点击显示 收起
1 From the National Institute of Public Health, Copenhagen, Denmark (MEJ and PB), and the Steno Diabetes Centre, Gentofte, Denmark (MEJ and KB-J)
2 The Greenland population study was funded by the Danish Medical Research Council, The Greenland Medical Research Council, The Commission for Scientific Research in Greenland, and the Karen Elise Jensen Foundation. 3 Reprints not available. Address correspondence to ME Jørgensen, Centre for Health Research in Greenland, National Institute of Public Health, Øster Farimagsgade 5A2, 1399 Copenhagen, Denmark. E-mail: mej{at}niph.dk.
ABSTRACT
Background: The association between obesity and cardiovascular disease risk differs across populations. Whether such differences in obesity-related risk factors exist within population groups of the same genetic origin but with differences in lifestyle remains to be determined.
Objective: The aim was to analyze whether obesity was associated with the same degree of metabolic disturbances in 2 groups of genetically homogeneous Inuit who were exposed to considerable differences in lifestyle.
Design: We studied obesity and cardiovascular disease risk factors in a cross-sectional population survey of 2311 Inuit living in Denmark (n = 995) or Greenland (n = 1316). The participants received an oral-glucose-tolerance test. Blood tests were supplemented by structured interviews and anthropometric and blood pressure measurements.
Results: The trend in the association between obesity and metabolic effects was not significantly different in the Inuit populations, but the values of several risk factors were significantly different. At any given level of obesity, Inuit residents in Greenland had lower blood pressure and lower concentrations of triacylglycerol and postchallenge plasma glucose and insulin than did the Inuit migrants in Denmark. The trend in the association with obesity categories was different only for HDL cholesterol, with higher concentrations observed in women Inuit migrants in Denmark than in women Inuit residents in Greenland.
Conclusions: The health risk associated with obesity clearly varies within groups of Inuit living in Greenland and Inuit migrants living in Denmark. The findings indicate that lifestyle factors modify the cardiovascular disease risk associated with obesity.
Key Words: Inuit • obesity • anthropometric measurements • metabolic syndrome • Greenland • migration
INTRODUCTION
Increasing evidence exists that the association of body mass index (BMI), percentage body fat, and body fat distribution (1-3) with obesity-associated cardiovascular disease risk differs across populations (4-8). Asian populations were found to be at a higher risk of diabetes and cardiovascular disease than were Europeans for a given level of BMI and waist circumference (9). In contrast, given levels of BMIs, waist-to-hip ratios (WHRs), and waist circumferences (WCs) were associated with lower blood pressure and lower concentrations of triacylglycerol and postchallenge plasma glucose and serum insulin and with higher concentrations of HDL cholesterol in Inuit populations of Greenland and Canada than in white populations (10, 11).
The associations between BMI and comorbidities may not be stable within populations over time, and it remains to be determined whether the differences in the obesity-related risk of cardiovascular disease are due to lifestyle, environmental, or genetic factors. The overall purpose of the present study was to analyze the effect of lifestyle factors on the association between obesity and cardiovascular disease risk factors with a comparison of Inuit migrants and the Inuit population of origin and to compare the data from these 2 groups with similar data from a European population. The specific aims were the following: 1) to analyze whether obesity was associated with the same degree of metabolic disturbances as reflected in blood pressure and concentrations of plasma glucose, insulin, HDL cholesterol, and triacylglycerols in 2 groups of Inuit exposed to considerable differences in lifestyle (Inuit migrants in Denmark and Inuit in Greenland) and to compare the data obtained from the Inuit groups with identical data from a previously published study conducted on Danes in Denmark; and 2) to assess whether differences, if any, in the obesity-associated risk of cardiovascular disease persisted after control for confounders with potential influence on the association between obesity and cardiovascular disease risk factors.
SUBJECTS AND METHODS
Participants
Data were collected from 1998 to 2002 in random samples of adult Inuit from Denmark and 3 selected areas in Greenland. The total population of Greenland is 56 000; of this population, 90% are ethnic Greenlanders (ie, Inuit). They are genetically, historically, culturally, and linguistically closely related to the Inuit and Yupik who live in Canada, Alaska, and Siberia (12). Approximately 7000 ethnic Greenlanders live in Denmark. This group of migrants is well integrated into Danish society and has a lifestyle similar to that of the general population of a Western industrialized country (13).
Until the 1950s, most Greenlanders made their living by small-scale hunting and fishing. Over the past decades, substantial changes have occurred in Greenland due to rapid modernization, especially in towns. However, hunting and fishing are still important leisure-time activities, and traditional Greenlandic food makes up 20–25% of the diet.
The study sample in Greenland comprised Greenlanders aged 18 y living in 3 areas of West Greenland: Nuuk (population: 14 000), Qasigiannguit (population: 1400), and 4 villages in the district of Uummannaq (population: 240–275 per village). In Nuuk, a random sample of the population was invited to participate, whereas in Qasigiannguit and Uummannaq everyone was invited. The details of this study have been described previously (14). In Denmark, Inuit migrants were initially identified through the Central Population Register as those persons born in Greenland but currently living in Denmark. A random sample of persons aged 18 y was drawn, and those with Greenlandic ancestry were invited to participate. In Denmark, only a subsample of participants aged 35 y was fasting and received an oral-glucose-tolerance test. Hence, the analyses of plasma glucose, insulin, and triacylglycerol in the combined dataset were restricted to this age group.
Genetic heritage was estimated from questions on the ethnicity of the 4 grandparents or, if this information was missing, of the parents. It was subsequently recoded as either full (ie, all grandparents were Greenlanders) or partial Inuit heritage. Only participants with full Inuit heritage were included in the study to exclude the potential influence of European genes on any observed differences. Informed consent was obtained in writing and orally from all participants. The relevant ethical review committees approved the study.
Data from a survey conducted in 6784 men and women in a Danish population-based study were included as reference (11). The study procedures in the 2 surveys were identical.
Interviews and questionnaires
The survey questionnaires were developed in Danish and subsequently translated into Greenlandic. The translation procedure included a translation by 2 interpreters followed by an independent back-translation into Danish and revision of the translation as needed. In Denmark, almost all information was obtained with the use of survey instruments in the Danish language, whereas in Greenland, almost all information was obtained in the Greenlandic language. Background information was obtained by mailed questionnaires or personal interviews.
Diet was recorded in a food-frequency questionnaire composed of 14 different traditional and imported food types; consumption of seal meat and fish combined and fresh fruit were included in these foods in the present analyses because they were found to be associated with cardiovascular disease risk factors in previous studies (15, 16). The frequency categories were the following: "daily," "4–6 times/wk," "1–3 times/wk," "2–3 times/mo," "1 time/mo," and "never"; these categories were recoded into weekly or none for seal and fish consumption and into daily or none for fresh fruit consumption. The food-frequency questionnaire has been used in several surveys in Greenland (14) and was validated with the use of biomarkers in one study (17). The frequency of alcohol consumption was reported. The participants were classified as current smokers, past smokers, or nonsmokers. The participants were asked about their physical activity levels within the past year. Physical activity was graded as sedentary, light, physical activity <4 h/wk, physical activity 4 h/wk, and heavy physical activity several times per week. Because of the small numbers of participants in each of the physical activity subgroups, physical activity was recoded on a 3-point scale: sedentary, moderate, and heavy. School education was recorded as the grade completed. The participants were asked whether a doctor had ever told them that they had diabetes.
Physical measurements
The participants underwent anthropometric measurements, blood sample tests, and a 75-g standardized oral-glucose-tolerance test. Weight and height were measured while the participants wore undergarments, and BMI was calculated as weight (in kg)/height2 (in m). WC was measured midway between the iliac crest and the costal margin and hip circumference was measured at its maximum while the participants were standing. Overall obesity was defined as a BMI 30; abdominal obesity was defined as WC >94 cm for men and >80 cm for women, according to the International Diabetes Federation criteria for Europeans (18). A large WHR was defined as >0.90 for men and >0.85 for women (19).
Three sitting blood pressures were measured by using a standard mercury sphygmomanometer with an appropriate cuff size after the participants rested 5 min. The mean of the 2 last blood pressures measured was used for the analyses.
The participants with known diabetes did not receive an oral-glucose-tolerance test, but their fasting venous plasma glucose concentration was measured. Plasma samples were immediately put on ice and separated by centrifugation for 10 min at 3000 x g at 4 °C within 30 min of sampling. Samples were stored frozen at –20 °C and shipped to Denmark for analyses. Plasma glucose was measured with the hexokinase G6P-DH method (Boehringer Mannheim, Mannheim, Germany). Serum insulin was analyzed with a flouroimmunoassay technique (code no. K6219; Dako Diagnostics Ltd, Ely, United Kingdom; measuring instrument: AutoDelfia; Perkin Elmer, Wallac, Finland). The interassay precision CV was <6%. The laboratory of the Steno Diabetes Center in Denmark performed the analyses of plasma glucose and serum insulin. Serum cholesterol was measured by using enzymatic calorimetric techniques (code no. K6219; Dako diagnosis, Boehringer Mannheim, Germany). Analyses were performed at the Department of Clinical Chemistry, Bispebjerg Hospital, University of Copenhagen, Denmark.
Statistics and data analysis
Analyses were performed with SAS version 8.2 (SAS institute, Cary, NC). Age-adjusted means and proportions of baseline characteristics were compared by population group with general linear models and chi-square tests, respectively. Age-adjusted means of insulin, plasma glucose, blood pressure, and lipids by category of BMI, WHR, and WC were calculated in general linear models. Fasting and 2-h insulin and triacylglycerol concentrations were log-transformed for analyses and transformed back to geometric means for presentation. The participants who were treated with antihypertensive or lipid-lowering medications were excluded from analyses of blood pressure and lipids, respectively. The participants with known diabetes were excluded from analyses that included plasma glucose and insulin concentrations, because of the possibility that lifestyle changes or glucose-lowering agents may influence the concentration of glucose and insulin as a function of obesity. Only data on the relation between BMI and the 8 risk factors for cardiovascular disease are shown in the figures, because BMI is the most commonly used anthropometrical measurement of obesity.
To formally test whether country of residence modified the association between obesity and metabolic risk factors for cardiovascular disease, we used an interaction term between country of residence and BMI, WHR, and WC. The following potential confounding factors were included as covariates in the linear models: diet, physical activity, smoking, amount of school education, and alcohol consumption.
RESULTS
The study population included in these analyses was composed of 2311 Greenlanders aged 18 y, 1316 of whom lived in Greenland and 995 of whom lived in Denmark. The participant rate was 67% in Greenland and 54% in Denmark (P < 0.001). The mean (±SD) age of the participants in Greenland was 43.4 ± 14.2 y compared with 44.2 ± 14.1 y for the nonparticipants; in Denmark, the mean age of the participants was 44.0 ± 12.1 y and that of the nonparticipants 43.9 ± 13.0 y. Men were underrepresented in both study populations: they made up 43% of the participants and 53% of the nonparticipants in Greenland (P = 0.01) and 27% of the participants and 30% of the nonparticipants in Denmark.
Mean values of metabolic characteristics for each population are shown in Table 1. A significant interaction between sex and country of residence was observed for most metabolic risk factors, which indicated that migration status influenced men and women differently. Hence, the analyses of levels of metabolic risk factors were run separately for men and women for those variables, and the P value for the interaction between sex and country of residence is given in Table 1. In the men, no significant difference in obesity measures was seen, whereas the mean values of fasting and 2-h insulin, blood pressure, and triacylglycerol were lower and HDL cholesterol was higher in the Greenland residents than in the migrants. Sex did not significantly modify the association between country of residence and glucose status; thus, the comparisons of glucose status for residents in Greenland and migrants were performed for men and women combined. However, the main association between country of residence and glucose status was not significant.
View this table:
TABLE 1. Metabolic characteristics of the 2 Inuit populations by sex1
Although the women from Greenland had the highest mean values of BMI, WHR, WC, and fasting insulin and the highest measures of large WHR and overall and abdominal obesity, the mean values of 2-h insulin, blood pressure, and triacylglycerol were lower than those in the women from Denmark. HDL-cholesterol concentrations were lowest in the women from Greenland.
The 2 population groups differed significantly with regard to some behavioral and socioeconomic factors (Table 2). Consumption of seal and fish was low in the migrants, whereas the consumption of fresh fruit was low in the participants from Greenland. Alcohol consumption was more frequent in the migrants. In contrast, there were fewer smokers in the migrant group than in the group of residents from Greenland. In the men, no significant difference in the distribution of physical activity categories was found, whereas in the women, the residents in Greenland were less physically active than the migrants. Finally, the length of school education received was generally shorter for the residents of Greenland than for the migrants. In Greenland, 45% of the population had attended school for 8 y compared with 23% of the migrants.
View this table:
TABLE 2. Distribution of behavioral and socioeconomic variables in Inuit from Greenland and Inuit migrants from Denmark1
To further investigate the differences in metabolic risk factors associated with obesity between the 2 population groups, the means for each risk factor were computed within categories of BMI. Migration status influenced fasting glucose, insulin, lipids, and systolic blood pressure associated with BMI differently in the men and women, and mean values are given for men and women separately. Fasting glucose was not significantly different in the 2 population groups (Figure 1). Both plasma glucose and insulin concentrations increased with increasing BMI in both Inuit groups, but, for each category of BMI, the migrant men had higher concentrations of 2-h plasma glucose and fasting- and 2-h insulin than did the men from Greenland (Figure 1 and Figure 2). In the women, a similar pattern was seen for 2-h glucose and 2-h insulin.
FIGURE 1.. Mean concentrations and 95% CIs of fasting and 2-h plasma glucose resulting from a general linear model with age included as a continuous variable by category of BMI in Inuit residents from Greenland and Inuit migrants from Denmark. , migrants (n = 995); , residents of Greenland (n = 1316); dotted line, Danes (reference; n = 6784). The number of men and women from Denmark and Greenland in each BMI subgroup ranged from 38 to 195. Data from Danes in Denmark were included as a reference. P value is for the hypothesis of no significant difference between the 2 groups of Inuit. A significant interaction between sex and country of residence in relation to fasting glucose (P = 0.048) was observed, whereas no significant effect modification was found on the interaction between sex and BMI for country of residence (sex x country of residence x BMI). No significant interaction was observed between sex and country of residence in relation to 2-h glucose concentrations.
FIGURE 2.. Mean concentrations and 95% CIs of fasting and 2-h insulin concentrations resulting from a general linear model with age included as a continuous variable by category of BMI in Inuit residents from Greenland and Inuit migrants from Denmark. , migrants (n = 995); , residents of Greenland (n = 1316); dotted line, Danes (reference; n = 6784). The number of men and women in Denmark and Greenland in each BMI subgroup ranged from 38 to 195. Data from Danes in Denmark were included as a reference. P value is for the hypothesis of no significant difference between the 2 groups of Inuit. A significant interaction was observed between sex and country of residence in relation to fasting insulin (P = 0.004) and 2-h insulin concentrations (P = 0.04), whereas no significant effect modification was found on the interaction between sex, BMI, and country of residence (sex x country of residence x BMI).
The concentration of triacylglycerol in both sexes was significantly lower in the residents in Greenland than in the migrants for most BMI categories (Figure 3). In the women, the migrants had higher concentration of HDL cholesterol than did the participants in Greenland for each category of BMI, whereas HDL cholesterol did not differ significantly between the 2 groups for the men. Finally, residents in Greenland consistently had the lowest levels of both systolic and diastolic blood pressure regardless of BMI category (Figure 4). For 2-h plasma glucose and insulin, triacylglycerol, and blood pressure, the figures indicated that Inuit migrants followed the pattern observed in Danes, ie, higher mean values for a given level of BMI. However, migrant women had higher concentrations of HDL cholesterol than did both the women from Greenland and Danish women.
FIGURE 3.. Mean concentrations and 95% CIs of triacylglycerol and HDL cholesterol concentrations resulting from a general linear model with age included as a continuous variable by category of BMI in Inuit residents from Greenland and Inuit migrants from Denmark. , migrants (n = 995); , residents of Greenland (n = 1316); dotted line, Danes (reference; n = 6784). The number of men and women in Denmark and Greenland in each BMI subgroup ranged from 38 to 195. Data from Danes in Denmark were included as a reference. P value is for the hypothesis of no significant difference between the 2 groups of Inuit. A significant interaction was observed between sex and country of residence in relation to triacylglycerol (P = 0.01) and HDL-cholesterol concentrations (P = 0.0002), whereas no significant effect modification was found on the interaction between sex, BMI, and country of residence (sex x country of residence x BMI).
FIGURE 4.. Mean levels and 95% CIs of systolic and diastolic blood pressure resulting from a general linear model with age included as a continuous variable by category of BMI in Inuit residents from Greenland and Inuit migrants from Denmark. , migrants (n = 995); , residents of Greenland (n = 1316); dotted line, Danes (reference; n = 6784). The number of men and women in Denmark and Greenland in each BMI subgroup ranged from 38 to 195. Data from Danes in Denmark were included as a reference. P value is for the hypothesis of no significant difference between the 2 groups of Inuit. A significant interaction was found between sex and country of residence in relation to systolic blood pressure (P = 0.04), whereas no significant effect modification was found on the interaction between sex, BMI, and country of residence (sex x country of residence x BMI). No significant interaction was observed between sex and country of residence in relation to diastolic blood pressure.
To formally test whether country of residence modified the effect of increasing BMI on cardiovascular disease risk factors, we included an interaction term between country of residence and BMI (data not shown). Although the levels of risk factors were significantly different, as shown in the figures, the interaction was not significant for any of the risk factors, thus showing that the trends in the association between obesity and metabolic effects in the Inuit populations were not significantly different.
The effect of migration status, including country of residence (Denmark compared with Greenland), on cardiovascular disease risk factors was analyzed including country of residence (Denmark or Greenland), BMI, and age as covariates in a general linear model (Table 3, model 1). To test whether lifestyle factors were responsible for the observed differences between the 2 Inuit groups, we added the following as covariates in the linear models: diet (seal or fish and fresh fruit) for model 2; and diet, physical activity, smoking, school education, and alcohol consumption for model 3. Adjustment for diet and other lifestyle factors attenuated the differences between the 2 groups; however, the difference remained significant after the adjustments for blood pressure, 2-h glucose, and triacylglycerol for all and 2-h insulin in the men and HDL cholesterol in the women.
View this table:
TABLE 3. Mean values of 8 cardiovascular disease risk factors according to country of residence and BMI resulting from general linear models
We repeated the analyses for categories of WC and WHR and found the same pattern as for BMI (data not shown). The differences in the levels of risk factors between the 2 groups of Inuit were even more pronounced for categories of WHR than for categories of WC.
DISCUSSION
Inuit residents in Greenland had lower levels of 2-h plasma glucose and insulin, blood pressure, and triacylglycerol than did the Inuit migrants in Denmark at any given level of obesity, even though there was generally an increasing trend in the levels of metabolic risk factors with obesity in both populations. The trend in the association with obesity categories was significantly different only for HDL cholesterol, with higher concentrations observed in the women migrants than in the women residents in Greenland. Fasting plasma glucose concentrations were not significantly different according to obesity categories in the 2 Inuit populations. Thus, Inuit migrants in Denmark seem to follow the same pattern as a general Danish reference population in the association between obesity and cardiovascular disease risk factors (11). The opposite association between migration status and the effect of obesity on HDL cholesterol observed in the women is particularly interesting. A previous study conducted on Inuit migrants found higher HDL-cholesterol concentrations in full-blooded Inuit compared with those of mixed origin, suggesting a genetically determined high sensitivity to environmentally induced changes in plasma lipids (15).
These findings indicate that lifestyle factors play a major role for observed differences in the levels of risk factors for cardiovascular disease at a given level of obesity. However, inclusion of lifestyle-related covariates influenced the findings only negligibly. Lifestyle differences between the groups may be present that we did not capture through our questionnaire; the 4-category physical activity questionnaire was never validated in this specific population. Also, the traditional Greenlandic diet, which is rich in n–3 polyunsaturated fatty acids, may influence metabolic activity in adipose tissue. Our results may be limited because we used a food-frequency questionnaire rather than direct measurements of dietary biomarkers. Furthermore, the food-frequency questionnaire did not provide detail on the amounts consumed, so specific nutrient intakes cannot be estimated.
Some limitations are inherent in our data. Because of the low response rate of the migrants, selection bias cannot be excluded. Unhealthy persons are less likely to participate in epidemiologic studies than are healthy persons (20), and if this trend was evident in both countries, the prevalence of risk factors would be underestimated in Denmark. However, even though selection bias may influence the general level of risk factors, it is not possible to determine the possible influence of selection bias on the association between obesity and other risk factors for cardiovascular disease. Aside from bias deriving from the collection of data, migrant studies share a common problem of interpretation, stemming from the fact that migration itself is a selective process. People migrate for social, economic, health-related reasons, etc. Thus, migrants do not constitute a representative sample of the population of origin.
Visceral fat distribution is recognized as an important risk factor for ischemic heart disease and type 2 diabetes more than is overall obesity (21, 22). Although differences in the amount of visceral fat for a given level of anthropometric measurements are suggested to explain observed differences in obesity-associated risk of cardiovascular disease across populations (23, 24), this is not likely to explain the observed differences between the 2 groups of Inuit. It has, however, been suggested that Inuit living in the Arctic may have a higher ratio of subcutaneous fat to intraabdominal fat as a mechanism of adaptation to the cold environments compared with people living in a warmer climate (25). However, only studies with direct measures of visceral fat accumulation by imaging techniques could provide additional insight into the causes of the observed variability between populations and population-groups and also into the reasons for a change in risk factors over time.
In the Inuit, alternative and higher cutoffs for BMI and WC have been suggested (26). The main purpose of defining cutoffs points for overweight and obesity is to allow comparisons within and between populations, and cutoffs allow for the identification for intervention purposes of persons who are at high risk. When population-specific cutoffs are defined, it should be noted that such cutoffs are only applicable to the population of origin and may reflect differences in lifestyle and environmental factors rather than genetic factors that influence disease risk associated with obesity. Hence, cutoffs should not be interpreted alone but in combination with other risk factors of morbidity and mortality. Our findings, however, were based on cross sectional data of mean values of risk factors for cardiovascular disease, and only prospective studies with diseases as outcomes can determine the most appropriate definition of overweight and obesity for treatment and intervention purposes and decide whether population-specific thresholds are warranted.
ACKNOWLEDGMENTS
MEJ collected and analyzed data and was the main writer of the manuscript. PB was principal investigator of the Greenland Population Study, provided advice on data analysis, and edited and revised the manuscript. KB-J was the principal coinvestigator of the Greenland Population Study, the principal investigator of the Danish study Inter99, provided advice on data analysis, and edited and revised the manuscript. The authors had no conflicts of interest.
REFERENCES