Spatial transcriptomics identifies a B cells–depleted profile underlying high-risk classical Hodgkin Lymphoma

Abstract Background Risk stratification in classical Hodgkin lymphoma (cHL) primarily relies on clinical and radiologic parameters, but outcomes are still heterogeneous after standard first-line chemotherapy. We investigated how, the spatial interplay between Hodgkin and Reed–Sternberg cells (HRS) and tumor microenvironment (TME) at diagnosis, impacts on clinical aggressiveness. Methods Spatial transcriptomics was performed to profile HRS (CD30+), non-tumoral B cells (CD20+) and TME (CD45+/CD30-) in a matched series of cHLs, comparing 4 relapsed/refractory (R/R+) and 4 relapse-free (R/R-) patients. A retrospective monocentric training cohort ( N = 155) and a multicentric retrospective validation cohort ( N = 89) of cHLs were analyzed using digital barcoding-based gene-expression profiling of 770 immune-related genes, to derive and validate a biologically relevant gene score predictive of progression-free survival (PFS). A B-cell–related gene score (B-score) was developed using elastic net regression. Results Spatial transcriptomics showed that R/R+ cHLs exhibited loss of B-cell lineage identity in HRS cells and marked depletion of naïve B cells in TME. Digital barcoding transcriptomic analysis in the training cohort (4-year PFS 80.4%, 95%CI 74.1–87.3) revealed that genes associated with improved PFS were enriched for B-cell programs. An 18-genes B-cell signature stratified patients into three groups with significantly different PFS (log-rank P < 0.001). A derived 8-genes B-score predicted PFS with an AUC of 0.73 (95%CI 0.62–0.84) and remained independently associated with PFS after adjustment for clinical and radiomic variables (HR 0.39, 95%CI 0.17–0.92). This association was confirmed in the independent validation cohort (4-year PFS 74.6%, 95%CI 67.0–83.1; AUC 0.64 95%CI (0.49–0.78); HR 0.30 95%CI 0.10–0.88). Conclusions Overall, these data showed that non-tumoral B cells depletion identifies patients at risk of progression. B-score provides a quantitative, reproducible, biologically grounded, independent tool for risk stratification improving performance of current prognostic criteria .