Pre-treatment B-cell activating factor (BAFF)–B-cell imbalance and multi-pathway immune phenotyping are associated with rituximab non-response in pediatric immune thrombocytopenia: a prospective cohort study

Background Rituximab is widely used as a second-line therapy for pediatric immune thrombocytopenia (ITP), yet approximately 40% of patients do not respond. The biological determinants of non-response are heterogeneous and incompletely understood. This exploratory study aimed to characterize baseline immune profiles associated with rituximab non-response and to generate hypotheses regarding the underlying mechanisms. Materials and methods In this single-center prospective study, pediatric patients with ITP who received rituximab were enrolled according to prespecified inclusion and exclusion criteria. Pre-treatment serum BAFF levels, peripheral blood immune cell subsets, and CD38 mean fluorescence intensity were compared between responders and non-responders. Correlation analyses and receiver operating characteristic (ROC) curve analyses were performed, and a multivariable logistic regression model combining BAFF, plasmablast CD38 mean fluorescence intensity, and CD3+T/lymphocyte percentage was explored. Results Of 103 patients screened, 43 patients were included in the final analysis (29 responders and 14 non-responders per 12-week efficacy assessment). Pre-treatment BAFF levels did not differ significantly between non-responders and responders (median 58.1 vs. 54.2 pg/mL, P = 0.385); however, within non-responders, BAFF was inversely correlated with peripheral CD19+B-cell proportions (Spearman ρ=−0.60, P = 0.029), consistent with a BAFF-sink mechanism. Non-responders also exhibited higher proportions of unswitched memory B cells (P = 0.033), elevated CD4+TEMRA proportions (P = 0.036), and increased naive T-cell CD38 mean fluorescence intensity (P = 0.004), with a borderline increase in plasmablast CD38 mean fluorescence intensity (P = 0.050). Within the responder group, BAFF levels were inversely correlated with plasmablast CD38 MFI (Pearson r=−0.39, P = 0.043), suggesting that BAFF exerts a residual regulatory influence on plasmablast CD38 expression in patients who ultimately respond. A multivariable model combining BAFF, plasmablast CD38 MFI, and CD3+/lymphocyte percentage achieved an apparent AUC of 0.743 (bootstrap optimism-corrected AUC 0.697), exceeding the discriminative performance of any single marker. Conclusion Rituximab non-response in pediatric ITP is associated with a multi-pathway baseline immune phenotype involving increased memory B-cell proportions, higher plasmablast CD38 surface density, and T-cell skewing, together with a BAFF–B-cell imbalance evident specifically in non-responders. These findings support a hypothesis-generating framework in which rituximab resistance arises from convergent but distinct immune mechanisms, and highlight the potential of combined immune profiling for risk stratification.