Abstract TMP44: The Novel Predictors Of Poor Functional Outcome In Acute Lacunar Infarction By Using Hierarchical Cluster Analysis Of Clinical Variables

• Published in: Stroke Vol. 53; no. Suppl_1; p. ATMP44
• Main Authors: Takahashi, Junichiro, Sakai, Kenichiro, Okumura, Motohiro, Kida, Hiroyuki, Maku, Takahiro, Nakada, Ryoji, Kitagawa, Tomomichi, Sato, Takeo, Takatsu, Hiroki, Komatsu, Teppei, Sakuta, Kenichi, Mitsumura, Hidetaka, Iguchi, Yasuyuki
• Format: Journal Article
• Language: English; Japanese
• Published: Ovid Technologies (Wolters Kluwer Health) 01.02.2022; Lippincott Williams & Wilkins
• ISSN: 0039-2499; 1524-4628
• DOI: 10.1161/str.53.suppl_1.tmp44

BackgroundLacunar infarct (LI) is a heterogeneous disorder of cerebral small vessel. Hierarchical clustering (HC) is useful for exploring potential underlying clinical phenotype in heterogeneous disease. Our aim is to identify the predictive factors of poor functional outcome in LI to classify the subjects into distinct subgroups using HC and compare the clinical variables including clinical outcome among groups. MethodsFrom October 2012 to April 2020, LI patients within 24 hours from the onset were enrolled. We used a HC k-means clustering algorithm on patients’ baseline data. We also compared clinical outcome among each cluster groups. ResultsAmong 187 patients (144 male, median age 66 years old, median NIHSS score2), three clusters were identified. Cluster1(58%), mostly comprised of younger patients (p<0.01), had fewer past history of ischemic stroke (p=0.007). The severity of deep subcortical white matter hyperintensity (DSWMH) and periventricular hyperintensity (PVH) were mild, and small number of cerebral microbleeds (MBs) are detected (p<0.001 for each). Cluster2(15%), mostly comprised of older patients (p<0.001), had the highest NIHSS score on admission (p=0.021). Serum D-dimer and Cr levels were relatively higher (p<0.001, p=0.003, respectively). Cluster3(28%), mostly comprised of men, represented the highest number of MBs (p<0.001) and increase of mean IMT (p =0.003). Poor functional outcome defined as mRS at 3 months > 2-6 was 4% of cluster1, 100% of cluster 2 and 12% of cluster 3 (p=0.001). In multivariate logistic regression analysis, the phenotype of cluster was the strongest predictor of poor functional outcome (OR 16.1, 95% CI 6.04-42.9, p<0.001, Figure). ConclusionThree clusters with different clinical outcome were identified. The phenotype of cluster was the strongest predictive factor of poor functional outcome. Further study to identify whether management of these components provide different clinical outcome was warranted.