Abstract A042: Genetic and microenvironmental determinants of site-to-site heterogeneity in follicular lymphoma

Abstract

Follicular lymphoma (FL) is highly heterogenous and can form distinct evolutionary trajectories within separate lesions, but the basis of this heterogeneity is unknown. Using pairs of nodal biopsies from 10 FL patients, we previously showed that many patients’ tumors exhibit site-to-site clonal divergence (Haebe, Shree et al., Blood, 2021). Here, we expanded the analysis to a unique dataset of spatially separated tumors from 20 FL patients and sought to investigate the genomic and microenvironmental basis for spatial divergence. Tumor biopsies were subject to single cell RNA sequencing for B-cell receptor (BCR), T-cell receptor (TCR) and whole transcriptome. We used the Renkonen similarity index (SI) to quantify similarity between sites for a given feature. Sequencing the paired heavy and light chain of the BCR, we generated phylogenetic lineages of each patient’s FL. Subclones were defined by clustering BCR sequences and then overlaid on phylogenies. We used CopyKat to infer copy number alterations and linear regression models to identify genes with the highest contribution to discordance. Paired chains provided greater granularity and a more accurate representation of FL lineage than heavy chain alone. Our expanded cohort confirmed BCR discordance was highly correlated with tumor gene expression discordance. Concordant patients had identical copy number profiles across sites, while discordant patient profiles were distinct. The most differential alteration in discordant patients was a copy number gain in 2p, harboring REL and BCL11A, enabling aggressive localized proliferation. Consistent with this, cell cycle genes ranked among the top differentially expressed genes in discordant patients. Interestingly, patients with concordant BCR subclones exhibited some divergent gene expression, driven by markers of B cell activation, such as CD69, suggesting that some site-to-site divergence arises from microenvironmental rather than genetic factors. Circulating tumor cells in one discordant patient formed their own subclone distinct from either site; in a concordant patient, circulating clones were representative of both sites. We sought to determine whether discordance resulted in altered T cell responses site-to-site. T cell clonotype SI correlated with BCR SI, suggesting an antigenic difference between lesions in discordant patients. T cell clonotypes private to one site were significantly more likely to belong to T follicular helper cells than clones shared between sites. Conversely, clonotypes belonging to activated effector CD4 and CD8 T cells were more likely to be shared across sites. Leveraging a unique dataset of simultaneously obtained tumors from FL patients, we demonstrate paired heavy and light BCR sequencing as a powerful tool for tracking subclonal evolution, confirm frequent site-to-site tumor subclone divergence, and identify potential genomic and microenvironmental contributors to phenotypic site-to-site heterogeneity in FL. We find divergent T cell clonotypes and phenotypes across sites, with implications for FL immunotherapy.