Abstract Body: Introduction: Eligibility for clinical trial enrollment is often determined by site investigators with subsequent re-adjudication by an imaging core lab. We sought to assess the differences between site-investigator reported CT ASPECTS with core lab adjudication, and how these differences affected trial eligibility and endovascular thrombectomy (EVT) treatment effect.

Methods: Absolute and relative differences between site and core-lab reported CT ASPECTS were recorded in the SELECT2 randomized trial. The agreement between measures was illustrated using Bland-Altman plots. Cases with extreme discordance were further examined for differences in baseline characteristics, as well as whether trial eligibility and EVT treatment effect were impacted.

Results: Of 352 enrolled, 346 patients had both site and core-lab ASPECTS available for evaluation. Median (IQR) values for site ASPECTS, core lab ASPECTS and the difference between site and core-lab ASPECTS were 4 (3, 5), 4 (3, 5) and 0 (-1, 1). Mean (95% CI) difference between site and core-lab ASPECTS was -0.03 (-0.19, 0.13), with higher limit of agreement at 2.83 and lower limit of agreement at -2.9 – Fig 1. Significant disagreement of ≥3-point difference between site ASPECTS and core-lab ASPECTS was observed in 25 (7%) patients, with 14 having site ASPECTS ≥ 3 points higher and 11 having core-lab ASPECTS ≥ 3 points higher.
EVT treatment effect was maintained based on site ASPECTS reads of 3-5 (aGenOR: 1.48, 95% CI: 1.16 to 1.89, p-value: 0.001) without significant heterogeneity as compared to other ASPECTS strata (p-interaction: 0.69) (Figure 2). In a sensitivity analysis excluding potential ineligible patients based on core lab reads [a) with core lab CT ASPECTS 0-2 and b) with core lab CT ASPECTS 6-10 and ischemic core estimates of <50ml], EVT treatment effect was maintained among patients with site read CT ASPECTS of 3-5 (aGenOR: 1.72, 95% CI: 1.26-2.34, p=0.001).

Conclusions: Site investigator reads and core lab reads for ASPECTS score largely mirrored each other, with very low mean and median differences. Significant disagreement (≥3 points) between site and core-lab reads was limited and occurred in ~7% of cases. EVT benefit and treatment effect were maintained based on eligibility by investigators’ reads without heterogeneity in different ASPECTS strata.