Myeloablative Conditioning for First Allogeneic Hematopoietic Stem Cell Transplantation in Children with ALL: Total Body Irradiation or Chemotherapy? - a Multicenter EBMT-PDWP Study

Introduction: Most pediatric patients (pts) with ALL receive total body irradiation (TBI) for myeloablative conditioning of allogeneic hematopoietic stem cell transplantation (allo-HSCT). It is unproven whether TBI can be replaced by chemotherapy (CHT).

Methods: To compare the outcomes of TBI- versus (vs.) CHT-based conditioning, we performed a retrospective EBMT-registry based study. Children between 2 and 18 years of age (y.) after myeloablative conditioning for first allo-HSCT of bone marrow (BM) or peripheral blood SC (PBSC) from matched sibling (MSD) or unrelated donors (UD) in first (CR1) or second remission (CR2) between 2000 and 2012 were included. Propensity score weighting was used to consider the covariate "distribution of TBI recipients" for pts who did not receive TBI.

Results: In total 3071 pts (CR1: 1504 (49%), CR2: 1567 (51%)) were included. CR1: 1045 pts (69%) received BM and 459 pts (31%) PBSC from MSD (760 (51%)) or UD (744 (49%)). CR2: 1067 pts (68%) received BM and 500 pts (32%) PBSC from MSD (675 (43%)) or UD (892 (57%)). Overall, conditioning was TBI- in 2647 (86%) and CHT-based in 424 pts (14%). Busulfan/Cyclophosphamide (Bu/Cy) and Bu/Cy/Etoposide (Bu/Cy/Eto) were the two most frequently applied CHT combinations in CR1 (68 (32%), 66 (31%)) and CR2 (68 (32%), 52 (25%)). The remaining conditionings included 5 different combinations of chemotherapeutics (other chemo).

1504 pts in CR1 were conditioned with TBI (1291), Bu/Cy/Eto (66) or other chemo (147) with a median follow-up of 4.4, 3.4 and 2.4 y. In weighted univariate analysis no significant differences were detected for LFS (5-y.-LFS, range: 62.4 to 67.5%) and relapse incidence (5-y.-RI, range: 24.0 to 29.0%). In pairwise testing, OS after Bu/Cy/Eto was significantly better compared with TBI (5-y.-OS, 78.7 vs. 66.8%, P=.006). Non-relapse mortality was significantly higher after other chemo (5-y.-NRM, 12.7%, P <.001) and TBI (12.5%, P <.001) compared with Bu/Cy/Eto (3.5%).

Other significant influencing factors on LFS were age (5-y.-LFS, 2-11 y. 65.8 vs. 12-18 y. 58.0%, P=.009), y. of HSCT (2008-2012 65.9 vs. 2000-2007 59.6%, P=.035) and donor type (MSD 64.9 vs. UD 59.5%, P=.007). RI was influenced by y. of HSCT (5-y.-RI, 2008-2012 21.7 vs. 2000-2007 26.8%. P=.026). OS was influenced by age (5-y.-OS, 2-11 y. 72.7 vs. 12-18 y. 61.7%, P <.001) and donor type (MSD 70.1 vs. UD 65.0%, P=.002). NRM was influenced by age (5-y.-NRM, 2-11 y. 9.0 vs. 12-18 y. 18.0%, P <.001), donor sex (male 10.9 vs. female 16.6%, P=.007) and donor type (MSD 8.5 vs. UD 17.8%, P <.001).

In weighted multivariable Cox model clustered on EBMT centers, TBI-based conditioning in CR1 was associated with lower RI (HR 0.70, P=.047), lower OS (HR 1.54, P=.017) and higher NRM (HR 3.97, P <.001).

1567 pts in CR2 were conditioned with TBI (1356), Bu/Cy (68) or other chemo (143) with a median follow-up of 3.9, 3.2 and 3.1 y. In weighted univariate analysis highly significant differences of survival were detected. TBI-based conditioning resulted in highest 5-y.-LFS of 52.2% (Bu/Cy 41.0%, other chemo 18.4%, P <.001), lowest 5-y.-RI of 33.2% (Bu/Cy 38.3%, other chemo 55.3%, P <.001), highest 5-y.-OS of 56.1% (Bu/Cy 43.6%, other chemo 23.6%, P <.001) and lowest 5-y.-NRM of 14.6% (Bu/Cy 20.8%, other chemo 26.2%, P <.001).

Other significant influencing factors on LFS were age (5-y.-LFS, 2-11 y. 53.5 vs. 12-18 y. 43.0%, P <.001), y. of HSCT (2008-2012 53.3 vs. 2000-2007 47.3%, P=.009) and SC source (BM 53.3 vs. PBSC 42.5%, P <.001). RI was influenced by SC source (5-y.-RI, BM 33.1 vs. PBSC 36.6%, P=.024). OS was influenced by age (5-y.-OS, 2-11 y. 59.3 vs. 12-18 y. 46.8%, P <.001), y. of HSCT (2008-2012 58.6 vs. 2000-2007 52.0%, P=.005), donor type (MSD 58.5 vs. UD 52.6%, P <.024) and SC source (BM 58.4 vs. PBSC 47.5%, P <.001). NRM was influenced by age (5-y.-NRM, 2-11 y. 12.5 vs. 12-18 y. 22.3%, P <.001), donor type (MSD 10.2 vs. UD 19.8%, P <.001) and SC source (BM 13.6 vs. PBSC 20.8%, P <.001).

In weighted multivariable Cox model clustered on EBMT centers, TBI-based conditioning in CR2 was associated with higher LFS (HR 0.48, P <.001), lower RI (HR 0.48, P <.001) and higher OS (HR 0.51, P <.001).

Conclusion: Conditioning by TBI demonstrated clear superiority in comparison to CHT for children with ALL undergoing HSCT in CR2. For pts in CR1, TBI- and CHT-based conditioning showed similar results. This retrospective data is currently re-evaluated in a prospective, randomized, international trial (ALL SCTped 2012 FORUM).