Real-World Outcomes of Total Body Irradiation versus Chemotherapy-Based Conditioning in Pediatric Acute Lymphoblastic Leukemia.
AI interpretation is pending for this paper.
Open original publication →What the AI sees
Not AI summarized yet.
Research significance
Pending deeper interpretation.
Source abstract
PURPOSE: To compare outcomes after total body irradiation (TBI)-based and chemotherapy-based conditioning in pediatric acute lymphoblastic leukemia (ALL). PATIENTS AND METHODS: This retrospective two-center PeARL registry cohort comprised 325 patients diagnosed with ALL at age ≤18 years who underwent first allogeneic hematopoietic stem cell transplantation during 2009-2023. Primary Cox and Fine-Gray models used full follow-up, adjusted for transplantation age, remission status and donor type, and were stratified by center. Propensity-score weighting was supplementary. Graft-versus-host disease (GVHD) analyses excluded recorded cases with unavailable or inconsistent onset dates and accounted for competing relapse and death; acute GVHD included any or unspecified grade. RESULTS: Of 325 patients, 210 received TBI-based and 115 chemotherapy-based conditioning; median follow-up by reverse Kaplan-Meier was 5.85 and 1.88 years, respectively. Unweighted 5-year overall survival was 65.5% versus 67.1%, and leukemia-free survival was 57.5% versus 53.4%. Adjusted hazard ratios were 1.30 (95% confidence interval [CI], 0.54-3.12) and 0.92 (0.43-2.01), respectively. Adjusted subdistribution hazard ratios were 0.79 (0.31-2.07) for relapse and 1.39 (0.43-4.49) for non-relapse mortality (NRM); supplementary propensity-score-weighted NRM estimates ranged from 4.20 to 6.48. Among patients included in the respective GVHD time-to-event analyses, 100-day acute GVHD cumulative incidence was 62.7% versus 32.2% (n=315), and 5-year chronic GVHD cumulative incidence was 24.9% versus 5.2% (n=314). CONCLUSION: Primary analyses did not detect statistically significant survival differences, but clinically important differences could not be excluded. NRM estimates varied across analyses. Non-random treatment selection, unequal follow-up, incomplete outcome recording and residual confounding limit causal interpretation.