Galinpepimut-S (GPS) is not based on an unexplored biological target. WT1 is one of the most extensively studied leukemia-associated antigens in AML, with decades of work spanning peptide vaccines, dendritic-cell approaches, and other WT1-directed immunotherapies. WT1 is frequently expressed in AML blasts, which has made it an attractive target for antigen-directed approaches.
GPS's Phase 2 study in patients treated in first complete remission (CR1) enrolled 22 patients. Median disease-free survival from CR1 was 16.9 months, while median overall survival from diagnosis was not reached and was estimated at ≥67.6 months. The vaccine was generally well tolerated, and immune responses were observed in a subset of patients.
Those observations provide a rationale for further development. They do not establish a treatment effect — the study was small and nonrandomized.
That distinction is central to REGAL, SELLAS's pivotal Phase 3 trial of GPS.
The balance sheet is no longer the primary near-term risk
As of June 30, 2026, SELLAS reported $138.3M in cash and cash equivalents. Second-quarter R&D expense was $6.3M (up from $3.9M a year earlier, driven by manufacturing, clinical consulting, and REGAL trial costs), G&A expense was $4.4M, and net loss was $9.6M. First-half 2026 net loss was $18.0M.
SELLAS also holds a $150M ATM equity facility, established for additional financing capacity; as of its most recent disclosure the company had not sold shares under it. Any future use would carry the usual dilution risk.
The current cash position materially reduces the risk that financing constraints prevent the company from reaching the REGAL readout. It does not eliminate financing risk indefinitely, and it says nothing about whether GPS is effective.
Historical base rates provide context, not a probability estimate
Randomized Phase 3 oncology trials frequently produce smaller effects than investigators assumed when designing the study. A 2025 JNCI analysis of 111 randomized Phase 3 oncology trials found that 82 of 143 primary survival endpoints — 57.3% — produced hazard ratios weaker than the effect sizes assumed in the original power calculations. In a prespecified subset of five journals, the figure was 70.2%.
That finding is relevant to REGAL, but it should not be converted into a probability estimate for this specific trial. REGAL has its own population, comparator, endpoint, statistical assumptions, and prior evidence. The more specific caution is that GPS's earlier AML evidence was generated largely outside a randomized concurrent-control setting — the 22-patient CR1 Phase 2 study provides useful historical context, but its 16.9-month DFS is not a randomized treatment effect. REGAL is the trial that actually tests whether GPS improves overall survival relative to investigator-selected best available therapy.
REGAL is a relatively constrained clinical population
REGAL enrolled adults with AML who had achieved CR2 or CRp2 (or later remission) following salvage therapy and were not candidates for allogeneic transplantation at study entry. The trial is an open-label, randomized comparison of GPS against investigator-selected best available therapy, with overall survival as the primary endpoint.
WT1-positive disease is not a current eligibility requirement. Earlier protocol language included WT1-expression criteria, but the enrolled population was broadened — WT1 status is tracked via IHC/PCR on screening samples, but it is not a gate to enrollment. WT1 remains central to the mechanism of GPS; it should not be described as a current enrollment filter. Separately, the washout period is not three months — the protocol requires the last dose of re-induction antileukemic therapy to have occurred at least 4 weeks or 10 half-lives, whichever is shorter, before study treatment.
The trial was first posted in 2020; its recorded study start date is February 8, 2021, and it ultimately enrolled 127 patients. This is a relatively constrained and clinically difficult population — patients have already relapsed, achieved a second remission, and are not proceeding to allogeneic transplant — which makes the setting clinically relevant but also limits how directly the eventual result generalizes to other AML populations.
Where might GPS fit if the biology is confirmed?
The more interesting development question isn't only whether GPS works — it's whether CR2 is the setting in which an antigen-directed maintenance strategy has the best chance of demonstrating benefit.
CR1 maintenance
CR1 is an earlier disease setting where patients generally have experienced less cumulative treatment and, in many cases, lower residual disease burden — characteristics that could favor an immune-based maintenance strategy. The counterargument is that the landscape is more established: oral azacitidine is an approved maintenance option, supported by an overall-survival benefit, for an important population of newly diagnosed AML patients who achieve remission after intensive chemotherapy and are not proceeding to transplant. A future CR1 trial would face a more established comparator and a higher evidentiary bar than a trial in a setting without a comparable approved standard.
Post-transplant maintenance
The post-HSCT setting offers a different biological hypothesis. WT1-directed vaccination after allogeneic transplant has been studied in small early-phase experiences, including vaccination initiated during the post-transplant period; these studies show that WT1 vaccination can induce immune responses and are consistent with the possibility of maintenance use. The mechanistic rationale is attractive — vaccination could potentially augment donor-derived graft-versus-leukemia immunity during immune reconstitution — but the clinical evidence remains limited. Small, uncontrolled studies cannot establish that post-transplant vaccination is superior to other maintenance strategies or that it would produce an overall-survival benefit.
CR2, transplant-ineligible
This is the population REGAL actually tests. Its development rationale is straightforward: substantial unmet need, and no established maintenance standard in this setting comparable to CR1's oral-azacitidine option — which is why REGAL uses investigator-selected best available therapy (observation, hypomethylating agents, venetoclax, and/or low-dose cytarabine under the protocol) as its comparator. The biological question is less straightforward: if GPS efficacy depends on immune competence and relatively low disease burden, a heavily treated CR2 population could present a greater biological challenge than an earlier maintenance setting. That is a hypothesis, not a conclusion the available data support. The most defensible reading is not that CR2 is the wrong population, but that a positive or negative REGAL result may itself provide information about how disease setting and immune state interact with WT1-directed vaccination.
Figure 1
How AML treatments act on the biology
Five broad treatment approaches, not a linear sequence — each acts on a different vulnerability in the leukemic cell or its environment.
Figure 2
General AML treatment algorithm
Simplified clinical pathway from diagnosis to maintenance, with REGAL's enrolled population marked at the point it actually occupies.
The read
SELLAS's current cash position materially reduces the risk that liquidity constraints prevent the company from reaching the REGAL readout: $138.3M in cash and cash equivalents at June 30, 2026, alongside a $150M ATM facility providing additional financing capacity if needed.
The biological rationale for targeting WT1 is extensive, and the earlier GPS studies provide a basis for continued investigation. But those data do not establish efficacy in a randomized setting — that is the distinction this whole piece rests on.
REGAL is a randomized Phase 3 test of whether GPS can improve overall survival relative to investigator-selected best available therapy in transplant-ineligible AML patients in CR2/CRp2 or later remission. The historical Phase 3 literature is a reason for caution, not a verdict: favorable earlier-stage data do not guarantee that the effect assumed at trial design will be reproduced in the final randomized analysis. The 80-event milestone matters because it brings the trial to its prespecified final analysis — not because the event count itself provides evidence about the outcome.
If REGAL is positive, it would provide the first randomized evidence supporting GPS in this specific maintenance setting. If it is negative, the result would not by itself establish that WT1-directed vaccination has no potential in other AML settings — but it also would not justify automatically attributing the outcome to immune depletion or to the CR2 population specifically. A negative result could reflect the target, the vaccine construct, treatment schedule, patient selection, disease setting, comparator performance, or simply an insufficient effect size to move overall survival.
For now, the most defensible view is straightforward: the financing question has become less important. The clinical efficacy question remains unresolved.
This piece reflects clinical and financial analysis based on public filings, trial records, and published literature available 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.
Sources:
SELLAS Q2 2026 financial results
ClinicalTrials.gov — REGAL / NCT04229979
Expected vs. observed effect sizes in Phase 3 oncology trials, JNCI
REGAL trial design, Future Oncology
GPS Phase 2 in CR1 AML, published trial
WT1 vaccine post-HSCT maintenance, Blood Cancer Journal
ASH 2026 abstract submission guidelines