A prognostic survival model based on metabolism-related gene expression in plasma cell myeloma

Accurate survival prediction of persons with plasma cell myeloma (PCM) is challenging. We interrogated clinical and laboratory co-variates and RNA matrices of 1040 subjects with PCM from public datasets in the Gene Expression Omnibus database in training ( N = 1) and validation ( N = 2) datasets. Genes regulating plasma cell metabolism correlated with survival were identified and seven used to build a metabolic risk score using Lasso Cox regression analyses. The score had robust predictive performance with 5-year survival area under the curve (AUCs): 0.71 (95% confidence interval, 0.65, 0.76), 0.88 (0.67, 1.00) and 0.64 (0.57, 0.70). Subjects in the high‐risk training cohort (score > median) had worse 5-year survival compared with those in the low‐risk cohort (62% [55, 68%] vs. 85% [80, 90%]; p