Risk-adjusted monitoring of veno-occlusive disease following Bayesian individualization of busulfan dosage for bone marrow transplantation in paediatrics.

In order to assess the performance of Bayesian individualization of busulfan (BU) dosage regimens, veno-occlusive disease (VOD) rate was monitored for paediatric patients undergoing allogeneic bone marrow transplantation (BMT). Consecutive patients undergoing allogeneic BMT with BU as conditioning regimen during 5-years period (January 2000-February 2006) were reviewed. VOD was a major outcome variable. Preconditioning risk of VOD was estimated for each patient using a scoring system that included type of transplant, recipient CMV-positive status and total parenteral nutrition (TPN) provided pretransplantation. A risk-adjusted cumulative sum method was used to compare observed versus predicted outcome by assigning a risk score, based on log-likelihood ratios, to each patient. These cumulative scores were sequentially plotted with preset control limits for 'signalling' where results were substantially different than expected (doubling or halving of odds ratio). Sixty-six children received BMT after oral busulfan-based conditioning regimen with median age 3.9 years, 63.6% of male. Median preconditioning risk of VOD was 0.34 ranging from 0.23 to 0.84. Observed VOD rate was 16.7% (n = 11) which was 60.7% (17) fewer than the expected number estimated by the risk score. The resulting risk-adjusted score for each patient was plotted sequentially. This plot adopted early a negative slope, crossing the lower control limit twice, after 27 and 66 patients, indicating improved results compared to those expected. Bayesian individualization of oral busulfan dosage regimens is useful to reduce the VOD rate in children undergoing allogeneic BMT.