Abstract
Allogeneic hematopoietic stem cell transplantation (HCT), defined by its underlying graft-versus-multiple myeloma (MM) effect, is a potentially curative approach for high-risk pts. However, relapse remains the main cause of treatment failure. Predictors for post-relapse survival after HCT are not well characterized, and we hypothesized that the time to relapse and the patterns of relapse are prognostic for post-relapse survival.
Methods: All pts with MM who underwent a CD34+-selected alloHCT at Memorial Sloan Kettering Cancer Center on protocol (NCT 01131169/01119066) or off study from 01/2010 to 12/2017 and progressed after alloHCT were included. Pts were conditioned with busulfan (0.8 mg/kg x 10), melphalan (70 mg/m2 × 2), and fludarabine (25mg/m2 × 5). The K-M method was used to estimate OS post-HCT relapse. Significant predictors were considered in a Cox model.
Results: Of the 115 transplanted pts, 60 (52%) relapsed and were analyzed. The median age was 56 yrs (35- 66), 62% male and 82% with high-risk cytogenetics (CG). KPS was ≥ 80 in 100% and ≥ 90 in 52%. Pts received a median of 4 lines of therapy pre-HCT (1-10), with 88% achieving at least a partial response (PR) prior to alloHCT. All pts had received at least one autoHCT. Of the 38% who received a planned or preemptive post-HCT therapy, 13 received DLI and 10 other strategies. Relapse was characterized as very early (<6 mo, 28%), early (6-24 mo, 50%), or late (>24 mo, 22%). At relapse, 27% presented with extramedullary disease (EMD). Median post-relapse OS was significantly different for pts who relapsed very early, early, or late; 4 mo, 17 mo, and 72 mo, respectively (p = 0.002). Age ≥55, relapse with EMD, lower KPS, <PR prior to HCT, <PR by day 100, and no maintenance were also prognostic for worse post-relapse OS in univariate analysis. There was no significant difference for post-relapse OS for number of lines, high-risk CG, or era of alloHCT. In multivariate analysis adjusting for sex and age, risk factors for worse post-relapse OS included very early (HR 4.4, 95%CI 1.4-13.5) or early relapse (HR 2.5, 95%CI 1.4-13.5), relapse with EMD (HR 5.2, 95%CI 2.1-12.9), and lack of DLI as planned post-HCT therapy (HR for DLI compared with no DLI = 0.11, 95%CI 0.01-0.9).
Conclusion: For this very high-risk group of MM pts, differences in time to relapse after alloHCT define very distinct post-relapse survival patterns. Very early relapses were associated with greater chemoresistance and dismal prognosis. In contrast, the median post-relapse OS for late relapses was 6 yrs, suggesting different underlying disease biology and/or host/donor characteristics. Similar to the post-auto setting, relapse with EMD conferred poor prognosis. Of note, post-relapse OS was superior for pts who received a DLI as planned post-HCT therapy prior to relapse. Efforts to characterize optimal post-relapse therapeutic strategies and mechanism(s) driving relapses are ongoing.