In this prospective observational registry study involving Japanese patients with recent stroke or TIA, the TyG index was significantly associated with prevalence of intra- and extracranial artery atherosclerosis and was an independent predictor of future MACE. The prognostic value of the TyG index differed between the etiologic subtypes of stroke; a significant correlation between the TyG index and MACE risk was observed in atherothrombotic stroke but not in lacunar or cardioembolic stroke. Our results suggest the usefulness of TyG index as a simple risk predictor, allowing for more targeted secondary prevention.
It is plausible that IR contributes to the development of vascular events through multiple mechanisms. The biological consequences of IR include hypertension, dyslipidemia, hyperglycemia, hyperinsulinemia, systemic inflammation, and vascular endothelial dysfunction, all of which can accelerate atherosclerosis [
1]. IR also promotes platelet adhesion, activation, and aggregation, [
22] leading to the occurrence of thrombotic events. In addition, IR can cause impairments of cerebral vascular reactivity and hemodynamic disturbances that may facilitate the onset of stroke [
23]. These mechanisms are able to explain our results, although the causal relationship between TyG index and vascular events needs to be confirmed in further studies.
The TyG index can be easily and inexpensively obtained in clinical practice using routine blood tests. A previous study showed that the TyG index is a highly sensitive (96.5%) and specific (85.0%) method to detect IR, compared to the euglycemic hyperinsulinemic technique [
3]. In addition, the TyG index has been shown to be superior to the HOMA-IR in assessing IR in individuals with or without diabetes [
2,
4]. After the TyG index was first proposed in 2008, [
2] a number of clinical studies found significant associations of the TyG index with arterial stiffness [
24,
25] and the presence of atherosclerotic diseases [
5‐
7]. Furthermore, epidemiological studies have demonstrated that the TyG index is predictive of future stroke in the general population [
8‐
10]. According to a meta-analysis including more than 5 million participants, the TyG index was independently associated with 1.4-fold and 1.3-fold increased risk of coronary artery disease and stroke, respectively [
8]. Another cohort study found a linear relationship between the TyG index and the risk of ischemic stroke, but not hemorrhagic stroke [
9]. In prior studies targeting acute stroke patients, the TyG index was associated with early recurrent ischemic lesions on brain imaging, [
14] early neurological worsening, [
15] and recurrent stroke, [
15] which is in agreement with our results. The predictive performance of TyG index observed in our study (sensitivity, 67.5%; specificity, 49.6%; AUC, 0.579) seems to be similar to that in previous studies reporting the sensitivity of 46.0–59.2%, specificity of 63.2–68.5%, and AUC of 0.560–0.633 to predict MACE in the general population [
6,
26] or coronary artery disease patients [
27]. We assumed that the prognostic impact of the TyG index may differ between etiologic subtypes of stroke, given the adverse effects of IR on the atherosclerotic process. As expected, the TyG index was strongly associated with the presence of ICAS and ECAS and with the risk of recurrent vascular events after atherothrombotic stroke or TIA.
Limitations
This study has several limitations. First, because the study participants were from a single-center cohort of Japanese patients, our results may not be generalizable to other populations. The annual MACE rate of 12.5% was higher than that reported in previous clinical trials [
28,
29]. This may be due to our consecutive enrollment of patients, regardless of their age, general condition, or comorbidities, whereas clinical trials usually select patients whose general condition is fair. In addition, patients were included within one week of onset, ensuring that early recurrent events were captured. Second, the TyG index measured during the acute phase of stroke might not precisely reflect the patient’s insulin sensitivity, given that acute stroke could lead to stress hyperglycemia. Moreover, we did not assess temporal changes in the index during the follow-up period. Some prior studies have measured the TyG index several times between specific intervals and demonstrated that an index of cumulative exposure to TyG is superior to a single measurement in risk prediction [
30,
31]. Hence, the use of the TyG index only at baseline may be less robust. Third, we could not calculate the HOMA-IR because our database lacked data on fasting insulin levels. It would be interesting to compare the predictive ability of the TyG index with that of the HOMA-IR. Forth, among the 105 MACEs observed at one year, 37 (35.2%) occurred during hospitalization. Medications at discharge shown in Additional file
3: Table S2 do not necessarily reflect the treatments when the patients had recurrent events. Finally, although the TyG index was an independent predictor, its sensitivity (67.5%) and specificity (49.6%) are still poor, suggesting that it is difficult to predict MACE based on the TyG index alone. The predictive performance of the TyG index needs to be validated in larger cohort studies.