Identification of Microbiome Associations with Tacrolimus Pharmacokinetics in Adult Hematopoietic Cell Transplantation Using Population Pharmacokinetic and Machine Learning.

PURPOSE: Tacrolimus (TAC) is known for its high pharmacokinetic variability which cannot be fully explained by pharmacogenomic (PGx) and clinical variables. We identified gut microbiome associated with TAC pharmacokinetic variability in allogeneic hematopoietic cell transplant (HCT) recipients. METHODS: In this observational study, metagenomic shotgun sequencing was used to analyze stool microbiome collected within ± 10 days from time of first oral TAC trough at steady state. TAC steady state concentrations (222 IV continuous infusion and 436 oral troughs) were modeled to estimate TAC clearance (CL) and oral bioavailability (F) using nonlinear mixed effects modeling. The effect of clinical covariates, PGx variants and concomitant medications on CL and F were evaluated. Machine learning was used to identify bacterial species associated with variability in F and CL. The identified species were incorporated into the final model, and simulations were conducted to estimate their clinical relevance on oral TAC troughs. RESULTS: TAC population CL was 6.91 L/h and population F was 64.4%. TAC CL was increased in those with CYP3A5*1 genotype and reduced with voriconazole use and if estimated glomerular filtration rate < 60 ml/min/1.73 m2. TAC F increased with laxative use and decreased with corticosteroid use. Limosilactobacillus fermentum had the strongest positive effect on oral TAC troughs while Bacteroides uniformis had the strongest negative effect. CONCLUSION: Gut microbiome contributes to the inter-patient variability in TAC CL and oral F.