HLA-matching with PTCy: a reanalysis of a CIBMTR dataset with propensity score matching and donor age

Abstract

Blood or marrow transplantation (BMT) outcomes using haploidentical donors (Haplo) and posttransplant cyclophosphamide (PTCy)-based graft-versus-host disease (GVHD) prophylaxis compare favorably to HLA-matched donors using calcineurin inhibitor-based prophylaxis. A recent Center for International Blood and Marrow Transplant Research analysis of patients receiving homogenous PTCy-based prophylaxis found that, with reduced intensity conditioning, Haplo BMTs had worse outcomes than matched unrelated donor (MUD) BMTs. Due to significant differences between groups, we reanalyzed the dataset using propensity score matching and, additionally, added a donor age variable. After matching MUD BMTs to Haplo BMTs in a 1:5 ratio, no significant differences were found between groups across all measured baseline characteristics. Outcomes analyses demonstrated no significant differences in overall survival (hazard ratio [HR] of mortality with MUD vs Haplo [95% confidence interval], 0.95 [0.65-1.16], P = .75), disease-free survival (HR of relapse or death, 0.98 [0.73-1.18], P = .89), relapse rate (HR, 1.06 [0.77-1.38], P = .69), or nonrelapse mortality (NRM) (HR, 0.85 [0.42-1.13], P = .49) between groups. After stratification by conditioning intensity, MUD BMTs in the reduced-intensity cohort had lower risk of NRM (HR, 0.56 [0.14-0.99], P = .05), with no significant difference in other clinical outcomes. These results suggest the effect of HLA matching on BMT outcomes with PTCy is less meaningful than previously reported. Timely identification of a young, at least half-matched (related or unrelated) donor may be more important than finding a fully matched donor if the latter leads to a delay in BMT or use of an older donor.

Graphical abstract

Figure 1.

Distribution of patient, disease, and BMT-related characteristics before and after propensity score matching. (A) Box plots demonstrate the distribution of age, donor age, and time from diagnosis to BMT by donor type before (left) and after (right) propensity score matching. Age and donor age, marked by an asterisk and pink shading, were significantly different between the 2 groups before propensity score matching but not afterward. (B-C) Bar graphs demonstrate the frequency of other covariables by donor type before (left) and after (right) propensity score matching. Race, year of BMT, and conditioning covariables marked with an asterisk and colored pink were significantly different between the Haplo and MUD groups before propensity score matching. There were no significant differences between the Haplo and MUD groups for any of the covariables after propensity score matching. BM/PB, bone marrow/peripheral blood; Condition, conditioning; DRI, disease risk index; Dx, diagnosis; HCT, recipient's Hematopoietic Cell Transplantation-specific Comorbidity Index; KPS, Karnofsky performance score; Recip. CMV, recipient's CMV status; TBI, total body irradiation.

Figure 2.

Outcomes for propensity-matched patients by haploidentical donor vs MUD. Data are for the whole cohort, including patients that received reduced-intensity and myeloablative conditioning. The HR for MUDs compared with Haplos: (A) OS was 0.95 (95% CI, 0.65-1.16), (B) DFS was 0.98 (95% CI, 0.73-1.18), (C) relapse was 1.06 (95% CI, 0.77-1.38), (D) NRM was 0.85 (95% CI, 0.42-1.13), (E) grade 3 to 4 aGVHD was 0.47 (95% CI, 0.16-0.85), and (F) cGVHD was 0.65 (95% CI, 0.48-0.92). aGVHD, acute graft-versus-host disease; BM/PB, bone marrow/peripheral blood; cGVHD, chronic graft-versus-host disease; Condition, conditioning; HCT, recipient Sorror comorbidity score; Recip. CMV, recipient CMV status.

Figure 3.

Outcomes for propensity-matched patients who received RIC. For patients that received RIC, the HR for MUDs compared with Haplos: (A) OS was 0.82 (95% CI, 0.54-1.26), (B) DFS was 0.88 (95% CI, 0.59-1.23), (C) relapse was 1.03 (95% CI, 0.69-1.58), (D) NRM was 0.56 (95% CI, 0.14-0.99), (E) grade 3 to 4 aGVHD was 0.62 (95% CI, 0.1-1.08), and (F) cGVHD was 0.65 (95% CI, 0.34-1.14). BM/PB, bone marrow/peripheral blood; Condition, conditioning; HCT, recipient Sorror comorbidity score; Recip. CMV, recipient CMV status.

Figure 4.

Outcomes for propensity-matched patients who received myeloablative conditioning. For patients that received myeloablative conditioning, the HR for MUDs compared with Haplos: (A) OS was 0.99 (95% CI, 0.59-1.48), (B) DFS was 1.10 (95% CI, 0.74-1.60), (C) relapse was 1.12 (95% CI, 0.68-1.85), (D) NRM was 1.07 (95% CI, 0.52-1.82), (E) grade 3 to 4 aGVHD was 0.63 (95% CI, 0.10-1.20), and (F) cGVHD was 0.66 (95% CI, 0.43-1.08). BM/PB, bone marrow/peripheral blood; Condition, conditioning; HCT, recipient Sorror comorbidity score; Recip. CMV, recipient CMV status.