Development and Validation of an Individual Nomogram to Predict 1-Year Cognitive Decline in Patients with Unruptured Intracranial Aneurysm Underwent Iatrogenic Cerebral Infarction After Stent Placement: A Multicenter Study

Abstract

Background: New iatrogenic cerebral infarcts (NICIs) are frequently seen on diffusion-weighted magnetic resonance imaging (DWI) after stent placement for unruptured intracranial aneurysms (UIAs). Early identification of individuals at risk of cognitive decline (CD) allows interventions to preserve cognitive function and maintain independence. Develop and validate a deep learning radiomics nomogram (DLRN) based on posttreatment DWI and clinical features to predict cognitive function at 1-year follow-up in UIA patients with NICIs after stent placement.

Methods: Multicenter observational study recruited 526 UIA patients with NICIs on post-treatment DWI were recruited in four institutions between January 2016 and December 2021. The study comprised a retrospective training cohort (251 participants), retrospective external validation cohort (167 participants), and a prospective validation cohort (108 participants). Posttreatment DWI was used to construct two imaging signatures that reflect the phenotypes of deep learning and handcrafted radiomics features. Radiomic features (1046) and deep learning features were extracted from NICI regions. After incorporating clinical factors, multivariable logistic regression was used to develop the integrated DLRN to predict risk of CD at 1-year. Predictive capability was evaluated in terms of discrimination, calibration, and clinical utility.

Findings: The rates of CD at the 1-year follow up in the training, external validation, and prospective cohorts were 18.7% (47/251), 19.8% (33/167) and 20.4% (22/108) respectively. The DLRN showed satisfactory discriminating ability for predicting 1-year CD and yielded areas under the receiver operating characteristic curve of 0.975, 0.942, and 0.841 in the training, external validation, and prospective cohorts, respectively. Calibration was good in all cohorts. Decision curve analysis confirmed the clinical utility of the DLRN.

Interpretation: Our DLRN exhibited promising performance for predicting CD in UIA patients with NICIs after stent placement, facilitating optimized and individualized treatment strategies.

Note:
Funding Information: This work was supported by Henan Province Medical Science and Technology Research and Joint Construction Project (grant number: LHGJ20220337), National Natural Science Foundation of China (grant numbers: 82201435, 82272092 and 81960330), China Postdoctoral Science Foundation (grant number:2022M712893), Scientific Research Cultivation Plan of Beijing Municipal Hospitals (grant number: PX2022022).

Declaration of Interests: All authors declare no competing interests.

Ethics Approval Statement: This study was approved by the institutional review board of all participating hospitals. The requirement for informed consent was waived because of the study’s retrospective observational design.

Keywords: unruptured intracranial aneurysms; new iatrogenic cerebral infarcts; cognitive decline; radiomics; stent placement