In Japan, unrelated bone marrow transplantations (U-BMT) for severe aplastic anemia (SAA) were first performed in 1993. Transplant regimens and patient selection have changed substantially since then. We aimed to determine the effect of these changes on transplant outcome. We retrospectively analyzed the outcome in patients with SAA who received U-BMT through the Japan Marrow Donor Program between 1993 and 2005. We selected 302 recipient-donor pairs in which molecular analysis of HLA-A, -B, -C, -DRB1, and DQB1 were performed. Patient ages ranged from 1 to 64 years (median, 17 years). Various preconditioning regimens were used by individual centers. Either tacrolimus with methotrexate or cyclosporine with methotrexate was used for the prophylaxis against graft-versus-host disease (GVHD) in 137 (45%) patients and 127 patients (42%), respectively. Of the 302 pairs, 104 (34%) were found to be matched at HLA-A, -B, -C, -DRB1, and DQB1; 97 (32%) were mismatched at a single HLA allele (23 HLA-A or -B, 42 HLA-C, 32 HLA-DRB1 or HLA-DQB1); 83 (28%) were mismatched at two HLA alleles (35 HLA-A or -B and HLA-C, 8 HLA-A or -B and HLA-DRB1 or -DQB1, 40 HLA-C and HLA-DRB1 or -DQB1); and 18 (6%) were mismatched at three HLA alleles. Currently, 210 of the 302 patients are alive with the median follow-up period of 722 days after transplantation. The incidence of graft failure was 2.2%; that of grade II/IV GVHD was 29.0%; that of III/IV acute GVHD was 13.8%; and that of chronic GVHD was 25.2%. Multivariate analysis revealed the following significant risk factors for survival: patients older than 15 years (RR, 1.98; range 1.17–3.35); HLA-A or -B + HLA-C or HLA-DQB1 or -DRB1 (RR 2.18; range 1.29–3.68); three loci mismatching (RR 3.14; range 1.47–6.69); prophylaxis against GVHD with tacrolimus with methotrexate (RR 0.45; range 0.28–0.73), ABO major mismatch (RR 1.67; range 1.05–2.63), and non-ATG-containing CY+TBI regimen (RR 2.17; range 1.16–4.03). Patients were divided into two cohorts based on years of transplantation and we compared patients transplanted within two time periods: 1993–2000 and 2001–2005. Five-year survival increased 55.9+/−4.6% in the 1993–2000 cohort (n=116) and 72.7+/−3.8% in the 2001–2005 cohort (n=186) (p=0.008). Patients and transplant characteristics that differed significantly between the two periods were patients’ age distribution, GVHD prophylaxis, HLA matching, and conditioning regimens. The percentage of patients older than 15 years was significantly larger in the recent period (p<0.001). However, the use of cyclosporine and non-ATG-containing CY+TBI regimen were more frequent in the earlier period (p<0.001). HLA-mismatching between patients and donors was more frequent in 1993–2000 (p<0.001). In conclusion, U-BMT for SAA has improved significantly in the recent 5-years period. Improved HLA matching, avoiding non-ATG-containing CY+TBI regimen, and the use of tacrolimus for GVHD prophylaxis account for this improvement.

Author notes

Disclosure: No relevant conflicts of interest to declare.