Share this article on:

Caffeine, CYP1A2 Genotype, and Endurance Performance in Athletes

GUEST, NANCI1; COREY, PAUL2; VESCOVI, JASON3; EL-SOHEMY, AHMED1

Medicine & Science in Sports & Exercise: August 2018 - Volume 50 - Issue 8 - p 1570–1578
doi: 10.1249/MSS.0000000000001596
Basic Sciences

Purpose Many studies have examined the effect of caffeine on exercise performance, but findings have not always been consistent. The objective of this study was to determine whether variation in the CYP1A2 gene, which affects caffeine metabolism, modifies the ergogenic effects of caffeine in a 10-km cycling time trial.

Methods Competitive male athletes (n = 101; age = 25 ± 4 yr) completed the time trial under three conditions: 0, 2, or 4 mg of caffeine per kilogram body mass, using a split-plot randomized, double-blinded, placebo-controlled design. DNA was isolated from saliva and genotyped for the −163A > C polymorphism in the CYP1A2 gene (rs762551).

Results Overall, 4 mg·kg−1 caffeine decreased cycling time by 3% (mean ± SEM) versus placebo (17.6 ± 0.1 vs 18.1 ± 0.1 min, P = 0.01). However, a significant (P <0.0001) caffeine–gene interaction was observed. Among those with the AA genotype, cycling time decreased by 4.8% at 2 mg·kg−1 (17.0 ± 0.3 vs 17.8 ± 0.4 min, P = 0.0005) and by 6.8% at 4 mg·kg−1 (16.6 ± 0.3 vs 17.8 ± 0.4 min, P < 0.0001). In those with the CC genotype, 4 mg·kg−1 increased cycling time by 13.7% versus placebo (20.8 ± 0.8 vs 18.3 ± 0.5 min, P = 0.04). No effects were observed among those with the AC genotype.

Conclusion Our findings show that both 2 and 4 mg·kg−1 caffeine improve 10-km cycling time, but only in those with the AA genotype. Caffeine had no effect in those with the AC genotype and diminished performance at 4 mg·kg−1 in those with the CC genotype. CYP1A2 genotype should be considered when deciding whether an athlete should use caffeine for enhancing endurance performance.

1Department of Nutritional Sciences, Faculty of Medicine, University of Toronto, Toronto, Ontario, CANADA;

2Department of Statistical Sciences, Dalla Lana School of Public Health, University of Toronto, Toronto, Ontario, CANADA; and

3Faculty of Kinesiology and Physical Education, University of Toronto, Toronto, Ontario, CANADA

Address for correspondence: Ahmed El-Sohemy, Ph.D., Department of Nutritional Sciences, University of Toronto, 150 College Street, Room 350 Toronto, ON M5S 3E2, Canada; E-mail: a.el.sohemy@utoronto.ca.

Submitted for publication December 2017.

Accepted for publication February 2018.

Caffeine is frequently used by athletes because of its reported performance-enhancing or ergogenic effects (1–15). Numerous studies have investigated the effect of supplemental caffeine on aerobic endurance performance, and while most reliably show performance enhancement with caffeine use (1–15), there is considerable interindividual variability as to the magnitude of these effects (16–20). For example, in a study of caffeine effects in runners, Graham et al. (5) reported that endurance benefits were associated with caffeine supplementation overall, but the magnitude of the improvements ranged from 5% to 87% and 10% to 156% in running and cycling time-to-exhaustion trials, respectively. Similarly, Doherty et al. (6) found that during treadmill running time-to-exhaustion, 9 out of 14 subjects improved, while 5 subjects did not, during the caffeine versus placebo trials. Wiles et al. (7) also found that the mean improvement during a 1-km cycling time trial (TT) was 3% overall under caffeine conditions, but individual results ranged from one subject performing worse to another, improving their performance by 6%.

In contrast to most caffeine–performance studies, no ergogenic effect of caffeine was reported by Roelands et al. (19) in a study involving trained male cyclists. The authors concluded that interindividual differences in response to caffeine might be responsible for the lack of overall performance improvement, as 50% of subjects improved while 50% worsened, in the caffeine compared with the placebo trial. Similarly, Skinner et al. (16) found no effect of caffeine at 2, 4, or 6 mg·kg−1 versus placebo in a rowing TT, which may have been due, in part, to the large variation in individual response to caffeine. The authors noted that this consideration is often overlooked in caffeine performance studies, and due to the infrequent reporting of individual data, it is difficult to determine the extent to which variation in responses may be occurring. The performance of some individuals is often in stark contrast to the average findings reported, which may conclude beneficial, detrimental, or no effect of caffeine on performance.

Studies that report on the effects of caffeine on performance have been inconsistent despite having similar study designs, subjects, and dose of caffeine. These inconsistencies might be due, in part, to interindividual differences in caffeine metabolism or caffeine response. Over 95% of caffeine is metabolized by the CYP1A2 enzyme, which is encoded by the CYP1A2 gene, and is involved in the demethylation of caffeine into the primary metabolites paraxanthine, theophylline, and theobromine (21). The −163A > C (rs762551) single nucleotide polymorphism (SNP) has been shown to alter CYP1A2 enzyme inducibility and activity (22,23) and has been used to categorize individuals as “fast” or “slow” metabolizers of caffeine. Individuals with the AC or CC genotype (slow metabolizers) have an elevated risk of myocardial infarction (24), hypertension (25), and prediabetes (26) with increasing caffeinated coffee consumption, whereas those with the AA genotype show no such risk. In addition, a few studies have shown that the rate of caffeine metabolism could also have implications for sports performance, but the findings remain equivocal (12,27–29).

The objective of this study was to determine the effects of low (2 mg·kg−1) or moderate (4 mg·kg−1) doses of caffeine supplementation on endurance performance, and whether variation in the CYP1A2 gene modifies these effects among competitive male athletes recruited from a variety of sports.

Back to Top | Article Outline

METHODS

Subjects and recruitment

Recruitment was conducted at the University of Toronto, Ryerson and York University campuses, the Canadian Sport Institute of Ontario, local running/triathlon clubs, and training gyms using posted flyers. A standardized e-mail with study details and contact information was also sent to head coaches, program directors of sports teams or clubs, and some professional sport organizations with eligible athletes. Ethics approval was obtained from the University of Toronto Institutional Review Board, and the study was registered with clinicaltrials.gov (NCT 02109783). All subjects provided written informed consent and were informed that they could terminate their participation in the study at any time.

A total of 113 competitive male athletes from a variety of sports participated in the present study. Subjects were recruited from a wide range of sports that could be classified into three categories: endurance (e.g., marathon, triathlon, cycling, and cross-country skiing), power (e.g., boxing, volleyball, dragon-boat, and powerlifting), or mixed (e.g., soccer, rugby, basketball, and swimming). All participants were training and/or competing for ≥8 h·wk−1, 9 out of 12 months per year, and for at least 3 yr in their given sport. Eight athletes dropped out of the study due to a sport-related injury (n = 3), school or work demands (n = 2), unwillingness to abstain from caffeine (n = 2), or relocation (n = 1). Four subjects were excluded because of incomplete data. The remaining 101 athletes had a mean ± SD age of 25 ± 4 yr and body mass of 81.3 ± 12.4 kg.

Back to Top | Article Outline

Experimental design

A split-plot randomized, double-blinded, placebo-controlled study design was used. Subjects completed four visits (~90–120 min each) that were approximately 1 wk apart, in the exercise laboratory at the Goldring Centre for High Performance Sport at the University of Toronto. During the first laboratory visit, each subject had descriptive and anthropometric data collected, completed a maximal aerobic capacity test (V˙O2peak) and completed a questionnaire on general health, caffeine intake habits, and sport history. Subjects also provided a saliva sample for DNA analysis. Testing took place on weekdays and weekends, and the treatment visits were scheduled at the same time of day, every 7 d, for each athlete. Participants were instructed to maintain their regular diet and sleeping habits, avoid strenuous activity 48 h before each visit, and abstain from caffeine 1 wk before the first visit and for the duration of the data collection (4 wk total). To ensure dietary consistency before testing across all visits, participants were advised on their first visit to consume meal(s) that could be easily replicated for all subsequent treatment visits. Participants were also reminded of their required meal composition by e-mail or text message 1 d before each visit. On treatment visits 2–4, subjects were randomly assigned to ingest capsules containing either anhydrous caffeine (A&C American Chemicals Ltd., Saint-Laurent, Quebec, Canada) at 2 or 4 mg·kg−1 body mass or placebo (PLAC). The PLAC (dextrose) capsule was tasteless and had the same volume and color as the caffeine. After ingestion, the subjects sat quietly (completing questionnaires or using e-devices) in the laboratory for 25 min before commencing their warm-up and four exercise tests. Blood pressure and HR were measured after capsule ingestion and 3 min of sitting quietly, and again 20 min later, just before warm-up. This protocol was repeated three times, one for each treatment (0, 2, or 4 mg·kg−1 caffeine).

Back to Top | Article Outline

Parameters of assessment

Before testing, athletes were led through a brief standardized warm-up that consisted of light cycling and stretching for approximately 7 min. Physical tests were conducted in a standard order to minimize fatigue: 1) vertical jump, 2) handgrip, 3) Wingate, and 4) 10-km cycling TT. Only the results of the 10-km cycling TT are reported here.

Back to Top | Article Outline

Anthropometry

Height was measured with a Harpenden stadiometer (Holtain Ltd., Crymych, UK), and body mass was measured by an electronic floor scale (AND FW-150K, A&D Ltd., Tokyo, Japan). Total body fat percentage was measured by BC-558 IronMan Segmental Body Composition Monitor (Tanita Corporation of America, Inc., Arlington Heights, IL).

Back to Top | Article Outline

Maximal exercise test (V˙O2peak)

Subjects began the test at a work rate of 50 W on a mechanically weighted and braked cycle ergometer (Monark Ergomedic 839E, Monark Exercise AB, Vansbro, Sweden), with load increases of 50 W each minute for the first 2 min, then 25 W each minute thereafter until volitional exhaustion. Gas exchange was measured by a portable metabolic system (Cortex Metamax 3B®, CORTEX Biophysik GmbH, Leipzig, Germany), and maximal oxygen uptake (V˙O2peak) was defined as the highest 1-min oxygen value obtained during the test. V˙O2peak power (W) was calculated by measuring the power output (W) at V˙O2peak, and end power W power was calculated as the power output (W) at volitional fatigue. HR was monitored using a Polar Heart Rate Monitor (Polar Electro Inc., Bethpage, NY).

Back to Top | Article Outline

TT

Subjects commenced the 10-km cycling TT (last exercise) when blood lactate levels reached <2.5 mmol·L−1 from the prior Wingate test. The TT was conducted by setting the Ergomedic 839 E stationary bike to a constant resistance or power output, and each subject cycled 10-km at the specified resistance (W). Resistance was set at 65% Wpower for all subjects as calculated from the V˙O2peak test, which was the equivalent of 65%–69% V˙O2peak (varying between subjects but identical % used within each subject for all three treatment visits). The on-board computer automatically controls the degree of resistance by applying varying amounts of braking force on the belt. The computer of the stationary bike calculates the speed of travel based on the cadence of pedalling (rpm), where a faster cadence would result in a faster speed. The 10-km TT requires 1667 rotations (6 m per rotation) to be completed; therefore, the power output did not affect the speed of the bike. Speed was altered only by how fast the subject pedalled (cadence). Therefore, different cadences (rpm) would result in different completion times for the 10-km TT. Subjects were blinded to time, speed, and HR but were able to see distance traveled. Water was available ad libitum throughout the TT. HR was monitored throughout the test using a Polar Heart Rate Monitor (Lake Success, NY). Subjects estimated their RPE on the basis of Borg’s rating scale (score ratings from 6 to 20, where 6 is no exertion and 20 is extremely difficult) at 5 and 9 km.

Back to Top | Article Outline

Genotyping

Saliva samples were collected on visit 1 using the Oragene ON-500 kit (DNA Genotek, Ottawa, Ontario, Canada) for DNA isolation using standard procedures, as previously described (30). Genotyping of the rs762551 SNP in the CYP1A2 gene was conducted using the Sequenom MassArray platform, as we have described previously (30). Because there is evidence of a difference in enzyme activity between the three CYP1A2 genotypes (22,23), we grouped individuals into AA (fast), AC (heterozygous slow), and CC (homozygous slow) for all analyses.

Back to Top | Article Outline

Statistical analyses

Data were analyzed using the SAS statistical package (SAS 9.4, SAS Canada, Toronto, Ontario, Canada) and are presented as mean ± SEM unless stated otherwise. Descriptive data (height, body mass, age, body fat, V˙O2peak [L·min−1], V˙O2peak [mL·kg−1·min−1], dietary caffeine or caffeine used for sport, and sport type distribution) were compared between genotypes using ANOVA or for sport type, using chi-square. Body mass was log transformed before analysis, as it was not normally distributed. Using a classical split-plot design, the between-subject variance was used to compare mean cycling times across the three genotypes, whereas the within-subject variance was used to compare placebo and the two caffeine doses. The order of the three visits was randomized across the subjects, and visit was included as a covariate in all analyses. Randomization was done using balanced permutations blocked by time of entry (randomization.com). The outcome variable was 10-km TT time, and the initial analysis included the three predictor variables, caffeine, gene, and visit, along with the three 2-factor interactions and the one 3-factor interaction. After identifying a significant caffeine–gene (P < 0.0001) interaction, each genotype was analyzed separately. This model was also used to assess RPE and HR between genotypes and within subjects between visits and caffeine treatments. The main effect of caffeine was assessed across all genotypes combined, which left two predictor variables: caffeine and visit, and the caffeine–visit interaction, with TT time to completion as the outcome variable. Post hoc Tukey adjustments for multiple comparisons were performed for all analyses. All P values are two tailed, and P < 0.05 was used as the threshold for significance. Effect sizes (ES) are presented as standardized differences between caffeine treatments (all subjects combined or for individual genotypes) using Cohen’s d = (M 2M 1)/SDpooled, where SDpooled = √((SD1 2 + SD2 2)/2) (31). Cohen (31) suggested that 0.2 be considered a “small” ES, 0.5 a “medium” ES, and 0.8 a “large” ES. For significant genotype and treatment P values, the analysis of the effect of caffeine dose on the mean 10-km TT time was completed with and without an adjustment for the visit variable to establish whether visit was a confounder. This was completed for the main effect of caffeine for all subjects, as well as the effect of caffeine within each of the three genotypes. Sample size was determined by power analysis calculations using a power of 0.8 and a medium ES of 0.5. A power calculation based on two caffeine doses and three genotypes revealed that a sample size of 110 athletes will provide sufficient power for our analysis, based on a potential subject dropout rate of 10% (31).

Back to Top | Article Outline

RESULTS

Subject Characteristics

Of the 101 participants, 49% (n = 49) were homozygous for the A allele (AA), 43% (n = 44) were heterozygous (AC), and 8% (n = 8) were homozygous for the C allele (CC). These distributions are in Hardy–Weinberg equilibrium and similar to frequencies reported previously in some other populations (22,24). The rs762551 polymorphism in the CYP1A2 gene was initially used to identify fast and slow metabolizers of caffeine. We discovered that another SNP in CYP1A2, rs2472300, is in 100% linkage disequilibrium with rs762551. As such, either polymorphism can be used to identify fast or slow metabolizers of caffeine. For rs2472300, GG corresponds to fast metabolizers, whereas GA and AA are considered slow metabolizers. For rs762551, AA corresponds to fast metabolizers, whereas AC and CC are considered slow metabolizers. In the present study, we genotyped subjects for both rs2472300 and rs762551 SNP and found 100% concordance, but we report the results for rs762551 because it is the one more commonly reported (12,22–26,29).

Descriptive characteristics of the three genotypes are shown in Table 1. There were no significant differences between the three genotypes for age, height, body fat, V˙O2peak (L·min−1), V˙O2peak (mL·kg−1·min−1), dietary caffeine, caffeine used for sport, or percent distribution of sport type (endurance, power or mixed; X2 [4, N = 101] = 3.31). The breakdown for sport type for all participants was as follows: endurance 42% (e.g., running, cycling, and rowing), power 42% (e.g., baseball, powerlifting, and boxing), and mixed 16% (e.g., basketball, rugby, and hockey).

TABLE 1

TABLE 1

Back to Top | Article Outline

TT Performance

All subjects

The average 10-km TT times (n = 101) under the three treatments (0, 2, or 4 mg·kg−1 caffeine) are shown in Figure 1. There was a significant (P = 0.04) main effect for treatment (2 or 4 mg·kg−1 caffeine vs placebo) for all subjects, where 4 mg·kg−1 caffeine decreased 10-km TT time by 3% (0.5 min) compared with placebo (17.6 ± 0.3 vs 18.1 ± 0.1 min, P = 0.01). There was no significant difference between 2 mg·kg−1 and either 4 mg·kg−1 caffeine or placebo.

FIGURE 1

FIGURE 1

Back to Top | Article Outline

By genotype

When subjects were stratified by caffeine dose (0, 2, and 4 mg·kg−1) and genotype (Fig. 2), there was a significant overall difference between genotypes (P = 0.002), as well as a caffeine–gene (P < 0.0001) interaction. Thus, the three genotypes were analyzed individually to determine the effects of caffeine within each genotype.

FIGURE 2

FIGURE 2

Back to Top | Article Outline

AA genotype (fast metabolizers)

Among those with the AA genotype, the caffeine effect remained significant, where 2 mg·kg−1 caffeine decreased TT time by 4.8% (0.8 min) compared with placebo (17.0 ± 0.3 vs 17.8 ± 0.4 min, P = 0.0005), and by 6.8 (1.2 min) in 4 mg·kg−1 compared with placebo (16.6 ± 0.3 vs 17.8 ± 0.4 min, P < 0.0001), but no difference was observed between 2 and 4 mg·kg−1 caffeine.

Back to Top | Article Outline

AC genotype (slow metabolizers)

In those with the AC genotype, there was no caffeine effect on TT performance for any of the treatments (18.6 ± 0.4, 18.4 ± 0.5, and 18.0 ± 0.5, for 0, 2, and 4 mg·kg−1, respectively; P = 0.43).

Back to Top | Article Outline

CC genotype (slow metabolizers)

Among those with the CC genotype, 4 mg·kg−1 caffeine increased cycling time by 13.7% (2.5 min) compared with placebo (20.8 ± 0.8 vs 18.3 ± 0.5 min, P = 0.04), but no difference was observed between 2 mg·kg−1 and either 4 mg·kg−1 caffeine or placebo.

Back to Top | Article Outline

Change in TT time: placebo versus 2 mg or 4 mg·kg−1 caffeine

Figure 3 shows the average change in TT time (mean ± SEM) to completion between (A) the 2-mg·kg−1 and (B) the 4-mg·kg−1 caffeine dose compared with placebo. In Figure 3A (2 mg·kg−1 vs placebo), there were no differences between any of the genotypes. Figure 3B shows a significant (P = 0.001) overall difference between genotypes, such that those with the CC genotype had the greatest change in 10-km time (although a worsening of performance with caffeine) compared with changes in time in the opposite direction in those with the AA (−2.5 ± 1.0 min vs 1.2 ± 0.3 P < 0.0001) and AC (−2.5 ± 1.0 min vs 0.6 ± 0.4 min, P = 0.0015) genotypes, respectively.

FIGURE 3

FIGURE 3

Back to Top | Article Outline

TT performance scatterplot by genotype

Figure 4 shows individual data points representing 10-km time to completion for placebo (x-axis) and either (A) 2 mg·kg−1 or (B) 4 mg·kg−1 (y-axis), for all subjects by genotype (AA, AC, and CC). Data points below the line indicate faster times with caffeine. For those with the AA genotype, 35 (71%) and 40 (82%) out of 49 subjects performed better during 2 or 4 mg·kg−1, respectively, compared with placebo. In those with the AC genotype, 26 (59%) and 28 (64%) out of 44 subjects performed better during 2 or 4 mg·kg−1 caffeine, respectively, compared with placebo. In those with the CC genotype, 2 (25%) and 1 (12%) out of 8 subjects performed better during 2 or 4 mg·kg−1 respectively, compared with placebo.

FIGURE 4

FIGURE 4

Back to Top | Article Outline

RPE and HR

At 5-km, those with the AA genotype reported a 3% lower RPE in the 4-mg·kg−1 TT compared with placebo (14.3 ± 0.3 vs 14.8 ± 0.2, P = 0.03), but there was no difference between 2 mg·kg−1 (14.5 ± 0.3) and either 4 mg·kg−1 caffeine or placebo. There were no differences in those with the AC genotype (15.1 ± 0.3, 15.5 ± 0.3, and 15.0 ± 0.3) or CC genotype (14.1 ± 0.6, 14.3 ± 0.6, and 15.5 ± 0.3) between any of the TTs at 0, 2, or 4 mg·kg−1 caffeine, respectively. At 9 km, there were no differences in RPE between any of the treatments within any of the genotypes. HR analysis was determined in those with the AA (n = 46), AC (n = 42), and CC (n = 6) genotypes. In those with the AA genotype, there were no significant differences in HR (mean ± SEM) between any of the doses (167 ± 1, 169 ± 1, and 168 ± 1 bpm for 0, 2, and 4 mg·kg−1 caffeine, respectively). In those with the AC genotype, there was a 2.5% (4 bpm) increase in HR in 4 mg·kg−1 compared with 2 mg·kg−1 caffeine and placebo, respectively (171 ± 2 vs 167 ± 2 bpm, P = 0.007 and 167 ± 2 bpm, P = 0.005). In those with the CC genotype, there was a 2% (3 bpm) decrease in HR in those taking 4 mg·kg−1 caffeine compared with both placebo and 2 mg·kg−1 (160 ± 5 vs 157 ± 5 bpm, P = 0.03; 160 ± 5 vs 157 ± 5 bpm, P = 0.05), respectively.

Back to Top | Article Outline

ES

The main effect for caffeine (n = 101) in the 10-km TT at the 4-mg·kg−1 dose resulted in a 3% (0.5 min) improvement and small ES, d = 0.27, compared with placebo. However, in those with the AA genotype (n = 49), the 4.8% (0.8 min) improvement with 2 mg·kg−1 and the 6.8% (1.2 min) improvement with 4 mg·kg−1 both correspond to a medium ES: d = 0.4 and d = 0.63, respectively. In those with the CC genotype, the 13.7% impairment in performance in 4 mg·kg−1 vs placebo resulted in a very-large ES, d = 1.3.

Back to Top | Article Outline

Treatment blinding

We collected responses from 86 subjects post-TT, who were asked whether they thought they had consumed caffeine. Out of 172 caffeine trials, 31% (54) were correctly identified as caffeine containing. Among the other 118 caffeine trials, 81% (96) reported “no caffeine” and 19% (22) reported “maybe caffeine.” Only 3% (3) of subjects correctly identified all three trials (i.e., 2 caffeine, 1 placebo).

Back to Top | Article Outline

Familiarization

A learning or visit effect due to familiarization with cycling on the Monark bike for the three treatment visits (plus cycling V˙O2peak test) was expected in this group of athletes, where less than 6% were experienced cyclists. Although we observed a well-balanced allocation of the three doses of caffeine across all three visits where χ2 (4, N = 101) = 2.01, P = 0.73, we assessed the effect of visit within each genotype. In those with the AA genotype, TT time decreased across visits, likely as a learning or familiarization effect. However, TT cycling time also decreased within each visit as caffeine dose increased from 0 mg to 2 to 4 mg·kg−1, where ~33% of subjects would have ingested one of the three caffeine doses at each particular visit. Therefore, at each visit, each group consisting of one third of the 101 total subjects improved their performance in a dose-dependent manner after caffeine ingestion from 0 mg to 2 to 4 mg·kg−1. Table 2 shows the TT time and caffeine dose by genotype with and without adjusting for visit. When analyzing the effect of caffeine and visit in each genotype individually, the caffeine effect remained significant in both the AA (P < 0.0001) and the CC (P = 0.04) genotypes. The predictive power (R 2) dropped from 0.85 in the model with all subjects (not shown) to 0.78 in the model with AA genotypes and 0.80 in the model with CC genotypes. When visit was not included in the model (Table 2, unadjusted), the caffeine effect remained significant in the AA genotype (P < 0.0001) but decreased the predictive power of the model (R 2 = 0.70). However, in those with the CC genotype, the caffeine effect was no longer significant, and R 2 decreased to 0.56.

TABLE 2

TABLE 2

Back to Top | Article Outline

DISCUSSION

The current study examined the effects of caffeine and a genetic modifier of caffeine metabolism, CYP1A2 genotype, on 10-km cycling TT performance in competitive male athletes after ingestion of caffeine at 0, 2, and 4 mg·kg−1 body mass. Our results indicate that in the total population, caffeine is ergogenic to endurance cycling performance, with a 3% improvement in TT time at 4 mg·kg−1, but not at 2 mg·kg−1, which is consistent with previous studies using similar doses (2,9). However, we observed a significant caffeine–gene interaction where the improvements in performance were seen with both 2 and 4 mg·kg−1 caffeine, but only in those with the AA genotype who are “fast” metabolizers of caffeine. In that group, the 6.8% improvement in cycling time at 4 mg·kg−1 is greater than the 2%–4% mean improvement seen in approximately 30 cycling TT studies using similar doses (2,8,9,14,15,18,32). The improvement in performance that we observed in the entire population at 4 mg·kg−1 corresponds to a small ES, d = 0.27, but in those with the AA genotype the effect corresponds to a medium ES d = 0.63. Contrary to the beneficial effects we observed among those with the AA genotype, we found that 4 mg·kg−1 caffeine impaired performance by 13.7% in those with the CC genotype who are “slow” metabolizers of caffeine, and this corresponds to a very large ES, d = 1.3. We found no effect of either dose in those who have the AC genotype.

Most studies on caffeine and performance do not explore the basis for the interindividual variation in response, which has been well-documented in several studies (5–7,14,19). For example, Jenkins et al. (14) examined the effects of caffeine on exercise performance in 13 cyclists, and the interindividual range for performance change with caffeine at 1, 2, or 3 mg·kg−1 compared with placebo was −7.9% to 17.8%. Although 11 of 13 cyclists benefited from the 3-mg·kg−1 dose, the authors noted that “the mean performance outcome did not reach statistical significance due to two ‘non-responders’ strongly influencing the mean and SEM,” in addition to 8 of the 13 subjects performing worse on at least one caffeine condition versus placebo. Similarly, Paton et al. (11) found that caffeinated (~3–4 mg·kg−1) chewing gum improved overall performance in 20 male and female cyclists, but only 13 (65%) of the cyclists were considered “positive responders,” 5 (20%) experienced “negative” responses, and the remaining 2 (15%) experienced no observable effect on cycling performance. The authors speculated that this variation in response may be related to differences in the rate of caffeine metabolism or absorption between individuals (11).

Acute caffeine ingestion has been shown to alter RPE, where effort may be greater under caffeine conditions, yet it is not perceived as such (1,33). Consistent with other studies (1,33), our results showed a 3% decrease in RPE for the AA genotype at 5 km after taking 4 mg·kg−1 caffeine, which coincides with the group that had the fastest 10-km TT time. Those with the CC genotype taking 4 mg·kg−1 had a nonsignificant increase in RPE, which is consistent with the impaired performance in that group. Our findings suggest that caffeine does not lower RPE in all individuals. Similarly, a recent study by Green et al. (34) showed that when subjects were instructed to cycle at specific RPE (effort) levels under caffeine conditions, the higher perceived intensity did not necessarily result in greater work and improved performance in all subjects equally. The authors noted that individual responses to the caffeine may explain their unexpected findings.

In the present study, only those with the AA genotype who are fast metabolizers of caffeine benefited from caffeine during the 10-km TT. There is some evidence that extended periods of blocked adenosine receptors may be detrimental to performance (35), and this may explain the lack of benefit or diminished performance in slow metabolizers. Slower clearance of caffeine and longer caffeine buildup in slow metabolizers have been associated with increased blood pressure (25), and this vasoconstriction may also have effects on both blood flow to the heart and muscles (36). Resting myocardial blood flow does not appear to be affected by caffeine ingestion, but exercise-induced myocardial blood flow has been shown to decrease after caffeine ingestion (36). Under exercise conditions, the expected adenosine-mediated coronary vasodilation and subsequent increase in myocardial blood flow to match augmentation in cardiac work is likely impaired by caffeine and could explain the impaired performance among slow metabolizers (35,36). It has also been postulated that caffeine metabolites, such as paraxanthine, may have ergogenic properties and would be generated more quickly in fast metabolizers, thereby providing benefits sooner than in slow metabolizers (12). The initial ergolytic effects of impaired adenosine-mediated vasodilation experienced by fast metabolizers may be outweighed by their ability to expedite the production of these metabolites, which may be the source of ergogenicity.

Our findings are consistent with a previous study by Womack et al. (12) who observed a caffeine–gene interaction and improved TT cycling performance with caffeine only in those with the AA genotype. By contrast, previous studies either did not observe any effect of the CYP1A2 gene on caffeine ergogenicity (27,28) or reported benefits only in slow metabolizers (29). Pataky et al. (29) reported an improvement in those with the AC genotype compared with the AA genotype after caffeine ingestion; however, the 3-km TT performed in that study was a much shorter duration than in the present study or in the previous study (12). Furthermore, that study (29) did not include any subjects with the CC genotype, which is the group that we found to have impaired performance after 4 mg·kg−1 caffeine. Algrain et al. (28) found no effect of 255 mg caffeine on a 15-min cycling performance trial and no modifying effect of CYP1A2 genotype. However, this dose of caffeine was likely too low to observe any effect because most previous studies report an effect only at higher doses. That study also had a small sample size of 20 subjects (AA genotype, n = 10; C-allele carriers, n = 9), did not differentiate between AC and CC genotypes among C-allele carriers, and included both males and females, which may have been a confounder due to potential gender differences in caffeine response by genotype (37). Similarly, no effects of CYP1A2 on caffeine ergogenicity were observed in the study by Salinero et al. (27), but the very small sample size of only 10 subjects with the AA genotype, compared with 49 subjects with this genotype in the present study, make it unlikely that any significant findings would be detected. Importantly, that study used a 30-s Wingate test, which is a measure of power or anaerobic capacity, and is not a valid measure of endurance (27). Consistent with this notion, a meta-analysis of caffeine and exercise performance (4) showed that larger ES with caffeine supplementation were more often reported in trials of longer duration. Therefore, the unmasking of the effects of genotype on performance may occur during exercise of longer duration and during an accumulation of fatigue, where caffeine often provides its greatest benefits, and where the adverse effects to slow metabolizers are more likely to manifest. Moreover, as previously mentioned, this may improve performance by allowing for a greater accumulation of the potentially ergogenic caffeine metabolites, in fast metabolizers.

Two concerns often raised in caffeine–performance studies using cycling TT protocols are 1) a learning effect and 2) the caffeine–placebo effect. To address the issue of familiarization, we included a visit variable in our statistical model to control for potential confounding due to a learning effect. Although performance improved with each successive visit, the improved performance with caffeine in those with the AA genotype occurred regardless of the order of treatment, and the findings were the same with and without adjusting for visit. The caffeine–placebo effect can introduce a psychological factor (32) outside of the expected physiological effect, when a subject is not blinded to the treatment. Treatment blinding in the present study was successful with less than one third of caffeine trials identified correctly. Although a placebo benefit has been reported to occur in subjects who believe they have ingested caffeine (32,38), this would not explain the benefits seen only in those with the AA genotype.

Our results also confirm the ergogenicity of lower caffeine doses (2–4 mg·kg−1) as previously reported (9,13), but only within a specific genetic subset of individuals. Lower caffeine doses are more desirable to avoid the potential adverse side effects of higher doses (6–9 mg·kg−1), such as sleep disturbances (39), especially for athletes training or competing at night, as well as other adverse effects such as anxiety and agitation (40).

Although the results from the present study suggest a potential role of CYP1A2 genotype in influencing the ergogenic response of caffeine in competitive athletes from a variety of sports, care should be taken in extrapolating these findings to female, nonathletic or older populations. It is unknown if there is a similar genetic influence for other modes of exercise of high intensity or short duration, or whether other polymorphisms in CYP1A2 or other genes involved in the response to or metabolism of caffeine may modify the effects of caffeine during exercise.

In summary, we found that caffeine improves endurance performance at a dose of 2 and 4 mg·kg−1 for fast metabolizers of caffeine who have the CYP1A2 AA genotype (rs762551). Among the slow metabolizers, there is either no effect (AC genotype) or impaired performance (CC genotype) under the caffeine conditions in this study. These results highlight the importance of considering CYP1A2 genotype when deciding whether athletes should use caffeine as an ergogenic aid to improve endurance performance.

The authors thank all participants in the study, as well as Alicia Jarosz, Peter Radonic, Nicholas Wojcik, and Angela Solomon for help with data collection. This study was funded by the Canadian Institutes of Health Research, the Canadian Foundation for Dietetic Research, Nutrigenomix Inc., The Coca-Cola Company, and Mitacs. Results of the present study do not constitute endorsement by the American College of Sports Medicine. Results are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation. A.E.-S. is the Founder and holds shares in Nutrigenomix Inc., and N.G. serves on the Scientific Advisory Board of Nutrigenomix Inc.

Back to Top | Article Outline

REFERENCES

1. Doherty M, Smith PM. Effects of caffeine ingestion on rating of perceived exertion during and after exercise: a meta-analysis. Scand J Med Sci Sports. 2005;15(2):69–78.
2. Ganio MS, Klau JF, Casa DJ, et al. Effect of caffeine on sport-specific endurance performance: a systematic review. J Strength Cond Res. 2009;23(1):315–24.
3. Hodgson AB, Randell RK, Jeukendrup AE. The metabolic and performance effects of caffeine compared to coffee during endurance exercise. PLoS One. 2013;8(4):e59561.
4. Doherty M, Smith PM. Effects of caffeine ingestion on exercise testing: a meta-analysis. Int J Sport Nutr Exerc Metab. 2004;14(6):626–46.
5. Graham TE, Spriet LL. Performance and metabolic responses to a high caffeine dose during prolonged exercise. J Appl Physiol (1985). 1991;71(6):2292–8.
6. Doherty M, Smith PM, Davison RC, et al. Caffeine is ergogenic after supplementation of oral creatine monohydrate. Med Sci Sports Exerc. 2002;34(11):1785–92.
7. Wiles JD, Coleman D, Tegerdine M, et al. The effects of caffeine ingestion on performance time, speed and power during a laboratory-based 1 km cycling time-trial. J Sports Sci. 2006;24(11):1165–71.
8. Skinner TL, Jenkins DG, Taaffe DR, et al. Coinciding exercise with peak serum caffeine does not improve cycling performance. J Sci Med Sport. 2013;16(1):54–9.
9. Desbrow B, Biddulph C, Devlin B, et al. The effects of different doses of caffeine on endurance cycling time trial performance. J Sports Sci. 2012;30(2):115–20.
10. Ganio MS, Johnson EC, Klau JF, et al. Effect of ambient temperature on caffeine ergogenicity during endurance exercise. Eur J Appl Physiol. 2011;111(6):1135–46.
11. Paton C, Costa V, Guglielmo L. Effects of caffeine chewing gum on race performance and physiology in male and female cyclists. J Sports Sci. 2015;33(10):1076–83.
12. Womack CJ, Saunders MJ, Bechtel MK, et al. The influence of a CYP1A2 polymorphism on the ergogenic effects of caffeine. J Int Soc Sports Nutr. 2012;9(1):7.
13. Talanian JL, Spriet LL. Low and moderate doses of caffeine late in exercise improve performance in trained cyclists. Appl Physiol Nutr Metab. 2016;41(8):850–5.
14. Jenkins NT, Trilk JL, Singhal A, et al. Ergogenic effects of low doses of caffeine on cycling performance. Int J Sport Nutr Exerc Metab. 2008;18(3):328–42.
15. Graham-Paulson T, Perret C, Goosey-Tolfrey V. Improvements in cycling but not handcycling 10 km time trial performance in habitual caffeine users. Nutrients. 2016;8(7):pii: E393.
16. Skinner TL, Jenkins DG, Coombes JS, et al. Dose response of caffeine on 2000-m rowing performance. Med Sci Sports Exerc. 2010;42(3):571–6.
17. Hunter AM, St Clair Gibson A, Collins M, et al. Caffeine ingestion does not alter performance during a 100-km cycling time-trial performance. Int J Sport Nutr Exerc Metab. 2002;12(4):438–52.
18. Bortolotti H, Altimari LR, Vitor-Costa M, et al. Performance during a 20-km cycling time-trial after caffeine ingestion. J Int Soc Sports Nutr 2014;11:45.
19. Roelands B, Buyse L, Pauwels F, et al. No effect of caffeine on exercise performance in high ambient temperature. Eur J Appl Physiol. 2011;111(12):3089–95.
20. Cohen BS, Nelson AG, Prevost MC, et al. Effects of caffeine ingestion on endurance racing in heat and humidity. Eur J Appl Physiol Occup Physiol. 1996;73(3–4):358–63.
21. Begas E, Kouvaras E, Tsakalof A, et al. In vivo evaluation of CYP1A2, CYP2A6, NAT-2 and xanthine oxidase activities in a Greek population sample by the RP-HPLC monitoring of caffeine metabolic ratios. Biomed Chromatogr. 2007;21(2):190–200.
22. Ghotbi R, Christensen M, Roh HK, et al. Comparisons of CYP1A2 genetic polymorphisms, enzyme activity and the genotype-phenotype relationship in Swedes and Koreans. Eur J Clin Pharmacol. 2007;63(6):537–46.
23. Djordjevic N, Ghotbi R, Bertilsson L, et al. Induction of CYP1A2 by heavy coffee consumption in Serbs and Swedes. Eur J Clin Pharmacol. 2008;64(4):381–5.
24. Cornelis MC, El-Sohemy A, Kabagambe EK, et al. Coffee, CYP1A2 genotype, and risk of myocardial infarction. JAMA. 2006;295(10):1135–41.
25. Palatini P, Ceolotto G, Ragazzo F, et al. CYP1A2 genotype modifies the association between coffee intake and the risk of hypertension. J Hypertens. 2009;27(8):1594–601.
26. Palatini P, Benetti E, Mos L, et al. Association of coffee consumption and CYP1A2 polymorphism with risk of impaired fasting glucose in hypertensive patients. Eur J Epidemiol. 2015;30(3):209–17.
27. Salinero JJ, Lara B, Ruiz-Vicente D, et al. CYP1A2 genotype variations do not modify the benefits and drawbacks of caffeine during exercise: a pilot study. Nutrients. 2017;9(3).
28. Algrain Haya A, TRM Carrillo Andres E, Ryan Emily J. The effects of a polymorphism in the cytochrome P450 CYP1A2 gene on performance enhancement with caffeine in recreational cyclists. Journal of Caffeine Research. 2016;6(1):34–9.
29. Pataky MW, Womack CJ, Saunders MJ, et al. Caffeine and 3-km cycling performance: Effects of mouth rinsing, genotype, and time of day. Scand J Med Sci Sports. 2016;26(6):613–9.
30. Josse AR, Da Costa LA, Campos H, et al. Associations between polymorphisms in the AHR and CYP1A1-CYP1A2 gene regions and habitual caffeine consumption. Am J Clin Nutr. 2012;96(3):665–71.
31. Cohen J. A power primer. Psychol Bull. 1992;112(1):155–9.
32. Saunders B, de Oliveira LF, da Silva RP, et al. Placebo in sports nutrition: a proof-of-principle study involving caffeine supplementation. Scand J Med Sci Sports. 2017;27(11):1240–7.
33. Killen LG, Green JM, O’Neal EK, et al. Effects of caffeine on session ratings of perceived exertion. Eur J Appl Physiol 2013;113(3):721–7.
34. Green JM, Olenick A, Eastep C, et al. Caffeine influences cadence at lower but not higher intensity RPE-regulated cycling. Physiol Behav 2017;169:46–51.
35. Namdar M, Schepis T, Koepfli P, et al. Caffeine impairs myocardial blood flow response to physical exercise in patients with coronary artery disease as well as in age-matched controls. PLoS One. 2009;4(5):e5665.
36. Higgins JP, Babu KM. Caffeine reduces myocardial blood flow during exercise. Am J Med. 2013;126(8):730.e1–8.
37. Denden S, Bouden B, Haj Khelil A, et al. Gender and ethnicity modify the association between the CYP1A2 rs762551 polymorphism and habitual coffee intake: evidence from a meta-analysis. Genet Mol Res. 2016;15(2); doi: 10.4238/gmr.15027487.
38. Beedie CJ, Stuart EM, Coleman DA, et al. Placebo effects of caffeine on cycling performance. Med Sci Sports Exerc. 2006;38(12):2159–64.
39. Bchir F, Dogui M, Ben Fradj R, et al. Differences in pharmacokinetic and electroencephalographic responses to caffeine in sleep-sensitive and non-sensitive subjects. C R Biol. 2006;329(7):512–9.
40. Silverman K, Griffiths RR. Low-dose caffeine discrimination and self-reported mood effects in normal volunteers. J Exp Anal Behav. 1992;57(1):91–107.
Keywords:

EXERCISE; GENES; CYCLING; TIME TRIAL; ERGOGENIC; NUTRIGENOMICS

© 2018 American College of Sports Medicine