Anne-Kristina Frobel (1), Mats O Karlsson (1), Janne T Backman (2), Kalle Hoppu (2), Erik Qvist (3), Paula Seikku (3), Hannu Jalanko (3), Christer Holmberg (3), Ron J Keizer (1), Samuel Fanta (2), Siv Jönsson (1)
(1) Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden; (2) Department of Clinical Pharmacology, University of Helsinki, and HUSLAB, Helsinki University Central Hospital, Helsinki, Finland; (3) Department of Pediatric Nephrology and Transplantation, Hospital for Children and Adolescents, Helsinki University Central Hospital, Helsinki, Finland
Objectives: Ciclosporin A (CsA) immunosuppression after paediatric renal transplantation remains challenging due to a narrow therapeutic range and limited information in children. Also, the target CsA exposure (AUC) is under debate. The aim of this study was to develop a model for the time to first acute rejection (AR) and to explore predictive factors for therapy outcome.
Methods: Data was extracted from patient records at the hospital for Children and Adolescents in Helsinki, Finland (1995-2006). A parametric survival model implemented in NONMEM 7.2 was used to describe the time to first AR. The influence of AUC and other covariates on the hazard were explored using stepwise covariate modelling (SCM), bootstrap-(boot-) SCM (1) and cross-validated (XV-) SCM (2). The clinical relevance of the potential effects was assessed with the time at which 90% of the patients were AR-free (T90), taking into account parameter uncertainty and covariate distributions.
Results: Data on 87 patients (0.7-19.8 years), whereof 54 experienced an AR, was analysed. An exponential survival model based on a function of discrete constant hazards changing in steps over time, (peak hazard day 5-8 after transplantation), described the data best. Dialysis time, sex and baseline weight were identified in the SCM as potential, but were not statistically significant covariates (p>0.01), which was confirmed by boot- and XV-SCM procedures. The boot-SCM demonstrated low inclusion rates for all relationships and selection bias in covariate effect sizes. The XV-SCM showed the best predictive performance for a model without covariates. The median T90 changed by about 1 day for different covariate values; for the strongest covariate found, dialysis time, median T90 was 5.8 days (90% confidence interval 5.1-6.8) for long (90. percentile) and 7.4 (6.4-11.7) days for short (10. percentile) times.
Conclusions: The data was described in a survival model based on a step function of hazards. Covariate effects were not statistically significant and the predicted clinical relevance of the effects was low. Boot-SCM and XV-SCM methods were successfully applied and discouraged inclusion of any covariates into the model. Daily CsA AUC was not identified as a covariate, suggesting that within the observed range (90% interval 1.13–8.40 h*mg/l), an increase in AUC did not increase protection from AR. This feedback on current therapy may help avoiding unnecessarily high, potentially toxic dosing of CsA in children.
Reference: PAGE 21 (2012) Abstr 2374 [www.page-meeting.org/?abstract=2374]
Poster: Paediatrics