AUTHOR=Wang Yajing , Mao Qianqian , Jiang Liang , Peng Mingyang , Chen Yu-Chen , Zhang Hong , Wang Liwei , Yin Xindao TITLE=Large vessel occlusion mediated fluid attenuated inversion recovery signal intensity ratio is associated with stroke within 4.5 h JOURNAL=Frontiers in Neurology VOLUME=Volume 15 - 2024 YEAR=2024 URL=https://www.frontiersin.org/journals/neurology/articles/10.3389/fneur.2024.1445017 DOI=10.3389/fneur.2024.1445017 ISSN=1664-2295 ABSTRACT=The primary objective was to investigate the value of the FLAIR signal intensity ratio (SIR) in identifying stroke within 4.5 hours. The secondary objective was to ascertain whether large vessel occlusion (LVO) mediated the relationship between the SIR and stroke within 4.5 hours.We analyzed 633 acute stroke patients within 24 hours of clear symptom onset. The SIR and DWI-FLAIR mismatch were evaluated. First, we determined whether demographic variables, vascular risk factors and LVO were related to stroke within 4.5 hours with multivariate logistic regression analyses and stratified regression analysis. Next, we used mediation analysis to determine whether LVO explained the association between SIR and stroke within 4.5 hours. Finally, we used receiver operating characteristic (ROC) analysis to assess the value of SIR, independent variable, and multiparameter models in identifying stroke within 4.5 hours and compared with DWI-FLAIR mismatch.Results: Hyperlipemia, LVO and SIR were associated with stroke within 4.5 hours.Mediation analysis revealed that LVO partially mediated the relationship between SIR and stroke within 4.5 hours (P < 0.001). The multiparameter model (hyperlipemia, LVO and SIR) showed significantly improved performance (AUC 0.869) in identifying stroke within 4.5 hours over DWI-FLAIR mismatch (0.684), hyperlipemia (0.632), LVO (0.667) and SIR (0.773) models.SIR is associated with stroke within 4.5 hours, and LVO partially mediates this relationship. A multiparameter model combining hyperlipemia, LVO and SIR can more accurately identify stroke within 4.5 hours than individual parameter models.