Dataset Information

A Prediction Model for Tacrolimus Daily Dose in Kidney Transplant Recipients With Machine Learning and Deep Learning Techniques.

ABSTRACT: Tacrolimus is a major immunosuppressor against post-transplant rejection in kidney transplant recipients. However, the narrow therapeutic index of tacrolimus and considerable variability among individuals are challenges for therapeutic outcomes. The aim of this study was to compare different machine learning and deep learning algorithms and establish individualized dose prediction models by using the best performing algorithm. Therefore, among the 10 commonly used algorithms we compared, the TabNet algorithm outperformed other algorithms with the highest R² (0.824), the lowest prediction error [mean absolute error (MAE) 0.468, mean square error (MSE) 0.558, and root mean square error (RMSE) 0.745], and good performance of overestimated (5.29%) or underestimated dose percentage (8.52%). In the final prediction model, the last tacrolimus daily dose, the last tacrolimus therapeutic drug monitoring value, time after transplantation, hematocrit, serum creatinine, aspartate aminotransferase, weight, CYP3A5, body mass index, and uric acid were the most influential variables on tacrolimus daily dose. Our study provides a reference for the application of deep learning technique in tacrolimus dose estimation, and the TabNet model with desirable predictive performance is expected to be expanded and applied in future clinical practice.

SUBMITTER: Zhang Q

PROVIDER: S-EPMC9197124 | biostudies-literature | 2022

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

A Prediction Model for Tacrolimus Daily Dose in Kidney Transplant Recipients With Machine Learning and Deep Learning Techniques.

Zhang Qiwen Q Tian Xueke X Chen Guang G Yu Ze Z Zhang Xiaojian X Lu Jingli J Zhang Jinyuan J Wang Peile P Hao Xin X Huang Yining Y Wang Zeyuan Z Gao Fei F Yang Jing J

Frontiers in medicine 20220527

Tacrolimus is a major immunosuppressor against post-transplant rejection in kidney transplant recipients. However, the narrow therapeutic index of tacrolimus and considerable variability among individuals are challenges for therapeutic outcomes. The aim of this study was to compare different machine learning and deep learning algorithms and establish individualized dose prediction models by using the best performing algorithm. Therefore, among the 10 commonly used algorithms we compared, the Tab ...[more]

PMID: 35712101

Similar Datasets

Project description:BackgroundCalcineurin inhibitors, including tacrolimus, remain a cornerstone of immunosuppressive therapy after kidney transplantation. However, the therapeutic window is narrow, and nephrotoxic side effects occur with overdose, while the risk of alloimmunization and graft rejection increases with underdose. Liquid chromatography-tandem mass spectrometry (LC-MS/MS) allows quantification of tacrolimus in biological samples from patients. This study investigates the feasibility of quantifying tacrolimus in scalp hair from kidney transplant (KT) recipients and correlates hair tacrolimus concentrations with tacrolimus dosage and blood trough levels. The aim was to provide proof-of-principle for hair tacrolimus drug monitoring in KT recipients.MethodSingle-center prospective study between September 9, 2021 and December 4, 2021, including KT recipients under tacrolimus. Minors, patients with active skin or hair diseases, and patients with scalp hair shorter than 4 cm were excluded from participation. Scalp hair was collected from the posterior vertex of patients, cut into segments, and analyzed for tacrolimus by LC-MS/MS. Patients filled out a questionnaire on hair treatments and washing habits. In parallel, tacrolimus trough levels were measured in whole blood and correlated with hair tacrolimus concentrations.ResultsIn total, 39 consenting KT recipients were included, and hair samples were collected at 53 visits. Tacrolimus was detected in 98% of hair samples from patients exposed to the drug. Tacrolimus hair levels and whole blood trough levels were correlated with a beta coefficient of 0.42 (95% CI: -0.22-1.1, p = n.s.). Age and dark hair affected hair tacrolimus measurements, while different tacrolimus formulations (immediate release vs. extended release), hair washes, and permanent coloring did not. Longitudinal measurements in a subgroup of patients indicate that long-term measurement of hair tacrolimus levels is feasible.ConclusionMeasuring tacrolimus in hair is a potentially reliable method to monitor drug exposure in KT patients. Rapid wash-in effects and consistent concentrations over time indicate that tacrolimus is incorporated into the hair matrix, allowing temporal resolution in the analysis of recent exposure and exposure history. This method provides a simple and low-risk alternative to regular blood sampling, sparing patients from frequent hospital visits through the self-collection of hair samples.

Project description:BackgroundTacrolimus is available as twice-daily Prograf® (Tac-BID) and the once-daily formulation, Advagraf® (Tac-OD). Although therapeutically equivalent, some transplant recipients require dose adjustments to achieve similar tacrolimus trough concentrations [Tac C0] after conversion between formulations. Tacrolimus is primarily metabolized by cytochrome P450 3A5 (CYP3A5). We sought to determine whether genetic polymorphisms in the CYP3A5 enzyme; CYP3A5 *1/*1 and CYP3A5 *1/*3 (expressers) compared to CYP3A5 *3/*3 (non-expressers) could account for discrepancies in dose requirements following conversion from Tac-BID to Tac-OD.MethodsA cohort of 60 renal transplant recipients (RTR) from our larger conversion study of 496 patients underwent additional testing for CY3A5 genetic polymorphisms. Analysis included demographics, tac dosing and [Tac C0] pre- and post-conversion and dosing changes relative to CYP3A5 genotypes. CYP3A5 genetic polymorphisms were identified through analysis of genomic DNA.ResultsConversion from tac bid to tac OD in this cohort required a mean (SD) dose increase from 3.1 (1.0) mg/day to 3.8 (1.3) mg/day (p = 0.007), to achieve similar [Tac C0]. The *1/*3 expresser group required a greater percentage dose adjustment (56.7 %) in converting from Tac-BID to Tac-OD as compared to the *3/*3 non-expresser group (26.6 %). Similar findings were observed with the both expresser groups combined (*1/*1 &*1/*3). The expressers were significantly more highly represented in the East Asian cohort.ConclusionsThe CYP3A5 expresser polymorphism necessitates an increase in dosing upon conversion from Tac-BID to Tac-OD, with the expresser genotypes contributing significantly to this finding. Given the variability in frequency of CYP3A5 genotypes in various ethnic groups, future studies should account for both isoenzyme polymorphism and ethnicity in optimizing dosing requirements.Trial registrationClinical trials.gov identifier: NCT01884480.

Project description:IntroductionFew studies have compared within-patient variability measures of tacrolimus trough levels by formulation and assessed within-patient variability on outcomes of kidney transplant recipients.Research questions(1) To compare within-patient variability of trough levels when converting from twice-daily to once-daily tacrolimus using standard deviation, coefficient of variation, and intrapatient variability percent. (2) To use the 3 measures of variability to examine the relationship between tacrolimus once-daily within-patient variability and total graft failure (i.e., return to chronic dialysis, pre-emptive retransplant, death with graft function).DesignIn this observational cohort study, within-patient variability of trough levels pre- and post-conversion from twice-daily to once-daily tacrolimus were compared using Wilcoxon matched-pairs signed-rank test. Graft outcomes were analyzed using Kaplan-Meier curves and multivariable Cox proportional hazards models.ResultsIn 463 patients, within-patient variability differences pre- and post-conversion of median standard deviation, coefficient of variation, and intrapatient variability percent were -0.16 (P = 0.09), -0.01 (P = 0.52), and -1.41 (P = 0.32), respectively. Post-conversion, every 1 unit increase in within-patient variability standard deviation and intrapatient variability percent and every 0.1 unit increase in the coefficient of variation was associated with an increased hazard ratio [1.19 (P = 0.004), 1.02 (P = 0.030), 1.13 (P = 0.001), respectively] of total graft failure. Post-conversion, within-patient variability above cohort medians using standard deviation and coefficient of variation had a significantly higher risk of total graft failure.DiscussionUnder a program-wide conversion, no significant difference was observed in within-patient variability post-conversion from twice-daily to once-daily tacrolimus using the three measures of variability. High within-patient variability was associated with adverse transplant outcomes post-conversion.

Dataset Information

A Prediction Model for Tacrolimus Daily Dose in Kidney Transplant Recipients With Machine Learning and Deep Learning Techniques.

Publications

A Prediction Model for Tacrolimus Daily Dose in Kidney Transplant Recipients With Machine Learning and Deep Learning Techniques.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets