P-186: Identifying a New Risk-Stratification Strategy by Including Single-Cell Biomarkers in Multiple Myeloma

Introduction: Despite advances in new therapies, a subset of multiple myeloma (MM) patients still faces poor prognosis. Identifying these ultra-high-risk patients is of paramount importance. Existing staging methods, such as R-ISS and R2-ISS, do not incorporate gene expression information of tumor cells. Our aim is to develop a novel staging method for MM that integrates traditional staging factors, cytogenetics, and gene expression information.

Methods: To accurately identify novel biomarkers associated with patient prognosis, we performed single-cell sequencing on bone marrow aspirates from four ultra-high-risk patients (OS < 24 months) and eight non-high-risk patients (OS > 24 months) undergoing VRD treatment. We utilized multivariate COX regression analysis to evaluate new gene markers and traditional staging factors, and developed a new staging strategy by employing a weighted scoring method.

Results: Firstly, we classified Ig light chain-restricted MM cells into 10 cellular subclusters by unsupervised clustering. Besides subcluster 4, the other subclusters of MM cells group together. We found that subcluster 4 was significantly increased in patients with overall survival less than 24 month. Of note, compared to other MM cells, subcluster 4 displayed higher proliferation and drug resistance scores. We further identified seven genes significantly increasing in subcluster 4 and constructed a seven-gene score to represent the characteristics of subcluster 4. Using the dataset of CoMMpass study of newly diagnosed MM patients, multivariate COX regression analysis revealed the seven-gene score is an independent prognostic factor. The seven-gene score more effectively identified ultra-high-risk patients compared to UAMS70 and SKY92. Based on the hazard ratios from the multivariate regression, we assigned weighted scores to the independent prognostic factors, including ISS stage, del(17p), 1q+, t(4;14), and the seven-gene score, and established a new staging model, MR-ISS. Compared to the ISS, R-ISS, and R2-ISS staging systems, the MR-ISS established in this study more evenly risk stratified patients (Low 16%, Intermediate 34%, High 38% and Ultra-high 11%), and better identified ultra-high-risk patients (OS < 24 months). We further validated the performance of MR-ISS using the GSE136337 dataset and an internal dataset. Additionally, we quantitatively assessed the 7-gene score in 139 newly diagnosed patients using both single-color and multi-color ddPCR, confirming its superior performance of the MR-ISS staging system and providing the possibility in clinical routine assay.

Conclusions: Our study identified an aggressive characteristic with seven genes highly expressed in ultra-high-risk patients by single-cell RNA-sequencing. A further risk stratification system (MR-ISS) incorporates ISS stage, del(17p), 1q+, t(4;14), and the seven-gene score was developed. Our study provides a more effective and convenient risk-stratify model in clinical practice to identify the ultra-high-risk group patient.