Genomic, Transcriptional, and Immunological Validation of Distinct Molecular Subtypes of Classic Hodgkin Lymphoma through Tissue-Based and Noninvasive Methods

Introduction: The scarcity of malignant Hodgkin and Reed-Sternberg (HRS) cells has hampered comprehensive genomic profiling of classic Hodgkin lymphoma (cHL) from tumor tissue. Multiple recent studies have demonstrated that plasma cfDNA profiling facilitates molecular characterization of cHL. Leveraging noninvasive genotypes and Latent Dirichlet Allocation, we recently defined 2 genetic cHL subtypes in a large international cohort comprising 366 patients including pediatric and adult patients of all ages (Alig et al., ASH 2022 and ICML 2023). Cluster H1 comprised ~2/3 of cases and was dominated by a high somatic mutational burden, and non-silent mutations in genes canonically involved in NF-κB, JAK/STAT and PI3K signaling as well as in B2M. Conversely, cluster H2 (~1/3 of cases) was primarily characterized by recurrent somatic copy number aberrations as well as mutations in TP53 and KMT2D. Herein, we validate these previously identified genetic subtypes in external datasets as well as through orthogonal methods including tissue-based and noninvasive transcriptional and immune profiling.

Methods: To validate genetic cHL subtypes, we first leveraged public cHL genotypes from 61 patients obtained through whole genome/exome sequencing (WGS/WES) of flow-sorted HRS cells (Maura et al, Blood Cancer Disc. 2023). Using our previously locked down probabilistic classifier, we assigned the H1/H2 subtype, and then correlated cluster assignments with clinical variables. To explore transcriptional differences between genetic subtypes, we profiled baseline plasma samples (n=113) from the larger plasma genotyping cohort (n=293) using EPIC-Seq (Esfahani et al, Nat Biotechnol 2022), which allows for noninvasive gene expression profiling from cfDNA fragmentation patterns at transcription start sites. Further, we used SABER (Sworder, Cancer Cell 2023) to enumerate T-cell receptor (TCR) rearrangements in cfDNA (cfTCR, n=292). Finally, to assess subtype-specific immune infiltrate patterns, we profiled cHL tumor specimens using RNA-Seq (n=64), and applied CIBERSORTx.

Results: Similar to observations in our cfDNA discovery cohort, H1 was found to be the more prevalent subtype in the external validation set comprising 56% of tumors, while 44% were classified as H2. Recurrence frequencies of genetic features were comparable to and significantly correlated with those from our plasma discovery cohort (R_S=0.59 [H1] and R_S=0.63 [H2], P<0.001 each). Of note, when considering the whole genome space, the higher mutational burden of H1 tumors could be confirmed (P<0.01), and this association was independent of the tumor EBV status. In agreement with the discovery cohort, the bimodal age distribution and increased EBV positivity of the H2 subtype could also be recapitulated (31% vs 6% EBV+, P<0.01).

To explore transcriptional differences between H1/H2 subtypes, we took advantage of the plasma enrichment of HRS cell ctDNA and utilized EPIC-Seq to noninvasively infer expression of 1,676 targeted genes. Tracking signatures derived from scRNA-Sequencing in plasma samples, we found that both HRS cells and the cHL tumor microenvironment can be successfully profiled by EPIC-Seq. Strikingly, we found substantial enrichment of a cytokine response signature in H1 tumors, while T-cell activation was among the top upregulated signatures in H2 tumors (Fig. A). Importantly, the T-cell signature found in H2 was accompanied by a higher abundance of T-cell clones as quantified by SABER in baseline plasma samples (P<0.001, Fig. B). Notably, cfTCR fragment length profiles resembled the mutant ctDNA profiles, strongly suggesting a tumor origin of the TCR rearrangements detected in plasma. Lastly, immune cell deconvolution of bulk RNA-Seq specimens indicated a higher abundance of CD8+ T-cells in H2 tumors (P<0.01), further confirming our prior observations.

Conclusions: Collectively, these results serve to validate H1 and H2 as distinct cHL subtypes, to confirm the characteristic genotypes defining them, and recapitulate their distinctive associations with key clinical and pathological variables including age and EBV status. We further validate the subtypes using orthogonal methods revealing dominant cytokine driven signaling in H1. Conversely, H2 tumors, which largely lack B2M mutations, despite their lower mutational burden, are rather immunogenic triggering T-cell responses.