Comparison of metabolic phenotypes and BMI as indicators of recurrent stroke in Chinese hospitalized stroke patients: a retrospective study

Metabolic status and body mass index (BMI) are known as apparent risk factors of recurrent stroke, but which one is more likely related to recurrent stroke remains uncertain. This study aimed to compare the metabolic phenotypes and BMI as indicators of recurrent stroke in Chinese hospitalized stroke patients. Methods In this retrospective population-based study, 856 hospitalized stroke patients from the Third Aliated Hospital of Soochow University were enrolled. Recurrent stroke was dened as newly-onset stroke patients with a history of previous stroke. Metabolic phenotypes were based on Adult Treatment Panel III criteria. BMI ≥ 25kg/m 2 was dened as obesity. means or for continuous variables, and numbers (percentages) for categorical variables. P values were calculated from one-way ANOVA for continuous variables and χ 2 test for categorical variables. Abbreviations: BMI, body mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure; TC, total cholesterol; TG, triglycerides; HDL-C, high- density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; FPG, fasting plasma glucose;


Introduction
Obesity as a worldwide epidemic has emerged as a critical public health issue, for its considerable impact on a number of diseases such as stroke, cardiovascular diseases, diabetes, hypertension, and cancer 1 . Abundant evidence have suggested that obesity is also a risk factor for stroke for the primary prevention, overweight and obesity is associated with an overt increasing risk of stroke 2 , and a signi cant 6% increase was concomitant with each unit increase of body mass index (BMI) 3 . However, for the secondary prevention, some prospective studies reported that obese patients after stroke had a lower risk of recurrent stroke 4 , overweight and mild obese patients had better 10-year survival after stoke, compared with normal weight patients 5 . The association between obesity and recurrent stroke is undetermined. Unfavorable metabolic status probably contributes to the distinctive impact on recurrent stroke between obese and non-obese stroke patients 6 .
Metabolic syndrome (MetS) as an integrated statue of metabolic disorders-elevated plasma glycemia, central obesity, hypertension and dyslipidemia, is an important predictor of recurrent stroke, with 52% increased risk of recurrent stroke 7 . However, among the components of Mets, obesity illustrates lower in uence than the other risk factors, no matter on stroke 8 or recurrent stroke 9,10 . Concretely, a metaanalysis of cohort studies showed that the hazard ratio of obesity for recurrent stroke was 1.12 (0.89-1.41), signi cantly lower than that of elevated glycemia and dyslipidemia 9 . A multiclinic study among mild-to-moderate stroke patients also did not nd the signi cant association of BMI with 5-year risk of recurrent stroke, which was lower than that of elevated blood pressure, elevated glycemia and low HDLc 11 . Therefore, we assume that recurrent stroke probably be in uenced by obesity in conjunction with the metabolic status.
Whether or not recurrent stroke is more likely related to poor metabolic health rather than to obesity remains uncertain, while con icting data are similarly examined among the new-onset stroke. Previous studies reported that obese individuals with otherwise metabolically healthy pro le, referred as "metabolically healthy obesity (MHO)", did not have increased risk of stroke than metabolically healthy non-obesity 10 . However, apparent lean but metabolically unhealthy experienced higher risk of stroke, and the risk increased as the number of metabolic syndrome components increased 12 .
Thus, the prevent study aimed to compare the metabolic phenotypes and BMI as indicators of recurrent stroke in Chinese hospitalized stroke patients, and proposed the hypothesis that metabolic unhealthy status is more closely associated with the higher risk of recurrent stoke than obesity.

Study population
This retrospective study was conducted among the hospitalized patients who were diagnosed with cerebrovascular diseases in the Third A liated Hospital of Soochow University. From January 2017 to January 2018, a total of 870 participants were recruited with the inclusion criteria: 1) who agreed to participate the present study with the informed form; 2) the cerebrovascular diseases were diagnosed according to computed tomography (CT) or magnetic resonance imaging (MRI) scans. Each patient underwent a face-to-face interview for a standard questionnaire and blood samples collection for biochemical measurements. For present analysis, the exclusive criteria were: 1) patients with subarachnoid hemorrhage, transient ischemic attack without progression to stroke or lacunar infarction (n=7); (2) missing information on lipid pro les, blood pressure and plasma glucose (n=4); (3) extreme value of body mass index (BMI) (n=3). Eventually, 856 participants were enrolled in present study.
The study protocol was approved by the Ethics Committee of the Third A liated Hospital of Soochow University. Each participants provided the written informed consent.

Data collection
A face-to-face interview was performed through a standard questionnaire for the collection of information on sociodemographic characteristics, lifestyle and medical history. Current smokers and drinkers were de ned as those who have smoked cigarettes and consumed alcohol in the past six months. Physical activity was divided into high physical activity or not, based on the International Physical Activity Questionnaire (IPAQ). Dietary habit was classi ed as omnivorous, vegetable activists, and meat activists, as well as the duration and frequency of consuming vegetables, fruits and dairy products were collected.
Anthropometric information of each participant including blood pressure, height and body weight was measured by trained personnel according to a standard protocol. Blood pressure was measured using an automated electric device (OMRON Model HEM-752 FUZZY, Omron Company, Dalian, China), on the nondominant arm. Two blood pressure measurements were taken with 1-minute interval, the mean of two values were de ned as the available blood pressure in the current analysis. BMI was calculated as weight in kilogram divided by squared height in meters.

Biochemical measurements
After at least 10-hour fasting, blood samples of each participants were obtained for the measurements of fasting plasma glucose (FPG) and lipid pro les (triglyceride, TG; low-density lipoprotein cholesterol, LDLc; high-density lipoprotein cholesterol, HDL-c; total cholesterol, TC), which were measured by an autoanalyzer (AU-5800 Chemistry System, Beckman, USA).

De nitions
Obesity and metabolic phenotypes Obesity phenotypes were de ned as BMI ≥25kg/m 2 and non-obesity as BMI<25kg/m 2 based on the BMI cut-point among Asian population 13 , which had been veri ed among Chinese population recently 14,15 .
According to Adult Treatment Panel III (ATP-III), the metabolic phenotypes were determined as: 1) elevated blood pressure: systolic blood pressure (SBP) ≥130mmHg and/or diastolic blood pressure (DBP) ≥90 mmHg, and/or taking of antihypertensive drugs; 2) elevated blood glucose: FPG ≥5.6mmol/l, and/or taking of antidiabetic drugs; 3) high TG: TG ≥1.7mmol/l, and/or taking lipid-lowing medication; 4) low HDL-c: HDL-c <1.04mmol/l in men and <1.29mmol/l in women. Individuals with two or fewer of the previously mentioned items were considered as being in a metabolically healthy status.

Recurrent stroke
Among the participants having a history of previous stroke, those with newly-onset stroke diagnosed by neurologists based on clinical symptoms, physical signs, and CT/MRI suggestions were considered as recurrent stroke.

Statistical analyses
The current analyses were conducted based on the combined obesity and metabolic phenotypes. Mean ± standard (SD) and median (interquartile range) were used as the normally distributed and skewed continuous variables. Numbers (proportions) were used as the categorical variables. One-way ANOVA for continuous variables and the Chi-square for categorical variables were performed for the comparisons among the four phenotypes; linear regression was used to analyze the trend across the four phenotypes.
Dependent on multivariate logistic regression analyses, odds ratios (ORs) and 95% con dence intervals (CIs) were presented as the risk of recurrent stroke across the variant BMI groups, ATP-III components, and metabolic phenotypes, as well as the four phenotypes across the obesity and metabolic status.
We calculated C statistics, net reclassi cation index (NRI), and integrated discrimination improvement (IDI) to evaluate the incremental prognostic values of individual BMI/metabolic status or the combined risk factors beyond conventional risk factors 16 .
We used SAS statistical software (version 9.3, Cary, NC) to perform all the statistical analyses, and considered two-tailed P values <0.05 were statistically signi cant.

Results
Characteristics of study population according to the obesity and metabolic phenotypes Sociodemographic, medical history and biochemical characteristics of the study population were presented in Table 1. Among all of the 856 participants, 44.8% (n=384) were obese and 77.7% (n=665) were metabolically unhealthy. The prevalence of different phenotypes were 14.5% for MHNO, 7.8% for MHO, 40.7% for MUNO, and 37.0% for MUO. Additionally, MHO accounted for 17.4% of obesity participants. Obese participants were younger, had higher blood pressure and serum lipids levels than non-obese individuals, no matter in metabolically healthy or unhealthy groups. Among the obesity/metabolic phenotypes, compared to the MHNO, individuals in MHO, MUNO, MUO were had gradually higher levels of TG, TC, LDL-c, FPG, proportions of hypertension, diabetes, hypercholesterolaemia, and higher usage of antihypertensive and antidiabetic drugs (all P values for trend <0.05).
Obesity/metabolic phenotypes and risk of recurrent stroke Table 2 showed the obesity/metabolic phenotypes and risk of recurrent stroke using multiple logistic regression. Divided BMI into two categories, comparing to non-obesity, the risk of recurrent stroke increased 50% (OR=1.50, 95% CI 1.06-2.13) in obese individuals after adjusting for age, sex, lifestyle, education, income, serum lipids, blood pressure, glucose, medication and family history. However, with the increase of BMI, signi cantly increased risk of recurrent stroke was not found. Among the ATPcomponents, elevated blood pressure and glucose and low HDL-c, but not high TG, were associated with increased risk of recurrent stroke in the multivariate-adjusted model (OR=3.55, 95% CI 1.33-7.93 for elevated blood pressure and OR=1.62 95% CI 1.14-2.28 for elevated blood glucose and OR=1.46 95% CI 1.01-2.11 for low HDL-c). In terms of metabolic status, referred to metabolically healthy, metabolically  Figure 1) did not nd the signi cant association between MHO and recurrent stroke, comparing to the nonobese counterparts. Likewise, individuals with MUNO did not have higher risk of recurrent stroke compared to MHNO. However, when combining the obesity and metabolic status, 1.14-fold increased risk of recurrent stroke was found in MUO compared to MHNO.
Incremental prognostic values of BMI/metabolic status Table 3 showed whether individual BMI or metabolic status could add the prognostic value of recurrent stroke to the conventional risk factors. The C statistic (95% CI) of the predictive models of conventional risk factors was 0.607 (0.574-0.640), and the addition of BMI or metabolic status did not increase the C statistic. Nevertheless, we found metabolic status, but not BMI, could signi cantly improve reclassi cation risk of recurrent stroke when adding to the conventional model. Adding the combined BMI and metabolic status to the conventional model, greater incremental prognostic values were found (all P values <0.05).

Discussion
In the present study, we found that among the Chinese hospitalized stroke patients, metabolically unhealthy status and ATP-components (elevated blood pressure and glucose and low HDL-c) signi cantly increase the risk of recurrent stroke, regardless of BMI, whereas no association existed between BMI and risk of recurrent stroke. Furthermore, we found metabolic status, but not BMI, signi cantly improved prognostic value of recurrent stroke when adding to the conventional risk factors.
Numerous previous studies have demonstrated metabolic syndrome was associated with increased risk of recurrent stroke 11,17,18 , the ACROSS-China study showing a hazard ratio of 1.94 (1.39-2.73) in patients with metabolic syndrome vs. no metabolic syndrome 18 , a longitudinal study calculating a higher risk of recurrent stroke with the hazard ratio of 1.43 (1.01-2.01) 11 . Further controlling for its components, the signi cant associations disappeared. In accordance with precious studies, our study found metabolic unhealthy individuals had 72% higher risk of recurrent stroke than metabolic healthy ones, as well as its components (elevated blood pressure and fasting glucose and low HDL-c levels). All these results underlined the crucial value of metabolic status on stroke prognosis.
Although obesity is a considerable risk factor of stroke, no matter de ned based on waist circumference or BMI, some certain studies interestingly reported that obesity might be a "protective" factor of stroke recurrence 4,19,20 . The obesity paradox is partly attributed to the selection bias (obesity patients are younger, more likely to be female and have less proportion of smokers) and the limitation of obesity de nition 20 . Inconsistent with the situation aforementioned, the present study detected that obesity patients (de ned as BMI ≥ 25kg/m 2 ), controlling metabolic confounder factors, had weakly higher risk of recurrent stroke. Howbeit BMI, as a continuous variable, was not found associated with recurrent stroke. The discrepancies might be attributed to the indiscriminate gender proportion (P = 0.42) and lifestyle (P for current smoker was 0.60, P for current drinker was 0.07) between obesity and non-obesity individuals, and further adjusting metabolic-related confounders, although the obesity patients were younger (P < 0.001).
This is the rst study to investigate and compare the effects of metabolic status and BMI on recurrent stroke. In order to con rm the impact of metabolic status and BMI, we strati ed the population based on metabolic phenotypes and different BMI group (BMI ≥ 25kg/m 2 and BMI < 25 kg/m 2 ) into four groups: MHNO, MHO, MUNO and MUO groups. Previous data about the obesity/metabolic phenotypes and stroke were focused on the stroke primary prevention eld, and drawn controversial conclusions [21][22][23][24] . A nationwide population-based study in Korea reported that MHO individuals were not at increased risk of stroke 21 , whereas another longitude study in 7122 participants found that MHO individuals were at increased risk of cardiovascular disease including stroke, and stroke separately 22 . Studies about the effect of obesity/metabolic phenotypes on stroke prognosis were sparsely. Results in present study showed that, compared to MHNO, MUO had statistically higher risk of recurrent stroke, and the OR of MUNO was higher than that of MHO, although the P values of ORs were less than 0.05. Above-mentioned results suggested that metabolic status, but not BMI per se, increased the risk of recurrent stroke. Furthermore, evidenced by NRI and IDI, metabolic phenotypes rather than BMI were associated with higher stroke recurrence risks, adding metabolic status to conventional model substantially improved the risk strati cation for recurrent stroke, while adding BMI did not.
Several suggested mechanisms could interpret metabolic status is more important than BMI in risk of recurrent stroke. Firstly, chronic in ammation was postulated to initiate recurrent stroke in metabolically unhealthy stroke patients, regardless of BMI. It has been well known that in ammation plays a critical role, through mechanisms such as the development of atherosclerosis, plaque instability, and triggering of plaque rupture, in the pathogenesis of cerebrovascular diseases 25 . Su cient population-based cohort studies similarly demonstrated the most extensively used blood marker of in ammation, C-reactive protein (CRP), signi cantly contributed to the recurrent stroke 26-28 . Recently, several previous studies reported metabolically unhealthy obese and nonobese individuals had unfavorable in ammation status than metabolically healthy counterparts [29][30][31] , with higher level of C-reactive protein, IL-6, tumor necrosis factor α (TNF-α), adipocyte fatty acid binding protein (A-FABP). Thus, the unfavorable in ammation status of metabolically unhealthy individuals, in all likelihood, mediated recurrent stroke. In addition, visceral adipose tissue may also play a substantial role. In ammation of visceral adipose tissue mediated metabolically unhealthy status regardless of generalized obesity 32 . Camhi SM, et al. also revealed that metabolically unhealthy individuals, no matter obese or non-obese, had higher amount of visceral adiposity or liver fat than metabolically healthy participants 33 . Visceral adiposity, suggested by su cient evidence, was a risk factor for recurrent stroke 7,34 . A retrospective study in Japan found that recurrent ischemic stroke patients had signi cantly higher visceral fat level than those without recurrence, and visceral fat level was associated with risk of recurrent stroke (OR = 2.44, 95% CI 1.08-5.49) 34 .
The present study had a few limitations of note. Firstly, medications which may affect the results were not fully adjusted. However, we had adjusted the antihypertensive/antidiabetic drugs and lipid-lowering drugs, which may have great in uence on the development of atherosclerosis and plaque instability, in the multivariate logistic regression. Secondly, our study was a retrospective study, a prospective study was warranted to further de nite the associations of metabolic phenotypes and BMI with recurrent stroke. Thirdly, our present study was performed in Chinese hospital patients, it should be cautious to generalize our ndings to other ethnicities.

Conclusions
In conclusion, our present study found a signi cantly increased association of recurrent stroke with metabolically unhealthy status, while the association between BMI and recurrent stroke was not detected. In addition, adding to the conventional risk factors, metabolic status statistically improved prognostic value of recurrent stroke, whereas BMI did not. It is suggested that recurrent stroke is, to a large extent, related to poor metabolic status rather than to BMI. Therefore, metabolic abnormalities controlling could be a more e cacious method to prevent recurrent stroke than weight controlling. Abbreviations ATP-III, Adult Treatment Panel III; A-FABP, adipocyte fatty acid binding protein; BMI, Body mass index; CT, computed tomography; CI, con dence intervals; CRP, C-reactive protein; DBP, diastolic blood pressure; FPG, fasting plasma glucose; HDL-c, high-density lipoprotein cholesterol; IDI, integrated discrimination improvement; IPAQ, International Physical Activity Questionnaire; LDL-c, low-density lipoprotein cholesterol; Mets, Metabolic syndrome; MHO, Metabolically healthy obesity; MHNO, metabolically healthy non-obesity; MUNO, metabolically unhealthy non-obesity; MUO, metabolically unhealthy obesity; MRI, magnetic resonance imaging; NRI, net reclassi cation index; OR, odds ratios; SBP, systolic blood pressure; SD, standard; TG, triglyceride; TC, total cholesterol; TNF-α, tumor necrosis factor α.

Declarations
Ethics approval and consent to participate The study protocol was approved by the Ethics Committee of the Third A liated Hospital of Soochow University. Each participants provided the written informed consent.

Consent for publication
Not applicable.

Availability of data and materials
The data in the current study arisen from a dataset of Department of Endocrine and Metabolic Diseases, the Third A liated Hospital of Soochow University, are not publicly available due to security consideration, but are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.