Higher habitual coffee consumption was associated with higher insulin sensitivity (1) and a lower risk for type 2 diabetes (26) in diverse populations. In contrast, short-term metabolic studies showed that caffeine intake can acutely lower insulin sensitivity (79) and increase glucose concentrations (1015). Randomized intervention studies are needed to examine whether tolerance to these acute effects develops after longer-term consumption (16). We therefore examined the effects of coffee and caffeine on fasting blood concentrations of glucose and insulin over 2–4 weeks in two crossover studies in healthy volunteers.

The studies were approved by the TNO Nutrition and Food Research Medical Ethics Committee, and all participants gave informed consent. The trials were originally designed to study the effects of coffee and caffeine on plasma concentrations of homocysteine, and the study designs have been reported in detail previously (17,18). Participants were regular coffee consumers (more than five cups/day) and did not have known diabetes.

The first study was a 4-week crossover study that compared the effects of regular paper-filtered coffee consumption with that of coffee abstinence. A total of 40 volunteers used 1 l of coffee (70 g coffee grounds) for 4 weeks and abstained from coffee for 4 weeks in random order. Fourteen participants did not complete the trial because of nausea and restlessness (n = 7), possible susceptibility to adverse effects of caffeine intake (n = 3), or reasons unrelated to treatment (n = 4). Thus, 26 participants were included in the analysis. The second study had a Latin-square design with three treatments given in random order for 2 weeks each: caffeine (a total of 870 mg in six capsules), regular paper-filtered coffee (52 g ground coffee/day in 0.9 l), and placebo (six capsules containing cellulose). Of the 54 volunteers, 6 subjects withdrew because of severe headaches (n = 2), study-related illness (n = 1), or reasons unrelated to treatment (n = 3). For the current analyses, we excluded participants because of missing blood samples (n = 1), not completing the whole caffeine intervention (n = 1), or who were clear outliers for an insulin concentration (n = 1). Thus, 45 subjects were included in the analysis. Caffeine-containing products (other than those provided) were prohibited during the entire trial. Venous blood samples were collected after an overnight fast. Plasma glucose concentrations were measured using the glucose hexokinase method. Serum insulin concentrations were measured using an immunoradiometric assay (Medgenix Biosource Diagnostics, Fleuris, Belgium).

In study 1, treatment responses were compared using paired t tests. In study 2, we tested for overall treatment effects using ANOVA. All reported P values were two sided, and P values <0.05 were considered statistically significant.

Study 1

Of the participants that completed the study, 62% were women, mean (±SD) age was 37 ± 12 years, and mean BMI was 23 ± 3 kg/m2. After 2 weeks, coffee consumption tended to lead to higher fasting glucose concentrations, but no appreciable effect was observed after 4 weeks (Table 1). Fasting insulin concentrations, measured only after 4 weeks, were higher after the coffee period than after the no coffee period (Table 1). Tests for carry-over effects did not indicate that these existed (insulin: P = 0.79; glucose: P = 0.27).

Study 2

Of the participants that completed the study, 56% were women, mean age was 40 ± 14 years, and mean BMI was 24 ± 3 kg/m2. Fasting glucose concentrations were similar after the caffeine, coffee, and placebo period (Table 1). Compared with the placebo period, fasting insulin concentrations tended to be higher after the coffee and caffeine periods (Table 1).

We found that high coffee consumption for 4 weeks increased fasting insulin concentrations compared with coffee abstinence. Consumption of somewhat weaker coffee and caffeine intake were nonsignificantly associated with higher fasting insulin concentrations. No substantial effects of coffee or caffeine on fasting glucose concentrations were observed.

The increased fasting insulin concentration after high coffee consumption in our study probably reflects decreased insulin sensitivity. In short-term metabolic studies, caffeine intake acutely lowered insulin sensitivity over 100–180 min (79). In a study of 5 days of caffeine intake, complete tolerance to the effects of caffeine on fasting glucose concentrations developed (19), but effects on norepinephrine and free fatty acid concentrations partly remained for the high-dose caffeine treatment. Thus, effects of high amounts of caffeine on catecholamines and free fatty acids may have contributed to a decrease in insulin sensitivity in our studies. However, we cannot completely exclude the possibility that the elevated insulin concentrations after coffee consumption were due to higher insulin secretion (20) or to reduced hepatic insulin clearance as a result of increased free fatty acid concentrations (21).

Our findings seem to be at variance with the inverse association between coffee consumption and risk for type 2 diabetes that has been observed in cohort studies (26). Several factors may contribute to this discrepancy. First, the results of cohort studies may reflect the effects of decades of regular coffee consumption, whereas the present study compared 2–4 weeks of coffee consumption with 2–4 weeks of coffee abstinence. Second, we cannot exclude the possibility that the rapid transition to high coffee consumption (equivalent to ∼13 conventional cups of coffee in study 1) in our studies had detrimental effects on insulin sensitivity. For example, experienced psychological stress may have lowered insulin sensitivity through increased stress hormone concentrations. Third, habitual coffee consumption may improve aspects of glucose metabolism that are not reflected in the outcome parameters of the present study (for example, postprandial glucose metabolism).

In conclusion, the present results indicate that tolerance to the adverse effects of high coffee consumption on insulin-glucose homeostasis does not develop within a 4-week period. This stresses that it is premature to advocate high coffee consumption as a means to lower risk for type 2 diabetes. Long-term trials of coffee consumption that include detailed measures of insulin sensitivity and glucose metabolism are warranted to elucidate the apparent discrepancy with studies that observed an inverse association between habitual coffee consumption and risk for type 2 diabetes.

Table 1—

Fasting glucose and insulin concentrations after paper-filtered regular coffee consumption, caffeine intake, and coffee/caffeine abstinence

Control periodCoffee periodCaffeine periodP*
Study 1 (n = 26    
    Plasma glucose (mmol/l)     
        After 2 weeks 5.0 ± 0.3 5.2 ± 0.6 — 0.08 
        Difference with no coffee — 0.2 (0.0–0.4) — — 
        After 4 weeks 5.1 ± 0.3 5.1 ± 0.4 — 0.94 
        Difference with no coffee — 0.0 (−0.1–0.1) — — 
    Plasma insulin (pmol/l)     
        After 4 weeks 57.5 ± 18.8 71.4 ± 31.2 — 0.002 
        Difference with no coffee — 13.9 (5.9–21.9) — — 
Study 2 (n = 45    
    Plasma glucose (mmol/l) 5.0 ± 0.4 5.0 ± 0.5 5.1 ± 0.4 0.42 
        Difference with placebo — 0.0 (−0.1–0.1) 0.1 (0.0–0.2) — 
    Plasma insulin (pmol/l) 51.5 ± 16.0 56.6 ± 22.4 54.2 ± 21.1 0.15 
        Difference with placebo — 4.9 (0.0–9.8) 2.8 (−2.1–7.7) — 
Control periodCoffee periodCaffeine periodP*
Study 1 (n = 26    
    Plasma glucose (mmol/l)     
        After 2 weeks 5.0 ± 0.3 5.2 ± 0.6 — 0.08 
        Difference with no coffee — 0.2 (0.0–0.4) — — 
        After 4 weeks 5.1 ± 0.3 5.1 ± 0.4 — 0.94 
        Difference with no coffee — 0.0 (−0.1–0.1) — — 
    Plasma insulin (pmol/l)     
        After 4 weeks 57.5 ± 18.8 71.4 ± 31.2 — 0.002 
        Difference with no coffee — 13.9 (5.9–21.9) — — 
Study 2 (n = 45    
    Plasma glucose (mmol/l) 5.0 ± 0.4 5.0 ± 0.5 5.1 ± 0.4 0.42 
        Difference with placebo — 0.0 (−0.1–0.1) 0.1 (0.0–0.2) — 
    Plasma insulin (pmol/l) 51.5 ± 16.0 56.6 ± 22.4 54.2 ± 21.1 0.15 
        Difference with placebo — 4.9 (0.0–9.8) 2.8 (−2.1–7.7) — 

Data are means ± SD or difference (95% CI).

*

P values are for paired t test comparison for study 1 and for existence of an overall treatment effect from ANOVA for study 2.

Numbers are lower because of missing blood samples for glucose data after 2 weeks in study 1 (n = 25 for the coffee period, n = 24 for the no coffee period) and for glucose data in study 2 (n = 41 for the coffee period, n = 43 for the caffeine period, and n = 42 for the placebo period).

Mean of values obtained after 25 and 28 days.

Supported by the Wageningen Center for Food Sciences, an alliance of Dutch food industries; TNO Nutrition and Food Research; Maastricht University; and Wageningen University and Research Center, with financial support from the Dutch government.

We thank the volunteers for their participation, all those involved at TNO Nutrition and Food Research for their dedication, and the laboratory staff at the Vrije Universiteit Medical Center for their careful analysis.

1
Arnlov J, Vessby B, Riserus U: Coffee consumption and insulin sensitivity.
JAMA
291
:
1199
–1201,
2004
2
van Dam RM, Feskens EJ: Coffee consumption and risk of type 2 diabetes mellitus.
Lancet
360
:
1477
–1478,
2002
3
Rosengren A, Dotevall A, Wilhelmsen L, Thelle D, Johansson S: Coffee and incidence of diabetes in Swedish women: a prospective 18-year follow-up study.
J Intern Med
255
:
89
–95,
2004
4
Salazar-Martinez E, Willett WC, Ascherio A, Manson JE, Leitzmann MF, Stampfer MJ, Hu FB: Coffee consumption and risk for type 2 diabetes mellitus.
Ann Intern Med
140
:
1
–8,
2004
5
Tuomilehto J, Hu G, Bidel S, Lindstrom J, Jousilahti P: Coffee consumption and risk of type 2 diabetes mellitus among middle-aged Finnish men and women.
JAMA
291
:
1213
–1219,
2004
6
Carlsson S, Hammar N, Grill V, Kaprio J: Coffee consumption and risk of type 2 diabetes in Finnish twins (Letter).
Int J Epidemiol
33
:
616
–617,
2004
7
Greer F, Hudson R, Ross R, Graham T: Caffeine ingestion decreases glucose disposal during a hyperinsulinemic-euglycemic clamp in sedentary humans.
Diabetes
50
:
2349
–2354,
2001
8
Keijzers GB, De Galan BE, Tack CJ, Smits P: Caffeine can decrease insulin sensitivity in humans.
Diabetes Care
25
:
364
–369,
2002
9
Thong FS, Derave W, Kiens B, Graham TE, Urso B, Wojtaszewski JFP, Hansen BF, Richter EA: Caffeine-induced impairment of insulin action but not insulin signaling in human skeletal muscle is reduced by exercise.
Diabetes
51
:
583
–590,
2002
10
Cheraskin E, Ringsdorf WM Jr, Setyaadmadja AT, Barrett RA: Effect of caffeine versus placebo supplementation on blood-glucose concentration.
Lancet
1
:
1299
–1300,
1967
11
Jankelson OM, Beaser SB, Howard FM, Mayer J: Effect of coffee on glucose tolerance and circulating insulin in men with maturity-onset diabetes.
Lancet
1
:
527
–529,
1967
12
Wachman A, Hattner RS, George B, Bernstein DS: Effects of decaffeinated and nondecaffeinated coffee ingestion on blood glucose and plasma radioimmunoreactive insulin responses to rapid intravenous infusion of glucose in normal man.
Metabolism
19
:
539
–546,
1970
13
Pizziol A, Tikhonoff V, Paleari CD, Russo E, Mazza A, Ginocchio G, Onesto C, Pavan L, Casiglia E, Pessina AC: Effects of caffeine on glucose tolerance: a placebo-controlled study.
Eur J Clin Nutr
52
:
846
–849,
1998
14
Mougios V, Ring S, Petridou A, Nikolaidis MG: Duration of coffee- and exercise-induced changes in the fatty acid profile of human serum.
J Appl Physiol
94
:
476
–484,
2003
15
Lane JD, Barkauskas CE, Surwit RS, Feinglos MN: Caffeine impairs glucose metabolism in type 2 diabetes (Brief Report).
Diabetes Care
27
:
2047
–2048,
2004
16
Robertson D, Wade D, Workman R, Woosley RL, Oates JA: Tolerance to the humoral and hemodynamic effects of caffeine in man.
J Clin Invest
67
:
1111
–1117,
1981
17
Urgert R, van Vliet T, Zock PL, Katan MB: Heavy coffee consumption and plasma homocysteine: a randomized controlled trial in healthy volunteers.
Am J Clin Nutr
72
:
1107
–1110,
2000
18
Verhoef P, Pasman WJ, Van Vliet T, Urgert R, Katan MB: Contribution of caffeine to the homocysteine-raising effect of coffee: a randomized controlled trial in humans.
Am J Clin Nutr
76
:
1244
–1248,
2002
19
Denaro CP, Brown CR, Jacob P 3rd, Benowitz NL: Effects of caffeine with repeated dosing.
Eur J Clin Pharmacol
40
:
273
–278,
1991
20
Bruton JD, Lemmens R, Shi CL, Persson-Sjogren S, Westerblad H, Ahmed M, Pyne NJ, Frame M, Furman BL, Islam MS: Ryanodine receptors of pancreatic beta-cells mediate a distinct context-dependent signal for insulin secretion.
FASEB J
17
:
301
–303,
2003
21
Wiesenthal SR, Sandhu H, McCall RH, Tchipashvili V, Yoshii H, Polonsky K, Shi ZQ, Lewis GF, Mari A, Giacca A: Free fatty acids impair hepatic insulin extraction in vivo.
Diabetes
48
:
766
–774,
1999

A table elsewhere in this issue shows conventional and Système International (SI) units and conversion factors for many substances.