Association between Triglyceride-Glucose Index and Early Neurological Outcomes after Thrombolysis in Patients with Acute Ischemic Stroke

Background: The triglyceride-glucose (TyG) index is a novel biomarker of insulin resistance which might plausibly influence endogenous fibrinolysis and thus early neurological outcomes in patients with acute ischemic stroke (AIS) treated with intravenous thrombolysis using recombinant tissue-plasminogen activator. Methods: We included consecutive AIS patients within 4.5 h of symptom onset undergoing intravenous thrombolysis between January 2015 and June 2022 in this multi-center retrospective observational study. Our primary outcome was early neurological deterioration (END), defined as ≥2 (END2) or ≥ 4 (END4) National Institutes of Health Stroke Scale (NIHSS) score worsening compared to the initial NIHSS score within 24 h of intravenous thrombolysis. Our secondary outcome was early neurological improvement (ENI), defined as a lower NIHSS score at discharge. TyG index was calculated using the log scale of fasting triglyceride (mg/dL) × fasting glucose (mg/dL)/2. We evaluated the association of END and ENI with TyG index using a logistic regression model. Results: A total of 676 patients with AIS were evaluated. The median age was 68 (Interquartile range, IQR (60–76) years old), and 432 (63.9%) were males. A total of 89 (13.2%) patients developed END2, 61 (9.0%) patients developed END4, and 492 (72.7%) experienced ENI. In multivariable logistic regression analysis, after adjustment for confounding factors, TyG index was significantly associated with increased risks of END2 (categorical variable, vs. lowest tertile, medium tertile odds ratio [OR] 1.05, 95% confidence interval, CI 0.54–2.02, highest tertile OR 2.94, 95%CI 1.64–5.27, overall p < 0.001) and END4 (categorical variable, vs. lowest tertile, medium tertile OR 1.21, 95%CI 0.54–2.74, highest tertile OR 3.80, 95%CI 1.85–7.79, overall p < 0.001), and a lower probability of ENI (categorical variable, vs. lowest tertile, medium tertile OR 1.00, 95%CI 0.63–1.58, highest tertile OR 0.59, 95%CI 0.38–0.93, overall p = 0.022). Conclusions: Increasing TyG index was associated with a higher risk of END and a lower probability of ENI in patients with acute ischemic stroke treated with intravenous thrombolysis.


Introduction
Intravenous thrombolysis treatment remains a first-line approach for acute ischemic stroke (AIS) [1,2]. Despite the beneficial effects of intravenous thrombolysis using recombinant tissue-plasminogen activator for AIS patients, about one-third may experience unfavorable early neurological outcomes [3,4]; 13.8% (95% confidence interval, [CI] 10.0% to 17.7%) of patients experienced early neurological deterioration (END) [5], and 20.9% had failure of early neurological improvement (ENI) [6]. Previous studies showed that END was related to unfavorable stroke outcomes, while ENI was associated with favorable prognosis [6][7][8]. Therefore, identifying factors for early neurological outcomes including END and ENI in the AIS population could provide important prognostic information with relevance for stroke management.
Insulin resistance is considered a main pathophysiological mediator of metabolic syndrome [9]. Despite the established importance of insulin resistance in cardiovascular and cerebrovascular disease [10], evidence of the link between insulin resistance and acute ischemic stroke outcomes is scarce [11]. Insulin resistance might be relevant to acute recanalization therapy through its associations with thrombosis and inflammation, with abnormal endogenous fibrinolysis and increased platelet activation. The hyperinsulinemiceuglycemic clamp test is the gold standard for evaluating insulin resistance. However, this labor-intensive and time-consuming procedure is not routinely measured in clinical practice [12]. Recently, the triglyceride-glucose (TyG) index, which is calculated using serum triglyceride and fast blood glucose levels, has been used as a reliable and novel biomarker of insulin resistance [13]. Aggravating data showed that the TyG index is related to arterial stiffness [14], a higher risk of the cardiocerebrovascular diseases and unfavorable outcomes in patients with cardiocerebrovascular disease [15,16]. Previous studies showed that elevated triglyceride and blood glucose levels were related to the incidence of END and adversely associated with ENI in ischemic stroke population [5,17,18]. Moreover, data from the UK Biobank cohort involving 273,368 individuals showed that the TyG index was superior to blood glucose and triglycerides alone in predicting stroke occurrence, suggesting that the TyG index may potentially be a good biomarker of insulin resistance to predict stroke outcomes [19]. We hypothesized that a higher baseline TyG index is associated with an increased risk of END and a lower probability of ENI in AIS patients who received intravenous thrombolysis. Therefore, we investigated the association of TyG index with the risk of END and ENI after intravenous thrombolysis in a retrospective observational study.

Study Design and Participants
We included consecutive adult AIS patients undergoing intravenous thrombolysis within 4.5 h at three certified stroke centers of Fujian Medical University between January 2015 and June 2022 in this multi-center retrospective cohort study. The exclusion criteria were as follows: (1) discharged within 24 h; (2) receiving arterial thrombolytic treatment; (3) interrupted intravenous thrombolysis due to prompt neurological function improvement or serious side effects.

Data Acquisition
Two authors extracted data regarding baseline demographic characteristics, vascular risk factors (smoking, drinking, history of stroke, hypertension, diabetes mellitus, dyslipidemia, atrial fibrillation, and coronary artery disease), pre-stroke medication-use history (antiplatelet, anticoagulation, statin, antihypertensive, and hypoglycemic agents), clinical features (admission stroke onset severity, admission systolic blood pressure and diastolic blood pressure, onset to treatment time, stroke subtypes), and laboratory findings. Stroke severity (presenting deficit severity) was assessed using the National Institutes of Health Stroke Scale (NIHSS) score. Stroke subtypes were classified into atherosclerosis (A), small vessel disease (S), cardioembolic (C), and others (O) [20]. Patients with missing data regarding the component of the TyG index and outcomes were excluded.

Triglyceride Glucose (TyG) Index Evaluation
Blood samples were collected after fasting for 8 to 12 h. Serum triglyceride and glucose levels were assessed with an automatic biochemical analyzer (Cobas c-system, Roche, Switzerland), and expressed in milligrams per deciliter (mg/dL). The TyG index was calculated using the log scale of fasting triglyceride (mg/dL) × fasting glucose (mg/dL)/2, as previously described [21].

Outcomes
Trained clinicians assessed the neurological deficit using the NIHSS score before and after intravenous thrombolysis. Our primary outcome was END. We applied two well-validated (available) definitions: (i) ≥2 NIHSS-point worsening (END 2 ); and (ii) ≥4 NIHSS-point worsening (END 4 ) compared to the initial NIHSS score within 24 h after intravenous thrombolysis [22,23]. Our secondary outcome was ENI, defined as a lower NIHSS score at discharge [6].

Statistical Analysis
Categorical variables were expressed as absolute counts with percentages. Continuous variables were expressed as means (standard deviation, SD) if normally distributed, or medians (interquartile range, IQR) if not normally distributed. TyG index was treated as both a three-level group (lowest as reference) and a continuous variable. A general linear model and chi-squared test were used to calculate the p for trend between variables for continuous and categorical TyG index data, respectively. To summarize the differences in baseline characteristics in patients with and without END or ENI, continuous variables were compared using the Student's t-test, the Mann-Whitney U-test, analysis of variance or the Kruskal-Wallis test, as appropriate, and categorical variables were compared using the chi-squared test or Fisher's exact test. We calculated the absolute risks and absolute risk differences between different TyG index tertile groups for early neurological outcomes. Five conventional multivariate adjusted logistic regression models were applied to assess the association of END or ENI with TyG index. Model 1: by incorporating those with p < 0.1 for END or ENI in the univariable analysis in addition to age and sex. Model 2: by incorporating the identified suitable minimally sufficient adjustment sets using a directed acyclic graph (DAG) [24] to minimize potential bias from intermediate variables when assessing the effect of the TyG index on END or ENI. Model 3: by incorporating those with p < 0.1 in different TyG index tertile groups in addition to age and sex. Model 4: by incorporating baseline NIHSS score and prespecified vascular risk factors (hypertension, atrial fibrillation, and coronary artery disease) that were shown to be related to the odds of END or ENI in addition to age and sex [5,25,26]. Model 5: by incorporating those variables that are associated with both exposure and outcome variables with p < 0.1 in the univariable analysis into the multivariable analysis. Considering the components of the TyG index, the fasting blood glucose, history of diabetes dyslipidemia, and previous hypoglycemic use were not simultaneously introduced along with the TyG index into the multivariable analysis. To test whether the effect of TyG index on END or ENI occurrence varied between age group (<65 vs. ≥65 years), sex (male vs. female), diabetes (yes vs. no), and stroke onset severity (baseline NIHSS <6 vs. ≥6), the interaction terms TyG-by-age, TyG-by-sex, TyG-by-NIHSS, and TyG-by-diabetes history were used as covariates. A priori p-value < 0.05 was considered significant for interactions. We conducted subgroup analyses, including age (<65 vs. ≥65 years), sex (male vs. female), baseline NIHSS (<6 vs. ≥6), and diabetes (yes vs. no). We conducted a sensitivity analysis limited to patients not receiving thrombectomy treatment after intravenous thrombolysis. Previous studies showed that symptomatic intracerebral hemorrhage (sICH) was a predictor for END, and END after thrombolysis treatment may be caused by sICH occurrence [6,27]. We further conducted a sensitivity analysis by defining END separately from sICH. We also conducted a separate analysis, investigating the relationship between continuous TyG index and continuous delta NIHSS (24 h NIHSS score-baseline NIHSS score). All variables with a p-value < 0.05 were considered statistically significant in this study. All statistical analyses were conducted using SPSS version 26.0 (IBM Corp., Armonk, NY, USA).

TyG Index and END
Patients who developed END 2 included 21 (23.6%) in the lowest tertile (range 6.78-8.30), 21 (23.6%) in the medium tertile (range 8.31-8.86), and 47 (52.8%) in the highest tertile (range 8.89-11.64), compared to 205 (34.9%), 206 (35.1%) and 176 (30.0%), respectively, in those without END 2 (overall p < 0.001). The risk of END 2 increases with increasing TyG index (continuous variable [per unit increase]; odds ratios: OR 1.77, 95%, CI 1.30-2.42, p < 0.001). The absolute risks for END 2 in the lowest, medium, and highest tertile groups were 9.3%, 9.3%, and 21.1%, respectively. The absolute risk difference for END 2 (highest vs. lowest) was 11.8% (95% CI, 5.2-18.3%). Similarly, the absolute risks for END 4 increased with increasing TyG index, while decreasing with increasing TyG index for ENI (Table S1). In the multivariable model 1, adjustment for age, sex, atrial fibrillation, coronary artery disease, initial NIHSS score, ASCO subtypes, and previous stroke use did not change the association between TyG index and the END 2  The DAG algorithm identified a minimal set of confounders: age, sex, current smoker, and regular alcohol user ( Figure 2). Adjustment for this set did not alter the association between a higher TyG index and an increased risk of END 2 (Table 3). This association remained in models 3,4, and 5 (Table S2). We performed a post hoc analysis by combining the lowest and medium groups due to the similar probabilities of END 2 in these two groups (both 23.6%). The results show that a TyG index in the highest tertile was associated with a higher risk of END 2 when compared to the lowest and medium groups (adjusted OR 2.87, 95%CI 1.78-4.62, p < 0.001, Table 3). Similar findings regarding the association between TyG index and END 4 risk are shown in Tables 4 and S3. tients with and without END4 ( Table 2).
Patients who developed END2 included 21 (23.6%) in the lowest tertile 8.30), 21 (23.6%) in the medium tertile (range 8.31-8.86), and 47 (52.8%) in the tile (range 8.89-11.64), compared to 205 (34.9%), 206 (35.1%) and 176 (30.0%), r in those without END2 (overall p < 0.001). The risk of END2 increases with inc index (continuous variable [per unit increase]; odds ratios: OR 1.77, 95%, CI 1 0.001). The absolute risks for END2 in the lowest, medium, and highest tertile 9.3%, 9.3%, and 21.1%, respectively. The absolute risk difference for END2 lowest) was 11.8% (95% CI, 5.2%-18.3%). Similarly, the absolute risks for EN with increasing TyG index, while decreasing with increasing TyG index for S1). In the multivariable model 1, adjustment for age, sex, atrial fibrillation, tery disease, initial NIHSS score, ASCO subtypes, and previous stroke use did the association between TyG index and the END2 risk (categorical variable, vs tile, medium tertile OR 1.05, 95% CI 0.54-2.02, highest tertile OR 2.94, 95%C overall p < 0.001; continuous variable [per unit increase], OR 1.91, 95% CI 1 0.001). The DAG algorithm identified a minimal set of confounders: age, smoker, and regular alcohol user (Figure 2). Adjustment for this set did not al ciation between a higher TyG index and an increased risk of END2 (Table 3). ation remained in models 3,4, and 5 (Table S2). We performed a post hoc anal bining the lowest and medium groups due to the similar probabilities of EN two groups (both 23.6%). The results show that a TyG index in the highes associated with a higher risk of END2 when compared to the lowest and med (adjusted OR 2.87, 95%CI 1.78-4.62, p < 0.001, Table 3). Similar findings regar sociation between TyG index and END4 risk are shown in Tables 4 and S3.    (Table S5). However, it was no longer significant when treating TyG index as a categorical variable for the risk of END 4 (Table S5). Another sensitivity analysis by defining END separately from sICH yielded similar results to those derived from the main analysis (Tables S6 and S7). Subgroup analyses showed that there were no significant interactions between TyG index and variables including age (<65 vs. ≥65 years), diabetes (yes vs. no), and stroke onset severity (baseline NIHSS <6 vs. ≥6) for the risk of END 2 and END 4 (all p for interaction > 0.05). There was a significant interaction between TyG index and sex (male vs. female) for the risk of END 2 and END 4 ( Figure S1A,B).

TyG Index and ENI
Differences in baseline characteristics in patients with and without ENI were shown in Table S8 (Table S9), but was lost when limited to patients who did not receive endovascular thrombectomy treatment (Table S10). Subgroup analysis showed that there were no significant interactions between the TyG index and variables including age (<65 vs. ≥65 years), sex (male vs. female), diabetes (yes vs. no), and stroke onset severity (baseline NIHSS <6 vs. ≥6) for the odds of ENI (all p for interaction > 0.05, Figure S1C).

TyG Index and NIHSS Score Change
Regarding the relationship between continuous TyG index and continuous delta NIHSS (24h NIHSS score-baseline NIHSS score), our data showed that the TyG index was associated with delta NIHSS (unadjusted Beta 0.116, p = 0.003, adjusted for the minimal sufficient adjustment set including age, sex, current smoker, and regular alcohol user, adjusted Beta 0.123, p = 0.001).

Discussion
The current study showed that a higher TyG index, a novel biomarker of insulin resistance, is associated with an increased probability of END and decreased odds of ENI in AIS patients who received intravenous thrombolysis; this suggests a role for insulin resistance in the unfavorable early neurological outcome in this population.
The potential role of the TyG index in AIS prognosis has been noted. A multi-center observational study showed that a higher TyG index was associated with 90-day unfavorable functional outcomes in AIS patients who received intravenous thrombolysis [28]. Moreover, TyG index was associated with a higher risk of in-hospital mortality in criticallyill stroke [29], and early stroke recurrence [30]. To our knowledge, whether TyG index is related to unfavorable early neurological functional outcomes is poorly understood. Our data showed that a higher TyG index (particularly being at the top tertile) was associated with a higher probability of developing END and lower odds of achieving ENI at discharge in AIS patients who received intravenous thrombolysis. In line with our finding, a previous study showed that TyG index was associated with END occurrence in patients with single subcortical infarctions [30]. However, that study included AIS patients within 72h of symptom onset applying only one END definition; therefore, the actual frequency of END might be underestimated [30]. Our study adds to previous studies by showing that the association between TyG index and END was consistent when applying two well-validated definitions for END. Moreover, when defining END separately from sICH, the association of END with TyG index remained in our cohort (Tables S6 and S7), since patients who had sICH were older, more likely to have vascular risk factors, higher blood glucose, and a more severe onset (Table S11). Another retrospective cohort study showed that TyG index was inversely associated with ENI (adjusted OR 0.68, 95% CI 0.52-0.89, p = 0.004) in AIS patients who received intravenous alteplase thrombolysis [31], which might also support our findings. The aforementioned findings may indicate that the TyG index is probably a biomarker for unfavorable early neurological functional outcomes. It is worth noting that the abovementioned studies were heterogeneous in study population and design; further large-sample size prospective studies are needed to validate the relationship between the TyG index and END risk.
There are several possible explanations for the relationship between TyG index and unfavorable early neurological outcomes after intravenous thrombolysis. First, patients with higher insulin resistance have elevated blood levels of fibrinolysis inhibitors, such as plasminogen activator inhibitors, which may reflect an impairment of endogenous fibrinolytic capacity [32]. Second, insulin resistance is also known to correlate with the worsening response to intravenous thrombolysis [33,34]. Insulin resistance may affect the structure of the offending clot itself, rendering it denser and more resistant to lysis in patients with AIS who received reperfusion treatments [35].
Our findings suggested a sex difference in the association between TyG index and END risk. Possible explanations might include sex disparities in glucose metabolism. For example, impaired glucose tolerance might be more prevalent in women [36]. Moreover, experimental data showed that endogenous estrogen may play a role in higher insulin sensitivity in a female rodent model [37]. Clinical evidence also demonstrated that menopausal hormone therapy may improve insulin sensitivity for postmenopausal women [38]. The above mentioned studies may highlight the need for a sex-specific risk management strategy.
The strengths of our study include using different models to validate the association between TyG index and END risk in a multicenter sample with two-well validated definitions of END. This study has some limitations. First, this is a retrospective observational study with a moderate sample size, inevitably introducing selection bias. Second, this study only included Chinese stroke patients; our findings may therefore not be generalizable to other populations. Third, because of practical limitations derived from our clinical setting, no direct measure to assess insulin resistance was used in our cohort. However, previous studies have shown that the TyG index has high sensitivity and specificity for assessing insulin resistance, suggesting that TyG index could be useful as a surrogate to identify insulin resistance [39]. Lastly, the early measurement may overestimate the prevalence of insulin resistance, since insulin resistance measurement is time-dependent during the acute ischemic stroke onset.

Conclusions
In conclusion, insulin resistance represented by increased TyG index is associated with higher odds of END and lower odds of ENI in acute ischemic stroke patients who received intravenous thrombolysis. Targeting the TyG index could be a potential obtainable biomarker for risk stratification in this stroke population during routine clinical practice.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/jcm12103471/s1, Figure S1: Interactions between the TyG index and END or ENI; Table S1: Absolute risks for early neurological outcomes in different TyG index tertile groups; Table S2: Association between TyG index and END 2 ; Table S3: Association between TyG index and END 4 ; Table S4: Association between TyG index and END 2 by limited to patients who did not receive thrombectomy treatment; Table S5: Association between TyG index and END4 by limited to patients who did not receive thrombectomy treatment; Table S6: Association between TyG index and END2 apart from sICH; Table S7: Association between TyG index and END 4 apart from sICH; Table S8: Differences in baseline characteristics in patients with and without ENI; Table S8: Differences in baseline characteristics in patients with and without ENI; Table S10: Association between TyG index and ENI by limited to patients who did not receive thrombectomy treatment; Table S11: Baseline characteristics in patients with and without early sICH.

Institutional Review Board Statement:
The study protocol was reviewed and approved by the Fujian Medical University Union Hospital Ethics Committees (2019KY076). Written informed consent was waived due to the nature of our retrospective study of routine clinical data.

Informed Consent Statement: Not applicable.
Data Availability Statement: Data are available from the corresponding author on reasonable request.