Outcomes of patients with newly diagnosed multiple myeloma treated with quadruplets and autologous stem cell transplant stratified by the 2025 ims/IMWG risk classification

Abstract

Abstract Introduction: Despite significant improvement in the survival of patients with newly diagnosed multiple myeloma (NDMM), a subset of patients continues to experience poor outcomes. Accurate classification of such high-risk patients is crucial to contextualize new therapies and develop risk-adapted strategies. One of the most frequently used classification tools identifies NDMM patients as high-risk when they harbor del(17p), t(4;14) or t(14;16), hereafter referred to as “old” high-risk (old-HiR). Recently, the International Myeloma Society (IMS) and International Myeloma Working Group (IMS/IMWG) published a new consensus high-risk (new-HiR) definition which combines certain cytogenetic abnormalities, next generation sequencing data (TP53 mutation, biallelic del1p), and elevated beta-2-microglobulin (B2M) without renal impairment. While the 2025 IMS/IMWG classification was predominantly developed among patients treated with triplet combinations, the MM treatment paradigm has rapidly evolved to include monoclonal antibody-based quadruplets (QUAD) with autologous stem cell transplant (ASCT) and further improved outcomes. We aimed to characterize the performance of the 2025 IMS/IMWG classification in the QUAD+ASCT population and compare the old-HiR vs. new-HiR models. Methods: We evaluated a cohort of NDMM patients who received QUAD induction, ASCT and risk-adapted post-ASCT therapy. We retrospectively reclassified patients as standard risk (new-StR) or high risk (new-HiR) using the 2025 IMS/IMWG classification. We evaluated the cumulative incidence of progression (CIP) at 18 and 36 months to demonstrate the ability of the new system to identify functional high-risk patients (patients experiencing disease progression within 18 and 36 months from start of initial therapy, respectively). We subsequently compared old-HiR vs. new-HiR classification. Best fitting models were determined using Akaike's information criterion (AIC), an information theory approach. Results: Among 310 patients, 89 (29%) were classified as new-HiR. Median age was 62 years (IQR 56-68), 55% were male, 33% of racial-ethnic minority, 6% harbored t(11;14), 56% any del(17p), 51% gain/amp(1q), 14% del(1p), 27% t(4;14), and 12% t(14;16). After a median follow up of 41.6 mo, the 4-year PFS was 49% vs. 84% for new-HiR vs. new-StR (HR 4.17, 95%CI 2.58-6.73; p<0.001). The CIP at 18 months was 15% vs. 3% for new-HiR vs new-StR and at 36 months was 33% vs. 10%, respectively. For the same population, 4-year PFS for old-HiR vs. old-StR was 55% vs. 83% (HR 3.34, 95% CI 2.08-5.36, P<0.001). Overall, the new model produced the best fitness, with delta AIC = 9.3 (delta AIC>2 being generally considered meaningful). Furthermore, we demonstrate that the presence of old-HiR abnormalities did not affect outcomes of patients with new-HiR (P=0.70) or new-StR (P=0.56). In contrast, when we applied the new criteria to patients with old-HiR we were able to successfully classify two cohorts with very distinct outcomes (median PFS NR vs 49.3 mo, for new-StR vs new-HiR, respectively, p=0.05) similar to when we applied new criteria for old-StR patients (median PFS NR vs. 40.7 mo for new-StR vs new-HiR; p=0.007). We further explored the frequently used 3-tier system based on cumulative number of high-risk chromosomal abnormalities [HRCA; del(17p), t(4;14), t(14;16) and gain/amp(1q)]. Among patients with new-HiR, the presence of 1 (P=0.99) or 2+ HRCA (P=0.49) did not affect outcomes. Conclusions: The 2025 IMS/IMWG consensus criteria for high-risk MM appears to be more prognostic in the setting of modern QUAD+ASCT (HR 4.17) compared to triplet+ASCT where the criteria were first created and validated (HR 2.17-2.94). Among patients with new-HiR, presence of old-HiR abnormalities or application of 3-tier HRCA system did not further improve the classification. Ultimately, the new simplified 2-tier system incorporating genomic data and B2M is most relevant in the era of QUADs and identifies patients at risk of early relapse with greater accuracy.