Onpode
Cover art for An AML leukemia vaccine just reduced death risk in Phase 3—but what does the win really mean for patients

An AML leukemia vaccine just reduced death risk in Phase 3—but what does the win really mean for patients

November 19, 2026 · 11 min

Eliza Ward & Brian Reed

Hey. You see the SELLAS announcement this morning? Yeah — I've been sitting with it. November 17th. SELLAS Life Sciences and Memorial Sloan Kettering put out Phase 3 results on a trial called REGAL, and the headline number is a 41% reduction in death risk for AML patients. In second complete remission, right? So these…

On November 17, 2026, SELLAS Life Sciences and Memorial Sloan Kettering Cancer Center (MSK) announced that the investigational cancer vaccine galinpepimut-S (GPS) met the primary endpoint of the pivotal Phase 3 REGAL trial in acute myeloid leukemia (AML).

0:0010:41
Get the next episode on Science →

Follow it free — new episodes land in your feed.

Or make your own — any topic, in minutes

More Onpode episodes on Science →

About this episode

On November 17, 2026, SELLAS Life Sciences and Memorial Sloan Kettering Cancer Center (MSK) announced that the investigational cancer vaccine galinpepimut-S (GPS) met the primary endpoint of the pivotal Phase 3 REGAL trial in acute myeloid leukemia (AML).

Grounded in 9 sources
Management of Acute Myeloid Leukemia: A Review ↗ · pmc.ncbi.nlm.nih.gov
Sellas Life Sciences and Memorial Sloan Kettering Cancer Center announce Phase 3 success for first AML leukemia vaccine ↗ · washingtonpost.com
SELLAS Life Sciences Announces Positive Pivotal Phase 3 REGAL Trial Data Showing Immunotherapeutic Cancer Vaccine, Galinpepimut-S (GPS), Demonstrated a Statistically Significant and Clinically Meaning ↗ · morningstar.com
SELLAS Life Sciences Reports Positive Phase 3 Results for AML Cancer Vaccine – Azat TV ↗ · azat.tv
SELLAS Cancer Vaccine Improves Overall Survival in Phase 3 AML Trial | BioPharm International ↗ · biopharminternational.com
Racura Oncology Receives First Site Approval for Seamless Phase 2b/3 EMILI Trial of RC220 in Relapsed or Refractory Acute Myeloid Leukemia - BioSpace ↗ · biospace.com
Sellas’ cancer vaccine slashes death risk by 41% in late-stage AML study - BioSpace ↗ · biospace.com
SELLAS Life Sciences Announces Positive Pivotal Phase 3 ↗ · globenewswire.com
Racura Oncology Receives First Site Approval for Seamless Phase 2b/3 EMILI Trial of RC220 in Relapsed or Refractory Acute Myeloid Leukemia ↗ · prnewswire.com
Read transcript

Brian Reed: Hey. You see the SELLAS announcement this morning?

Eliza Ward: Yeah — I've been sitting with it. November 17th. SELLAS Life Sciences and Memorial Sloan Kettering put out Phase 3 results on a trial called REGAL, and the headline number is a 41% reduction in death risk for AML patients.

Brian Reed: In second complete remission, right? So these are patients who already had their cancer come back once.

Eliza Ward: Right. And the survival numbers — median overall survival in the galinpepimut-S arm was 24 months. Control arm, best available treatment, was 14.7. That's nine extra months in a population that almost never gets good news.

Brian Reed: Nine months is — I mean, in this population that's not a small thing. And the stock moved, what—

Eliza Ward: More than 150%. Same day. SELLAS on NASDAQ, single session.

Brian Reed: So the market registered it as real.

Eliza Ward: The market registered something. Whether that's the same as real is — well, that's what I want to get into.

Brian Reed: But registered what, exactly? Because 'first AML vaccine' is what I keep seeing in the headlines, and that framing is doing a lot of work.

Eliza Ward: It is. And it's not accurate — or at least it's not the whole picture. GPS, galinpepimut-S, it targets something called WT1. Wilms Tumor 1 protein. It's overexpressed in AML cells. What GPS does is train your T cells to recognize that protein and attack cells carrying it. Think of it like handing your immune system a wanted poster for one specific type of cancer cell — it only works if WT1 is the criminal your body is hunting.

Brian Reed: Okay, so it's not 'vaccine' in the flu-shot sense.

Eliza Ward: Not even close. And the population — wait, this is the part people are skipping. REGAL enrolled 126 patients. Second complete remission after relapsed AML. That means these are people whose cancer came back, they responded to treatment again, and now they're trying to hold that response. That's not all AML. AML five-year survival is already around 30% broadly. Second-remission patients are a subset of a subset.

Brian Reed: So 'first AML vaccine' is — I mean, it's more like 'first GPS result in this one very specific, very hard-to-treat corner of AML.'

Eliza Ward: Right. And then there's the comparator — the control arm was Best Available Treatment, BAT. Investigator-selected. Which sounds reasonable until you realize that's not a single standardized regimen. Some patients might have been getting venetoclax combinations, others something else entirely. It's 126 people spread across different clinical approaches.

Brian Reed: So GPS beat a moving target, not a fixed one.

Eliza Ward: That's — yeah, that's one way to put it. It doesn't erase the result. Nine months median survival difference in this population is real. But 'first AML vaccine' as a headline implies something GPS isn't claiming to be — a broad intervention. It's a narrowly targeted tool for a tightly defined group, and the trial design means the comparison is messier than the press release suggests.

Brian Reed: People are running with this morning — I keep seeing 'confirmed result, publication-ready' — and that's the take I want to push on. Because the topline announcement didn't include confidence intervals for that hazard ratio of 0.59.

Eliza Ward: Right. No CIs.

Brian Reed: And that matters because — hang on, let me put it concretely. Say you're a hematologist. Tuesday morning. Your patient comes in with a printout of this headline. You want to check: does 0.59 sit in a band between 0.40 and 0.78, or is it tighter, like 0.52 to 0.67? Those are radically different stories about how solid this actually is. And right now, you literally cannot check. The full data aren't out.

Eliza Ward: Wait — is that just normal lag before the BLA filing, or is that a choice SELLAS made?

Brian Reed: That's what I can't answer yet. And — okay, by itself, maybe it's lag. But then you layer on that the REGAL analysis plan was changed during the trial. The control arm, BAT, outperformed the pre-specified assumptions. Someone looked at accumulating data and adjusted. That's not automatically disqualifying, but it means the omission of confidence intervals is harder to wave away. Two data points, same direction.

Eliza Ward: p=0.035 clears 0.05, but only narrowly. If the confidence interval is wide, that number starts looking — I mean, it doesn't fail, but it doesn't inspire the same confidence the headline implies.

Brian Reed: Exactly — and subgroup analyses, quality-of-life measures, durability data, none of that's public yet either. So 'confirmed and ready to act on' is the wrong frame. What's confirmed is a p-value and a hazard ratio. The shape of the effect is still pending.

Eliza Ward: And the stock offering — that same-day $150 million raise — that's actually where this gets more complicated, and we should get into what it signals about whether SELLAS itself believes the FDA path is settled.

Brian Reed: The offering is the tell. Because — hang on — the 150% single-day surge happened during market hours. SELLAS waited until after-hours to announce the $150 million raise. J.P. Morgan and TD Cowen running the book. And shares fell roughly 10% the moment that hit. So retail investors who bought in on the REGAL headline absorbed the dilution.

Eliza Ward: Ten percent down. Same day.

Brian Reed: And there's a $22.5 million overallotment option on top of the $150 million. So the question isn't whether they needed capital — they probably do, FDA review, manufacturing scale-up, that's real. The question is the timing. If you are genuinely confident this 126-patient trial satisfies FDA registration standards as a standalone package, do you dilute shareholders the same afternoon?

Eliza Ward: I mean — that's the implied bet, right? The offering says: we think there's a non-trivial chance FDA asks for more. A confirmatory trial, post-marketing conditions, something. And that takes runway we don't currently have.

Brian Reed: Fast Track designation helps — GPS got that back in 2018 for AML. But Fast Track expedites review. It doesn't change what the FDA actually requires to approve.

Eliza Ward: Right. And the BLA — SELLAS is targeting H1 2027 to file. That's the actual clock. But the FDA hasn't seen the full data yet. The changed analysis plan, the BAT control arm outperforming assumptions — those are exactly the flags a reviewer flags.

Brian Reed: So what do we actually watch for? Like concretely.

Eliza Ward: Full data release — the confidence interval around 0.59 and the subgroup breakdowns. If the CI is wide, the p=0.035 starts looking thin. That's the first gate. Then whether FDA's pre-BLA meeting — which they'll have before the H1 2027 filing — produces any signal about whether a single trial of 126 patients clears the bar without confirmatory conditions attached.

Brian Reed: And patients in second remission right now can't wait for 2027. That's the part that doesn't fit cleanly into any of this — the data gap and the patient timeline aren't running on the same clock.

Eliza Ward: And that's — I mean, that's the thing that keeps landing for me. The same REGAL data set gets written up two completely different ways depending on what comes next. If GPS wins FDA approval and becomes standard maintenance for transplant-ineligible patients in second complete remission, this trial is the first step of a longer validation story. If the full data disappoint or the FDA asks for a confirmatory trial — same 126 patients, same p=0.035, same hazard ratio of 0.59 — it gets remembered as a narrowly positive signal that the company moved on too fast.

Brian Reed: And we won't know which story it is for months.

Eliza Ward: At minimum. The BLA isn't even filed yet. And the thing that would actually settle my read — not the approval itself, the confidence interval. When the full publication comes out and I can finally see the band around 0.59, that's when I know if this result is tight or if it's sitting on a wide range that just happened to clear 0.05.

Brian Reed: So for you that's the actual gate — not the BLA, not the FDA decision, but whether the CI, when it finally publishes, holds up under a real look. I'm watching the same number, but the question I can't answer is simpler than that. What does a hematologist at MSK tell a second-remission patient today? Not in 2027 — today. Because the announcement happened, the stock happened, the offering happened, and none of that changes what's available to that patient right now.

Eliza Ward: No. And that gap — between November 17th and whenever a BLA actually clears — that's not a gap the REGAL data closes. It's genuinely still open.