| |
Association of Serum C-Reactive Protein Level with Sex-Specific Type 2 Diabetes Risk: A Prospective Finnish Study
|
| |
| |
Journal of Clinical Endocrinology & Metabolism June 2009, doi:10.1210/jc.2008-2260
"the present study demonstrated that elevated plasma level of CRP was associated with an increased risk of type 2 diabetes among both men and women, but this association was stronger in women than men. Our results support the notion that low-grade systemic inflammation may be associated with the pathogenesis of type 2 diabetes. The mechanisms remain unknown. Nevertheless, our data suggest that the measure of CRP might provide an adjunctive method for the early detection of type 2 diabetes risk."
Gang Hu, Pekka Jousilahti, Jaakko Tuomilehto, Riitta Antikainen, Jouko Sundvall and Veikko Salomaa
Departments of Health Promotion and Chronic Diseases Prevention (G.H., P.J., J.T., V.S.) and Health and Functional Capacity (J.S.), National Public Health Institute, FIN-00300, Helsinki, Finland; Department of Public Health (G.H., J.T.), University of Helsinki, FIN-00014, Helsinki, Finland; Chronic Disease Epidemiology Laboratory (G.H.), Pennington Biomedical Research Center, Louisiana State University System, Baton Rouge, Louisiana 70808; South Ostrobothnia Central Hospital (J.T.), 60220, Seinajoki, Finland; Oulu City Hospital (R.A.), 90015 Oulu, Finland; and Imperial College, Division of Medicine, Hammersmith Campus (R.A.), London W12 ONN, United Kingdom
Abstract
Objective: Our objective was to assess whether the association of serum C-reactive protein (CRP) with type 2 diabetes risk is modified by sex.
Design and Subjects: We prospectively followed 12,861 Finnish men and women who were 35-74 yr of age, and free of diabetes, coronary heart disease, stroke, and cancer at baseline. Hazard ratios of type 2 diabetes were estimated for different levels of serum CRP.
Results: During the follow-up, 208 men and 113 women developed diabetes. The multivariable-adjusted (age, physical activity, education, smoking, alcohol and coffee drinking, family history of diabetes, use of antihypertensive drugs, cholesterol-lowering agents, and hormone replacement therapy in women, systolic blood pressure, serum high-density lipoprotein cholesterol, serum triglycerides, and body mass index) hazard ratios of diabetes at three different levels of CRP (0.05-0.99, 1.0-2.99, and >/=3.0 mg/liter) based on the recommendation by Centers for Disease Control and the American Heart Association were 1.00, 1.46, and 1.85 (P for trend = 0.006) in men, and 1.00, 3.83, and 8.37 (P for trend <0.001) in women, respectively. CRP had a stronger association with diabetes risk in women than men (P for interaction: X2 = 6.42; 1 df; P < 0.025). This positive association between CRP and diabetes risk did not change when participants were stratified by age group, smoking status, level of obesity, alcohol drinking habit, or family history of diabetes.
Conclusions: High baseline level of serum CRP was associated with an increased risk of diabetes among both men and women, but this association was stronger in women than men.
Introduction
Serum C-reactive protein (CRP), a marker of systemic inflammation, has predicted cardiovascular disease (1). Recent cross-sectional studies have indicated a clear association of elevated levels of CRP with obesity, insulin resistance, or the metabolic syndrome (2, 3). In several cohort (4, 5, 6, 7, 8, 9, 10) and nested case-control studies (11, 12, 13, 14), elevated levels of CRP have also been identified as a predictor of the development of type 2 diabetes. This association is markedly reduced but still remains statistically significant after adjustment for body mass index (BMI) or waist circumference in some studies (5, 7, 8, 10, 11, 13, 14), whereas in other studies it has become nonsignificant (9, 12). Moreover, the Hoorn study has indicated that adjustment for BMI attenuated the positive association between CRP and diabetes risk more than the waist to hip ratio did (6). Very limited prospective studies have assessed whether this positive association between CRP levels and type 2 diabetes is modified by sex (4, 5, 6, 7). The study samples in such studies have been relatively small, and the results from them are inconsistent. The aim of this study was to examine the sex-specific role of CRP on the risk of type 2 diabetes independently of three indicators of obesity (BMI, waist to hip ratio, or waist circumference) and other established risk factors among Finnish men and women.
Discussion
In this large prospective study, the elevated baseline level of CRP was significantly and directly associated with an increased risk of developing diabetes among both Finnish men and women. This association was stronger in women than men. This association was independent of any indicator of obesity (BMI, waist to hip ratio, or waist circumference) and some known risk factors for type 2 diabetes.
Although CRP levels are much higher in women than men in general, to our knowledge, only four prospective epidemiological studies (4, 5, 6, 7) and one nested case-control study (28) have evaluated a potential sex difference in the association of CRP levels with regard to the risk of diabetes, and the results are inconsistent. In the Mexico City Diabetes Study including 515 men and 729 women (4), a significantly direct association between CRP and the risk of diabetes was found in women but not men, and this association was significantly different between genders. In the Hoorn Study including 140 Dutch men and 139 women, a significantly direct association between CRP and the risk of diabetes was found in men but not women (6). In two other prospective studies with Japanese origin, one Japanese-American population study including 396 men and 551 women in Hawaii and Los Angeles (5) and the Hisayama Study including 694 men and 1065 women in Japan (7), an equally strong direct association between CRP and diabetes risk was seen in both men and women. In the MONItoring trends and determinants in CArdiovascular disease/KORA Augsburg nested case-control study (28), elevated concentrations of CRP showed a clearly stronger association with the risk of type 2 diabetes in women than men (P for sex interaction <0.001). However, these associations considerably attenuated after further adjustment for BMI and other metabolic risk factors, and became nonsignificant in men but remained significant in women (28). Our study, with a much larger sample size (5755 men and 6280 women) and the larger number of diabetic cases during follow-up (208 men and 113 women) compared with the aforementioned studies (4, 5, 6, 7, 28), found a strong and independent association between CRP levels and the risk of diabetes in both men and women, but this association was stronger in women than men. The inconsistency of these studies with our study can be due to the relatively small sample sizes and few cases of incident diabetes in previous studies (4, 5, 6, 7, 28).
Several possible reasons might explain the stronger association of CRP with the risk of diabetes in women than men in our study. First, because the correlations between CRP and any indicator of adiposity were stronger in women than men in our study, adiposity related inflammation may have a greater role in the development of insulin resistance and type 2 diabetes in women. Second, an interaction between endogenous sex hormones and inflammation with the risk of diabetes may also exist. A recent meta-analysis indicated that high testosterone levels in women and conversely, low testosterone levels in men are associated with a higher risk of type 2 diabetes (29). High concentrations of SHBG are associated with a decreased diabetes risk in both men and women, however, this inverse association was stronger in women than men. High endogenous estradiol levels may be associated with a higher risk of type 2 diabetes both in men and postmenopausal women (29). The results from other studies demonstrated that high concentrations of SHBG were associated with lower CRP levels in postmenopausal women (30). In our study, 24.3% of women used hormonal replacement therapy; however, adjustment for the use of hormone replacement therapy, use of hypertensive drugs, and use of cholesterol-lowering agents did not change the results.
The strengths of our study included the large sample size of both men and women from a homogeneous population, the large number of incident diabetes cases and standardized measurements of three different indicators of adiposity, and a large number of other related risk factors and potential confounders. Several limitations should be considered. First, our study did not perform either fasting glucose test or glucose tolerance test at baseline and follow-up surveys. Therefore, we could have missed some cases of asymptomatic and diet-treated diabetes, although the clinical diagnosis of diabetes from the hospital discharge register may in part avoid this potential underdiagnosis. Second, we used a single baseline measurement of anthropometric parameters and serum CRP. Therefore, we could not evaluate possible effects of changes in adiposity indicators and CRP over time. Third, no instructions were given to the participants to avoid smoking or alcohol before blood sample collection. Finally, we cannot completely exclude either the effects of residual confounding due to measurement error in the assessment of confounding factors, or some unmeasured factors.
In conclusion, the present study demonstrated that elevated plasma level of CRP was associated with an increased risk of type 2 diabetes among both men and women, but this association was stronger in women than men. Our results support the notion that low-grade systemic inflammation may be associated with the pathogenesis of type 2 diabetes. The mechanisms remain unknown. Nevertheless, our data suggest that the measure of CRP might provide an adjunctive method for the early detection of type 2 diabetes risk.
Results
General characteristics of the study population at baseline are presented in Table 1. Age, BMI, waist circumference, waist to hip ratio, systolic and diastolic blood pressure, serum triglycerides, family history of diabetes, physical inactivity, use of antihypertensive drugs, and use of hormone replacement therapy (in women only) were positively associated with baseline CRP level in both men and women. Serum HDL cholesterol and education were inversely associated with CRP level in both men and women. Alcohol consumption and current smoking showed positive associations with CRP level among men only, whereas coffee consumption showed an inverse association, and use of cholesterol-lowering agents showed a positive association with CRP level among women only. CRP correlated significantly with BMI in both men and women (r = 0.35 and 0.51, respectively), waist to hip ratio (r = 0.35 and 0.38), and waist circumference (r = 0.38 and 0.50) (all P <0.001).
The multivariable-adjusted [age, study year, physical activity, education, smoking, alcohol and coffee drinking, family history of diabetes, use of antihypertensive drugs, cholesterol-lowering agents and hormone replacement therapy, systolic blood pressure, serum HDL cholesterol, and serum triglycerides (multivariable model 2)] hazard ratios of diabetes at three different levels of CRP (0.05-0.99, 1.0-2.99, and >/=3.0 mg/liter) based on the recommendation by Centers for Disease Control and the American Heart Association were 1.00, 1.87, and 2.51 (P for trend <0.001) in men, and 1.00, 5.46, and 14.9 (P for trend <0.001) in women, respectively (Table 2). This positive association attenuated but remained significant after further adjustment for BMI (multivariable model 3), waist to hip ratio (multivariable model 4), or waist circumference (multivariable model 5) in both men and women (all P for trend <0.05), except for multivariable model 5 in men (P for trend = 0.072).
The multivariable-adjusted (model 2) hazard ratios of diabetes across quartiles of CRP were 1.00, 2.05, 3.46, and 3.84 (P for trend <0.001) in men, and 1.00, 2.63, 6.37, and 16.4 (P for trend <0.001) in women, respectively (Table 2). When CRP was examined as a continuous variable, the multivariable-adjusted (model 2) hazard ratios of diabetes for each 1-unit increase in log-transformed CRP were 2.30 [95% confidence interval (CI) 1.73-3.06] in men, and 4.66 (95% CI 3.21-6.75) in women. Additional adjustment for BMI, waist to hip ratio, or waist circumference attenuated these associations, but they still remained statistically significant in both sexes (all P for trend <0.05). When further analyses were conducted based on the 1997 survey or 2002 survey separately, the positive association between CRP and the risk of diabetes did not change in these two different surveys (data not shown), and there was no significant interaction between study year and CRP with the risk of diabetes.
After exclusion of participants with CRP levels more than 10 mg/liter (n = 430), the multivariable-adjusted (model 3) hazard ratios of diabetes at three different levels of CRP (0.05-0.99, 1.0-2.99, and 3.0-10.0 mg/liter) were 1.00, 1.48 (95% CI 1.03-2.11), and 1.79 (95% CI 1.20-2.67) (P for trend = 0.015) in men, and 1.00, 3.77 (95% CI 1.58-8.98), and 8.53 (95% CI 3.61-20.2) (P for trend <0.001) in women. After exclusion of participants with coronary heart disease, stroke, or cancer during the follow-up (n = 697), the multivariable-adjusted (model 3) hazard ratios of diabetes at three different levels of CRP (0.05-0.99, 1.0-2.99, and >/=3.0 mg/liter) were 1.00, 1.56 (95% CI 1.06-2.28), and 1.89 (95% CI 1.25-2.85) (P for trend = 0.009) in men, and 1.00, 2.65 (95% CI 1.12-6.27), and 5.83 (95% CI 2.51-13.5) (P for trend <0.001) in women.
There were significant interactions between sex and CPR as a continuous variable with the risk of diabetes in multivariable analyses: model 3 (P for interaction: X2 = 6.42; 1 df; P <0.025); model 4 (P for interaction: X2 = 9.98; 1 df; P <0.005); and model 5 (P for interaction: 2 = 4.89; 1 df; P <0.05). These results indicated that the association between CRP and the risk of diabetes was stronger in women than men.
In the multivariable analyses (model 3), the association between CRP and the risk of diabetes did not change when participants were stratified by age group, smoking status, level of obesity, alcohol drinking habit, or family history of diabetes (Fig. 1). No significant interactions of CRP and any of these parameters with the risk of diabetes were identified. However, there were significant interactions between sex and CPR at three different levels (0.05-0.99, 1.0-2.99, and >/=3.0 mg/liter), with the risk of diabetes in participants aged 55-74 yr old, ever or current smokers, nonobese participants, nonalcohol drinkers, alcohol drinkers, and participants without a family history of diabetes (all P for interaction <0.05) (Fig. 1).
Subjects and Methods
Subjects
Two independent cross-sectional population surveys were performed in five geographical areas of Finland in 1997 and in six geographical areas in 2002. The samples included subjects who were 25-74 yr of age. The original sample was drawn at random after stratification by sex and five equally large 10-yr age groups according to the World Health Organization MONItoring trends and determinants in CArdiovascular disease protocol (15). In the analyses we included participants in the 1997 survey only if they by chance happened to participate in both 1997 and 2002 surveys. The total sample size of the two surveys was 17,107. The participation rates were 74% in 1997 and 65% in 2002 (16). Our analysis included participants of 35-74 yr of age (13,794 participants) due to the few cases of type 2 diabetes in participants of 25-34 yr of age during the follow-up. Of 13,794 participants, we excluded participants with known diabetes (n = 543), coronary heart disease (n = 415), stroke (n = 269), or cancer (n = 169) at baseline, participants with type 1 diabetes during follow-up (n = 1), and participants with incomplete data on serum CRP (n = 334) or any other variables required for this analysis (n = 28), leaving 5,755 men and 6,280 women in the present analysis. These surveys were conducted according to the ethical rules of the National Public Health Institute, and the investigations were performed in accordance with the Declaration of Helsinki. All participants provided their written informed consent before participation in the study.
Measurements
A self-administered questionnaire was sent to the participants to be completed at home and returned to the survey site. The questionnaire included questions on medical history, history of drug treatments, socioeconomic factors, smoking habits, physical activity, and dietary habits. Education level, measured as the total number of school years, was divided into birth cohort-specific tertiles. Based on the responses the participants were classified as never, ex-smokers, and current smokers. Current smokers were categorized into those who smoked less than 20, or 20 or more cigarettes per day. Physical activity included occupational, commuting, and leisure-time physical activity, and was merged and regrouped into three categories: low, moderate, and high (17, 18, 19, 20, 21, 22). Information on the use of antihypertensive drugs, cholesterol-lowering agents (statins), hormone replacement therapy in women, and personal and parental history of diabetes (yes and no) was obtained. Coffee consumption was categorized into four categories: zero to two, three to four, five to six, and seven or more cups per day (23, 24, 25). Alcohol consumption was categorized into four groups: none, 1-50, 51-100, and more than 100 g/wk (24).
At the study site, specially trained research nurses checked the questionnaire and measured the height, weight, waist and hip circumferences, as well as blood pressure using a standardized protocol (15). Body weight of the participants wearing usual light-indoor clothing without shoes was measured with a 0.1-kg precision. Height was measured to the nearest 0.5 cm. BMI was calculated as weight in kilograms divided by the square of the height in meters. Waist circumference was measured midway between the lower rib margin and iliac crest. Hip circumference was measured at the level of widest circumference over greater trochanters. Blood pressure was measured with a standard sphygmomanometer from the right arm of the participant who was seated for 5 min before the measurement. After blood pressure measurement, a venous blood specimen was taken. Blood samples were collected after at least a 4-h fast and mailed to a central laboratory.
Serum high-density lipoprotein (HDL) cholesterol concentration was measured enzymatically (Roche Molecular Biochemicals, Wiesbaden, Germany) after precipitation of non-HDLs with dextran sulfate/magnesium sulfate in 1997. HDL cholesterol method was homogenous enzymatic test in 2002 (Thermo Scientific, Vantaa, Finland). Serum triglycerides were measured using enzymatic method from Roche Molecular Biochemicals in 1997 and from Thermo Scientific in 2002. Measurements were performed using Olli-C analyzer (Thermo Scientific) in 1997 and Optima analyzer (Thermo Scientific) in 2002. A high-sensitivity serum CRP concentration was determined using a latex immunoassay (Sentinel Diagnostics, Milan, Italy) with the Architect c8000 clinical chemistry analyzer (Abbott Laboratories, Abbott Park, IL). The lowest detection level was at 0.1 mg/liter. All values below the detection level (<0.5% of determinations) were coded as 0.05. The coefficient of variation varied between 4.7 and 2.3%, depending on the level of CRP. Due to a skewed distribution of CRP values, log transformations were used in some analyses. All measurements were performed at the laboratory of the National Public Health Institute.
Diagnosis of diabetes
We ascertained incident cases of diabetes from the National Hospital Discharge Register and the National Social Insurance Institution's Drug Register (17, 18, 23, 24, 25). These register data were linked to the risk factor survey data with the unique personal identification numbers assigned to every resident of Finland. Antidiabetic drugs prescribed by a physician are free of charge in Finland, subject to approval of the application to the Social Insurance Institution with a case history prepared by the treating physician attached. The diagnosis of diabetes is confirmed on the basis of the World Health Organization criteria: one or more classical symptom plus a fasting plasma glucose level of 7.0 mmol/liter or more, or the oral glucose tolerance test of 11.1 mmol/liter or more; at least one increased plasma glucose concentration and a fasting plasma glucose level of 7.0 mmol/liter or more, or the oral glucose tolerance test of 11.1 mmol/liter or more in the absence of symptoms; or treatment with a hypoglycemic drug (oral antidiabetic agents or insulin) (26). All patients who are entitled to free of charge medication for diabetes are entered into a register maintained by the Social Insurance Institution. Data on type 1 and 2 diabetes have been reported separately from the National Hospital Discharge Register since 1987. Participants with diabetes at baseline identified retrospectively from the National Social Insurance Institution's Drug Register, from the National Hospital Discharge Register, and by a survey question about diabetes diagnosed by a physician were excluded from the present analysis. Participants with type 1 diabetes during follow-up were also excluded from the present analysis. Follow-up of each cohort member continued until the date of the diagnosis of type 2 diabetes, death, or until the end of year 2006. The median follow-up periods were 9.8 yr for the 1997 survey and 4.8 yr for the 2002 survey. When these two surveys were combined, 208 men and 113 women developed incident type 2 diabetes during an average of 7.0 yr of follow-up.
Statistical analyses
Differences in risk factors based on different levels of CRP were tested using ANOVA after adjustment for age and study year. The Cox proportional hazards model was used to estimate the association of serum CRP with the risk of type 2 diabetes. CRP was evaluated in the following three ways: 1) as the three categories (0.05-0.99, 1.0-2.99, and 3.0 mg/liter) recommended by the Centers for Disease Control and the American Heart Association (1); 2) as sex-specific quartiles; and 3) as a continuous variable after log transformation. Different levels of CRP were included in the models as dummy variables, and the significance of the trend over different categories of CRP was tested in the same models by giving an ordinal numerical value for each dummy variable. The proportional hazards assumption in the Cox model was assessed with graphical methods, and with models including time by covariate interactions (27). In general, all proportionality assumptions were appropriate. The analyses were first performed adjusting for age and study year, and then further for education, physical activity, smoking, alcohol and coffee consumption, family history of diabetes, use of antihypertensive drugs, cholesterol-lowering agents, and hormone replacement therapy, systolic blood pressure, serum HDL cholesterol, serum triglycerides, and any one indicator of obesity (BMI, waist to hip ratio, and waist circumference). The likelihood ratio test for interaction was performed to determine
whether the effect of CRP [as a continuous variable in whole sample analyses or into three categories (0.05-0.99, 1.0-2.99, and >/=3.0 mg/liter) in subgroup analyses] on diabetes risk was different in men and women. Statistical significance was considered to be P <0.05. All statistical analyses were performed with SPSS for Windows 16.0 (SPSS, Inc., Chicago, IL).
|
|
| |
| |
|
|
|
