Interaction between overweight, obesity and smoking on the risk of prediabetes and type 2 diabetes in Guangdong, China

Pre-diabetes mellitus (PDM) is considered an early warning signal of type 2 diabetes mellitus (T2DM). However, most studies only analyze the risk factors of diabetes, ignore the exploration of PDM. The aim of this study was to investigate the independent and combined impacts of overweight obesity and smoking on the risk of PDM and T2DM. T2DM:type 2 diabetes mellitus; PDM:pre-diabetes mellitus; NGT:normal glucose tolerance; IFG:impaired fasting glucose; IGT: impaired glucose tolerance;IGR: impaired glucose Regulation; TC:Total cholesterol; TG:Triglyceride ; LDL-C:Low Density Lipoprotein Cholesterol;HDL-C:High-Density Lipoprotein Cholesterol; DBP:Diastolic Blood Pressure;SBP:Systolic Blood Pressure;BMI:Body Mass Index ;FPG:Fasting venous blood glucose;2hPG:2h plasma glucose concentration.


Participants
This study based on a health checkup of chronic non-communicable diseases among residents aged 18 in the Pearl River Delta region of Guangdong Province in 2017. It covered Dongguan, Guangzhou, Shenzhen, Zhuhai, and Jiangmen cities in the Pearl River Delta, Guangdong Province. Five cities, each of which was divided into layers according to the overall population size, urban population ratio, mortality and complex multi-stage sampling. The samples were collected by surveyors with uniform training, who conducted a comprehensive questionnaire survey including general demographic characteristics, disease history, life style and other factors. Laboratory tests and physical measurements were carried out by professionals and the information was recorded in the physical examination records. According to the principle of samegender, age difference less than 5 years, living in the same region, 28,208 patients with PDM and 28,208 NGT were matched with T2DM, respectively.
The questionnaire content The health examination questionnaire were covered (1) General situation investigation: gender, age, marriage, physical exercise, smoking: smoking during the examination and smoking history before the examination is de ned as smoking, never smoking is de ned as no smoking, etc. ; (2) Laboratory tests: fasting intravenous blood glucose test and oral glucose tolerance test to diagnose diabetes, and detect Total Related diagnostic criteria WHO de nition was used to diagnose diabetes and pre-diabetes (intermediate hyperglycaemia) [2] T2DM group Fasting venous blood glucose (FPG)≥7.0 mmol/L or 2h plasma glucose concentration (2hPG) after OGTT≥11.1 mmol/L. PDM group: Impaired Fasting Glucose (IFG) is 6.1≤FPG 7.0mmol/L, or Impaired Glucose Tolerance (IGF) is 7.8≤2hPG 11.1mmol / L. NGT group: FPG 6.1 mmol/L and 2hPG after OGTT 7.8 mmol/L Overweight and obesity: According to the "Guidelines for the Prevention and Control of Overweight and Obesity in Chinese Adults" [12], develop standards: Overweight: 24.0≤BMI 28.0kg/m 2 , Obesity: BMI≥28.0kg/m2.

Statistical analysis
Continuous variables conforming to the normal distribution were expressed as mean standard deviation `c±s , non-conforming variables were expressed as the median interquartile range (MQ), and the categorical variables were analyzed by chi-square test for continuous variable analysis of variance. The development process of diabetes was an ordered categorical variable, so an ordered multi-class logistic regression model was used: NGT was used as the control group, PDM and T2DM were the case groups, and univariate analysis was performed rst to incorporate the statistically signi cant variables into the multifactorial in the model. Then the unconditional logistic regression models of NGT-T2DM and NGT-PDM were established respectively to study the independent and comprehensive effects of smoking and overweight and obesity and diabetes. The multiplicative interaction was determined by P value less than 0.05, and the additive interaction was conducted by nonlinear mixed effect model by the relative excess risk ratio (RERI), attribution ratio (AP) and interaction index (S). These three indicators determined the additive interaction of smoking, overweight and obesity on impaired glucose tolerance and diabetes.

Characteristics of study participants
A total of 84,624 subjects were enrolled, with 28,208 subjects in each group. There were statistical difference in age, education, marriage, occupation, smoking, drinking, exercise, diet and BMI among the three groups. By pairwise comparison there were no difference in the distribution of education and marriage in the comparison NGT and PDM. In the comparison of NGT and T2DM, there were no signi cant difference in the distribution of occupations between the two groups too (Table 1).
Biochemical indicators of NGT, PDM, T2DM Table 2 shows the biochemical indicators between the three groups of NGT, PDM and T2DM. From the indicators, it might be found that HDL-C level decreased the risk to development PDM and T2DM, however, other indicators like TC, TG and SBP increase the risk of PDM and T2DM.
Analysis the risk factors of NGT, PDM, T2DM by ordered Multi-Classi cation Logistic Results NGT, PDM, T2DM were be regarded as dependent variables, and age, education, marriage, occupation, sports, alcohol, diet, BMI, hypertension and other factors were be regarded as risk factors, we utilized an orderly multi-class logistic regression analysis to nd the risk factors of PDM and T2DM. Smoking and BMI have an impact on the development of diabetes. Adjusting factors such as age, education, marriage, exercise, alcohol, diet, hypertension, and other factors show that smoking also affects the development of diabetes (P <0.001, OR (95% CI) = 1.161 (1.113 1.212)), BMI were in uential factors for the progression of diabetes too(P <0.05, OR overweight (95% CI) = 1.427 (1.388 ~ 1.468), OR obesity (95% CI) = 1.829 (1.753 ~ 1.908)) ( Table 3).
Unconditional Logistic Regression results between NGT-PDM and NGT-T2DM Table 4 showed the results of unconditional logistic regression between NGT-PDM and NGT-T2DM. In the NGT-PDM group, adjusting for age, education and other confounding factors showed that smoking, obesity, and overweight were risk factors for PDM (P < 0.001), in the NGT-T2DM group, adjusting for mixed factors such as age and education showed that smoking, obesity and overweight were risk factors for T2DM (P <0.001). By comparing the two models, we found that smoking and obesity affect the development both PDM and T2DM. Table 4 showed that overweight and obesity were correlated with PDM and T2DM, and the risk had a dose-response relationship. While underweight was a protective factor of PDM in the NGT-PDM, but it was insigni cance by comparison NGT-T2DM groups. Therefore, overweight and obesity were considered as risk group, and normal and underweight were considered as control group for additive interaction and multiplicative interaction model analysis. Table 5 showed the multiplicative and additive interactions effects of obesity and smoking on PDM after strati cation. The risk of PDM among overweight/ obesity and smokers was 2.262(2.091~2.448) times than that of non-overweight/non obesity and non-smokers. The results showed that overweight/ obese and smoking increased the risk of PDM than those exposed single risk factor alone. However, we did not nd an multiplicative and additive interactions between overweight/ obesity and smoking on PDM.

Results of multiplicative and additive interactions between NGT-PDM and NGT-T2DM
To found multiplicative and additive interactions of overweight/ obesity and smoking to develop T2DM, the risk of T2DM among overweight/obesity and smokers was 2.2(2.036~2.377) times than that of non-overweight /non obesity and non-smokers. We found that subjects who were overweight/ obesity and smoking had greater risk to develop T2DM than those exposed to a single risk factor alone.
However, there was no multiplicative interaction between smoking and overweight/obesity and T2DM, but an additive interaction was existence ( Table 6).

Discussion
This study found that smoking, overweight and obesity were independent factors of PDM and T2DM. Moreover, overweight/obesity and smoking might affect the development of T2DM. These results were consistent with other previous studies. For example [13], the Strong Heart Study conducted among 1677 American Indians found that obesity signi cantly increased the risk for T2DM in those with PDM by 2.7 times after 7.8 years of follow-up. A systematic review of 25 prospective cohort studies (1.2 million participants) showed that smoking was associated with 44% increased risk of developing T2DM (relative risk [RR] 1.44, 95%CI 1.31-1.58) [14].However, the interaction between obesity and smoking on incident of T2DM is still unclear, so we studied the interaction between overweight obesity and smoking on PDM and on T2DM.
In this study, we did not nd an interaction between overweight/obesity and smoking on PDM, but the risk of PDM among overweight and obese smokers was 2.262 times than that of non-overweight and obese non-smokers. Previous studies had shown that smoking increases blood glucose concentration after oral glucose tolerance [15] and may impair insulin sensitivity [16].However, smoking is associated with higher energy expenditure and decreased appetite, which might cause smokers to lose weight and gain weight after quitting smoking [17].But, smoking and body mass index interact to in uence the diabetes, and the interactions are complex. A follow-up study in Japan suggests that, light smoking reduced the risk for T2DM in lean men [18].Furthermore,a prospective cohort study conducted among 3,598 Chinese found that although there was a signi cant interaction between smoking and abdominal obesity in patients with T2DM, even no signi cant interaction was found between smoking and overall obesity, but T2DM was also associated with a higher incidence among overall obese smokers [19].
In this study, there was an additive interaction between smoking, overweight obesity and T2DM, and the interaction between smoking and overweight obesity accounted for 9.1% of the occurrence of diabetes. Furthermore, overweight and obese smokers was 2.2 times to develop T2DM than non-overweight /non obesity and non-smokers. Multiple biological mechanisms could explain the link between overweight obesity, smoking and diabetes.
First, smoking affects the neuroendocrine system. Smoking and nicotine directly act on the surrounding tissues (mainly mediated by catecholamines) and indirectly affect the neuroendocrine circuit in the central nervous system [20], reducing food intake by inhibiting the signal of hypothalamus appetite, and increasing energy consumption, thus reducing body weight. However, smoking increases the risk of central obesity by increasing the 2-hydroxylation of estradiol or by inducing an imbalance of androgen to estrogen activity in smokers [21].Central obesity in the form of abdominal fat accumulation is closely related to insulin resistance and diabetes [22].Smoking is associated with increased levels of anti-regulatory hormones and increased sympathetic activity, which may be the cause of impaired insulin sensitivity caused by smoking [23].
Secondly, the insulin resistance of most obese patients [24] is related to the signi cantly increased level of free fatty acids in the blood [25].
Smoking aggravates the insulin resistance of obese patients by increasing free fatty acids. Smoking has been shown to be associated with insulin resistance in non-diabetic [26] and type 2 diabetic [27], with long-term smokers having insulin resistance, hyperinsulinemia and dyslipidemia. Nicotine promotes adipobreakdown and transports free fatty acids to the liver and skeletal muscles, which are associated with the secretion of very low density lipoprotein in the liver, lipid saturation in muscle cells, and peripheral insulin resistance [28].
Third, both smoking and obesity might affect mitochondrial function. Smoking increases oxidative stress and in ammation, thereby impairs endothelial function, leading to insulin resistance and diabetes [29]. Smoking is associated with carbon monoxide exposure [30]. It had been reported that carbon monoxide exposure increases oxidative stress, leading to impaired mitochondrial function, in ammation, and endothelial function. A series of studies had found that in the obesity-induced IR population, the mRNA level and protein content of mitochondrial genes, the size and number of mitochondria, and the activity of oxidase [31][32] were all lower than those in the normal control group.
Fourth, in ammation plays a role in the development of T2DM. Prospective nested case-control studies [33] showed that the baseline levels of IL-6 and CRP in DM cases were signi cantly higher than those in the control group, and the elevated levels of CRP and IL-6 predicted the development of T2DM. Adipose tissue produces about 25% of systemic il-6 in the body [34]. In ammatory properties of IL-6 include stimulating the liver to produce acute phase proteins [35,36]. The release of IL-6 from adipose tissue may lead to low-level systemic in ammation in people with excessive body fat. Another sensitive marker of systemic in ammation is acute c-reactive protein (CRP). A higher BMI [37] was found to be associated with a higher CRP concentration, even in young people aged 17-39 years, suggesting a low-grade systemic in ammatory state in overweight and obese people. Studies had shown [38] that current smokers had signi cantly higher CRP levels (2.53 vs 1.35 mg/L) than those who had never smoked. The double effects of overweight and obesity and smoking increase chronic in ammatory responses then led to PDM or T2DM.

Advantage:
The mechanisms by which overweight, obesity and smoking contribute to diabetes are still being investigated. It has been documented that smoking is a risk factor for diabetes [39,40], and overweight and obesity may affect the occurrence of diabetes [41].However, there is few literatures on the in uence of overweight/obesity and smoking on the progression of the disease, there is few reports on whether overweight/obesity and smoking are risk factors for PDM. The combined effects of overweight and smoking on pre-diabetes and T2DM have been less well reported. The main advantage of this study is a large sample case-control study design. In this study, we aimed to explore the in uence of overweight, obesity and smoking on incident T2DM,and the independent and comprehensive effects on PDM and T2DM respectively, so as to provide reference for early controlling diabetes risk factors.

Limitations
The limitation of this study is that the smoking status is collected through self-report of the respondents, and there may be smokers who do not admit that they smoke. Secondly, fewer women are smokers, so the interaction between smoking and overweight and obesity among female smokers might be veri ed by expanding the sample. Third, smokers, including former and current smokers, have been reported to reduce the risk of diabetes after quitting smoking. Fourth, smoking intensity was not recorded. Fifth, the selection of The Pearl River Delta region in Guangdong has certain limitations, which might expand the sample to increase the representation of other Cities and regions in China.

Conclusions
The main ndings of this study is that the overweight/obesity and smoking are important in uence factors of PDM and T2DM, which affect the degree of disease development. It also found an interaction between overweight/obesity and smoking for T2DM.It is suggested that early weight control and active control is helpful to prevent and delay the onset of T2DM. Overweight and obese people should control their weight in a reasonable range. Smokers take quitting smoking as an important way to improve their lifestyle. In other words, people with smoking habits and overweight and obese should not only stop smoking and control their weight, but also reduce the inhalation of second-hand smoke, so as to reduce the occurrence of diabetes. Note: a: represents the differences between the three groups of NGT, PDM, and T2DM, b: represents the differences between the two groups of NGT,PDM, and c: represents the differences between the two groups of NGT, T2DM