No abstract

Background: Evidence demonstrated diffusion-weighted imaging (DWI) lesions of different age predicted future stroke in transient ischemic attack (TIA) patients. The aim of this study is to develop a novel score (mABCD3-I) by replacing dual TIA with DWI lesions of different age and externally validate it in a multicenter TIA cohort. Methods: We conducted a prospective cohort study. The predictive outcome was stroke/all-cause death. Cox proportional hazards model was used to evaluate the predictive value of risk factors. The area under the curves (AUC) and net reclassification improvement (NRI) were used to present the discriminative and risk stratification abilities. Results: 2283 patients were included: 1077 in the derivation cohort and 1206 in the validation cohort. We derived mABCD3-I score by replacing dual TIA with DWI lesions of different age based on results of the multivariate analysis (Table 1). In the derivation cohort (Table 2), the AUCs for the mABCD3-I score were 0.78 at 90-day (95% confidence interval [CI]: 0.75-0.80; compared with ABCD3-I score, p=0.002), 0.78 at 6-month (95%CI: 0.75-0.81; p=0.008) and 0.78 at 1-year (95%CI: 0.75-0.81; p=0.018). The mABCD3-I score improved the reclassification at 90-day (p=0.006) and 6-month (p=0.033). In the validation cohort, the mABCD3-I had equal discriminative abilities compared with ABCD3-I score at 90-day (p=0.056) and 6-month (p=0.054), and higher ability at 1-year (p=0.044). Conclusions: Replacing dual TIA with DWI lesions of different age in ABCD3-I score improved risk prediction for stroke/death risk after TIA.