Metabolic and Performance Effects of Yerba Mate on Well-trained Cyclists : Medicine & Science in Sports & Exercise

Journal Logo


Metabolic and Performance Effects of Yerba Mate on Well-trained Cyclists


Author Information
Medicine & Science in Sports & Exercise 50(4):p 817-826, April 2018. | DOI: 10.1249/MSS.0000000000001482


Ilex paraguariensis is a South American plant of growing scientific interest worldwide, consumed in beverages by millions on a daily basis. Yerba Mate (YM), as named colloquially, has been used since pre-Columbian times ritually and for its alleged medicinal properties by the Guaraní ethnic group. YM is rich in phenolics (chlorogenic acid and other caffeoyl derivatives) with high antioxidant potency, saponins, and xanthines (caffeine and theobromine) (1). Biomedical research on YM is scarce when compared with coffee and tea (1), but several reports support the role of YM as a modulator of metabolism. For instance, YM has been reported to be antiobesogenic (2), a regulator of Akt and AMPK signaling pathways in different tissues (3,4), cardioprotective (4), hypocholesterolemic (5), and able to increase thermogenesis (6), and to shift substrate utilization during exercise towards higher fat oxidation (7,8). Some of these effects position YM as a potential exercise enhancer (9) and ergogenic aid. However, studies to date have only been conducted in rodents and unfit or unhealthy humans. Hence, whether YM can have an effect in an already optimized metabolic machinery of well-trained endurance athletes is yet unknown.

Specifically, recent reports of YM increasing fat utilization during exercise of different intensities and durations (7,8) and its potential neurological effects (10,11) suggest YM as a suitable supplement to improve endurance performance. Enhancing metabolic flexibility could be an important effect of YM to optimize the use of endogenous fuel; that is, to increase the use of fat during low-intensity exercise and spare the limited carbohydrate (CHO) stores for performance-determining high-intensity bouts (12,13). In addition, independent of the metabolic effects, a neuromodulatory effect through synergism between the phenolics and xanthines could drive CNS stimulation (10,11). YM has additionally been reported to have analgesic and anxiolytic effects in rodents (14), which could also be a factor resulting in positive performance outcomes.

Some of the metabolic effects of YM are comparable to those of other well-studied plants also with a high content of phenolics, such as green tea and coffee (15), but the phytochemical profile of YM is unique. Green tea and YM share, among others, compounds such as caffeine, but different from green tea, YM contains a large quantity of caffeoyl derivatives (chlorogenic acids) and saponins with different moieties (5). Some of the metabolic effects of phenolics have been attributed to their down-regulation of the catecholamine-O-methyltransferase (COMT), but markers of effects on this enzyme have not been shown in humans (15). Importantly, YM has been reported to lack (−)epigallocatechin-3-gallate and other catechins, allegedly the main compounds for increasing fat oxidation in green tea (15). Therefore, it is possible that the phytochemicals unique to YM, or the effects resulting from their interactions, are the direct key for its metabolic effects.

The aim of the current study was to test the effect of YM ingestion daily for 5 d and 1 h before tests to assess substrate utilization during submaximal exercise and performance during a short (~30 min) simulated cycling time trial (TT) in well-trained cyclists. The dose of YM would be such as to contain a low (nonergogenic) amount of caffeine. Our hypotheses were that YM would increase fat utilization during submaximal exercise and improve TT performance.


YM Samples

To determine dosage and because of variability of phytochemical content in plants (16), four lots of YM (leaves mixed with stems) of different origin (Taragüi, Regular Blend, Las Marias, Argentina; La Merced, de Campo, Las Marias, Argentina; Rosamonte, Regular blend, Diez hermanos, Argentina; Guayaquí, Biodynamic, Argentina—see Tables, Supplemental Digital Content 1, Phytochemical content of 4 brands of YM, prescreened for the study,, and Supplemental Digital Content 2, Origin details of YM brands used in this study, were screened for xanthine and total phenol content as well as radical scavenging activity. On the basis of lower relative caffeine content and higher phenol content and scavenging activity, La Merced, de Campo (lot number L 24515 09:10 A 3370), was selected as treatment and for further analysis of xanthine and phenol content (Table 1). The caloric content of YM doses was estimated to be negligible (<4 kcal).


Eleven well-trained competitive male endurance cyclists/triathletes were recruited. The subjects’ age, body mass (BM), maximal oxygen uptake (V˙O2max), peak aerobic power output (PAPO), and regular training load were 30 ± 3 yr, 75 ± 7 kg, 71 ± 6 mL·kg−1·min−1, 403 ± 32 W, 11 ± 2 h·wk−1, respectively. Before giving their written consent, all subjects were informed of the nature of the study and possible risks involved.

Pretesting: Incremental Cycling Test

Approximately 2 wk before commencing their first experimental trial, subjects underwent an incremental cycling test to exhaustion on an electronically braked cycle ergometer (Lode Excalibur Sport, Groningen, the Netherlands) for determination of PAPO and V˙O2max as previously described (17). During this test, subjects breathed through a two-way, low-resistance non-rebreathing valve (Hans Rudolph Inc., Kansas City, KS) and mouthpiece attached to a calibrated Oxycon Pro metabolic system (Jaeger, Hochberg, Germany) interfaced to a computer that calculated the instantaneous rates of O2 consumption (V˙O2), CO2 production (V˙CO2), and the RER. Before each test, analyzers were calibrated with commercially available gasses of known and certified O2 and CO2 content. V˙O2max was defined as the highest uptake a subject attained during any 30 s of the test, whereas PAPO was calculated from the last completed work rate plus the fraction of time spent in the final noncompleted work rate (17). All exercise sessions were conducted under standardized laboratory conditions of temperature and humidity, and subjects were fan cooled (Sealey HVSF30, Suffolk, United Kingdom) throughout all tests.

Study Overview

An overview of the experimental protocol is shown in Figure 1. The study was evaluated by the Regional Ethics Committee of Norway in accordance to the Norwegian Act on Medical and Health Research and allowed for implementation. Briefly, each subject completed two experimental trials in a double-blind, placebo (PL)-controlled, randomized, counterbalanced crossover design. Five days before each trial day, subjects ingested per day 5 g of YM or PL (maltodextrin). Two days before each trial, physical activity was controlled, and food for the day before each trial was provided. On each trial day, subjects reported to the laboratory fasted at ~7.00 AM and an 18 G catheter (BD, Franklin Lakes, NJ) was inserted into the antecubital vein of one arm to allow for serial blood sampling. Immediately after baseline blood sampling, subjects ingested YM or PL capsules, 1 h before starting a 10-min warm-up (7 min at 70% of V˙O2max, followed by 3-min ramp-down to 30% of V˙O2max) before a submaximal step test (SST). After SST, subjects recovered for 5 min and completed a laboratory-based cycling TT. The washout period between trials was 1–2 wk (18,19).

Schematic overview of experimental trials. Participants ingested 5 g of YM or PL (maltodextrin) per day for 5 d before experimental trials. Exercise was controlled for 48 h before trials and diet was provided for 24 h before trials. On trial days, venous blood samples were obtained at indicated time points before and after ingestion of YM or PL. SST consisted of consecutive 5-min steps at 30%, 40%, 50%, 60%, 70%, and 80% of V˙O2max. After 5-min recovery, participants undertook a ~30-min TT to complete a preset amount of mechanical work. Expired gas samples were collected continuously during the SST and at indicated time points during TT.


During this test, participants were instructed to keep cadence constant at 85 rpm while completing consecutive 5-min-long stages of cycling at power outputs to elicit 30%, 40%, 50%, 60%, 70%, and 80% of V˙O2max. Throughout SST, the respiratory parameters were measured and the last minute of each stage was analyzed. Rate of perceived exertion (RPE; Borg scale) was assessed at the end of each stage.


Simulated laboratory cycling TT were undertaken with the ergometer set in a cadence-dependent power output (linear) mode for participants to take ~30 min to complete a set amount of mechanical work. A custom-determined alpha value was assigned to each individual on the basis of the preferred cadence during baseline testing and the total workload to be completed. On the basis of previous experiments (20) and extensive pilot testing, we determined that subjects would average ~80% of PAPO for 30-min TT. Accordingly, the mechanical work to be completed for each individual was determined as follows:

During TT, subjects were blinded for time, power output, and V˙O2/V˙CO2 but provided with cadence and total amount of mechanical work to be completed (as a percentage of total and continuous real-time kilojoules countdown to zero).

Subjects were instructed to complete TT as fast as possible, as they would do on a race. No encouragement was provided during TT, and neutral verbal feedback was provided at specific time points. All tests were conducted by the same researcher (J.L.A.). No music was played during trials, and participants had no cues of time elapsed. Participants were tested one at a time on separate days. To assess blinding, immediately after the last trial, participants were asked to match first and second experimental trials with “YM,” “PL,” or “I don’t know.”

Respiratory gasses were measured four times for 3 min at the start of TT: start, 1/3, and 2/3 of total workload and during the last ~70 kJ of work. RPE was assessed at the end of each measurement. An overall weighed average for the four TT measuring points was obtained. The weight of each measure was determined on the basis of duration and timing of each sampling point and total duration of the trial; that is, first and last measures were considered to be only representative of the time during which samples were obtained, whereas samples 2 and 3 where considered to be representative of the times where respiratory air was not collected.

Exercise Control

In the 48 h before each trial, subjects abstained from any vigorous physical activity but were allowed to exercise at <70% maximal heart rate (HRmax) for a maximum of 2 and 1 h, and 2 and 1 d, respectively, before each trial. The training completed before the first trial was kept on a training log and repeated before the second trial.

Dietary Control

A custom-made prepackaged diet containing 45 kcal·kg−1 BM in the form of 8, 0.8, and 1.5 g·kg−1 of high glycemic index CHO, fat, and protein, respectively, was provided to each participant for the 24 h before each trial. Participants were instructed to ingest a portion of their diet (snack) containing 0.9 g·kg−1 BM CHO before sleeping ~9 h before experimental trials to eliminate the chances of hypoglycemia during experimental trials, which were undertaken in the fasted state. A participant’s checklist of food items was handed back to the researcher served also as control for dietary compliance. On trial days, ad libitum water intake was measured and recorded from the moment of ingestion of the capsules until the end of TT.

Capsules Preparation and Delivery

For each dose, dark gelatin capsules (12 units, size 00) were filled with 5 g of YM brought to fine powder in a grinder (MKM6003, Bosch, Germany) or with maltodextrin. All capsules were tasteless and treatments visually undistinguishable from each other (tested in pilot tests before study). Participants were indicated to have capsules together with lunch. A reminder was set on participants’ phone with a dedicated app (Medisafe; Medisafe Inc.) and/or online calendar, which also allowed the main researcher to track compliance with capsule ingestion schedule.

Blood Sampling

Blood samples were collected in 6-mL EDTA tubes (BD) upon insertion of catheter (baseline), before SST, after SST, every 1/3 of TT, and immediately after TT. After each blood sample and at regular 20- to 30-min intervals, catheters were flushed with 5 mL saline solution (0.9% NaCl g·L−1) to maintain patency in the cannula. Immediately upon collection, samples were spun at 3000 rpm for 10 min at 4°C. The resultant plasma was aliquoted and stored at −80°C for later analyses.

Gastrointestinal Distress Questionnaire

A gastrointestinal distress questionnaire as detailed previously (21) was used to assess any potential negative symptoms of the treatment. The questionnaire was completed by participants before capsule ingestion, before SST, and after TT.

Calculations of Substrate Utilization, Maximal Fat Oxidation, FATmax, and Gross Efficiency During SST

For all calculations, the data of the last minute of each stage were used. Whole-body rates of CHO and fat oxidation (g·min−1) were determined for each steady-state gas measurement point from V˙CO2 and V˙O2 values using the nonprotein RER calculation (22). Gross efficiency was calculated as reported by Moseley and Jeukendrup (23). For each SST, a third-degree polynomial line of best fit was incorporated using fat oxidation (g·min−1) as a function of measured intensity (% of V˙O2max) including origin (0,0) according to previously described methods (24). The turning point (local maximum) of the curve was used to determine maximal fat oxidation (MFAO; g·min−1) and FATmax (the intensity at which MFAO is elicited).

Analytical Techniques


Free fatty acid (FFA) concentrations were measured using a nonesterified fatty acid assay kit (NEFA-HR (2); Wako Pure Chemical Industries, Ltd, Osaka, Japan). Glycerol was analyzed using a kit coupling enzyme assay involving glycerol kinase and glycerol phosphate oxidase (MAK117; Sigma-Aldrich, St. Louis, MO). Lactate was analyzed using a YSI 1500 SPORT (Yellow Springs Instruments, Yellow Springs, OH). Glucose was analyzed using a Biosen C-Line (EKF Diagnostics, Magdeburg, Germany). Plasma adrenaline was analyzed using an enzyme-linked immunosorbent assay kit (EIA-4306; DRG Instruments, Marburg, Germany). Caffeine and paraxanthine were analyzed by validated liquid chromatography coupled to tandem mass spectrometry methodology described elsewhere (25). Briefly, the liquid chromatography coupled to tandem mass spectrometry system was composed of a Shimadzu (Shimadzu Scientific Instruments, Columbia, MD) LC20AD system and an ABSciex triple-quadrupole mass spectrometer equipped with a Waters Xbridge C18 column (2.1 × 100 mm, 3.5 μm). MilliQ water (0.1% formic acid) and acetonitrile (0.1% formic acid) were used as mobile phases. Total runtime was 9 min. Column oven was operated at 10°C and 30°C, respectively. The mass spectrometry parameters were as follows: curtain gas, 16; collision gas, 8, ionspray gas, 5500; temperature, 575°C; nebulizing gas, 60; and drying gas, 50. The mass spectrometer was operated in positive mode with electrospray ionization and multiple-reaction monitoring. Data collection and analysis were handled by Analyst Software 1.5.1 (Applied Biosystems).

Seven concentrations of each analyte were prepared as calibration standards in blank plasma, and four different concentrations were prepared as quality controls. Limit of quantitation (LOQ) was based on the lowest concentration of the linear calibration curve that gave an acceptable accuracy and precision (+20%). Linear range and LOQ are given in nanograms per milliliter (compound: (linear range, LOQ): caffeine (5–500, 5) and paraxanthine (5–250, 5).

YM extract preparation for phytochemical analysis

Dry YM plant material was grinded to a powder (<1 mm) and extracted on an Accelerated Solvent Extraction system (ASE 350; Dionex, Sunnyvale, CA). Diatomaceous earth (Dionex) was mixed with 3.5 g of plant material and loaded in 100-mL steel cartridges. The cartridges were fitted onto the system, and exhaustive extraction was performed with two cycles of 100% methanol followed by two cycles of 50% methanol at 60°C followed by two cycles of 50% methanol at 100°C. Preheating time was 5–7 min, static extraction per cycle was 5 min, and the extraction was carried out under a pressure of 1500 PSI (10 MPa). The extraction was performed three times. The extracts from each cell were combined and diluted to 250 mL with methanol. The diluted extract was used for xanthine determination. For quantification of total phenol content, DPPH scavenging activity, and extraction yield, 150 mL of the diluted extract was taken to dryness in a rotary evaporator.

Quantitative analysis of xanthines in plant extract

One milliliter of 0.525 μg·μL−1 8-chlorotheophylline 98% (Sigma-Aldrich) as internal standard was added to 5 mL of the diluted extract (26), dried on a rotary evaporator, and subsequently dissolved in 10 mL of mobile phase before filtering into high-performance liquid chromatography (HPLC) vials (polytetrafluroethylene, 0.22 μm). Five microliters of the filtrate was analyzed by using HPLC (LaChrom Elite; Hitachi, Tokyo, Japan) equipped with a reverse-phase C18 column (Atlantis T3, 3 μM, 150 × 4.6 mm; Waters, Dublin, Ireland) and an L-2455 diode array detector. Elution was performed using isocratic eluent, acetonitrile/0.1% formic acid in distilled water (1:9 vol/vol; HPLC grade; Sigma-Aldrich) (26). The flow rate was 1.0 mL·min−1. The absorbance was recorded at 272 nm, and the separation was carried out at 25°C. The average value of three parallels was used for the amount calculation. Calculation of xanthine amount was based on a linear regression model with internal standard. The calibration curve was obtained using caffeine ReagentPlus® 99% (Sigma-Aldrich) and theobromine of ≥98.5% (Sigma Life Science, St. Louis, MO, USA). Stock solution of caffeine (1.95 μg·μL−1 in methanol) and theobromine (0.53 μg·μL−1 in 50% methanol) were used for a calibration in the range of 0.01–0.12 μg·μL−1. The calculated amount of caffeine and of theobromine were expressed as percent (wt/wt). Theophylline was not detected in the plant material and standard curves were therefore not obtained.

Total phenolic content

Dried extract was dissolved in dimethyl sulfoxide (5 and 2.5 mg·mL−1) in triplicates. A linear calibration curve of gallic acid (≥97.5%; Sigma-Aldrich) was obtained in the range of 0.3–2.5 mg·mL−1. The experiment was performed according to Singleton et al. (27). Briefly, 40 μL of test solution was mixed with 3160 μL of distilled water (MilliQ) and added 200 μL of Folin-Ciocalteu reagent (Merck, Darmstadt, Germany). After 5 min, 600 μL of 20% Na2CO3 solution was added and incubated in the dark at room temperature for 2 h. The absorbance was measured at 765 nm on a Biochrom Libra S32 PC UV/Vis spectrophotometer (Biochrom Ltd., Cambridge, United Kingdom).

DPPH radical scavenging

Reaction with the DPPH radical was carried out as previously described (28). Briefly, the dried extract (0.05 mL, in dimethyl sulfoxide) was mixed with a solution of DPPH (Sigma-Aldrich) in methanol (A517 = 1.0; 2.95 mL), and the UV absorbance at 517 nm was measured for 5 min. Samples were assayed in triplicate, and result was given as the effective concentration to give 50% scavenging of the DPPH radical (EC50) ± SD. Quercetin (Sigma-Aldrich) was used as a positive control.

Statistical Analysis

Data were analyzed using two-way repeated-measures ANOVA with Student–Newman–Keuls post hoc analysis to correct for the family-wise error during multiple post hoc tests (Sigmaplot for Windows; Version 13). Grouped data were analyzed using paired t-tests, and performance data were analyzed using Glass Δ effect sizes (ES) with an online available tool ( following guidelines outlined by Hopkins (29). Linear regressions were calculated using the least-squares method. All data are presented as mean ± SD, except that for ES—and elsewhere specified—data are reported as ±95% confidence interval (CI). The level of statistical significance was set at P < 0.05.


Plant Material

Details of plant material are reported in Table 1. On the basis of a 7.9% water content, the total content per 5-g dose was 52 mg of caffeine (relative to participants’ BM: 0.70 ± 0.06 mg·kg−1) and 456 mg of total phenolics (relative to participants’ BM: 6.1 ± 0.56 mg·kg−1). Absence of catechins in the material was confirmed using HPLC-DAD and proton nuclear magnetic resonance analysis.

Phytochemical content of dried YM plant material.


Two participants were unable to execute maximal performance efforts, one on YM and one on PL trial day because of individual incident and to a technical issue, respectively. Because of participants’ reported incapacity for best-performance, large performance variation (in excess of 11%) and being outliers following Chauvenet’s criterion, these were excluded from TT analysis before breaking the blinding code and performance data are reported for n = 9. The SST data for these individuals were unaffected and therefore kept for the analysis.


All subjects complied with dietary requirements. Total water ingested on trial days was 948 ± 401 and 894 ± 249 g in PL and YM, respectively (P = 0.56).

Performance Test

There was a small (ES = 0.38 ± 0.33) but significant (P = 0.028) performance improvement with YM in time to complete TT from 30.1 ± 1.8 to 29.4 ± 1.4 min (delta 0.40 ± 0.45 s), and a concomitant small but significant 1.7% ± 2.1% increase in average power as percentage of PAPO (ES = 0.36 ± 0.33; P = 0.035; Fig. 2A) and a 2.3% ± 2.6% increase in absolute power (ES = 0.2 ± 0.19; P = 0.042). Individual performance difference as average percentage of PAPO ranged from −1.8% to 6.4% (Fig. 2B). Analysis of pacing (Fig. 2C) indicates that there were no differences between treatments in power output until 30% of the total completed workload, from which point the power output in YM was significantly higher (P < 0.05) consistently until the end of the TT. There were no differences in RPE between treatments (Fig. 2D).

Performance TT test results showing average % PAPO (columns) and individual responses (dots and lines; A), individual differences in performance as delta % PAPO (B), pacing in both groups (C), and rates of perceived exertion at different time points (D). Data in panels C and D are means ± SD. *Significantly different from PL at same point (P < 0.05). Data were analyzed using paired t-tests for pairwise comparisons and two-way repeated-measures ANOVA with Student–Newman–Keuls post hoc analysis for multiple comparisons.

Respiratory Results


There was a main effect of treatment (P = 0.05) and intensity (P < 0.001) on fat oxidation. Fat oxidation was increased in YM compared with PL at 30%, 40%, and 50% of V˙O2max (P = 0.008) by 0.10, 0.11, and 0.09 g·min−1, respectively, representing an average 23% higher fat oxidation over this range (Fig. 3A). Fat oxidation tended to be higher in YM (P = 0.071) at 60% of V˙O2max. Accordingly, CHO oxidation was higher in PL (P = 0.01) over the same range of intensities. MFAO (0.67 ± 0.1 g·min−1 YM, 0.60 ± 0.16 g·min−1 PL) and FATmax (55% ± 3% V˙O2max YM, 55 ± 3% V˙O2max PL) were not significantly different, but there was a trend (P = 0.1) for higher MFAO in YM. Despite the difference in substrate utilization, there were no differences in efficiency at any workload (Fig. 3B).

Respiratory responses during SST (A and B) and during simulated TT (C and D). Calculated fat utilization (A) and efficiency (B) during SST at power to elicit 30%, 40%, 50%, 60%, 70%, and 80% of V˙O2max. TT oxygen consumption as % of V˙O2max (A), RER (B), and HR (C). Data are Means ± SD (n = 11; SST, n = 9; TT). *Significantly different from PL at same point (P < 0.05). Data were analyzed using two-way repeated-measures ANOVA with Student–Newman–Keuls post hoc analysis.


Oxygen uptake was higher (P = 0.018) in YM (84.1% ± 3.4% V˙O2max) than in PL (81.8% ± 4.3% V˙O2max; Fig. 3C), but there were no differences between treatments in RER during TT (PL, 0.95 ± 0.03; YM, 0.95 ± 0.02; Fig. 3D), average HR (PL, 168 ± 10 bpm; YM, 169 ± 8 bpm), or maximal HR (PL, 179 ± 10 bpm YM, 182 ± 9 bpm) during TT.


There were no differences in HR and RPE between treatments during SST (Table 2).

RPE, HR, RER, and V˙O2 during SST (n = 11).

Blinding Assessment

Two of nine participants correctly identified the trial in which they were under the YM treatment. The remaining seven individuals were unable to match the capsules ingested to PL or YM treatment.

Gastrointestinal Distress Questionnaire

There were no differences between time points or treatments in any of the items of the gastrointestinal distress questionnaire.

Plasma Metabolites

Details of plasma metabolites are outlined in Table 3.

Plasma metabolites (n = 9).

Lactate and Glucose

Lactate showed a trend (P = 0.07) for main treatment effects and a clear main effect of time (P < 0.001) with a treatment–time interaction (P = 0.029). Lactate was significantly higher in YM compared with PL at 2/3 (P = 0.018) and 3/3 of TT (P = 0.001). Glucose showed a main effect for time (P < 0.001) but no main effect for treatment or interactions. Glucose was significantly higher in YM pre-SST compared with PL (P = 0.029) and raised above resting values in YM during 2/3 of TT (P = 0.029) and in both groups at 3/3 of TT (P < 0.001).

FFA and Glycerol

FFA showed main effects for time (P = 0.03) and treatment–time interactions (P = 0.002). FFA values were lower in YM compared with PL pre-SST (P = 0.002) and lower in pre-SST in PL compared with resting values (P = 0.04). Glycerol showed main effect of time (P < 0.001) and treatment–time interactions (P = 0.04). At post-SST (P = 0.046) and 1/3 of TT (P = 0.033), glycerol was higher in YM than in PL. A large correlation (r = 0.59, P = 0.0057; μM FFA ox·kg−1·min−1 = 0.012 (±95% CI, 0.008) × [FFA] + 10.65 (±95% CI, 4.83)) was observed between the pre-SST plasma FFA and fat oxidation during SST, and a very large correlation (r = 0.76, P < 0.001; μM FFA ox·kg−1·min−1 = 0.086 (±95% CI, 0.037) × [Glycerol] + 6.63 (±95% CI, 4.83)) between post-SST plasma glycerol and fat oxidation during SST.

Caffeine, Paraxanthine, and Adrenaline

Caffeine showed main effects for treatment, time, and time–treatment interactions (P < 0.001). Caffeine concentration was higher in YM than in PL at all time points (P < 0.001). Only in YM, there was a marked increase above resting values before SST (P < 0.001) and remained elevated throughout the rest of the trial. Adrenaline concentration was higher in YM than in PL at baseline (P = 0.049), pre-SST (P = 0.002), and at 3/3 of TT (P = 0.002). In both treatments, adrenaline increased compared with baseline at all time points after SST (P < 0.001).


The main findings of this study were that supplementing well-trained cyclists with 5 g of YM daily for 5 d and 1 h before experimental trials resulted in: 1) a 23% increase in fat oxidation, on average, compared with PL during cycling at intensities between 30% and 50% V˙O2max, and 2) a small but significant performance improvement in a ~30-min TT compared with PL. During submaximal exercise, we could detect clear metabolic differences between treatments on substrate oxidation, plasma FFA, and glycerol, but no differences in gross efficiency. During the performance test, there was a clear CHO dependence in both groups with no differences between treatments, indicating that the performance improvements in YM were due to factors other than a shift in substrate utilization. In addition, we report for the first time increased plasma adrenaline in humans in response to a supplement low in caffeine and rich in phenolics.

The current study represents the first scientific investigation, to our knowledge, to assess the effects of YM on metabolism during submaximal exercise and TT performance in well-trained cyclists. Previous studies assessing the metabolic effect of YM during exercise were limited to a single preexercise dose of ≤2 g in untrained population and included no dietary control, assessment of performance, or plasma markers of fat metabolism, focusing mainly on respiratory measures to assess substrate utilization (7,8). These studies show that YM increased fat oxidation by 24% during a step test and at intensities less than 70% V˙O2peak (7) and by ~18% during 30-min continuous exercise at 37% of V˙O2peak (8) compared with PL treatments. Despite the differences in population and experimental protocols, our findings of 23% increase in fat utilization during submaximal exercise are in line with what has been reported previously. Moreover, it is also possible that effects of YM on increased fat utilization at higher intensities during SST (or indeed at FATmax) were not detected because of a small RER overestimation due to CO2 contribution from body HCO3 stores at higher intensities (13,30). Nonetheless, we provide further insights supporting the use of YM as a new aid to manipulate substrate utilization during submaximal exercise in addition to further novel insights into its potential effect on metabolic regulation.

The metabolic markers assessed in plasma provide new clues on the causes behind changes in substrate utilization. We specifically assessed plasma FFA and glycerol because of their reliability as markers of lipolysis and their relationship to fat oxidation (31). We found no effects of 5-d YM supplementation on resting blood FFA or glycerol, and plasma FFA remained unchanged 1 h after YM ingestion. Unexpectedly, FFA decreased slightly in PL resulting also in a significant difference from YM at the same time point. Blood glycerol also showed no differences between groups before SST, but instead it was higher in YM compared with PL after SST and at 1/3 of TT, and proved to have a stronger association with fat oxidation than pre-SST plasma FFA.

Provided that fat metabolism is very sensitive to small increases in blood insulin via a decreasing lipolysis and plasma FFA oxidation (32), it should be considered that the insulin response to 5 g of maltodextrin in PL may have affected substrate utilization and partially explain the drop in plasma FFA before SST in PL. However, because this dose represents one-fifth of the minimal dose that has been previously shown to affect substrate utilization (33) and also likely to elicit a minimal insulin response (34), a suppression of fat oxidation seems unlikely. As shown by Achten and Jeukendrup (35), a dose of 75 g of glucose (13 times the amount we used) seems to be necessary to suppress fat oxidation by ~25% during an incremental test. Moreover, fat oxidation does not seem to be affected by changes in FFA in the range observed (31).

Instead, it is possible that the observed increase in lipolysis and fatty acid oxidation is a consequence of a catecholamine-mediated response. Indeed, we observed an increase in baseline and pre-SST adrenaline, which could be mediated by inhibition of the COMT enzyme as a consequence of the content of xanthines, chlorogenic acid, and other caffeoyl derivatives in mate (11). Such mechanism has been hypothesized in the past but never observed in humans (15).

Taken together, these results suggest that circulating pre-SST FFA and lipolysis during SST were related to the rates of fat oxidation observed but do not completely explain our results. Therefore, it is likely that other loci of control and mechanisms such as intermediary metabolism and substrate transport into metabolically active tissues are important to explain our findings. These potential mechanisms should be further evaluated in the future by direct assessment of muscle metabolism in response to YM in vitro and in vivo. In addition, future studies assessing the effect of very low doses of CHO before exercise on fat oxidation will allow for establishing with better precision the magnitude of the effect of YM on fat oxidation reported here and shed further light on the points of metabolic control. In the mean time, our results suggest that YM indeed increases fat oxidation during exercise at submaximal intensities. The enhanced fat oxidation observed during SST, however, does not seem to be a reason behind the performance improvement.

The cycling performance improvement and parameters assessed during TTs provide valuable insights into the physiological response to YM supplementation. TTs were an average 40 s faster in YM as a consequence of an average 2.3% increase in absolute power output. The magnitude of improvement observed is in line with a recent meta-analysis showing an average 1.9% increase in performance by short-term polyphenol supplementation (9). Substrate selection was unlikely a factor affecting performance as we show a clear CHO dependence in both groups (Fig. 3D), which is in accordance to the metabolic demands of intense cycling TT: mild differences in plasma substrate availability are overridden and normalized by high-intensity exercise (20). In relation to physiological and other metabolic variables, a higher power output was concomitant to higher-average V˙O2 as well as higher plasma lactate and adrenaline by the end of TT in YM. Pacing strategy showed a higher power output in YM during the last two-thirds of TT, indicating increased tolerance to fatigue. There were no differences in other physiological parameters such as average heart rate or HRmax, RPE, or gross efficiency (measured during the submaximal test). Some of these results, in particular the pacing strategy and increased power output with similar RPE, suggest a performance improvement such as those observed for the effect of caffeine (36), but caffeine on its own is an unlikely candidate for explaining our findings.

The low amounts of naturally occurring caffeine (52 mg, or ~0.7 mg·kg−1) provided to our participants with YM treatment were deliberately intended to fall under an ergogenic dose (37). The caffeine content in the plant material was in fact critical to determine treatment dosage so as to provide the highest possible amounts of other phytochemicals together with a nonergogenic amount of caffeine, provided that it is widely accepted that 3 mg·kg−1 of caffeine represents an ergogenic dose for endurance sports, an amount that results in plasma caffeine concentration of 15–20 μM. Under this threshold, caffeine is unlikely to have ergogenic effects (37). In the current study, we observed a peak plasma caffeine of 2.3 μM, which represents approximately one-eighth of the alleged minimum plasma concentration representative of an ergogenic dose. Accordingly, it has been shown that doses of caffeine of 1 (38), 1.5 (39), and even 2 (40) mg·kg−1 are not ergogenic in endurance tests lasting ~30 min. Moreover, no difference in the rates of perceived exertion and HR during the submaximal test is another strong indicator of the lack of a caffeine-mediated effect (41). Therefore, our findings match those of previous studies and allow us to conclude that caffeine was not ergogenic, at least not in the way that is commonly observed.

Instead, the ergogenic effects of YM could be explained through a synergism between chlorogenic acids and caffeine in stimulating the CNS. The effect of phenolic compounds on brain-specific COMT could diminish the breakdown of dopamine and increase its bioavailability (10). As a consequence of the increased CNS dopamine levels, doses of caffeine that would normally not be ergogenic could have a potentiation effect (42) on the—already higher—dopamine levels and have ergogenic effects during exhaustive exercise (10,42).

In addition, the ergogenic effects of YM could also be related to its effects on regulation of blood flow (4). YM has been associated with increased endothelial nitric oxide synthase activation and consequent endothelial nitric oxide production and vasorelaxation (4,43). Although the effect of YM on this pathway has been studied only in rodent cardiac muscle (4), endothelial nitric oxide synthase is present in human skeletal muscle (43) and has been proven to be activated by polyphenols (9). Although these ideas remain speculative, they provide suitable explanations matching our observations to the shown in vivo physiological and metabolic effects of YM. In particular, the effect of polyphenols on COMT represents a suitable mechanism linking both the metabolic and performance effects through its effects on liver and brain COMT isoforms.

In conclusion, YM increased fat oxidation at low-exercise intensities (30%–50% V˙O2max) and increased performance in a short (30-min) TT. Although the performance improvement in the TT was likely due to factors other than increased fat utilization, enhanced fat oxidation by YM could potentially be of use for manipulating substrate use during training in conditions of low CHO availability (44). The performance effect of YM should be addressed in more “real-life” racing conditions including before- and during-exercise CHO-rich nutrition. In the meantime, our findings add to a growing body of information on the importance of phytochemicals for performance, and we provide valuable new physiological and metabolic insights to understand the mechanisms behind the response to YM.

We thank Prof. Jostein Hallén for assistance during the preparation of the study, Øyvind Skattebo for language assistance in ethics application process, and Bent Håvard Hellum from PROMEC (Norwegian University of Science and Technology) for plasma caffeine and paraxanthine analyses.

The current study was internally funded by the department of physical performance of the Norwegian School of Sport Sciences. All authors report no conflicts of interest.

The results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation. The results of the present study do not constitute endorsement by the American College of Sports Medicine.

J. L. A. contributed to the conceptualization of the study. J. L. A., I. A., and H. W. analyzed plant material samples. J. L. A. conducted clinical trials. J. L. A., I. A., H. W., and C. C. analyzed data. J. L. A. wrote the original draft. J. L. A., I. A., H. W., and C. C. wrote, reviewed, and edited the manuscript. J. L. A. and C. C. supervised the study. J. L. A. and C. C. have primary responsibility for the final content. All authors have read the final version of the manuscript.


1. Bracesco N, Sanchez AG, Contreras V, Menini T, Gugliucci A. Recent advances on Ilex paraguariensis research: minireview. J Ethnopharmacol. 2011;136(3):378–84.
2. Gambero A, Ribeiro ML. The positive effects of yerba maté (Ilex paraguariensis) in obesity. Nutrients. 2015;7(2):730–50.
3. Arcari DP, Santos JC, Gambero A, Ferraz LF, Ribeiro ML. Modulatory effects of yerba maté (Ilex paraguariensis) on the PI3K-AKT signaling pathway. Mol Nutr Food Res. 2013;57(10):1882–5.
4. González Arbeláez LF, Fantinelli JC, Ciocci Pardo A, et al. Effect of an Ilex paraguariensis (Yerba Mate) extract on infarct size in isolated rat hearts: the mechanisms involved. Food Funct. 2015;7(2):816–24.
5. Heck CI, de Mejia EG. Yerba Mate tea (Ilex paraguariensis): a comprehensive review on chemistry, health implications, and technological considerations. J Food Sci. 2007;72(9):R138–51.
6. Martinet A, Hostettmann K, Schutz Y. Thermogenic effects of commercially available plant preparations aimed at treating human obesity. Phytomedicine. 1999;6(4):231–8.
7. Alkhatib A. Yerba Maté (Illex Paraguariensis) ingestion augments fat oxidation and energy expenditure during exercise at various submaximal intensities. Nutr Metab (Lond). 2014;11:42.
8. Alkhatib A, Atcheson R. Effects of Yerba Maté ingestion on fat utilisation during prolonged moderate intensity exercise. Proc Nutr Soc. 2015;74:E297.
9. Somerville V, Bringans C, Braakhuis A. Polyphenols and performance: a systematic review and meta-analysis. Sports Med. 2017:1–11.
10. Roelands B, Meeusen R. Alterations in central fatigue by pharmacological manipulations of neurotransmitters in normal and high ambient temperature. Sports Med. 2010;40(3):229–46.
11. Jatana N, Apoorva N, Malik S, Sharma A, Latha N. Inhibitors of catechol-O-methyltransferase in the treatment of neurological disorders. Cent Nerv Syst Agents Med Chem. 2013;13(3):166–94.
12. Burke LM. Re-examining high-fat diets for sports performance: did we call the ‘nail in the coffin’ too soon? Sports Med. 2015; 45(1 Suppl):S33–49.
13. Rowlands DS, Hopkins WG. Effects of high-fat and high-carbohydrate diets on metabolism and performance in cycling. Metabolism. 2002;51(6):678–90.
14. Santos EC, Bicca MA, Blum-Silva CH, et al. Anxiolytic-like, stimulant and neuroprotective effects of Ilex paraguariensis extracts in mice. Neuroscience. 2015;292:13–21.
15. Hodgson AB, Randell RK, Jeukendrup AE. The effect of green tea extract on fat oxidation at rest and during exercise: evidence of efficacy and proposed mechanisms. Adv Nutr. 2013;4(2):129–40.
16. Marx F, Janssens M, Urfer P, Scherer R. Caffeine and theobromine composition of mate (Ilex paraguariensis) leaves in five plantations of Misiones, Argentina. Plant Foods Hum Nutr. 2003;58(3):1–8.
17. Hawley JA, Noakes TD. Peak power output predicts maximal oxygen uptake and performance time in trained cyclists. Eur J Appl Physiol Occup Physiol. 1992;65(1):79–83.
18. Harwood M, Danielewska-Nikiel B, Borzelleca JF, Flamm GW, Williams GM, Lines TC. A critical review of the data related to the safety of quercetin and lack of evidence of in vivo toxicity, including lack of genotoxic/carcinogenic properties. Food Chem Toxicol. 2007;45(11):2179–205.
19. Davis JM, Carlstedt CJ, Chen S, Carmichael MD, Murphy EA. The dietary flavonoid quercetin increases VO(2max) and endurance capacity. Int J Sport Nutr Exerc Metab. 2010;20(1):56–62.
20. Torrens SM, Areta JL, Parr EV, Hawley JA. Carbohydrate dependence during prolonged simulated cycling time trials. Eur J Appl Physiol. 2016;116(4):781–90.
21. Pfeiffer B, Cotterill A, Grathwohl D, Stellingwerff T, Jeukendrup AE. The effect of carbohydrate gels on gastrointestinal tolerance during a 16-km run. Int J Sport Nutr Exerc Metab. 2009;19(5):485–503.
22. Peronnet F, Massicotte D. Table of nonprotein respiratory quotient: an update. Can J Sport Sci. 1991;16(1):23–9.
23. Moseley L, Jeukendrup AE. The reliability of cycling efficiency. Med Sci Sports Exerc. 2001;33(4):621–7.
24. Stisen AB, Stougaard O, Langfort J, Helge JW, Sahlin K, Madsen K. Maximal fat oxidation rates in endurance trained and untrained women. Eur J Appl Physiol. 2006;98(5):497–506.
25. Thu OK, Spigset O, Nilsen OG, Hellum B. Effect of commercial Rhodiola rosea on CYP enzyme activity in humans. Eur J Appl Physiol. 2016;72(3):295–300.
26. Hackett J, Telepchak MJ, Coyer MJ. Analysis of total caffeine and other xanthines in specialty coffees using mixed mode solid-phase extraction and liquid chromatography-diode-array detection after microwave digestion. J Anal Toxicol. 2008;32(8):695–701.
27. Singleton VL, Rossi JA. Colorimetry of total phenolics with phosphomolybdic–phosphotungstic acid reagents. Am J Enol Vitic. 1965;16(3):144–58.
28. Wangensteen H, Samuelsen AB, Malterud KE. Antioxidant activity in extracts from coriander. Food Chem. 2004;88(2):293–7.
29. Hopkins WG. A spreadsheet for analysis of straightforward controlled trials. Sport science. 2003;7.
30. Rowlands DS. Model for the behaviour of compartmental CO(2) stores during incremental exercise. Eur J Appl Physiol. 2005;93(5–6):555–68.
31. Montain SJ, Hopper MK, Coggan AR, Coyle EF. Exercise metabolism at different time intervals after a meal. J Appl Physiol (1985). 1991;70(2):882–8.
32. Bonadonna RC, Groop LC, Zych K, Shank M, DeFronzo RA. Dose-dependent effect of insulin on plasma free fatty acid turnover and oxidation in humans. Am J Physiol Endocrinol Metab. 1990;259(5):E736–50.
33. Jentjens R, Cale C, Gutch C, Jeukendrup A. Effects of pre-exercise ingestion of differing amounts of carbohydrate on subsequent metabolism and cycling performance. Eur J Appl Physiol. 2003;88(4–5):444–52.
34. Martin-Du Pan R, Mauron C, Glaeser B, Wurtman R. Effect of various oral glucose doses on plasma neutral amino acid levels. Metabolism. 1982;31(9):937–43.
35. Achten J, Jeukendrup AE. The effect of pre-exercise carbohydrate feedings on the intensity that elicits maximal fat oxidation. J Sports Sci. 2003;21(12):1017–25.
36. Stadheim HK, Kvamme B, Olsen R, Drevon CA, Ivy JL, Jensen J. Caffeine increases performance in cross-country double-poling time trial exercise. Med Sci Sports Exerc. 2013;45(11):2175–83.
37. Spriet LL. Exercise and sport performance with low doses of caffeine. Sports Med. 2014;44(2):175–84.
38. Jenkins NT, Trilk JL, Singhal A, O’Connor PJ, Cureton KJ. Ergogenic effects of low doses of caffeine on cycling performance. Int J Sport Nutr Exerc Metab. 2008;18(3):328.
39. Desbrow B, Barrett CM, Minahan CL, Grant GD, Leveritt MD. Caffeine, cycling performance, and exogenous CHO oxidation: a dose–response study. Med Sci Sports Exerc. 2009;41(9):1744–51.
40. Ryan EJ, Kim CH, Muller MD, et al. Low-dose caffeine administered in chewing gum does not enhance cycling to exhaustion. J Strength Cond Res. 2012;26(3):844–50.
41. Doherty M, Smith P. 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.
42. Solinas M, Ferré S, You ZB, Karcz-Kubicha M, Popoli P, Goldberg SR. Caffeine induces dopamine and glutamate release in the shell of the nucleus accumbens. J Neurosci. 2002;22(15):6321–4.
43. Frandsen U, Lopez-Figueroa M, Hellsten Y. Localization of nitric oxide synthase in human skeletal muscle. Biochem Biophys Res Commun. 1996;227(1):88–93.
44. Lane SC, Areta JL, Bird SR, et al. Caffeine ingestion and cycling power output in a low or normal muscle glycogen state. Med Sci Sports Exerc. 2013;45(8):1577–84.


Supplemental Digital Content

© 2018 American College of Sports Medicine