点击显示 收起
1 From the Department of Nutrition, Institute of Basic Medical Sciences, University of Oslo, Oslo, Norway (LFA, DRJ, MHC, and RB), and the Division of Epidemiology and Community Health, School of Public Health, University of Minnesota, Minneapolis, MN (DRJ)
2 The contents of this manuscript are solely the responsibilities of the authors and do not necessarily represent the official view of the funding bodies. 3 Supported by grants from the Norwegian Research Council, The Johan Throne Holst Nutrition Research Foundation, and The Norwegian Cancer Society (RB) and by a grant (RO1 CA39742) from the National Cancer Institute (Iowa Women's Health Study; DRJ). 4 Address reprint requests to DR Jacobs Jr, Division of Epidemiology and Community Health, School of Public Health, University of Minnesota, 1300 South 2nd Street, Suite 300, Minneapolis, MN 55454-1015. E-mail: jacobs{at}epi.umn.edu.
ABSTRACT
Background: Coffee is the major source of dietary antioxidants. The association between coffee consumption and risk of death from diseases associated with inflammatory or oxidative stress has not been studied.
Objective: We studied the relation of coffee drinking with total mortality and mortality attributed to cardiovascular disease, cancer, and other diseases with a major inflammatory component.
Design: A total of 41 836 postmenopausal women aged 55–69 y at baseline were followed for 15 y. After exclusions for cardiovascular disease, cancer, diabetes, colitis, and liver cirrhosis at baseline, 27 312 participants remained, resulting in 410 235 person-years of follow-up and 4265 deaths. The major outcome measure was disease-specific mortality.
Results: In the fully adjusted model, similar to the relation of coffee intake to total mortality, the hazard ratio of death attributed to cardiovascular disease was 0.76 (95% CI: 0.64, 0.91) for consumption of 1–3 cups/d, 0.81 (95% CI: 0.66, 0.99) for 4–5 cups/d, and 0.87 (95% CI: 0.69, 1.09) for 6 cups/d. The hazard ratio for death from other inflammatory diseases was 0.72 (95% CI: 0.55, 0.93) for consumption of 1–3 cups/d, 0.67 (95% CI: 0.50, 0.90) for 4–5 cups/d, and 0.68 (95% CI: 0.49, 0.94) for 6 cups/d.
Conclusions: Consumption of coffee, a major source of dietary antioxidants, may inhibit inflammation and thereby reduce the risk of cardiovascular and other inflammatory diseases in postmenopausal women.
Key Words: Oxidation • infection • antioxidants • mortality • women
INTRODUCTION
Inflammation is a beneficial host response to foreign challenge or tissue injury that leads to restoration of tissue structure and function. If prolonged, however, inflammation can contribute to pathogenesis in diseases of infectious origin and in diseases such as rheumatoid arthritis, gout, chronic obstructive pulmonary disease, emphysema, asthma, ischemia-reperfusion, ulcerative colitis, Crohn disease, type 1 and type 2 diabetes, several types of neurodegenerative diseases, cancer, and atherosclerosis (1-4).
Inflammation is closely related to oxidative stress. Many cells produce reactive oxygen and nitrogen species during inflammation. NADPH oxidase, which is activated during inflammation, mediates the respiratory burst (5, 6). Its primary product, superoxide (O2–), can lead to other reactive oxygen species. Superoxide is also formed by the mitochondrial electron transport chain during inflammation (5). Furthermore, induction of nitric oxide synthases will increase the production of the · NO radical and subsequent reactive nitrogen species (5, 6). Antioxidant therapy (reactive oxygen species or reactive nitrogen species decomposition catalysts or selective antioxidant enzyme mimics) has been shown to prevent in vivo tissue injury during inflammation (7-9).
We recently analyzed the total antioxidant capacity of several thousand foods (10-13) and were surprised to learn that in a healthy Norwegian population coffee contributes >60% of total dietary antioxidants (on the basis of in vitro analysis of foods) (13). Several different compounds contribute to coffee's antioxidant capacity, eg, caffeine (14); polyphenols, including chlorogenic acids (15); volatile aroma compounds (16); and heterocyclic compounds, including pyrroles, oxazoles, furans, thiazoles, thiophenes, imidazoles, and pyrazines (16, 17). Many of these are efficiently absorbed, and plasma antioxidants increase after coffee intake (16, 18). Epidemiologic studies have found that coffee is associated with reduced biomarkers of oxidative stress (19).
Additional interest in coffee in relation to inflammatory diseases arises because caffeic acid efficiently inhibits in mice in vivo the activation of the transcription factor nuclear factor B, which is a central mediator of the inflammatory response (1, 20). Recently, Schulze et al (21), in a nested case-control study of nurses aged 30–55 y who were followed for 24 y for incident diabetes, observed that coffee intake was inversely associated with inflammatory cytokines. Similarly, Honjo et al (22) showed that biomarkers of liver inflammation were inversely associated with coffee consumption in 7313 Japanese men aged 48–59 y. Fleming et al (23) found that serum ferritin was inversely associated with coffee intake among 634 men and women aged 67–93 y. In contrast, Zampelas et al (24) reported higher concentrations of inflammatory markers among coffee drinkers in 3042 Greek men and women with a much wider age range of 18–89 y. In the present report, we studied the association of coffee with total and cause-specific mortality in postmenopausal women in Iowa who answered a dietary questionnaire in 1986 and were followed for 15 y.
SUBJECTS AND METHODS
The Iowa Women's Health Study cohort comprises 41 836 women aged 55–69 y who were recruited through a baseline questionnaire mailed in 1986 and were followed up through December 31, 2001. A previous report explained how self-reported baseline risk factors were assessed and defined (25). Baseline histories of physician-diagnosed cancer (other than skin cancer), heart disease, and angina were obtained. In the present study, we excluded women who were not postmenopausal (n = 569) and those with self-reported baseline heart disease (n = 3459), nonskin cancer (n = 3497), diabetes (n = 2386), chronic colitis (n = 2357), or liver cirrhosis (n = 313), as well as those who reported <600 or >5000 kcal/d or left >30 items blank on the diet questionnaire (n = 3096). Some exclusions overlapped; the final sample included 28 246 women. Data were missing for waist-hip ratio for 94 women, estrogen use for 100, physical activity status for 405, and smoking status for 380, with some overlap; therefore, the models referred to below contain 27 312 women. The study was reviewed and approved by the Institutional Review Board of the University of Minnesota.
Dietary intake at baseline was assessed by using a 127-item food-frequency questionnaire (26). The participants were asked how often on average during the previous year they had consumed a cup of coffee; the questionnaire included one question for regular coffee and one for decaffeinated coffee. The frequency responses were as follows: never, 1–3 cups/mo, 1 cup/wk, 2–4 cups/wk, 5–6 cups/wk, 1 cup/d, 2–3 cups/d, 4–5 cups/d, and 6 cups/d. Coffee is quite reliably reported; as reported elsewhere, the 3-mo test-retest r for caffeine was 0.95, and the correlation of caffeine intake as reported on the food-frequency questionnaire with that in the average of five 24-h dietary recalls was 0.82 (27).
Most deaths were identified by annual linkage of cohort identifiers to Iowa statewide death records. To find additional deaths, identifiers of nonrespondents to the follow-up surveys were sent to the National Death Index. Cause of death was that assigned as the single underlying cause by state health departments as follows: cardiovascular disease [(CVD) International Classification of Diseases, Ninth Revision (ICD-9) codes 390–459; International Classification of Diseases, Tenth Revision (ICD-10) codes I00-I99] and cancer (ICD-9 codes 140–239; ICD-10 codes C00-D48) (28) or all other codes. The other codes were further split a priori into diseases in which inflammation plays a major role and all remaining diseases. We categorized all infectious diseases as inflammatory diseases (specifically in these women, 170 cases: ICD-9 classifications 11.6, 28.9, 31, 38, 38.1, 38.4, 38.9, 42.9, 46.1, 54.3, 54.7, 70.3, 70.5, 75, 78.5, 117.5, 117.9, 137, 322.9, 323.9, 473.9, 480.9, 481, 482, 482.1, 482.3, 482.4, 482.8, 485, 486, 487.1, 540, 555.9, 566, 567.2, 614.4, and 682.6 and ICD-10 classifications A04.4, A04.7, A31.9, A41.9, A49.0, A81.0, B18.2, B44.1, B49, B94.8, G03.9, J11.0, J11.1, J15.2, J15.9, J18.0, J18.1, J18.9, J22, K65.9, L02.9, and N39.0). Additional categories that were defined as inflammatory diseases were chronic neurodegenerative diseases (105 cases: ICD-9 codes 331, 332, 340, 341.9, 348.3, 355.9, and 357 and ICD-10 codes G20, G30.1, G30.9, G31.9, and G35), type 1 and type 2 diabetes (35 cases: ICD-9 codes 250, 250.1, 250.3, and 250.6 and ICD-10 codes E10.2, E10.7, E10.9, E11.2, E11.5, E11.9, E14.4, E14.5, E14.7, and E14.9), chronic lower respiratory diseases (265 cases: ICD-9 codes 491.2, 491.8, 491.9, 492, 493.1, 493.9, 494, and 496 and ICD-10 codes J40, J42, J43.9, J44.0, J44.1, J44.8, J44.9, J45.9, J46, and J47), chronic liver disease and cirrhosis (23 cases: ICD-9 codes 571.1–6 and 573.3 and ICD-10 codes K73.2 and K74.6), chronic renal failure (12 cases: ICD-9 codes 585 and 586 and ICD-10 codes N12, N18.0, N18.9, and N19), rheumatic diseases (32 cases: ICD-9 codes 710, 710.1, 710.4, 714, 715.3, and 725 and ICD-10 codes M05.1, M06.9, M19.9, M30.1, M31.3, M34.9, and M46.2), and several other diseases of the respiratory system (49 cases: ICD-9 codes 135, 136.3, 466, 507, 515, and 530.1 and ICD-10 codes J69.0, J80, and J84.1), digestive system (15 cases: ICD-9 codes 541, 556, 558, 562.1, and 574.1 and ICD-10 codes K50.9, K52.9, K81.9, and K85), and some others (7 cases: ICD-9 codes 279.3, 303, and 707 and ICD-10 codes E85.3, E85.4, E85.9, L40.5, and N32.1).
Statistical analyses were performed with SAS software, version 8.2 (SAS Institute, Inc, Cary, NC). Coffee consumption was categorized into 5 groups: 0, <1, 1–3, 4–5, and 6 cups/d. An F test with 4 and 27 307 degrees of freedom was used for testing baseline differences among the coffee intake categories. Cox proportional hazards regression analysis was used to assess the association between coffee consumption and death, presented as estimates of hazard ratios (HRs) and their 95% CIs. Two models are presented, a minimally adjusted model that included adjustment only for age, smoking (5 categories: no, current smokers of <15 cigarettes/wk, current smokers of >15 cigarettes/wk, past smokers of <15 cigarettes/wk, past smokers of >15 cigarettes/wk), and intake of alcohol (continuous, g/d, as a quadratic function), and a fully adjusted model that included the following other possible confounders: body mass index, waist-hip ratio, education (lower than high school, high school, higher than high school), physical activity (low, moderate, pr high), use of estrogen (never used, current use, or past use), use of multivitamin supplements (yes, no, or don't know), energy intake (continuous, kcal), and intakes of whole grain (continuous, servings/d), refined grain (continuous, servings/d), red meat (continuous, servings/d), fish and seafood (continuous, servings/d), and total fruit and vegetables (continuous, servings/d). We performed a test for trend by treating coffee intake as a continuous variable.
RESULTS
The mean (±SD) consumption of coffee in the Iowa Women's Health Study cohort was 2.7 ± 2.0 cups/d (237 mL per cup), of which 1.7 ± 1.9 and 1.0 ± 1.5 cups/d were regular and decaffeinated coffee, respectively. Higher coffee consumption was strongly associated with cigarette smoking and greater alcohol intake (Table 1). Weaker inverse associations were seen with age, body mass index, physical activity, educational attainment, intake of fruit and vegetables, intake of whole grains, and history of arthritis; positive associations were seen with energy intake, intake of refined grains, and red meat intake.
View this table:
TABLE 1. Baseline characteristics according to total coffee consumption of 27 312 women in the Iowa Women's Health Study who were free at baseline of cancer (other than skin), heart disease, diabetes, colitis, and liver cirrhosis, 1986
Although the unadjusted association of coffee intake with total mortality was flat (data not shown), the unadjusted total mortality rate across coffee consumption categories was downsloping when stratified into 5 levels of smoking (Figure 1). The figure may suggest a larger absolute difference in mortality according to coffee intake in current smokers (from 45% for nondrinkers of coffee to 35% in all other coffee-drinking categories) than in nonsmokers (decreasing from 15% for nondrinkers of coffee to 12% or 13% in other coffee-drinking categories), but there was no significant interaction (P = 0.43 with 4 df and coffee as a continuous variable and 5 categories of smoking). Consistent with the limited sample sizes for these detailed comparisons, most pairwise comparisons were not statistically significant. Furthermore, the percentage reduction was more homogeneous across smoking categories. The findings reported below were similar when current smokers were omitted. Thus, the primary message of Figure 1 is that an inverse relation between coffee intake and total death was present even within smoking categories. Consequently, we used the 5-level smoking variable as a covariate in all further analyses.
FIGURE 1.. Total mortality rate by coffee consumption and smoking (adjusted for age) in the Iowa Women's Health Study, 1986–2001 (n = 27 312). , never smokers (n = 18 102); , past smokers of <15 cigarettes/d (n = 2569); , past smokers of 15 cigarettes/d (n = 2528); , smokers of <15 cigarettes/d (n = 1245); , smokers of 15 cigarettes/d (n = 2868).
After adjustment for age, smoking, and alcohol, the hazard ratio for total death rates showed an inverse association with total coffee consumption. In the fully adjusted model, the hazard ratios were 0.85 (95% CI: 0.76, 0.94) and 0.81 (95% CI: 0.72, 0.91) for 1–3 cups/d and 4–5 cups/d, respectively, and were slightly higher in those drinking 6 cups/d (Table 2). A U-shapedassociation was found between coffee intake and death from CVD (P for quadratic trend = 0.005) in which the hazard ratios were 0.76 (95% CI: 0.64, 0.91) and 0.81 (95% CI: 0.66, 0.99) for 1–3 and 4–5 cups/d, respectively, compared with 0.87 (95% CI: 0.69, 1.09) for 6 cups/d. Death from cancer showed no significant relation with intake of coffee per day. Death from causes other than CVD and cancer showed a decrease in risk with increasing coffee intake. In the fully adjusted model, the hazard ratios were 0.80 (95% CI: 0.65, 0.99) and 0.76 (95% CI: 0.60, 0.97) for 1–3 and 4–5 cups/d, respectively, compared with 0.79 (95% CI: 0.60, 0.97) for 6 cups/d.
View this table:
TABLE 2. Hazard ratios for total and cause-specific mortality in the Iowa Women's Health Study according to total coffee consumption, 1986–20011
To explore the relation between non-CVD and noncancer deaths and coffee intake further, we subdivided this group a priori into death caused by inflammatory diseases and death caused by injury and other causes. After adjustment for age, smoking, and alcohol, an inverse association was observed between coffee consumption and the 713 deaths from inflammatory disease (Table 3). Further adjustments did not significantly influence the hazard ratios. In the fully adjusted model, the hazard ratios for daily consumption of 1–3, 4–5, and 6 cups/d were 0.72 (95% CI: 0.55, 0.93), 0.67 (95% CI: 0.50, 0.90), and 0.68 (95% CI: 0.49, 0.94), respectively. Excluding the first 2 y of death did not substantially alter the findings. No significant associations were observed between coffee intake and death caused by injury or other causes.
View this table:
TABLE 3. Hazard ratios for death attributed to inflammatory diseases, infectious diseases, injury, and other diseases in the Iowa Women's Health Study according to total coffee consumption, 1986–20011
Infectious diseases, a major subgroup of the inflammatory diseases, are most often characterized by a more acute pathogenesis than the other inflammatory diseases. As shown in Table 3, we did not observe any significant association between coffee intake and the 170 deaths attributed to infectious diseases, whereas the inverse relation with the 543 deaths attributed to noninfectious inflammatory diseases was significant.
We did the same type of analysis for regular and decaffeinated coffee and for other common beverages (Table 4). We observed similar effects for the 2 subcategories of coffee as reported above for total coffee intake. The coffee associations were not repeated for other beverages, including tea, fruit juice, sugar-sweetened drinks, diet soda, and skim, low-fat, and whole milk. Total and CVD mortality were higher in those who reported consuming more than one sugar-sweetened beverage per day than in those who reported any other consumption level. Correspondingly, there was a significant upward trend in CVD mortality with a significant increase in those reporting >2.5 drinks/d. Non-CVD and noncancer inflammatory death risk was lower in each category of those who reported drinking skim or low-fat milk than in nondrinkers, but the association was not graded and the trend was not significant. Further adjustment for diet soda and skim or low-fat milk did not substantially alter the findings described in Tables 2 and 3.
View this table:
TABLE 4. Hazard ratios for total death and death attributed to cardiovascular and other inflammatory diseases in the Iowa Women's Health Study according to total intake of regular coffee, decaffeinated coffee, and other diet beverages, 1986–20011
DISCUSSION
We prospectively studied the relation of coffee drinking with total and cause-specific mortality, with particular interest in deaths attributed to inflammatory diseases in a cohort study of postmenopausal women. Similar to total death, death rates from CVD showed a U-shaped association with coffee intake, whereas death attributed to inflammatory diseases other than CVD and cancer decreased significantly and consistently across categories of increasing coffee consumption. In the final fully adjusted model that included 27 312 women, the hazard ratio for death from inflammatory diseases was 28% lower among coffee drinkers of 1–3 cups/d than in nondrinkers; there were further small decreases suggestive of a threshold to 33% and 32% reductions in the hazard ratio among those drinking 4–5 cups and 6 cups/d. These trends were not seen for the consumption of other beverages.
We did not observe any significant inverse relation between coffee intake and death attributed to infectious diseases, whereas the inverse association between coffee intake and noninfectious inflammatory diseases remained strong and consistent. These data suggest that the acute pathogenesis that often characterizes most of the infectious diseases is less prone to the protective effect of coffee. Individual infectious diseases may be affected by coffee, but our data set was not large enough to justify analysis of smaller subcategories.
We developed an a priori classification of noncardiovascular, noncancer diseases in which inflammation plays a major role. We acknowledge that this classification remains somewhat subjective, but the inclusion was based on the following arguments. In addition to all infectious diseases, we included chronic degenerative diseases in which tissue degradation is causally linked to the progression of the diseases. CVD and cancer also involve inflammation, for example, leukocyte recruitment and expression of proinflammatory cytokines. Because they are themselves major disease categories, we examined CVD and cancer separately from these other inflammatory diseases.
A plausible candidate mechanism for the observed inverse association of death attributed to inflammatory diseases with coffee may be related to recent studies indicating that coffee is the major source of antioxidants in the diet (13, 18, 29). Chlorogenic acid, the ester of caffeic acid with quinic acid, is the most abundant polyphenol in coffee, is readily absorbed, and has relatively high bioavailability (15, 18).
The inverse association of coffee with death attributed to inflammatory diseases is consistent with recent findings in epidemiologic studies of an inverse association of coffee with the risk of inflammatory diseases such as Parkinson disease (30), gallstones (31), liver cirrhosis (32), and diabetes (21). Epidemiologic studies have also found that coffee is associated with reduced plasma concentrations of -glutamyl transpeptidase, a biomarker for oxidative stress (19), and that dietary antioxidants may reduce the risk of inflammatory diseases (9) and diabetes (33).
Surprisingly few studies have explored the possible effect of coffee consumption on mortality. A meta-analysis that analyzed studies available up until 1992 of coffee use and coronary death was nonconclusive (34). A more recent study of 11 000 men and women aged 40–53 y who took part in the Scottish Heart Health Study observed that increasing coffee consumption was associated with reduced all-cause mortality (35). Similar observations were obtained in a recent Finnish study consisting of 20 179 men and women aged 30–59 y (36). In that study, the association of coffee drinking with all-cause mortality was U-shaped, with mortality being highest among those who did not drink coffee at all.
A study by Jazbec et al (37) examined the association of coffee consumption with all-cause mortality in Croatia. Analyses were based on data obtained from participants recruited in 1969 with follow-ups in 1972 (1571 men and 1793 women) and 1982 (1093 men and 1330 women). During the observation period until 1999, 568 men and 382 women died. Jazbec et al observed that men and women who regularly drank 1–2 cups of coffee daily had a significantly lower risk of all-cause death than did those who did not drink coffee. No significant effects were observed for those drinking >2 cups/d. In a study from Norway, Tverdal and Skurtveit (32) examined mortality due to liver cirrhosis at follow-up of 51 306 men and women who underwent screening for CVD from 1977 to 1983. After 17 y, the total number of deaths from liver cirrhosis was 53. Death from liver cirrhosis, adjusted for sex, age, and alcohol use, was inversely associated with coffee consumption.
Caution is needed in generalizing our results and the results of other cited studies because the method of preparation and the source and type of coffee beans can vary considerably and a variety of noncoffee foods are added to coffee in different amounts. The method of coffee preparation has also changed with time and varies geographically (11). In the Iowa Women's Health Study population, the percolator method and the filter method are likely to have been the main coffee preparation methods, with an increasing use of the filter method over time. This probably led to less exposure to atherogenic coffee lipids (38) during later parts of the study period than during early study periods. The U-shaped association between coffee intake and CVD may also be related to how the coffee is prepared: at higher intakes of coffee prepared by the percolator method, the intake of atherogenic coffee lipids may outweigh the potentially beneficial compounds in coffee. Reverse causality remains a possibility, because there are inflammatory conditions such as rheumatoid arthritis about which we did not ask at baseline. However, we did exclude women with many of the most prevalent inflammatory conditions. Residual confounding remains a possibility, particularly for smoking, alcohol, and coffee intake at the highest levels of each. Loss to follow-up, however, was minimal. Mortality follow-up was close to complete given the use of both the Iowa state records and the National Death Index.
In conclusion, our results are consistent with a protective effect of intake of 1–3 cups of coffee per day on total death and death from cardiovascular and other inflammatory diseases in a group of postmenopausal women. If our observation is reproduced in other studies and proves to be causal, the implications are considerable, being that coffee is the second most widely consumed drink worldwide.
ACKNOWLEDGMENTS
LFA and DRJ had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. DRJ and RB were responsible for the study concept and design. MHC, LFA, and DRJ were responsible for data acquisition and analysis and interpretation of data. RB was responsible for drafting the manuscript. LFA, DRJ, MHC, and RB contributed to critical revision of the manuscript for important intellectual content. LFA and DRJ performed the statistical analysis. DRJ and RB obtained funding and supervised the study. None of the authors had personal or financial conflicts of interest.
REFERENCES