Bone marrow CD4 + T cell depletion predicts relapsed/refractory DLBCL and is associated with a putative lactate-mediated systemic immunosuppressive axis

Diffuse large B-cell lymphoma is a highly heterogeneous disease with 30–40% of patients progressing to relapsed/refractory disease. The bone marrow immune microenvironment may provide early immune predictors of treatment failure, but its association with the subsequent development of relapsed or refractory disease remains underexplored. We retrospectively analyzed clinical and bone marrow flow cytometry data from 145 newly diagnosed DLBCL patients. Transcriptomic (GSE10846, n = 420) and single-cell (GSE182434, ~ 28,000 cells) datasets were integrated to explore upstream mechanisms. Patients who subsequently developed relapsed or refractory disease exhibited an inverted CD4/CD8 ratio (P = 0.001) and concurrent depletion of CD4⁺ T cells (P < 0.001), NK cells (P = 0.005), and NKT cells (P = 0.005). In public multi-omics datasets, HIF1A -driven glycolysis and MCT4-mediated lactate export were among the dominant metabolic features associated with systemic immunosuppression, although these findings were not validated in paired clinical samples, and baseline serum LDH did not correlate with BM CD4⁺ T-cell counts in our cohort (Spearman ρ = − 0.098, P = 0.240). Multivariable analysis confirmed that a high absolute CD4⁺ T-cell count (≥ 4837 cells/ µL) was an independent protective factor against relapsed or refractory disease (OR = 0.207, 95% CI 0.086–0.467, P < 0.001) and an independent favorable prognostic factor for PFS (adjusted HR = 0.448, 95% CI 0.227–0.885, P = 0.021). A clinical-immunological nomogram integrating CD4⁺ T-cell count, B symptoms, IPI score, and BCL-2 expression achieved an AUC of 0.797 for predicting subsequent R/R disease. Bone marrow CD4⁺ T-cell depletion represents an independent early indicator of treatment failure in DLBCL, potentially reflecting a lactate-mediated systemic immunosuppressive axis that remains to be experimentally validated. The proposed nomogram offers a practical pre-treatment tool for individualized risk stratification.