点击显示 收起
1 From the Departments of Nutrition (EL-G, RMvD, LQ, and FBH) and Epidemiology (FBH), Harvard School of Public Health and the Channing Laboratory (FBH), Harvard Medical School, Boston, MA
2 Supported by NIH research grants HL65582 and DK58845. FBH was partially supported by an American Heart Association Established Investigator Award. 3 Reprints not available Address correspondence to E Lopez-Garcia, Department of Preventive Medicine, Universidad Autonoma de Madrid, Arzobispo Morcillo, 2 Madrid, Spain. E-mail: esther.lopez{at}uam.es.
ABSTRACT
Background: In several short-term studies, coffee consumption has been associated with impairment of endothelial function.
Objective: The objective was to assess the relation between long-term caffeinated and decaffeinated filtered coffee consumption and markers of inflammation and endothelial dysfunction.
Design: We conducted a cross-sectional study of 730 healthy women and 663 women with type 2 diabetes from the Nurses’ Health Study I cohort, who were aged 43–70 y and free of cardiovascular disease and cancer at the time blood was drawn (1989–1990). Dietary intake was assessed with a validated food-frequency questionnaire in 1986 and 1990.
Results: About 77% of the healthy women consumed 1 cup (237 mL) caffeinated coffee/mo and 75% consumed 1 cup decaffeinated coffee/mo; the corresponding intakes for women with type 2 diabetes were 74% and 63%, respectively. In healthy women, no appreciable differences in plasma concentrations of the markers were found across categories of caffeinated coffee intake. In women with type 2 diabetes, higher caffeinated coffee consumption was associated with lower plasma concentrations of E-selectin (adjusted percentage change per 1 cup/d increment = –3.2%; P = 0.05) and C-reactive protein (adjusted percentage change = –10.2%; P < 0.001). Higher decaffeinated coffee consumption was associated with lower plasma concentrations of E-selectin (adjusted percentage change = –2.5%; P = 0.08) and C-reactive protein (adjusted percentage change = –7.9%; P = 0.02) only in healthy women. The results were similar when we also adjusted the models for other dietary factors and blood lipids and when we excluded participants with hypertension or hypercholesterolemia.
Conclusions: These results indicate that neither caffeinated nor decaffeinated filtered coffee has a detrimental effect on endothelial function. In contrast, the results suggest that coffee consumption is inversely associated with markers of inflammation and endothelial dysfunction.
Key Words: Coffee • inflammation • C-reactive protein • CRP • endothelial dysfunction • cross-sectional study • Nurses’ Health Study • type 2 diabetes • women
INTRODUCTION
Increasing evidence points to inflammation and endothelial dysfunction as common features in the group of alterations constituting the metabolic syndrome: hyperglycemia, insulin resistance, hyperinsulinemia, hypertension, obesity, and dyslipidemia (1), all of which are risk factors for cardiovascular disease. Evidence also exists that dietary factors might influence the risk of cardiovascular disease through modulation of endothelial function (2-4).
Several studies have found that coffee consumption is associated with an impairment of the flow-mediated dilatation of the brachial artery (5), aortic stiffness, and wave reflections (6). In addition, a cross-sectional study found that moderate coffee consumption is related to an increase in plasma concentrations of several markers of general inflammation (7). On the other hand, prospective studies have reported that coffee consumption is not associated with an increased risk of coronary heart disease (8-10), and epidemiologic studies have consistently linked coffee consumption with a lower risk of type 2 diabetes mellitus (11). In the present study, we evaluated the associations between long-term consumption of caffeinated and decaffeinated filtered coffee and plasma markers of inflammation and endothelial dysfunction in healthy and diabetic women in the Nurses’ Health Study (NHS).
SUBJECTS AND METHODS
Subjects
The NHS was established in 1976 and included 121 700 female registered nurses residing in the United States. Every 2 y, participants are mailed a follow-up questionnaire to update information about their medical history, lifestyle, and other risk factors. The present study included 2 subcohorts. One subcohort consisted of 730 women selected as control subjects for an earlier nested case-control study of type 2 diabetes (12) and who had not been diagnosed with cardiovascular disease, cancer, or type 2 diabetes mellitus at the time of the blood collection in 1989–1990. The second subcohort included 663 women who provided blood samples and had a confirmed diagnosis of type 2 diabetes mellitus but not of cardiovascular disease or cancer (13). We also excluded those women who reported insulin use, because this hormone has antiinflammatory properties (14). The average ages of the women at the time of the blood collection in the 2 subcohorts were 56 and 58 y, respectively (range: 43–70 y). The Harvard School of Public Health and Brigham and Women’s Hospital Human Subjects Committee Review Board approved the study protocol.
Blood collection and assessment of markers
Blood was collected between 1989 and 1990. Women willing to provide blood specimens were sent instructions and a phlebotomy kit. Blood specimens were returned by overnight mail on ice, centrifuged (1200 x g, 15 min) on arrival to separate plasma from buffy coat and red cells, and frozen in liquid nitrogen until analyzed. Ninety-seven percent of the samples arrived within 26 h of phlebotomy. Quality-control samples were routinely frozen along with study samples to monitor changes due to long-term storage and assay variability. All markers were measured in the Clinical Chemistry Laboratory at Children’s Hospital in Boston. Concentrations of soluble intercellular adhesion molecule 1 (sICAM-1), E-selectin, and soluble tumor necrosis factor receptor 2 (sTNF-R2) were measured by commercial enzyme-linked immunosorbent assay (R&D Systems, Minneapolis, MN). High-sensitivity C-reactive protein (CRP) concentrations were measured by latex-enhanced turbidimetric assay on a Hitachi 911 (Denka Seiken, Tokyo, Japan). In the subcohort of healthy women, the interassay CVs for each biomarker were as follows: sICAM-1, 6.1–10.1%; E-selectin, 6.4–6.6%; CRP, 3.4–3.8%; and sTNFR-2, 3.6–5.1%. In the diabetic subcohort, the interassay CVs were as follows: sICAM-1, 3.3–4.8%; E-selectin, 5.7–8.8%; CRP, 2.8–5.1%; and sTNF-R2: 2.6–4.8%. Processing times did not substantially affect the concentrations of the markers (15).
Assessment of coffee consumption
In 1986 and 1990, a semiquantitative food-frequency questionnaire (FFQ) was mailed to the participants. The FFQ included 116 food items with specified serving sizes that were described by using natural portions or standard weight and volume measures of the servings commonly consumed in this study population. On each questionnaire, the participants were asked how often on average during the previous year they had consumed caffeinated or decaffeinated coffee. The participants could choose from 9 responses (never, 1–3 times/mo, 1 time/wk, 2–4 times/wk, 5–6 times/wk, 1 time/d, 2–3 times/d, 4–5 times/d, and 6 times/d). The method of coffee preparation was assessed in 1990. Using the US Department of Agriculture food-composition sources (16) supplemented with other sources, we estimated the nutrient content of the foods. We assessed the total intake of nutrients by summing the nutrient content for the specific amount of each food multiplied by a weight proportional to the frequency of its use. These values were adjusted for total energy intake by the residual approach (17).
In our validation study, we obtained high correlations between consumption of coffee from the FFQ and consumption estimated from repeated 1-wk diet records (r = 0.78) (18). The mean coffee intakes (cups/d; 1 cup = 237 mL) estimated by diet records according to categories of the FFQ were 0.24 for <1 cup/mo, 0.97 for 1 cup/d, and 1.85 for 2–3 cups/d (18). For the present analyses, we used the average coffee consumption and nutrient intakes in 1986 and 1990 to represent long-term dietary consumption and to reduce measurement error. Coffee consumption was categorized into 4 groups: <1 cup/mo, 1 cup/mo to 4 cups/wk, 5–7 cups/wk, and 2 cups/d.
Assessment of other variables
Height was assessed on the baseline questionnaire in 1976, and body weight was assessed in 1990 (19, 20). Body mass index (BMI) was calculated as weight (kg)/height2 (m). Physical activity in 1990 was assessed in hours per week spent on common leisure-time physical activities expressed as metabolic equivalent hours per week (MET-h/wk) (21). Alcohol consumption was measured as the average intake (g/d) between 1986 and 1990 (22). Hormone therapy use was ascertained among postmenopausal women, who were classified as never, past, or current users in 1990. The concentrations of total cholesterol and triacylglycerols were measured simultaneously from the blood samples with the Hitachi 911 analyzer with reagents and calibrators from Roche Diagnostics (Indianapolis, IN); the CV for both measurements was <1.8%.
Statistical analysis
We used PROC GLM in SAS (23) to calculate the age-adjusted geometric means and their 95% CIs for the markers in each category of coffee consumption. We log transformed plasma concentrations of the markers to better approximate normal distributions. Then we calculated the exponential values of the means and the CIs. We also calculated the means of the markers adjusted for age (5-y categories), BMI (<23.0, 23.0–24.9, 25.0–29.9, 30.0–34.9, and 35.0), quintiles of physical activity (MET-h/wk), smoking status [never smoker, past smoker, and current smoker (1–14 and 15 cigarettes/d)], alcohol consumption (nondrinker and 0–4.9, 5.0–9.9, 10–14.9, and >15.0 g/d), current aspirin use, and postmenopausal hormone therapy (premenopausal and never, past, and current user).
Multiple linear regression analyses were used to assess the relation between continuous coffee intake and plasma concentrations of endothelial markers. Regression coefficients were expressed as percentage differences in dependent variables for a specified difference in coffee consumption (1 cup/d). In addition, we performed sensitivity analyses by further adjusting the models for nutrients and foods that had previously been associated with endothelial function or diabetes risk (trans fatty acids, total n–3 fatty acids, cereal fiber, glycemic load, magnesium, dairy products, processed meat, nuts, and fruit and vegetables). We also examined whether plasma lipids (total cholesterol and triacylglycerols) might modify the associations. Finally, we repeated the analyses excluding the participants who reported high blood pressure or hypercholesterolemia, because they could have modified their dietary intake when the disease was diagnosed.
RESULTS
Among 730 healthy women, 77% consumed at least 1 cup of caffeinated coffee per month and 75% women consumed at least 1 cup/mo of decaffeinated coffee. The corresponding percentages for the 663 diabetic women were 74% for caffeinated coffee and 63% for decaffeinated coffee. There were no significant differences among the groups.
In Table 1, we show the characteristics of both populations stratified by the categories of total coffee consumption. Women in the higher categories tended to smoke more cigarettes and drink more alcohol than those in the lower ones. When comparing the extreme categories, caffeine intake varied from 84 to 356 g/d among healthy women, and from 83 to 377 g/d among diabetic women; also, magnesium intake increased from 285 to 315 mg/d and from 284 to 322 mg/d, respectively. In addition, coffee consumption was associated with lower glycemic load and lower tea consumption.
View this table:
TABLE 1. Characteristics of the population in 1990 by categories of total coffee consumption (average between consumption in 1986 and 1990)1
We calculated mean plasma concentration markers by categories of coffee consumption and observed higher values for all markers in women with type 2 diabetes (Tables 2 and 3). In healthy women, no appreciable differences across categories of caffeinated coffee were found, except a marginally inverse association between increasing coffee intake and decreasing plasma concentrations of sTNF-R2 (adjusted percentage change per 1 cup/d increment = –2.1%; P = 0.08 from the linear regression model) (Table 2). In women with type 2 diabetes, we found an inverse association of caffeinated coffee with plasma E-selectin concentrations (adjusted percentage change = –3.2; P = 0.05) and a stronger inverse association with CRP (adjusted percentage change = –10.2; P < 0.001). These results were similar when we used caffeine intake instead of caffeinated coffee consumption. For decaffeinated coffee (Table 3), we observed a marginally inverse association between coffee consumption and the concentration of E-selectin (adjusted percentage change = –2.5; P = 0.08) and CRP (adjusted percentage change = –7.9; P = 0.02) in the subcohort of healthy women. Finally, no association between decaffeinated coffee and the studied markers was seen in diabetic women.
View this table:
TABLE 2. Adjusted geometric mean (and 95% CIs) plasma concentrations of inflammation and endothelial dysfunction markers by categories of caffeinated coffee consumption1
View this table:
TABLE 3. Adjusted geometric mean (and 95% CIs) plasma concentrations of inflammation and endothelial dysfunction markers by categories of decaffeinated coffee consumption1
In additional analyses, we observed similar results when we further adjusted the models for the nutrients and foods that could confound the association. Similarly, adjustment for total cholesterol and plasma triacylglycerols did not modify the results, nor did the exclusion of participants with a diagnosis of high blood pressure or hypercholesterolemia.
DISCUSSION
We found no evidence for adverse effects of caffeinated or decaffeinated coffee on markers of inflammation and endothelial dysfunction. By contrast, we found an inverse association of caffeinated coffee consumption with E-selectin and CRP in diabetic women and an inverse association of decaffeinated coffee consumption with CRP in healthy women.
The role of inflammatory and endothelial dysfunction markers in the atherogenic process has been well established. CRP is a marker of systemic inflammation and has been shown to play a direct role in atherosclerosis by mediating LDL uptake by macrophages, which stimulates monocyte release of inflammatory cytokines, and also by mediating induction of monocyte chemotactic protein 1 in endothelial cells (24). The soluble TNF receptor, which is induced by TNF- and other cytokines, may attenuate the activity of TNF- but also promotes inflammation in the absence of free TNF ligand (25). More specific markers of endothelial function are E-selectin and soluble vascular adhesion molecule 1, which are surface and soluble leukocyte adhesion molecules, respectively, which are overexpressed when the endothelium encounters inflammatory stimuli (26). All these molecules have been found to predict cardiovascular disease (27) and to be associated with cardiovascular events (28, 29).
Few previous studies have examined the effect of coffee on the endothelium. Zampelas et al (7) performed a cross-sectional study with 3000 healthy participants from the ATTICA study, in which coffee consumption was assessed once with a FFQ. They found that regular coffee consumption was related to higher plasma concentrations of interleukin 6, CRP, and TNF-. Because unfiltered coffee was included when coffee consumption was measured, it is plausible that compounds in this beverage that increase cholesterol concentrations also contribute to increased inflammation and endothelial dysfunction (30). Another possible explanation for the differences between our study and theirs is that we used data on coffee consumption from 2 different FFQs, which reflect 4 y of consumption—a longer duration than was used in the ATTICA study.
Our findings are somewhat consistent with the fact that coffee consumption may reduce inflammation and markers of endothelial activation. Yukawa et al (31) conducted an in vivo study with 11 healthy males who drank 24 g coffee/d for 1 wk and found that regular coffee ingestion reduced LDL oxidation susceptibility. Coffee may favorably affect endothelial atherosclerotic plaques through this pathway, because oxidized LDL is present in atherosclerotic lesions enhancing the process (32). Additionally, the phenolic compounds of coffee (chlorogenic acid, ferulic acid, and p-coumaric acid) have a strong protective antioxidant effect in the endothelium. One of them, chlorogenic acid, is hydrolyzed in the body into caffeic acid and quinic acid. In a study performed with endothelial cells (33), a metabolite of caffeic acid, dihydrocaffeic acid, was found to increase nitric oxide synthase activity and to scavenge intracellular reactive oxygen species.
Our study had several limitations. First, it was cross-sectional, so we cannot infer causality from our results. Second, we cannot exclude a certain degree of residual confounding in the analyses. For example, it is possible that women with hypertension might have limited their coffee consumption after the diagnosis, confounding the association. However, we performed analyses in which those women who reported hypertension or hypercholesterolemia were excluded, and the results did not change. Third, some degree of error is inherent in the measurement of food consumption as well as in biochemical measures, although the dietary questionnaire has been shown to reflect long-term intake, and the validation data showed that coffee was among the most accurately reported foods in the FFQ (18). In addition, the use of repeated measurements of food consumption enabled us to reduce within-person random error. Finally, the biomarker measures were stable and statistically consistent for 36 h from collection to processing (15).
In conclusion, our data did not support a detrimental effect of coffee consumption on the endothelium. Instead, we observed associations between higher coffee consumption and lower plasma concentrations of several markers of inflammation and endothelial dysfunction in both healthy and diabetic women. These results support previous findings that filtered coffee consumption does not increase the risk of cardiovascular disease (34).
ACKNOWLEDGMENTS
EL-G, RMvD, LQ, and FBH conceived and designed the study, analyzed and interpreted the data, and critically revised the manuscript for important intellectual content. FBH acquired the data, obtained funding, supervised the study, and provided administrative, technical, and material support. EL-G and FBH drafted the manuscript and provided statistical expertise.
None of the authors declared a conflict of interest.
REFERENCES