Coronary artery disease has a multifactorial origin, including hereditary and acquired risk factors which may be the direct cause of the disease or merely associated with it(Reference Grundy, Cleeman, Merz, Brewer, Clark, Hunninghake, Pasternak, Smith and Stone1, 2). Disorders in lipid metabolism such as high levels of plasma or serum total cholesterol (TC)(3), TAG(Reference Satoh, Nishino, Tomita and Tsutsui4), LDL cholesterol (LDL-C)(Reference Achari and Thakur5) and apo B(Reference Kwiterovich, Coresh, Smith, Bachorik, Derby and Pearson6), and low levels of HDL cholesterol (HDL-C)(Reference Boden7), play a relevant role in the progression of atherosclerosis(Reference Sharrett, Ballantyne, Coady, Heiss, Sorlie, Catellier and Patsch8); thus the laboratory assessment of lipids is of fundamental importance to diagnose, treat and prevent this condition(9). Although a great deal of research effort has been focused on the determinants of these lipid phenotypes in different ethnic groups and diseases(Reference Bermudez, Velez-Carrasco, Schaefer and Tucker10–Reference Nethononda, Essop, Mbewu and Galpin12), very few data are available about the prevalence of hyperlipidaemia and its risk factors in isolated minority groups.
There are fifty-six ethnic groups in China. Han is the largest group and Yao is the eleventh largest minority among the fifty-five minority groups according to population size. Bai Ku Yao (White-trouser Yao), an isolated subgroup of the Yao minority, is so named because all of the men wear white knee-length knickerbockers. The population size is about 30 000. Because of their isolation from other ethnic groups, special customs and cultures including their clothing, intra-ethnic marriages, dietary habits, and corn wine and rum intakes are still completely preserved to the present day. Little is known about the prevalence of hyperlipidaemia and its risk factors in this population. Therefore, the present study was undertaken to compare the differences in hyperlipidaemia prevalence and its risk factors between the Guangxi Bai Ku Yao and Han populations.
Methods
Subjects
The subjects of Bai Ku Yao residing in Lihu and Baxu villages in Nandan County, Guangxi Zhuang Autonomous Region, People’s Republic of China, were surveyed by a stratified randomized cluster sampling. First, teams (resident communities; inhabitants) were selected randomly from the two villages. Then, sexes in each team were separated for the survey. Finally, the sampled resident sections were determined from the local population registers. A total of 1230 subjects from sixteen teams were asked to participate in the study and 1193 subjects actually participated. The response rate was 97·0 %. Twenty-three persons (1·9 %) with a history or evidence of other diseases were excluded from the analysis. A total of 1170 subjects were included in the present study. The ages of the subjects ranged from 15 to 85 years, with an average age of 37·5 (sd 16·7) years. There were 588 males (50·3 %) and 582 females (49·7 %). All subjects were peasants. The age distribution was as follows: <20 years, 185 people (15·8 %); 20–29 years, 261 people (22·3 %); 30–39 years, 239 people (20·4 %); 40–49 years, 210 people (18·0 %); 50–59 years, 139 people (11·9 %); 60–69 years, eighty-three people (7·1 %); ≥70 years, fifty-three people (4·5 %). The subjects accounted for 3·9 % of the total Bai Ku Yao population.
During the same period, a total of 1235 subjects of Han Chinese from fifteen teams were asked to participate in the study by the same method and 1205 subjects completed the survey. The response rate was 97·6 %. Thirty-two people (2·7 %) with a history or evidence of other diseases were excluded. A total of 1173 subjects were included in the present study. The mean age of the subjects was 38·3 (sd 16·8) years (range 15 to 89 years). There were 589 men (50·2 %) and 584 women (49·8 %). All of them were also peasants. The age distribution was as follows: <20 years, 180 persons (15·3 %); 20–29 years, 241 persons (20·5 %); 30–39 years, 251 persons (21·4 %); 40–49 years, 208 persons (17·7 %); 50–59 years, 139 persons (11·8 %); 60–69 years, eighty-five persons (7·2 %); ≥70 years, sixty-nine persons (5·9 %).
All study subjects were essentially healthy and had no evidence of any chronic illness, including hepatic, renal or thyroid disease, heart attack or myocardial infarction, stroke, congestive heart failure or diabetes mellitus. They were not taking medications known to affect serum lipid levels (lipid-lowering drugs such as statins or fibrates, beta-blockers, diuretics or hormones). The present study was approved by the Ethics Committee of the First Affiliated Hospital, Guangxi Medical University. Informed consent was obtained from all subjects after they received a full explanation of the study.
Epidemiological survey
The survey was carried out using internationally standardized methods, following a common protocol(13). Information on demographics (age, gender, residential area), socio-economic status (education level achieved, marital status, annual household income), cigarette smoking, alcohol consumption and physical activity was collected with standardized questionnaires. The 24 h dietary recall method was used to determine the dietary intakes of each subject(Reference Lyu, Yeh, Lichtenstein, Li, Ordovas and Schaefer14). Detailed descriptions of all foods, beverages and supplements consumed during the 24 h period before the interview, including the quantity, cooking method and brand names, were recorded by a chief physician. The interviewer used food models and pictures depicting portion sizes and followed a standardized protocol for determining the weight of food consumed. The intakes of macronutrients from the ingredients were determined by using the 2002 Chinese Food Composition Table (Reference Yang, Wang and Pan15). Although a 24 h dietary recall may be inaccurate when diets are highly variable, the Bai Ku Yao diet is consistent throughout the year and among individuals because of the Bai Ku Yao’s reliance on a limited number of locally available food items. Overall physical activity was ascertained with the use of a modified version of the Harvard Alumni Physical Activity Questionnaire(Reference Paffenbarger, Wing and Hyde16), which included questions about the number of hours per day (mean of a regular weekday and a regular weekend day) spent sleeping and in sedentary, light, moderate and vigorous activities; the interviewer ensured that the total time added up to 24 h. The alcohol information included questions about the number of grams of rice wine, wine, beer or liquor consumed during the preceding 12 months. At the physical examination, several parameters including height, weight and waist circumference were measured. Sitting blood pressure was measured three times with the use of a mercury sphygmomanometer after a rest of at least 15 min, and the average of the three measurements was used for the level of blood pressure. Systolic blood pressure was determined by the first Korotkoff sound and diastolic blood pressure by the fifth Korotkoff sound. Body weight, to the nearest 50 g, was measured using a portable balance scale. Subjects were weighed without shoes and in a minimum of clothing. Height was measured, to the nearest 0·5 cm, using a portable steel measuring device. From these two measurements BMI (kg/m2) was calculated. Waist circumference was measured with a non-stretchable measuring tape, at the level of the smallest area of the waist, to the nearest 0·1 cm.
Measurements of lipids and apolipoproteins
A venous blood sample was drawn from an antecubital vein in all subjects after an overnight fast. The blood was transferred into glass tubes and allowed to clot at room temperature. Immediately following clotting serum was separated by centrifugation for 15 min at 3000 rpm. The levels of TC, TAG, HDL-C and LDL-C in samples were determined enzymatically using commercially available kits: Tcho-1, TG-LH (RANDOX Laboratories Ltd, Crumlin, UK), Cholestest N HDL and Cholestest LDL (Daiichi Pure Chemicals Co., Ltd, Tokyo, Japan), respectively. Serum apo A1 and apo B levels were measured by an immunoturbidimetric assay (RANDOX Laboratories Ltd). All determinations were performed with an autoanalyser (type 7170A; Hitachi Ltd, Tokyo, Japan) in the Clinical Science Experiment Center of the First Affiliated Hospital, Guangxi Medical University(Reference Yin, Chen and Pan17, Reference Yin, Wang, Chen, Lin, Yang and Pan18).
Diagnostic criteria
The normal values of serum TC (mmol/l), TAG (mmol/l), HDL-C (mmol/l), LDL-C (mmol/l), apo A1 (g/l), apo B (g/l) and the ratio apo A1:apo B in the Clinical Science Experiment Center were 3·10–5·17, 0·56–1·70, 0·91–1·81, 2·70–3·20, 1·00–1·78, 0·63–1·14 and 1·00–2·50, respectively. The individuals with TC > 5·17 mmol/l and/or TAG > 1·70 mmol/l were defined as hyperlipidaemic(Reference Yin, Chen and Pan17, Reference Yin, Wang, Chen, Lin, Yang and Pan18). Hypertension was diagnosed according to the criteria of the 1999 WHO/International Society of Hypertension guidelines for the management of hypertension(Reference Ruixing, Limei, Yuming, Dezhai, Weixiong, Muyan, Fengping, Jinzhen, Guangqing and Zhenbiao19, Reference Ruixing, Jiaqiang, Dezhai, Weixiong, Shangling, Jinzhen, Jiandong and Xiuyan20). The diagnostic criteria of overweight and obesity were according to the Cooperative Meta-analysis Group of China Obesity Task Force. Normal weight, overweight and obesity were defined as BMI < 24, 24–28 and >28 kg/m2, respectively(21).
Statistical analysis
The data were organized and analysed using Excel XP (Microsoft, Seattle, WA, USA) and the SPSS for Windows statistical software package version 11·5·0 (SPSS Inc., Chicago, IL, USA). Means and standard deviations as well as frequency distributions of participant characteristics were calculated. The difference in mean values of parameters between Bai Ku Yao and Han was tested by Student’s unpaired t test. The difference in percentage values was tested by the χ 2 test. The influences of sex, age, physical activity, BMI, waist circumference, alcohol consumption and cigarette smoking were adjusted for in the statistical analyses. In order to evaluate the risk factors for hyperlipidaemia, unconditional logistic regression analysis was also performed in the combined population of Bai Ku Yao and Han, the Bai Ku Yao and the Han, respectively. The backward multiple logistic regression method was used to select the risk factors significantly associated with hyperlipidaemia. Total intake of each nutrient was summed over all foods consumed. The Matlab 5·0 software (The MathWorks, Inc., Natick, MA, USA) was used for processing these procedures by the method of multiplication of matrix(Reference von Kamp and Schuster22). A P value of less than 0·05 was considered statistically significant.
Results
Comparison of general characteristics between Bai Ku Yao and Han
The demographic, dietary and other lifestyle characteristics of Bai Ku Yao and Han are shown in Table 1. The level of physical activity and the intakes of carbohydrate, vegetal protein and total dietary fibre were higher in Bai Ku Yao than in Han (P < 0·001 for all), whereas educational level, height, weight, BMI, waist circumference, blood pressure levels including systolic, diastolic and pulse pressure, hypertension and the intakes of total energy, total fat, total protein, dietary cholesterol and salt were higher in Han than in Bai Ku Yao (P < 0·05 to 0·001). In addition, there were also significant differences in the staple and subsidiary foods and drinks between the two ethnic groups. For the great majority of Bai Ku Yao people, corn (gruel or tortillas) was the staple food and rice, soya, buckwheat, sweet potato and pumpkin products were the subsidiary foods all the year round. Approximately 90 % of the beverages comprised corn wine and rum that they brewed themselves, the alcohol content is about 15 % (v/v). The subjects of Bai Ku Yao were also accustomed to drinking hempseed soup. In contrast, rice was the staple food and corn, broomcorn, potato and taro products were the subsidiary foods in Han Chinese. About 90 % of the beverages comprised rice wine with alcohol content of about 30 % (v/v). There were no significant differences in the age structure, the percentages of subjects who consumed alcohol or smoked cigarettes, or the ratio of men to women between the two ethnic groups (P > 0·05).
Comparison of demographic, dietary and lifestyle characteristics between the Bai Ku Yao and Han populations
Prevalence of hyperlipidaemia in Bai Ku Yao and Han
As shown in Table 2, the prevalence of high TC (hypercholesterolaemia), LDL-C, apo B and apo A1:apo B ratio was lower in Bai Ku Yao than in Han (P < 0·05 to 0·001), whereas the prevalence of low HDL-C and apo A1 was higher in Bai Ku Yao than in Han (P < 0·001 for each). There was no significant difference in the prevalence of high TAG (hypertriacylglycerolaemia) between the two ethnic groups (P > 0·05). In the subjects with hypercholesterolaemia and hypertriacylglycerolaemia, there were thirty-six subjects in Bai Ku Yao (3·1 %) and eighty-three subjects in Han (7·1 %; χ 2 = 19·430, P < 0·001) with both high TC and TAG. Thus, the prevalence of hyperlipidaemia in Bai Ku Yao and Han was 24·4 % (285/1170) v. 33·9 % (398/1173; χ 2 = 25·981, P < 0·001).
Associations of demographic, dietary and lifestyle characteristics with the prevalence of hyperlipidaemia in the Bai Ku Yao and Han populations
TC, total cholesterol; HDL-C, HDL cholesterol; LDL-C, LDL cholesterol.
*P < 0·05, **P < 0·01, ***P < 0·001 for comparison with men, BMI ≤ 24 kg/m2, normotensive, non-drinkers or non-smokers of the same ethnic group.
†P < 0·05, ††P < 0·01, †††P < 0·001 for comparison with the same subgroup of Bai Ku Yao.
‡Values for the comparison of <20, 20–29 and 30–39 years age subgroups in Bai Ku Yao.
§Values for the comparison of <20 and 30–39 years age subgroups in Han Chinese.
Associations between general characteristics and hyperlipidaemia in Bai Ku Yao and Han
The associations of sex, BMI, hypertension, alcohol consumption, cigarette smoking and age with hyperlipidaemia in Bai Ku Yao and Han are also shown in Table 2. The prevalence of high TAG in both ethnic groups, high apo A1:apo B ratio in Bai Ku Yao, and low HDL-C and apo A1 in Bai Ku Yao was higher in males than females. The prevalence of high TC, TAG, LDL-C and apo B in both ethnic groups was higher in subjects with BMI > 24 kg/m2 than those with BMI ≤ 24 kg/m2. The prevalence of high TC, TAG and apo B in both ethnic groups was higher in drinkers than in non-drinkers. The prevalence of high apo B and apo A1:apo B ratio in Bai Ku Yao and high TAG in Han was higher in smokers than in non-smokers. There was also a significant difference in the prevalence of high TC, LDL-C and apo B, and of low HDL-C and apo A1 between age subgroups in both ethnic groups.
Risk factors for hyperlipidaemia in Bai Ku Yao and Han
The results of multiple logistic regression analysis are shown in Table 3. Hyperlipidaemia was positively correlated with BMI, waist circumference, total energy and total fat intakes, and negatively associated with physical activity and total dietary fibre intake, in Bai Ku Yao (P < 0·05 to 0·001). Hyperlipidaemia was positively associated with age, alcohol consumption, BMI, waist circumference, total energy and total fat intakes, and inversely correlated with physical activity and total dietary fibre intake, in Han (P < 0·05 to 0·001). There was no significant correlation between hyperlipidaemia and sex, education level, dietary cholesterol, cigarette smoking and hypertension in both ethnic groups (P > 0·05).
Comparison of the risk factors for hyperlipidaemia between the Bai Ku Yao and Han populations
For the multiple logistic regression analysis, the data were recorded as follows: ethnic group: Bai Ku Yao = 0, Han = 1; sex: female = 0, male = 1; age (years): <20 = 1, 20–29 = 2, 30–39 = 3, 40–49 = 4, 50–59 = 5, 60–69 = 6, ≥70 = 7; BMI (kg/m2): ≤24 = 0, >24 = 1; blood pressure: normotensive = 0, hypertensive = 1; alcohol consumption (g/d): non-drinkers = 0, <25 = 1, 25–49 = 2, 50–99 = 3, ≥100 = 4; cigarette smoking (cigarettes/d): non-smokers = 0, <10 = 1, 10–19 = 2, 20–39 = 3, ≥40 = 4.
Comparison of hyperlipidaemia between Lihu and Baxu villages
The prevalence of hypertriacylglycerolaemia, low HDL-C and apo A1 in Bai Ku Yao was higher but the prevalence of hypercholesterolaemia was lower in subjects from Lihu village than from Baxu villages (P < 0·05 to 0·001). There was no significant difference in the prevalence of high apo B, apo A1:apo B ratio and hyperlipidaemia between the two villages (P > 0·05, Table 4).
Comparison of hyperlipidaemia in Bai Ku Yao between Lihu and Baxu villages
TC, total cholesterol; HDL-C, HDL cholesterol; LDL-C, LDL cholesterol; hyperlipidaemia, TC > 5·17 mmol/l and/or TAG > ·70 mmol/l.
Discussion
The present study shows that the prevalence of hypercholesterolaemia and hyperlipidaemia was lower in Bai Ku Yao than in Han. This discrepancy may mainly be attributed to the differences in dietary habits and lifestyle choices between the two ethnic groups. The staple and subsidiary foods are more favourable for lipid profiles in Bai Ku Yao than in Han. Corn contains abundant dietary fibre and high-quality plant protein(Reference Dong, Ma, Zhang and Yu23). Consumption of dietary fibre, specifically the soluble type such a pectins and guar gum, can result in a decrease of serum cholesterol levels in healthy and hyperlipidaemic subjects(Reference Jenkins, Kendall, Axelsen, Augustin and Vusksan24, Reference Lairon25). Plant protein can promote the transportation and excretion of free cholesterol. Corn oil is rich in PUFA and MUFA(Reference Liu, Zhang and Wu26). Suitable intakes of PUFA and MUFA can lower serum levels of cholesterol and LDL-C(Reference Liu, Zhang and Wu26, Reference Zhang, Zhou, Wu and Zhang27). Soya protein intake is effective in reducing TC by 9·3 %, LDL-C by 12·9 % and TAG by 10·5 %, and in increasing HDL-C by 2·4 %. Furthermore, the changes in serum TC and LDL-C concentrations are directly related to the initial serum TC concentration(Reference Anderson, Johnstone and Cook-Newell28). The hypocholesterolaemic activity of buckwheat protein products is far stronger than that of soya protein isolate and corn(Reference Zhang, Zhang, Lu, Tong and Cao29). Daily ingestion of 4 g caiapo (the extract of white-skinned sweet potato) for 6 weeks reduces TC and LDL-C in type 2 diabetic patients previously treated by diet alone(Reference Ludvik, Mahdjoobian, Waldhaeusl, Hofer, Prager, Kautzky-Willer and Pacini30). The long-term beneficial effects of caiapo on plasma cholesterol levels are further confirmed in a recent report(Reference Ludvik, Neuffer and Pacini31). Studies have also demonstrated that pumpkin or pumpkinseed oil may be a useful therapy for hypercholesterolaemia by reducing oxidative stress and cholesterol levels(Reference al-Zuhair, Abd el-Fattah and Abd el Latif32). The people of Bai Ku Yao are accustomed to drinking hempseed soup. There are more than twenty-nine fat-soluble constituents in hempseed, with fatty acid methyl esters making up 99·3 %(Reference Zhang and Wang33, Reference Ross, Mehmedic, Murphy and Elsohly34). The main fatty acid components of hempseed oil are palmitic acid (8·4 %), γ-linolenic acid (1·3 %), linoleic acid (58·7 %), linolenic acid (14·0 %), oleic acid (10·1 %), stearic acid (3·8 %) and arachidic acid (1·0 %)(Reference Zhou35). A number of experimental and clinical studies have demonstrated that the beneficial effects of hempseed or hempseed oil on lipid profiles include decreasing the levels of TC, TAG and LDL-C, inhibiting lipid peroxidation, reducing the atherogenic index and increasing the levels of HDL-C(Reference Ren, Sun, Ma, Zhang, Yi, Wu, Liu and Li36–Reference Schwab, Callaway, Erkkila, Gynther, Uusitupa and Jarvinen38). The intake of animal fat, body weight and BMI were all lower in Bai Ku Yao than in Han. For nearly 50 years it has been widely accepted that high-fat diets, particularly those that contain large quantities of SFA, raise blood cholesterol concentrations and predispose individuals to CVD(Reference Yu-Poth, Zhao, Etherton, Naglak, Jonnalagadda and Kris-Etherton39). In addition, strict intra-ethnic marriages have been performed from time immemorial in the Bai Ku Yao. Therefore, the hereditary characteristics, genotypes and phenotypes of lipids in the Bai Ku Yao population may be different from those in Han Chinese. However, this still needs to be determined.
It is well documented that androgens induce disadvantageous effects on lipid profiles, whereas oestrogens are held to have opposite effects(Reference Mudali, Dobs, Ding, Cauley, Szklo and Golden40, Reference Cheung41). In the present study, we also showed that the prevalence of high TAG in both ethnic groups and low HDL-C and apo A1 in Bai Ku Yao was higher in males than in females. These discrepancies in dyslipidaemia between men and women might be partially attributable to the effects of gonadal hormones.
Epidemiological studies have provided abundant evidence that lipid levels are closely related with age(Reference Karki, Neopane, Pradhan and Magar42, Reference Guize, Benetos, Thomas, Malmejac and Ducimetiere43). In the present study, we showed that there was a significant difference in the prevalence of high TC, LDL-C and apo B, and low HDL-C and apo A1, for the age subgroups in both ethnic groups. This is consistent with previous studies. The exact mechanisms of age on hyperlipidaemia are not well known. They may due, in part, to the hereditary characteristics and ageing of the population.
The link between obesity and dyslipidaemia has been clearly documented(Reference Li, Yu, Zhou, Zhang, Yao and Sinclair44, Reference Mataix, Lopez-Frias, Martinez-de-Victoria, Lopez-Jurado, Aranda and Llopis45). Obesity not only induces dyslipidaemia, but also is directly associated with diabetes, hypertension and coronary artery disease. The current study also shows that the prevalence of high TC, TAG, LDL-C and apo B in both ethnic groups was higher in subjects with BMI > 24 kg/m2 than those with BMI ≤ 24 kg/m2. This is in agreement with the findings of previous studies(Reference Li, Yu, Zhou, Zhang, Yao and Sinclair44–Reference Kawada46). Dyslipidaemia in obesity may result from insulin resistance(Reference Houston, Basile and Bestermann47, Reference Pei, Kuo, Wu, Lin, Hseih, Lee, Hsu, Chen, Sheu and Li48). Insulin resistance can increase plasma NEFA concentrations and stimulate VLDL synthesis and release. At the same time, insulin resistance can also suppress lipoprotein lipase activity and increase plasma VLDL levels.
The association of hyperlipidaemia with hypertension is poorly understood. Studies have shown that the levels of TAG, VLDL cholesterol and apo E are higher and levels of apo AI, apo AII and apo CII are lower in untreated hypertensives than in controls(Reference Catalano, Aronica, Carzaniga, Seregni and Libretti49). TC and non-HDL-C levels increase significantly with increasing systolic or diastolic blood pressure in both sexes(Reference Bonaa and Thelle50). Some authors believe this relationship may be a kind of random phenomenon(Reference Li, Niu, Li, Wang and Zhao51). In the present study, we found that the prevalence of hypercholesterolaemia in both ethnic groups and hypertriacylglycerolaemia in Han was higher in hypertensives than in normotensives. These findings suggest that there may be a complicated interrelation between hyperlipidaemia and hypertension(Reference Borghi, Dormi, Veronesi, Sangiorgi and Gaddi52).
Both case–control and cohort studies have described a J- or U-shaped association between alcohol intake and CHD and between alcohol intake and mortality(Reference Corella, Tucker, Lahoz, Coltell, Cupples, Wilson, Schaefer and Ordovas53). Although moderate alcohol consumption appears to be protective, heavy consumption of alcohol is associated with subclinical impairment of left ventricular function and occasionally results in overt cardiomyopathy. High alcohol intake and cigarette smoking have disadvantageous effects on lipid profiles. Smoking also disturbs lipoprotein metabolism by raising insulin resistance and lipid intolerance, and is implicated in the production of small dense LDL-C. In the present study, we showed that the prevalence of high TC, TAG and apo B in both ethnic groups was higher in drinkers than in non-drinkers. The prevalence of high apo B and apo A1:apo B ratio in Bai Ku Yao and high TAG in Han was higher in smokers than in non-smokers. These findings indicate that both alcohol consumption and cigarette smoking may be risk factors for hyperlipidaemia in the two populations.
In the present study, we also showed that the prevalence of high TAG, low HDL-C and apo A1 in Bai Ku Yao was higher, but the prevalence of high TC was lower, in subjects from Lihu village than from Baxu village. The reason for this discrepancy is not yet known. The customs and cultures in both villages are similar. One difference is the residential locus. The distance between the two villages is about 50 km. It is unclear whether geographical or other unknown factors might be involved in this discrepancy between the two villages.
Conclusion
The present study reveals that the prevalence of hypercholesterolaemia and hyperlipidaemia was lower in Bai Ku Yao than in Han. These differences between the two ethnic groups may result from different demographic characteristics, dietary patterns, lifestyle choices and physical activity level, as well as genetic background.
Acknowledgements
Conflict of interest: No conflicts.
Funding support: The study was supported by the National Natural Science Foundation of China (No. 30660061).
Authorship responsibilities: Y.R. conceived the study hypothesis, participated in the design, carried out the epidemiological survey, collected the samples, supervised data analyses and drafted the manuscript; Y.D., L.S., P.S. and L.W. carried out the epidemiological survey and biochemical analyses; C.Y. carried out the epidemiological survey and helped to carry out biochemical analyses; F.Q. performed the statistical analyses; Y.H., T.J. and L.Y. carried out the epidemiological survey. All authors read and approved the final manuscript.
