The Differential — Diligence Note

Reading an event-driven Phase 3: what "78 of 80" actually tells you

Event-driven survival trials report progress in a way that looks precise and is almost entirely uninformative about the result. A field guide for the non-clinician.

Clinical signalN/A
Reg. pathExplainer
Balance sheetN/A

SELLAS Life Sciences has been putting out a steady drumbeat of press releases about its pivotal Phase 3 REGAL trial in AML: 60 events reached the interim analysis threshold in late 2024, 72 events by December 2025, 78 events by mid-May 2026. Each release is framed as progress. Each time, the stock has moved on the news. And each time, the update is genuinely, deliberately uninformative about whether the drug works.

That's not a criticism of SELLAS's disclosure — it's a structural feature of how event-driven survival trials work, and it's worth understanding before you react to the next headline that says "X of Y events reached."

What "event-driven" means

REGAL is a randomized trial testing galinpepimut-S (GPS) as maintenance therapy in AML patients who've achieved a second complete remission. Its primary endpoint is overall survival (OS). Rather than reading out at a fixed calendar date, the trial's final analysis triggers once a pre-specified number of events — in this case, deaths — has accumulated across the trial population: 80 events, following an interim analysis at 60.

This design exists because survival trials need enough deaths to have statistical power to detect a difference between arms. Read out too early, with too few events, and even a real treatment effect can be statistically invisible. The event count is a proxy for statistical maturity, not for how the drug is performing.

Why the running tally tells you almost nothing

Here's the part that trips people up: the company remains blinded to which events are occurring in which arm. SELLAS has been explicit about this in every REGAL update — the contract research organization reports a pooled, aggregate event count, and neither the company nor investors know whether those events are disproportionately occurring in the treatment arm, the control arm, or split evenly.

Diligence note
A pooled event count climbing from 60 to 72 to 78 tells you the trial is approaching maturity. It tells you nothing about the treatment effect. A trial can hit its full event count with a strongly positive result, a null result, or a negative one — the counter moves identically in all three scenarios. Trading on the event count alone is trading on noise.

What can be informative, and what SELLAS has also disclosed, is the outcome of the pre-specified interim analysis: in January 2025, the Independent Data Monitoring Committee (IDMC) completed its interim review at 60 events and recommended the trial continue without modification, having confirmed the drug exceeded pre-specified futility criteria with no safety concerns. That's a real, if limited, signal — an unblinded committee looked at the data and didn't see grounds to stop the trial for futility or harm. It is not equivalent to a positive efficacy read; futility thresholds are set deliberately low so that only a clearly failing trial gets stopped early.

What to actually watch for

The read

Every event-count update is real information about trial timeline and essentially no information about trial outcome. Treat "78 of 80" the way you'd treat a countdown clock, not a scoreboard — it tells you roughly when you'll know the answer, not what the answer is.

This piece reflects clinical and financial analysis based on public filings and disclosures as of August 2026. It is not investment advice, and Stevenson HemOnc Advisors does not hold, and has not been compensated by, any company named above.

Trying to interpret a specific trial update?

A same-day read on what a press release actually says — and doesn't.

Or just get the next one