Expanded safety analysis of pevonedistat, a first-in-class NEDD8-activating enzyme inhibitor, in patients with acute myeloid leukemia and myelodysplastic syndromes

Dear Sir, We previously reported the therapeutic potential of pevonedistat (TAK-924/MLN4924), a novel inhibitor of protein neddylation, in patients with acute myeloid leukemia (AML). Pevonedistat is a small-molecule inhibitor of the neural cell developmentally downregulated 8 (NEDD8)-activating enzyme, which processes NEDD8 for binding to target substrates. The most characterized NEDD8 target in cells are the Cullin-RING E3 ubiquitin ligases (CRLs), which direct the degradation of specific substrates through the proteasome (for example, p27, CDT1, Nrf-2). In the absence of NEDD8, CRL substrates accumulate causing antiproliferative effects in AML. In a phase I trial (C15003 study), patients with relapsed/refractory myelodysplastic syndrome or AML were treated with 1-h infusions of pevonedistat every 21 or 28 days across five schedules. On Schedule A, dose escalation commenced at 25 mg/m on Days 1, 3 and 5 in a standard ‘3+3’ manner to determine the maimum tolerated dose (MTD). Following this, four additional schedules were evaluated with pevonedistat: Schedule B (Days 1, 4, 8 and 11 starting at 147 mg/m); Schedule C (Days 1, 8 and 15 starting one dose level lower than the MTD determined for Schedule A (44 mg/m)); Schedule D (Days 1, 3 and 5 stating at 45 mg/m to be combined with azacitidine 75 mg/m on Days 1–5 and 8–9); and Schedule E (Days 1, 3 and 5 at fixed dose of 50 mg/m). Schedules A–C and E were administered every 21 days, whereas schedule D was administered every 28 days. Adverse events (AEs) on Schedules A and B were taken into account when designing Schedules C–E; based on Schedule A MTD, an expansion dose of 50 mg/m was selected for Schedule E. Results for Schedules A and B were published by Swords et al., and the reported MTDs were 59 and 83 mg/m, respectively. Hepatotoxicity was dose limiting for Schedule A and multiorgan failure (MOF) for Schedule B. The objective response rate in patients treated at or below the MTD was 17% for Schedule A and 10% for Schedule B, confirming single agent activity in refractory AML patients. Considering the promising safety and activity of pevonedistat reported thus far, we carried out a comprehensive safety analysis of all 72 patients treated across five schedules (A–E) in the C15003 study. Seventy-two patients were evaluable for toxicity: Schedule A (n= 27), Schedule B (n= 26), Schedule C (n= 2), Schedule D (n= 1), and Schedule E (n= 16). Most patients had AML (N= 66, 92%), 5 had myelodysplastic syndrome (7%) and 1 patient had ALL (1%). Median age was 65.5 years (range: 19–84). The most frequently reported all-grade AEs were: pyrexia (49%), diarrhea (43%), decreased appetite (35%), peripheral edema (33%), dyspnea and fatigue (each 32%), febrile neutropenia (31%) and nausea (31%). These were generally mild and transient. The most common reported laboratory toxicities were elevations in aspartate aminotransferase (25%) and alanine transaminase (24%). Treatment-related myelosuppression was infrequent with thrombocytopenia reported in 18% of patients. Fifteen (21%) patients discontinued the study owing to AEs (Table 1). Overall, 56 (78%) patients experienced at least 1 serious adverse event (SAE): Schedule A= 20 (74%), Schedule B = 20 (77%), Schedule C = 2 (100%), Schedule D= 1 (100%), and Schedule E = 13 (81%). The most common SAE overall was febrile neutropenia reported in 17 (24%) patients (A = 8, B = 6 and E= 3). Other SAEs occurring in 42 patients were pneumonia (n= 12, 17%), disease progression (n= 9, 13%), fever (n= 6, 8%), sepsis (n= 6, 8%), intracranial hemorrhage (n= 5, 7%), hypotension (n= 4, 6%), gastrointestinal hemorrhage (n= 3, 4%) and mental status changes (n= 3, 4%). There were 28 (39%) on-study deaths and 4 of these deaths were due to sepsis (1 patient on Schedule A at 78 mg/m), MOF (2 patients on Schedule B at 110 mg/m and 147 mg/m) and cardiopulmonary failure related to leukastasis (1 patient on Schedule D at 45 mg/m). Eighteen deaths, including three of the four deaths due to MOF/ cardiopulmonary failure occurred during Cycle 1. The on-study deaths unrelated to pevonedistat (n= 24) occurred owing to underlying disease (n= 11), sepsis (n= 3), intracranial hemorrhage (n= 5), pneumonia (n= 3), and 1 case each of pulmonary hemorrhage and pancytopenia. Overall, there were 10 dose-limiting toxicities (DLTs) in 6 patients: transaminase elevation (Schedule A, 78 mg/m); orthostatic hypotension (Schedule B, 110 mg/m); cardiac failure (Schedule B, 147 mg/m); MOF (Schedule B, 147 mg/m); rash (Schedule E, 50 mg/m); and elevated alanine transaminase (Schedule E, 50 mg/m). With the exception of 1 DLT, all events occurred within 2 days of first exposure to pevonedistat. There were no DLTs at doses o50 mg/m, and all Grade 4 DLTs occurred at the highest dose of pevonedistat administered (147 mg/m). Mean pevonedistat plasma concentration–time profiles obtained on Day 1 of Cycle 1 (and Cycle 3 on Schedule E) are shown in Figure 1. No obvious differences in exposures were observed between Cycles 1 and 3, suggesting that pevonedistat PK remains stationary over time. Of the 55 response evaluable patients, there were 2 complete remissions (4%) on Schedule A, and 5 partial remissions (9%) across Schedule A (n= 2), Schedule B (n= 2) and Schedule E (n= 1). The overall complete remission/ partial remission rate was 13%. Thirty-four patients (62%) had stable disease after treatment with ⩾ 4 courses of pevonedistat (Schedule A = 14, Schedule B = 13, Schedule E = 7). All responses were in patients with AML. The principal serious and potentially drug-related toxicities noted in this trial were hepatotoxicity and sepsis syndromes with MOF. The exact mechanisms accounting for the hepatotoxicity observed in some patients treated with pevonedistat is unclear. In vitro RNA interference to disrupt NEDD8 conjugation induces shape changes in HEK293 and HeLa cells (edema and swelling). This effect may relate to the accumulation of the cullin-dependent GTPase RhoA, which regulates cytoskeleton proteins. This effect is reproducible with pevonedistat treatment in vitro and is reversible by RhoA knockdown. It is interesting to speculate that transient leakage of transaminases from edematous hepatocytes in treated patients may have accounted for some of the laboratory toxicities observed in this study. A more compelling perspective on pevonedistat-induced liver toxicity was presented by Citation: Blood Cancer Journal (2017) 7, e520; doi:10.1038/bcj.2017.1