Elucidating a Proteomic Signature for the Detection of Intracerebral Aneurysms. (1st September 2019)

• Record Type: Journal Article
• Authors: Nistal, Dominic A; Martini, Michael L; Hardigan, Trevor; Fernandez, Nicolas; Kim-Schulze, Seunghee; Song, Rui; Spica, Natalia Romano; Kleitsch, Julianne; Mocco, J D; Kellner, Christopher P
• Is Part Of: Neurosurgery. Volume 66(2010)Supplement 1
• Publication Date: 2019-09-01
• Subjects: Nervous system -- Surgery -- Periodicals; 617.48005
• DOI: 10.1093/neuros/nyz310_170
• Languages: English
• ISSNs: 0148-396X

Abstract

Abstract: INTRODUCTION: Intracranial aneurysms (IA) occur in approximately 2% of the general population and are the leading cause for spontaneous subarachnoid hemorrhage (SAH). Recent studies have shown that inflammatory and cell adhesion molecules are associated with the formation and progression of aneurysmal growth. In this study, we utilized proteomic data from patients with known intracranial aneurysms and age, sex, and comorbidity matched controls to identify a proteomic signature that is highly consistent with the presence of an intracranial aneurysm. METHODS: A total of 56 patients were prospectively enrolled in this study, 28 of which had unruptured intracranial aneurysms. Protein was isolated from peripheral blood samples and sent for Proseek multiplex immunoassay processing. Of the 92 analytes in the selected Olink Inflammatory Panel, 70 had variable expression across our cohort and were included for analyses. Multivariate regression models were constructed to predict the presence of an aneurysm. Support Vector Machine (SVM) learning and naïve Bayes algorithms were developed with an 80/20: training/test data separation to determine the precision and reliability of the proteomic signature. RESULTS: Of the 28 patients, 82.1% (n = 23) were female, with a mean aneurysm size of 8.9 mm. Logistic regression analysis revealed 8 highly sensitive analytes that were predictive of the presence of an aneurysm at a threshold of P < .0001. The support vector machine learning and naïve Bayes classification algorithms performed well with a positive predictive value of 100% and 85.7% and a sensitivity of 100% and 100%, respectively (Brier score = 0.032, 0.083). CONCLUSION: This study leveraged individualized data from patients with IA to identify specific protein analytes that predict the presence of an aneurysm. To our knowledge this is the first proteomic signature developed for the purpose of identifying IA's prior to rupture. Future research with increased sample size will allow for the validation and strengthening of this predictive signature.