Functional and Molecular Analyses Reveal Impaired HSPCs in Multiple Myeloma Patients Post-Induction

Multiple Myeloma (MM) is an incurable disease due to relapse, despite development of novel agents and progress in autologous stem cell transplantation (ASCT) (Rajkumar, 2022). There remains a high variability in patient ASCT response in MM. We investigated the function and the transcriptome of CD34⁺ hematopoietic stem and progenitor cells (HSPC) of newly diagnosed, treatment-naïve MM (NDMM) patients and that of patients that underwent induction treatment prior to ASCT (treated MM) to inform us on possibilities to further improve ASCT outcomes. HSPCs from treated MM patients, when transplanted into NBSGW recipients, presented with a significantly impaired function compared to control HSPCs from healthy aged-matched donors (CD45⁺ frequency in recipients 0.8% vs. 16.7%, P < 0.0001), while HSPCs from NDMM patients showed a function similar to control HSPCs. In addition, there was a significant lymphoid differentiation deficiency of HSPCs from NDMM patients (CD19⁺ frequency 57.7% vs. 75.5%, P 0.02). To delineate likely mechanisms, single-cell RNA-sequencing analyses of CD34⁺ cells were performed. Interestingly, the transcriptomic profile of HSPCs from MM patients (both before, but also especially after treatment) already present with a MM gene signature, despite the fact that HSPCs do not comprise the target malignant cell population of MM (CD138⁺ plasma cells). Using machine learning (ML) algorithms, we developed a model to identify gene signatures among distinct types of HSPCs from MM patients. Based on a blinded dataset, the model correctly predicted 55.2%, 94.2% and 77.5% cells to healthy control, NDMM or treated MM group, respectively, revealing that NDMM HSPCs show the highest gene signature specificity. Interestingly, 16.8% of the treated MM cells showed a NDMM-like transcriptomic profile while 5.7% were assigned to the healthy control. We next asked which specific HSPC cell types, including HSCs, (Calvanese et al., 2022; Bandyopadhyay et al., 2024) present with the ML model MM signatures. GMPs showed the highest correlation whilst HSCs showed no association and remained unaffected. Furthermore, gene set enrichment analysis of EMC92 signature (Kuiper et al., 2015) based on ML gene ranking revealed a high correlation to a high-risk MM profile in the HSPCs from treated MM patients. In summary, we show that CD34⁺ HSPCs from MM patients present with MM gene signatures. We identified that HSPCs acquire a high-risk MM expression profile upon MM treatment, which likely adds to the risk of relapse. This MM profile resides primarily within GMPs, while HSCs stay unaffected by transcriptional changes. Our findings imply that ASCT for MM treatment might be improved by either harvesting CD34⁺ cells already before treatment or transplanting specific types of CD34⁺ cells that do not show an MM signature.