BACKGROUND AND AIMSDespite improved clinical outcomes of patients undergoing Hematopoietic Stem Cell Transplantation (HSCT) in recent years, chronic GVHD (cGVHD) and its associated complications have a significant impact on non-relapse mortality and quality of life. Loss of immune tolerance in the post-transplant period plays an important role in the emergence of cGVHD, and while imbalances in CD4+ T cell subsets, such as regulatory T cells, are known to contribute to its pathobiology, the role of different subsets of CD8+ T cells is poorly characterized.PATIENTS AND METHODSWe prospectively monitored a group of 56 patients who underwent unrelated donor HSCT after reduced intensity conditioning containing ATG. GVHD prophylaxis consisted of cyclosporine plus mycophenolate mofetil for all patients. 15 patients were excluded due to early disease relapse and/or death secondary to infection or acute GVHD in the first 9 months post-HSCT. cGVHD diagnosis and staging was performed according to the 2014 NIH criteria. Patients were divided into 2 groups according to cGVHD status: 17 patients (41%) with cGVHD (cGVHD+) and 24 patients (59%) without cGVHD (cGVHD-).Peripheral blood was analyzed by flow cytometry to characterize T cell subsets at fixed time points over a 2 year period. CD8+ T subsets were identified as follows: True Naïve cells as CD45RA+CD62L+CD95- (TN), Stem Cell Memory cells as CD45RA+CD62L+CD95+ (SCM), Central Memory cells as CD45RA-CD62L+ (CM), Effector Memory cells as CD45RA-CD62L- (EM) and terminally differentiated effector cells as CD45RA+CD62L- (EMRA).Frozen PBMC from patients at months 9 and 12 post-HSCT were used to purify CD8+ T cells by cell sorting to quantify Signal-Joint T Cell Receptor Excision Circles (TREC) by Real-Time PCR and to evaluate TCR diversity by TCR Vbeta complementarity-determining region 3 spectratyping.RESULTSPatients with cGVHD had significantly increased percentages and absolute counts of TN CD8 in the first 6 months post-HSCT compared with cGVHD- patients (p≤0.04 at months 1 and 3 for percentages; p≤0.02 for months 1-6 in absolute counts).Furthermore, cGVHD+ patients had increased SCM CD8 cells throughout the post-HSCT period (significances of p<0.05 at months 6 and 12 for percentages and 18 and 24 for counts). The increased prevalence of these two subsets was not associated with increased proliferation as assessed by Ki67, which was reduced in cGVHD+ in the total CD45RA+CD62L+ gate (TN plus SCM cells) (p≤0.04 at months 6, 12 and 18). No significant differences were observed in susceptibility to apoptosis as measured by BCL2 or CD95 expression levels.CM and EM subset counts were reduced in cGVHD+ as compared to cGVHD- (p=0.04 and 0.02 at month 12). EM CD8 were the dominant population in cGVHD- patients throughout the follow up. Significant increases in the percentage of EM CD8 cells in cGVHD- were observed at months 2, 9, 12 and 18 (p≤0.04).cGVHD+ patients had increased percentages of EMRA (p<0.04 at months 9, 12 and 18) and absolute counts (p<0.04 in months 2, 9 and 24). This was not correlated to altered Ki67, Bcl-2 or CD95 levels.The increase in TN in cGVHD+ patients was not associated with increased TREC numbers within CD8+ T cells. TREC content was significantly reduced in both patient groups at month 9 as compared healthy controls (HC) (p<0.05). This could be due to the low prevalence of these cells and/or to rapid differentiation into other memory subsets such as the SCM and EMRA. Despite the differences in subset prevalence in the two patient groups, we did not observe any significant differences in TCR Vbeta diversity in CD8+T cells. Reduced TCR diversity was only observed when patients were compared to HC (p=0.0001).DISCUSSIONOur observations of increased TN, SCM and EMRA CD8 subsets very early after HSCT in patients who subsequently develop cGVHD suggest the possibility that these may be relevant biomarkers for cGVHD development. SCM T cells have previously been shown to cause lethal GVHD in mouse models and to represent a reservoir of antigen-specific T cells capable of differentiating in vivo into other memory subsets with effector function. We speculate that thymic tissue damage resulting from conditioning regimens and/or acute GVHD, together with other yet unidentified factors, may lead to the emergence of self-reactive naïve CD8+ T cells that differentiate to SCM and later to EMRA cells potentially mediating cGVHD. DisclosuresNo relevant conflicts of interest to declare.