A gene signature to predict the survival of patients with multiple myeloma

Abstract

Background: Multiple Myeloma (MM) is an incurable hematological malignancy. Genomic studies have provided solid foundation for understanding the molecular mechanisms underlying progression of multiple myeloma. This study aimed to identify novel molecular markers associated with MM prognosis. Methods: The GSE136337 dataset was downloaded from GEO database and taken as the training cohort. The GSE24080 and GSE57317 datasets also obtained from GEO database and were set as the external validation cohorts. In training cohort, Kaplan–Meier analysis and univariate Cox regression analyses were performed to preliminary identify prognostic genes. Multivariate Cox regression analysis was applied to build prognostic signature, which was then verified in the validation cohorts through Kaplan–Meier, Cox, and ROC analyses. Gene Set Enrichment Analysis (GSEA) and tumor immunity analysis were used to elucidate the molecular mechanisms and immune relevance. The half-maximal inhibitory concentration (IC50) of chemotherapeutic drugs in MM patients were estimated by pRRophetic algorithm. Results: A total of fifteen genes were individually associated with overall survival via Kaplan–Meier analysis and univariate Cox regression analyses. A novel five-gene signature (CTPS1, FABP5, STK26, HECA and TMEM167B) was constructed for MM prognosis prediction. The ROC curve showed that the prognostic signature performed well at predicting overall survival in both the training cohort and the external validation cohorts. The K-M curve revealed low risk group had significantly higher survival rates compared with the high risk group. This signature was further proven to be an independent prognostic factor compared to other clinic-pathological parameters via the multivariate Cox regression analysis. Gene set enrichment analysis revealed the pathways enriched in the high risk group were mainly associated with the cell proliferation and glycolysis. Tumor-infiltrating cells had notably differential levels of expression between high and low risk groups. Moreover, the patients in the high risk group were more sensitive to chemotherapy. Conclusions: The five-gene signature could accurately predict patients’ prognosis and had close relationship with the tumor immune microenvironment and drug sensitivity, which may provide new insights for prognosis and treatment of MM patients and help doctors make more rational treatment decisions.