A cancer drug 'works.' But did patients live longer, or did a tumor just shrink? The answer is which endpoint the trial made its primary measure. Here is how to read it.
Only one number proves a cancer drug helps patients live longer, and that is overall survival. Progression-free survival, response rate, and pCR are faster stand-ins that may or may not convert. A plain guide to reading a trial's primary endpoint, the hazard ratio, the survival curve, and the surrogate-endpoint trap, with real trials from KEYNOTE-189 and daraxonrasib to the INTerpath-001 mRNA vaccine.
Several companies want to stretch IBD antibodies from a shot every few weeks to one every three months, or even six. Keeping the drug in the blood that long is the easy part. Whether the disease stays controlled right up to the next dose is the hard part, and the bet the newest programs are making.
“Long-acting” hides three claims: the drug stays in the blood (PK), keeps blocking its target (PD), and keeps the disease controlled (durability). Only the last proves a long interval works. Across 1,147 IBD trials, TL1A is the most crowded emerging target, with 7 antibodies, 6 sponsors and 36 trials, and 3 already in Phase 3. The leading long-acting candidates report 56–85 day half-lives, a 3–5× leap. But the trough, how much drug is left at the next dose, tells the real story: at the proposed quarterly and six-month intervals it sits at roughly 19–47% of peak, inside the range today’s biologics already tolerate. The PK was never the wall; durability is, and only XmAb942 is yet testing a long interval in patients.
Two of the newest PsA drugs block IL-17F as well as IL-17A; the older ones block only IL-17A. A plain walk through what the extra target actually buys: clear help on the skin, no proven help on the joints yet, and a fungal cost in the middle.
Sonelokimab met its Phase 3 endpoint in psoriatic arthritis in August 2026, the second drug engineered to neutralise IL-17F alongside IL-17A. The first, bimekizumab, is already approved; secukinumab and ixekizumab block IL-17A alone. On the joints, the dual drugs post ACR50s around 43–44% versus 35–40% for the A-only incumbents — but the trials aren't comparable, so the gap isn't proof. On the skin the story firms up: PASI90 around 61–69% versus 33–49%, plus a head-to-head in psoriasis where bimekizumab out-cleared secukinumab. Blocking IL-17F looks like a skin benefit, not yet a joint one.
The TNF receptor superfamily has produced fifteen approved medicines, and every one of them works by shutting something down. Six receptors have been pushed the other way for twenty years, and not one agonist antibody has been approved.
Adalimumab, etanercept, denosumab and belimumab target different ligands, but they all modulate signalling through the TNF receptor superfamily, and every one works by turning a pathway off. For about twenty years the field has also tried the opposite, activating CD40, OX40, 4-1BB, GITR, TRAIL-R and CD27 with agonist antibodies. That has produced nothing approved, in the United States or Europe. These receptors only signal when several copies are clustered into a particular arrangement, and the clustering has to be supplied by the patient's own FcγRIIB, so potency depends on the tissue as much as on the molecule. Romiplostim shows agonism is not impossible — its receptor just asks for something a two-armed drug can deliver.
Of 214 industry IBD trials terminated or withdrawn over two decades, only ~23% stopped for efficacy or safety. The rest ran out of patients, money, or corporate priority — the trial failed, not the biology.
We read the sponsor’s stated reason — the ClinicalTrials.gov whyStopped note — on all 214 terminated or withdrawn industry IBD trials. Business and portfolio decisions are the single largest cause (36%); inability to enroll is next (22%). Lack of efficacy (20%) and safety (4%) together are only ~23% — so when an IBD trial is pulled early, ~77% of the time the reason says nothing about whether the drug worked.
Alnylam's two siRNA drugs hit the same liver target. One failed an FDA review in 2023. Two years later the other was approved for both forms of ATTR amyloidosis. The mechanism didn't change. The trial design did.
In September 2023 an FDA advisory committee voted 9-3 that patisiran's benefit supported approval for cardiac ATTR amyloidosis. The FDA rejected it anyway. Eighteen months later it approved vutrisiran — a near-identical Alnylam siRNA against the same liver target — for both forms of ATTR at once. Patisiran's APOLLO-B was 360 patients, 12 months, a six-minute-walk endpoint (+14.7 m, p=0.0162) with a secondary composite that missed at a win ratio of 1.27. Vutrisiran's HELIOS-B was 655 patients, up to 36 months, a hard mortality-plus-CV-events composite cut by 28%. Same mechanism, opposite verdicts — the molecule never changed, the trial did.
