Scholarlink

Background <p>Tacrolimus, a primary immunosuppressant in kidney transplantation, exhibits variable pharmacokinetics. While model-informed precision dosing (MIPD) using population pharmacokinetic (PopPK) models improve dosing accuracy, models developed in specific populations lack applicability in other populations. We aimed to evaluate the predictive performance of published models integrated in clinical decision support systems (CDSS) in Indian adult renal transplant recipients.</p> Methods <p>Predictive performance of four published PopPK models were evaluated using an independent dataset of 120 trough concentrations from 60 adult renal transplant recipients. Models were assessed using prediction based and simulation-based diagnostics. Maximum <i>a posteriori</i> (MAP) Bayesian prediction was also used to evaluate the model performance.</p> Results <p>A priori predictions (PRED) for all four models demonstrated systematic underprediction, with relative mean prediction error (rME) ranging from − 25.0% to -64.0%, failing clinical acceptability thresholds. However, integrating patient-specific data via MAP Bayesian estimation corrected this bias and improved precision. Notably, all models achieved higher accuracy and precision, with more than 60% of individual predictions falling within <InlineEquation ID="IEq1"><EquationSource Format="TEX">\(\:\pm\:20\text{\%}\)</EquationSource></InlineEquation> (<InlineEquation ID="IEq2"><EquationSource Format="TEX">\(\:{F}_{20}\)</EquationSource></InlineEquation>) of the observed values. Additionally, scaling down the clearance parameter by 30% mitigated the systematic bias observed in the initial a priori predictions.</p> Conclusions <p>Published PopPK models exhibited significant systematic bias when applied to Indian renal transplant recipients, posing a substantial risk of overdosing if used without adjustment for initial dose selection. However, these models adapted using MAP Bayesian approach or scaling of clearance parameter demonstrate better predictive performance in this cohort, suggesting they can guide proactive precision dosing in the Indian cohort. Prospective studies evaluating the target attainment and clinical outcomes are required to establish their clinical applicability.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

External evaluation of population pharmacokinetic models for tacrolimus in kidney transplant recipients: implications for model-informed precision dosing

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月22日

查看全文
  • Arun Prasath Raju,
  • Priyanka Naithani,
  • Deepesh Kenwar,
  • Sarbpreet Singh,
  • Shiva Kumar SP,
  • Savita Verma Atri,
  • Vivek Kumar,
  • Anjali Agarwal,
  • Ravindra Prabhu Attur,
  • Mahadev Rao,
  • Ashish Sharma,
  • Smita Pattanaik,
  • Surulivelrajan Mallayasamy

摘要

Background

Tacrolimus, a primary immunosuppressant in kidney transplantation, exhibits variable pharmacokinetics. While model-informed precision dosing (MIPD) using population pharmacokinetic (PopPK) models improve dosing accuracy, models developed in specific populations lack applicability in other populations. We aimed to evaluate the predictive performance of published models integrated in clinical decision support systems (CDSS) in Indian adult renal transplant recipients.

Methods

Predictive performance of four published PopPK models were evaluated using an independent dataset of 120 trough concentrations from 60 adult renal transplant recipients. Models were assessed using prediction based and simulation-based diagnostics. Maximum a posteriori (MAP) Bayesian prediction was also used to evaluate the model performance.

Results

A priori predictions (PRED) for all four models demonstrated systematic underprediction, with relative mean prediction error (rME) ranging from − 25.0% to -64.0%, failing clinical acceptability thresholds. However, integrating patient-specific data via MAP Bayesian estimation corrected this bias and improved precision. Notably, all models achieved higher accuracy and precision, with more than 60% of individual predictions falling within \(\:\pm\:20\text{\%}\) (\(\:{F}_{20}\)) of the observed values. Additionally, scaling down the clearance parameter by 30% mitigated the systematic bias observed in the initial a priori predictions.

Conclusions

Published PopPK models exhibited significant systematic bias when applied to Indian renal transplant recipients, posing a substantial risk of overdosing if used without adjustment for initial dose selection. However, these models adapted using MAP Bayesian approach or scaling of clearance parameter demonstrate better predictive performance in this cohort, suggesting they can guide proactive precision dosing in the Indian cohort. Prospective studies evaluating the target attainment and clinical outcomes are required to establish their clinical applicability.