Ipilimumab plus decitabine for patients with MDS or AML in posttransplant or transplant-naïve settings

TO THE EDITOR:

CTLA-4 blockade has generated long-lasting remissions for subsets of cancer patients for whom traditional chemotherapies have failed. We have previously demonstrated clinical activity with ipilimumab (IPI) monotherapy at a 10 mg/kg dose in the treatment of relapse after allogeneic hematopoietic stem cell transplantation (HSCT) (including leukemia cutis) without excessive graft-versus-host disease (GVHD) induction.^1,2 Epigenetic modifiers enhance expression of CTLA-4, CD80, and tumor antigen expression on leukemia cells and can induce cytotoxic CD8⁺ T-cell activation.^3-5 We set out to determine if combining decitabine (DEC) with IPI could augment responses without causing unacceptable immune toxicity. Here, we report the results of a multicenter phase 1 trial with combination IPI + DEC in patients with relapsed, refractory (R/R), or secondary myelodysplastic syndromes (MDS)/acute myeloid leukemia (AML) in both the post-HSCT and transplant-naïve settings (NCT02890329; CTEP 10026).

This phase 1 trial used a 3 + 3 dose-escalation design and was approved by central and local institutional review boards. All patients provided written informed consent. Eligible patients with morphologic relapse, refractory, or untreated secondary MDS/AML were stratified by prior transplant status into post-HSCT (arm A) or transplant-naïve (arm B) arms. Both arms received a lead-in priming cycle of DEC (cycle 0), and in the absence of GVHD or disease progression, patients received combination IPI + DEC in 28-day cycles for 1 year (supplemental Figure 1, available on the Blood website). IV DEC at 20 mg/m² was administered on days 1 through 5. The 3 IV IPI dose-escalation cohorts included 3, 5, and 10 mg/kg on day 1 of each “ipilimumab induction” cycle (cycles 1-4) and every other “ipilimumab maintenance” cycle (cycles 5-12). IPI infusion was withheld for immune-related adverse events (irAEs) and resumed only if the irAE resolved within 8 weeks. Dose-limiting toxicities (DLTs) were assessed for 8 weeks after the first IPI dose and defined as being any of the following: treatment-related death; ≥grade 3 acute GVHD; ≥grade 3 nonhematologic toxicity; or grade 4 hematologic toxicity without recovery. Overall response rate (ORR) included complete remission (CR) and CR with incomplete count recovery for AML, and CR and marrow CR with or without hematologic improvement for MDS.^6,7 

From September 2017 to August 2021, a total of 54 patients enrolled, including 6 who withdrew (n = 3) or progressed (n = 3) during cycle 0. Ultimately, 48 patients received combination IPI + DEC, including 25 post-HSCT patients (23 AML and 2 MDS; all R/R) and 23 transplant-naïve patients (15 AML and 8 MDS; including 20 R/R and 3 previously untreated), and are included in the following safety and efficacy analyses (supplemental Figure 2; supplemental Tables 1-3).

The most frequent grade ≥3 treatment-emergent adverse events, in the post-HSCT and transplant-naïve arms, respectively, were neutropenia, 32% and 48%; thrombocytopenia, 28% and 48%; and febrile neutropenia, 36% and 61% (supplemental Table 4A). The overall irAE rate was 44% (11 of 25) in the post-HSCT and 48% (11 of 23) in the transplant-naïve settings (Figure 1A; supplemental Tables 4B-C). Most irAEs occurred at an IPI dose of 10 mg/kg (7 of 11 [63%] post-HSCT patients; 9 of 11 [82%] transplant-naïve patients) and were less frequent at lower doses (n = 2 per arm at an IPI dose of 3 mg/kg; n = 2 in the post-HSCT arm at an IPI dose of 5 mg/kg). Steroid-responsive chronic GVHD developed at each tested IPI dose level in the post-HSCT arm, including 1 severe and 3 moderate cases. In the transplant-naïve arm, the most common irAEs were dermatitis (n = 6) and colitis (n = 3). A median of 3 IPI doses (range: 1-8) and 4 treatment cycles were received in both arms (range: 1-13). No DLTs were observed at IPI dose levels of 3 and 5 mg/kg. Three DLTs occurred at an IPI dose of 10 mg/kg, including 2 toxic deaths in post-HSCT patients receiving immune suppression (1 case of late-onset grade 3 acute GVHD of the colon/liver, and 1 case of pneumonitis), and 1 case of steroid-refractory immune-mediated grade 4 thrombocytopenia without bleeding in a transplant-naïve patient. In total, 31 patients were treated at the IPI dose level of 10 mg/kg, including 15 post-HSCT and 16 transplant-naïve patients. In combination with DEC, the maximum tolerated dose/recommended phase 2 dose (MTD/RP2D) of IPI was determined to be 10 mg/kg for both arms.

Objective responses were achieved more frequently in the transplant-naïve group, which had a 52% ORR (n = 12 of 23), compared with the post-HSCT group, which had a 20% ORR (n = 5 of 25; P = .034; Figure 1B). Subsequently, 1 transplant-naïve patient was bridged to HSCT, and 2 post-HSCT patients underwent a second transplant. Although responses occurred at each IPI dose level, most occurred at 10 mg/kg (3 of 11 patients at 3 mg/kg, 2 of 6 patients at 5 mg/kg, and 12 of 31 patients at 10 mg/kg). Of the 6 cases of myeloid sarcoma without morphologic marrow involvement, 3 achieved CR. We identified 2 historical cohorts (2016-2022), including 46 patients with post-HSCT morphologic relapse and 44 with previously untreated AML, who were treated with single-agent DEC at our institution. Although no difference was detected in the post-HSCT relapse setting (P = .5), a higher response rate was observed with IPI + DEC compared to single-agent DEC in the transplant-naïve setting (P = .019).

The median follow-up for patients in the study was 9.7 months (range: 4.4-35.4) in the post-HSCT arm, and 14.3 months (range: 5.1-25.0) in the transplant-naïve arm. No statistical difference occurred in overall survival (P = .38) or progression-free survival (P = .36) between the transplant-naïve and the post-HSCT arms (Figure 1C-D). Univariate analyses did not reveal any significant association of response with history of prior GVHD, presence of a TP53 mutation, blast burden, disease histology (AML vs MDS), or prior hypomethylating agent–therapy exposure.

The median duration of response was 4.46 months (range: 1.0-40.7) and 6.14 months (range: 1.2-16.9) in patients with and without prior HSCT, respectively (supplemental Figure 3). Durable remission >1 year with concomitant irAE development occurred in 3 transplant-naïve patients treated at an IPI dose of 10 mg/kg and 1 post-HSCT patient (number 1006) treated at an IPI dose of 3 mg/kg. Patient 1006 achieved CR and developed chronic GVHD after one IPI + DEC cycle and continues to maintain a durable CR >3.5 years later. We thus asked if irAE development was associated with outcome (supplemental Figure 4). No differences were observed in median overall survival in either post-HSCT or transplant-naïve patients based on irAE occurrence (Figure 2A). However, among the transplant-naïve patients, 1-year overall survival was significantly longer in those with irAE development than in those without (72.7% vs 33.3%; P = .039), suggesting that IPI + DEC–induced immune activation was associated with survival benefit. However, to significantly lengthen progression-free survival among transplant-naïve patients, subsequent consolidation with transplant is still necessary when possible.

Despite striking examples of individual responders with increased cytotoxic T-cell infiltrate (CD3⁺CD8⁺granzyme B⁺) using multiplexed immunofluorescence staining on serial bone marrow biopsies and globally increased CD3⁺ density after IPI + DEC administration, no distinct pattern of local T-cell infiltration pre- or posttreatment was associated with response, highlighting underlying tumor and immune heterogeneity (Figure 2B; supplemental Figures 5 and 6; supplemental Tables 5 and 6). We further evaluated systemic effects of treatment on circulating T-cell subsets (supplemental Table 7; supplemental Figures 7-9). Following IPI infusion, we observed upregulation of the inducible costimulator (ICOS) molecule on CD4⁺ T cells (mean 13.3% vs 25.7%, P = .021) and on CD8⁺ T cells (mean 3.5% vs 7%, P = .011), both independent of clinical response (Figure 2C; supplemental Figure 9D).

Our prior IPI monotherapy study demonstrated that responses were associated with decreased activation of regulatory T cells.¹ Here, we detected a global increase in regulatory T cells following IPI + DEC treatment (mean 4.2% vs 10.8%; P = .04; Figure 2D), which has been described previously after hypomethylating-agent maintenance treatment in the HSCT setting^8-10 and may have been induced as a compensatory mechanism to immune activation.

In summary, combination IPI + DEC treatment has an acceptable safety profile and has meaningful clinical activity in patients with R/R MDS/AML that does not appear to require T cell–mediated alloreactivity. IPI + DEC treatment may serve as a less-intensive bridge to transplant among potential transplant candidates. Future studies are warranted to identify rational IPI-based treatment strategies to generate prolonged responses without severe immune toxicity.