Clonal Hematopoiesis Mutations and Cell-Free DNA Characteristics in Patients with High-Risk Large B-Cell Lymphomas

Introduction: The prevalence and clinical impact of clonal hematopoiesis (CH) mutations in peripheral blood (PB) of patients with aggressive large B-cell lymphomas (LBCL) has remained poorly characterized. Moreover, CH mutations partly overlap with lymphoma driver mutations, potentially compromising the specificity of emerging biomarker assays for LBCLs based on plasma circulating tumor DNA (ctDNA). Hence, we set out to discover the impact of CH mutations in PB of young high-risk LBCL patients and correlated the results with clinical characteristics, outcome, and ctDNA features. CH and lymphoma mutations were traced longitudinally in PB and cell-free DNA (cfDNA) compartments.

Methods: We sequenced 65 myeloid driver genes in PB samples of 163 healthy donors and 247 primary high-risk (94% with age-adjusted IPI ≥ 2) LBCLs patients treated in Nordic Lymphoma Group (NLG) phase II protocols with anthracyclin-based intensified immunochemotherapy and central nervous system prophylaxis (NLG-LBC-04, -05 and -06). The patients were young (median age, 55 years; range, 19-64) and the median follow-up was 47 months. We included CH mutations with variant allele frequency (VAF) of 0.5% for single-nucleotide mutations and 1% for frameshift variants. ctDNA data was available from 227 patients (Meriranta et al. 2022, Leppä et al. 2024), and CH mutations were evaluated in pretreatment and follow-up plasma cfDNA of 126 patients with targeted sequencing incorporating duplex error correction. Associations between PB and cfDNA-derived VAFs were analyzed using linear regression. Uni- and multivariable analyses were performed to evaluate associations between CH mutations, patient phenotypes, and outcomes.

Results: CH mutations were found in 38% (95/247) of the pretreatment PB samples from lymphoma patients with a median VAF of 1.2% (range 0.5-41%). Correspondingly, 37% (61/163) of the healthy donors had a CH mutation. Lymphoma patients were more likely to have a CH mutation compared with healthy donors when adjusting for age and sex (OR=1.7, P=0.023). Of the most recurrently mutated genes, TET2 mutations were enriched in LBCL patients compared with healthy donors both in univariable (P=0.016) and multivariable models (OR=4.4, P=0.004).

All 62 CH mutations with cfDNA available were also detected in cfDNA using force genotyping. Notably, two variants in EZH2 and one in TP53 showed strikingly high cfDNA VAFs compared with whole PB VAFs reflecting lymphoma driver mutations, and were excluded from the subsequent analyses for the clinical associations with CH mutations. For the remaining CH mutations, VAFs in PB showed a strong association with cfDNA-derived VAFs (R2=0.74, P<0.001).

The presence of CH mutations was associated with older age (median, 58 vs 52 years, P<0.001). We observed no differences in other clinical characteristics, such as sex, symptom burden, IPI score, or ctDNA burden, between patients with and without CH mutations. Furthermore, we observed no difference in overall survival (OS) based on the presence of CH mutations in the univariable (P=0.87) or the multivariable model adjusting for age, sex, and IPI score (P=0.66). Also, we detected no difference in progression-free survival by CH mutation status (P=0.84). Patients with TP53 mutations in PB at diagnosis had inferior OS compared to patients without TP53 mutations (3-year OS 50% vs 88% P=0.04).

Out of 43 LBCL patients with follow-up PB samples available, seven (16%) showed the emergence of new TP53/PPM1D clones (median VAF, 1.6%) following therapy. DNMT3A and TET2 mutations showed stable clones during follow-up (median VAF change, 0.1% and -0.1%, respectively). In 25 patients with pretreatment and end-of-therapy cfDNA samples available, DNMT3A, TET2, and ASXL1 mutations remained stable or showed decrease in clone size (median VAF change, 0.9%, -0.7%, -3%, respectively), whereas TP53 mutation VAFs increased in cfDNA (median VAF change, 1.7%) despite the clearance of the lymphoma ctDNA.

Conclusion: Collectively, our results suggest that CH mutations may influence cfDNA characteristics, with potential implications for longitudinal use of cfDNA-based analyses in clinical contexts. CH mutations were not associated with survival; however, TP53 mutations in PB conferred worse OS. Additional studies are warranted to assess the long-term impact of CH mutation dynamics on outcomes, such as the risk of therapy-related myeloid malignancies.