{
  "abstract": "Introduction/Purpose Despite increasing adoption of advanced predictive modeling tools, clinicians continue to face challenges in forecasting longterm neurological outcomes following large vessel occlusion (LVO) stroke treated with mechanical thrombectomy. The highly variable trajectory of neurological recovery further complicates prognosis, as a meaningful subset of patients exhibit delayed yet substantial improvement despite initially poor postprocedural status. A clearer understanding of factors associated with early versus delayed neurological improvement may enhance clinical decisionmaking, guide goalsofcare discussions, and better inform patient and family expectations.Materials and Methods This retrospective study analyzed the mechanical thrombectomy arm of the prospective, multicenter SELECT cohort study database with an aim of predicting which patients would achieve functional independence by 90 days despite a discharge mRS of 3-5. Early neurological improvement (ENI) was defined as discharge mRS 0-2, while delayed neurological improvement (DNI) was defined as 90day mRS 0-2 among patients discharged with mRS 3-5. Candidate predictors were categorized into preprocedure, day1, and day7/discharge variables. Backward stepwise regression and Lasso modeling (with crossvalidation and adaptive methods) were employed to identify significant predictors.Results A total of 149 patients met inclusion criteria with complete data across all time points. Baseline variables included age, admission NIHSS, serum glucose, occlusion location, ischemic core volume, penumbral volume, and ASPECTS. Day1 variables included number of passes, anesthesia use, achievement of TICI ≥2b, and day1 NIHSS. Day7/discharge variables included followup diffusionweighted imaging infarct volume, midline shift, and day7 (or discharge) NIHSS. Baselineonly models demonstrated poor predictive performance; however, incorporating day1 and day7/discharge data significantly improved model calibration. ASPECTS, day1 NIHSS, and day7/discharge NIHSS emerged as the strongest predictors of both ENI and DNI. Midline shift ≥5 mm at 24 hours was associated with poor 90day outcomes.Conclusion The mechanisms underlying DNI remain poorly understood, yet this phenomenon affects 22.5-30% of nonENI patients undergoing thrombectomy. Consequently, predictive models for DNI are less precise than those for ENI, with clinical neurological status exerting a greater influence than demographic or imaging factors. This may be due to the complex interplay of several mechanisms such as delayed microcirculatory recovery, endothelial and interstitial edema with capillary collapse, ultrastructural microvascular damage, the impact of blood elements, and other mechanisms seen in no-reflow phenomenon. Continued investigation may clarify how clinical, imaging, and even biomarker predictors can better guide prognostication and clinical communication for ischemic stroke patients treated with mechanical thrombectomy.Disclosures K. Duncan: None.",
  "authors": [
    {
      "affiliations": [
        "Case Western Reserve Univ/ Univ Hospitals Cleveland Medical Center, Cleveland, OH"
      ],
      "name": "K Duncan"
    },
    {
      "affiliations": [
        "Neurology, McGovern Medical School, Houston, TX"
      ],
      "name": "D Pujara"
    },
    {
      "affiliations": [
        "University of Texas at Houston, Houston, TX"
      ],
      "name": "A Sarraj"
    }
  ],
  "title": "E-285 Unraveling delayed neurological improvement after thrombectomy: predictive factors and physiologic considerations",
  "uid": "3e91973e-2b71-5b51-8132-eec472a9a7e4"
}