A Comprehensive Genetic Study of Classical Hodgkin Lymphoma Using Ctdna

Introduction. We leveraged advanced analytical methods focusing on ctDNA to provide an in-depth overview of the genetic landscape of classical Hodgkin lymphoma (cHL) and its connection to disease pathophysiology and clinical course. Methods. Our study investigated somatic mutations, somatic copy number abnormalities (SCNA), fusions, and whole genome duplication (WGD) in ctDNA from 297 patients enrolled in the IOSI-EMA003-NCT03280394 (treatment-naïve - TN: 215; relapsed/refractory - RR treated with checkpoint inhibitors: 37) and FIL-RougeBIO-NCT05066555 studies (N=45, all TN) using LyV4.0 CAPP-seq. Results. The genes most frequently affected by non-synonymous mutations were SOCS1, TNFAIP3, STAT6, B2M, GNA13, and ITPKB. When considering SCNA, we identified gains in 2p16.1 (REL, XPO1), 3q29 (BCL6), 7q36.3 (EZH2), 9p24.2 (JAK2, CD274, PDCD1LG2), 8q24.23 (MYC), 15q26.3 (IGF1R), and deletions in 1p36.13 (TNFRSF14), 1p12 (CD58), 3q27.1 (KLHL6), 6q23.3 (TNFAIP3), 13q13.3 (FOXO1), 14q12 (NFKBIA), 15q25.3 (B2M), 16p13.3 (CREBBP), 20q13.13 (PTPN1). Recurrent fusions were observed in 32% of cases. No mutations or SCNA were found to correlate with reduced progression-free survival (PFS) in TN cHL. WGD was detected in 24% of patients, which is higher than the frequency observed in other mature B-cell malignancies. cHL with WGD showed a higher incidence of CCNE1 gain, which affects the checkpoint associated with polyploidization. In contrast, multi-hit TP53 lesions or RB1 abnormalities were infrequent and did not correlate with WGD. WGD was significantly linked to a shorter PFS in TN cHL of the IOSI-EMA003 training cohort by univariate and multivariate analyses (HR: 2.4; p=0.02) and further confirmed as a prognostic factor for PFS in the FILRougeBIO validation cohort (HR: 8.9, p=0.01). EBV infection was detected in 18% of cases. EBV+ cases were mostly mutated in SOCS1, infrequently mutated in STAT6, and lacked the loss of PTPN1 and NFKBIA. EBV+ cHL exhibited a significantly lower burden of mutations and SCNA. Two major classes of microenvironments emerged through digital cytometry deconvolution and tissue microarray analysis of cHL biopsies. One class was marked by macrophages (52% of cases) and the other by T-cells and immune checkpoints (48% of cases). Mutations, SCNA, and WGD were evenly distributed across both microenvironmental classes. Conversely, cHL cases with a T-cell-enriched microenvironment had fewer subclonal MHC-I neoantigens, indicating selective pressure exerted from T-cells. Consistently, RR cHL treated with checkpoint inhibitors had longer PFS if they lacked subclonal MHC-I neoantigens. LyV4.0 CAPP-seq together with NMF clustering, reproduced the classification of the C1-C5 genetic subtypes in 235 untreated DLBCL patients. In cHL, two clusters were identified using the same approach. Although the two clusters had similar baseline characteristics, EBV infection rates, ctDNA load, and outcomes, the two clusters differed in their underlying mechanisms of genetic instability. One cluster was defined by aneuploidy and a higher number of SCNA (36% of cases), while the other was characterized by mutations in genes targeted by AID-hypermutation (64% of cases) and had a higher fraction of mutations associated with AID activity signatures (SBS39, SBS84, SBS85). Both clusters shared hallmark lesions of cHL, including STAT6 mutations, and gains in 2p/2p16.1 and 9p/9p24.2. In multivariate regression analysis, MTV and WGD were the most important variables influencing pre-treatment ctDNA quantity. Conversely, the fraction of tumor cells and proliferating cells in the biopsy, and levels of DNASE1L3 showed minimal importance in explaining ctDNA concentration. While we could confirm pre-treatment ctDNA levels above 2.5 Log10 hGE/ml of plasma as a prognostic factor for DLBCL patients, they did not impact PFS in either the IOSI-EMA003 or in the FILRougeBIO cohorts. Conclusions. WGD is the only confirmed genetic factor for predicting the prognosis of TN cHL. MHC-I neoantigen load correlates with the microenvironment characteristics and the response to checkpoint inhibitors rather than specific genetic mutations. The genetic subtypes of cHL are mainly defined by the mechanisms driving genetic instability rather than by individual genetic lesions or EBV infection. The ctDNA load has no prognostic significance in cHL.