RESUMEN
PURPOSE: This study aimed to summarize and meta-analyze existing evidence regarding the influence of CYP1A2 genotypes on the acute effects of caffeine for exercise performance and to investigate the interaction between genotype, dosage, and timing of caffeine supplementation. METHODS: Six databases were searched for studies determining the effect of caffeine (except mouth rinsing) on exercise performance between CYP1A2 genotypes. Three-level meta-analyses were performed using standardized mean differences (SMD; Hedge's g ) to determine the effect of caffeine on exercise outcomes within and between CYP1A2 genotypes (AA, AC, and CC). Meta-regressions were performed for dose, timing, and presence of reported conflict of interests (RCOI). A meta-analysis was also performed with placebo values to assess for imbalances between genotypes. RESULTS: Thirteen studies, totaling 119 outcomes and 440 participants, were included (233 AA, 175 AC, ad 34 CC). Caffeine improved performance for AA (SMD = 0.30, 95% confidence interval [CI] = 0.21-0.39, P < 0.0001) and AC (SMD = 0.16, 95% CI = 0.06-0.25, P = 0.022) but worsened performance for CC (SMD = -0.22, 95% CI = -0.44 to -0.01, P < 0.0001). Dose affected only CC, with greater doses generating more positive SMD (CC-dose estimate: +0.19/1 mg·kg -1 body mass, 95% CI = 0.04-0.33, P = 0.01). Timing influenced only CC, with better performance with later onset of exercise after supplementation (CC-timing estimate: +0.01/min, 95% CI = 0.00-0.02, P = 0.02). RCOI only affected SMD of CC (CC-RCOI estimate: -0.57, 95% CI = -1.02 to -0.12, P = 0.01). After excluding studies with RCOI, no influence of genotype was seen (all P ≥ 0.19). Small, nonsignificant differences were seen in placebo between genotypes (SMD AA vs CC: -0.13; AA vs AC: -0.12; AC vs CC: -0.05; all P ≥ 0.26). CONCLUSIONS: Caffeine improved performance for AA and AC but worsened performance for CC. Dose and timing moderated the efficacy of caffeine for CC only. Caution is advised because baseline differences and studies with RCOI could have influenced these results.
