Bayesian Networks: A New Approach to Predict Therapeutic Range Achievement of Initial Cyclosporine Blood Concentration After Pediatric Hematopoietic Stem Cell Transplantation

Abstract

BackgroundPediatric hematopoietic stem cell transplantation (HSCT) allows the treatment of numerous diseases, both malignant and non-malignant. Cyclosporine, a narrow therapeutic index drug, is the major immunosuppressant used to prevent graft-versus-host disease (GVHD), but may also cause severe adverse effects in case of overdosing.ObjectiveThe objective of this study is to predict the initial cyclosporine residual blood concentration value after pediatric HSCT, and consequently the dose necessary to reach the therapeutic range, using a mathematical individual predictive model.MethodsClinical and biological data collected from the graft infusion for 2 months after transplantation in 155 pediatric patients undergoing HSCT between 2008 and 2016 were used to generate synthetic data for 1000 subjects which were used to build a Bayesian network model. We compared the characteristics and sensitivity to clinical or biological missing data of this model with four other methods.ResultsThe tree-augmented Naïve Bayesian network showed the best characteristics, with no missing data (area under the curve of the receiving operator characteristics curve [AUC-ROC] of 0.89 ± 0.02), 18.9 ± 2.6% of patients misclassified, and positive and negative predictive values of 85.9 ± 3.4% and 74.2 ± 5.1%, respectively, and this trend is found in the synthetic dataset from no to 10% missing data. The most relevant variables that could influence whether the initial residual cyclosporine concentration is in the therapeutic range are the last dose before measurement and the mean dose before measurement.ConclusionsWe developed and cross-validated an online Bayesian network to predict the first cyclosporine concentration after pediatric HSCT. This model allows simulation of different dosing regimens, and enables the best dosing regimen to reach the therapeutic range immediately after transplantation to be found, minimizing the risk of adverse effects and GVHD occurrence.

A Systematic Review of Machine Learning Techniques in Hematopoietic Stem Cell Transplantation (HSCT)

“…However, drawbacks of RF include complexity, including interpretability of the findings. Interestingly, we found that BN was used in two studies where they were used to predict optimal cyclosporine drug dosing to prevent GVHD [34,42]. The unique ability of BN to encode domain knowledge and consider various variability factors make them attractive in the HSCT setting.…”

“…Leclerc et al [34] proposed an ML approach to predict appropriate cyclosporine drug dosage after pediatric HSCT to prevent acute GVHD. The authors employed a Bayesian learning model on clinical and biological data collected from 155 pediatric patients.…”

Goal-Oriented Monitoring of Cyclosporine Is Effective for Graft-versus-Host Disease Prevention after Hematopoietic Stem Cell Transplantation in Sickle Cell Disease and Thalassemia Major

“…Further investigations are required, but this study opens new perspectives to improve survival and safety of HSCT from alternative donors both in TM and SCA to levels comparable with that obtained with MSDs. Moreover, the recent description of a Bayesian network to predict therapeutic range achievement of initial CsA blood concentrations after pediatric HSCT offers interesting perspectives to continue to improve monitoring of this drug characterized by a narrow therapeutic index [38].…”

Generation of Virtual Patients for in Silico Cardiomyopathies Drug Development

“…The lack of significant computational complexity followed by an increase in the quality of the virtual data is an advantage of the proposed methodology, especially in the case of largescale in silico clinical trials where the number of virtual patients to be generated is significantly larger. As a future work additional methods for virtual population generation, such as the Bayesian networks [17], [18] and the modified genetic function [19], along with neural network-based strategies [20] will be employed for comparison.…”