Noninvasive testing for mycophenolate exposure in children with renal transplant using urinary metabolomics.

Abstract

Despite the common use of mycophenolate in pediatric renal transplantation, lack of effective therapeuic drug monitoring increases uncertainty over optimal drug exposure and risk for adverse reactions. This study aims to develop a novel urine test to estimate MPA exposure based using metabolomics. Urine samples obtained on the same day of MPA pharmacokinetic testing from two prospective cohorts of pediatric kidney transplant recipients were assayed for 133 unique metabolites by mass spectrometry. Partial least squares (PLS) discriminate analysis was used to develop a top 10 urinary metabolite classifier that estimates MPA exposure. An independent cohort was used to test pharmacodynamic validity for allograft inflammation (urinary CXCL10 levels) and eGFR ratio (12mo/1mo eGFR) at 1 year. Fifty-two urine samples from separate children (36.5% female, 12.0± 5.3 years at transplant) were evaluated at 1.6± 2.5 years post-transplant. Using all detected metabolites (n= 90), the classifier exhibited strong association with MPA AUC by principal component regression (r= 0.56, p< .001) and PLS (r= 0.75, p< .001). A practical classifier (top 10 metabolites; r= 0.64, p< .001) retained similar accuracy after cross-validation (LOOCV; r= 0.52, p< .001). When applied to an independent cohort (n= 97 patients, 1053 samples), estimated mean MPA exposure over Year 1 was inversely associated with mean urinary CXCL10:Cr (r= -0.28, 95% CI -0.45, -0.08) and exhibited a trend for association with eGFR ratio (r= 0.35, p= .07), over the same time period. This urinary metabolite classifier can estimate MPA exposure and correlates with allograft inflammation. Future studies with larger samples are required to validate and evaluate its clinical application.