Development and Internal Validation of a Nomogram for Subtherapeutic Total Aripiprazole Concentrations: A...

A straightforward and user-friendly tool for predicting whether the total aripiprazole concentration in a patient falls below 150 ng/mL is currently unavailable. In this study, we identified independent risk factors that influence the subtherapeutic total aripiprazole concentration and developed a nomogram prediction model, thereby providing a reference for individualized dosing.

This was a retrospective cohort study that included 409 therapeutic drug monitoring samples. According to established guidelines, the recommended plasma total concentration range for aripiprazole and its active metabolite, dehydroaripiprazole, is between 150 and 500 ng/mL. Therefore, samples with concentrations below 150 ng/mL are classified as belonging to the subtherapeutic concentration group, whereas those within the 150-500 ng/mL range are categorized as the therapeutic concentration group. A multivariate logistic regression analysis was performed to identify independent factors that affect the subtherapeutic total aripiprazole concentration and a nomogram prediction model was developed. The Net Reclassification Improvement (NRI) and the Integrated Discrimination Improvement (IDI) were employed to assess whether the nomogram model provides statistically significant incremental value over the daily dose-only model.

Among the 409 samples, 94 (22.98%) had concentrations below 150 ng/mL. Multivariate logistic regression analysis revealed that the male (OR = 3.922, 95% CI: 1.732-9.169, P = 0.0012), lower daily dose (OR = 0.525, 95% CI: 0.444-0.608, P < 0.001) and the use of combined valproate (OR = 4.360, 95% CI: 2.013-9.883, P < 0.001) were independent factors affecting the total aripiprazole concentration less than 150 ng/mL. Based on the results, a risk nomogram model was developed to predict the subtherapeutic total aripiprazole concentration. The nomogram showed robust predictive performance, achieving an area under the curve (AUC) of 0.916. The concordance index (C-index) was found to be 0.924 based on 1,000 bootstrap resamples. Additionally, the model showed good internal calibration (calibration intercept = 0.00, slope = 1.00, Brier score = 0.088), decision curve analysis indicated favorable clinical utility, with a wide range of threshold probabilities (5-95%). Compared with the daily dose-only model, the full nomogram achieved a NRI of 0.631 (95% CI: 0.305-0.930; P < 0.001) and an IDI of 0.0472 (95% CI: 0.0188-0.0963; P < 0.001).

This study established, for the first time, a visualized nomogram model designed to preliminarily predict total aripiprazole plasma concentrations that are below 150 ng/mL. In the future, we will conduct external validation to further evaluate the predictive performance of the model.