Deutsche Studiengruppe Multiples Myelom (DSMM) Evaluation of the Risk of Relapse and Survival Using a Conditional Survival Approach in 815 MM Patients with Extensive, Complete and Repeated and Long-Term Follow-up

Introduction: CS describes probabilities of surviving t additional years (ys) given patients (pts) survived s ys, and generates information, how prognosis evolves over time. We have previously adopted this approach in MM pts with complete follow-up, which is mandatory for CS calculations (Hieke et al., Clin Cancer Res 2015). The results were refined here by accumulating time-dependent MM- and pt-specific information, including cytogenetics and response to treatment over time, again with extensive, complete and repeated long-term follow-up.

Methods: We evaluated 815 consecutive MM pts treated from 1997 to 2011, with follow-up for this study until 6/2016 in a DSMM study initiative. We assessed 20 variables, including age, stage, gender, admission period, response to therapy and others both at initial diagnosis (ID) and repeated time points during pts' regular follow-up. We calculated probabilities to survive another 5ys (5y-CS) according to disease- and host-related risks with a landmark approach. Multivariate Cox models were employed to assess the impact of baseline and of updated prognostic factors on OS.

Results: Median OS in the entire cohort and median follow-up were 5.1 and 10.3ys, respectively. Our pts showed typical characteristics for tertiary centers with a median age of 62ys (27-93), Karnofsky performance status (KPS) of 70% (20-100), hemoglobin (Hb) of 11g/dl (2.4-18.2) and bone marrow infiltration of 30%. When comparing 5y-CS probabilities from data derived at ID with those updated at the respective landmark of 1, 2, 3, 4, 5y survival, we observed substantial differences: for KPS ≥80% vs. <80% at ID, 5y-CS in pts who had survived 5ys was similar (0.58 [95%CI 0.49-0.66] vs 0.52 [95%CI 0.43-0.61]), whereas with repeated KPS follow-up data was largely different with 0.63 (95%CI 0.56-0.69) vs. 0.33 (95%CI 0.22-0.45), respectively. Similarly impressively, we observed 5y-CS probabilities of 0.55 (95%CI 0.47-0.62) vs. 0.50 (95%CI 0.39-0.60) for initial Hb levels >10 vs. ≤10g/dl, and 0.60 (95%CI 0.53-0.66) vs 0.33 (95%CI 0.22-0.46) with repeated Hb data. Likewise, for creatinine levels <1.4 vs. ≥1.4mg/dl at baseline, 5y-CS probabilities were 0.55 (95%CI 0.48-0.62) vs. 0.49 (95%CI 0.36-0.60), and with repeated follow-up 0.58 (95%CI 0.51-0.64) vs. 0.36 (95%CI 0.23-0.50), respectively. Similar results were obtained for albumin, ß2-micorglobuline (ß2-MG) but not for LDH. Median CS for pts surviving 1, 3 and 5y differed for pts without vs. with progression documented before the landmark with 6.0, 6.8 and 7.8ys vs. 2.7, 3.2 and 3.8ys, respectively.

In univariate analyses, age, stage, cytogenetics, organ function, KPS, albumin, ß2-MG, LDH, k/l-ratio and comorbidities showed significance for survival, whereas gender and admission period did not (1997-2007 vs. 2007-2011: HR 1 vs. 1.09 (95%CI 0.89-1.32). Multivariate Cox regression models for cohorts surviving 0 to 5ys showed initially increased HRs of 1.68 (95%CI 1.38-2.05) and 3.17 (95%CI 2.55-3.94) for age groups 60-69 and >70y compared to <60y. These HRs declined: pts surviving 5ys had HRs 1.42 (95%CI 1.02-1.96) and 2.06 (95%CI 1.29-3.27), respectively. In line, HRs for Durie&Salmon stage II/III decreased from 2.06 (95%CI 1.67-2.49) at ID to 1.3 (95%CI 0.97-1.76) for pts surviving 5y, and for unfavorable cytogenetics from 1.29 (95%CI 1.01-1.65) to 1.07 (95%CI 0.68-1.70). Differently, progressive disease (PD) showed slightly increasing HRs in pts with 1, 3 and 5ys of survival with 1.85 (95%CI 1.47-2.31), 2.11 (95%CI 1.64-2.71) and 2.14 (95%CI 1.54-2.98), respectively. Repetition of these analyses, including covariates, such as creatinine, KPS, Hb and κ/λ-ratio in turn into the Cox model, did not change the results with regard to age, stage, cytogenetics and PD. Review of the literature on CS in various tumors revealed improved CS with prolonged survival in lymphoma GI-, lung-, breast-cancer and after stem cell transplants. As compared to the age-matched general population, CS in cancer pts remained lower at the 5y-landmark and to the best of our knowledge has not been described with repeated risk factor analysis over time in MM nor other tumors.

Conclusions If baseline risks are compared to those repeatedly assessed over time with regular, complete follow-up, much more impressive effects on CS are observed with the latter. This is the first analysis on CS at ID and after 1 to 5ys using baseline and follow-up risk parameters in MM.