Skip to main content
IMS Events Main banner
Rate ePoster

    Myeloma Genomics and Microenvironment and immune profiling

    Category: Myeloma Genomics and Microenvironment and immune profiling

    Expression Profile–Based Dissemination Score for Myeloma Prognostication

    (PA-214) Expression Profile–Based Dissemination Score for Myeloma Prognostication

    Wednesday, September 17, 2025
    Has Audio
    Introduction: Disseminated myeloma, defined by the presence of circulating plasma cells (CPCs) and/or extramedullary disease (EMD), is associated with poor prognosis and resistance to standard therapies. Epithelial-to-mesenchymal transition (EMT) is marked by the downregulation of adhesion molecules such as CD138, CD56, and integrins. EMT has been implicated in both CPCs and EMD by previous studies and unifies the understanding of disseminated myeloma.

    Methods: We analyzed baseline RNA-seq data from 754 newly diagnosed multiple myeloma patients enrolled in the MMRF CoMMpass study. A total of 1,357 candidate genes were selected by integrating the EMTome gene set (1,114 genes) with genes differentially expressed in association with CPC percentage (n=243) and PET-CT–confirmed plasmacytomas (n=34), identified using the limma-voom pipeline (FDR ≤ 0.05, effect size >5 SD). Among these, 365 genes demonstrated significant associations with overall survival (p ≤ 0.05 in univariate Cox regression) and were entered into an Elastic Net Cox model (α = 0.5), yielding a 30-gene signature to compute the weighted Dissemination Score(D-S).

    Results: The D-S included positively weighted genes: PHF19, FABP5, APOBEC3B, TOP2A, HTR2C, FOXM1, Novel lnRNA(ENSG00000237422), BUB1, NAP1L3, WEE1, PFKP, SHROOM3, KIR3DX1, KIF21B, PKDCC, JAG2, FOXD1, MAGEA1, CDCA7, ELOVL6, HHAT, CRIP2, SCRIB, and ADAM15., and negatively weighted genes: LTBP1, DENND2D, GRHL1, LAMA5, TJP1, and KIAA1191, representing a balance between mesenchymal and epithelial traits. The D-S demonstrated strong prognostic discrimination with a C-statistic of 0.71 (SE = 0.02). Median overall survival (OS) was 55.5 months (95% CI: 47.2–65.0) in the high D-S group, versus 102.5 months (95% CI: 89.3–NA) in the low D-S group. The score further stratified patients within IMWG risk groups (log-rank χ² = 77.83, df = 3, p < 0.001), with the worst outcomes seen in those with both high DS and high IMWG risk (median OS: 52.3 months, 95% CI: 37.8–70.9). Incorporating the D-S into existing genomic models improved performance: C-statistics increased from 0.68 to 0.76 for EMC92 and from 0.69 to 0.75 for UAMS70 when D-S was added alongside age and R2-ISS (p < 0.001 for both comparisons, Wilcoxon signed-rank test, 1,000 bootstraps). Dissemination Score was significantly elevated in patients harboring high-risk cytogenetic lesions, including 1q21 gain, 17p deletion, t(4;14), and GPRC5D loss (all p < 0.001), while it was lower in hyperdiploid cases, suggesting its alignment with aggressive biological subtypes. In multivariable Cox regression, the D-S remained an independent predictor of overall survival (HR 1.09, 95% CI: 1.05–1.14, p < 0.001), outperforming other gene expression–based scores including UAMS70 and EMC92.

    Conclusions: The Dissemination Score is a robust metric that captures the biology of plasma cell dissemination, a phenotype analogous to metastatic behavior in solid tumors and may help guide therapeutic decisions.