An Analysis of the Potential Relationship of Triglyceride Glucose and Body Mass Index With Stroke Prognosis

Background: The inverse association between obesity and outcome in stroke patients (known as the obesity paradox) has been widely reported, yet mechanistic details explaining the paradox are limited. The triglyceride glucose (TYG) index has been proposed as a marker of insulin resistance. We sought to explore possible associations of the TYG index, body mass index (BMI), and stroke outcome. Methods: We identified 12,964 ischemic stroke patients without a history of diabetes mellitus from the China National Stroke Registry and classified patients as either low/normal weight, defined as a BMI <25 kg/m2, or overweight/obese, defined as a BMI ≥ 25 kg/m2. We calculated TYG index and based on which the patients were divided into four groups. A Cox or logistic regression model was used to evaluate the association between BMI and TYG index and its influence on stroke outcomes, including stroke recurrence all-cause mortality and poor outcome (modified Rankin Scale score of 3–6) at 12 months. Results: Among the patients, 63.3% were male, and 36.7% were female, and the mean age of the patient cohort was 64.8 years old. The median TYG index was 8.62 (interquartile range, 8.25–9.05). After adjusting for multiple potential covariates, the all-cause mortality of overweight/obese patients was significantly lower than that of the low/normal weight patients (6.17 vs. 9.32%; adjusted hazard ratio, 0.847; 95% CI 0.732–0.981). The difference in mortality in overweight/obese and low/normal weight patients with ischemic stroke was not associated with TYG index, and no association between BMI and TYG index was found. Conclusion: Overweight/obese patients with ischemic stroke have better survival than patients with low/normal weight. The association of BMI and stroke outcome is not changed by TYG index.


INTRODUCTION
Stroke is a leading cause of death and disability worldwide (1). Obesity is known to raise the risk of stroke (2). However, numerous reports have shown an inverse association between obesity and outcome in stroke patients, which is in contrast to the general population and is known as the stroke-obesity paradox (3)(4)(5). Although it is unknown if the stroke-obesity paradox is related to insulin sensitivity/resistance (IR), IR has been theorized to contribute to this phenomenon (6).
IR is a physiologic state in which a normal amount of insulin produces a subnormal physiologic response (7). IR is closely associated with obesity (8) and is an independent risk factor for mortality and major disability after stroke (9)(10)(11). Several mechanisms have been proposed to explain this association, including enhanced local inflammatory and prothrombotic responses and accelerated atherosclerotic development (12). The gold standard for measuring IR is the euglycemic hyperinsulinemia clamp (13). However, this method is difficult to apply in large population studies and clinical settings because of its costly, time-consuming and complex nature. In response to this limitation, large-scale epidemiological studies have resorted to the homeostasis model assessment of IR (HOMA-IR) index (9,14), while the application of HOMR-IR is limited by the need for the fasting immunoreactive insulin.
A potential surrogate method of measuring IR has been developed, known as the triglyceride glucose (TYG) index (15). TYG index can be used in large-scale observational and/or interventional cohorts by comparing with the hyperinsulinemiaeuglycemic clamp (16). Previous studies have shown that TYG index is positively correlated with ischemic stroke, and ischemic stroke patients with a high TYG index have a higher risk of dying or poorer outcome (17). Therefore, the aim of our study is to evaluate if the association between body mass index (BMI) and stroke outcomes is modified by TYG index.

Study Population
Launched in 2012 by the Chinese Ministry of Health, the China National Stroke Registry II (CNSR II) is a prospective cohort study conducted nationwide aimed at establishing a reliable national stroke database for evaluating stroke care delivery in clinical practice (18). The criteria for site selection in China National Stroke Registry I (CNSR I) have been previously published (19) and were replicated in the CNSR II study. Patient's included in the study must have met the following criteria: age ≥ 18 years old, diagnosed within 7 days of the index event of ischemic stroke, transient ischemic attack (TIA), spontaneous intracerebral hemorrhage, or subarachnoid hemorrhage confirmed by brain imaging, or direct hospital admission from a physician's clinic or emergency department. Of the 25,018 patients in the CNSR II, 15,544 were diagnosed with ischemic stroke without a history of diabetes mellitus. Ultimately, 12,964 were analyzed after excluding patients without necessary clinical data including weight, height, fasting triglyceride or fasting glucose at admission (n = 848), as well as those who were lost to follow-up at 12 months (n = 1,732; Figure 1).
The study was approved by the central Institutional Review Board at Beijing Tiantan Hospital and prior to the enrollment, all patients were provided written informed consent.

Data Collection and Management
Patient baseline data were collected by trained research coordinators at each study center including basic patient characteristics, laboratory data, vascular risk factors, prior diagnoses and treatments. Vascular risk factors included stroke history, diabetes mellitus, hypertension, coronary heart disease, atrial fibrillation and prior/current smoking history. Past medication history included antihypertensive, antiplatelet drugs, anticoagulation, lipid-lowing, and hypoglycemic drugs. BMI was measured by trained nurses at each clinical center at the time of patient enrollment. The delineation between low/normal and overweight/obese patients was set at 25 kg/m 2 , based on World Health Organization standards (20). Fasting blood samples were collected for all participants within 24 h of admission. Using an automated enzymatic method, triglyceride and glucose levels were tested at each research center. TYG index was calculated as ln[fasting triglycerides (mg/dl) × fasting glucose (mg/dl)/2] (21).

Patient Follow-Up and Outcome Assessment
The primary outcomes of this study were stroke recurrence, all-cause mortality, and modified Rankin Scale (mRS) score, which were obtained by trained research coordinators that were blinded to the patients' baseline characteristics. Instances of recurrent stroke included ischemic stroke, intracranial hemorrhage, and subarachnoid hemorrhage. Poor functional outcomes were defined as a modified Rankin Scale score of 3-6. Modified Rankin Scale scores range from 0 (no symptoms) to 6 (death). At 3, 6, and 12 months after stroke, trained research personnel contacted patients by telephone following standard scripts, and collect information on death, stroke recurrence, and disability. During the follow-up, to ensure reliable diagnosis, stroke recurrences associated with rehospitalization were sourced to the corresponding hospitals. Adjudication was performed by the research coordinators together with the principal investigator in case of a suspected recurrent cerebrovascular event without hospitalization. Follow-up procedures for this study performed as previously described (22).

Statistical Analysis
Continuous variables are presented as mean ± standard deviation or median with interquartile range (IQR). Categorical variables are presented as frequencies and percentages. We used a student's t-test to compare groups with normally distributed parameters, a Wilcoxon test for non-normally distributed parameters, and a χ 2 test for categorical variables. A Cox proportional hazard analysis was conducted to estimate the hazard ratio (HR) and the 95% confidence interval (CI) of death and stroke recurrence. Odds ratios (OR) with 95% CIs were calculated for poor functional outcomes at 12 months and were estimated using a logistic regression model. Adjusted variables for HR with 95% CIs were sex, age, NIHSS score at admission, IV thrombolytic administration, prior/current smoking history, medical history (i.e., myocardial infarction, atrial fibrillation, hypertension, hyperlipidemia), medications (from the previously listed categories), and laboratory examination (i.e., total cholesterol, high density lipoprotein, low density lipoprotein). The impact of BMI and IR status on the risk of developing poor outcomes was also tested using Cox or logistic regressions. Kaplan-Meier analyses were used to assess survival from time of initial diagnosis until stroke recurrence during the 12-month follow-up period and a log-rank test was used to compare groups. We also examined the potential impact of age on the association between BMI and outcomes by separating patients into two groups using a cutoff of ≥65 years old. We assessed interactions between weight status in each age category using weight status × age category in multivariable models.
Statistical analyses were performed using SAS software (version 9.4, SAS Institute Inc, Cary, NC). Results were deemed statistically significant at a two-sided (p < 0.05).

Baseline Patient Characteristics
Twelve thousand nine hundred sixty-four patients were included in the study, with an average age of 64.83 ± 12.2 years old and the gender distribution was 65.17% male and 34.83% female. The median TYG index was 8.62 (IQR, 8.25-9.05). We separated patients into four groups based on TYG index. Other baseline patient characteristics are summarized in Table 1. In comparing to low/normal patients, obese/overweight patients were typically younger, more likely to have a history of hypertension and received the treatment for hypertension prior to admission. Within the overweight/obese cohort, patients with a high TYG index were more likely to have a history of hyperlipidemia. Significant differences in the sex, NIHSS score, and history of anticoagulation therapy were also observed within groups.

Association of BMI With Outcome
One thousand seventy-nine patients died in the 12 months following their initial stroke occurrence, including 832 (9.32%) overweight/obese and 247 (6.12%) low/normal weight patients. The prognosis of patients with ischemic stroke at 12 months in   each of the BMI categories was shown in Table 2. Compared to patients classified as low/normal weight, overweight/obese patients had a lower risk of death at 12 months after adjusting for potential covariates (9.32 vs. 6.12%; adj. HR 0.847, CI 95% 0.732-0.981). No significant association was found between being overweight/obese and poor functional outcome or stroke recurrence (p = 0.2738 and 0.3078). There was an interaction effect of age group by BMI for the risk of stroke or death (p = 0.0431 and 0.0173, respectively; Supplementary Table 1). An inverse relationship between BMI and stroke mortality (obesity paradox) was observed in patients older than 65 years, whereas no association between them was observed in young patients. . Similar results were found for the endpoints indicating a poor outcome. No significant association was found between BMI and stroke recurrence in the four groups. There was also no interaction between BMI and TYG index on the risk of stroke recurrence (p = 0.2449), mortality (p = 0.7887) and poor functional outcome (p = 0.7452; Table 3). In the analyses of continuous TyG index, BMI was associated with stroke mortality (adj HR 0.841, CI 95% 0.725-0.974, p = 0.0210), and there was no interaction between TYG index and BMI on stroke outcome ( Table 4). Multivariate analysis of stroke outcomes by BMI and TYG index showed that overweight and obese patients had a lower risk of death at 12 months, and TYG index was associated with poor stroke outcomes (p = 0.0237, p = 0.0260; Table 5).

DISCUSSION
We found that overweight/obese stroke patients had a remarkably lower mortality compared to patients with normal/low BMI. There is an interaction between age and BMI. We found that association between BMI and stroke outcome was modified by age. However, we found no difference in outcomes in overweight/obese and normal/low-weight stroke patients that were stratified by TYG index and we did not observe an association between BMI and TYG index in ischemic stroke patients. A previous study examined IR using HOMA-IR suggested that the obesity paradox for mortality exists in insulin-resistant patients but not insulin-sensitive patients (6), suggesting IR may be part of the underlying mechanism of the obesity-stroke paradox. In this study, however, we did not observe a relationship between IR and the obesity-stroke paradox. Possible explanations for this result include that obesity is a fundamental risk factor for the onset and development of insulin resistance, and the majority of the population with IR is obese or overweight. IR results in excessive release of free fatty acids (FFAs) into circulation by unrestrained lipolysis (23). The increased flux of FFAs increases triglyceride synthesis in the liver and the synthesized triglyceride is then released into plasma, which raises the risk of hypertriglyceridemia and subsequent metabolic syndrome (23). When the storage capacity of adipose tissue reaches full capacity, the energy surplus presenting in obese patients leads to an increased efflux of FFAs and an ectopic accumulation of fat in the liver (24). As such, numerous studies have demonstrated that the TYG index is highly associated with the incidence of non-alcoholic fatty liver disease (NAFLD) (25)(26)(27). An epidemiological survey of a Chinese population also found that an elevation of the TYG index might predict an increased incidence of NAFLD (28). A high TYG index is more likely to reflect metabolic abnormalities after the occurrence of the IR. Previous studies have also shown that visceral fat (VAT) rather than subcutaneous fat (SAT) is correlated with the occurrence of cardiovascular diseases (29). Kim et al. found that a low VAT proportion is associated with favorable and excellent outcomes in acute ischemic stroke patients treated with intravenous thrombolysis (30). Another study found that better clinical outcomes in obese patients are associated with a lower proportion of VAT (30). Conversely, higher visceral and liver fat amounts have been found in IR obesity (31), suggesting that IR may counteract the paradoxically protective effects of obesity in stroke patients. Second, the presence of metabolically healthy obese (MHO) individuals, a previously identified subset of obese subjects without metabolic abnormalities (including IR and a proatherogenic lipoprotein profile) may also be responsible for the controversial results. Although MHO criteria have not been clearly defined, normal glucose and lipid metabolism parameters, in addition to the absence of hypertension, is considered typical MHO diagnostic criteria (32) (15,21,34,35). However, these findings lack consistency and a systematic review deemed the evidence for the usefulness of the TYG index as a surrogate biochemical marker of IR as "moderate-to-low quality." We identified the obesity paradox in all four groups before the covariates were adjusted, but the results were no longer significant after adjustment. Indeed, whether or not the obesity paradox is a true phenomenon is still the subject of much debate. Some have proposed that the paradox is caused by physiological factors that reduce risk in obese participants (36), whereas others suggested that it is an artifact resulting from selection bias in observational studies (37). From clinical and epidemiological observations, many contributing factors have been identified that lead to lower mortality, such as demographic characteristics (e.g., age, sex, educational status) and physical conditions (e.g., severity and duration of heart disease, initial neurological severity, and cachexia). In the present study, we found that overweight/obese patients are more likely to use antihypertensive, lipid-lowering, and antiplatelet drugs at baseline. We adjusted for patient characteristics that include sex, age, NIHSS score at admission, IV thrombolytic therapy, smoking, medical history (i.e., myocardial infarction, atrial fibrillation, hypertension, hyperlipidemia), and medication (i.e., antihypertensive, antiplatelet, anticoagulation, lipid-lowing, and hypoglycemic drugs). Despite the adjustments for these potentially confounding factors, the obesity paradox remained, with an adjusted HR for overall mortality in the overweight/obese group being 0.847 (0.732-0.981). A previous study including a cohort of 21,884 patients from the Danish National Indicator Project registry, found that post-stroke mortality in the 5-year follow-up period is lower in overweight (BMI 25.0-29.9 kg/m 2 ) and obese (BMI 30.0-34.9 kg/m 2 ) patients compared to those with normal weight (BMI 18.5-24.9 kg/m 2 ) and underweight (BMI<18.5 kg/m 2 ) (5). We also identified a BMI-age interaction since obese or overweight stroke patients older than 65 have a lower mortality rate, while the interaction is not presented in patients younger than 65. In contrast to our results, Vemmos et al. found that the protective effect of BMI on mortality decreases with increasing age (3). However, a post-hoc analysis of a randomized controlled trial in China (Clopidogrel in High-risk patients with Acute Non-disabling Cerebrovascular Events) did not identify the obesity paradox or observe an influence of age on the association of weight status with the prognosis of stroke (38). In a national wide representative survey of U.S. adults, an effect of age among stroke survivors was discovered, whereby a higher BMI was associated with a marginally increased mortality in younger individuals that reduced linearly with increasing age. This is similar to our findings (39).
We acknowledge several limitations with our findings. In particular, as a post-hoc analysis, several variables of interest were not accessible for analysis because they were not recorded. Additionally, a comparison of BMI with the waist-to-hip ratio cannot be evaluated. Our findings also use baseline BMI, fasting triglyceride and glucose, which may have confounding effects on the results as the collection of consecutive data about body weight, fasting glucose, and triglyceride at the 12 months followup is unavailable. Measurement errors across multiple centers may also slightly impact patients' fasting triglyceride and glucose levels, yet the laboratory values of these studies should still be comparable as the measurement techniques in all centers are based on the recommendation of the International Federation of Clinical Chemistry and Laboratory Medicine. A final limitation of our study is the relatively short follow-up period of 12 months. Several other studies have tracked patients much longer. For example, Korean Stroke Registry included 2,317 patients with ischemic stroke with 7.5-year follow-up (40) and the Prospective investigation of Greece included 2,785 first-ever acute stroke patients with a 10-year follow up (3).
In conclusion, we found that overweight and obese stroke patients have significantly lower mortality compared to patients with normal BMI, in line with previous reports of the stroke obesity paradox. Additionally, our findings suggest that age may influence the relationship between BMI and stroke prognosis. Stratification by TYG index eliminated the presence of the obesity paradox in this patient cohort, and thus TYG index, or its correlate IR, may not be beneficial in understating the metabolic mechanisms responsible for the stroke obesity paradox.

DATA AVAILABILITY STATEMENT
The data analyzed in this study is subject to the following licenses/restrictions: The datasets generated for this study are available on request to the corresponding author.
Requests to access these datasets should be directed to YW, yongjunwang@ncrcnd.org.cn.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by The Central Institutional Review Board at Beijing Tiantan Hospital. The patients/participants provided their written informed consent to participate in this study.

AUTHOR CONTRIBUTIONS
ZH, YP, YY, XY, and YoW contributed to the conception and design of the study. YiW and XZ assisted with data acquisition and interpretation. ZL and XM coordinated the study. YP and XX conducted the statistical analysis. ZH contributed to drafting. YoW is the guarantor for this paper. All authors read and approved the final manuscript.