Human genetic evidence and the success of drug programmes in clinical development

Evidence Research Format (ERF) v0.1 · Investment · first edition

Research status

AI-assisted draft Claims checked against publicly available abstracts and records; some details taken from the main text by automated extraction and marked as such. Not reviewed by a qualified drug-development or investment professional.

Last verified

2026-10-24

Scope

Drug programmes (target–indication pairs) that have already entered clinical development. Probability of advancing through clinical phases. Not company valuation, not share-price return.

Format

ERF v0.1 (SYCDS)

Not investment advice. This page summarises published research for decision analysis. It does not evaluate any company, asset or security. Probability of clinical success, company valuation and investment return are three different questions; this page addresses only the first.

Research Question

Among drug programmes already in clinical development, how much incremental information does human genetic evidence give about the probability of success, and what are the limits of that information for biotech due diligence?

Key Findings

1. Direction is consistent. Three analyses from 2015 to 2024 find that drug mechanisms with human genetic support progress through clinical development more often than those without. [E1, E2, E3]
2. Size, as a relative measure. The largest analysis (Minikel et al., 2024) reports in its abstract that the probability of success is 2.6 times greater for drug mechanisms with genetic support than without. According to its Methods, this is a ratio of success probabilities (a relative risk, not an odds ratio) for target–indication pairs in one commercial pipeline database; the phase transition behind the 2.6 figure is not stated in the abstract. It is not the probability of success of any programme. [E3, E4]
3. It depends on the kind of evidence. The independent re-analysis reports more than two-fold effects when the causal gene is clear (abstract); its main text reports weaker effects for ordinary GWAS associations, often not distinguishable from no effect. The 2024 analysis reports a similar ordering. (Main-text details: automated extraction, manual review pending.) [E2, E5, C1]
4. It is reported to vary by phase and therapy area (main-text detail, manual check pending). [E6]
5. A relative figure is not a programme probability. The 2024 abstract states that about 10% of clinical programmes reach approval. That baseline covers programmes with and without genetic support, so it cannot be combined with the 2.6 figure to give an absolute probability for a single programme. [E7, E9]
6. Assessment. Genetic support is a well-evidenced due-diligence factor for programmes already in the clinic, provided the type of evidence and the confidence in the causal gene are assessed. It is not a predictor of investment return and should not be used alone.

Evidence Table

Abstract-checked checked against the source abstract or public record only (not the full text) · Partly verified / Inferred · Interpretation judgment, not a checkable fact
No claim on this page is full-text-checked or independently reviewed.

Evidence Limitations

• How this page was checked: abstracts and publication records of R1–R5 were checked. Statements in this section and under Counterevidence that go beyond those abstracts (for example funding, sensitivity analyses and evidence-type contrasts in R2 and R3) come from automated extraction of the main text; manual review is pending.
• All studies are retrospective and observational. Programmes with genetic support may differ in other ways (target quality, sponsor, disease area), so the association is not proof that genetic evidence causes success.
• Reverse causation is possible: curated databases such as OMIM may record gene–disease links because a drug mechanism was already known [R2].
• Results are sensitive to method choices, such as the trait-similarity threshold and how GWAS variants are mapped to genes [R2, R3].
• RS covers only programmes that entered clinical development. It does not estimate how often a disease-associated gene yields a successful drug [R3].
• The direction of the genetic effect (gain or loss of function) was not analysed; whole-exome and whole-genome studies were largely not included [R3].
• The main analyses were funded by industry or investors (R2: AbbVie; R3: Deerfield, with authors employed by or consulting for it). This does not make the results wrong, but it is disclosed here.
• Success here means phase progression. Development cost, time, market size, competition, pricing and patents, which determine investment return, are outside these studies.
• A company-level report that one pharmaceutical company’s success rate rose after it changed its R&D framework was identified only through a secondary report, with an unconfirmed definition of “success”. It is not used on this page.

Counterevidence

• C1 — GWAS evidence is weak on its own. In the independent re-analysis, the GWAS effect was lower than originally reported and often not distinguishable from no effect in validation sets; it approached the Mendelian effect only when restricted to variants linked to coding changes [R2].
• C2 — The wrong kind of genetic association can be a negative signal. Genetic associations with traits dissimilar to the drug’s indication were associated with lower approval odds than having no association [R2].
• C3 — Early-phase effect is smaller. Phase I-to-II estimates were lower than originally reported [R2]; the effect is smallest in phase I [R3].
• C4 — Practical offsets. Genetically supported targets may need upregulation or harder modalities, or may be useful in fewer indications [R3].
• C5 — Retrospective counts are not predictions. The 66% figure for 2021 approvals (E8) uses a broad definition of support and is descriptive only [R4].

Decision Implications

• For due diligence on a programme already in the clinic, the evidence supports asking: is there human genetic support, of which kind (Mendelian, GWAS, somatic), and how confident is the link to the causal gene?
• The evidence supports treating Mendelian-level support and ordinary GWAS hits as different strengths of evidence (E5, C1).
• The 2024 analysis reports the largest effect in phases II and III (E6, manual check pending); treat this as a working hypothesis, not a settled result.
• The evidence does not support converting the 2.6 figure into a probability for a single programme, an expected return, or a valuation multiple (E9).
• The evidence does not support using a cross-area average for a specific indication without checking the therapy-area estimate.
• Example: PCSK9 is a case where genetics anticipated a successful drug (see the Biomedical PCSK9 review). A single success is not frequency evidence.

What Would Change the Assessment

• Prospective, pre-registered tracking of programmes classified by genetic support at a fixed date, followed to their actual outcomes.
• Replication in an independent pipeline database, or by authors without industry or investor funding, showing a materially different RS.
• Evidence that the advantage shrinks as genetic support becomes standard practice (crowding into the same targets).
• Large sequencing-based studies showing higher or lower predictive value than array-based GWAS.
• Manual full-text checking of R3 (planned), or review by a qualified professional, which would change the check labels and research status above.

Discuss with Tac Yang

Last Verified / Update History

• 2026-10-24 — Revised after content review (REVISE): the estimand, unit and scope of the 2.6 figure stated precisely; the earlier inferred row “even with genetic support, most programmes still fail” withdrawn because it combined baselines from different studies; findings that rest on main-text details marked as pending manual review.
• 2026-10-24 — First edition (ERF v0.1). Sources checked at abstract or record level; main-text details marked as automated extraction pending manual recheck. Last verified: 2026-10-24.

Sources

1. [R1] Nelson MR, Tipney H, Painter JL, et al. The support of human genetic evidence for approved drug indications. Nat Genet. 2015;47(8):856–860. doi:10.1038/ng.3314 · PMID 26121088
2. [R2] King EA, Davis JW, Degner JF. Are drug targets with genetic support twice as likely to be approved? Revised estimates of the impact of genetic support for drug mechanisms on the probability of drug approval. PLoS Genet. 2019;15(12):e1008489. doi:10.1371/journal.pgen.1008489 · PMID 31830040
3. [R3] Minikel EV, Painter JL, Dong CC, Nelson MR. Refining the impact of genetic evidence on clinical success. Nature. 2024;629(8012):624–629. doi:10.1038/s41586-024-07316-0 · PMID 38632401
4. [R4] Human genetics evidence supports two-thirds of the 2021 FDA-approved drugs. Nat Rev Drug Discov. 2022 (Open Targets / EMBL-EBI). doi:10.1038/d41573-022-00120-3. Authors, volume and pages not checked.
5. [R5] Wong CH, Siah KW, Lo AW. Estimation of clinical trial success rates and related parameters. Biostatistics. 2019;20(2):273–286. doi:10.1093/biostatistics/kxx069 · PMID 29394327. A corrigendum exists; not checked. Link unconfirmed automatic link check returned HTTP 403; pending manual check in a browser.

SYCDS Evidence Research · ERF v0.1 · Investment · AI-assisted draft, not professionally reviewed · Not investment advice