T-cell tolerant fraction as a predictor of graft-versus-host disease in murine and human allogeneic hematopoietic cell transplantation

Abstract Background: Allogeneic hematopoietic cell transplantation (HCT) is curative in a variety of hematologic malignancies, through reconstitution of protective donor T cells capable of mediating anti-cancer and anti-infective effects in the recipient. Broad application of HCT is however hampered by graft-versus-host disease (GvHD), caused by donor T cells targeting recipient tissues. Traditional methods of donor selection rely on human leukocyte antigen (HLA) matching to reduce GvHD risk, however, this does not fully eliminate GvHD. Here we propose and assess a novel metric—the T-cell receptor beta (TRB) tolerant fraction—as an additional predictor for GvHD. Tolerant Fraction: To predict GvHD risk from TRB sequences, we identify TRB sequences presumed tolerant of the recipient from those potentially alloreactive. Recipient productive TRB sequences (in-frame, no stop codon) encode functional T-cell receptor chains and thus represent a tolerant repertoire. Donor TRB sequences matching these productive recipient sequences are therefore presumed tolerant. In contrast, recipient non-productive TRB sequences (out-of-frame or containing stop codons) do not encode functional receptor chains and thus represent a repertoire component that is not necessarily tolerant. Donor TRB sequences matching these non-productive recipient sequences are presumed potentially alloreactive. The relative proportion of donor TRB sequences matching productive vs. non-productive recipient sequences defines the TRB tolerant fraction. Thus, a higher TRB tolerant fraction could indicate greater donor-recipient tolerance. Methods: Based on the hypothesis that GVHD-free patients would have higher TRB tolerant fraction values, this metric was initially assessed in murine GvHD models (2 C57BL/6 mice, 4 BALB/c mice). Specifically, we utilized spleen-derived TRB sequences (Adaptive Biotechnologies) from an established murine GvHD model based on a fully MHC-mismatched pair C57BL/6→BALB/c, compared to the pair BALB/c→C57BL/6, as well as syngeneic pairs, BALB/c→BALB/c, C57BL/6→C57BL/6. Next, the metric was assessed in a human HCT cohort. TRB sequences (Adaptive Biotechnologies) obtained pre-transplant from 19 donor-recipient pairs previously published were analyzed (Kankary et al, JCI insight. 2016; Pagliuca et al, JCI insight. 2021) . Thirteen donors were HLA-matched related (MRD) and six were haploidentical donors. Clinical outcomes assessed included acute and chronic GvHD. Statistical testing was conducted using one-sided Mann-Whitney U tests to evaluate prespecified directional hypotheses, and receiver operating characteristic (ROC) analyses were performed to assessed predictive performance. Results: The murine GvHD model (C57BL/6→BALB/c), known to result in severe GvHD, exhibited significantly lower tolerant fractions (mean 58.2%) compared to milder (BALB/c→C57BL/6, mean 62.5%) and syngeneic control pairings (BALB/c→BALB/c mean 61.1%, C57BL/6→C57BL/6 mean 62.6%; p=1.7×10⁻⁷), confirming the ability of the TRB tolerant fraction to distinguish GvHD severity. In the human HCT cohort, patients developing acute GvHD demonstrated lower mean TRB tolerant fraction (82.6%, n=8) compared to those without acute GvHD (85.2%, n=11), although the difference did not reach statistical significance (p=0.087). ROC analysis yielded moderate predictive capability (AUC=0.69). In contrast, chronic GvHD patients had significantly lower mean TRB tolerant fraction (51.3%, n=8) than those without chronic GvHD (59.6%; n=9 p=0.019). ROC analysis demonstrated predictive capability (AUC=0.81), with a threshold of 68.7% yielding 78% sensitivity for no chronic GvHD and 100% specificity for chronic GvHD. Conclusions: The TRB tolerant fraction measuring overlap between pre-transplant donor and recipient T cell repertoire may differentiate GvHD from non-GvHD states in mouse models and in human HCT recipients. Integrating the pre-transplant TRB tolerant fraction with existing HLA typing approaches may substantially improve donor selection, inform GVHD prophylaxis strategies, and possibly reduce GvHD incidence, enhancing the safety of HCT.