Abstract 013: Infarct density: a novel imaging biomarker of long‐term functional outcome after endovascular thrombectomy

Abstract

Introduction Endovascular thrombectomy (EVT) dramatically improves clinical outcomes, but the reduction in final infarct volume only accounts for 10‐15% of the treatment benefit. There is a clear need for imaging biomarkers that are strongly associated with long‐term functional outcome after EVT, both in clinical practice and clinical trial design (surrogate outcome measure). Here we aimed to quantify the relationship between functional outcome and a range of ADC thresholds on post‐EVT MRI. Methods A single‐center cohort of consecutive acute stroke patients with anterior circulation large vessel occlusion, successful recanalization via EVT (mTICI ≥ 2b), and MRI of the brain between 12 hours and 7 days after EVT was evaluated. Imaging was processed via RAPID software. Final infarct volume was based on the traditional ADC <620 threshold. Logistic regression quantified the association of lesion volumes and good functional outcome (90‐day modified Rankin Scale ≤ 2) at a range of lower ADC thresholds (<570, <520, and <470). Infarct density was calculated as the percentage of the final infarct volume below the ADC threshold with the greatest effect size. Univariate and multivariate logistic regression quantified the association between clinical/imaging variables and functional outcome. A receiver operating characteristics (ROC) analysis was used to calculated areas under the curve (AUC) with 95% confidence intervals for the final model. The Delong test compared this AUC with two additional models: (1) the final multivariate model without infarct density and (2) the final model without infarct density and final infarct volume. Results Of the 120 patients who underwent MRI after successful EVT, lesion volume based on the ADC threshold <470 had the strongest association with good outcome (OR: 0.81 per 10mL; 95% CI: 0.66–0.99). In a multivariate model, infarct density (volume <470/volume <620 * 100) was independently associated with good outcome (aOR 0.68 per 10%; 95% CI: 0.49–0.95). In the multivariate model, final infarct volume was no longer associated with outcome (aOR 0.98 per 10mL; 95% CI: 0.85–1.14). The ROC analysis of the multivariate model with only clinical variables (age, sex, and NIHSS) yielded a good ability to distinguish patients with good and bad outcome (AUC = 0.77; 95% CI: 0.69 – 0.84). Adding infarct volume improved classification performance (AUC = 0.82; 95% CI: 0.75 – 0.88), but this was further improved by adding infarct density (AUC = 0.84, 95% CI: 0.78 – 0.91); p=0.02 comparing all three AUCs (Figure 1). Conclusion The degree of tissue injury after EVT is topographically heterogeneous, and more profound tissue injury manifests as lower ADC values. Here, we operationalized the severity of infarct as infarct density, which is independently associated with long‐term clinical outcome and provides greater prognostic information than final infarct volume. This technique may hold value as a surrogate outcome measure in early phased clinical trials.