Survival Analysis from the CALGB Study of Lenalidomide Maintenance Therapy in Newly Diagnosed Multiple Myeloma Post-Autologous Stem Cell Transplantation Adjusted for Crossover (Alliance 100104)

Background: The phase 3 double-blind, CALGB/Alliance 100104 study assessed the efficacy and safety of lenalidomide (LEN) or placebo maintenance therapy until relapse in post-autologous stem cell transplantation (ASCT) patients with newly-diagnosed multiple myeloma (NDMM). Interim analyses demonstrated superiority for LEN versus placebo for time to progression (TTP), thus meeting the primary TTP endpoint of the trial. The study was unblinded in December 2009 and, based on the TTP and overall survival (OS) benefit observed with LEN, patients in the placebo arm (who had not yet progressed) at the time of unblinding were given the option of crossing over to receive LEN. Data have been reported for a follow-up of 91 months based on the planned intention to treat (ITT) analysis, yielding a TTP hazard ratio (HR) 0.57 and an OS difference of HR 0.61 (Holstein SA, et al. Lancet Haematol. 2017;4:e431-42). Crossover to active treatment enabled patients to benefit from LEN maintenance and collection of additional efficacy and safety data; however, crossover can confound the estimates of OS for longer follow-up. We report the results of an analysis adjusted for crossover, thus aiming to provide a more accurate estimate of the survival benefit achieved with LEN maintenance therapy.

Methods: Results are reported for an analysis based on an updated data cutoff, post-October 2016. The rank-preserving structural failure time model (RPSFTM) was considered the most appropriate method to adjust for crossover (Robins JM, Tsiatis AA. Commun Stat Theory Methods. 1991;20:2609-31). This method requires that the assumption that the treatment effect is the same for patients who receive LEN at randomization as those who receive LEN as crossover treatment is not violated; a landmark analysis was performed to explore this. The iterative parameter estimation (IPE; Branson M, Whitehead J. Stat Med. 2002;21:2449-63) algorithm was used as validation. Both the RPSFTM and IPE analyses assume a residual LEN effect after discontinuation (base case).

Results: A total of 76 NDMM patients who had not progressed, as determined by central adjudication, crossed over from placebo to LEN and were included in the analysis. The landmark analysis of OS from the date of unblinding indicated that the treatment effect for the crossover versus the non-crossover placebo group was HR 0.57 (95% confidence interval [CI] 0.29-1.15) (Figure). This also provided a measure of the benefit of LEN maintenance in patients who started maintenance therapy after the trial-specified 90-100 day window post-ASCT (median 11.5 months post-randomization; range 3.2-51.0 months). Adjustment for crossover using the RPSFTM or IPE resulted in an increase in the relative treatment effect of LEN maintenance (vs placebo) on OS from 30.8 months for the ITT analysis (HR 0.61; 95% CI 0.47-0.81) to 40.1 months (HR 0.52; 95% CI 0.36-0.73) for the RPSFTM analysis and 38.8 months (HR 0.52; 95% CI 0.37-0.74) for the IPE analysis (Table). Results were consistent across the two methods.

Discussion: This analysis was performed on an updated version of the data set reported in the Holstein article. Once we adjusted for crossover, depending on the methodology used, there was an additional gain of ~40 months of OS. Previously, a pre-planned, pooled meta-analysis of 3 studies that included CALGB/Alliance (McCarthy PL, et al J Clin Oncol. 2017;35:3279-89) indicated that the OS gain was 2.5 years; however, our analysis from the CALGB study alone shows that LEN maintenance may provide a survival benefit of > 3 years. In diseases where prolonged follow-up is required to demonstrate survival benefits, allowing patients to crossover to active treatment may be important to facilitate this. However, the results should be analyzed appropriately to enable assessment of the value of the therapy. As this analysis indicates, the value of active treatment may be underestimated if adjustment for crossover is not performed. In conclusion, the data reported here provide further insight into the survival benefits of LEN maintenance therapy post-ASCT and support guideline recommendations to offer LEN maintenance therapy to all patients post-ASCT.

Support: U10CA180821, U10CA180882, U10CA180820; ClinicalTrials.gov Identifier: NCT00114101.

View largeDownload slide

View largeDownload slide

Close modal