Targeted capture sequencing identifies driver mutations and enables prognostic modeling in multiple myeloma Free

Introduction Multiple myeloma (MM) remains incurable with pronounced prognostic heterogeneity, driven by extensive genomic complexity. Incomplete understanding of molecular-clinical outcome links limits precise risk stratification and treatment optimization. Advances in high-throughput omics (e.g., whole-exome sequencing) now enable systematic dissection of MM pathogenesis. This study employs targeted capture sequencing to: (1) Identify progression-driver genomic features and prognostic markers; (2) Integrate molecular determinants with clinical indices (e.g., R-ISS); (3) Construct an optimized prognostic model refining risk stratification for personalized therapy.

Method CD138+ plasma cells from newly diagnosed MM patients underwent targeted capture sequencing (Illumina NovaSeq X Plus; >2000× depth). High-confidence variants were filtered by: (1) Excluding synonymous SNVs, dbSNP(v147)/COSMIC overlaps, benign ClinVar entries, and variants with MAF≥0.001; (2) Retaining SNVs/Indels with Phred-scaled QUAL>30, VAF>1%, and depth>500 (custom pipeline). (3) Associations with clinicopathology used Mann-Whitney U/Spearman tests. Survival differences (Kaplan-Meier, log-rank) and Cox regression-based prognostic modeling were assessed at p<0.05 (R/GraphPad Prism).

Results Targeted capture sequencing of 81 newly diagnosed MM patients identified 926 somatic mutations, predominantly missense variants (87%). OncodriveCLUST analysis identified 31 tumor driver genes, among which KRAS, NRAS, TP53, DNAH5, USP6, TRAF2, and CHEK2 exhibited statistically significant mutation frequencies (p < 0.05). Gene Set Enrichment Analysis revealed that these drivers converge on four core oncogenic pathways: RTK-RAS (55.6%), Hippo (35.8%), NOTCH (21.0%), and TP53 (12.4%), suggesting that dysregulation of these pathways represents a key molecular mechanism in MM pathogenesis.

Co-occurrence/mutual exclusivity analysis revealed significant synergistic interactions between driver gene pairs (p < 0.05), with the MAML2–ROS1 pair demonstrating the strongest cooperation (p < 0.01), suggesting Notch–tyrosine kinase signaling crosstalk. The NOTCH1–DNAH5–ROS1 cluster further established NOTCH pathway dysregulation as a key progression driver, associated with bone marrow microenvironment remodeling and aberrant proliferation. Collectively, these cooperative networks provide a mechanistic basis for combinatorial therapeutic targeting.

Integration of established prognostic markers (e.g., 1p/1q21/17p deletions, R-ISS stage, age) revealed critical molecular–clinical associations: MAX co-mutation with CDKN2C/TP53 (p < 0.05), indicating cell cycle/stress response synergy; 17p deletion linked to TP53/ADAMTS9/JAK2 alterations (p < 0.01), confirming tumor suppressor inactivation and JAK-STAT dysregulation; MAX/KMT2A mutations enriched in R-ISS stage III (p < 0.001), defining high-risk molecular drivers; and age-dependent decreases in USP6/TP53 mutation frequency (p < 0.05). These interconnected associations construct a clinically actionable molecular subtyping framework for risk-adapted management.

Finally, regression coefficients (β) of significant gene variables in the multivariate Cox model were used as weights to construct an individual prognostic risk score model. Kaplan-Meier survival analysis showed significantly shorter median PFS in the high-risk group versus the low-risk group (18 months vs. 22 months; log-rank test χ² = 24.198, p < 0.001), confirming the model's efficacy in predicting disease progression.

Conclusions By integrating targeted genomics, bioinformatic assessment of gene co-occurrence/exclusivity, and rigorous clinical correlation, this study identified key oncogenic drivers and interaction patterns in MM. We developed a validated personalized prognostic risk score via Cox regression to improve clinical stratification. These findings bridge genomic discovery with clinical application, providing mechanistic insights into MM pathogenesis and actionable tools for prognostication and therapeutic targeting. Future efforts will focus on independent model validation and functional characterization of prioritized targets.