Establishment and multicenter external validation of a risk prediction model for de novo intracranial aneurysms based on a systematic review and meta-analysis of 19 cohorts

Abstract

OBJECTIVE A de novo intracranial aneurysm (IA) is a second, new IA that develops in patients with IAs distant from where the initial IA was detected. This study aimed to identify risk factors for de novo IA formation and establish and externally validate a multicenter risk prediction model for de novo IAs. METHODS A systematic review and meta-analysis of existing de novo IA cohorts was conducted to form the derivation cohort. The risk ratios and 95% CIs of each risk factor were calculated. In addition, risk scores included in the model were calculated based on the statistically significant risk factors with their weightings. Then the model was validated in a multicenter external cohort of Chinese patients, and receiver operating characteristic and calibration curves, decision curve analysis, and Kaplan-Meier curves were used to evaluate the model. RESULTS Nineteen studies with 9351 patients, of whom 304 patients (3.25%) developed de novo IAs, were included in the derivation cohort. These patients developed de novo IAs at 2.5–18.5 years during a total follow-up of 3.3–18.8 years. The statistically significant risk factors were age < 60 years, female sex, smoking history, family history of IAs, multiple IAs at initial diagnosis, and initial IAs in the middle cerebral artery, with risk scores of 4, 5, 2, 6, 3, and 3, respectively. Then, a multicenter external cohort comprising 776 patients, of whom 45 patients (5.80%) developed de novo IAs, was included in the validation cohort. De novo IAs formed in these patients at a mean of 5.25 years during a mean follow-up of 6.19 years. The area under the curve of the model was 0.804, with a sensitivity of 0.667 and specificity of 0.900, at a cutoff value of 13. The calibration curve, decision curve analysis, and Kaplan-Meier curves also indicated good performance of the model. CONCLUSIONS This prediction model is a convenient and intuitive tool for identifying high-risk patients with de novo IAs. Reasonable use of the model can not only aid in clinical decision-making but also play a positive role in the prevention of aneurysmal subarachnoid hemorrhage to a certain extent.