Gene expression risk signatures maintain prognostic power in multiple myeloma despite microarray probe set translation

Summary Introduction Gene expression profiling (GEP) risk models in multiple myeloma are based on 3′‐end microarrays. We hypothesized that GEP risk signatures could retain prognostic power despite being translated and applied to whole‐transcript microarray data. Methods We studied CD138‐positive bone marrow plasma cells in a prospective cohort of 59 samples from newly diagnosed patients eligible for high‐dose therapy (HDT) and 67 samples from previous HDT patients with progressive disease. We used Affymetrix Human Gene 1.1 ST microarrays for GEP. Nine GEP risk signatures were translated by probe set match and applied to our data in multivariate Cox regression analysis for progression‐free survival and overall survival in combination with clinical, cytogenetic and biochemical risk markers, including the International Staging System (ISS). Results Median follow‐up was 66 months (range 42–87). Various translated GEP risk signatures or combinations hereof were significantly correlated with survival: among newly diagnosed patients mainly in combination with cytogenetic high‐risk markers and among relapsed patients mainly in combination with ISS stage III. Conclusion Translated GEP risk signatures maintain significant prognostic power in HDT myeloma patients. We suggest probe set matching for GEP risk signature translation as part of the efforts towards a microarray‐independent GEP risk standard. (ClicinalTrials.gov identifier: NCT00639054).