Dissecting pirtobrutinib resistance in Mantle Cell Lymphoma through single-cell multi-omics

Abstract

Abstract Introduction Although the non-covalent BTK inhibitor Pirtobrutinib (PBN) has demonstrated clinical efficacy in treating relapsed and refractory mantle cell lymphoma (MCL) patients, resistance emerges. Inter-patient genetic variability and tumor intrinsic plasticity give rise to diverse resistant phenotypes under therapeutic pressure, posing a major challenge to understanding and overcoming drug resistance. Methods To dissect the molecular dynamics driving PBN resistance, we performed integrative single-cell multi-omic profiling on pre-treatment and relapsed primary MCL patient samples. Our approach combined single-cell RNA sequencing (scRNA-seq) to measure gene expression, single-cell assay for transposase-accessible chromatin using sequencing (scATAC-seq) to assess chromatin accessibility, and single-cell DNA sequencing (scDNA-seq) to define genomic alterations, including copy number variations and clonal architecture. Results We analyzed 24 peripheral blood or apheresis samples collected from 13 MCL patients at both drug-sensitive and resistant timepoints, yielding a high-quality dataset of 113,143 single cells. Malignant B cells, marked by high levels of CD79A, CCND1, MS4A1, and IGHM, comprised 50.28% of the dataset (n= 56,889). Our analyses revealed both genetic and non-genetic routes to resistance. In some patients, resistance involved stepwise acquisition of copy number gains, with 2p and 8q gains detected in early resistant clones and additional 1q gains emerging in late-stage resistant clones (adjusted p < 0.05). Polygenetic tree reconstructed from scDNA-seq data confirmed this clonal evolution, and metaphase karyotyping further validated these findings. In contrast, some patients showed no large-scale genomic alterations and developed resistance through transcriptional and epigenetic reprogramming alone, reflecting non-genetic adaptation to therapeutic pressure. Pathway analysis revealed enrichment of NF-κB signaling and gene programs associated with chromosome segregation, mitosis, and protein localization to chromosomes (enrichment ratio >0, adjusted p < 0.05). Integration of scATAC-seq and scRNA-seq enabled gene regulatory network inference. RAD21, a core component of the cohesin complex, was identified as a key regulator by binding to 987 resistance-specific DNA regions and regulating 816 genes, potentially driving PBN resistance (adjusted p < 0.05). Simulated in silico knockout of RAD21 using a random forest model shifted resistant cells toward a sensitive-like transcriptional state, supporting RAD21 as a candidate therapeutic target for overcoming PBN resistance. Additionally, we identified a stem-like malignant B cell population enriched in resistant samples, displaying features of metabolic reprogramming and epithelial-mesenchymal transition. Within this population, two distinct subclusters emerged. Cluster 1 demonstrated epithelial-like, progenitor-associated, while cluster 2 showed mesenchymal-like features and metabolic dysregulation (enrichment ratio >0, adjusted p < 0.05). This aligns with a recently proposed model published in Nature describing a resistance continuum, in which treatment-resistant tumors are not static but contain a range of stem-like states with varying degrees of lineage plasticity and metabolic flexibility. Conclusion In summary, our study presents the first longitudinal single-cell multi-omic characterization of PBN resistance in primary MCL patients. We identified coordinated transcriptional and epigenetic remodeling during resistance evolution across patients. Our integrative analyses highlight RAD21 and the cohesin complex, along with stem-like heterogeneity, as potential drivers of therapeutic resistance.