Distinct archetypes of clonal dynamics underlie response and resistance to diverse CLL therapies Free

Patients with chronic lymphocytic leukemia (CLL) experience variable clinical course and duration of therapeutic response. While prior studies have shown that specific genetic drivers can shape leukemia growth kinetics and influence the natural course of CLL, the longitudinal patterns and genetic determinants of clonal evolution underlying response and resistance to time-limited therapies are not fully defined. To address this, we performed longitudinal analyses of 107 patients treated with time-limited frontline CLL therapies, either chemotherapy or chemoimmunotherapy (CIT, n=62), or fixed-duration venetoclax-obinutuzumab (VO, n=30) or ibrutinib-venetoclax (IV, n=15). We combined quarterly monitoring of measurable residual disease (MRD) levels with genetic characterization of CLL from blood samples collected serially before and during therapy, at MRD sampling time points after therapy, and at clinical relapse. A median of 6 samples were genetically characterized per patient (range 4 to 15), including a median of 4 post-therapy MRD samples (range 2-11). Genetic analysis of MRD samples was carried out using ultra-deep, patient-specific targeted sequencing focusing on mutations representing distinct subclones of each CLL. Subclones were defined based on branches of the phylogenetic tree inferred using PhylogicNDT from paired pretreatment and relapse whole-exome sequencing data. Altogether, 3058 baits capturing subclone-specific mutations (median 6 baits/subclone) were deployed to track 561 subclones (median 5 subclones/CLL) across intervening MRD time points. We used duplex sequencing to reduce sequencing errors at MRD time points and obtained an average duplex depth of 2870x (range 156-5181x) (~110,000x raw depth) per sample, with individual variants reaching duplex depths >10,000x allowing detection of subclones with cancer cell fractions (CCF) of ≥10% at MRD ≥4 x 10^-3. We integrated serial white blood cell counts, MRD and clone-specific CCF data and applied the Markov Chain Monte Carlo method to model decay and repopulation rates of each subclone during and after therapy, respectively. Overall, we discerned 4 archetypes of clonal dynamics present across the CIT and VO cohorts. Archetypes 1-3 were eachmarkedby the outgrowth of a particular subclone while on therapy, consistent with their relative insensitivity to treatment, but differed with respect to the post-therapy repopulation kinetics of the therapy-insensitive subclone relative to other subclones. For Archetype 1 (26% CIT; 33% VO), the therapy-insensitive subclone remained stably dominant after therapy and was the sole basis of CLL relapse. While chemoresistant subclones typically harbored mutations in canonical CLL driver genes (primarily TP53, ATM, SF3B1, POT1, CHEK2, SAMHD1, IKZF3 and DIS3), therapy-insensitive subclones in VO-treated cases exhibited greater genetic heterogeneity, encompassing mutations in CLL drivers such as SPEN and ARID1A,mutations in cancer drivers not recurrently seen in CLL, and copy-number events. Archetype 2 (35% CIT; 48% VO) was marked by the continued expansion of the therapy-insensitive subclone, reflecting its accelerated regrowth kinetics relative to other subclones. After CIT, these subclones typically harbored chemoresistance-conferring mutations, while after VO, repopulating subclones were genetically diverse and carried mutations in such genes as NFKBIE, SPEN, NOTCH1, BIRC3, MGA, DYRK1A and RFX7. In contrast, Archetype 3 (21% CIT; 14% VO) was typified by the initial therapy-insensitive subclone being outcompeted by more therapy-sensitive ones that rapidly regrew, driving relapse. For example, in 3 of 5 patients where TP53-, ATM-, or SF3B1-mutant subclones were relatively sensitive to VO, CLL relapse coincided with preferential post-therapy expansion of these subclones, in line with their fitness advantage during disease progression. Finally, Archetype 4, observed predominantly in CIT-treated patients (18% CIT; 5% VO), featured emergence and expansion of novel subclones post-therapy, presumably induced by new mutation events, that were undetectable at the time of treatment initiation. Analysis of the IV cohort is in progress. Collectively, these archetypes delineate the diverse evolutionary trajectories that CLL subclones can follow post-therapy, providing a framework for understanding CLL relapse and guide strategies to anticipate clonal evolution, counteract subclonal selection, and ultimately prevent relapse.