Comparison Between Related T-Cell Depleted HLA-Haploidentical Stem Cell Transplantation (TCD-Haplo) and Umbilical Cord Blood Transplantation (UCBT) in Pediatric Patients with Acute Leukemia, a Eurocord, PDWP-EBMT Study

Both TCD-Haplo and UCBT are used for treating children with either acute lymphoblastic (ALL) or myeloid (AML) leukemia in need of an allograft and lacking a suitable donor. Although both these types of HSCT have been shown to be effective in curing children with acute leukemia, to date, no study has compared the outcomes of these two types of transplant.

We performed a retrospective registry-based study on children (less than 18 years) with either ALL or AML, receiving, after a myeloablative conditioning regimen, a TCD-Haplo (CD34⁺ cell positive selection, CD3⁺ negative or TCR alpha/beta⁺ cell depletion) or single unit UCBT. Patients given pharmacological graft-versus-host disease (GVHD) prophylaxis after graft infusion in haplo HSCT were excluded.

Transplants were performed from 2001 to 2012 in EBMT centers; 1067 patients received single UCBT and 266 TCD-Haplo for AML (n=478) or ALL (n=855). Median follow up was 28 (range 1-150) and 20 (range 1-152) months for UCBT and TCD-Haplo, respectively.

Compared to TCD-Haplo, UCBT recipients were younger (median age 5.96 years vs 9.6 years, p=<0.0001), were transplanted more frequently in CR1 (42% vs 24%, p=<0.001), less frequently in advanced disease (10% vs 21%, p=<0.001) and had more often negative cytomegalovirus (CMV) serology (45% vs 31%, p=<0.001).

Conditioning regimen was mainly total body irradiation (TBI)-based, 56% versus 50% for TCD-Haplo and UCBT, respectively. For UCBT, patients CSA+steroids was the most commonly used (72%) GVHD prophylaxis. A higher proportion of patients transplanted with TCD-Haplo received ATG in the conditioning regimen (90% vs 74%, p=<0.001).

Acute GVHD (grade II-IV) incidence was 16% and 28% (p<0.001) while that of chronic GVHD was 14% and 16%, (p=0.40) for TCD-Haplo and UCBT respectively. Since diagnosis was the most important factor influencing outcome, the analysis was performed separately for patients with ALL and AML.

For ALL (n=855) in univariate analysis, the 2-year probability of leukemia-free survival (LFS) was 35% and 43% (p=0.08), for TCD-Haplo and UCBT, respectively. The cumulative incidence (CI) of non-relapse mortality (NRM) was 28% and 29% (p=0.57), and relapse incidence (RI) was 36% and 28% (p=0.01) for TCD-Haplo and UCBT, respectively.

According to disease status, for patients in CR1 (n=300), 2-year RI was 21% versus 20%, p=0.75; NRM 30% versus 24%, p=0.59; and 2- year LFS 49% versus 56%, p=0.87; for TCD-Haplo and UCBT, respectively.

For patients in CR2 (n=478), 2-year RI was 34% versus 29%, p=0.75; NRM 28% versus 32%, p=0.29; and 2- year LFS 38% versus 39%, p=0.63, for TCD-Haplo and UCBT, respectively. For patients in advanced disease (n=77), 2-year LFS was 5% versus 8%, p=0.08 for TCD-Haplo and UCBT, respectively.

For patients with AML (n=478), 2-year probability of LFS was 21% and 58% (p=<0.0001), for TCD-Haplo and UCBT, respectively. CI of NRM was 43% and 19% (p=<0.001), and RI was 36% and 23% (p=0.01) for TCD-Haplo and UCBT, respectively.

According to disease status, for patients in CR1 (n=207), 2-year RI was 31% versus 14%, p=0.06; NRM 45% versus 16%, p=0.002; and 2-year LFS 24% versus 69%, p=<0.001; for TCD-Haplo and UCBT, respectively.

For patients in CR2 (n=180), 2-year RI was 28% versus 24%, p=0.49; NRM 44% versus 21%, p=0.004; and 2- year LFS 28% versus 55%, p=0.0005, for TCD-Haplo and UCBT, respectively. For patients in advanced disease status (n=91), 2-year LFS was 10% versus 22%, p=0.44 for TCD-Haplo and UCBT, respectively.

In multivariate analysis adjusted for differences between the 2 groups, for patients with ALL, no statistically significant differences were observed according to the graft source. Disease status was the only factor associated with better LFS (HR 4.84, p<0.0001). TCD-Haplo was associated with greater risk of RI (HR 1.58, p=0.01).

For AML, in multivariate analysis, TCD-Haplo was associated with greater risk of relapse (HR 1.67, p=0.05) and of NRM (HR= 1.94; p=<0.001), and worse LFS (HR 1.94, p=<0.001) when compared to UCBT. Advanced disease at transplantation (HR 2.89, p=<0.001) was the other factor associated with LFS and RI.

This retrospective analysis demonstrates that children with ALL have comparable probability of LFS after either UCBT or TCD-Haplo. By contrast, in children with AML, UCBT is associated with lower risk of relapse and NRM than TCD-Haplo, this translating into better LFS. These results may help guide physician choices for transplanting children with acute leukemia.