This is a hospital-based observational study on patients with ischemic stroke who were prospectively registered to a stroke registry from January 2012 to December 2018. The registry enrolled consecutive patients with acute ischemic stroke within 7 days of onset. During admission, all patients were tested for TBI with ABI measurements and evaluated with brain MRI and/or CT, as well as cerebral angiography (magnetic resonance angiography, CT angiography, or digital subtraction angiography). Systemic evaluation included 12-lead electrocardiography, chest radiography, standard blood tests, and lipid profile during admission in the stroke unit. To determine the cardioembolic source and aortic atherosclerosis, all study patients were thoroughly evaluated via transesophageal echocardiography (TEE) and continuous electrocardiography monitoring. TEE was conducted within 2 weeks of the initial stroke. TEE was a part of the standard evaluation for all patients, except in those with decreased consciousness, impending brain herniation, poor systemic conditions, tracheal intubations, or failure of introducing an esophageal transducer. A patient was also excluded from undergoing TEE if informed consent for TEE was not obtained from either the patient or family members (13, 14). Patients were treated using standard treatment protocols based on the guidelines for acute ischemic stroke (15, 16). Stroke classification was determined during weekly conferences. Based on the consensus of three stroke neurologists, stroke subtypes were classified according to the Trial of ORG 10,172 in Acute Stroke Treatment classification (17). The Institutional Review Board of Severance Hospital, Yonsei University Health System, approved this study and waived the need for informed consent because of the retrospective design and observational nature of this study (IRB: 4-2021-0023).

We collected data on baseline characteristics [age, sex, and National Institutes of Health Stroke Scale (NIHSS) score for neurological deficit] upon admission, presence of risk factors, and laboratory data (glucose, high-density lipoprotein, and low-density lipoprotein). Hypertension was defined as systolic blood pressure ≥140 mmHg or diastolic blood pressure ≥90 mmHg after repeated measurements during hospitalization or currently taking antihypertensive medication. Diabetes was defined as fasting plasma glucose levels ≥126 mg/dL or taking an oral hypoglycemic agent or insulin. Hypercholesterolemia was defined as serum total cholesterol ≥240 mg/dL, low-density lipoprotein ≥160 mg/dL, or any history of use of lipid-lowering agents after a diagnosis of hypercholesterolemia. Current smoking was defined as having smoked a cigarette within 1 year before admission. Coronary artery disease was diagnosed when patients had a previous history (acute myocardial infarction, unstable angina, coronary artery bypass graft, or percutaneous coronary artery stent/angioplasty) or the presence of significant stenosis (≥50%) in any of the three main coronary arteries.

Aortic plaques were identified via TEE and classified as complex or simple aortic plaques. Complex aortic plaques consisted of plaques protruding into the lumen by ≥4 mm and mobile lesions located in the proximal aorta. Simple aortic plaques were plaques <4 mm in the proximal aorta or plaques of any size located in the descending aorta (13, 14). The total aortic plaque score was defined as the sum of the number of complex and simple aortic plaques.

The TBI, ABI, and brachial–ankle pulse wave velocity (baPWV) were measured in the supine position using an automatic device (VP-1,000; Colin Co., Ltd., Komaki, Ja-pan), which has been validated previously (18). The TBI was calculated as the ratio of the systolic blood pressure of the great toe divided by the higher systolic blood pressure of the arms. The lower TBI values of both sides were used for the analysis. The cutoff value of TBI was set to < 0.6 based on a previous study (9), showing a higher kappa value than TBI < 0.7 for ABI < 0.9 (k = 0.369 vs. 0.187, all p < 0.001). The ABI was calculated as the ratio of the ankle systolic blood pressure divided by the higher systolic blood pressure of the arms. The lower ABI values of both sides were used for the analysis. Patients with low ABI were defined as having ABI < 0.9 on either side (19). Lower limb calcification is usually defined as ABI > 1.4; thus, patients with high ABI were defined as having ABI > 1.4 on either side (2, 19). Therefore, normal ABI was defined as 0.9 ≤ ABI ≤ 1.4. The baPWV on each side was automatically calculated as the transmission distance divided by the transmission time and expressed in centimeters per-second. Transmission distance from the arm to each ankle was automatically calculated according to the patient's height. Transmission time was defined as the time interval between the initial increase of brachial and tibial waveforms. The average of both baPWV was used for the analysis. The average of limb systolic and diastolic blood pressure was also used for the analysis. TBI and ABI measurements were performed by three investigators. Intra-rater reliability for TBI and ABI parameters was good to excellent as follows: right TBI [intraclass correlation coefficient (ICC) = 0.939, p < 0.001], left TBI (ICC = 0.978, p < 0.001), right ABI (ICC = 0.983, p < 0.001), and left ABI (ICC = 0.987, p < 0.001). Inter-rater reliability for TBI and ABI parameters was also good to excellent as follows: right TBI (ICC = 0.961, p < 0.001), left TBI (ICC = 0.919, p < 0.001), right ABI (ICC = 0.945, p < 0.001), and left ABI (ICC = 0.926, p < 0.001). TBI and ABI measurements were usually performed at one time. We repeated tests in patients with ABI < 1.0, interarm or interankle blood pressure difference ≥ 10 mmHg, or TBI ≤ 0.6 for reliability.

Patients were followed up in the outpatient clinic or by structured telephone interview at 3 months and yearly after discharge. Short-term functional outcomes at 3 months were determined via a structured interview using the modified Rankin Scale (mRS). Poor functional outcome was defined as mRS ≥ 3. Stroke recurrence was defined as newly developed neurologic symptoms with relevant lesions on brain CT and/or MRI 7 days after an index stroke or hospital discharge. MACE was defined as stroke recurrence, myocardial infarction occurrence, or death. The censoring date was December 31, 2019.

Software packages such as SPSS (version 25, SPSS, Chicago, IL, United States) and R package (version 4.0.5, http://www.R-project.org) were used for the statistical analysis. Intergroup statistical analyses were conducted to compare the demographic characteristics and risk factors in all study patients. The statistical significance of intergroup differences was assessed using the chi-squared test for categorical variables and independent two-sample t-test or Mann–Whitney U-test for continuous variables. Data are expressed as means ± SD or median (interquartile range) for continuous variables and number (%) for categorical variables. The chi-squared test for trend was conducted to investigate the association between TBI < 0.6 and the number of risk factors. The correlations between TBI and vascular markers were determined using Pearson's correlation analysis. Multivariable logistic regression analysis was conducted after adjusting for age, sex, initial stroke severity, and variables with p < 0.05 in univariate analysis to investigate the association between TBI and short-term functional outcome. Survival curves were generated according to the Kaplan–Meier method and compared using the log-rank test. Multivariable Cox proportional hazard regression was conducted to investigate the independent association between TBI and long-term outcomes. Harrell's C-index was used to assess whether low TBI improves predictive models of long-term outcomes. All p-values were two-tailed, and differences were considered significant at p < 0.05.

A total of 2,449 consecutive ischemic stroke patients were enrolled during the study period. After the exclusion of 752 patients without TBI measurements, 1,697 patients were included for further analysis. Table 1 summarizes the baseline characteristics of the study population. The mean age of all study patients was 64.6 ± 13.5 years, and 1,033 (60.9%) patients were men. The median NIHSS score at admission was 2.0 (interquartile range, 1.0–5.0) (Table 1).

In all study patients, 18.3% patients showed low TBI (TBI <0.6). Among risk factors, TBI < 0.6 was associated with hypertension and diabetes (all p < 0.05). Atrial fibrillation was more frequently found in patients with TBI < 0.6 (25.7%) compared with TBI ≥ 0.6 patients (16.3%) (p < 0.001) (Supplementary Table 1). The presence of TBI < 0.6 was increased as the number of risk factors increased (p < 0.001 for trend) (Figure 1). For central and peripheral atherosclerotic markers, a continuous variable of TBI was negatively correlated with complex (r = −0.220), simple (r = −0.184), and total aortic plaque scores (r = −0.210) (all p < 0.001). TBI had a negative correlation with baPWV (r = −0.115) and a positive correlation with ABI (r = 0.530) (all p < 0.001) (Supplementary Table 2).

Poor functional outcome at 3 months was observed in 395 (23.3%) patients. In univariable analysis, the poor functional outcome at 3 months was associated with old age, female sex, initial stroke severity, hypertension, diabetes, non-smoker, high glucose level, aortic plaque scores (complex, simple, and total), heart rate, baPWV, ABI > 1.4, 0.9 ≤ ABI ≤ 1.4, and lower TBI (continuous TBI value and TBI < 0.6 cutoff) (all p < 0.05) (Table 1). Multivariable logistic regression analysis showed that a continuous variable of TBI was independently associated with poor functional outcome at 3 months [odds ratio (OR) 0.294, 95% confidence interval (CI) 0.114–0.759; p = 0.011] in all patients (n = 1,697). A continuous variable of TBI was also associated with poor functional outcome (OR 0.293, 95% CI 0.095–0.906; p = 0.033) in patients with normal ABI (n = 1,534) (Table 2). However, the TBI < 0.6 cutoff was not significant in all study patients and normal patients with ABI.

The study population was followed up for a median of 39.7 (interquartile range, 25.7–54.6) months. A total of 305 patients suffered MACE (18.0%), including 171 stroke recurrences (10.1%), 78 myocardial infarction occurrences (4.6%), and 98 all-cause deaths (5.8%) during the study period. The Kaplan–Meier survival curves showed that the TBI < 0.6 cutoff was significantly associated with high stroke recurrence, all-cause mortality, and MACE in all study patients (log-rank test; all p < 0.05). In patients with normal ABI, TBI < 0.6 cutoff was also associated with high stroke recurrence, all-cause mortality, and MACE (log-rank test; all p < 0.05) (Figure 2).

To adjust for covariates (age, sex, initial stroke severity, hypertension, diabetes, smoking, atrial fibrillation, glucose, total aortic plaque score, heart rate, and baPWV), multivariable Cox proportional hazards regression analysis was conducted (Table 3). In all study patients, a continuous variable of TBI was independently associated with stroke recurrence [hazard ratio (HR) 0.270, 95% CI 0.090–0.810; p = 0.019], all-cause mortality (HR 0.173, 95% CI 0.045–0.671; p = 0.011), and MACE (HR 0.254, 95% CI 0.110–0.584; p = 0.001). The TBI < 0.6 cutoff was also independently associated with stroke recurrence (HR 1.651, 95% CI 1.135–2.400; p = 0.009), all-cause mortality (HR 2.105, 95% CI 1.343–3.298; p = 0.001), and MACE (HR 1.838, 95% CI 1.396–2.419; p < 0.001). In patients with normal ABI, a continuous variable of TBI was independently associated with all-cause mortality (HR 0.103, 95% CI 0.017–0.618; p = 0.013). The TBI < 0.6 cutoff was also independently associated with stroke recurrence (HR 1.681, 95% CI 1.080–2.618; p = 0.022), all-cause mortality (HR 2.075, 95% CI 1.180–3.651; p = 0.011), and MACE (HR 1.619, 95% CI 1.149–2.281; p = 0.006).

In sensitivity analysis, TBI < 0.6 had significant interactions with hypertension (p = 0.040) in stroke recurrence. TBI < 0.6 had significant interactions with hypertension (p = 0.006) and ABI status (p = 0.049) in MACE. The TBI < 0.6 was associated with stroke recurrence in normotensive patients. The TBI < 0.6 was also associated with MACE regardless of the presence of hypertension. Particularly, TBI < 0.6 was significantly associated with stroke recurrence, all-cause mortality, and MACE in patients with normal ABI (Figure 3 and Supplementary Table 3).

The Harrell's C-index was calculated to compare risk models in survival analysis (Table 4). The baseline model (Model 1) consisted of age, sex, initial stroke severity, hypertension, diabetes, smoking, atrial fibrillation, glucose, total aortic plaque score, heart rate, and baPWV. In all study patients, the C-index for the baseline model was 0.631 (95% CI 0.590–0.672) for stroke recurrence, 0.721 (95% CI 0.674–0.768) for all-cause mortality, and 0.619 (95% CI 0.588–0.650) for MACE. Adding the TBI < 0.6 cutoff (Model 2) improved the prognostic utility of the baseline model in stroke recurrence (C-index 0.647, 95% CI 0.606–0.688), all-cause mortality (C-index 0.744, 95% CI 0.699–0.789), and MACE (C-index 0.640, 95% CI 0.609–0.671). Significant differences were found in all-cause mortality (p = 0.037) and MACE (p = 0.009). In patients with normal ABI, the C-index for the baseline model was 0.623 (95% CI 0.578–0.668) for stroke recurrence, 0.723 (95% CI 0.664–0.782) for all-cause mortality, and 0.608 (95% CI 0.573–0.643) for MACE. Similarly, adding the TBI < 0.6 cutoff (Model 2) improved the prognostic utility of the baseline model in stroke recurrence (C-index 0.641, 95% CI 0.596–0.686), all-cause mortality (C-index 0.735, 95% CI 0.676–0.794), and MACE (C-index 0.625, 95% CI 0.590–0.660). However, there was no statistically significant difference.