PD-L1 Blockade with Atezolizumab in Higher-Risk Myelodysplastic Syndrome: An Initial Safety and Efficacy Analysis

Introduction: Patients (pts) with higher-risk myelodysplastic syndrome (HR-MDS) and those who fail to respond to or relapse/progress after treatment with hypomethylating agents (HMA) have limited therapeutic options and poor prognosis. PD-L1 expression is upregulated in HR-MDS pts (compared with lower-risk MDS pts) and in those who fail HMA therapy. Combining inhibition of the PD-L1/PD-1 pathway with azacitidine may improve outcomes in MDS.

Methods: We conducted a Phase Ib trial of the anti-PD-L1 monoclonal antibody atezolizumab, with or without azacitidine, in HMA-failure and HMA-naive MDS pts (NCT02508870). The primary objective was to determine the safety and tolerability of atezolizumab as a single agent or in combination with azacitidine. An initial safety evaluation was performed in three cohorts. Cohort A1 (10 pts) consisted of HMA-failure HR-MDS pts treated with atezolizumab alone (1200mg IV q3w). Cohort B1 (11 pts) consisted of HMA-failure HR-MDS pts treated with atezolizumab (840mg IV q2w) in combination with azacitidine (75mg/m² qd for 7 days q4w) for 6 cycles, followed by atezolizumab maintenance alone (1200mg IV q3w). Cohort C1 (6 pts) consisted of HMA-naive HR-MDS pts treated with atezolizumab (840mg IV q2w) in combination with azacitidine (75mg/m² qd for 7 days q4w). If atezolizumab alone or in combination with azacitidine was deemed safe and tolerable in Cohorts A1 and B1, an additional 1:1 randomization into two cohorts of 30 pts each (Cohorts A2 and B2) was planned. If the combination of atezolizumab and azacitidine was found to be safe and tolerable in Cohort C1, an additional expansion cohort (Cohort C2) of 14 pts with HMA-naive HR-MDS was planned. Primary endpoints included determining the safety and tolerability of atezolizumab-based regimens in HR-MDS and defining the recommended Phase II dose for the combination.

Results: As of January 2018, 42 HR-MDS pts had been treated with atezolizumab-based regimens: Cohort A, 10 pts; Cohort B, 11 pts; Cohort C, 21 pts. Median age for the entire pt cohort was 76 years (range: 63−89). Median treatment duration for Cohorts A, B, and C was 4.2, 5.5, and 5.8 months, respectively. The overall response rate for Cohorts A, B, and C was 0%, 9% (hematologic improvement [HI]: 9%), and 62% (CR, 14%; mCR, 19%; mCR + HI, 10%; HI, 19%), respectively. All pts in Cohorts A and B have discontinued therapy, with a median overall survival (OS) of 5.9 months and 10.7 months, respectively. For pts in Cohort C, 8/21 pts remain on therapy, and median OS has not been reached. Grade 3−5 adverse events (AEs) in >10% of pts were primarily hematologic; grade 3−5 febrile neutropenia occurred in 29% of all pts and was particularly common in pts receiving the atezolizumab-azacitidine combination (Cohorts B [36%] and C [33%] compared with Cohort A [10%]; Table 1). In Cohort A, 70% (7/10) of pts died, as did 64% (7/11) of pts in Cohort B and 29% (6/21) of pts in Cohort C. Median time to death was 160 days (Cohort A), 299 days (Cohort B), and 53 days (Cohort C). Timing and causes of death were different in the three cohorts. Causes of death were more commonly from disease progression in Cohorts A and B, while serious AEs accounted for all deaths in Cohort C (Table 2). In addition, deaths within 3 months occurred in 10%, 18%, and 29% of pts in Cohorts A, B, and C, respectively. The high early death rate compared with historical controls observed in HMA-naive HR-MDS patients (Cohort C) led to early termination of the study prior to completing recruitment. Biomarker assessment demonstrated PD-L1 expression on variable proportions of AML blasts in samples from all pts analyzed. However, PD-L1 expression was not associated with clinical response (Figure).

Conclusions: Combination of atezolizumab plus azacitidine in HMA-naive HR-MDS pts had an unfavorable safety profile, which led to early termination. Limited responses were observed with atezo-based regimens (with or without azacitidine) in HMA-failure HR-MDS pts, without excessive or unexpected toxicity. Better understanding of the reasons associated with the differential toxicity profile observed between HMA-naive versus HMA-failure HR-MDS pts will be crucial for potential future developments of this combination.

View largeDownload slide

View largeDownload slide

Close modal