Efficacy and Safety of Pembrolizumab Added to Azacitidine Plus Venetoclax for Patients with Acute Myeloid Leukemia: Results from an Investigator-Initiated, Multi-Center, CTEP-Sponsored Randomized, Phase II Trial (BLAST AML-2)

AMZ, RMS Contributed equally and serve as co-senior authors

Background

AZA + VEN is a standard of care treatment for pts with AML who are unfit for intensive chemotherapy. However, long-term survival remains limited. Addition of anti-PD-1 antibody to AZA+VEN may activate T-cells against blasts, induce higher rates of measurable residual disease (MRD)-negativity (neg), and improve outcomes. We report results of a randomized, phase II trial of AZA+VEN +/- PD-1 inhibition for pts with AML ineligible for intensive induction (NCI10334, NCT04284787).

Methods

BLAST AML-2 is an investigator-initiated, multi-center, CTEP-sponsored, open-label, randomized phase II trial. We randomized (1:1) pts aged ≥60 years with non-core binding factor, FLT3 wild-type, newly diagnosed AML who were ineligible for/declined intensive induction to receive AZA (75 mg/m2 IV/SQ x7 days [D], D1-7 or D5-2-2) + VEN (400 mg/d, D1-28) or AZA+VEN and pembrolizumab (PEM; 200 mg IV, D8 of cycle [C] 1, then every 3 weeks). VEN was adjusted at investigator discretion. The primary endpoint was complete remission (CR)/CR with incomplete count recovery (CRi) with MRD-neg assessed by centralized flow cytometric analysis at 0.1% threshold within the first 6 cycles. Disease and MRD status were evaluated after cycles 2, 4, 6, then every 3 months (mo). Multiplex immunofluorescence, whole exome sequencing, RNA/T-cell receptor sequencing, CyTOF, and Olink cytokine analyses are planned in collaboration with the Cancer Immune Monitoring and Analysis Centers (CIMACs). This protocol was IRB approved, all patients provided informed consent, and the trial was conducted according to the Declaration of Helsinki.

Results

During a planned safety run-in period, the first 6 pts who received PEM had no dose limiting toxicities. 60 pts were then randomized (31 to AZA+VEN, 29 to AZA+VEN+PEM) between 2/16/21-6/5/23. 2 pts in control arm withdrew consent prior to receiving study treatment and were not included in analysis. For this data cut, median follow up was 12.2 months. Median age was 73 (range: 62-82) years; 38 (66%) were male. The majority of pts were white (48, 83%), 5 (9%) were African-American. Most pts had intermediate cytogenetic risk (N=42, 72%) and ELN 2022 adverse risk (N=39, 67%). There were no significant differences in relevant baseline patient- or disease-specific characteristics. Both arms received a median of 3 cycles of AZA+VEN (interquartile range: 2-7). 2 pts in each arm did not complete C1 (withdrew consent [N=3], suicide [N=1]).

Pts who received any treatment dose on trial were evaluable for safety and efficacy analyses (N=58). The most frequent treatment-related AEs (TRAEs) (≥10%) were hematologic (89.5% vs 79.3% of pts on AZA+VEN and AZA+VEN+PEM, respectively) and gastrointestinal (62.1% vs 65.6%). Five pts on PEM had G≥3 transaminase elevations (2 required permanent discontinuation) vs 1 pt on AZA+VEN (resolved). Six pts died during treatment; 2 deaths on AZA+VEN attributed to sepsis, and 4 deaths on AZA+VEN+PEM with one death attributed to PEM (PEM-related respiratory failure [N=1], suicide [N=1], sepsis [N=1], and disease progression [N=1]).

A pre-planned interim futility analysis led to halting enrollment and discontinuing PEM amongst pts on AZA+VEN+PEM. In an intention-to-treat analysis of pts who received >1 dose of AZA+VEN, 22/29 (76%) pts on the AZA+VEN arm and 20/29 (69%) on AZA+VEN+PEM arm achieved CR/CRi within 6 cycles (p=0.528). Median time to best response was also similar (58 vs 56 days in AZA+VEN and PEM arms, respectively). The rate of MRD-neg CR/CRi (the primary endpoint) was not different between the 2 arms (13 pts patients in each arm; p=1.000). Five pts on AZA+VEN proceeded to allogeneic stem cell transplant vs 4 in the PEM group. At data cutoff, 10 pts in each group had relapsed. By 7/8/24, there were 37 deaths (16 pts on AZA+VEN, 21 on AZA+VEN+PEM). Pts on AZA+VEN+PEM had a trend toward worse OS compared to AZA+VEN (8.9 vs 17.9 months, p = 0.101). PFS also trended in favor of the control arm (7.0 vs 14.8 months, p = 0.127). Correlative studies are in progress.

Conclusions

To our knowledge, this is the first randomized study of an anti-PD1 antibody added to AZA+VEN in newly diagnosed pts with AML who are ineligible for/refused induction therapy. The addition of PEM to AZA+VEN did not improve MRD-neg CR/CRi and was associated with a trend for worse outcomes. Immune and correlative analyses are being conducted to understand these results.