Identifying Clinical Predictors of Disease-Free Survival after Haploidentical Transplant at a Single-Center Level

Patient age, HCT-CI, and high-res matching at HLA-B, DRB1, DQB1, DPB1, and donor/recipient CMV serostatus have previously been validated as predictors of 1 and 2-year Disease-Free Survival (DFS) after haploidentical transplant with PTCy for patients with leukemia or MDS in a CIBMTR analysis. DFS was optimized with a donor who is HLA-B leader-matched, -DRB1–mismatched, -DQB1–matched, and -DPB1 TCE-non permissive–mismatched, CMV +/+, low HCT-CI, and young recipient age. We hypothesized that covariates found to predict DFS in very large datasets may not be predictive in smaller cohorts at institutional level.Patient data and demographics were sourced from the EMR. Data on haploidentical patients with leukemia or MDS who received MA or RIC conditioning at The James Cancer Center between 2009 and 2021 was utilized for both cohorts. The IPD-IMGT/HLA HLA-B Leader Mismatch Calculator was used to determine the genotype and degree of mismatch at the HLA-B leader. We analyzed the associations between mismatches and DFS among subset of patients with available HLA mismatch data and institutional data to study associations between other clinical variables and DFS. DFS was calculated from date of transplantation to the date of relapse or death, whichever occurred earlier, censoring those alive without relapse at last follow up. Cox proportional hazard models were used to estimate associations between patients’ data and DFS.Total 74 patients (Table 1) were identified with the inclusion criteria and evaluable mismatch data, among which none of the previously identified variables was predictive for DFS aside from CMV donor seropositivity (p=0.042) in univariable analysis (UVA) (Table 2). HLA-B leader match did not predict incidences of relapse, acute or chronic GVHD, or OS (data not shown). After expanding our analysis to 114 patients with more clinical variables (Table 3), we identified several possible predictive markers in UVA. Factors with p<0.1 were included in multivariable analysis (MVA) with age and comorbidity index included as continuous variables. Conditioning intensity, CMV donor seropositivity, and CIBMTR DRI were significant risk factors for DFS in MVA.Our single center analysis of haploidentical patients with leukemia or MDS receiving PTCy is not consistent with the CIBMTR analysis. We did not identify significant differences in DFS for HLA matching at DRB1, DQB1, DPB1, or HLA-B leader. Donor but not recipient CMV seropositivity was predictive of DFS. MA conditioning, CMV donor seropositivity, and CIBMTR DRI were predictive in MVA at our institutional level. The low number of DRB1, DQB1 match and DPB1 mismatch may be attributable. HLA-B leader mismatch was well distributed and did not significantly impact DFS. This may be due to the limited impact of B-leader mismatch. We are further analyzing the impact of CMV donor seropositivity on other outcomes within our haploidentical patients.