MYD88 L265P Mutations Influence Clinical Outcome and Identify a Pathway for Targeted Inhibition in Chronic Lymphocytic Leukemia

The L265P somatic mutation in the Myeloid Differentiation Primary Response 88 (MYD88) gene is recurrently observed in CLL; although this mutation has been demonstrated to have functional effects in multiple hematologic malignancies, its role in CLL is largely unknown. To address this gap in knowledge, we examined the clinical and biological impact of MYD88 L265P mutations in CLL by analysis of gene expression, cell viability and Toll-like Receptor 9 (TLR9)-induced signaling and cytokine production.

Out of 160 CLL patient samples subjected to whole-exome sequencing and previously reported by our group, 10 were found to harbor MYD88 L265P mutations, all of which possessed mutated immunoglobulin heavy chain variable (IGHV) regions (p = 0.006). While IGHV mutated patients are generally known to exhibit better prognosis compared to IGHV unmutated patients, the presence of MYD88 L265P within the IGHV mutated subset was associated with earlier age of disease onset (p = 0.04) and worse overall survival (OS; p = 0.00017), comparable to IGHV unmutated samples with wild-type (WT) MYD88. No association with the presence of chromosome 13q deletions (p = 0.26) or prior treatment at the time of sampling (p = 0.10) was observed.

Gene expression microarray analysis restricted to the IGHV mutated subset (MYD88 WT: n = 76; MYD88 L265P: n=10) and conducted using a PAM-based approach demonstrated that MYD88 L265P mutation was associated with differential expression of 28 genes, whose expression was then examined across all CLL samples with available gene expression data (n = 150). Using Cox modeling, a composite gene signature score was determined for each patient, who were subsequently dichotomized based on median signature. This method was able to predict both OS and event free survival (EFS) in a univariable analysis (OS: p = 1.2E-06; EFS: p = 7.6E-13). Statistical significance was maintained when a multivariable analysis was conducted, adjusting for known CLL risk factors including age, IGHV status, ZAP70 expression, cytogenetics and prior treatment (p < 0.0001 for OS and EFS). The univariable (OS:  p = 1.6E-05; EFS: p = 5.7E-10) and multivariable findings (p < 0.003 for OS and EFS) were further confirmed in an independent validation cohort (n = 87). To identify a more parsimonious gene set, we applied a L1 penalized proportional hazards model to the discovery and validation cohorts, separately.  This approach identified 5 overlapping genes (BCAT1, BMP6, CHAD, IKZF2, and TRIO) between the two cohorts that appear to be the main drivers of the predictive signature.

To inhibit MYD88 signaling in CLL cells, we treated MYD88 L265P and WT cells (n = 6/group, matched for clinical characteristics: IGHV, ZAP70, cytogenetic, and treatment status) in vitro with a highly-selective small molecule IRAK4 inhibitor, ND-2158 (Nimbus Therapeutics).  ND-2158 significantly reduced cell viability in a dose dependent manner in both MYD88 WT and L265P primary CLL cells, either alone or in combination with a fixed concentration of the B-cell receptor (BCR) pathway inhibitor, ibrutinib. 

The TLR9 agonist CpG was used to stimulate signaling through the MYD88 pathway in vitro. ND-2158 inhibition of CpG-induced IRAK4 activation in CLL cells (n = 3/group, matched for clinical characteristics) blocked IRAK1 and IκBα degradation and led to a dose-dependent decrease in the ratio of phospho/total proteins. No significant differences were noted between MYD88 WT and L265P samples, consistent with our cell viability results. 

CLL-secreted levels of IL-6, IL-10 and CCL3 were measured in culture supernatants treated with ND-2158 +/- CpG stimulation (n = 4/group, matched for clinical characteristics). CpG stimulated cytokine levels (p < 0.0001 for all cytokines +/- CpG) were significantly inhibited in a dose-dependent manner by ND-2158. Again, no significant differences were observed between MYD88 WT and L265P CLL with respect to cytokine production, either at baseline or in CpG-stimulated DMSO treated control cells.

In conclusion, the differences in clinical outcome and gene expression observed between MYD88 WT and L265P IGHV mutated CLLs indicate a functional role for MYD88 L265P in CLL. The inferior clinical outcome in IGHV mutated CLL with L265P mutation suggests that MYD88 signaling may be a relevant target in CLL. ND-2158 inhibits signaling in the MYD88 pathway, suggesting potential therapeutic utility of IRAK4 inhibitors in CLL.