METHODS

Study sample and protocol

Twelve normal healthy habitual caffeine drinkers (four men) of mean age of 33.2 (9.3) years were recruited to the study. All were non-smoking non-asthmatic subjects with normal spirometric parameters and no recent history of respiratory tract infections. Subjects abstained from dietary methylxanthines for at least 8 hours before the study. The study was approved by the institutional ethics committee and all subjects gave written informed consent.

There were three arms to the study, which were randomised in a single blind, placebo controlled, crossover manner and comprised: (1) coffee 2.4 g with an approximate caffeine content of 100 mg per 200 ml cup (Espresso filtered coffee, Harris Ltd, Sydney) and a matched placebo capsule (Glucodin (glucose), Boots Healthcare, Australia); (2) decaffeinated coffee (<3 mg caffeine per 200 ml cup, Moccona, Douwe Egberts Joure, Holland) and a capsule of 200 mg caffeine (No Doz, Key Pharmaceuticals, Sydney); or (3) decaffeinated coffee and a placebo capsule. Subjects consumed a further aliquot of the same beverage after the first hour.

At baseline, eNO and serum caffeine levels were measured. Exhaled NO was then measured every hour for 4 hours. Serum caffeine samples were taken 1 hour after caffeine intake to sample peak serum levels of caffeine.

Measurement of exhaled NO

Exhaled NO was measured by chemiluminescence (Model 2107, Dasibi Environmental Corporation, Glendale, CA, USA) with subjects exhaling slowly at a flow rate of 5 l/min from total lung capacity over 30–40 s with constant raised oral pressure avoiding nasal contamination, following ERS recommendations.⁸ The mean of three successive peak plateau readings was used for analysis. Ambient NO levels were recorded and, if >40 ppb, the analyser was flushed with NO-free gas.

Measurement of serum caffeine levels

Serum caffeine concentrations taken at baseline and 1 hour were analysed by gas chromatography-mass spectroscopy.

Spirometric testing

Spirometric tests were performed using a dry wedge spirometer (Vitalograph, UK) to ensure that all subjects had a normal forced expiratory volume in 1 second (FEV₁).

Statistical analysis

Exhaled NO values were log transformed to the normal distribution and analysed by two way repeated measures ANOVA to compare the differences between the three treatment groups over time. Exhaled NO levels were expressed as mean (SE) parts per billion (ppb) and changes in exhaled NO displayed as mean (SE) % change from baseline. Serum caffeine levels were reported as mean (SE) mg/l and analysed by paired t test. Significance was taken as p<0.05, with post hoc Bonferroni corrected paired t test applied to significant ANOVA results.

RESULTS

Subjects

Eleven of the 12 subjects completed the study; one subject was withdrawn because of a respiratory tract infection.

Exhaled NO

Baseline eNO results were not significantly different between the study days in any of the three phases. After coffee or caffeine consumption there was a significant decrease in eNO levels compared with placebo (p=0.008, two way ANOVA). Mean eNO levels decreased significantly in the first hour after coffee or caffeine consumption compared with the placebo phase and caused a mean percentage fall of 13.5 (4.0)% and 19.0 (3.8)% (p=0.009 and p=0.001 respectively, post hoc t tests, fig 1). Mean eNO levels during the coffee and caffeine phase remained consistently lower than baseline levels during the following 3 hours, although the difference was not statistically significant.

Serum caffeine levels

Baseline serum caffeine levels did not differ significantly between study days. The mean (SE) serum caffeine level changed after 1 hour during the three phases: (1) caffeinated coffee and placebo from 0.6 (1.3) mg/l to 2.8 (0.8) mg/l, p=0.002; (2) decaffeinated coffee and caffeine from 0.6 (1.2) mg/l to 3.7 (0.9) mg/l, p=0.003; (3) decaffeinated coffee and placebo from 0.5 (1.3) mg/l to 0.4 (0.3) mg/l, p=0.21. No relationship was found between serum caffeine levels and change in eNO.

DISCUSSION

This study shows that caffeine significantly decreases eNO levels, either as 200 mg caffeine or as coffee with an average caffeine content of 100 mg. There are many other additional components of coffee, but none have a known mechanism which would cause NO to fall. The mechanism for this reduction in eNO is not certain, but it is known that at doses similar to those used in our study caffeine antagonises adenosine receptors. At higher concentrations caffeine causes inhibition of the breakdown of cAMP and cGMP, but it may also inhibit guanylate cyclase.^5,6,9

Adenosine enhances NO production from endothelial cells through adenosine A_2a receptors, while adenosine A₁ receptors decrease it.⁷ Thus, caffeine could antagonise A_2a receptors if these are active in airway epithelium, reducing the generation of NO.⁷ If caffeine acts by adenosine receptor antagonism, then A₁ receptors are less likely to be involved. Direct airway AMP challenge has also been associated with a fall in eNO, but this was interpreted as being related to airway bronchoconstriction.¹⁰

An alternative mechanism by which caffeine could alter NO generation is through the inhibition of the phosphodiesterases and hence the breakdown of the secondary messengers cGMP and cAMP. However, theophylline inhibits soluble guanylate cyclase thus reducing the formation of cGMP, so the effects of caffeine on cGMP may be mixed.^5,9 Cyclic GMP upregulates ecNOS generation while cAMP may downregulate it,^5,6 but these effects are at the level of protein translation which would not be expected to be seen for some hours. It would appear therefore that the immediate effect of caffeine observed in this study is more compatible with a reduction in cGMP via inhibition of soluble guanylate cyclase.⁹

Spirometric tests were not performed so as not to alter the levels of eNO spuriously, as some reports have suggested a fall in eNO after repeated spirometry alone.¹¹ Silkoff et al measured exhaled NO both after repeated spirometric testing and β₂ adrenergic agonist inhalation in asthmatic and normal subjects and found that spirometric testing decreased eNO levels while the β₂ agonist increased them.¹¹ Others have reported that eNO levels are reduced shortly after bronchoconstriction to direct and indirect stimuli in subjects with asthma, and suggested that eNO levels can be modulated by bronchial tone.¹⁰ The first measurement after intervention was at the end of the first hour, which was at the time when previous studies have indicated that eNO is no longer significantly different from baseline. In addition, the control phase of the experimental design indicates that the effects on eNO are due to caffeine and not to repeated spirometric tests (fig 1).

Decaffeinated coffee had no effect on eNO levels, suggesting that caffeine is the principal agent causing the decrease in eNO. Caffeine is associated with a brief reduction in eNO levels over 4 hours in normal subjects, which suggests that it should be withheld for at least this time period before taking measurements in normal and probably in asthmatic subjects; however, this latter group was not assessed in this study.