Abstract:
Objective To evaluate the predictive ability and influencing factors of individualized drug administration adjuvant decision-making system Java PK® for Desktop (JPKD) for tacrolimus blood concentration in kidney transplant recipients.
Methods The monitoring data of tacrolimus blood concentration from 149 recipients early after kidney transplantation were collected. The trough blood concentration of tacrolimus was predicted by JPKD. The absolute weighted deviation and relative prediction deviation between the actual and predicted concentration were calculated. The influencing factors of the absolute weighted deviation were analyzed by univariate and multivariate logistic regression analyses, and the predictive values of these influencing factors on the accuracy of software prediction were assessed by delineating the receiver operating characteristic (ROC) curve.
Results Two hundred and sixty-six samples of tacrolimus blood concentration data were collected from 149 patients. The measured blood concentration of tacrolimus was (6.5±3.0) ng/mL (1.1-16.6 ng/mL), and the predicted value calculated by JPKD was (5.6±2.5) ng/mL (1.4-14.4 ng/mL). The absolute weighted deviation of the calculated data was 28.38%, and the relative prediction deviation was −13.55%. Univariate analysis showed that gender, albumin, changes in hematocrit, cytochrome P450 (CYP)3A5*3 genotype and C3435T genotype were associated with the inaccurate prediction results. Multivariate logistic regression analysis found that CYP3A5*3 genotype of AA and the changes in hematocrit were the independent risk factors affecting the accuracy of tacrolimus blood concentration predicted by JPKD. ROC curve analysis showed that when the changes in hematocrit exceeded 2.25%, the risk of inaccurate software prediction was increased.
Conclusions JPKD possesses certain accuracy in predicting the blood concentration of tacrolimus in kidney transplant recipients, which may improve the qualified rate of tacrolimus blood concentration. Nevertheless, CYP3A5*3 genotype and the changes of hematocrit may affect the accuracy of predictions.