Lower Risk of Graft Versus Host Disease after Exposure to Checkpoint Inhibitors with the Use of Post-Transplant Cyclophosphamide Prophylaxis

Background: Allogeneic hematopoietic stem cell transplantation (alloHCT) remains a curative approach for patients with AML/MDS. Still, up to 50 % of patients relapse after alloHCT. Immune checkpoint inhibitors (ICI) are being used in the post-transplant setting to reverse the immune dysfunction that contributes to AML/MDS relapse. However, a major concern with using ICI post alloHCT is the increased risk of graft versus host disease (GVHD). A variety of pre- and post- transplant factors can play a crucial role in exacerbating GVHD. We performed a retrospective analysis to report our experience with ICIs used after alloHCT in AML/MDS patients with a focus on the impact of post-transplant cyclophosphamide (PTCy).

Methods: In this study, we included 16 AML and 5 MDS patients treated with single-agent anti-PD-1 (nivolumab, n=16) or anti-CTLA-4 (ipilimumab, n=5) for disease relapse after alloHCT at our institution. The primary outcome was acute GVHD (aGVHD) following ICI initiation post alloHCT. Secondary outcomes were progression-free survival (PFS) and overall survival (OS) that were estimated using the Kaplan-Meier method.

Results: Median patient age was 54 years (range, 29-77 years). Patients received a median of 2 doses of ICI therapy (range, 1 to 12). The median time from alloHCT to ICI initiation was 9.1 months (range 1.3-134.4 months). All patients had received myeloablative conditioning regimens. Donors were matched related in 9 (43%), matched unrelated in 9 (43%), and haploidentical in 3 (14%). Stem cell source was peripheral blood stem cells in 14 (67%) and bone marrow in 7 patients (33%). Twelve patients (57%) received PTCy as GVHD prophylaxis (PTCy group). Transplantation characteristics were comparable between the two groups (PTCy vs no-PTCy), except for lower median score of hematopoietic stem cell comorbidity index in PTCy group (median score of 1 (range,0-3) vs. median score of 3 (range,1-5), P=.04), and a shorter median time from alloHCT to ICI initiation (5 vs. 26 months, P=.04) noted in PTCy group. Before initiation of ICI, 5 patients (24 %) had previously had GVHD.

In the entire cohort, 4 patients (19 %) developed aGVHD following ICI initiation. All the 4 cases of aGVHD were reported following nivolumab initiation. Over a median follow up of 5.3 months, the frequency of grade 2-4 aGVHD was 19%. The frequency of grade 1 to 2 and grade 3 to 4 aGVHD were 9.5% and 9.5%, respectively, with a median time from initiation of nivolumab to onset of aGVHD of 24 days (range, 12-31 days), and a median time from alloHCT to nivolumab-related aGVHD of 188 days (range,109-419 days). The median number of nivolumab doses to aGVHD was 2 doses (range, 1-8 doses). No chronic GVHD or grade > 2 immune-related adverse events were reported.

The frequency of grade 2-4 aGVHD was not influenced by stem cell source, donor type, age at transplant, and previous history of GVHD, but was affected by post-transplant GVHD prophylaxis. Specifically, patients receiving PTCy had a lower observed cumulative incidence of grade 2 to 4 aGVHD than patients who did not receive PTCy (16% vs 22%, P=.07). After controlling for comorbidities and time from transplant to ICI initiation, we found that PTCy was significantly associated with a 90% reduced risk of aGVHD (HR= 0.1; 95% CI: 0.02 - 0.6; P=.01).

Of the 13 evaluable patients who received nivolumab, 4 patients (31%) had an objective response, including 3 patients achieving a complete response (CR) and 1 patient achieving a partial response. The other 9 patients (69%) had no response. Of the 5 patients who received ipilimumab, only 1 patient (20 %) achieved a CR. The other 4 patients (80%) had progression of disease. There was no significant difference between the clinical responses in patients who received PD-1 inhibitor versus CTLA-4 inhibitor (P= 0.65). Patients receiving PTCy had a significantly longer median PFS (22.4 vs. 5.2 months, P=.02), and a trend toward longer median OS (22.4 vs. 5.2 months, P=.08). Fourteen patients (67%) died, and no deaths were related to aGVHD.

Conclusions: The use of ICI for AML/MDS relapse after alloHCT may be a safe and feasible option. PTCy appears to decrease the incidence of aGVHD in this cohort of patients. Given the small number of patients involved in our study, further investigations with a larger number of patients in prospective trials are needed to determine the magnitude of the of PTCy effect on reducing aGVHD after ICI.