Introduction Current ASPECTS scoring used for mechanical thrombectomy patient selection assumes each neuroanatomic region as having an equal contribution to clinical outcomes. However, each ASPECTS-based region is known to have a variable effect on determining clinical outcomes. This study aims to identify which ASPECTS-based neuroanatomic region best predicts clinical outcomes.
Methods A retrospective analysis was performed on consecutive mechanical thrombectomies at an urban, tertiary care academic comprehensive stroke center. Baseline imaging and 24 hour post-thrombectomy imaging were reviewed by two ASPECTS-trained physicians for affected ASPECTS-based regions. Clinical data, thrombectomy performance and modified Rankin scores at 90 days were also collected. Multivariate logistic regression analysis was performed to determine whether certain APSECTS regions could predict clinical outcomes.
Results Between June 2013 through September 2016, 109 patients underwent stroke thrombectomy with a mean age of 63 years. Overall, 42% were female, mean NIHSS was 19 (SD +/-5), and TICI 2b/3 rate was 85%. Mean 24 hour ASPECTS was 5.5 (SD 2.2). Forty percent had mRS ≤2 at 90 days. Among the cohort with mRS ≤2, the most common ASPECTS-hypodensity regions were the insula (68%), M5 (63%), lentiform nuclei (58%), and caudate (55%). Among those with mRS >2, the most common ASPECTS-hypodensity regions were the insula (55%), M5 (54%), M2 (46%), and lentiform nuclei (43%). In logistic regression analysis, the ASPECTS regions most significantly associated with poor outcomes was M4 (OR=5.6, 95% CI 1.3–23.9, p=0.02) and M6 (OR=5.7, 95% CI 1.2–27.5, p=0.03).
Conclusions In the entire cohort, the insula, M5, and lentiform nuclei were the most commonly affected ASPECTS-based regions but were not significantly associated with poor clinical outcomes. The M4 and M6 ASPECTS–based regions were most predictive of decreased long-term functional independence. These results may assist the future use of weighted ASPECTS-based region scoring to improve the utility of this tool for stroke thrombectomy patient selection.
Disclosures K. Fukuda: None. K. Keppetipola: None. M. Davison: None. M. Chen: None.