A Multivariate prediction model for amlodipine therapeutic efficacy in pediatric primary hypertension.

Abstract

Background: There are currently no biomarker-based prediction models for amlodipine therapeutic efficacy in pediatric hypertension. This study aimed to identify potential biomarkers and establish a biomarker-based model for predicting amlodipine therapeutic efficacy in pediatric primary hypertension (PH).

Methods: From January 2022 to December 2023, 165 children and adolescents with PH prescribed amlodipine were recruited at our department for a prospective observational study. Patients were grouped into Responders and Non-responders after one month treatment. The baseline data in the two groups were analyzed to identify variables associated with amlodipine treatment responsiveness; furthermore, a nomogram prediction model was established based on those potential predictors derived from multivariate regression analysis. This model's discrimination and calibration were evaluated by a series of statistical methods and internal validation was done using the bootstrap sampling method (1000 resamples).

Results: Eighty-nine patients responded to amlodipine while 76 did not. After statistical adjustment, 4 variables were found to be independently associated with therapeutic efficacy, including hyperinsulinemia (OR = 3.000, 95% CI: 1.409-6.386, p  = 0.004), insulin resistance (OR = 2.354, 95% CI: 1.032-5.370, p = 0.042), the baseline plasma Endothelin-1 level (OR = 0.627, 95% CI: 0.532-0.740, p < 0.001) and amlodipine dosages (OR = 1.743, 95% CI: 1.400-2.169, p <0.001). Compared to the baseline model, the full model with the four variables had a good calibration with an area under the curve (AUC) of 0.967 (95% CI: 0.945-0.990), yielding a sensitivity and a specificity of 91.0% and 92.1%, respectively; the clinical decision curve showed a positive net benefit. Additionally, a nomogram model was established based on the four variables and evaluated by bootstrap internal validation with the c-statistic of 0.865 and the calibration curve being close to the ideal line (p > 0.05).

Conclusion: A nomogram model with high predictive value for amlodipine therapeutic efficacy in pediatric PH was established. This model may be potentially applied to guide the selection of amlodipine for the treatment of pediatric PH.