A Randomized Phase II Study of Azacitidine (AZA) Alone or with Lenalidomide (LEN), Valproic Acid (VPA) or Idarubicin (IDA) in Higher-Risk MDS: Gfm's 'pick a Winner' Trial

Background:AZA improves overall survival (OS) in higher risk MDS, but only 50-60% of the patients respond, and median OS with AZA is only 20-24 months. As OS improvement is obtained at modest response rates, OS rather than response should probably remain the primary endpoint for all combinations with AZA, requiring large phase III trials with significant follow up. On the other hand, combinations that do not increase response will likely not improve OS. We therefore tested, based on a "pick the winner" approach, AZA combinations with the HDAC inhibitor VPA, LEN or IDA to identify, based on response, the most promising combination with AZA in higher risk MDS, that could be subsequently compared with AZA alone in a larger phase III study.

Methods: AZA-PLUS (#NCT01342692)was an adaptive two-stage phase II trial based on Jung design (Stat Med.2008;27:568) that randomly assigned higher-risk MDS, low blast count AML (20-30%) and CMML to: AZA (75 mg/m²/d d1-7 of 28-day cycles); AZA plus LEN (10 mg/d on d1-14); AZA plus VPA( 50 mg/kg/d on d1-7; 35 mg/kg/d in patients> 60y) or AZA plus IDA (10 mg/m²on d1 for the first 9 cycles). The primary end point was response rate (RR, including CR, PR, marrow CR, based on IWG 2006) of the combination arms vs AZA alone. Given a 30% RR with AZA alone, we considered that a ≥45% RR would make combination(s) promising. Controlling for type I and type II errors at 0.15 and 0.20, 40 patients per arm were to be enrolled at each stage. Any experimental arms with RR lower than those observed in the AZA arm at the first stage should be stopped. At the second stage, any arm with > 6 more responses than AZA alone should be selected for further testing. Secondary endpoint were ORR (RR+ stable disease with HI (HI) and OS.

Results: After inclusion of 40 pts/arm (first stage) all experimental arms had at least the same number of responses as the control arm and were continued in second stage. Overall, 322 pts were enrolled from 06/2011 to 07/2017: 81, 80, 80, 81 in the AZA, AZA+VPA, AZA+LEN and AZA+IDA arms, respectively. Baseline characteristics were well-balanced across arms. Median age was 74.6 y, 213 pts were male, IPSS was INT-2 in 54% and High in 46%. IPSS Karyotype was fav, int and poor in 40%, 26% and 34%, respectively. Pts received a median of 7 cycles and median follow-up was 15.1 months. Prevalence of trial discontinuation due to adverse events was 32%, 29%, 28% and 31% in the AZA , AZA+VPA , AZA+LEN and AZA+IDA arms, respectively (p=0.95). Rates of hospitalization during the first 6 cycles were 38%, 44.7% , 55.1%, 59.7% in the AZA, AZA +VPA, AZA+LEN and AZA+IDA arms, respectively (p=0.028), suggesting increased myelosuppression in the experimental arms, especially in the LEN and IDA arm.

In the control arm, 29 responses (CR+PR+mCR) after 6 cycles were observed, with 29, 25 and 29 responses observed in AZA+VPA , AZA+LEN and AZA+IDA arms, respectively. Thus, no combination demonstrated benefit over AZA. The RR was estimated at 34.8% (18.6% CR, 3.1% PR, and 13.0% mCR) and the ORR after 6 cycles was 40.4%. The RR after 6 cycles (35.8% for AZA, 36.2% for AZA+VPA, 31.2% for AZA+LEN, and 35.8% for AZA+IDA) and the ORR after 6 cycles (41.9% for AZA; 41.2% for AZA+VPA, 40.0% for AZA+LEN and 38.3% for AZA+IDA) were close across study arms. By multivariate analysis, factors associated with better ORR were higher Hb level (p=0.05), low fibrinogen (p=0.008) and low LDH (p=0.01). 17 (5%) pts were bridged to allogeneic SCT: 6 on AZA, 5 on AZA+VPA, none in the AZA+LEN arm and 6 on AZA+IDA arm (p=0.03).

At the reference date of July 2018, median EFS was 16.6 months for in AZA, 14.5 months for in AZA+VPA, 15.1 months for in AZA+LEN and 13.2 months for in AZA+IDA (p=0.74) (Fig A). Multivariable Cox model selected Hb level (p=0.02), presence of circulating blasts (p<0.0001), LDH (p=0.006) and high IPSS (p<0.0001) as prognostic.Median OS was 24.5 months for AZA, 18.9 months for AZA+VPA, 17.5 months for AZA+LEN and 20.1 months for AZA+IDA (p=0.50) (Fig B). Factors associated with OS were circulating blasts (p=0.003) and high IPSS (p<0.0001).

Conclusion: Although OS differences may occur with longer follow up, the combination of VPA, LEN or IDA to AZA did not improve response or OS over AZA alone and worsened myelosuppression. With newer, potentially more potent drugs that can be combined with AZA, the "pick the winner " approach may still be useful to select promising combinations based on response in phase II trials. Molecular data of the pt cohort will be presented at the meeting.

Figure.

View largeDownload slide

Figure.

View largeDownload slide

Close modal