing the steep decline in pharma R&D efficiency. The future entails vastly more ambitious scientific moonshots (but equally as rigorous as ambitious).

As a 1st gen approach, small molecules will likely be the first way to target system-level dysfunction and delay aging in humans. Since initial proof of lifespan extension in mice and rats via caloric restriction, multiple pioneering studies have found ways to alter organismal aging using genetic interventions (ex: via modifying age-1 and daf-2 genes in C.elegans) and small molecules (ex: rapamycin and metformin in mice via NIH’s Interventions Testing Program). But if we are to improve the clinical efficacy in which we can target aging, we’ll need to pursue 2nd gen approaches that replace, restore, or pause aging. 

2nd gen approaches might sound like sci-fi, but most have deep historical roots and are widely accepted in different contexts. Tissue replacement originated from bone autografts and skin allografts date back to 1668. For whole organ replacement, the first successful kidney transplant was performed in 1954 by Dr. Joseph E. Murray, who later received a shared Nobel Prize for Medicine in 1990. Similarly, regenerative medicine was built on the discovery of transplantable blood stem cells in 1961 by James Till and Ernest McCulloch. Reprogramming has its roots in John Gurden’s work from 1962, where he discovered that the specialization of cells is reversible. Cryobiology dates back even further, to 1665 with Robert Boyle’s studies on the effects of freezing temperatures on living animals. And today, freezing oocytes and sperm is standard practice in reproductive health.

From this historical perspective, the novelty in 2nd gen approaches will mostly come from developing technical maturity in these science areas and demonstrating clinical utility for aging or longevity. These interventions are (and should be) grounded in decades, if not centuries, of foundational scientific work. 

Categories of interventions for aging 

1. Delay (delay or prevent aging): most existing approaches fall within this category, which are hallmarks of aging-esque.  

2. Replace (replace aged, damaged cells and tissues with youthful ones): possible approaches include cell therapy, mechanical parts, and tissue/organ transplantation (supplied by xenotransplantation, growing human organs ex vivo, 3D bioprinting, etc). Replacing and restoring aging are not mutually exclusive (ex: the replacement theory of aging as an approach to enhance neural plasticity and unlock neurogenesis).

3. Restore (turn the clock back to restore the age of cells and tissues to a youthful stage): possible approaches include full or partial cellular reprogramming using transcription factors, though more studies are required to validate that “reversal” of age has occurred through such mechanisms without inducing dedifferentiation to harmful levels and extensive, subsequent teratoma formation. Broadly, most approaches within regenerative medicine could be categorized as possible ways to restore aging. 

4. Pause (pause the aging clock in human cells and tissues by inducing stasis): possible approaches include torpor, hibernation, and cryopreservation at the cellular, tissue/organ, and organism levels. This is especially key for enabling longer-term storage of organs before transplantation without significant tissue damage. 

For more detailed categorization (non-exhaustive), here are some resources: 

• Longevity biotech landscape by Ada Nguyen 

• Synthesis of categories from the hallmarks of aging, 7 pillars of aging, and SENS damage types by Karl Pfleger 

• An up-to-date overview of bottlenecks and underlooked areas by Amaranth Foundation (not a categorization)

Comparing 1st gen vs 2nd gen approaches

In our view, there are three ways in which 1st and 2nd gen approaches differ (for more considerations at this stage of development, see Celine Halioua’s How to Build a Biotech): 

1. Indications: the FDA has not (yet!) given official guidance on a possible regulatory path for an indication representing aging or a proxy of aging. In the interim, those for age-related diseases are most relevant. 1st gen approaches can take on greater regulatory risk to forge these new indications, as they require less innovation on modality. Loyal took the first step and has shown it is possible to obtain an indication for lifespan extension in large dogs. The logical next step is to obtain a similar indication for humans.  

2. Modality: we are unlikely to design successful interventions that can replace, restore, or pause aging with well-established modalities like small molecules. To achieve functional improvements with a drastic effect on aging, these 2nd gen approaches will require innovation on emerging modalities with less technical maturity. Examples include cell and gene therapies, which need further development in safety and efficacy. 

3. Possible effect size: 2nd gen approaches are more likely to achieve a greater effect size, which likely translates to increased healthy lifespan extension (effect size is the magnitude of difference between two groups, independent of the sample size). With a greater effect size, we also do not need such large samples to achieve statistical significance, which could help reduce clinical trial costs. When a drastic effect is observed, it will be obvious the therapeutic works without extensive statistical modeling (ex: Neurona’s regenerative cell therapy for epilepsy treatment where a significant effect was observed in the first 2 patients). 

Regardless of differences, the parallel development of 1st gen and 2nd gen is crucial. Success in 1st gen acts as a proof point of real progress in the field, critical for attracting big pharma, institutional investors, and top talent. 1st gen approaches also pave the way for appropriate indications, so there is a path forward for pushing 2nd gen approaches to indications more closely relevant to aging. 

Next, to test modalities and quantify effect size on the indication(s) of interest, we need good models and measures of aging. 

Models & measures of aging 

Creating robust models and measures of aging that are clinically relevant.

Many current models and measures of aging do not meet the bar for clinical translation. If a model and measure have low predictive validity, we are more likely to get a false positive or negative (bad!). We want good models and good measures of aging to increase the likelihood of a true positive. 

• False positive (type I errors): We may observe a desired effect size, but the intervention has limited or no clinical utility → Translation fails in phase I to III clinical trials (and $$ goes down the drain).

• False negative (type II errors): We may not observe a desired effect size, but the intervention has some clinical utility → The desired effect size can be observed with alternative models/measures.

Disease models that don’t align with human disease outcomes and measurements via biomarkers with limited validation result in misguided candidate selection. This problem is universal to bio R&D, but aging adds the element of time to models and measurements, which complicates the design of experimental controls and appropriate statistical modeling (no p-hacking!). 

NIH’s Interventions Testing Program (ITP) serves as a gold standard. ITP tests single drug interventions using genetically heterogenous mice (UM-HET3, four-strain cross), with replicates across three sites and statistical power of 80% for a 10% change in median lifespan extension. Despite the rigorous design, the ITP is limited in throughput and combinatorial testing.

Models of aging

Preclinical models span theoretical models (in silico), engineered models (in vitro/ex vivo), and model/non-model organisms (in vivo). A table of notable model and non-model organisms is in the appendix.

In silico and in vitro models can help drive faster iteration. One advantage is the ability to use human data and samples, which addresses a key limitation of in vivo models. However, the predictions are only as good as the data and the modeling as good as our understanding of biology. So, either we improve our models (which will take more time) or compensate with large volumes of data and massively parallel, high-throughput screening. To do so, we need more biobanks for human cell/tissue samples paired with de-identified and longitudinal genetic, lifestyle, and health data like the UK Biobank. In the long run, the development of microphysiological systems (ex: organ-on-a-chip systems) with physiological relevance for aging is also necessary, given the FDA Modernization Act 2.0.

In contrast, in vivo models can better capture the biology as we scale up to more complex organisms. However, no model organism compares to a full human, so we similarly need more high-throughput approaches to compensate. This includes low-cost production of genetically diverse and naturally-aged organisms, with automation where possible, biobanks for model and non-model organisms like the mouse gut microbial biobank, and more studies like the Zoonomia Project but focused on aging biology. 

Above all, more models do not equate to better modeling. A single model with good predictive validity will always beat out an indefinite combination of bad models.

Measures of aging 

There are endless possibilities of what to measure at the molecular, cellular, tissue/organ, or organism levels (ex: possible measurements for age-related diseases and emerging areas for molecular/cellular damage accumulation assays).

Beyond what metrics to measure, how the measure is correlated with aging is also critical. We want to capture state changes over time, such as shifts in survival curves, changes in composite scores, or fluctuations detectable by assays or aging clocks. For FDA interest, measures should capture function, feeling, and survival. These phenotypic outcomes are crucial for the design of clinical trial endpoints. A table of examples of molecular and non-molecular measures of aging is in the appendix.

Endpoints serve as benchmarks to evaluate the relevance and reliability of biomarkers and proof of concept measures. A measure that accurately reflects the biology of aging needs to show a rigorous correlation with endpoints such as survival rates, disease progression, and symptom improvement.

Appendix

Table of model organisms

Table of non-model organisms (examples)

Table of measures of aging (examples)

Acknowledgements

All resources related to longevity/aging referenced in the piece or during the writing process are compiled here. The longevity funnel framework was largely inspired by the Amaranth Foundation’s “Bottlenecks of aging” and Celine Halioua’s “What the aging field needs”.

Special thank you to everyone on the age1 team: Alex Colville for extensive brainstorming support, Alan Tomusiak for inspiring V1 of the LFF, and Kat Kajderowicz, Alex Kesin, Carol Magalhaes, and Lily Clayton for helpful feedback.

My mom started perimenopause in 2023. As she experienced headaches, night sweats, hot flashes, and brain fog, she would tell me she didn't feel like herself. As she started hormone replacement therapy (HRT) – the only treatment for menopause – I took a deep dive into this world. I was already in the field of longevity, but menopause wasn't something I often saw discussed, which puzzled me.

The life expectancy for women in the U.S. is around 80 years, while the average menopause age is 51. As medicine continues to develop and life expectancy increases, the number of years post-menopause will continue to increase. Despite this fact, reproductive aging is significantly under-researched. Only 10.8% of the NIH's 2020 budget went to women's health research, and reproductive biology and aging are largely siloed fields. However, menopause could be an important lever we can pull to extend years of healthy life in women, and I hope more longevity researchers will start considering the field of reproductive aging.

This post will consider how longevity and menopause might be connected, their implications, and possible approaches to delaying menopause.

Table of contents:

• Women have a finite pool of oocytes

• Menopause is about fertility but also quality of life

• Two non-mutually exclusive theories could explain the connection between menopause and aging

• Theory 1: Healthy aging can influence the age of menopause

• Theory 1 implication: menopause could be an indication in clinical trials for "repair" aging drugs

• Theory 2: Menopause directly impairs health

• Theory 2 implication: the body is a network, and estrogen is a major node to impact overall health

• How to delay menopause

• Conclusion

• Appendix: Interventions that extend fertility and protect the ovarian reserve

Women have a finite pool of oocytes

While men continue to produce sperm throughout their lives, women are born with all the eggs they will ever have. These eggs are part of the primordial follicle pool and remain immature oocytes until puberty. It’s pretty mindblowing that the egg cell that made you was produced by your mother when she was a developing fetus in your grandmother!

A female fetus typically has around 6–7 million oocytes at 20 weeks of gestation, dropping to 1–2 million at birth. When puberty begins, she only has about 400,000 oocytes, meaning she’s already lost over 90% of the oocytes she started with. During much of her adult lifetime, about 1,000 immature oocytes will be activated every month to start maturation; and only one mature oocyte will be ovulated. This egg, if fertilized by a sperm, will become a baby 9 months later. The cells surrounding oocytes – granulosa and theca cells – form follicles, which produce important hormones for health, specifically progesterone and estrogen. With a decline in the number (as well as quality!) of oocytes, these hormones also decline with age. 

Although it is not clear how many follicles a woman has at menopause, that number is expected to be around 1,000. The oocytes left are of poor quality and can’t grow to become mature eggs. At this point, a woman becomes unable to have children, and it is said that the ovarian reserve is depleted. 

Menopause is about fertility but also quality of life

Menopause is, in part, about fertility and the autonomy to decide when to start a family. If you’re a woman and you'd like to be a mother, you are also most likely familiar with the biological clock ticking in the background of your decisions. Initially faint in one's 20s, this tick grows louder each year, reminding you that your fertility is declining with age. From dating decisions to major career options, this ticking is subtle but present. 

The chances of having a natural pregnancy decline in a woman's early 30s, when she still has plenty of immature eggs left, but the general trend in developed nations is that women are having children at a later age.

However, menopause is also about quality of life. Along with menopause comes a wave of unfavorable symptoms – hot flashes, insomnia, and mood swings. Not surprisingly, menopausal women have lower quality-of-life scores compared to pre-menopausal women. 

Other health effects pertain to underlying health and aging. With the depletion of the ovarian reserve, the ovaries no longer produce much estrogen and progesterone – hormones that are key to various biological functions. This begs the question - how are menopause and aging connected? 

Two non-mutually exclusive theories could explain the connection between menopause and aging

Later menopause is associated with longer survival, and centenarian women are four times more likely to have children in their 40s compared to women who survived only to 73. However, it is unclear if menopause causes more biological aging or if the age of menopause correlates with the underlying rate of aging. There are two reasonable hypotheses that are not necessarily mutually exclusive:

1. Healthy aging→ Delayed menopause

   1. An underlying biology of aging dictates the age of menopause. This suggests that if one has accelerated aging, this would be reflected in early menopause. If slow aging, then later menopause. 

1. Delaying menopause → Healthy aging 

   1. Menopause can accelerate the underlying biology of aging. This suggests that menopause (to some extent) causes aging and that if menopause were artificially delayed, there would be less significant physiological aging.

Theory 1: Healthy aging can influence the onset of menopause

Women who are in worse health by various metrics, such as type 2 diabetes and heart disease, are at risk of earlier menopause. For example, higher total cholesterol levels and blood pressure in pre-menopause years are associated with an earlier age of menopause. This suggests that one's health status before menopause can indeed dictate the age of menopause.

If menopause is a marker of underlying aging, women who undergo menopause earlier are aging at a faster rate than women who experience menopause at a later age. It then becomes possible that some longevity interventions will also delay menopause if they are administered early enough. So, even if a drug is not developed with the intent to delay menopause, it could. For example, calorie restriction and rapamycin consistently extend lifespan and prolong reproductive span (the period that female mice can have offspring). 

Theory 1 implication: menopause could be an indication in clinical trials for "repair" aging drugs

If some aging drugs have the potential to delay menopause, one exciting implication is that menopause could be an indication for a clinical trial. This, however, heavily depends on the type of aging drug.

The three paradigms of aging drugs are delay, repair, and replace. Specifically, drugs that delay aging may be the primary candidates also to delay menopause. These have the potential to delay the decline in overall health markers, which could, in turn, delay aging in the ovaries. If administered before the onset of menopause, they may delay menopause entirely. In contrast, interventions that repair or replace damaged tissue would likely be administered after the onset of menopause, when damage has accumulated. Unless the ovaries are directly targeted, repairing or replacing damaged tissues in the body would not affect an aged ovary with few oocytes and low levels of hormone production.

If menopause is a marker for biological aging, there’s a potential it could be used as an endpoint for aging drugs, which allows for the design of a relatively straightforward clinical trial. For instance, Columbia University is currently sponsoring a clinical trial studying the effects of Rapamycin on perimenopause.

Overall, it seems that the age of menopause is strongly connected to underlying biological aging, and there’s even the potential that a drug for aging might delay menopause. But instead of just being a marker of aging, could changing the age of menopause be a lever to extend healthy years of life that women enjoy?

Theory 2: Menopause directly impairs health

Menopause is connected to a variety of age-related diseases, including muscle loss, osteoporosis, arthritis, cardiovascular disease, obesity, diabetes, and neurodegenerative diseases. Specifically, the deficiency in estrogen post-menopause seems to contribute to this decline in health. 

Importantly, rodents don't have the same hormonal profile as menopausal women. Even as they age, some will still have sustained estradiol levels, which doesn't recapitulate low human levels. A standard model system to study menopause includes mice that have been ovariectomized – the ovaries are surgically removed, mimicking the human inactive ovaries after menopause. The effect of ovariectomy, as well as the role of estrogen in various age-related diseases, are discussed next:

Cardiovascular disease 

At a glance: estrogen supports cardiovascular health.

Numerous studies have suggested that the risk of cardiovascular diseases rises after menopause. Also, compared to men, women have lower blood pressure until they reach their 60s. Estrogen might be a significant culprit in this decline in cardiovascular health.

For example, some studies have shown that estrogen can prevent cardiac fibrosis by blocking the fibroblast to myofibroblast transition. In animal models of cardiovascular disease, young females have lower rates of vascular injury, a slower progression to heart failure, and lower mortality than males. However, these differences can be reduced or eliminated by estrogen deficiency or removal of the ovaries. This suggests that the higher estrogen levels produced by the ovaries in females than in males might be a protective factor. 

At a molecular level, estrogen significantly inhibits the expression of adhesion molecules in endothelial cells, which play a significant role in the development of atherosclerosis. It also substantially influences vascular injury response through increased nitric oxide production.  

In humans, studies have suggested that the risk of cardiovascular mortality is higher for women with early menopause than for those with late menopause. This aligns with the hypothesis that the longer women are exposed to estrogen produced by the ovaries, the better. Hormone replacement therapy also decreases the risk of cardiovascular disease when started within 10 years of menopause onset (although such results are controversial). 

These studies suggest that the reduction in estrogen removes protective female mechanisms that support cardiovascular health. 

Obesity 

At a glance: estrogen is protective against obesity.

More men are overweight than women. Similarly, male mice have a higher susceptibility to obesity than female mice, suggesting that this gap might be partly due to biology and not just environmental factors. Removing the ovaries in female mice eliminates the protection against obesity, and estrogen supplementation restores this protection. Ovariectomized mice on a high-fat diet are also protected against obesity when administered estrogen. This indicates that estrogen from the ovaries is a major player in this protection against obesity. 

In humans, a study following healthy women for 4 years determined that all women gained subcutaneous abdominal fat over time; however, only postmenopausal women had a significant increase in visceral fat, known as "toxic fat." Also, obesity and metabolic syndrome happen more often postmenopause than before. Estrogen deficiency is likely a major causal factor that increases the risk of obesity with age in the case of women.

Diabetes

At a glance: estrogen regulates glucose.

A similar pattern to that observed in cardiovascular disease is also seen with diabetes. There are more diabetic males before puberty but more diabetic females after the age of menopause. Research has suggested that estrogen regulates glucose homeostasis, which explains why pre-menopausal women have a low incidence of type 2 diabetes compared to age-matched men. In animal studies, ovariectomized mice have higher fasting blood glucose, serum insulin, triglycerides, and LDL cholesterol than mice with functioning ovaries. 

This indicates that estrogen deficiency post-menopause is likely a causal factor leading to diabetes in old age, which, in turn, leads to a decline in general health and increased risk of various age-related diseases.

Neurodegenerative diseases 

At a glance: estrogen has neuroprotective effects.

Undoubtedly, menopause comes with significant changes to the brain. Early age of menopause is associated with an increased risk of dementia, and multiple studies show that postmenopausal women have less grey matter in the brain compared to pre-menopausal women. 

The question becomes how much of the changes seen in the brain are due to estrogen deficiency (and thus menopause itself) or other aspects of aging. Although the numbers vary in the literature, some studies suggest that women have lower stroke incidence until the late 50s and early 60s, when they catch up to men. In animal models, ovariectomized mice do worse in maze learning, and estrogen recovers reference memory. Similarly, ovariectomized rhesus monkeys have impaired cognitive function, which is improved with estrogen replacement. These studies suggest that while estrogen deficiency doesn't directly cause neurodegenerative diseases, it likely has neuroprotective effects that can improve cognition and prevent (to some extent) neurodegeneration.

The neuroprotective effects of estrogen have been widely reported. For example, ovariectomized mice that are treated with estrogen have more new cells in the hypothalamic region compared to mice without the treatment. Ovariectomized mice treated with specifically an estrogen receptor alpha agonist have improved learning. Also, estrogen has been shown to contribute to ischemic neuroprotein in mice and has been strongly implicated in the prevention of Alzheimer's. In humans, women on HRT are at lower risk of Alzheimer's, and estrogen increases gray matter volume in the hippocampus of postmenopausal women.

Although the specific molecular pathways involved in estrogen-mediated neuroprotection are not well elucidated, estrogen likely plays an essential role in maintaining proper cognitive function. In turn, its deficiency might make the brain more vulnerable to cognitive decline and neurodegenerative diseases.

Muscle loss 

At a glance: ovarian hormones support muscle growth.

Loss of muscle strength happens in both males and females. In males, there is a more gradual process, and in females, there is a sharp decline around menopause. However, hormone replacement therapy (HRT) can prevent this decline of strength in women. One study compared monozygotic twins on and off HRT and observed that those who took the therapy had greater relative muscle area and smaller relative fat area.

In animal models, ovariectomized mice experience muscle atrophy in a time-dependent manner. Also, ovariectomized mice with induced muscle atrophy experience limited muscle regrowth unless supplemented with estrogen. 

At a molecular level, estrogen helps with satellite cell activation and proliferation, which are crucial for muscle maintenance and restoration. In humans, some studies have suggested that testosterone and progesterone (which also decline with menopause), but not estrogen, stimulate muscle protein synthesis in women after menopause. 

Although the connection between estrogen, progesterone, and muscle synthesis is hazy, menopause seems to cause a sharp decline in strength, which can be partly explained by the loss of ovarian hormones. This suggests another major biological function strongly connected to the ovaries. 

Osteoporosis and arthritis 

At a glance: estrogen deficiency causes osteoporosis

Estrogen promotes the activity of osteoblasts – cells that make new bone– and estrogen deficiency directly causes osteoporosis in both older men and postmenopausal women. Osteoporosis can dramatically decrease the quality of one's life, and there are about 1.5 million osteoporotic fractures in the U.S. each year. 

Similarly, estrogen deficiency has also been implicated in arthritis through the regulation of cartilage metabolism. It's been reported that ovariectomized rats have increased cartilage turnover and surface erosion. Ovariectomized macaques with one ovary removed have similar symptoms. Various clinical studies have also shown that osteoarthritis is related to estrogen levels. Hormone replacement therapy can preserve and even increase BMD (bone mineral density) at all skeletal sites in postmenopausal women. Additionally, the administration of hormonal oral contraceptives can also restore bone turnover and normal bone density. 

Estrogen deficiency directly causes osteoporosis, making postmenopausal women especially vulnerable to age-related frailty. It is also strongly associated with arthritis, indicating that menopause is a significant factor in decreasing the quality of life of older women. 

Theory 2 implication: the body is a network, and estrogen is a major node to impact overall health

Estrogen signaling influences numerous systems in the body and seems to have protective effects that are lost with menopause. In this framework, estrogen deficiency is a domino piece that, when pushed over, impacts various downstream biological mechanisms that become more vulnerable and contribute to unhealthy aging. 

Ovariectomies in both animal models and humans suggest damaged overall health. Ovariectomized dogs have shortened lifespans, and the same is seen in mice. (It's noteworthy that while ovariectomized mice are a common model, these animals are also deficient for any other signals the ovaries would otherwise produce.) In humans, women who underwent ovariectomy before the age of 46 were more likely to develop depression, coronary artery disease, arthritis, asthma, osteoporosis, and chronic obstructive pulmonary disease, among others. Finally, mice who had their ovaries surgically replaced with younger ones experienced a 40% increase in life expectancy.

The first wave in women's health that led to profound social changes entailed contraceptives, specifically the pill. The second wave in reproductive health will target ovarian aging and delay menopause to improve underlying health.   

How to delay menopause

If menopause is a major node in the aging process that marks a pivotal moment in a cascade of health decline, then delaying it could significantly increase healthy years of life. This could be done by directly targeting oocytes. Alternatively, menopause could also be avoided by correctly mimicking hormonal signals. 

Targeting oocytes to delay menopause

• Inhibiting activation (delay)

The trigger of menopause is the depletion of the ovarian reserve, so what if we could delay the loss of oocytes? One major way to do this is by halting their activation. The female body activates about 1,000 follicles each month to get 1 egg ovulated. If only 500 were activated, this would theoretically double the years women have between puberty and menopause. Various interventions that have been shown to extend fertility in mice work by inhibiting the activation of oocytes - such as rapamycin, AMH (anti-mullerian hormone), and even calorie restriction. 

This approach focuses on the number of oocytes, but the ovarian environment undergoes physiological changes with age, negatively impacting their quality. For example, ovaries become more inflammatory and fibrotic with age. When considering extremes, it's possible that a woman could have many oocytes at an old age due to their growth inhibition. Still, if oocytes are of bad quality, they might not produce hormones properly or be competent for fertilization. 

Pro: most upstream cause of menopause

Con: quality of oocytes might be a limitation 

• Targeting oocyte quality (repair)

Another approach to protect oocytes is to target their quality directly. There are likely various ways to do this, but one potential option is to increase DNA damage repair. Oocytes in the ovaries accumulate mutations with age, and as the ovaries become more inflammatory and fibrotic, the oocytes experience more DNA damage. Oocytes that cannot grow due to an improper environment or poor quality go through cell death and are called atretic. More DNA damage means more oocytes become unviable, leading to quicker depletion. More substantial DNA damage repair could protect oocyte quality and prevent atresia. 

Age at menopause is highly heritable, and its associated genes in humans are strongly implicated in DNA repair. This is consistent with the fact that old populations also have enhanced DNA repair activity. 

Pro: ensures that oocytes are competent for fertilization

Con: even if oocytes are repaired, they will still be lost in monthly cycles

• Providing more oocytes (replace)

If the trigger of menopause is the depletion of high-quality oocytes in the ovaries, these could be theoretically replaced. Although replacing an entire ovary with a younger one is unfeasible, there's a potential that sections of the ovaries could be frozen at a young age and re-implanted later. This is called ovarian tissue cryopreservation and is done in prepubertal patients undergoing chemotherapy. These individuals can't undergo egg freezing and are likely to have their oocytes damaged during the chemotherapy treatment. Instead, they can have strips of their ovaries cryopreserved and then surgically re-implanted as adults. In the context of delaying menopause, this is unpreferable due to two surgeries. It is also unclear how much of an effect small sections of the ovaries would have in delaying menopause. 

Another potential way to replace oocytes is by making new ones ex vivo. An adult cell, such as a blood or skin cell, could be taken and then transformed into an induced pluripotent stem cell (iPSC). We have yet to learn how to make these into oocytes, but various groups are working on it. If this becomes possible, these could be potentially re-implanted into the ovaries. The quality of the ovaries would still be a concern, as they would need to provide a proper environment for the oocytes. 

Pro: unlimited supply of oocytes

Con: unclear if reimplanted oocytes would produce proper hormones in an aged environment

Regardless of how oocytes are protected, the ovaries must also be considered an environment to house the oocytes. How to prevent inflammation and fibrosis will be significant questions that need to be answered for an effective strategy to delay menopause. 

Targeting hormonal signals to treat menopause

Alternatively, instead of focusing on the oocytes that produce the hormones, we could focus on the hormones themselves. While this wouldn't extend fertility, it could reverse the health decline due to menopause. Currently, menopausal women are given HRT (hormone replacement therapy), which usually is composed of estrogen and progesterone. While this undoubtedly helps with menopause symptoms, it doesn't reverse its effects on the body entirely. 

There's little research in this area due to the lack of model organisms that recapitulate human menopause, which makes it challenging to develop these interventions. Two specific reasons for the inefficacy of HRT include the difficulty in dosage optimization and the decline in estrogen receptors with age. Estrogen and progesterone in a woman pre-menopause are tightly regulated in cycles. However, hormone replacement therapy gives a constant dosage (in the case of patches, for example) or a spike dosage (in the case of oral pills). Additionally, the expression of estrogen receptors also declines with age, so even if estrogen is available, receptors might not be functional in cells at their proper levels.

A better understanding of how to mimic endogenous hormones more closely, as well as their signaling through receptors, could provide better treatment for menopause and reverse its effects on underlying health. 

Pro: The number or quality of oocytes doesn't matter

Con: Downstream effect of menopause, and difficult to perform research due to the lack of model systems that recapitulate menopause

Tables can be found here and will be continuously updated. If you know of other relevant labs/companies email me at carol[at]age1.com.

Conclusion 

Menopause is a significant node that can be influenced to improve numerous other systems in the body and ensure healthy aging. Delaying menopause will be one of the first steps to achieving a world where women can live healthily and actively, regardless of an arbitrary chronological age. 

Unfortunately, reproductive aging is significantly under-researched, and within biotech, although 50% of the population experiences menopause, there are only a handful of companies focused on reproductive longevity. As aging biology becomes mainstream, the aging of the reproductive system and its effects on overall health must be carefully considered to enable longer, healthier lives. 

I want to create a future where my grandmother feels healthy and active. Where my mother has full autonomy to do all the things that she loves and isn't limited by physical weakness or mental constraints. A world where I have the choice to have children when it feels right and don't need to sacrifice my career because of a biological clock. Whatever your reasons are, I hope you'll join me in building this future.

Appendix

Tables can be found here and will be continuously updated. If you know of other interventions, email me at carol[at]age1.com.

Interventions that extend fertility in rodents:

Interventions that protect the ovarian reserve when rodents are treated with chemotherapy:

Acknowledgments

I stand on the shoulders of giants and would like to acknowledge some of the names which greatly inspire me.

First, Jennifer Garrison, a Professor at the Buck Institute, is a major leader who brought much-needed attention to this research. Dr. Garrison is the faculty director at the Global Consortium for Female Reproductive Longevity and Equality (GCRLE), which is currently looking for interns! The GCRLE is creating the ecosystem to facilitate and accelerate reproductive aging research. Some of Dr. Garrison's contributions to the field include the Reproductive Aging Conference (2024 registration is open!) and the GCRLE grants.

I would also like to acknowledge Francesca Duncan for my time at her lab, where I learned under the mentorship of Farners Amargant about the ovaries and their aging process. The mentorship I received from Dr. Duncan and Dr. Amargant, as well as the lab creativity and passion, are major reasons that I fell in love with this research.

Finally, a HUGE thanks to Maggie Li for creating all the great graphics used in this piece, as well as the rest of the age1 team (Alan Tomusiak, Lily Clayton, Alex Colville, Kat Kajderowicz, Alex Kesin) for reviewing this post many times and helping me develop my ideas.

2023 marked a meteoric rise in the public's interest in their own longevity. Maybe not so coincidentally, 2023 was marked by the notable rise in the number of direct-to-consumer enterprises that specialize in selling supplements and offering dedicated longevity concierge services. Alongside these prominent developments, the year also witnessed remarkable progress in the longevity biotechnology sector; 2023 brought the longevity biotech field unprecedented regulatory advances, scientific advances, validation from big pharma, and an accumulation of top talent into the field.

In 2023, we backed four stellar founders dedicated to improving the way we age. Our first investment, Aperture Therapeutics, is committed to curing neurodegenerative diseases and delaying brain aging through their drug discovery platform. We're equally excited about the other three ventures, which are presently in stealth mode, and we'll be sharing more about them soon. Moving into 2024, our focus remains on investing in extraordinary founders who are passionate about extending healthy lifespan and aspire to make a lasting impact in this field.

As we ring in the new year with that in mind, the age1 team has put together a New Year’s Company Wish List of ideas we’d love to see future founders work on. 

Our 2024 wish list

Clinical trial innovation, drug repurposing & combinatorial aging targets: We’re seeking founders to innovate on clinical trial design and take on regulatory risk to get a drug approved and labeled for healthspan or age-related diseases as opposed to one specific disease, as Loyal is pioneering in dogs. This could involve:

1. Developing novel drugs for exciting potential geroprotector targets

2. In-licensing (either an asset entirely or indication-specific rights) to repurpose existing drugs that have evidence of a role in longevity, such as in these tables

3. Developing combinations of potential longevity therapies. This could increase potential efficacy beyond what any one small molecule or antibody could accomplish or improve the safety/tolerability profile.

Genome stability interventions: As we age, our genome undergoes many stressors leading to cellular damage. The mechanisms used to repair the genome become compromised as well. Genome instability may be the most important causal factor in human aging, yet the number of translational efforts and longevity companies working on the problem is minuscule. Here are some things that could be done in the shorter term that we’re excited about: 

1. Improving DNA repair capabilities in humans (for example, in HSCs to protect from serial chemotherapy regimens)

2. Improving skincare by targeting core genome stability aging mechanisms

3. Tackling neurodegenerative diseases via ameliorating oxidative damage that destabilizes the neuronal genome

Reproductive longevity & women’s health: Reproductive longevity and women’s health have long been underfunded. Historically, women’s health has been neglected in medical practice, and many longevity therapeutics similarly do not tailor to women’s unique medical needs. While we are generally excited about increasing funding in this space, we are even more excited about making significant strides in improving the low-hanging fruit in this field. We are extremely excited to support approaches that delay menopause and improve fertility. 

1. Approaches to delay menopause or extend the fertility window

2. Technologies to improve IVF

3. Artificial wombs to develop embryo models derived from human pluripotent stem cells (“synthetic embryos”) ex vivo 

Organ chimeras & stasis to replace or pause aging: Innovative engineering with hardware/software that works with biology’s inherent feedback loops can help accelerate clinical translation. We’re excited to apply this thinking to the replacement theory of aging with chimeras. Think: growing human organs in host animals for transplantation or ex vivo cultivation for preclinical human models. This would require making large animal chimeras and further humanizing the host animals, with strong consideration of bioethics when designing approaches. 

1. Vascular engineering for supporting tissue grafts and organ compatibility, to improve the success rate of transplants.

2. Stasis to keep chimera organs ‘banked’ in host animals until needed for retrieval.

Regeneration, glial reprogramming & neurotechnology to keep the brain young: The brain is our control center — it both receives and sends inputs and is arguably the most important organ, crucial to our existence. We are excited by approaches that work towards improving function in the brain and nervous system.

1. Improving cognitive function in the aging brain — neuromodulatory approaches such as deep brain stimulation and glial reprogramming

2. Regenerating severed or degenerating nerves (spinal cord, eye axons, etc)

Tissue engineering to restore aging: with aging and age-related disease, we lose optimal function and control of many body systems. We are excited about approaches for improving functional outcomes to prolong physical independence. 

1. Generating fully rejuvenated and functional HSCs 

2. Fully rejuvenating an aged human thymus

3. Strategies for restoring musculoskeletal function

Developing new therapeutic modalities inspired by nature: Naked mole rats, bowhead whales, and Galapagos tortoises: what makes these critters have such unusual longevity and resistance to disease? Fortunately, we are starting to have a clue - and that means it’s only a matter of time before we start harnessing their superpowers. We might not be able to create PlanariaMan just yet, but we ought to try heading in that direction!

1. Developing transgenic animal models with longevity genes from different long-lived species (e.g. bowhead whales, bats, Galapagos tortoises)

2. Extrapolating the characteristic properties of certain tissues to other tissue/organ types (e.g. the regenerative ability of the liver or cancer resistance of the heart)

3. Harnessing unique aspects of biology from other species (e.g. hemoglobin of lugworms) to apply to treated age-related diseases

   ---

This is far from an exhaustive list of ideas that would propel the longevity field forward. We expect that a large portion of our investments will be in ideas and theses that do not match those we already hold internally. We are firm believers in founder-driven visions.

We encourage you to brainstorm with us — reach out, we would love to chat. We need exceptional founders to work on the hardest problem: achieving healthy lifespan extension within the next decade. 

Build now, age later. 

This size is ~4x the peak projected annual revenue of GLP-1s at $50 billion by 2030, ~6x the peak projected revenue of America’s soon-to-be best-selling drug, Keytruda (PD-1 cancer immunotherapy) at $30 billion annually by 2028, ~10x current bestseller Humira’s peak annual revenue (anti-TNF) at $21.2 billion in 2022, and ~15x the peak annual revenue of enormous blockbuster Lipitor (a statin) at $13 billion in 2013. 

It’s also promising that the clinical trial to get this drug approved for aging and age-related disease on the label could potentially be as short as 3-6 years in length and cost $50-$100 million, within an order of magnitude of most Phase III clinical trials. Furthermore, just in the past month, we learned it was possible to get a drug approved and labeled for healthy lifespan extension in dogs by the FDA.

In this piece, we walk through an estimate of the market potential of an aging drug. We’ve also built an interactive model on Streamlit to allow anyone to play with key assumptions and numbers.

The aging drug TAM model

To build a model for the TAM of an aging drug, we need to answer: 

1. How big is the market? 

2. Who will pay for it?

3. What is the price of the drug? 

4. What does adoption look like?

How large is the addressable market? 

The total addressable market (TAM) for an aging drug refers to the entire potential customer base that could generate revenue for the drug, encompassing all individuals who might benefit from or be interested in using it. We based our foundational TAM size on the total number of adults aged 65 or older in the US from the 2017 US census data (see Appendix D for a table of values). This is a conservative estimate as it only reflects the initial target population. Over time this initial population may be expanded to those above 50 or 40 years old which would dramatically increase the peak market.

We also included non-US regions in this model. Given the US accounts for about half of global pharmaceutical sales, we approximated the inclusion of non-US populations by doubling the US revenue–avoiding the need for further estimation of population sizes and drug accessibility outside of the US. Also, we only included non-US populations for the reimbursement model, as regulations for DTC markets can vary drastically. 

How big is the total addressable market in the US: Numbers range between 73 million people in 2030 to 83 million people in 2045 (see Appendix D).

Who will pay for an aging drug?

To know how much a drug is worth, we need to start with understanding who will pay for it. In the US, the majority of prescription medication costs are paid for by:

• Commercial health insurers

• Medicare Part D

Prescription drugs covered by insurance are far more accessible and frequently used by patients than those without coverage. Given that aging is closely linked with the increased prevalence of various diseases, typically managed with prescription drugs, a drug capable of effectively preventing these conditions could have significant implications. Such a drug would not only improve health outcomes by preventing the deterioration of the patient's health, but it would also reduce expenses for insurers by addressing the root cause rather than just treating symptoms.

Another path for an FDA-approved drug targeting aging could include the new direct-to-consumer (DTC) telemedicine approach in which users pay for services and medications out-of-pocket. By utilizing apps and online platforms, telemedicine companies give users access to convenient, on-demand care from licensed healthcare providers. 

The rise of Hims, Hers, and Ro, three DTC telemedicine brands which started in reproductive health, exemplifies the potential of the approach in healthcare. By leveraging digital health, employing an effective consumer marketing strategy, and destigmatizing traditionally sensitive health topics, Him, Hers, and Ro have shifted the Overton window of how consumers access and perceive healthcare solutions. If a drug designed to slow down the aging process receives FDA approval, it could initially pursue a DTC route, especially before long-term health outcomes and economics research can show financial benefits for insurers and other payers.

To convey the potential scope of future market developments, the increasing interest in obesity drugs serves as an insightful example. This trend hints at the emergence of a pharmaceutical product with revenue potential comparable to the most lucrative products currently in the market. For instance, it's conceivable that drugs like GLP-1 could surpass even the iPhone in sales and profitability. This is particularly relevant when considering the willingness of the average American to invest $500 monthly for weight management. For a more in-depth analysis, the Stifel report, particularly the aging total addressable market (TAM) section on slide 126, offers a comprehensive overview of this market's potential.

Who will pay for it: 1) commercial health insurers and Medicare (reimbursement model) and potentially 2) DTC markets (cash-pay model)

What is the price of an aging drug?

To predict the price of an aging drug, we estimated the monthly list price for a cash-pay model and the monthly net price for a reimbursement model. 

The list price is what the manufacturer sets for a drug before negotiations or discounts, which might reflect DTC market prices. We estimated based on existing medications taken by older adults at scale, such as antidiabetics and cardiovascular drugs (Appendix D). The average out-of-pocket cost of the name-brand version of these drugs is set as the monthly list price, at $618 per month (Appendix C). 

In reality, a consumer pays something closer to the net price after discounts and rebates, determined via negotiations between pharma industry stakeholders. In the US, this comprises patients, insurers, pharmaceutical benefit managers (PBMs), pharmacies, wholesalers, manufacturers, and government entities like the Center for Medicare & Medicaid Services (CMS). Figure 3 provides a summary of how US drug pricing works, with complementary details in Appendix C.

Following negotiations with PBMs, drug manufacturers usually absorb a portion of the drug costs. Currently, the reduction from the monthly list price ranges between 40-60%, with an average of around 50%. With shifting policies around drug rebates, this reduction might escalate to between 70% and 75%. Therefore, we considered an average reduction of 65% from the monthly list price for our calculations.

For the reimbursement model, the monthly net price would be $216, which is 35% of $618. 

Estimate of the price of an aging drug: $216 per month (reimbursement model) or $618 per month (cash-pay model)

What does adoption over time look like? 

For the maximum penetration rate (percentage of the TAM captured), we considered the following patient adherence numbers to approximate a range of 25% to 50%: 

• Half of all American adults get the flu vaccine every year.

• Patient adherence to chronic medications is about 50%.

• Patient adherence to cardiovascular medication (treatment dependent) ranges from 25-50%.

• In one study: of 400,000 people eligible to take statins, about 20% were chronically taking them.

For the annual penetration rate, we applied the Diffusion of Innovation model to approximate a nonlinear increase in market capture as the adoption rate grows over time; detailed values and estimations for annual penetration rates are provided in Appendix D. 

What is the adoption of an aging drug over time: Assuming a 25% penetration rate, the numbers range from 0.006% in 2030 to 25% in 2045. Assuming a 50% penetration rate the numbers range from 0.012% in 2030 to 50% in 2045.

Calculating the annual revenue of an aging drug

The duration of projection is from 2030 to 2045, with 2045 being the end year plus the duration of drug patent exclusivity. This period ranges from 10 to 20 years, so we took 15 years as an average. Table 1 summarizes the key assumptions and variables considered. Now, we can calculate annual revenue as

= total number of adults × annual penetration rate × monthly price/person × 12 months/year

Figures 5 and 6 show annual revenue for the cash pay model and reimbursement models, respectively, at 25% and 50% maximum market penetration rates. 

Between the cash-pay and the reimbursement model, the reimbursement model will likely dominate–reduced prices mean greater accessibility and with proven efficacy, greater accessibility will enable broader adoption. So, we used a reimbursement model with a 50% maximum market penetration rate to compare against other blockbuster drugs in terms of peak or peak-projected revenue (Figure 1).

Conclusion

The approval of a drug targeting the aging process could lead to a seismic shift in healthcare. The potential long-term healthcare savings are vast, as was with statins, which significantly reduced expenditure for inpatient care from hospitalization (see Appendix E).

With the potential to yield conservatively up to $200 billion annually, a company that owned only this drug would be more valuable than the top two big pharma companies (J&J and Pfizer) in terms of revenue… combined. We believe that this drug has the potential to become the largest product in human history, even larger than the iPhone. 

We fully acknowledge that this estimation exercise is simplified. One factor we didn’t account for is the impact of the Inflation Reduction Act (IRA) in the later revenue lifecycle (roughly 9 years into the patent exclusivity period for a small molecule, and 13 years for a biologic). While ongoing Medicare negotiations will likely lower the price of such drugs, this hopefully increases their accessibility to the general public. So, even with a reduced price, the overall market size could increase.

It’s crucial to underline that our discussions remain hypothetical; we're venturing into uncharted waters, grappling with vast numbers and concepts that few yet perceive. However, we hope that the concept of an aging drug with some quantifiable economic value moves the needle closer from the abstract to the tangible. 

A key question remains–is it even possible to get a drug approved and labeled for aging and age-related diseases in humans? We’ll explore this topic in depth in a future piece.

Written by: Alex Kesin, Maggie Li, and Alex Colville.

Acknowledgements: Wen Kin Lim, Dr. Joan Mannick, Laura Deming, Lily Clayton, Ada Nguyen, Carol Magalhaes, and Raiany Romanni.

Further reading

• Aging drugs should be boring

• The Secret Cambrian Bio Master Plan to Build Drugs to Treat Aging (just between you and me)

• Quantifying risk/reward in biotech investing

• What Could New Anti-Obesity Drugs Mean for Medicare? | KFF

• Medicare Coverage of Weight Loss Drugs Could Significantly Reduce Costs | USC Schaeffer

• Inside the push to get weight loss treatment covered by Medicare | The Hill

(back to top)

Appendix A. List of drugs commonly taken by older adults

Antidiabetics (for type 2 diabetes)

• GLP-1 receptor agonists (current approved versions listed here are not small molecules, but a must-have contender for age-related disease. Stay tuned to an upcoming piece for why they are important!)

  • Semaglutide (Rybelsus, Wegovy, Ozempic)

  • Tirzepatide (Mounjaro)

• SGLT2s

  • Canagliflozin (Invokana)

  • Dapagliflozin (Farxiga)

  • Empagliflozin (Jardiance)

• Biguanides

  • Metformin (Glucophage)

Cardiovascular health

• Beta-blockers

  • Metoprolol (Lopressor)

  • Atenolol (Tenormin)

  • Bisoprolol (Zebeta)

• Statins

  • Atorvastatin (Lipitor)

  • Rosuvastatin (Crestor)

  • Simvastatin (Zocor)

• Angiotensin-converting-enzyme (ACE) inhibitors

  • Enalapril (Vasotec)

  • Lisinopril (Zestril)

  • Captopril (Capoten)

Bone health

• Bisphosphonates

  • Alendronate (Fosamax)

  • Risedronate (Actonel)

  • Ibandronate (Boniva)

Central nervous system 

• Donepezil (Aricept)

• Levodopa (Rytary)

(back to top)

Appendix B. Price of brand name drugs commonly taken by older adults 

Values for the out-of-pocket price per month are from GoodRx. 

(back to top)

Appendix C. Breakdown of drug pricing and payer coverage in the US 

In the U.S. pharmaceutical industry, a complex web of financial transactions connects patients, insurers, pharmaceutical benefit managers (PBMs), pharmacies, wholesalers, manufacturers, and government entities like the Center for Medicare & Medicaid Services (CMS). Figure 7 provides an overview with examples.

Patients fund insurers through premiums; insurers then pay for drugs and services to pharmacies and manufacturers, with CMS mediating financial activities for Medicare/Medicaid. Manufacturers grant rebates to PBMs and pharmacies, influencing drug pricing and insurance formularies, with a share of rebates going back to insurers. CMS's role in rebates, affected by legislation like the Inflation Reduction Act (IRA), and the involvement of Group Purchasing Organizations (GPOs) in price negotiations, further complexify the system. The entire process dictates medication pricing and accessibility, underlining the intricacy of healthcare economics. Here is a quick video that explains Rx drug rebates and premiums.

Coverage by payers for patients 

The older adult population is split in terms of coverage by payers: among older adults (aged 65 and over), 40.9% were covered by private insurance (with or without Medicare), 28.0% had Medicare Advantage, 13.6% had traditional Medicare only, 8.9% had some other coverage (including military coverage without Medicare), 7.6% were covered by Medicare and Medicaid (dual-eligible), and 1.0% were uninsured. Of all Medicare recipients, around 74% are enrolled in Part D. 

(back to top)

Appendix D. Growth of drug price, TAM size, and market adoption over time 

The monthly list price and the monthly net price correspond to the drug pricing used for the cash-pay and reimbursement models, respectively. 

The total number of older adults in the US are interpolated values from 2017 US census projections of total population size and percentage of the total population aged 65 and older until 2060.

The annual market penetration rate is found by multiplying the annual cumulative percentage of total adopters by the maximum market penetration rate. Following the Diffusion of Innovation model, the cumulative percentage of total adopters is found by approximating the area under the bell curve. Over time, as evidence of the drug's efficacy grows, more individuals will likely adopt it, with a non-linear growth trend similar to the red curve in Figure 2.

(back to top)

Appendix E. Learnings from statins for long-term Medicare savings

If massive long-term savings in terms of improved quality of life and reduced healthcare costs due to chronic conditions become clear, then the way Medicare spends on a drug approved and labeled for aging and negotiates pricing with drug manufacturers could look very different. This similarly applies to commercial health insurers. Insurers at the end of the day care about cost savings and value so it’s important to think about this part of an aging drug.

In our analysis, we presented a simplified model for projecting reimbursement model revenue based on cash-pay revenue. However, the actual situation is more complex, especially when considering Medicare. Medicare's reimbursement is contingent upon cost-effectiveness; they reimburse for treatments that demonstrate savings, meaning an aging drug must prove to be cost-efficient.

Here, we can learn from the story of statins.

Imagine you are a person aged 65+ with high LDL (bad cholesterol) in the early 1980s. Your doctor is concerned about your cholesterol level’s potential impact on your heart health; they recommend you improve your lifestyle. If there’s an exceptional concern, they might put you on a fibrate, a class of cholesterol-altering drugs. 

A few years into the late 80s; the first statins have been approved and are now cautiously being prescribed to those who most need them. While it is known they are effective at lowering LDL cholesterol, they’re so new on the scene that you’re not likely to get your hands on one. If you were, you’d likely have to pay at the full list price: not only did Medicare Part D – the payer that would make statins affordable for patients–not even exist yet, but nobody knew what these things were! Good luck trying to get private insurance to cover a medication nobody had even heard about.

Fast forward to 1994. The 4S study–a multi-year trial of a statin called simvastatin – is completed and its findings published in the Lancet, and the results are stupefying. 

The few thousand patients with a history of angina/heart attacks who were on a statin saw a ~35% reduction in LDL-C and a ~30% decrease in overall mortality. Perhaps most relevant of all: hospitalization costs went down 32% in the statin-taking group, demonstrating that statins were cost-effective in addition to significantly improving the quality of life in patients with coronary heart disease. 

Essentially, payers would be saving money if they paid to get their patients on statins, as opposed to the status quo. Since then, statins have become some of the most commonly prescribed drugs in America. Lipitor has grossed $163 billion in sales since its release in 1997, making it the third-best-selling drug of all time. Backtesting savings for payers on simvastatin (a statin) in the 90s, the estimated price per patient via savings comes out relatively close to the actual price at the time for the drug.

According to this 1995 paper, the estimated expenditure for inpatient care (e.g. hospital care, nursing home care) was $8 billion. If you recall from the 4S paper, simvastatin reduced hospital costs by ~30%, so we could crudely estimate savings by the following formula:

$ savings = $ expenditures * % savings → $8 billion * 0.3 = $2.4 billion / year in savings

Divide the savings by the number of patients (~2.6 million hospitals admit patients with heart failure per year according to the paper) and we should get a reasonable number for the pricing of the brand name Simvastatin (Zocor). 

Inflation-adjusted $2.4 billion from 1995 dollars to 2023 dollars is $4.84 billion

$4.84 billion savings / 2.6 million patients = ~$2017 in savings per patient every year

A yearly prescription of $3,718.96 per year is ~$168 per month for patients, cash pay.

In 1995, the wholesale list price for a 30-tablet supply of simvastatin (Zocor) was $54.00 for 20mg and $97.80 for 40mg. Adjusted for inflation, that would be $111.87 on the low end (20mg) and $202.61 on the high end (40mg). A price of $202.61 per month falls in the same ballpark as $168 per month. 

The main takeaway from statins? The potential cost-effectiveness of an aging drug matters a lot: patient adoption depends on insurance coverage, which highly depends on the savings of said drug. 

Using Medicare spending data to validate the reimbursement model 

Based on Medicare spending data from 2018, we can complete a similar calculation. For starters, here is the spending data from that year per disease:

Notice that a lot of these diseases are comorbid–which is likely why the final total is quite big. Instead of opting for this number, we could instead refer to the number that Medicare quotes as their annual spending of $829 billion in 2021. That would put spending at $942 billion, adjusted for inflation.

Taking statins as an example, which decreased hospitalization expenses for heart disease by 30%–if an aging drug could achieve a similar reduction in the expenditure related to age-associated diseases for Medicare–it could result in savings of approximately 30% of $942 billion, equating to around $283 billion in savings annually.

When considering our reimbursement model, which estimates US. revenues of $50 to $100 billion annually from such a drug, it appears that the spending by Medicare could shift some spending from a range of drugs to an aging drug. Nevertheless, this would still result in a net saving of around $183 to $233 billion each year for Medicare.

Through our investments, we strive for counterfactual impact. We are creating a future where all people have agency over how long they live in good health. 

Our vision is to make aging and age-related disease optional for all.

Why does it matter?

We’re at a historic inflection point in biotech. In the next decade, the preservation of a healthy life will go from a futuristic ideal to an actionable right. 

In the original draft of The Declaration of Independence, Thomas Jefferson famously wrote that all humans “derive rights inherent and inalienable, among which are the preservation of life, liberty, and the pursuit of happiness.”

This is our goal at age1: the preservation of healthy life, to enable a more dignified future globally for humans of all ages. 

Some of the biggest human accomplishments progressed us towards this goal. Since the inception of antibiotics in the 1910s, they have extended the global life expectancy by 23 years. These now-taken-for-granted solutions combat the propensity of humans to develop infections constituting modern medicine, yet the thought of democratizing such therapeutics was once laughable. Similarly, we foresee a future with universal access to a healthy life enabled by longevity biotech.

The preservation of healthy life is critical because it will enable:

• Reduced suffering: Loss of identity, clarity of mind, physical independence, and grief over the decline of our loved ones are all pains we’ve come to expect as necessary counterparts to life. We strive for a future where this is not the norm.

• Increased time with family, friends, and yourself: To maintain the ability to spend quality time with family, friends, pets, nature – and yourself – for longer. 

• Diminished financial burden: By 2029, the United States will spend $3 trillion dollars yearly – half its total federal budget– on the medical and social care of adults aged 65 and older. Health, quality of life, and longevity are foundational to global prosperity. 

The future age1 wants to create

Currently, there are zero FDA-approved interventions for aging or longevity. We know of 0 interventions that extend maximum lifespan in humans. And 0 clinical trials are underway to test if a therapeutic can intervene in human aging, holistically. 

Zero.

Most of the field is focused on the “hallmarks of aging”, a useful but reductionist framework that doesn’t address the systematic, multifactorial nature of aging biology. As a result, the field is dogmatic and not growing at the pace it should.

We have tunnel-visioned on the same shortlist of mostly small-molecule interventions, candidates like rapamycin and senolytics, without substantive progress in determining exactly how they affect aging in humans. After 45 years of research in the modern era of the field, we only have evidence to rigorously support that lifestyle/exercise and nutrition can extend the average life expectancy in humans. 

We need to change this current trajectory. Our goal is to:

• Finally put small molecule interventions to the test in a rigorous way to answer once and for all if they can affect human aging. 

• Push the aging field to embrace newer modalities such as biologics, cell therapy, gene therapy, and tissue engineering to fulfill the promise of regenerative medicine.

The future age1 wants to create is one with effective and safe moonshot interventions, approved for human use. We focus on interventions which:

• Delay: Delay or prevent aging

• Replace: Replace aged or damaged cells and tissues with youthful ones

• Restore: Turn the clock back to restore the age of cells and tissues to a youthful stage

• Pause: Pause the aging clock in human tissues and cells by inducing stasis

.

Below is how we change the status quo – to make the above a reality.

Build now, age later

We must partner with the most ambitious, talented founders and the best LPs in order to tackle one of medicine’s biggest problems.

This inflection point in history may never present itself again. In a new letter from Laura, she shares her predictions and why this is the decade to build for the future of longevity. 

age1 will seed the best founders the field has ever seen, equipping them with the right resources to finally attain proven safe, equitable interventions to delay, restore, replace, and pause aging in humans. With the first-ever FDA-approved longevity drugs from these companies, they’ll scale to obtain $10B+ valuations and impact global health systems.

For these founder-led biotechs to succeed and thrive, we need world-class founders, talent, and investors.

That’s you. 

It is only through a collective effort that we can deliver the best returns for our future. Every investment in the aging field – time, effort, and capital – doesn’t just extend the runway of a company or team, but the runway of healthy human life.

Visit the age1 website