AUTHORS
Natalia Timofeeva, MD, Evan Cohen, PhD2, Lakesla R. Iles, BS3, Akinori Arimura, PhD4, Masaaki Sawa, PhD5, Wissam Assaily, PhD6, William G. Wierda, MD, PhD7, Varsha Gandhi, PhD8
AFFILIATIONS
1 MD Anderson Cancer Center, Houston, Texas, United States
2 MD Anderson Cancer Center, Houston, Texas, United States
3 MD Anderson Cancer Center, Houston, Texas, United States
4 Carna Biosciences, Inc., Kobe, Japan
5 Carna Biosciences, Inc., Kobe, Japan
6 AbbVie Bay Area, South San Francisco, California, United States
7 MD Anderson Cancer Center, Houston, Texas, United States
8 MD Anderson Cancer Center, Houston, Texas, United States, MD Anderson Cancer Center, Houston, Texas, United States
ABSTRACT
Background
BTK inhibitors (BTKi) and BCL-2 inhibitors (BCL-2i) are highly effective in chronic lymphocytic leukemia (CLL) when used sequentially. However, acquired resistance remains a major problem, driven by therapy-emergent BTK and BCL-2 mutations, often with TP53, NOTCH1, and SF3B1 co-mutations. Genotype alone poorly predicts drug response, as distinct BTK kinase-domain mutations can converge on shared cross-resistance patterns to covalent and non-covalent BTKi. We applied ex vivo drug profiling to primary relapsed/refractory (R/R) CLL samples to define the functional consequences of BTK and BCL-2 resistance mutations and identify combinations that can overcome resistance.
Methods
Peripheral blood samples were collected from 31 R/R CLL patients progressing on BTKi and/or BCL-2i therapy under an IRB-approved protocol. We profiled 31 primary samples including 11 patients with BTK mutations (C481x, T474x and L528W), 2 with BCL-2 mutations (G101V, A113G), and 5 with co-occurring BTK/BCL-2 mutations. Samples were treated ex vivo with covalent and non-covalent BTKi (ibrutinib, pirtobrutinib, docirbrutinib), BCL-2i (venetoclax, sonrotoclax, lisaftoclax, surzetoclax), and the MCL-1 inhibitor AZD5991, as single agents and pairwise combinations. Viability was measured by Annexin V/Propidium Iodide staining with CD19/CD5 gating at 24, 48, and 72 hours, and IC50 values of BCL-2i were determined by CellTiterGlo assay. BCL-2 family protein expression (BCL-2, BCL-XL, MCL-1) was assessed by immunoblot. Unsupervised clustering (Principal component analyses) stratified samples by drug response.
Results
Clustering identified three functional response groups: a sensitive cluster enriched for wild-type (WT) BTK/BCL-2 with PLCG2 and SF3B1 co-mutations, an intermediate cluster with BTK C481x/non-C481x (T474I, I432I), G113A BCL-2 and NOTCH1 mutants, and a resistant cluster defined by BTK L528W or BCL-2 G101V. BTK-WT samples were significantly more sensitive to venetoclax (p=0.03) and AZD5991 (p=0.05) than BTK-mutant samples. BTK mutation status affected sensitivity to all BCL-2i tested: BTK-mutant samples exhibited increased mean IC₅₀ values compared with BTK-WT samples, with 10.8-fold, 165-fold, 45-fold, and 24.6-fold increases observed for venetoclax, sonrotoclax, surzetoclax, and lisaftoclax, respectively. Newer agents (sonrotoclax, surzetoclax) retained lower IC50 than venetoclax and lisaftoclax in both genotypes. BCL-2 family protein expression was heterogeneous even within a single BTK mutation group, with variable MCL-1, BCL-2, and BCL-XL levels among samples of the same genotype. To determine whether pairing BTKi with BCL-2 or MCL-1 inhibition could overcome single-agent resistance, we calculated Δ-viability (mean combination viability - single agent) for each BTKi paired with venetoclax or AZD5991, stratified by mutation-defined groups. BTK-WT samples remained highly sensitive to single-agent venetoclax and AZD5991, with minimal additional viability reduction across all combinations, reflecting already-effective monotherapy. Venetoclax-based combinations with ibrutinib or docirbrutinib produced consistent viability reduction across BTK non-C481x samples (Δ-viability up to -24%). In BTK L528W and BCL-2-mutant samples, venetoclax-based combinations produced a smaller absolute reduction in viability than other BTK-mutant clones, with benefit largely restricted to ibrutinib and docirbrutinib pairings (Δ-viability up to -16.5%). This suggests that a BCL-2i switch, rather than BTKi pairing alone, may be needed to overcome resistance in these groups. AZD5991 combinations, in contrast, produced more uniform benefit across BTK and BCL-2 genotypes, with strongest synergy for AZD5991 plus docirbrutinib or ibrutinib, yielding average Δ-viability reductions of approximately -29% in non-C481, -22% in L528W, and -20 to -21% in G101V and dual BTK/BCL-2 mutant samples.
Conclusions
Our ex vivo profiling of a clinically annotated R/R CLL cohort shows that acquired BTK and BCL-2 mutations produce convergent resistance signatures that cannot be predicted by genotype alone. BCL-2 G101V and BTK L528W define a distinct high-resistance phenotype, rescued preferentially by MCL-1-directed combinations rather than venetoclax-based regimens. These findings support combining functional drug profiling with genomic data to the mechanistic rationale for a prospective biomarker-driven combination strategy and warrant further validation in a larger prospective cohort.
DISCLOSURES
Natalia Timofeeva, MD: No relevant conflicts of interest to declare.