Plasma proteomic analysis as a tool for monitoring the development of ibrutinib resistance in Waldenström macroglobulinemia.

Abstract About a decade ago, the BTK inhibitor ibrutinib (IBR) became the first approved agent for the treatment of patients with symptomatic Waldenström Macroglobulinemia (WM), and despite the high response rates and durable remissions, about a third of patients fail to respond to or maintain their response to therapy while there is notable heterogeneity in the depth and the timing of response. Clonal characteristics are important in elucidating the reasons for this heterogeneity, but peripheral blood biomarkers could be able to capture clonal and microenvironmental heterogeneity of the disease, providing new insights and predictive and prognostic tools. In this study, we conducted comprehensive plasma proteomic profiling of WM patients before and after IBR treatment, aiming to identify plasma-based prognostic biomarkers that could improve prediction of response to BTK inhibitor therapy. We performed high-throughput profiling of >5400 proteins using the Olink® Explore HT library and Proximity Extension Assay (PEA) technology. We profiled 84 peripheral blood plasma samples from 35 WM patients including longitudinal samples from patients at diagnosis and 6-month after start of IBR therapy. After a median follow-up of 1.7 years (range 0.58-4.08 years), 5 (14%) patients achieved very good partial response (VGPR); 17 (49%) a partial response (PR), 4 (11%) patients had achieved a minor response (MR) whereas 9 (26%) patients had stable or progressive disease (SD and PD) on IBR therapy. For the purposes of the analysis, we classified 22 patients as responders to IBR (R group, those that achieved at least partial response) and 13 patients were classified as non-responders (NR group, those with minor response, stable or progressive disease after 6 months of therapy). We initially compared the proteomic profile among the two response groups to identify factors associated with response to IBR. Our analysis identified 40 significantly upregulated proteins in the NR group including MAPKAP1, SF3B3, AKT2, PSMG3-AS1, VPS4B and PPP2R5A involved in pathways such as PI3K, MAPK, MYC and regulation of CXCR4. On the other hand, 24 proteins were significantly upregulated in the R group including SELE, MAP3K19, HDAC9 and RGS12. To determine robust predictors of responsiveness, before treatment with IBR, we performed ElasticNet regression, which retained 61 features including ARID5A, MDM1, PCNA and BRD3 that could best separate the responders from the non-responders with significant impact in PFS (p=0.002). To identify proteins that significantly changed over time and between response groups, protein levels were fit to an ANOVA model with main effects of time and of response. Hierarchical clustering of all patients by levels of response effect and treatment time, identified 248 proteins as significant for time-effect and highlighted the separation of this subset. These proteins were mainly involved in the inflammatory response, chemokine secretion and chemokine signaling acting on TNFα and JAK/STAT signaling pathways. When observing change from baseline to 6 months, proteins significant for time-effect were separated into two distinct clusters. Focusing on cytokine and chemokine secretion in both time points, we observed that amongst plasma proteins higher in NR group compared to the R group at baseline were cytokines such as TNFRSF13B, TNFRSF17, CXCL5, CXCL11 and IL2RA. On the other hand, CXCL13, MMP1 and CCL17 were higher in the R group. By 6 months of treatment the reduction of CXCL13 expression levels were more pronounced in the R compared to the NR group while MMP12 was significantly increased in the NR group.Proteins such MMP8 (member of matrix metalloproteinases family) and TNFSF13B (B cell activating factor / BAFF/BLyS) significantly increased upon treatment with IBR. These changes were more pronounced in the NR group compared to the R group. Conversely, we observed a significant reduction of vital proteins involved in chemokine signaling as well as T cell recruitment and activity such as CXCL6, CXCL13, CXCL11 and CCL5 upon treatment with IBR. In conclusion, this study represents the first analysis of the plasma proteome in WM aiming at identifying potential biomarkers for predicting response to IBR. Furthermore, we identified plasma proteins that change over time in both response groups, which could provide valuable insights into the resistance mechanisms involving both tumor and tumor microenvironment markers.