Acute graft-versus-host disease (aGvHD) is an unpredictable immunological complication that affects the skin, gut, or liver following allogeneic hematopoietic cell transplantation (HCT). Donor Th1 cells can initiate aGvHD after recognition of recipient alloantigens and migration to target tissues via homing receptors.1 Paradoxically, experiments in animal models of transplantation suggest that deficiency of the inflammatory Th1 cytokine IFN-γ can exacerbate aGvHD severity and mortality.32 Investigators have interpreted these results to mean that other cell types are capable of generating aGvHD or that loss of a protective function related to IFN-γ was responsible for the intensified alloreactivity observed in these animal models.
Indeed, IL-17 secreting Th17 cells can initiate aGvHD with skin as the principal target under certain experimental conditions.4 IFN-γ also up-regulates the inhibitory molecule, programmed cell death 1 ligand [PD-L1, (CD274)].65 Binding of PD-L1 to its receptor programmed cell death 1 [PD1, (CD279)] on T cells leads to apoptosis and promotes immune tolerance.7 Thus, IFN-γ seems to have opposing functions during HCT. IFN-γ expression is increased at sites of active aGvHD while complete Th1 deficiency intensifies aGvHD severity. Although such findings present an apparent contradiction, it is possible that differing migration patterns by the various cell types or expression of effector molecules in the systemic versus local compartments might explain the contrasting outcomes of such studies. We tested the hypothesis that effector T cells can be inflammatory [(mediating pro-GvHD and graft-versus-leukemia effects (GvL)] or inhibitory (mediating anti-GvHD/GvL effects) depending on their expression of tissue-homing molecules on T cells, such as cutaneous lymphocyte antigen (CLA) or the gut-homing marker α4β7.
Patients with hematologic malignancies undergoing T-cell-replete, matched-related donor (MRD) HCT were accrued to an aGvHD biomarker study (n=42). All patients received HLA-identical peripheral blood stem cell grafts followed by prophylaxis with cyclosporine and either methotrexate or mycophenolate mofetil. By day+100, grade 2–4 or grade 3–4 aGvHD was diagnosed clinically in 31 or 4 patients, respectively (Table 1).8 aGvHD was confirmed by biopsy in the majority (n=23).
Flow cytometry, soluble suppression of tumorigenicity 2 (sST2), and statistical analysis are presented in the Online Supplementary Methods.
Tissue-specific T cells were identified in blood from healthy volunteers (n=21) or from recipients of HLA-identical sibling donor transplants at 2 time points: neutrophil engraftment (n=42, median 19 days post HCT) and at day+30 (n=37) (Online Supplementary Figure S1). Among HCT recipients, circulating Th17 cells at engraftment preferentially expressed a skin-homing (CLA) phenotype (10.5% vs. 2.14%; P<0.001), while Th1 cells demonstrated an increased frequency of gut-homing (α4β7) phenotype (27.1% vs. 14.0%; P<0.001). Similar homing patterns were seen at day+30 and among healthy volunteers, supporting the importance of Th17 and Th1 subsets for cutaneous or gastrointestinal immunity, respectively (data not shown).
Among patients with grade 2–4 aGvHD (n=31), T-cell analysis could be performed before clinical alloreactivity in all but 3 individuals when sampling occurred during hyperacute symptoms. Both percentage and absolute number of circulating Th1 cells at engraftment were significantly lower in patients developing grade 2–4 aGvHD when compared to recipients with grade 0–1 disease, while Th1 cell percentages were similar between healthy volunteers and HCT recipients without alloreactivity (Table 1 and Figure 1). After adjusting for conditioning intensity and day+30 sST2, decreased percentages of total Th1 cells at engraftment continued to be a significant predictor of moderate to severe aGvHD (Table 1).9 We were unable to find an association between total Th1 cells at day+30 and Th17 cells with the development of aGvHD (Table 1 and Figure 1). The remainder of analyses focused on Th1 subsets at engraftment.
We hypothesized that the observed decrease in Th1 cells in patients with clinical alloreactivity could represent the migration of effector cells from the systemic circulation to an aGvHD target tissue. Consistent with this idea, patients with grade 2–4 aGvHD had increased frequencies of circulating skin-homing (CLA) Th1 cells and gut-homing (α4β7) Th1 cells, but lower numbers of conventional (CLA-α4β7) Th1 cells than patients with grade 0–1 aGvHD (Table 1 and Figure 1). When analyzing patients with isolated skin aGVHD, CLA Th1 cells were increased in patients with more severe skin involvement (stage 3–4) versus those with stage 0–2 (6.51% vs. 0.78%; P=0.013). After excluding individuals with skin aGvHD, α4β7+ Th1 cells tended to be higher in patients with any stage of gut-only aGvHD (32.8% vs. 21.8%; P=0.08). The absolute decrease in conventional (CLA- α4β7-) Th1 cells in the blood of patients developing aGvHD could result from upregulation of homing-markers within this lymphocyte population followed by migration from the circulation to a peripheral compartment, thus explaining why the relative frequency of tissue-specific T cells in blood changed but not the absolute numbers (Figure 1).
Since alloimmunity can lead to both detrimental aGvHD and beneficial GvL effects, exploratory analyses were performed to determine whether circulating total Th1 subsets at engraftment could impact the cumulative incidence of malignancy relapse in patients with MDS/AML (n=16; 7 relapses post HCT). The majority of individuals had standard risk disease (n=13), and all patients had less than 5% blasts prior to HCT. As suggested by the aGvHD data, decreased percentages of total Th1 cells at engraftment predicted a lower risk of MDS/AML relapse after adjustment for conditioning intensity and disease risk (HR 0.84; 95%CI: 0.71–0.99; P=0.047). To investigate the potential inhibitory mechanisms of IFN-γ on GvL in our study, we examined PD1 and PD-L1 expression on circulating CD4 T cells and CD14 monocytes, respectively, between engraftment and day +30 after HCT (Online Supplementary Figure S2).105
Since cryopreserved specimens at the time of relapse were not available, PD-L1 expression by monocytes was used as a surrogate measure of IFN-γ effects on residual MDS/AML cells persisting after HCT. In the 14 MDS/AML patients with accessible samples, total Th1 cells at engraftment correlated positively with both PD1CD4 T cells and PD-L1CD14 monocytes (Figure 2). Increased PD-L1 expression by CD14 monocytes also predicted a greater risk for MDS/AML relapse (HR 1.11; 95%CI: 1.01–1.24; P=0.048).
Early increases in Th1 cells were associated with decreased alloreactivity and greater PD-L1 expression following HCT. Furthermore, we expanded upon the findings from previous animal studies by identifying tissue-homing subsets within this Th1 population that are critically important for either generating or preventing aGvHD.
Clinical manifestations of aGvHD often occur early after engraftment and are restricted to particular tissues, suggesting tissue-specific expression of effector molecules or peripheral localization of lymphocytes are responsible for aGvHD. Similar to our previous Treg data, our current report supports lymphocyte tissue-trafficking as an important factor for alloreactivity, as patients developing grade 2–4 aGvHD had increased skin-homing (CLA) and gut-homing (α4β7+) Th1 cells and tended to have more cutaneous or gastrointestinal symptoms, respectively.1211 As opposed to lymphocyte migration determining aGvHD, it is also plausible that conventional (CLA-α4β7-) Th1 cells possess immunoregulatory properties. When we analyzed Foxp3 expression, the Foxp3 cells appeared separate and distinct from the Th1 subsets, indicating that these IFN-γ+ lymphocytes likely were not traditional Tregs (Online Supplementary Figure S3). IFN-γ induces expression of programmed death ligand 1 (PD-L1) on antigen presenting cells, which leads to apoptosis of PD1-expressing T cells and anergy.7 Consistent with the idea of IFN-γ mediated immune regulation, we have shown a very strong, positive correlation between the percentage of total Th1 cells at engraftment with the expression of PD1 and PD-L1 by CD4 T cells or CD14 monocytes, respectively. This preliminary data indicate that early Th1 cell proliferation primarily within the CLA-α4β7-compartment could promote peripheral tolerance by up-regulating the inhibitory protein PD-L1. Although we showed a correlation between IFN-γ and PD-L1, it is unknown whether the expression of PD-L1 on monocytes is a valid surrogate for IFN-γ effects on AML cells or the cause for relapse, and further experimentation is required.
Some, but not all, research supports a role for Th17 cells in the generation of cutaneous aGvHD.134 We found that CLA was expressed more commonly by Th17 cells than by Th1 cells; however, the frequency of cells in Th17 subsets did not correlate with aGvHD. These analyses were possibly limited by the low incidence of cutaneous alloreactivity. Strikingly, about 70% of our MRD cohort developed aGvHD with gastrointestinal symptoms. Reassuringly, sST2, a validated biomarker for aGvHD, was elevated in patients developing alloreactivity.9 The high incidence of gastrointestinal aGvHD likely was due to early endoscopy, as previously reported.14
Malignancy relapse and severe aGvHD are the primary causes for early HCT failure. Blockade of immune checkpoints including CTLA-4 with ipilimumab or PD1 with nivolumab and pembrolizumab could improve GvL, but also may increase aGvHD.1510 Concerns regarding aGvHD exacerbation may have dampened enthusiasm for PD1 blockade after allogeneic HCT.15 Our data confirm preclinical research that early Th1 immunity is associated with immune tolerance and increased PD-L1 expression. It also potentially explains the paradoxical nature of IFN-γ during HCT in terms of lymphocyte compartmentalization (i.e. inflammatory tissue-homing Th1 cells vs. inhibitory conventional Th1 cells). If validated, these results could identify patients with low aGvHD risk who may benefit from therapies targeting immune checkpoints to promote GvL and prevent malignancy relapse.
- Krenger W, Ferrara JL. Graft-versus-host disease and the Th1/Th2 paradigm. Immunol Res. 1996; 15(1):50-73. PubMedhttps://doi.org/10.1007/BF02918284Google Scholar
- Murphy WJ, Welniak LA, Taub DD. Differential effects of the absence of interferon-gamma and IL-4 in acute graft-versus-host disease after allogeneic bone marrow transplantation in mice. J Clin Invest. 1998; 102(9):1742-1748. PubMedhttps://doi.org/10.1172/JCI3906Google Scholar
- Yang YG, Dey BR, Sergio JJ, Pearson DA, Sykes M. Donor-derived interferon gamma is required for inhibition of acute graft-versus-host disease by interleukin 12. J Clin Invest. 1998; 102(12):2126-2135. PubMedhttps://doi.org/10.1172/JCI4992Google Scholar
- Carlson MJ, West ML, Coghill JM, Panoskaltsis-Mortari A, Blazar BR, Serody JS. In vitro-differentiated TH17 cells mediate lethal acute graft-versus-host disease with severe cutaneous and pulmonary pathologic manifestations. Blood. 2009; 113(6):1365-1374. PubMedhttps://doi.org/10.1182/blood-2008-06-162420Google Scholar
- Berthon C, Driss V, Liu J. In acute myeloid leukemia, B7-H1 (PD-L1) protection of blasts from cytotoxic T cells is induced by TLR ligands and interferon-gamma and can be reversed using MEK inhibitors. Cancer Immunol Immunother. 2010; 59(12):1839-1849. PubMedhttps://doi.org/10.1007/s00262-010-0909-yGoogle Scholar
- Lee SK, Seo SH, Kim BS. IFN-gamma regulates the expression of B7-H1 in dermal fibroblast cells. J Dermatol Sci. 2005; 40(2):95-103. PubMedhttps://doi.org/10.1016/j.jdermsci.2005.06.008Google Scholar
- Freeman GJ, Long AJ, Iwai Y. Engagement of the PD-1 immunoinhibitory receptor by a novel B7 family member leads to negative regulation of lymphocyte activation. J Exp Med. 2000; 192(7):1027-1034. PubMedhttps://doi.org/10.1084/jem.192.7.1027Google Scholar
- Przepiorka D, Weisdorf D, Martin P. 1994 Consensus Conference on Acute GVHD Grading. Bone Marrow Transplant. 1995; 15(6):825-828. PubMedGoogle Scholar
- Vander Lugt MT, Braun TM, Hanash S. ST2 as a marker for risk of therapy-resistant graft-versus-host disease and death. N Engl J Med. 2013; 369(6):529-539. PubMedhttps://doi.org/10.1056/NEJMoa1213299Google Scholar
- Weber J. Immune checkpoint proteins: a new therapeutic paradigm for cancer–preclinical background: CTLA-4 and PD-1 blockade. Semin Oncol. 2010; 37(5):430-439. PubMedhttps://doi.org/10.1053/j.seminoncol.2010.09.005Google Scholar
- Engelhardt BG, Jagasia M, Savani BN. Regulatory T cell expression of CLA or alpha(4)beta(7) and skin or gut acute GVHD outcomes. Bone Marrow Transplant. 2010; 46(3):436-442. PubMedGoogle Scholar
- Engelhardt BG, Sengsayadeth SM, Jagasia M. Tissue-specific regulatory T cells: biomarker for acute graft-vs-host disease and survival. Exp Hematol. 2012; 40(12):974-982. PubMedhttps://doi.org/10.1016/j.exphem.2012.08.002Google Scholar
- Broady R, Yu J, Chow V. Cutaneous GVHD is associated with the expansion of tissue-localized Th1 and not Th17 cells. Blood. 2010; 116(25):5748-5751. PubMedhttps://doi.org/10.1182/blood-2010-07-295436Google Scholar
- Martin PJ, McDonald GB, Sanders JE. Increasingly frequent diagnosis of acute gastrointestinal graft-versus-host disease after allogeneic hematopoietic cell transplantation. Biol Blood Marrow Transplant. 2004; 10(5):320-327. PubMedhttps://doi.org/10.1016/j.bbmt.2003.12.304Google Scholar
- Saha A, Aoyama K, Taylor PA. Host programmed death ligand 1 is dominant over programmed death ligand 2 expression in regulating graft-versus-host disease lethality. Blood. 2013; 122(17):3062-3073. PubMedhttps://doi.org/10.1182/blood-2013-05-500801Google Scholar