Genomic Evolution By Serial Single-Cell Sequencing Identifies High Tumor Expression of LYN As a Promoter of Resistance in Waldenstrom’s Macroglobulinemia Undergoing Ibrutinib Monotherapy

Background: Ibrutinib, as a first-line treatment for Waldenstrom macroglobulinemia (WM), has demonstrated a good safety profile and efficacy, particularly in MYD88^mut patients. However, clinical responses are variable and resistance to ibrutinib remains a significant clinical challenge.

Methods: To characterize the molecular correlates of response and resistance to ibrutinib, we performed high-throughput droplet-based 5′-single-cell RNA sequencing (scRNA-seq) paired with B cell receptor sequencing (scBCR-seq) on 75 longitudinal bone marrow samples collected at multiple timepoints from WM patients treated with ibrutinib monotherapy, with a median follow-up of 50 months (NCT02604511 clinical trial). Whole-exome sequencing (WES) was performed on 34 samples collected at baseline and post-treatment in 17 patients. Bulk RNA-seq was performed on 51 samples collected at baseline patients treated with ibrutinib monotherapy.

Results: To study the impact of ibrutinib on tumor clonal dynamics, we identified malignant cells leveraging our scBCR-seq data, inferred somatic copy number variants (CNVs) in tumor cells and constructed phylogenetic trees for each tumor separately. We then traced the frequency of each clone across timepoints and characterized evolutionary patterns in response to ibrutinib. Clonal evolution, whereby daughter subclones increased in abundance post-treatment, was observed exclusively in patients who progressed on treatment and was associated with significantly shorter progression-free survival (PFS). Remarkably, in some cases, clonal evolution was observed several months before clinical progression was noted, suggesting that genomic surveillance could help diagnose progression long before the clinical criteria are met. Notably, several patients who responded to ibrutinib demonstrated clonal de-evolution, whereby the frequency of more genomically evolved daughter subclones decreased in abundance post-treatment resulting in less complex clonal architecture. Clonal de-evolution may help explain a particular clinical phenomenon in WM, whereby patients may experience long PFS despite significant residual disease post-therapy, and indicate that ibrutinib can help prune genomically mature subclones while sparing the root clone. Both clonal evolution and de-evolution patterns were validated by WES (n=18). Notably, clonal evolution or de-evolution was not associated with a particular CNV nor with CXCR4 mutation status, suggesting that it may be non-genetic cellular states that predict whether a subclone will dominate post-treatment (i.e., clonal evolution) or recede (i.e., clonal de-evolution). To understand whether it is possible to predict these patterns prior to administering treatment, we systematically compared the transcriptional profiles of “winner” and “loser” clones within patients who exhibited clonal evolution and derived an evolution signature out of genes that were recurrently up- and down-regulated in “winner” clones. We then used bulk RNA-seq to score a larger cohort of patients for our evolution signature (n=51) and observed that patients with higher scores had significantly shorter PFS compared to patients with lower scores. Interestingly, the top upregulated gene in “winner” clones was the BCR signaling component LYN. To verify the correlation of high LYN levels with ibrutinib resistance, we knocked down LYN in BCWM.1 and MWCL-1 cells and observed significant sensitization to ibrutinib compared to LYN-WT cells. Collectively, these results suggest that it may be possible to predict how tumors will respond to ibrutinib based on their transcriptomic status at baseline, and unveil novel targets that could be therapeutically relevant in WM.

Conclusion: We demonstrated that genomic evolution underlies and precedes clinical progression by several months in patients with WM receiving ibrutinib, supporting that active genomic surveillance may help improve monitoring of response and early identification of resistant clone emergence. Furthermore, we showed that ibrutinib may genomically down-grade WM tumors, and adopt a novel transcriptional signature that can help predict a tumor’s response to ibrutinib. Lastly, we identified novel therapeutic targets, including LYN, that could help circumvent ibrutinib resistance.