Peripheral plasmablast expansion as a predictor and therapeutic target in chronic GVHD

Abstract

Abstract Chronic GVHD (cGVHD) following allogeneic stem cell transplantation (allo-SCT) is marked by disruption of the adaptive immune system, including aberrant B cell homeostasis, and numerous studies suggest that B cells play a role in cGVHD pathogenesis. However, which specific B cell populations contribute to cGVHD development, their pathogenic role in the disease, and how these cells could potentially be targeted remains uncertain. To gain insight into B cell dysregulations before onset of cGVHD, peripheral blood samples of 172 allo-SCT patients were examined at defined time points between 56 days and 365 days after allo-SCT using multiparameter flow cytometry in a prospective study design. Retrospective grouping categorized the patients into three cohorts: no GVHD (n=17), acute GVHD without later cGVHD (n=32), and cGVHD (n=59; de novo n=11; quiescent n=49), which emerged at a median of day 180. Patients who later developed cGVHD revealed a markedly higher frequency of antibody-producing CD21low CD20neg CD38hi plasmablasts at day 90 than those who never experienced GVHD or patients with only acute GVHD (median 5.9% vs 2.2% vs 2.2% of CD19+ cells; p=0.0016 and p=0.0304, respectively) and the elevation persisted until day 365. The expansion of the plasmablast population already occurred prior to the onset of cGVHD and a Cox proportional hazards model with cGVHD diagnosis as an endpoint identified an elevated frequency of plasmablasts higher than 3% of CD19+ cells 90 days after allo-SCT as an independent risk factor for subsequent cGVHD development taking into account multiple clinical transplant parameters as co-variables (hazard ratio, 3.07; 95% CI, 1.05 to 8.99; p value, 0.04). As demonstrated by scRNA sequencing data (n=5), expanded plasmablasts of cGVHD patients revealed clonal expansions, a high frequency of somatic mutations, and a selection of certain VH genes in their antibody repertoire. Most plasmablasts were found class switched to IgG and even more frequently to the mucosa-associated isotype IgA. Many IgA+ plasmablasts additionally expressed surface markers for mucosal homing like CCR9 or α4β7-integrin. Notably, 12 out of 24 IgA+ antibody clones derived from cGVHD plasmablasts recognized specific short-chain fatty acid-producing taxa of intestinal commensals such as Bacteroides, Blautia, Faecalibacterium, or Ruminococcus indicating a connection between peripheral plasmablasts and the microbiome. In a new patient cohort, we investigated whether plasmablast frequencies were elevated at disease onset and how first-line therapy for cGVHD affected B cell subpopulations. Samples from 20 patients (de novo, n=6; quiescent, n=14) were collected at the day of cGVHD onset (median day 183, range 98 to 402 days) and 4 and 8 weeks after disease onset and initiation of steroid therapy using prednisolone. At the day of cGVHD onset a remarkably elevated frequency of plasmablasts compared to patients 180 days after allo-SCT who never experienced GVHD was found by multiparameter flow cytometry (median 7.0% vs 1.9% of CD19+ cells; p=0.009). Steroid therapy led to a reduction of cGVHD severity with a concomitant decrease of plasmablast frequencies already 4 weeks after onset treatment (median 7.0% vs 1.6% of CD19+ cells; p=0.008). This was accompanied by a substantial decline in serum BAFF levels under steroid therapy (median 5.4 ng/ml vs 1.9 ng/ml; p=0.009). Interestingly, the frequency of other B cell subsets such as CD11c+ T-bet+ B cells remained unaltered in the first 8 weeks after treatment (median 3.1% vs 3.0% of CD19+ cells), suggesting a complex dynamic of B cell dysregulation at the onset of therapy. Overall, our new findings confirm the prominent role of plasmablasts as important predictors for the onset of cGVHD.