Dynamic prediction based on variability of a longitudinal biomarker

Area of Science:

• Nephrology
• Immunology
• Pharmacometrics

Background:

• Tacrolimus is essential for immune suppression after kidney transplantation.
• High Tacrolimus variability may indicate poor adherence and increase adverse events.
• Assessing Tacrolimus variation is critical for patient outcomes.

Purpose of the Study:

• To compare methods for assessing Tacrolimus variability's influence on de novo Donor Specific Antibodies (dnDSA) risk.
• To evaluate joint models for predicting dnDSA development post-transplant.
• To determine if individual-specific variability improves predictive performance.

Main Methods:

• Utilized data from the University of Colorado transplant program.
• Compared multiple joint models, focusing on predictive accuracy.
• Incorporated patient-specific random error terms into longitudinal Tacrolimus models.

Main Results:

• Models accounting for individual Tacrolimus variability demonstrated superior predictive performance.
• Higher Tacrolimus variability (variance and CV) was associated with an increased hazard of dnDSA.
• Individual-specific variability improved model fit and prediction of time-to-dnDSA.

Conclusions:

• Individual Tacrolimus variability is a key predictor of long-term adverse events in kidney transplant recipients.
• This finding supports personalized Tacrolimus dosing to enhance post-transplant outcomes.
• Understanding drug variability is vital for optimizing immunosuppression strategies.