Tracing Clonal Hematopoiesis and Tumor-Specific Mutations in Hematopoietic Progenitors of B Cell Non-Hodgkin Lymphoma (NHL) Patients

While most lymphomas are mature B-cell neoplasms deriving from germinal center B-cells, a stem cell origin has been debated for indolent NHL (iNHL). Lymphoma-initiating stem cell pools may serve as reservoirs for further clonal evolution, contributing to frequent relapses in iNHL such as follicular lymphoma (FL). Thus, identifying and characterizing lymphoma-initiating hematopoietic progenitors is crucial. Here, we map the molecular ontogeny of NHLs and profile genetic precursor lesions potentially contributing to lymphomagenesis.

First, we obtained genetic profiles of tumor/normal pairs of 25 aggressive NHLs (aNHL: 18 DLBCL, 4 MCL, 2 PMBCL, 1 BL) and 13 iNHLs (12 FL, 1 MZL) by whole exome sequencing (274x). iNHLs exhibited lower mutation numbers than aNHLs (41 vs 63), fewer chromosomal aberrations (3 vs 7) and a higher prevalence of mutations in chromatin modifying genes, particularly KMT2D (OR 37; CI 2.9-471.7; p=.005). Moreover, cancer cell fraction analysis identified mutations in KMT2D, EZH2 and CREBBP as clonal events in at least 80% of patients.

We then analyzed clonal hematopoiesis (CH) as a potential predisposition for lymphomagenesis by sequencing 45 CH-associated genes in whole blood lysates (3470x). CH mutations were found in 55% of our cohort, with significant enrichment in iNHL (p=.005). Most frequently affected genes were DNMT3A and TET2 (42%; median VAF [VAF_med] 2.4% and 4.6%). Identified CH mutations were screened in flow-sorted peripheral blood (PB) and bone marrow (BM) populations using ultra-deep targeted amplicon sequencing (12226x). Lymphoma patients showed lymphoid skewed propagation compared to non-NHL patients (n=39), with higher VAF_med for PB B-cells (6.4% vs 3.8%) and significantly lower VAF_med for PB monocytes (2.7% vs 13%; p=.027). In lymphoma patients, DNMT3A mutated clones showed clonal shrinkage (VAF ratio≤1) in both myeloid [monocytes/myeloid progenitors] and lymphoid [B-cells/lymphoid progenitors] lineages, suggesting impaired hematopoietic differentiation. Conversely, TET2 mutations showed clonal expansion (1.2-66.4x) in both lineages. Notably, one TET2 and one EZH2 mutation were also part of the dominant clone in the respective FLs, indicating their participation as lymphoma-initiating events. Thus, we hypothesized that other lymphoma-specific mutations might also be detectable along the hematopoietic differentiation tree.

Selected tumor-specific mutations (range: 21-45 per patient) were analyzed in mature PB populations (CD19⁺ B-cells, CD3⁺ T-cells, CD56⁺ NK-cells, CD14⁺ monocytes) for 7 aNHL and 7 iNHL (all FL) cases. Strikingly, lymphoma-specific mutations were detected in 6 out of 7 FL cases, with 58% of screened mutations found in at least one cell fraction, compared to only 1% in aNHLs. Mutations were mainly detected in B-cells (8.3%-92.9%; VAF_med 2-34%), with one patient harboring low VAF mutations in T-cells (VAF_med 3.5%). Next, we analyzed hematopoietic stem/progenitor cells (HSPCs) from FL patients in the BM. We excluded B-cells displaying light chain expression matching the lymphoma phenotype to account for potential circulating FL cells. Mutation fractions increased from CD34⁺/C38^-/CD45RA^- multipotent progenitors (3%; VAF_med 1.3%) to CD34⁺/CD38^-/CD45RA⁺ lymphoid progenitors (31%; VAF_med 4%), CD34^-/CD38⁺/CD19⁺ pre B-cells (42%; VAF_med 8%) and B-cells (49%; VAF_med 24.4%). We identified three potential FL origins: 4 patients had mutated multipotent or lymphoid progenitors, 2 patients showed mutation acquisition in committed pre- or mature B-cells, and 1 case had no detected mutations, suggesting all mutations were gained at a single stage of maturation, similar to aNHL. Early mutations in HSPCs affected lymphoma-associated genes i.e. KMT2D, IGLL5, BTG2 and GNAI2. We conducted single-cell RNA sequencing combined with genotyping of selected mutations by long-read sequencing (nanoranger) of BM from FL patients with suspected HSPC involvement. This allowed us to confirm the presence of early mutations in HSPCs (1.5-30% mutated out of 354 genotyped cells).

Our study demonstrates that, unlike aNHL, FL is often preceded by CH and the accumulation of somatic mutations in the normal HSPC compartment, setting the stage for neoplastic transformation. By identifying mutated HSPCs we lay ground for the characterization of FL-initiating cells, which may provide new opportunities for their therapeutic targeting.