Coffee consumption and reduced risk of developing

Lead Article

Coffee consumption and reduced risk of developing type 2 diabetes: a systematic review with meta-analysis

Context: Type 2 diabetes (T2D) is a major health problem worldwide that is associated with increased morbidity and mortality. There is increased interest in the value of different nutrition-based strategies for preventing the development of T2D. Objective: This review aims to cover current knowledge regarding the effects of coffee consumption on development of T2D or modulation of adverse complications. A meta-analysis on coffee consumption and the risk of T2D was conducted. Moreover, bioactive components in coffee, polymorphisms, and potential underlying mechanism(s) in relation to T2D and adverse complications are discussed. Data sources: PubMed was searched up to December 1, 2017, and prospective cohort and nested case–control studies of the association between coffee consumption and T2D risk were selected. Data extraction: Two investigators independently extracted data from included studies. Results: A total of 30 prospective studies with 1 185 210 participants and 53 018 incident T2D cases were included in the meta-analysis. The pooled relative risk (RR) was 0.71 (95% confidence interval [CI], 0.67–0.76) for the highest category of coffee consumption (median consumption, 5 cups/d) vs the lowest category (median consumption, 0 cups/d). The risk of T2D decreased by 6% (RR ¼ 0.94; 95%CI, 0.93–0.95) for each cup-per-day increase in coffee consumption. Results were similar for caffeinated coffee consumption (per additional cup of coffee per day: RR ¼ 0.93; 95%CI, 0.90–0.96) and decaffeinated coffee consumption (corresponding RR ¼ 0.94; 95%CI, 0.90–0.98). Conclusions: Available evidence indicates that coffee consumption is inversely associated with risk of T2D. Possible mechanisms behind this association include thermogenic, antioxidative, and anti-inflammatory effects; modulation of adenosine receptor signaling; and microbiome content and diversity.

INTRODUCTION Aging, physical inactivity, and obesity are closely linked to type 2 diabetes (T2D), which has become a global health problem.1 Type 2 diabetes is characterized by abnormal insulin secretion or insulin resistance, resulting in hyperglycemia. According to the Global Report on Diabetes from the World Health Organization

(WHO),2 an estimated 422 million adults worldwide were living with diabetes (ie, a prevalence of 8.5% among the adult population) in 2014, compared with 108 million in 1980. According to WHO, the number of people with T2D continues to rise, which reflects the increase in associated risk factors, such as being overweight or obese. The World Health Organization estimated that the largest numbers of people with

Affiliation: M. Carlstro¨m is with the Department of Physiology and Pharmacology, Karolinska Institutet, Stockholm, Sweden. S.C. Larsson is with the Unit of Nutritional Epidemiology, Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden. Correspondence: M. Carlstro¨m, Department of Physiology and Pharmacology, Karolinska Institutet, Nanna Svartz V€ag 2, SE-17177 Stockholm, Sweden. E-mail: [email protected]. Key words: cardiovascular, coffee, diabetes, inflammation, metabolic syndrome, oxidative stress, renal, type 2. C The Author(s) 2018. Published by Oxford University Press on behalf of the International Life Sciences Institute. V

All rights reserved. For permissions, please e-mail: [email protected]. doi: 10.1093/nutrit/nuy014 Nutrition ReviewsV Vol. 76(6):395–417 R

395

Downloaded from https://academic.oup.com/nutritionreviews/article-abstract/76/6/395/4954186 by Karolinska Institutet Library user on 24 September 2018

Mattias Carlstro¨m and Susanna C. Larsson

396

consumed around the world every day.12 Due to the broad consumption of coffee, numerous studies have examined the potential link between coffee intake and its affect on the modulation of metabolic functions and the development of diabetes. Considering the high consumption of and the many bioactive molecules in brewed coffee, further knowledge on both the positive and negative health effects of coffee is important. This review aims to cover current knowledge regarding the effects of coffee consumption on the risk of T2D or modulation of adverse complications. In this meta-analysis, results obtained using caffeinated versus decaffeinated coffee and filtered versus unfiltered coffee are discussed, along with potential differences between the sexes and among geographical regions. Moreover, bioactive components in coffee and underlying mechanisms, as well as how coffee consumption may impact on the triad of metabolic, cardiovascular, and renal disorders and associated morbidity and mortality, are discussed. CURRENT EVIDENCE FROM EPIDEMIOLOGICAL AND CLINICAL STUDIES ON TYPE 2 DIABETES Very early studies that investigated the potential effects of coffee on metabolic regulation were published in the late 1960s.13,14 In contrast with the early observation made by Jankelson and colleagues11 in men with maturity-onset diabetes, the study by Feinberg et al.14 clearly demonstrated that coffee ingestion considerably improved glucose clearance following an oral glucose tolerance test in healthy human participants. Since these initial studies, numerous cohort studies and crosssectional studies, along with a few systematic reviews and meta-analyses on coffee consumption and the risk of T2D, have been published.15–19 As reported herein, a new analysis of coffee consumption and risk of T2D was conducted, which expands beyond previously published systematic reviews and meta-analyses by also discussing potential underlying mechanisms. RESEARCH DESIGNS AND METHODS A flow-chart of study selection is shown in Figure 1. The implementation and reporting of this meta-analysis followed the Meta-analyses of Observational Studies in Epidemiology (MOOSE) guidelines (Appendix S1).20 PubMed was used to identify studies that evaluated the association between coffee consumption and risk of T2D and were published between January 1966 and December 1, 2017. No language or other restrictions were imposed. The computer-based search included the keyword “coffee” combined with “diabetes.” The reference lists of relevant articles were reviewed to search for Nutrition ReviewsV Vol. 76(6):395–417 R


diabetes live in the South-East Asia and Western Pacific Regions, accounting for approximately half of the diabetes cases in the world. Type 2 diabetes increases the risk of both renal and cardiovascular diseases (eg, kidney failure due to nephropathies, myocardial infarction, and stroke) and of adverse complications (eg, retinopathy, lower limb amputation, neuropathy, hearing impairment). Although renal disease is commonly described as a complication of metabolic syndrome and diabetes, recent studies suggest that compromised renal function may not only contribute to hypertension but also antecede metabolic dysfunction with abnormal insulin and glucose regulation.1,3,4 Major benefits of early diagnosis and treatment of hyperglycemia and cardiovascular risk factors in T2D have been proposed.5 However, this might be difficult because, in the early phase, there are often no obvious symptoms. Although diabetes-associated severe complications affect many people, there are currently very limited global estimates of diabetes-related end-stage renal disease and cardiovascular events, which can partly be explained by variations in population characteristics and methodological dissimilarities between studies.6 Further studies are urgently needed to better understand the underlying mechanisms and to reduce morbidity and mortality associated with this triad of metabolic, cardiovascular, and renal disorders. Given the substantial burden of diabetes, much attention has been given to the development of new pharmacological or nutritional approaches to prevent the development of T2D or to alleviate its complications. Several risk factors for T2D have been identified, including age, sex, obesity, low physical activity, smoking, diets with a low amount of fiber and high amount of saturated fat, ethnicity, family history, history of gestational diabetes mellitus, elevated blood pressure, dyslipidemia, and different drug treatments (eg, diuretics, nonselective b-blockers, statins, etc). There is also ample evidence that T2D has a strong genetic basis. The concordance of T2D in monozygotic twins is 70%, compared with 20%–30% in dizygotic twins.7 Although genetic factors have been associated with the etiology of T2D,8 accumulating evidence demonstrates important influence of modifiable lifestyle factors. It is well known that increased physical activity and lowering of caloric intake reduce the incidence of T2D,9–11 but more knowledge is needed to understand how different dietary approaches or food constituents may contribute to the prevention of T2D. Coffee is among the most widely consumed beverages worldwide and is a central and popular part of different cultures. According to the European Coffee Federation, an estimated 3.5 billion cups of coffee are

further studies. No contact was made with researchers to search for unpublished data. The Participants, Intervention Comparators, Outcomes, Study design (PICOS) criteria are shown in Table 1. Criteria for inclusion in this meta-analysis were 1) the study had a prospective cohort or nested case–control design; 2) exposure was coffee consumption (including total, caffeinated, decaffeinated, filtered, or unfiltered coffee); and 3) outcome was T2D incidence. Studies with a case–control or cross-sectional design without information on quantity of coffee consumption or with mortality from T2D or gestational diabetes as the outcome were excluded from the analysis. No restrictions were made with regard to study population. If results from 1 study population were reported in > 1 publication, the study with the longest follow-up or the one that provided results separately for men and women (if same follow-up) was included. Where results were reported separately for caffeinated and decaffeinated coffee but not for total coffee consumption, results for caffeinated coffee in the main analysis were included. Characteristics extracted from each study included first author’s name, publication year, study location, study name, sex and age of participants, sample size, number of cases, follow-up time, assessment of T2D, variables adjusted for in the full multivariable model, and the most fully adjusted relative risk (RR) estimates for each category of coffee consumption. Study quality was evaluated using the 9-star Newcastle-Ottawa Scale (NOS).21 The quality assessment was based on 3 broad perspectives, including selection, comparability, and outcome. The literature search, assessment of eligibility for inclusion in the meta-analysis, data extraction, and assessment of study quality were carried out Nutrition ReviewsV Vol. 76(6):395–417 R

RESULTS Twenty-nine articles with data from 30 prospective cohorts (1 article presented results from 2 cohorts) or nested case–control studies were included in the analysis (38 studies if separated by sex),27–55 with a total of almost 1.2 million participants (n ¼ 1 185 210, including 53 018 incident T2D cases; incidence of 4.5%) (Figure 1 and Table 2). Most studies adjusted for major potential confounders, including age, sex, body mass index, physical activity, and smoking, and the majority of studies also adjusted for alcohol and dietary factors (Table 2).27–55 Only 1 study adjusted for milk and sugar consumed with coffee or separately. Overall results The pooled RR for incident T2D was 0.71 (95%CI, 0.67–0.76) for the highest category of coffee consumption (median consumption, 5 cups/d) compared with the lowest category (median consumption, 0 cups/d) (Figure 2).27–55 There was heterogeneity among studies (P < 0.001; I2 ¼ 57.9%), which was mainly explained by the strength of the inverse association because all but 1 of the 38 RR estimates were < 1 and 20 were statistically significant. No evidence of publication bias was detected (Egger’s test: P ¼ 0.27). In a dose–response meta-analysis, the risk of T2D decreased by 6% (RR, 0.94; 95%CI, 0.93–0.95) for each cup-per-day increase of coffee consumption 397


Figure 1 Flow diagram of the literature search process

independently by both authors, and any disagreements were resolved by open discussion. Results were combined using a random-effects model, which takes into account both within-study and between-study variance. In addition to combining results for the highest versus lowest category of coffee consumption, a dose–response meta-analysis was conducted using the methods described previously.22–24 A P value for nonlinearity was computed by testing the null hypothesis that the coefficient of the second spline was equal to zero.24 Heterogeneity among studies was tested with the Cochran Q and I2 statistics.25 To evaluate sources of heterogeneity, stratified analyses were performed according to sex (men vs women), geographic region, study quality (NOS score 6–7 [moderate quality] vs 8–9 [high quality]), and type of coffee (caffeinated vs decaffeinated). Publication bias was evaluated using Egger’s test.26 The P value for difference across strata was obtained using meta-regression. The statistical analyses were conducted with GraphPad Prism 6 (Mac OS X, version 6.0 b, 2012, La Jolla, CA, USA) and Stata (version 14.2, College Station, TX, USA).

Table 1 PICOS criteria for inclusion and exclusion of studies Parameter

Inclusion criteria

Participants

Study design

Prospective cohort and nested case–control studies

Studies on caffeine intake only

Studies without information on quantity of coffee consumption Mortality from T2D, gestational diabetes Case–control and cross-sectional studies, case reports

Abbreviation: T2D, type 2 diabetes.

(Figure 3),27–55 without evidence of departure from a linear-response model (P ¼ 0.18). There was heterogeneity among studies (P < 0.001; I2 ¼ 66.5%). When consumption of >8 cups of coffee per day was excluded, the RR of T2D for each cup-per-day increase of coffee consumption was 0.93 (95%CI, 0.92–0.95) with no evidence of departure from a linear-response model (P ¼ 0.56). Although future prospective, well-designed clinical studies are warranted, the existing knowledge from epidemiological studies emphasizes prevention of T2D development with increasing intake of coffee. The association between coffee consumption and T2D risk appears to be linear, at least up to about 8–10 cups/day of coffee. Data on the influence of very high coffee consumption (>8–10 cups/d) on T2D risk are limited. Stratified analyses by sex, geographical region, and study quality An inverse association between coffee consumption and risk of T2D was observed in both men and women in studies conducted in the United States, Europe, and Asia, both in studies of moderate quality and those of high quality (Table 3).27–55 There was indication of a stronger association in women in the dose–response analysis per additional cup per day of coffee (P for difference by sex ¼ 0.03) but not in the analysis comparing high versus low coffee consumption (P ¼ 0.18) (Table 3). A previous dose–response meta-analysis18 also showed a somewhat stronger inverse association between coffee consumption and risk of T2D in women than in men (Figure 4A). In the present meta-analysis, 398

Differences by age or body mass index Given that aging and obesity are closely linked to the development of T2D, it is of interest to distinguish potential differences in the association between coffee consumption and incident T2D in young versus old, as well in lean versus obese people. Moreover, higher coffee consumption has generally been associated with less healthy lifestyle (eg, cigarette smoking, low level of physical activity, less healthy diet).54 Thus, the true association between coffee and diabetes risk might be stronger than observed. Previously published metaregression analysis studies with obesity (BMI >25) or increased age (>50 y) as independent variables found similar association between coffee consumption and risk of T2D.15,18

Caffeinated versus decaffeinated coffee Results for both caffeinated and decaffeinated coffee were available in 10 studies. Comparing the highest versus the lowest category, both caffeinated coffee consumption (RR, 0.73; 95%CI, 0.64–0.83) and decaffeinated coffee consumption (RR, 0.80; 95%CI, 0.69–0.94) were inversely associated with risk of T2D. This association was dose-dependent, with no evidence of departure from a linear-response model (P for nonlinearity ¼ 0.72 and 0.39, respectively, for caffeinated and decaffeinated coffee). The risk of T2D decreased, respectively, by 7% and 6% per cup-per-day increment of caffeinated and decaffeinated coffee consumption (P for difference ¼ 0.69) (Table 3). In agreement with a previous study,18 this meta-analysis also shows that caffeinated and decaffeinated coffee consumption, as well as caffeine intake, had similar inverse associations with risk of T2D (Figure 4B). Nutrition ReviewsV Vol. 76(6):395–417 R


Adults, without regard to sex or ethnicity Intervention Coffee consumption or exposure (including total, caffeinated, decaffeinated, filtered, or unfiltered coffee) Comparison Highest vs lowest category of coffee consumption; per 1-cupper-day increase in coffee consumption Outcome Incident T2D

Exclusion criteria

there was evidence that the association was stronger in studies with moderate quality than in those with high quality (P for difference ¼ 0.02), but this difference was only observed in the analysis between the highest versus the lowest category of coffee intake (Table 3). Heterogeneity among studies was observed in most subgroups, except in men (dose–response analysis only) and in studies conducted in Asia (Table 3). The data show no clear differences in association between coffee consumption and risk of T2D by sex, geographic region, or study quality, although small differences (eg, a possible stronger association in women) cannot be ruled out.

R

Age, y

Both 32–60

Both 40–65

Hoorn Study

North America (USA) NHANES-1

JACC Study

JACC Study

Europe (Netherlands)

Asia (Japan)

Asia (Japan)

Greenberg et al. (2005)30

Iso et al. (2006)31

Iso et al. (2006)31

Both 40–65

Both 50–74

39–65

van Dam et al. (2004)29

W

BEDA Study

Europe (Sweden)

Rosengren et al. (2004)28

Both 30–60

Finnish Twin Cohort

10 686

6727

7006

1312

1361

10 652

2680

Both 15

Europe (Finland)

19 518

17 111

No.

Both 20–98

Both 30–60

Sex

Carlsson et al. (2004)27

Saremi et al. (2003)26

NA

Study name

Mobile Clinic Health Examination Survey North America (USA) Pima Indians Study

Europe (Netherlands)

Region (country)

Reunanen et al. Europe (Finland) (2003)25

van Dam and Feskens (2002)24

References

185

200

309

128

74

408

824

855

306

5

5

8.4

6

18

20

11

16

7

Cases Follow-up years

Table 2 Studies included in the meta-analysis of coffee consumption and risk of type 2 diabetes

SR

SR

SR

OGTT

CSR/NR

CSR

OGTT

NR

SR

Assessment of T2D

7

7

7

8

7

6

6

7

6

NOS score

0