A gene expression based predictor for high risk myeloma treated with intensive therapy and autologous stem cell rescue

Abstract

Myeloma is characterized by a highly variable clinical outcome. Despite the effectiveness of high-dose therapy, 15% of patients relapse within 1 year. We show that these cases also have a significantly shorter post-relapse survival compared to the others (median 14.9 months vs. 40 months, p = 8.03 × 10(- 14)). There are no effective approaches to define this potentially distinct biological group such that treatment could be altered. In this work a series of uniformly treated patients with myeloma were used to develop a gene expression profiling (GEP)-based signature to identify this high risk clinical behavior. Gene enrichment analyses applied to the top differentially expressed genes showed a significant enrichment of epigenetic regulators as well as "stem cell" myeloma genes. A derived 17-gene signature effectively identifies patients at high risk of early relapse as well as impaired overall survival. Integrative genomic analyses showed that epigenetic mechanisms may play an important role on transcription of these genes.

Keywords: Myeloma; gene expression profiling; predictor; risk.

Figure 1.

Impact of remission duration on post-relapse survival according to data from Myeloma IX. (A) Analyses on 423 relapsed cases show a cut-off effect of relapsing within 1 year on post-relapse survival: median 4.1 months (< 6 months), 16.1 months (6 months–1 year), 40 months (1 year–18 months), 33.4 months (18 months–2 years) and not reached (> 2 years). (B) When combined, patients who relapsed within 1 year post-HDT had median post-relapse survival of 14.9 months in contrast to 40 months with those who relapsed at later point (log-rank test p = 8.03 × 10^{− 14}).

Figure 2.

Effect of risk groups derived from the REL-17 signature on PFS and OS. In the training set, 15.1% cases being identified as having more than 60% chance to relapse within 1 year had significantly shorter PFS (A; median 13.8 vs. 34.8 months, p < 10^{− 16}) and OS (B; median 29.9 vs. 88.1 months, p = 2.39 × 10^{− 14}) in contrast to those at lower risk. Using the same criteria 12.3% patients being identified at high risk in the test set also had significantly shorter PFS (C; median 15.9 vs. 40.5 months, p = 10^{− 7}) and OS (D; median 53 months vs. not reached, p = 0.0003) compared to rest of the cases.