Prior studies have established the ergogenic benefits of caffeine for both high-intensity short-duration performances [1–3], as well as endurance performance [4–6]. However, based on two studies that have reported individual data [3, 6], approximately 30% of participants derive no ergogenic effects from caffeine ingestion. Doherty et al. [3] observed that four out of 14 subjects had no appreciable change in time to fatigue during running at a supramaximal workload following ingesting of caffeine. Meyers and Cafarelli [6] investigated the effects of acute caffeine supplementation on time to fatigue during repetitive quadriceps contractions. Three out of the 10 study participants did not respond to the caffeine or exhibited a worse performance under caffeine versus the placebo. Furthermore, not all studies report a significant ergogenic effect [7–9]. Beck et al. [7] did not observe any effect of caffeine on either maximal bench press strength or time to fatigue at 85% VO₂max. Jacobson et al. [8] observed that caffeine had no additive effect on time trial performance when administered with pre-exercise carbohydrate or fat feedings. Finally, caffeine had no effect on peak power output or total work in a short-duration maximal cycling test [9]. Thus, the ergogenic effect of caffeine, while evident, is highly variable. The cause(s) of this variability across individuals remains unclear, and it is unknown if any of this variance is accounted for by genetic polymorphisms.

Cytochrome P450 is a hepatic enzyme that is a key component of caffeine metabolism. A (C/A) single nucleotide polymorphism at intron 1 of the cytochrome P450 gene influences the inducibility of this enzyme, with the C variant affecting a slower caffeine metabolism following caffeine ingestion in smokers [10]. This polymorphism has clinical importance, as caffeine increases risk for cardiovascular disease in individuals who possess the C variant, but not in individuals homozygous for the A variant [11, 12], presumably due to a slower caffeine clearance in the former group. In contrast, Hallstrom et al. [13] observed that coffee consumption contributes to low bone mineral density in individuals homozygous for the A variant, and not those who possess the C allele. Therefore, this genetic polymorphism is a potential candidate to explain variability of ergogenic response to caffeine supplementation. The purpose of the present study was to determine if this specific CYP1A2 polymorphism influences the ergogenic effect of caffeine supplementation in trained cyclists.

The major finding of the present study is that caffeine affects 40-kilometer time trial performance in cyclists homozygous for the A variant to a greater degree than those who possess the C variant. Specifically, caffeine decreased 40-km time by an average of 3.8 minutes in the AA homozygotes as compared to 1.3 minutes in the C allele carriers. To our knowledge, this is the first study to implicate a specific polymorphism as a potential cause of the variation in the ergogenic effect of caffeine supplementation.

Sachse et al. [10] observed slower caffeine metabolism in C allele carriers who smoke, suggesting that this CYP1A2 polymorphism may affect the inducibility of the Cytochrome P450 enzyme. Caffeine has also been shown to increase risk of heart disease in C allele carriers but not AA homozygotes [11, 12], ostensibly because caffeine is metabolized at a higher rate in the AA homozygotes. Given these prior findings, it could be hypothesized that a slower metabolism would be advantageous for maximizing the ergogenic benefit of caffeine. Alternatively, Hallstrom et al. [13] found that coffee consumption was associated with decreased bone mineral density in AA homozygotes, but not C allele carriers. The authors speculated that the rapid accumulation of caffeine metabolites may have been responsible for this finding [13]. In support of this contention, paraxanthine and theophylline (downstream metabolites of caffeine metabolism) have higher binding affinities with adenosine receptors than caffeine [16]. Thus, it is possible that a faster caffeine metabolism in AA homozygotes created a more rapid production of paraxanthine and/or theophylline and therefore enhanced the ergogenic effect. This possibility is speculative as no markers of caffeine metabolism were available. Future studies should determine caffeine metabolism during exercise across these genotypes to better determine the mechanism of the observed effect.

Despite the fact that there was a significant Genotype × Treatment interaction for 40-km time, it should also be noted that the AA homozygotes had a slower placebo 40-km time and the caffeine supplementation served to decrease 40-km time for AA homozygotes to a level comparable to C allele carriers (Figure 1). This raises the concern that the results were driven by a difference in cycling performance capabilities between the two groups, rather than the genetic polymorphism. Collomp et al. [17] observed that caffeine improved swimming velocity in trained, but not untrained swimmers. O'Rourke et al. [18] observed a similar 5-km performance improvement from caffeine in both well-trained and recreational runners. Thus, one would expect performance capabilities to have no effect on caffeine response, or to affect it in the opposite direction of what was observed in the present study. Nonetheless, we further addressed the difference in performance between the two groups in a follow-up analysis. Because the higher 40-km time in AA homozygotes was primarily driven by four cyclists whose 40 k times during the placebo trial were greater than 80 minutes (see Figure 2), we removed these four subjects from the dataset for the follow-up analysis. This resulted in similar 40-km times in the placebo condition between the two groups, yet caffeine still had a significantly (p = 0.047) greater effect in AA homozygotes (caffeine = 70.5 ± 3.0 min, placebo = 73.5 ± 3.8 min) compared to the C allele carriers (caffeine = 70.9 ± 4.3 min, placebo = 72.2 ± 4.2 min). Caffeine resulted in at least a 1-minute improvement in 40 k time in all but one of the AA homozygotes; whereas only about half of C allele carriers responded to that extent (Figure 2). Thus, our data support the contention that it is the genetic polymorphism and not the performance capabilities of the respective groups that explain our observations.

Although data from the present study clearly suggest a potential role of this polymorphism in influencing the ergogenic response of caffeine in cyclists, care should be taken in extrapolating these findings. It is unknown if there is a similar genetic influence for other modes of exercise and/or for short-duration high-intensity exercise. Furthermore, we used trained cyclists in the present study and our findings cannot be extrapolated to sedentary individuals. Neither can it be suggested that this polymorphism is the only source of variation or even the only source of genetic variation involved. Finally, although we have outlined a potential mechanism that explains the current findings, it should be emphasized that the mechanistic causes of our findings cannot be determined from the present data. Future studies should determine whether these findings can be replicated using other modes of exercise and in other populations. Other candidate polymorphisms should also be identified and evaluated.

In summary, data from the present study suggest that caffeine potentiates a larger ergogenic effect for cycling performance in individuals homozygous for the A variant of the studied CYP1A2 polymorphism. The mechanism(s) of this selective ergogenic effect are unknown and future studies should seek to establish the impact of this polymorphism on caffeine metabolism during exercise. While these findings elucidate a possible source of variance in the ergogenic effect of caffeine, other factors, including other genetic polymorphisms, may also influence caffeine responses during exercise.