Abstract TP160: Minimum Intracerebral Hemorrhage Volume to Maximize Diagnostic Accuracy of Simplified Edinburgh Criteria

Abstract

Introduction: Simplified Edinburgh criteria, based on plain CT, facilitates etiology diagnosis of ICH on cerebral amyloid angiopathy (CAA). Excellent performance in autopsy cohort though, the criteria show disparate discrimination ability amongst survivors with smaller hematoma size in external validation. On assumption that diagnostic accuracy may depend on ICH volume, cutoff volume could likely be found to better the application of Edinburgh criteria. Methods: Primary lobar ICH patients with CT and one of the following: pathology OR qualified MRI available for probable CAA diagnosis per Boston 2.0 criteria were retrospectively enrolled. Clinical characteristics, ICH volume, histology and markers of CT and MRI were collected for validation test. Additionally, likelihood-ratio curves for rule-in and rule-out criteria were plotted, taking 3.0 and 0.3 as the threshold of positive likelihood ratio (PLR) and negative likelihood ratio (NLR) respectively. Results: The criteria were tested among 52 lobar ICH patients with histological proof and 120 with MRI data. Patients with pathology by hematoma evacuation were characterized by larger ICH volume (31.45(31.45), 12.61(16.07), P<0.001) and younger age (67±7, 71±9, P=0.002), compared to those tested on MRI. Discrimination of pathology-evidenced CAA was good (AUC 0.74) for both rule-out (NLR 0.26) and rule-in (PLR 2.97) criteria, while less satisfying when validated against probable CAA on MRI (AUC 0.58, NLR 0.61, PLR 1.42), evincing volume-dependent performance of Edinburgh criteria. According to likelihood-ratio curves (Figure 1), cutoff volume of 30ml may maximize positive likelihood ratio for rule-in criteria and optimize rule-out criteria with NLR≤0.3. Conclusion: Simplified Edinburg criteria was initially validated against pathology- or MRI-attested CAA in Asian population. Dependent on ICH volume, the diagnostic accuracy may be considerably augmented when cutoff value of 30ml was applied.