A Novel Hypoxia Gene Signature from the HOVON-65/GMMG-HD4 Phase 3 Study Prognostic of Survival and Associated with Immune Suppression in the Tumor Microenvironment
Multiple myeloma (MM) thrives in the bone marrow, a niche that provides myeloma cells with essential nutrient and growth signals. This hostile immune suppressive tumor micro-environment (TME) limits immune cell function and contributes to poor treatment responses. Hypoxia, a key feature of the TME, regulates cancer cell metabolic homeostasis, differentiation, and progression. We developed a MM hypoxia gene signature using data from the HOVON65/GMMG-HD4 trial that compared bortezomib during induction and maintenance versus standard of care in newly diagnosed transplant eligible MM patient, (Sonneveld et al, JCO 2012), prognostic of poor patient outcome.
MM cell lines (U266, JJN-3, RPMI-8226, OPM-2, L363) were cultured under normoxic (20% O₂) and hypoxic (0.02% O₂) conditions. Bulk RNA sequencing identified hypoxia-associated differentially expressed genes (DEGs) with a false discovery rate (FDR) cutoff of 0.01. DEGs common to ≥3 cell lines were selected as “seed” genes for cluster analysis in the HOVON-65/GMMG-HD4 training cohort consisting of 326 patients with 10 year follow up (Mai et al, Hemasphere 2024). To identify a prognostic signature, two machine learning shrinkage algorithms—LASSO and PAMR—were applied independently, and prognostic genes were selected for Cox proportional hazards regression analysis. The signature was validated in publicly available datasets: COMMPASS, GSE57317, and GSE4452. Immune cell infiltration was assessed using MM whole bone marrow RNA data from GSE136324 and Single Cell RNA Seq COMMPASS and GSE232988.
Eighty hypoxia-induced DEGs were identified, with the top Gene Ontology (GO) pathways including“response to hypoxia” and “response to oxygen levels.” K-means cluster analysis assigned 121 patients to the hypoxia-high cluster and 205 to the hypoxia-low cluster. Cox proportional hazards regression analysis revealed a median PFS of 34.7 months [95% CI 30.8- 38.1], in the hypoxia-low and 20.8 months [95% CI 18.7-26.1], in the hypoxia-high group with significant overall survival (OS) (HR 0.46 [95% CI 0.31-0.68], p< 0.0001) benefit for the hypoxia-low cluster. Both LASSO and PAMR reduced the number of predictive genes to five, which were significantly associated with worse PFS (HR 0.39 [95% CI 0.26-0.59] p< 0.0001) and OS (HR 0.59 [95% CI 0.44-0.78], p< 0.001) for the hypoxia cluster. In multivariable analysis, the hypoxia gene signature was independently associated with worse PFS and OS and this association could be improved by combining it with ISS stage and TP53 mutations. The signature was validated in the COMMPASS, GSE57317, and GSE4452 datasets. Using whole bone marrow and single cell RNA, the signature revealed genes associated with a more immune-suppressive TME, characterized by decreased neutrophil and NK cell infiltration and upregulation of immune checkpoints.
We developed a hypoxia-based gene signature for MM, predictive of survival and associated with an immune-suppressive TME.