What a Phase 2 Result Does and Does Not Tell You

The one-sentence version

A phase 2 trial is a company asking itself which dose to take into phase 3, in a small group of patients, over a short window. It is a question, not an answer.

That is why a phase 2 headline is worth reading closely and worth discounting heavily — and why the two things are not in tension.

What phase 2 is built to do

Find a dose. Most phase 2 trials run several arms at once precisely because the right dose is unknown.

Detect a signal. The question is whether the effect is big enough to be worth a phase 3, not how big it is.

Surface the common side effects. Sample sizes in the low hundreds can find the side effects that happen often. They cannot find the ones that happen rarely.

Nothing in that list is "establish that the drug works". That is the next trial's job.

Worked example: retatrutide, phase 2 into phase 3

The phase 2 record is NCT04881760: 338 adults with obesity or overweight, randomised across a placebo arm and six retatrutide arms, sponsored by Eli Lilly. Its stated primary endpoint is percent change in body weight at week 24¹.

The published result at 48 weeks was a least-squares mean weight change of −24.2% in the 12 mg arm against −2.1% on placebo¹. That −24.2% is the number that travelled. It is a secondary endpoint. The primary was the week 24 comparison, where the same arm showed −17.5% against −1.6%¹.

The phase 3 that has since reported is TRANSCEND-T2D-1: 537 adults with type 2 diabetes inadequately controlled by diet and exercise, 40 weeks, at 48 sites in the USA, Mexico and India. Its primary endpoint is not weight at all — it is change in HbA1c. Weight was a key secondary, and came in at −15.3% in the 12 mg arm against −2.6% on placebo².

Now hold the two next to each other. −24.2% and −15.3% are not the same drug performing differently. They are different populations (obesity without diabetes versus type 2 diabetes), different durations (48 weeks versus 40), and different endpoint hierarchies. Reading one as a downgrade of the other is the mistake. Reading them as answering different questions is the skill. Our retatrutide page tracks that programme as it reports.

Worked example: semaglutide, where the shape held

The phase 2 was NCT02453711 — 957 participants, daily subcutaneous dosing, 52 weeks, with the highest arm producing an estimated mean weight loss of −13.8% against −2.3% on placebo³.

STEP 1, the phase 3, enrolled 1,961 adults without diabetes, moved to once-weekly dosing at 2.4 mg, ran 68 weeks, and reported −14.9% against −2.4%⁴.

The number held. The schedule did not: the phase 2 studied a daily injection and the approved product is weekly. A dose read off a phase 2 arm can be obsolete by the time the same molecule reaches the label. See semaglutide for where that programme ended up.

The base rate, and why it is contested

The largest public analysis of drug-development success used 406,038 trial entries covering 21,143 compounds registered between January 2000 and October 2015. It estimated that 58.3% of programmes transition from phase 2 to phase 3, and that 13.8% of programmes entering phase 1 eventually reach approval⁵.

The same paper's comparison table is the more useful artefact. Earlier analyses of the same question, using a different counting method, put the phase 2 to phase 3 transition at 30.7% and 32.4%⁵.

So the honest statement is a range, not a number: somewhere between roughly a third and roughly six in ten phase 2 programmes advance, depending on how you count. Either way, a meaningful share of phase 2 results are the high-water mark of that molecule's public record.

What a phase 2 does not establish

That the effect size will replicate. Different population, different duration, different comparator — all three move the number, and all three change between phase 2 and phase 3.

That the drug is safe. Rare harms are, by definition, invisible at n in the hundreds. That is what sample size buys you.

That the endpoint matters to a patient. A phase 2 usually measures whatever moves fastest. Whether that translates is the surrogate endpoint question.

That anything is approved. Phase 2 is a stage of investigation. It carries no regulatory status whatsoever.

Reading a phase 2 headline in four moves

Find the primary endpoint on the registry record, not in the press release. If the quoted number is a secondary endpoint at a later timepoint, say so out loud — as with retatrutide's −24.2%.

Check the comparator. Against placebo, against an active drug, or against baseline are three different claims.

Check who was enrolled. The retatrutide pair above differs on population, duration and primary endpoint at once, which is exactly why two trials cannot tell you how much of the gap each one caused.

Check whether a phase 3 exists yet. If it does not, the phase 2 is the ceiling of what is known, and it should be quoted that way.

When a peptide has no phase 2 at all

Most compounds marketed as "peptides" are not in this conversation. A ClinicalTrials.gov search for BPC-157 returned 3 studies in August 2026. The oldest is a phase 1 safety and pharmacokinetics trial in 42 healthy volunteers, last updated in December 2015 and carrying a status of unknown⁶. The largest is a phase 2 hamstring-strain trial listed as recruiting, with an estimated primary completion of February 2027⁷. The third is a 40-participant study of peptide gummies with no phase assigned.

That is the state of the file, and it is a different kind of statement from "the phase 2 was disappointing". There is no phase 2 result to discount. Our BPC-157 page reads that record in full.

What would move a phase 2 finding up a tier

A phase 3 in the population the compound is actually marketed to, powered for an endpoint a patient can feel, with the result posted rather than described. Until that exists, a phase 2 is a well-conducted reason to keep looking — and no more than that.