Dose Effect of Caffeine Ingestion on Cardiac Parasympathetic Reactivation Following an Acute Bout of Anaerobic Performance Exercise in Recreational Athletes


 Background: To examine the dose effects of caffeine ingestion on autonomic reactivation following anaerobic exercise. Methods: Recreationally active males (N=20; 24±2y) participated in a randomized, double-blind, placebo-controlled, crossover study where participants ingested: (1) Control (CON; no supplement), (2) a non-caffeinated placebo (PLA), (3) 3-mg∙kg-1 of caffeine (CAF3) or (4) 6-mg∙kg-1 of caffeine (CAF6) prior to a Wingate testing. Parasympathetic (lnRMSSD, primary outcome) and global HRV (lnSDNN, secondary outcome) were assessed at rest (i.e., pre-ingestion), 45-min post-ingestion, and 5-min and 35-min post-exercise recovery. We used a GLM to assess mean (95% CI) changes from pre-ingestion baseline.Results: Overall, we observed a significant trend for lnRMSSD and lnSDNN (both, p=0.001, ηp2=0.745). Forty-five minutes after treatment ingestion, we observed a significant increase in lnRMSSD for CAF3 (0.15ms, 95%CI, 0.07,0.24) and CAF6 (0.16ms, 95%CI, 0.06,0.25), both being significant (both, p<0.004) vs. CON (-0.02ms, 95%CI, -0.09,0.04). Five-minutes after exercise, all treatments demonstrated significant declines in lnRMSSD vs. baseline (all, p<0.001). After 35-min of recovery, lnRMSSD returned to a levels not significantly different than baseline for CAF3 (0.03ms, 95%CI, -0.05, 0.12) and CAF6 (-0.03ms, 95%CI, -0.17, 0.10), while PLA (-0.16ms, 95%CI, -0.25, -0.06) and CON (-0.17ms, 95%CI, -0.28, -0.07) treatments remained significantly depressed. A similar pattern was also observed for SDNNConclusion: Caffeine ingestion increases resting cardiac autonomic modulation and accelerates post-exercise autonomic recovery after a bout of anaerobic exercise test in recreationally active young men. However, no differences between caffeine doses on cardiac autonomic reactivity were observed.


Introduction
Adaptations to exercise training require an appropriate training stimulus accompanied by su cient recovery (1). One factor associated with exercise training and recovery is the balance between the sympathetic and parasympathetic nervous branches of the autonomic nervous system (ANS) (2). Accordingly, an imbalance in ANS between the training stimuli and recovery can may lead to ANS dis-regulation and negatively impact exercise performance (1,3).
Fortunately, ANS can be assessed easily and non-invasively assess heart rate variability (HRV) via heart rate monitors, thus providing for a useful tool for sports performance and recovery from strenuous exercise.
During exercise, heart rate increases via parasympathetic withdrawal and increased sympathetic activity (2). HRV becomes useful during training and as it a validated tool to assess internal training load during and following exercise allowing for the individual evaluation surrounding training stimuli (1,3,4). Despite the potential implications of HRV, the time course of HRV following exercise is multifactorial and varies based on genetics, and the intensity, duration, and mode of exercise (2,(4)(5)(6). Accordingly, exercise intensity seems to play the greatest role relative an athlete during acute post exercise HRV recovery, in particular parasympathetic reactivation, is delayed following high intensity exercise (4,5,7). A potential, yet relatively unexplored means of improving parasympathetic activity following strenuous exercise is the ingestion of caffeine.
Caffeine is one of the most popular ergogenic compounds used in sport and supported by a large body of scienti c evidence for improving anaerobic and aerobic activities (8)(9)(10)(11). The ingestion of caffeine in uences the ANS via an increase in catecholamine secretion, which subsequently increases heart rate and mean arterial blood pressure at rest and during exercise (12). Caffeine acts as a sympathetic stimulus during exercise and has been shown to attenuate autonomic recovery post-exercise (13,14). However, research ndings are mixed, with some studies showing that 300-400 mg of caffeine delays postexercise parasympathetic reactivation (12,14,15), while others have found no effect from caffeine at doses associated with < 3 mg•kg − 1 body mass (16,17).
While it is di cult to compare and contrast results across studies, these inconsistent ndings may be related to the caffeine dosage used, which warrants further investigation within athletic populations. The purpose of this study was to examine the effects of two dosages of caffeine (3 mg•kg − 1 and 6 mg•kg − 1 ) on HRV indices of the ANS following a single bout of high intensity exercise. The primary outcome measure was root mean square of the successive differences (rMSSD), a time domain HRV index re ecting parasympathetic tone. The secondary outcome for our study was the standard deviation of the NN (R-R) intervals (SDNN). We hypothesized that caffeine ingestion would reactivate parasympathetic markers following a reduction immediately following exercise.
questionnaires, such as the PAR-Q and HHQ. Participants also recorded their dietary intake prior to the rst experimental day and were asked to replicate this diet on subsequent visits. Height (cm) was measured with an electronic stadiometer (SECA 217, Seca Ltd, Hamburg, Germany) to the nearest 0.01 cm without shoes and with each participant standing erect against a wall and body mass (kg) was measured to the nearest 0.01 kg using a calibrated digital scale (Seca 770-oor, Seca Ltd, Hamburg, Germany). During the second visit participants completed a 30-sec all-out sprint test Wingate Anaerobic Test (WAnT) for familiarization and to reduce the learning effect (19,20).
During the remaining experimental visits (3)(4)(5)(6), participants performed the experimental treatments based on computer generated randomization of four treatments: (1) control (no treatment), (2) placebo (PLA, non-caffeinated treatment matched for taste and color), (3) 3 mg•kg − 1 of caffeine (CAF3) and (4) 6 mg•kg − 1 of caffeine (CAF6). All experimental visits were conducted at the same time of day (between 11:00 and 14:30). To minimize potential gastrointestinal distress, participants consumed a standardized snack (white bread and boiled eggs) containing 3 g•kg − 1 of carbohydrates, 20 g of protein, and 10 g fat 180-240 min before each session. A schematic of the testing protocol is presented in Fig. 1 and is detailed below: (Time 1) Resting: Pre-Ingestion. Prior to treatment ingestion, participants sat quietly for 5-min before providing a resting blood sample to assess blood lactate (BLa; Lactate Scout plus analyzer, SensLab GmbH, Germany). A resting HRV was also measured. Following these baseline measurements participants then ingested their respective treatments.
(Time 2) Resting: 45-min Post Ingestion. Forty-ve minutes after treatment ingestion, participants were assessed again for HRV and BLa. Following this assessment, participants engaged in a standardized warm-up before performing WAnT testing (detailed below).

Wingate Anaerobic Test
Participants performed a Wingate Anaerobic Test (WAnT) using a friction-loaded cycle ergometer (MONARK 894E, Stockholm, Sweden) connected to a computer as previously described (19) (20,21). The ergometer was calibrated before each test. Brie y, the WAnT was a 30-sec test requiring participants to pedal as fast as possible against a xed resistance. Each test began with a 4-min standardized warm-up against a xed load of 1 kilopond and three separate 2 second sprints were performed against a load of 0.075 kp•kg − 1 of body mass were interspersed with 45-sec of active recovery (20). After the warm-up, one minute of light dynamic stretching was performed. The test protocol began with a verbal command of "faster, faster, 3,2,1, go", during which the load of 0.075 kp•kg − 1 body mass (22) was applied. At "go" the participants were pedaling as fast as possible and each participant was verbally encouraged to maintain as high of a cadence as possible over the entire 30-sec. No visual or verbal feedback regarding the time to complete the test was provided.
Power output in Watts was calculated as the product of resistance and ywheel revolutions, which was recorded every 1-sec, peak power output was determined from the average of the rst 5-sec a mean power output was assessed as the average over the entire 30-sec of the test. Fatigue index was calculated from the difference between peak 5 second power output and the power output that occurred during the nal 5-sec of the test divided by the peak power output multiplied by 100 (23). The seat height and handlebars were adjusted such that the knee would be slightly bent at maximal leg extension and kept constant throughout the remaining experimental sessions. likelihood of Type I error (SPSS ® version 25, IBM North America, New York, NY, USA). Normality was examined using Kolmogorov-Smirnov test and con rmed for all exercise related variables. However, we observed that HRV indices were not normally distributed. Therefore, we performed a natural log transformation for all HRV measures. Effect sizes are presented as partial Eta squared (ηp 2 ) for the general linear models and Cohen's D for simple effects.
All calculations were accomplished using) and the probability level for statistical signi cance was pre-set at P = 0.05, while ES were calculated using means and pooled standard deviations (SD). Effect sizes for partial eta squared were interpreted as: 0.01, small; 0.06, medium and > .14, large (28). Effect sizes for Cohen's D were interpreted as 0.20 = small, 0.50 = moderate, 0.80 = large (29). Data throughout the manuscript are presented as mean (SD) or mean change (95% CI).

Results
We have presented the results for WAnT testing in Table 1, inclusive of RPE and blood lactate. Further, we have presented a schematic representation of the study procedures in Fig. 1 and our ndings for lnRMSSD and lnSDNN in Fig. 2. In brief, we observed that overall, participants generated a peak power output of 773.84 ± 178.41 W, a mean power of 502.96 ± 86.46, achieved a maximum exercise heart rate of 155.04 ± 7.36 b/min − 1 , a maximal RPE of 16.37 ± 1.68, a maximal blood lactate level of 4.80 ± 0.65 mmol/L at peak exercise and 5-min post exercise 8.26 ± 1.08). Between group comparisons showed a signi cantly and CAF6 treatments increased lnRMSSD signi cantly 45-min after treatment ingestion (both, p < 0.001) both CAF treatments to be signi cant vs. CON (p < 0.004). Thirty-ve minutes after the completion exercise, both CAF3 and CAF6 treatments demonstrated a return of lnRMSSD to values not signi cantly different to baseline, while the CON and PLA treatments remained signi cantly depressed (both, p < 0.001). For this latter assessment, the CAF3 treatment was signi cantly higher than CON (p < 0.004) and PLA (p < 0.011) treatments (Fig. 2).
Secondary Outcome: lnSDNN: For lnSDNN, we observed signi cant treatment effects 45-min post-treatment ingestion (p < 0.023, ηp 2 = 0.117) and 35-min post exercise performance (p = 0.017, ηp 2 = 0.124), but not at 5-min post exercise (p = 0.784, ηp 2 = 0.014), BMI 23.3 (SD). Forty-ve minutes following treatment ingestion and before exercise testing, we observed a signi cant increase from baseline for lnSDNN in the CAF3 and CAF6 treatment conditions (both, p < 0.001). Both the CAF3 and CAF6 treatments were signi cant vs. the CON treatment (p < 0.004). Following WAnT testing, we observed a signi cant reduction in lnSDNN for all treatments after 5-min of recovery compared to baseline, pre-ingestion (all, p < 0.001). After 35-min of exercise, we observed a continued reduction in lnSDNN for the CON and PLA treatments (both, p < 0.001); however, no signi cant reductions in lnSDNN vs. baseline, pre-ingestion were noted for the CAF3 and CAF6 treatments. For this latter assessment, lnSDNN was signi cantly greater for the CAF3 and CAF6 treatments vs. PLA (p < 0.003). In summary, the HRV indices measured in this study returned to baseline conditions for both caffeine treatments, while the same indices remain reduced compared to baseline following a single bout of strenuous exercise.

Discussion
The primary objective of this study was to analyze the effects of different caffeine dosages on resting and post-exercise cardiac autonomic modulation.
While higher cardiac parasympathetic and global modulations were observed after CAF3 and CAF6 ingestion during the resting condition, no such effects were noted for the PLA and CON groups. Further, while all treatment groups demonstrated a signi cant reduction in lnRMSSD and lnSDNN 5-min following exercise, no between treatment effects were noted. Finally, given the continual HRV suppression for the PLA and CON groups at 35-min post-exercise, compared to the restoration of said indices for the CAF3 and CAF6 treatments to levels not signi cantly different from baseline, we conclude that the CAF ingestion in the quantities used in this study are su cient to accelerates post-exercise autonomic recovery following a single bout of strenuous exercise.
Based on these observations we accept our research hypothesis.
The effects of caffeine ingestion on resting HRV are con icting, with studies reporting increases (30-32), reduction (33) and no changes (34) of resting parasympathetic and/or global modulation markers. Establishing a cause of these divergences is not an easy task since several variables can affect HRV analysis, such as sex (35), body position (36), body mass index (37), nutritional status (38), functional condition (39), corresponding heart rate (40), cardiorespiratory tness (41) and age (42). In that same sense, the physiological and functional response to caffeine ingestion also depends on various factors such as individual caffeine habituation (43), caffeine dosage (44), sex (45), functional condition adopted to analysis (34), genetic pro le (46), caffeine expectancies (47) and some other neuromuscular characteristics (48). Thus, it is plausible to infer that the autonomic response to caffeine ingestion is dependent on several independent variables, and the increase of cardiac parasympathetic and global modulations observed in this study may be limited to our study design and participants' characteristics.
Regarding the caffeine dosage effect, no differences between CAF3 and CAF6 on resting autonomic dynamics were observed by this study. Previous studies showed that both 2 (30) and 5 mg•kg − 1 (32) of caffeine, dosage close to those adopted in this study, were able to increase cardiac parasympathetic modulation. In this scenario, our results reinforce the possibility of increasing parasympathetic modulation after caffeine intake and add important information suggesting that the relationship between caffeine dosage and parasympathetic reactivity is not linear. No changes on lnRMSSD and lnSDNN were observed after 45 min of resting on PLA and CON groups, suggesting no effect of resting time on rest cardiac autonomic modulation. In opposition, a signi cant effect of resting time on supine and orthostatic cardiac parasympathetic and global modulation was previously observed after 60 minutes of resting in the supine position in young men (34). In this same scenario, Zimmermann-Viehoff et al. (2015) observed a signi cant effect of resting time on HRV parameters in a sample of young men and women from 30 to 50 minutes of rest at the seated position (49). Despite ours and Zimmermann-Viehoff et al (2015) studies using the seated position to analyze HRV, some differences between them and may explain the con icting results observed. These differences include the amplitude of R-R interval segments used for HRV analysis, rest time before the nutritional intervention and sample characteristics. Thus, these data indicate that the effect of resting time on HRV may be protocol-dependent and should be considered in studies involving the effect of different pharmacological and non-pharmacological interventions on HRV.
In the initial post-exercise analysis, all treatment groups demonstrated a signi cant reduction in lnRMSSD and lnSDNN, but no differences between treatments were noted. However, after 35 min of passive recovery no differences between rest and pos-texercise lnRMSSD and lnSDNN were identi ed in CAF3 and CAF6 protocols, while a persistent depression of these autonomic markers was identi ed in CON and PLA groups. Thus, these results con rm our initial hypothesis (please, check it) that caffeine intake can boosts post-exercise cardiac autonomic recovery. Corroborating our results, Rolim et al. (2018) observed a higher post-exercise cardiac parasympathetic reactivation after a submaximal exercise test in young men after caffeine uptake (3 mg•kg − 1 ), despite no changes in resting markers of cardiac autonomic modulation (34). On the other hand, Kliszczewicz et al. (2018) underwent ten physically active young males to Wingate anaerobic test, and no effect of a complex containing caffeine (100 mg) + Citrus Aurantium (100 mg) was observed on post-exercise parasympathetic and sympathetic activity, despite higher resting sympathetic activity compared to the placebo condition (17). Otherwise, Bunsawat et al (2014) suggest that caffeine can promote a sympathetic over activation after maximal exercise, a hypothesis based mainly on higher absolute heart rate and blood pressure after an exercise test (14). However, higher training load and maximum heart rate were observed in Bunsawat's study after caffeine intake, with no differences in heart rate recovery at the rst and the second-minute post-exercise, which makes the caffeine-induced sympathetic over activation hypothesis questionable. In other words, higher post-exercise absolute heart rate and blood pressure may occur due to a higher training load and not necessarily a direct effect of caffeine ingestion.
From a physiological perspective, the autonomic response to caffeine uptake is complex and may be bilateral. Was previously reported that caffeine ingestion could promote a signi cant increase in plasma levels of catecholamines (50,51) and inhibits the enzymatic degradation of cyclic adenosine monophosphate by phosphodiesterases, which potentiates postsynaptic neurotransmission in the sympathetic nervous system (52) On the other hand, despite parasympathetic response to caffeine uptake remain underexplored, it has been shown that caffeine can stimulate acetylcholine receptors and acts as an inhibitor of acetylcholinesterase (53,54), which explains, at least in part, the caffeine-induced increase in parasympathetic activity reported in our and previous studies. In addition, it has been hypothesized that the caffeine-induced parasympathetic activation may be a result of barore ex activation due to an increase in peripheral vascular resistance and blood pressure resulting from antagonistic caffeine effect on adenosine receptors (34), which need to be con rmed in future studies.
Notwithstanding a lack of difference between group mean power observed in caffeine and placebo protocols, higher peak power was observed in CAF3 and CAF6 compared to PLA and control reveal an ergogenic effect of caffeine on anaerobic performance. Also, higher peak power in CAF6 compared to CAF3 indicate that this ergogenic effect is dose dependent. Previously, a lack of effect and even reduction in anaerobic performance after caffeine consumption has already been reported in the literature (50). However, a recent meta-analysis using studies of good and excellent methodological quality reveal that caffeine intake can augment mean and peak power output on the Wingate anaerobic test by 3% and 4%, respectively (11). Interestingly, in our study, higher cardiac parasympathetic reactivation after caffeine intake was observed even in the face of higher peak power in CAF3 and CAF6 compared to control and PLA protocols. This nding strengthens the favorable effect of caffeine on post-exercise parasympathetic reactivation since an inverse relationship between exercise intensity and the magnitude of parasympathetic reactivation is expected (55,56).
While a higher fatigue index in caffeine protocols compared to control, no differences between CAF3/CAF6 and placebo protocol were identi ed. Thus, claiming that caffeine supplementation reduces exercise tolerance in this study is not a correct approach. Of note, examining the effect of caffeine supplementation on repeated bouts of Wingate tests (four 30-s Wingate tests with 4 min of rest between each exercise) after caffeine (6 mg•kg − 1 ) or placebo ingestion, Greer, McLean, and Graham (1998) observed that caffeine ingestion had an ergolytic effect in the latter two exercise bouts (50). Otherwise, it was recently reported that caffeine supplementation (6 mg•kg − 1 ) increased the peak power during Wingate anaerobic test and diminished neuromuscular fatigue, shown by attenuation of decrease in countermovement jump performance after Wingate test (57). Thus, since increase (57, 58) and reduction (50,59) of different markers of exercise tolerance after caffeine supplementation already been reported, the recommendation of caffeine supplementation to improve recreational or athletic performance should be made cautiously.
Despite no observed difference between RPE in caffeine and placebo during warm-up, the main effect of treatment and lower RPE observed after CAF6 compared to control and PLA indicates that caffeine may reduce the exercise-induced psychological stress. Interestingly, Duncan et al (2019) observed a reduction of RPE during Wingate test for the upper-body, but not for the lower-body segment, suggesting that caffeine's effect on RPE depends on body segment exercised (60). Despite the absence of caffeine effect on RPE during lower-body Wingate test observed in some studies (60)(61)(62), our ndings reveal that this bene t can be acquired with caffeine supplementation in this condition. We note that lower RPE identi ed in CAF6 protocol was accompanied by high peak power and mean power, which reinforce the psychostimulant effect of caffeine. It is an interesting approach since increases in exercise performance without altering RPE mean a higher power output without the increase in psychological stress per se; this positive effect should also be investigated in future studies.
As expected, an increase of BLA was observed after WAnT in all protocols indicating the vital contribution of anaerobic metabolism to the energy requirements during the exercise test. Despite increase (63) and maintenance (60) of BLA levels are commonly reported after caffeine intake, lower BLA concentration was observed in CAF3 compared to other protocols after ve minutes of recovery. Unfortunately, the only lactate analysis performed in the initial phase of recovery does not permit to detect the exact moment with the highest lactate concentration, which makes any inference about the effect of caffeine on lactate production or clearance questionable. In the nal phase of post-exercise recovery, we observed higher BLA levels in CAF6 compared to PLA, but the absence of difference between CAF6 and control prevents the attribution of higher blood lactate to caffeine supplementation. Of note, blood lactate re ects the balance between lactate production and clearance and the precise mechanisms that explain the small differences observed in this study is unclear and it may be just a inter day variation of BLA response to exercise, hypothesis previously reported in the literature (64,65).

Limitations
A major strength of our study is our randomized, cross-over design. A limitation of our study is the use of a single, acute bout of WAnT testing. Therefore, we cannot generalize our ndings to higher exercise volume conditions, such as multiple WAnT testing, multiple sets of resistance training, interval style workouts. We also cannot generalize our ndings to women. The absence of ventilatory, sympathetic activity, and post-exercise blood pressure analysis, variables that in uence HRV could also contribute to a better physiological interpretation of our data. We believe that a particular strength of our study was the use of a non-supplemented CON condition in addition an inert PLA and support this contention that a number of between group comparisons in our study were signi cant vs. the CON, but not the PLA treatments. Lastly, we assessed cardiac parasympathetic reactivation during 35 minutes of recovery, which limits our conclusions to this time window. However, despite the mentioned limitations, the analysis of autonomic response to different caffeine supplementation dosages on resting and post-exercise conditions adopted in this study adds robust information to current scienti c debate about the autonomic effect of caffeine ingestion. The post-exercise time window adopted in this study allow fast and slow parasympathetic reactivation analysis and is within of window of opportunity for sudden death in young observed 30 minutes after vigorous exercise, which can be partially attributed to post-exercise cardiac autonomic dysfunction (66) and add clinical relevance to our results.

Conclusion
We conclude that caffeine ingestion increases resting cardiac autonomic modulation and accelerates post-exercise autonomic recovery after a bout of anaerobic exercise test in recreationally active young men. However, no differences between caffeine doses (3 or 6 mg•kg -1 ) on cardiac autonomic reactivity were observed.  Values in the same row and subtable not sharing the same subscript are signi cantly different at p < 0.05 in the two-sided test of equality for column means. Tests are adjusted for all pairwise comparisons within each row using a Bonferroni correction. Figure 1 Schematic representation of the study procedures.