Kinetics and Risk Factors of Relapse after Allogeneic Stem Cell Transplantation in Children with Leukemia: A Long-Term Follow-Up Single-Center Study

Abstract Allogeneic hematopoietic stem cell transplantation (allo-HSCT) is an established treatment for high-risk hematological malignancies in the pediatric population, but relapse remains the leading cause of death. We analyzed risk factors associated with relapse. Data from 353 allo-HSCTs from 1989 to 2015 in our center were studied retrospectively. We performed a multivariate analysis of pre- and postransplantation variables and developed a predictive risk score for relapse using the significant factors in this training cohort. The results were confirmed in a validation cohort of 90 allo-HSCTs done in our institution from 2016 to the present. A total of 104 patients relapsed after allo-HSCT, with a relapse cumulative incidence of 31 ± 2%. In multivariate analysis, only 2 variables influenced relapse: disease phase (advanced versus early, HR, 2.84; 95% CI, 1.76 to 4.57; P  = .001) and presence of chronic graft-versus-host disease (GVHD) (acute GVHD versus chronic GVHD [HR, 4.27; 95% CI, 1.99 to 9.15; P  = .0001] and no GVHD versus chronic GVHD [HR, 6.86; 95% CI, 3.63 to 12.97] P  = .0001]. Applying the personalized risk score (0 to 3), the relapse cumulative incidence was 70 ± 5% in patients with a score of 3 (without GVHD and in the advanced phase) compared with 6 ± 4% in patients with a score of 0 (with chronic GVHD and in an early phase). This score has been verified in the validation set. With a median follow-up of 54 months, the disease-free survival (DFS) and overall survival rate were 37 ± 3% and 45 ± 4%, respectively. The association of GVHD with the graft-versus-leukemia effect is clearly established in our study, and the form of GVHD associated with less relapse and the best DFS is the classical form of chronic GVHD according to the National Institutes of Health classification. The proposed relapse risk score was validated in an independent cohort and allows personalization of the prognosis.