Prognostic Factors of Response and Overall Survival (OS) in Higher-Risk MDS (including RAEB-t) Treated with Azacytidine (AZA): Results of the French ATU Program.

Abstract

Abstract 3820Poster Board III-756 BackgroundAZA provides 50-60% responses and improves OS in higher-risk MDS (Lancet Oncol 2009) but prognostic factors of response and OS remain largely unknown. MethodsAn AZA compassionate program (ATU) was opened in France between Dec 2004 and Dec 2008 for higher risk MDS, and for AML not candidates or refractory to intensive chemotherapy (IC). We retrospectively analyzed the outcome of higher risk MDS (including RAEB-t) patients (pts) included in this program in the 42 centers with complete patient (pt) reporting, having received ≥ 1 cycle of AZA, and excluding those who had received prior allo SCT, IC, or a hypomethylating agent. Results233 pts (M/F: 145/88) with a median age of 71y (range 20-91) were included. Diagnosis (WHO classification) was: RA±RS: 2, RCMD±RS: 7, RAEB-1: 48, RAEB-2: 123, and RAEB-t: 53; IPSS cytogenetic risk was favorable (fav) in 75 (32%), int (intermediate) in 35 (15%), and unfavorable (unfav) in 112 (48%) (failure in 11, 5%); IPSS was: int-2 in 133 (57%), high in 97 (42%), NA in 3 (1%). Median time since diagnosis was 5 months (range 0–209). 29 patients (12%) had previously received low dose Arac (LD AraC). 162 pts (70%) were RBC transfusion dependent, including 49% requiring ≥ 4 RBC units/8 weeks. Median follow-up was 13 months. Patients received a median of 5 cycles (range 1-26) of AZA, at FDA/EMEA-approved schedule (75 mg/m2/d x7d /4 week) in 70% patients and a less intensive schedule (5d/4w, or <75 mg/m2/d) in 30% patients, while 16% received concomitant valproic acid (VPA). First response was evaluated after 3-4 cycles. Best response (IWG 2006 criteria) was CR in 22 pts (9%), PR in 6 (3%), marrow CR (mCR) in 25 (11%), stable disease (SD) with HI in 34 (15% including HI-E ± HI-N/HI-P in 13, HI-P±HI-N in 17, and isolated HI-N in 4) leading to an overall response rate of 38%. Moreover, SD without HI was observed in 26%, progression in 26%, death before 4 cycles in 7% and AZA discontinuation before 4 cycles in 3%. Overall, HI-E was obtained in 48/162 anemic pts (23%). 38% of the responders progressed after a median time of 13 months (range 3-26), and 62% remained responders after a median time of 11 months (range 1-35). Median response duration was 10.2 months in CR pts, not reached in PR pts, 13 months in mCR pts, and 25 months in pts who achieved SD with HI. No pre-treatment characteristic including age, sex, time since diagnosis, WHO, karyotype, importance of cytopenias, BM or PB blast %, IPSS, RBC transfusion dependency, nor VPA addition was predictive of response, but pts with prior LD-AraC had a significant lower reponse rate (24% vs 50% in LD-AraC naive pts, p=0.01). Considering specific cytogenetic abnormalities, response rates were 42% in del5q/-5 (n=59), 38% in del7q/-7 (n=65), 38% in +8 (n=26), 39% in complex karyotype (n=67) and 49% in normal karyotype (n=57) (p=NS). 1 and 2-year OS were 57.9% and 29.7%, respectively, and median OS was 13.7 months. By univariate analysis, pre-treatment parameters negatively influencing OS were transfusion dependency, presence of PB blasts, cytogenetic risk (IPSS), and IPSS high risk. By multivariate analysis, OS was independently influenced by baseline transfusion dependency (≥4 RBC units/8 weeks; 1y-OS 57.5% vs 71.5%, p=0.01), presence of PB blasts (1y-OS 44.6 % vs 74.4%, p=0.02) and cytogenetics (1 y OS 78.2%, 55.8% and 46.4% for fav, int, and unfav, respectively (p=0.009)). In a landmark analysis of OS from first evaluation date, pts with CR, PR or SD with HI-E had improved OS compared to those with marrow CR or SD without HI-E (1-y OS: 71.5% vs 49.6%, p=0.03) and to those who progressed (1y OS=24.4% p=0.002). Overall Survival was significantly longer in pts who achieved HI-E (n=48; 1-y OS from evaluation: 65.4% vs 35.4%, p=0.003), but not in pts achieving HI-N (n=33, p=0.2) or marrow response (ie< 5% blasts with >50% decrease; n=47; p=NS), while reaching HI-P had a borderline favorable effect on OS (n=51, p=0.07). ConclusionNo pre-treatment parameter (apart from prior LD-AraC treatment) predicted response to AZA. Regarding OS, presence of PB blasts, intermediate or unfavorable cytogenetics and transfusion dependency were associated with poorer survival. Improvement of anemia (and to a lesser extent of platelets) rather than BM blast reduction were associated with survival improvement. Whether other tests, including methylation profiles, may provide further information predicting response to AZA is being evaluated. Disclosures:Fenaux:Celgene: Honoraria, Research Funding; Roche: Honoraria, Research Funding; Ortho Biotech: Honoraria, Research Funding; Amgen: Honoraria, Research Funding; Cephalon: Honoraria, Research Funding; Merck: Honoraria, Research Funding; Novartis: Honoraria, Research Funding.