Abstract WP14: Identifying Potential Sliding Dichotomy Cutpoints for Thrombectomy Trials Using Baseline Age, Deficit Severity, and Core Lesion Volume
Introduction: In sliding dichotomy (SD) stroke trial analysis, each patient’s win criterion is tailored to their prognosis. SD is generally more powerful in detecting treatment effects than crude dichotomy (CD), in which the same win criterion is used for all patients. Potential SD cutpoints for endovascular therapy (ET) trials have not been well-delineated.
Hypothesis: In ET patients, we tested whether the most informative cutpoints on the modified Rankin Scale (mRS) of global disability vary with age, deficit severity, and pretreatment core lesion volume.
Methods: Among DEFUSE 2 trial patients, we compared 3 month outcomes between reperfusers (TICI 2b/3) and non-reperfusers (0-2a). Treatment effect profile maps delineated how prognostic subgroups influenced 5 dichotomizations of the 90D mRS.
Results: Among 105 patients, mean age was 65.4 (±16.1), NIHSS 15.2 (±5.3), and median pretreatment DWI core lesion volume 15.4 ml (IQR 7-32.8). Treatment effect profile maps (Figure) showed better prognosis features tended slightly to increase the absolute difference in outcomes assessed by less disabled mRS cutpoints, including for younger vs older age, milder vs more severe NIHSS, and intermediate vs large core volume. However, crude dichotomization at mRS 0-3 performed well across all subgroups, indicating strong clustering of reperfusion effect across the population. Very small DWI core lesion patients showed a paradoxical effect, with greater treatment benefit at the more disabled mRS levels; high rates of good outcome among both recanalizers and nonrecanalizers made better mRS cutpoints less informative.
Conclusions: Thrombectomy trials tend to enroll homogenously severe patients in whom reperfusion benefits cluster, especially at the mRS 4 to 3 transition; so sliding dichotomies based on single variables do not substantially increase trial power. Further analysis will examine if combinations of prognostic variables confers additional sliding dichotomy power.
Author Disclosures: J. Tarpley: None. G. Albers: Ownership Interest; Significant; iSchemaView. Consultant/Advisory Board; Significant; iSchemaView. M. Mlynash: None. M. Lansberg: None. J. Gornbein: None. J. Saver: Consultant/Advisory Board; Modest; Medtronic/Covidien, Styker, Neuravi, Brainsgate, Pfizer, Squibb, Boehringer Ingelheim (prevention only), ZZ Biotech, St. Jude Medical.
- © 2016 by American Heart Association, Inc.