Purpose With thrombectomy for anterior circulation large vessel occlusion (ACLVO) stroke, time to recanalization is considered important, but collateral status may be a more important driver of outcome than time. Here we used data from an institutional cohort to test the hypothesis that small DWI core volume on presenting MRI, a biomarker of robust collateral perfusion, is the dominant predictor of favorable outcome after thrombectomy.

Methods A cohort of 56 ACLVO patients treated with thrombectomy at our institution between 11/1/2012 and 5/15/2015 was studied by logistic regression using final infarct volume ≤50 mL, a validated predictor of good clinical outcome, as the outcome variable. The following univariate predictor variables were assessed: age, NIHSS score, time to CTA, time to CTA ≤ 6 hours, presenting diffusion-restricting core volume, presenting core ≤ 50 mL, time to recanalization, time to recanalization ≤6 hours, time to recanalization ≤8 hours, and recanalization ≥ TICI2B. Finally, multiple logistic regression models were created comparing time and presenting core as predictors adjusting for age and TICI2B recanalization status.

Results Of the univariate predictors studied, only presenting core volume, presenting core volume ≤50 mL, age, and post procedure TICI ≥ 2 B were significant predictors of the outcome variable at the α = 0.15 level (Table 1). Paradoxically, the administration of IV tPA was associated with larger core volumes in our dataset (β = 1.53, p = 0.100). Of the time-based predictors, time to recanalization and time to recanalization ≤8 hours performed best, although none of them approached significance at the α = 0.15 level. The performances of time, core volume, age and TICI ≥ 2 B as predictors of small final infarct were then evaluated in a multiple logistic regression model. In that model, presenting core volume ≤50 mL, not time to recanalization, was the dominant predictor of small final infarct volume (OR 15.2 [95CI 1.3, 180] vs. 0.94 [95CI 0.78, 1.1]). Finally, a 3 variable logistic regression model incorporating presentation core volume, age, and TICI ≥ 2 B as covariates was constructed. The resulting model was statistically significant χ²(2) = 22.65, (p < 0.001) and accounted for 52.9% of the variance in outcome with a percent accuracy of classification of 85.4%.

Conclusion Presentation volume of diffusion-restricted core was a strong predictor of final infarct volume post-thrombectomy in this cohort of ACLVO patients, whereas time-based covariates showed little if any significant predictive value. For every 1 mL increase in presentation infarct volume, the odds of a small completed infarct were reduced by about 6%. A presentation core volume ≤50 mL improved the odds of a small final infarct by at least 26%, probably more.

Disclosures B. Cristiano: None. M. Pond: None. S. Basu: None. U. Oyoyo: None. J. Jacobson: 4; C; GeneLux.