A B-Cell like Phenotype Is Associated with Sensitivity to Venetoclax in Multiple Myeloma

As B cells differentiate into plasma cells they switch from Bcl-2 dependent to Mcl-1 dependent. Multiple myeloma should therefore be resistant to the Bcl-2 inhibitor venetoclax. However an ongoing phase I trial of venetoclax has reported an ORR of 40% in a subset of myeloma characterized by the t(11;14) translocation. Nevertheless, 60% of patients with t(11;14) failed to respond to treatment, suggesting that factors beyond t(11;14) determine response. Our own testing of 15 cell lines and approximately 50 myeloma patient samples detected both venetoclax resistant t(11;14) samples as well as venetoclax sensitive non-t(11;14) samples. We have also used siRNA to knockdown cyclin D1, the protein overexpressed in t(11;14), and demonstrated that knockdown has no effect on venetoclax response in three venetoclax sensitive t(11;14) cell lines. Together these findings indicate that t(11;14) is neither necessary nor sufficient for response to venetoclax, and is therefore unlikely to be driving Bcl-2 dependence. To identify other factors that predict for response to venetoclax in myeloma, we used RNA sequencing data to analyze the differential gene expression in sensitive vs. resistant cell lines. We identified 148 genes whose mean expression differed by greater than 2 fold and with an fdr < 0.05. Interestingly, numerous B cell related genes were present among the genes up-regulated in the sensitive lines, including MS4A1 (CD20), CD79A, STAT5A, RASGRP2, FCGR2B, and CCR7. These B cell genes were not uniformly expressed in all of the sensitive cell lines, suggesting that no single marker is likely to be predictive. Expression of CD20, CD79A, and PAX5 have previously been reported in t(11;14) myeloma, however our analysis revealed expression of these markers in the non t(11;14) lines OCI-My5 and PCM6, both of which are t(14;16). A second set of genes highly upregulated in the sensitive lines includes interferon response genes. Notably, none of the master plasma cell transcription factors including Blimp1, IRF4, or XBP1 differed significantly between the sensitive and resistant lines. More importantly, neither Cyclin D1 nor any of the Bcl-2 family genes was differentially expressed, providing additional evidence that expression of the Bcl-2 family and the presence of t(11;14) cannot account for differences in venetoclax response. In order to further analyze the B cell-like properties of the venetoclax-sensitive cell lines, we applied gene set enrichment analysis using genes previously reported to be differentially expressed between mouse B cells and plasma cells. This revealed that B cell genes were enriched in venetoclax sensitive cell lines while plasma cell genes were depleted among the same cell lines. We also used the 148 differentially expressed genes associated with venetoclax response for linear discriminate analysis (LDA) on a larger panel of myeloma cell lines to predict their response to venetoclax. Our analysis correctly predicted the response for at least one cell line previously reported to be venetoclax sensitive (MOLP2) and multiple cell lines reported to be resistant (JJN3, L363, LP1, NCIH929, MOLP8, KMS26, KMS34), suggesting that this scoring system may be valid. We applied the same analysis to the CoMMpass data set, which currently includes RNAseq expression data from nearly 700 newly diagnosed myeloma patients. Many of the t(11;14) are predicted to be venetoclax sensitive, which is consistent with cell line and patient responses. Notably, a number of t(14;16) patients are predicted to be sensitive, mirroring our cell line results. In contrast, t(4;14) patients are predicted to be almost exclusively venetoclax resistant. Together this analysis demonstrates that there are patients with expression patterns similar to venetoclax-sensitive cell lines and more importantly, that such a scoring system may be clinically useful as well.