Post-Transplant Cyclophosphamide Ameliorates Lethal Thymic GVHD By Restoring Regulatory and Effector T Cell Homeostasis in Recipients with Prior PD-1 Blockade Therapy

Allogeneic HSCT has a curative potential for patients with hematological malignancies. However, graft-versus-host disease (GVHD) remains to be a significant cause of morbidity and mortality after HSCT. Regulatory T cells (Tregs) are critical mediator for immune tolerance after HSCT and we recently reported that PD-1 plays an essential role for Treg survival (Asano et al, Blood 2017). Clinical studies suggested that PD-1 blockade prior to HSCT could be a risk of increasing severe GVHD. However, the mechanisms about GVHD induced by PD-1 blockade have largely unclear and there remains a paucity of data on appropriate GVHD prophylaxis for patients who undergo HSCT after PD-1 blockade. To address these issues, we investigated the impact of PD-1 expression on donor T cells on immune reconstitution with murine BMT models. First, lethally irradiated B6D2F1 mice were transplanted with 10 million of C57BL/6-background PD-1^+/+ or PD-1^-/- spleen cells with 5 million of bone marrow cells from normal C57BL/6, and GVHD scores and overall survival was monitored. Recipients receiving PD-1^-/- graft developed severe GVHD resulting in a significant shorter survival than recipients receiving PD-1^-/- graft (P<0.0001). We analyzed lymphocytes in spleen and thymus on day3, 7, and 14. We found that CD8 T cells in PD-1^-/- group showed markedly higher Ki67 expression and CFSE-dilution until day3. Interestingly, PD-1^-/- Tregs increased aggressively at day3 but it could not maintain until day14, while PD-1^-/- CD8 T cells and conventional CD4 T cells (CD4 Tcons) continued to increase until day+14, resulting in the significant higher CD8/Treg ratio in PD-1^-/- group (P<0.05, vs PD-1^+/+ group). PD-1^-/- Tregs showed significantly higher expression of Annexin V on day+7 and thymus CD4- and CD8- double-positive (DP) cells were in the extremely low levels in PD-1^-/- group on day+14 (P<0.05, vs PD-1^+/+ group). Thymic analysis showed that donor PD-1^-/- graft-derived CD8 T cells infiltrated thymus in PD-1^-/- group, suggesting reconstruction of thymic function was critically disturbed by severe GVHD. These data suggest that loss of PD-1 signaling resulted in unbalanced reconstitution of donor-derived T cell subsets as a consequence of continuous CTL expansion and increased Treg apoptosis. Next, to evaluate the impact of post-transplant cyclophosphamide (PTCy) on the abnormal reconstitution after PD-1 blockade, we administered 50mg/kg of Cy or control vehicle on day3. PTCy efficiently ameliorated GVHD in PD-1^-/- group and extended overall survival by safely regulating the proliferation and apoptosis of T cell subsets. Of note, after PTCy, Tregs regained the ability of continuous proliferation in the first 2 weeks, resulting in well-balanced reconstitution of donor-derived T cell subsets. Thymic DP cells on day 14 was markedly increased in PD-1^-/- group with PTCy intervention as compared to without PTCy, suggesting PTCy could rescue thymus from PD-1 blockade-related severe GVHD. Finally, to evaluate GVL activity, we performed BMT with co-infusion of P815L tumor cells on day0 and we confirmed that PTCy treatment for PD-1^-/- recipients reduced the severity of GVHD with maintaining sufficient GVL effect. In summary, our data suggested three insights about the impact of PD-1 signaling on immune reconstitution. First, PD-1 inhibition influenced graft-derived T cells very differently within T cell subsets. PD-1^-/- Tregs increased transiently but it was counterbalanced by accelerated apoptosis, while PD-1^-/- CD4+Tcons and CD8 T cells continued the drastic expansion. Second, we found that PD-1^-/- donor T cells developed severe GVHD in thymus. Few reports have concentrated on the impact of donor graft PD-1 expression to thymus after BMT and acute GVHD in thymus could lead late central immune disturbance. Third, PTCy successfully ameliorated GVHD induced by PD-1^-/- donor T cells preserving GVL effect. Cell proliferation study implied that PD-1^-/- graft-derived CD8 T cells might be more susceptible for PTCy because of the high-rate proliferation. In conclusion, PD-1^-/- graft cause lethal thymic GVHD and PTCy successfully ameliorated it. The influence of PD-1 inhibition was different within T cell subtypes. PTCy might be appropriate GVHD prophylaxis strategy for patients who had prior usage of PD-1 blockade.