Development of a prediction model for aneurysmal events to guide imaging surveillance in autosomal dominant polycystic kidney disease.

Objective

Patients with autosomal dominant polycystic kidney disease (ADPKD) face a higher prevalence of intracranial aneurysms. However, the risk factors for aneurysmal events-particularly subarachnoid hemorrhage (SAH) or aneurysm growth, morphological changes, or de novo formation-remain poorly understood, leading to a lack of standardized policy of follow-up. This study aimed to identify predictors of such high-risk events and develop a risk prediction model to guide follow-up strategies.

Methods

This retrospective cohort study was conducted at 2 ADPKD referral centers in Japan. A total of 1267 adult patients with ADPKD who underwent brain imaging between 2003 and 2024 at 2 Japanese referral centers were eligible, and 132 patients with untreated intracranial aneurysms and follow-up imaging were finally included. The primary outcome was the composite of aneurysm growth, morphological change, de novo formation, or rupture. A multivariable logistic regression analysis was used to pick up variants for prediction model. Internal validation was performed in a validation cohort of 638 patients and also via bootstrapping (n = 2000).

Results

Multiple aneurysms (OR 5.56, p < 0.001), hypertension (OR 3.60, p = 0.063), and family history of ADPKD (OR 2.24, p = 0.131) were chosen as the scoring variables, resulting in the 4-point scoring model. This model demonstrated good prediction both in derivation and validation cohort discrimination (apparent AUC = 0.724; optimism-corrected AUC = 0.726). The Kaplan-Meier curves divided by the risk groups showed clear stratification, and the 5-year incidence of the composite outcome were 0% (low risk), 12% (moderate risk), and 46% (high risk) in the validation cohort.

Conclusions

The authors developed an internally validated, easily applicable risk score for predicting aneurysm-related events in ADPKD patients. Generalization requires caution because of potential selection bias, and external validation is warranted.