Resolving the Microarchitecture of Classic and Transformed Follicular Lymphoma By Single Cell Alignment to Spatial Transcriptomes

Background: Despite significant advances in single-cell and bulk transcriptome profiling of classic and transformed follicular lymphomas (FL), their tumor microenvironments remain insufficiently understood, limiting the discovery of treatment strategies that target local survival niches. The emerging field of spatial transcriptomics (ST) aims to elucidate the gene-level microarchitecture of lesional biopsies. However, most current spatial methods fail to effectively resolve the high tissue cellularity characteristic of indolent lymphomas while simultaneously maintaining full transcriptome coverage. We optimized tissue preparation and bioinformatic workflows to enhance FL-specific spatial transcriptomics. Methods: We first evaluated conditions for whole transcriptome recovery using the 10x Visium ST platform with 50 µm sized spots and optimized permeabilization time (40 mins) and section thickness (10 µm). We next applied these optimizations to profile frozen excisional lymph node biopsies from representative patients with Classic and Transformed FL, as well as non-malignant reactive lymphoid tissue. We developed custom FL-specific cell signature matrices by curating and reannotating publicly available 10x single-cell RNA-seq datasets from FL, then extracted single-cell resolution data utilizing the novel CytoSPACE algorithm (Vahid et al, Nat Biotech 2023). In addition to spatial and cell subset specific gene expression, we evaluated spatial copy number variation (CNV) inference using inferCNV and Numbat. Results: We recovered a median of 10.7k UMI counts per 50 µm spot across samples, allowing evaluation of a median 3,716 unique genes. Single cell mapping with CytoSPACE revealed 35±10 cells per spot totaling to 77±20k cells per sample. In reactive tissue the single cells maps show a classic patterns of polarized follicles and interfollicular T-cell zones. Interestingly, within the interfollicular zones there were immune hotspots of T-regulatory (20±4%), T follicular helper (TFH, 18±6%) and naïve CD8+ T cells (11±5%) situated adjacent to the follicle dark zones. In classic FL tissues we observed enlarged dysmorphic follicles with diminished polarization and interfollicular spaces densely populated with myeloid cells. However, similar appearing but smaller sized immune hotspots were preserved, with T-regulatory cells as the most prevalent subset. Compared to immune hotspots in reactive lymphoid tissue we found lower numbers of TFH cells (p-adj<0.01) and increased non-cytotoxic CD8 T-cells (p-adj=0.02) and NK cells (p-adj<0.05). Interestingly in both classical and transformed FL these hotspots were highly infiltrated by macrophages, activated dendritic cells (ADC), and plasma cells; while these cell types were virtually absent in reactive tissue immune hotspots. Spatial CNV profiling resolved a tumor-specific somatic aneuploidy fingerprint of FL cells for all profiled cases, with per-spot CNV signals highly correlated with malignant B-cell counts as defined by CytoSPACE. Conclusion: Our study demonstrates the efficacy of Visium ST combined with CytoSPACE single-cell mapping in elucidating the complex microarchitecture of FL tumors. The single cell maps faithfully reproduced the cellular architecture as confirmed by classical histology. We detected different patterns of spatially localized immune hotspots between reactive, classic- and transformed FL tissues. These immune hotspots were marked by a local abundance of T-regulatory cells in both reactive and FL tissues, but enriched in malignant tissue for effector subsets including ADCs, CD8+ T, and NK cells. Modulating immune hotspots may represent a therapeutic opportunity. Spatial tumor fingerprinting using CNV inference enabled high-resolution identification of malignant cells in all patients. Our approach offers a comprehensive view of FL biology within the tumor microenvironment, and additional analyses including spatial T- and B-cell receptor repertoire will be presented.