Molecular Evolution of Classical Hodgkin Lymphoma Revealed Though Whole Genome Sequencing of Hodgkin and Reed-Sternberg Cells

Introduction: Classical Hodgkin lymphoma (cHL) is characterized by a small fraction of Hodgkin and Reed-Sternberg (HRS) tumor cells (~1%) which are surrounded by an extensive immune infiltrate. The rare nature of HRS cells limits the ability to study the genomics of cHL using standard platforms. To circumvent this, our group has optimized fluorescence-activated cell sorting to isolate HRS cells and intratumor B- and T- cells and to perform whole exome sequencing (WES; Reichel, Blood 2015). To date, however, there have been no reports on whole genome sequencing (WGS) of cHL.

Methods: We performed flow-sorting of HRS cells and WGS to define the genomic landscape of cHL including: i) mutational processes involved in pathogenesis, ii) large and focal copy number variants, iii) structural variants including complex events, iv) the sequence and evolution of molecular events in cHL. We interrogated WGS from 25 cases of cHL: 10 pediatric patients (age<18), 9 adolescents and young adults (AYA, age 18-40), and 6 older adults (age>40). Intra-tumoral T-cells were used as germline control. An additional 36 cHL cases were evaluated by WES.

Results: The average depth of coverage among the 25 WGS cases was 27.5x. After having identified and removed amplification-based palindromic sequencing artifacts, we observed a median of 5006 single base substitutions (SBS; range 1763-18436). Pediatric and AYA patients had a higher SBS burden compared to older adults (median 5279 vs. 2945, p=0.009). Five main SBS signatures were identified: SBS1 and SBS5 (aging), SBS2 and SBS13 (APOBEC), and SBS25 (chemotherapy, in a relapsed case). A dNdScv driver discovery analysis performed on the combined WES and WGS cases identified 24 driver genes including BCL7A and CISH which had not been previously reported as drivers in cHL.

An investigation of copy number alterations (CNAs) confirmed high ploidy in cHL (median 2.95, range 1.66-5.33). Whole genome duplication was identified in 64% cases. We also observed clear evidence of complex events such as chromothripsis (n=4), double minutes (dm, n=2), breakage-fusion-bridge (bfb; n=4). Some of these events were responsible for the acquisition of distinct drivers. For example, we observed one dm and one bfb responsible for CD274 and REL gains, respectively (>10 copies).

Leveraging the high prevalence of large chromosomal gains, we performed an investigation of the relative timing of acquisition of driver mutations. Clonal mutations within chromosomal gains can be defined as duplicated (VAF~66%; acquired before the gain) or non-duplicated (VAF~33%; acquired before or after the gain). Sixty-one percent (152/249) of driver genes were duplicated suggesting that they were acquired prior to large chromosomal gains. Next, we used the corrected ratio between duplicated and non-duplicated mutations within large chromosomal gains to estimate the molecular time of each duplicated segment (Rustad, Nat Comm 2020). In 11/22 genomes the final CNA profile was acquired through at least two temporally distinct events. To convert these relative estimations into absolute timing (i.e., the age at which events occurred), we leveraged the clock-like mutation signatures (SBS1, SBS5). We first confirmed that the SBS1 and SBS5 mutation rate were constant over time (R ²=0.84; p<0.0001 in Peds/AYA; R ² =0.82; p=0.002 in older adults). We observed a higher mutation rate in Pediatric/AYA cases compared to older adults (p=0.01), which is consistent with the higher mutational burden observed in this age group. By estimating the SBS1- and SBS5-based molecular time for large chromosomal gains and converting relative estimates to absolute time, we are able to estimate the age in years at the time of the first multi-chromosomal gain event. We observed that the first multi-chromosomal gain in cHL is often acquired several years before the diagnosis/sample collection: median latency of 19.5 (range 12-27) and 5.6 (range 1.8-16) years in older adults and pediatric/AYA patients respectively.

Conclusion: Here we report the first WGS in cHL. We identify novel drivers and genomic mechanisms involved in cHL pathogenesis. We found that mutations in driver genes are often acquired earlier then chromosomal gains, potentially preceding the cHL diagnosis by several years. In addition, we observed key differences in biology of cHL across age groups including accelerated mutagenesis and increased mutational burden among younger patients.