Abstract T048

A Reproducible Mutation- versus Copy-Number-Enriched Genomic Axis Underlies Molecular Subtypes of Classic Hodgkin Lymphoma

Background: Several independent groups have recently proposed genomic subtype classifications of classic Hodgkin lymphoma (cHL), either by ctDNA profiling or analysis of microdissected tumor tissue: two subtypes by Alig et al. (Nature 2023, H1/H2), four by Aoki et al. (Cancer Discov 2026, CST/CN913/STB/CN2P), three by Heger et al. (JCO 2024), and two by Pirosa et al. (Blood 2025, C1/C2). Whether these independently derived systems capture the same biology or identify orthogonal structure remains unknown.

Methods: We applied the BC Cancer (Aoki) subtype classifier to 293 cHL patients from the Stanford ctDNA targeted sequencing cohort (Alig et al., Nature 2023). Subtype concordance was assessed by cross-tabulation, cluster enrichments by Fisher's exact test, and prognostic associations by log-rank test. Co-association analysis of cluster-defining features by observed vs expected (O/E) frequencies was performed across all 4 classification systems.

Results: The BC Cancer classifier resolved the H1/H2 axis into four subtypes distributed as CST (39%), STB (23%), CN913 (15%), CN2P (5%), and unclassified (18%), with CST and STB (mutation-dominated) mapping predominantly within H1 (87% and 80% H1, respectively) and CN913 and CN2P (copy number aberration [CNA]-dominated) within H2 (73% and 63% H2; p<0.00001). CN913 was enriched for older age (median 59 vs 29 years for CST/STB; p=0.002) and EBV positivity (60% vs 15% for CST/STB; p<0.00001). PFS differed significantly across subtypes (p=0.046), with inferior outcomes in CN913/CN2P vs CST/STB. Co-association analysis of cluster-defining features showed strong within-cluster co-enrichment (log2 O/E CST=0.60, CN913=0.49, STB=0.31, CN2P=-0.03), with mutual exclusivity largely restricted to features crossing the mutation- versus CNA-dominant axis. Co-association analyses using defining features from the Heger and Pirosa classifications also each independently reproduced the same CNA- versus mutation-dominated pattern.

Conclusion: A binary mutation- versus CNA-dominant axis represents the primary organizing feature of cHL tumor genome biology and appears highly reproducible across independent studies supporting at least two molecular subtypes. More granular models may provide additional resolution within this framework, with their additional clinical impact an area of active investigation. Harmonization efforts across studies are warranted to define a consensus classification for prospective clinical application.