Introduction: Classic Hodgkin Lymphoma (CHL) is currently classified into four subtypes based on histomorphologic characteristics. However, additional molecular features might help improve disease taxonomy to guide treatment strategies and provide insights into treatment response. Here, we aimed to uncover disease heterogeneity and develop a new classification framework based on multi-dimensional profiling.
Methods: We performed whole exome/targeted sequencing on enriched HRS cells from 116 fresh-frozen CHL biopsies at BC Cancer. In addition, we constructed tissue microarrays from the same cohort and performed GeoMx® Whole Transcriptome Assay of HRS cells and imaging mass cytometry to delineate the spatial tumor microenvironment (TME) ecosystem.
Results: Mutation and copy number analyses identified known recurrent driver events including mutations and copy number changes in SOCS1, STAT6, TNFAIP3, B2M, REL, and the PDL1 locus. ZNF217 mutations was significantly associated with progression-free survival (PFS) (P<.01), and STAT6 mutation +/- amplification was the most significant feature associated with unfavorable PFS in younger patients (< 45) (P=.013).
To define molecular subtypes of CHL, we applied non-negative matrix factorization consensus clustering and discovered four robust subsets of tumors (clusters) using recurrent genomic events; Cluster1 (C1): mutations in TNFAIP3 and CSF2RB, younger age and loss of MHC-I, C2: old age, EBV and upregulation of the IFN-g pathway; C3: REL and STAT6 gain, and upregulation of a DNA repair signature; and C4: mutations in STAT6 and B2M. TME analyses further identified correlations between each mutational NMF cluster and TME composition (Figure): C1:FOXP3+Tregs, C2:LAG3+Tregs and CD68+macrophages, C3: PD1+CD4+Tcells, C4 = no correlation. We then translated our mutational clustering model into a ctDNA-based classification assay using independent validation cohorts from BC Cancer/UHN (N = 78) and Stanford (Alig et al, Nature 2024), and validated the robustness of our model and correlations with clinical features; C2: EBV (P=6.30E-04); and C4: younger age (P=.037).
Conclusion: Our multi-dimensional profiling approach delineated molecular profiles of HRS cells linking mutational clusters to distinct TME patterns. These linkages have implications for molecular subtyping of CHL, and cellular vulnerabilities that might be therapeutically exploitable via targeting of HRS cell phenotypes and/or immune escape mechanisms.
Tomohiro Aoki, Gerben Duns, Shinya Rai, Andrew Lytle, Yifan Yin, Aixiang Jiang, Stefan K. Alig, Mohammad Shahrokh Esfahani, Clementine Sarkozy, Stacy Hung, Katy Milne, Adele Telenius, Makoto Kishida, Michael Li, Luke O` Brien, Celia Strong, Talia Goodyear, Juan Patino Rangel, Michael Hong, Shaocheng Wu, Katsuyoshi Takata, Tomoko Miyata-Takata, Merrill Boyle, Susana Ben-Neriah, Andrew P. Weng, Andrew Roth, Michael Crump, John Kuruvilla, Anca Prica, Robert Kridel, Brad H. Nelson, Pedro Farinha, Ash A. Alizadeh, Kerry J. Savage, David W. Scott, Christian Steidl