Abstract
BACKGROUND: The present study intended to establish a predictive nomogram for early relapse (ER) (<12 months) after autologous stem cell transplantation (ASCT) in the novel drug era for multiple myeloma (MM). PATIENTS AND METHODS: The nomogram was designed and constructed to a retrospective clinical data of newly diagnosed MM patients received novel agent induction therapy and subsequent ASCT at three centers in China from July 2007 to December 2018. The retrospective study was conducted among 294 patients in the training cohort and 126 in the validation cohort. The nomogram's predictive accuracy was evaluated by the concordance index, calibration curve and decision clinical curve. RESULTS: The study cohort included 420 newly diagnosed MM patients, and 100 (23.8%) were identified as having ER, including 74 in the training cohort and 26 in the validation cohort. According to the result of multivariate regression in the training cohort, the prognostic variables included in the nomogram were high-risk cytogenetics, LDH > UNL, and response less than very good partial response (VGPR) after ASCT. The calibration curve showed good fitness between the nomogram predictions and the actual observations and the nomogram was further validated by a clinical decision curve. The nomogram's C-index achieved 0.75 (95% CI, 0.70-0.80), which was higher than that of the Revised International Staging System (R-ISS) (0.62), ISS (0.59), and Durie-Salmon (DS) staging system (0.52). The discrimination ability of the nomogram in the validation cohort was superior to that of the other staging systems (C-index: 0.73 vs. R-ISS (0.54), ISS (0.55), and DS staging system (0.53)). DCA showed the prediction nomogram adds much more clinical utility. Different scores of the nomogram draw a distinction of OS. CONCLUSION: The present nomogram could serve as a feasible and accurate prediction of ER in novel drug induction transplantation-eligible MM patients, which could help modify the post-ASCT strategy for patients at high risk of ER.