Clinical and biological subtypes of follicular lymphoma revealed by tumor and immune cell states.

Abstract

7070 Background: Follicular lymphoma (FL) exhibits considerable variability in biological features and clinical trajectories, compounded by a complex tumor microenvironment (TME) populated with nonmalignant immune cells. In this study, we comprehensively characterized FL by integrating multi-omics data that considered both tumor cells and microenvironmental components. This analysis revealed four subtypes of FL, each exhibiting distinct biological characteristics and differing clinical behaviors. Methods: Our study included 1,203 samples, 53 bulk transcriptomic samples from our center, 1,150 samples from public databases, and five single-cell RNA sequencing (scRNA-seq) cohorts. The Ecotyper algorithm was employed to identify immune cell states. Unsupervised clustering analysis using non-negative matrix factorization (NMF) was performed to identify distinct FL subtypes based on cellular infiltration. Results: A total of 30 unique cell states were identified from nine annotated cell populations (B cells, plasma cells, CD4 T cells, CD8 T cells, Tregs, NK cells, Tfh, monocytes/macrophages, and dendritic cells). Validation was performed across three independent scRNA-seq and bulk transcriptomic cohorts to ensure robustness. Four distinct B cell states were identified, each exhibiting unique gene expression profiles and varying prognostic implications. B cell S2 and S3 were associated with adverse outcomes. Notably, PRDM15 in B cell S2 had the most significant prognostic impact, remaining an independent adverse prognostic factor even after adjustment for m7-FLIPI. Mutational profile analysis demonstrated distinct mutation patterns across B cell states: mutations in CCND3, SPEN, and ARID1A were enriched in B cell S1; CREBBP and STAT6 mutations were prominent in B cell S2; and IRF8 and BCL2 mutations were observed in B cell S4, while no significant mutation enrichment was detected in B cell S3. Similarly, immune cells of the same type yet in different states exhibited functional and clinical heterogeneity. Four subtypes of follicular lymphoma characterized by distinct tumor cell states and varying immune infiltration were identified through unsupervised clustering analysis. FLE1 exhibits characteristics of a "cold tumor" with a high abundance of B cell S2. In contrast, FLE2 and FLE3 show moderate immune infiltration, while FLE4 is marked by abundant immune infiltration and elevated expression of immunosuppressive checkpoint molecules, resembling an "inflammatory" tumor phenotype. Prognostic analysis indicates that FLE1 has the most unfavorable prognosis, followed by FLE4. Conclusions: Our study stratifies FL patients based on the heterogeneity of tumor cells and the immune microenvironment, proposing an immune-based classification for FL. Targeting CREBBP and PRDM15 may offer promising new strategies for the clinical management of patients exhibiting a "cold tumor" phenotype.