Abstract WP189: Evaluating Accuracy in Defining Large Core Infarction: A Comparison of Non-Contrast Head CT ASPECTS and CT Perfusion Core Infarct Volume

• DOI: 10.1161/str.56.suppl_1.wp189
• Publication Date: 2025-01-30T10:35:27Z
• Authors: Ngoc Mai Le; Anjan Nagesh Ball...; Hussain Azeem; Ananya Iyyangar; Ritesh Bajaj; Eunyoung Lee; Emmanuel Ebirim; Sunil Sheth

Background: Imaging definitions of large core infarction are highly variable. Here, we assess the performance characteristics of non-contrast head CT versus CT perfusion at predicting final infarct volume (FIV) in patients with large core acute ischemic stroke (AIS) undergoing endovascular therapy (EVT).

Methods: From our prospectively collected multi-center registry across four comprehensive stroke centers (CSC) in the Greater Houston area, we identified patients from 2017 to 2023 with LVO AIS who received EVT. FIV was defined using MRI DWI at 48 hours. Large infarct core was defined by CTP as exceeding 70 ml or by CT ASPECTS < 6. Kappa statistics assessed agreement between NCHCT ASPECTS, and CTP-RAPID predicted core values (rCBF < 30%). Receiver operating characteristic (ROC) analyses were used to compare the ability of NCHCT ASPECTS and CTP Core to identify large core infarct populations against MRI FIV.

Results: Among 264 patients with LVO AIS who underwent thrombolysis and met the inclusion criteria, median NIHSS was 16 [IQR, 11-20], MRI FIV was 17.67 [IQR, 6.53-56.45], ASPECTS was 8 [IQR, 6-9], CTP-RAPID predicted core was 6.5 [IQR, 0-29]. 15.2% of patients had an MRI FIV greater than 100 ml. CTP Core measurements and NCHCT ASPECTS agreed 87.5% of the time, with a kappa statistic of 0.3346 (p-value < 0.0001). In 33 cases with discordant results between CTP Core and NCHCT ASPECTS, MRI FIV identified 21 patients as having a large core, aligning with CTP Core but not NCHCT ASPECTS. Conversely, in 12 patients, MRI FIV agreed with NCHCT ASPECTS but not with CTP Core. Both NCHCT ASPECTS and CTP Core showed moderate accuracy in identifying large core infarcts, with similar AUC values of 0.63. Additionally, both methods demonstrated similar discriminatory power (AUC = 0.62) in identifying patients with a modified Rankin Scale (mRS) score of 0-3 after adjusting for age and NIHSS. Sensitivity analysis confirmed similar discriminatory power between NCHCT ASPECTS and CTP Core in identifying large core infarcts (ROC = 0.62) and mRS 0-3 (ROC = 0.59) after adjusting for age and NIHSS, with an 86.58% agreement rate (p-value < 0.001).

Conclusions: Both NCHCT ASPECTS and CTP core volumes are similarly effective in distinguishing large core infarcts and predicting good outcomes suggesting that either NCHCT or CTP imaging techniques are equally useful for identifying patients with large core infarcts in clinical practice.