Systematic molecular profiling to identify determinants of response to ibrutinib

Abstract

Abstract BACKGROUND Whilst the broad clonal architecture of naturally progressing chronic lymphocytic leukemia (CLL) has been described, a comprehensive picture of how chemotherapy and targeted agents reshape that landscape is lacking. Here we integrate clone-specific growth kinetics with mutational profiles, transcriptomic subtypes, and CpG-methylation-based epigenetic classes to capture the multidimensional evolutionary responses of CLL under native conditions and during treatment. METHODS We previously reported a systematic whole-exome sequencing (WES) analysis of 417 leukemia–germline pairs from 169 treatment-naïve patients ≥ 65 y (Karisani et al., Blood 2024). The treatment arm comprised 83 patients who received ibrutinib and 39 who received chlorambucil in the RESONATE-2 trial (NCT01722487). Forty-seven age-matched “watch-and-wait” patients from the CLL Research Consortium served as untreated controls. Peripheral-blood samples were collected at baseline and again ~300 d and ~600 d after therapy initiation (or diagnosis in the watch-and-wait cohort), enabling longitudinal clonal tracking. WES data were processed with PhylogicNDT to reconstruct clonal architectures, define phylogenies, and identify significant clonal shifts.  MutSig tools identified mutations enriched in expanding or contracting clones. Pretreatment RNA-sequencing profiles were grouped into transcriptional subtypes by consensus clustering based on the Louvain algorithm. Differentially methylated regions plus epitypes were called from reduced representation bisulfite sequencing data. RESULTS Across the 169 patients with WES, we resolved 579 subclonal clusters. Significant clonal shifts (defined as a distribution shift of cancer cell fraction in a subclone of >95% between two timepoints) were uncommon during watch-and-wait (36%) but markedly higher after therapy — chlorambucil  67% (q = 0.01), and ibrutinib 77% (q < 10-4). In ibrutinib-treated CLL, regressing subclones were enriched for KRAS (q<10-3) and SF3B1 mutations (q<10-3), whereas expanding subclones preferentially harboured mutations in BIRC3 (q<10-4), NOTCH1 (q <10-3),  TP53 (q=0.02), and POT1 (q=0.098). In chlorambucil-treated patients, no specific mutations were found to be enriched in expanding or regressing subclones. Using the recently defined CLL expression clusters (ECs) in patients treated with chemoimmunotherapy as defined by Knisbacher et al. (Nat Genet 2022), we were unable to identify any relationship between ECs and outcomes. We therefore sought to identify de novo ECs with prognostic relevance in the context of BTK-inhibition (BTKi-ECs). We identified a transcriptionally defined patient subgroup (C4, n=20) associated with inferior progression-free survival (median PFS = 63 months, log-rank q=0.05), independent of known prognostic clinical and molecular features, including epitypes, in multivariate analysis (Cox HR=3.7, 95% CI: 1.6-8.6, q=0.02). Gene-set enrichment in this cluster highlighted TNF-ɑ signalling via NF-κB (REL, TNFAIP3, NFKB2) and G2M checkpoint pathways (SMC4, EZH2). We also identified a distinct transcriptionally defined patient subgroup (C3, n=18) that was enriched for patients harboring at least one contracting subclone and no expanding subclones (q=0.005). This cluster showed a relative depletion of mutations within the chromatin modification pathway (6% vs 50%, q<0.01). Gene-set enrichment in this cluster highlighted a downregulation of genes in the TNF-ɑ signalling via NF-κB pathway (SOCS3, ICOSLG, NFKB2). CONCLUSION BTK inhibition and chemotherapy drive distinct evolutionary trajectories in CLL. Multi-omics profiling reveals mutation-defined subclones with divergent sensitivity to targeted therapy and delineates two informative expression states: a high-risk NF-κB/inflammatory cluster (C4; poor PFS) and a cluster enriched with contracting subclones with depleted TNF-ɑ signalling via NF-κB (C3). Integrating clonal growth kinetics with genomic, transcriptomic, and methylation data can potentially identify patients likely to progress on BTK inhibitors and identify actionable pathways for adaptive or combination therapy.