Abstract
Background: Although cHL patients who receive standard therapy have a high rate of event-free and overall survival, 5% - 10% of patients will be refractory to initial therapy and up to 1/3 will relapse. Thus, improved prognostication and identification of new treatment targets are needed. High throughput sequencing can identify recurrent somatic mutations that potentially drive lymphomagenesis and impact treatment response. However, Hodgkin-Reed-Sternberg (HRS) cells have a low (~1%) abundance in cHL biopsies, creating a challenge for comprehensive detection of somatic mutations in bulk lymphoma biopsies. We hypothesized that ultra deep exome sequencing can address this challenge, by providing a more widely available approach to HRS cell characterization compared to HRS cell purification.
Methods: We performed exome sequencing on 32 fresh frozen samples from 31 cHL patients obtained prior to treatment (27) or after relapse (4) with paired normal skin samples (31). Sequencing was performed using the Illumina HiSeq platform (2 x 150bp reads) with a >1,000X median coverage goal. Sequence data was aligned to GRC38 and SNVs and INDELs were called with multiple algorithms. We employed several variant filtering strategies, including manual review, to remove common polymorphisms, sequencing artifacts, and false positives.
Results: In total 7,000Gb were sequenced across all samples with an average of 1,100Gb/sample. The median depth of coverage was 1,009X for lymphoma and 926X for normal tissue. After filtering, 4,734 somatic SNVs and INDELs were identified, with an average of 33 protein-coding variants/case, excluding a single hyper-mutated case. The hyper-mutated patient had 3,684 variants, several of which are in DNA mismatch repair genes including: EXO1, MLH3, and MUS81 . We identified 296 genes with mutations observed in two or more patients. We confirmed previously identified genes with recurrent mutations in cHL (e.g., TNFAIP3 [18.8%], B2M [12.5%], XPO1 [9.4%]). We identified one B2M start loss mutation, which is consistent with other studies (Reichel et al. Blood 2015). However, our overall detection of B2M mutations was lower than previous reports (70%; Reichel et al. Blood 2015). Several recurrently mutated genes in our dataset are not well characterized in cHL and may contribute to disease development or progression. IGLL5 [21.9%] has been shown to be mutated in DLBCL and CLL, with a mutation pattern consistent with off-target activation-induced deaminase (AID) activity. ITPKB [12.5%], which functions as a negative regulator of PI3K, has been recently discussed as a tumor suppressor in DLBCL and detected in cfDNA from cHL patients. RYR1 [15.6%] encodes a Ca+ channel expressed in normal B cells, and was previously reported as recurrently mutated in DLBCL, Burkitt, and mantle cell lymphoma. Our work confirmed that components of the JAK/STAT pathway including SOCS1 [18.8%] and STAT6 [9.4%] are mutated in cHL. We also identified mutations associated with SWI/SNF chromatin remodeling complexes (e.g., BCL7A [6.3%], ARID1A [6.3%], SMARCA4 [3.1%], SMARCA2 [3.1%], SMARCD1/2 [3.1%]), which have not previously been reported as recurrently mutated in cHL. Finally, we observed mutations in 11 histone genes (e.g., HIST1H1B [6.3%]), extending the importance of histone gene mutations in lymphomas from germinal center B cells.
Conclusions: Recurrent somatic mutations withincHL were identified via ultra deep exome sequencing, providing a strategy to identify somatic variation that does not require HRS cell purification. We have confirmed known recurrent genes within cHL and identified previously unreported mutated genes and pathways. These data suggest that cHL genomes harbor recurrent SNVs and INDELs that can inform new targets for cHL treatment and prognostication
Mehta-Shah: Celgene: Research Funding; Verastem: Research Funding; Bristol Myers-Squibb: Research Funding. Kahl: Gilead: Consultancy; Celgene: Consultancy; ADC Therapeutics: Research Funding; Seattle Genetics: Consultancy; Genentech: Consultancy.
