In fl uence of aerobic fi tness on the correspondence between heart rate variability and ventilatory threshold

Lúcio Flávio Soares-Caldeira */** Carla Cristiane da Silva *** Priscila Chierotti * Nicolle de Souza Dias * Fábio Yuzo Nakamura */**** *Sport and Physical Educat ion Center, L o n d r i n a S t a t e University, Londrina, PR, Brazil. **Research Centre on Health Sciences, Department of Physical Education, University of Northern Paraná, Londrina, PR, Brazil. * * * D e p a r t m e n t o f Physical Education, S t a t e U n i v e r s i t y of North of Paraná, J a c a r e z i n h o , P R , Brazil. ****Nucleus of High Performance in Sport, São Paulo, SP, Brazil. Abstract


Introduction
Infl uence of aerobic fi tness on the correspondence between heart rate variability and ventilatory threshold Th e autonomic nervous system (ANS) is the primary physiological pathway which regulates heart rate (HR) by means of sympathetic and parasympathetic activities on the sinus node 1 . In response to exercise stress, HR is accelerated due to cardiac vagal withdrawal concomitant with increased sympathetic outfl ow 2 . Th e beat-to-beat oscillation of HR [i.e., heart rate variability (HRV)] is a non-invasive method to assess changes in the balance between the ANS branches [3][4][5] . Heart rate variability at rest and during low-intensity exercise has been shown to be positively correlated with aerobic fi tness 6 and responsive to aerobic and sports training eff ects 4,7,8 .
In response to the physiological stress caused by a progressive exercise test, pulmonary ventilation increases in a non-linear fashion, as a refl ection of the subtle increase in the muscular anaerobic contribution to ATP resynthesis 9 . In theory, the ventilatory threshold (VT) corresponds to the rise in blood lactate above resting levels [i.e., lactate threshold (LT)] 10 . Recently, HRV responses during incremental tests have been used to determine the heart rate variability threshold (HRVT), which is purported to coincide with VT and LT 11 . According to Karapetian 11 , HRVT can be defi ned as the point at which there is no further decline in time-domain HRV indices, thus indicating vagal withdrawal, however, the correspondence between HRVT, VT, and LT was tested in subjects with a wide variation in aerobic fi tness. A close examination of fi gure 3 in of the Karapetian's 11 paper reveals considerable individual discrepancies in the comparison between autonomic and metabolic thresholds. Th is suggests that the high correlation between HRVT and VT (r = 0.89; ranging from ~0.8 -2.5 L/min of oxygen output consideringmeans of HRVT and VT) may be weaker in a more homogeneous groups (i.e., it is possible that there is a low correlation between the VT and HRVT when analyzed separately between individuals with low and high means of HRVT and VT). It is suggested that increases in sympatheticadrenal activity and subsequent catecholamine release are responsible for stimulating glycogenolysis and ventilation breakpoints 11 . Although it is tempting to mechanistically link HRVT, LT, and VT, it is accepted that is some cases this correspondence may fail, indicating that factors other than sympathetic outflow and anaerobic glycolysis can stimulate pulmonary ventilation 12,13 , such as the vagal autonomic tone resulting in fast parasympathetic withdrawal in oscillations of HR in individuals with higher aerobic fi tness 14,15 , during progressive tests. Th is hypothesis could indicate reduced infl uence of the respiratorymetabolic mechanisms linked to the occurrence of HRVT in individuals with higher aerobic fi tness.
Th e lack of correspondence between HRVT and VT may reveal diff ering mechanistic bases underlying these threshold intensities and raises concerns on the use of the former as a simple and non-invasive alternative to aerobic capacity evaluation and training intensity prescription, to replace VT. Th erefore, the aims of this study were threefold: 1) to compare the HRV changes during progressive exercise performed by groups possessing low and high aerobic fi tness levels; 2) to compare the HRVT and VT between groups with low and high aerobic fi tness levels and; 3) to verify the correspondence between HRVT and VT within each aerobic fi tness group.

Methods
Th e study involved two groups, divided according to aerobic fi tness levels (low and high). For each group, oxygen consumption (VO 2 ) at VT and at HRVT were determined from same progressive cycling test. One subject from the low maximal oxygen consumption (VO 2 max ) group was excluded from the analyses due to abnormal ventilatory responses which precluded determination of VT. Th e VO 2 at VT and HRVT was compared within and between groups to ascertain the role of aerobic fi tness, as assessed by VO 2 max, on the cardiac autonomic response to exercise and on the correspondence between HRVT and VT. According to the median of (48.8 ml•kg −1 •min −1 ), the participants were evenly split into two groups: low cardiorespiratory fi tness group (range: 37.2 -44.2 ml•kg −1 •min −1 ; n = 10) and high cardiorespiratory fi tness group (range: 49.5 -61.3 ml•kg −1 •min −1 ; n = 10). Th e cut-off value established for high cardiorespiratory fi tness group was near the 95% percentile of assumed as ~50 and 47.3 ml•kg −1 •min −1 for men aged 25-34 and 35-44 16 , respectively.
Twenty male volunteers aged between 20 and 44 years of age (29.5 ± 6.2 years) were recruited. Prior to data collection, all participants were initially surveyed to gather information on their general characteristics. Individuals were included in the study if they met the following inclusion criteria: were healthy with physically active or sedentary lifestyles in the previous six months [using the International Physical Activity Questionnaire criteria (http://www.ipaq.ki.se/)], nonsmokers, presented no cardiovascular dysfunction (asked in an anamnesis procedures considering the last medical appointment), and were not under drug administration within at least four weeks before of the incremental test. All participants signed an informed consent and the procedures were approved by the local Ethics Committee (process number 091/2013). TABLE 1 displays the physical and physiological characteristics of the low and high cardiorespiratory fi tness groups. Th ere were no signifi cant diff erences between groups regarding age, height, and HRmax. Th e cardiorespiratory fi tness (and maximal power output Procedures -Wmax) was higher for the group with high aerobic fi tness in comparison with the group with low aerobic Physical and physiological characteristics of the low cardiorespiratory fi tness group; (n = 10) and high cardiorespiratory fi tness group; (n = 10) from the median of VO 2 max (48.7 ml•kg −1 •min −1 ). Mean values, (± standard deviation) and confi dence intervals [CI 95%].

Exercise testing
Subjects were instructed to avoid food or beverages containing caff eine for 24 hours and vigorous exercise for 48 hours prior to testing. Upon arrival at the laboratory, body mass and height were measured with a calibrated scale and a stadiometer to the nearest 0.02 kg and 0.1 cm, respectively. Th e body mass index (BMI) was calculated by dividing the weight (kg) by the square of the height (cm).
The progressive test was performed on an electromagnetically braked cycle ergometer (LODE, Excalibur Sport, Groningen, Netherlands) in a laboratory with a controlled temperature (~23º C), and a fan directed to the participant when required, in order to improve thermal comfort during the test. Th e testing protocol consisted of 3-min stages 11 beginning at 25 watts and increasing 25 watts every 3-min 15 . fi tness. Body weight and BMI were also higher in the lower aerobic fi tness group.

Statistical analysis
Th e 3-min stages were chosen for the progressive test as this enables a higher quality stationary signal of RR intervals to be obtained. Subjects were instructed to maintain a pedal cadence of between 60 and 90 revolutions per min (rpm). Th e exercise test duration ranged from 15 to 40 min, depending on each subject's cardiorespiratory fi tness. Subjects fi nished the test when they reached volitional exhaustion or felt unable to maintain the pedal cadence >60 rpm. Th roughout the test, participants were verbally motivated to give their best performance.

Determination of heart rate variability threshold (HRVT)
Heart rate variability was measured using a telemetric device (Polar RS800 Electro Oy, Kempele, Finland) used to record each subject's RR intervals (beat-tobeat fl uctuation in HR) throughout the test. Th e RR interval data were stored in the receiving watch and subsequently uploaded to a computer for analysis using Polar Pro Trainer software (version 5.0). All analyses were performed using Kubios HRV Analysis Software 2.0 (Biosignal Laboratory, University of Kuopio, Finland). Th e RR intervals from the fi nal 2 min of each progressive test stage were used for analysis of HRV 11 , avoiding the infl uence of non-stationary RR intervals on HR oscillations, commonly observed in the early stages. Th e RR interval was automatically interpolated in cases where it deviated from the previous interval by >30%, followed by visual inspection to correct for noise 15 . Th e standard deviation of all RR intervals (SDNN), which represents both sympathetic and parasympathetic activity and the square root of the mean squared diff erences of successive RR intervals (RMSSD) in milliseconds (ms), as a parasympathetic activity indicator, were retained for analysis 1 . To determine the HRVT, the RMSSD and SDNN for each stage were plotted against power output (W) or percentage of maximum workload (%W). Th e exercise intensity at which no further decline in SDNN and RMSSD values is observed defi nes the occurrence of HRVT (HRVT SDNN and HRVT RMSSD , respectively). Two experienced researchers (with at least 2 years of practical experience in assessments of cardiopulmonary parameters in exercise tests) with no access to the identity of the subjects performed the HRVT analyses. In the case of disagreement between power outputs, a third independent researcher performed the HRV determination. In all cases, the third result coincided with one of the previous results and this intensity was thus defi ned as the individual's HRVT, or the median those three outcomes. High inter-rater intraclass correlation coeffi cient (0.80) between the researchers considering HRVT was found.

Determination of ventilatory threshold (VT) and VO 2 max
Respiratory gas exchange was measured continuously breath-by-breath using a metabolic cart (Quark CPET, Cosmed, Italy) and averaged each 20-s in order to obtain VT and VO 2 max . Prior to each test, the equipment was calibrated using ambient air and gases of known O 2 (16%) and CO 2 (5%) concentrations. Th e turbine fl ow-meter was calibrated using a 3-L syringe. Th e calibration procedures followed the manufacturer's instructions.
Th e individual VO 2 max was determined as the highest value achieved in 20-s averages close to the end of the test. Although no secondary criterion was used to validate VO 2 max 17 , all subjects reported >19 rating of perceived exertion on the 6-20 Borg scale, determined in the fi nal 30 seconds of each stage 18 . Th e VT was determined at the power output corresponding to the increase in VE / VO 2 without a concomitant increase in VE /VCO 2 16 . Th e power output and VO 2 at VT have previously been considered a valid measure with high intraclass correlations using tests with increments of 25W per stage 19 . Each individual's VT was assessed by two experienced researchers involved in the study, who were blinded to the participants' identity and group allocation 20 . In cases of disagreement between the two researchers, a third researcher arbitrated. Th e HR, VO 2 , and RPE at VT were retained for analyses. Th e heart rate (HR) and VO 2 at VT and HRVT were calculated as the mean values from the fi nal 2-min of the stage.
The Shapiro-Wilk's test indicated normal distribution of data. Th us, the independent Student's t test was applied for comparisons between groups (low versus high cardiorespiratory fitness group) for age, anthropometric variables (weight, height, and BMI), physiological responses from progressive testing (VO 2 max, HRmax, and Wmax), physiological variables at VT, HRVT RMSSD , and HRVT SDNN (VO 2 , %VO 2 max, HR, %HRmax, RPE), load (watts), and relative load (watts•kg -1 ). One way ANOVA was used to compare the physiological and perceptual responses within each group at VT, HRVT RMSSD , and HRVT SDNN (VO 2 , %VO 2 max, HR, %HRmax, RPE), load (watts), and relative load (watts•kg -1 ). Th e two-way repeated measures ANOVA with Bonferroni correction was used to compare the values of RMSSD and SDNN (ms) at the percentages of maximum workload corresponding to from 10% to 100% between the high and low cardiorespiratory fi tness groups. Th e sphericity was checked using Mauchly's test and whenever the test was violated the necessary technical corrections were performed with the Greenhouse-Geisser test. Data are presented as mean±standard deviation (SD) and the signifi cance level was set at 5% (P<0.05). Pearson's correlation was performed to assess the relationship between VO 2 (ml•kg −1 •min −1 ) values at ventilatory

Results
The HRV (RMSSD and SDNN parameters) responses to the progressive test between the groups with different cardiorespiratory fitness levels are displayed in FIGURE 1. Th e pairwise comparisons between low and high cardiorespiratory fi tness showed there were no main eff ect signifi cant diff erences for RMSSD and SDNN at the percentage of maximum workload corresponding to from 10% to 100% of VO 2 max. Two-way repeated measures ANOVA demonstrated main eff ects only when considering the percentage of maximum workload factor for RMSSD (F=27.3; P<0.01) and SDNN (F=54.4; P<0.01). Th e group and interaction eff ects between the percentage of maximum workload and group were not signifi cant for either RMSSD (F=0.81; P>0.05) or SDNN (F=0.02; P>0.05). FIGURE 1 illustrates the HRVT identifi cation of two typical subjects from the low (Panel C) and high (Panel D) cardiorespiratory fi tness groups. In the low cardiorespiratory fi tness group, it was easier to visually determine the HRVT RMSSD and HRVT SDNN , and in this case they were coincident. On the other hand, in the high cardiorespiratory fi tness group, it was more diffi cult to clearly determine the HRVT thresholds, and the HRVT RMSSD and HRVT SDNN did not coincide in most cases. TABLE 2 presents the comparisons between VO 2 at VT and at the heart rate variability thresholds (HRVT RMSSD and HRVT SDNN ). Th e VO 2 at VT (+32%), HRVT SDNN (+31%), and HRVT RMSSD (+27), as well as the relative workloads to body mass at VT (+41%), HRVT SDNN (+33%), and HRVT RMSSD (+39), were higher in the group with high cardiorespiratory fi tness compared to the group with low cardiorespiratory fi tness. Th ere were no signifi cant diff erences between the low and high cardiorespiratory fi tness groups for the percentage of Heart rate expressed both in absolute values (bpm) and as a percentage of maximal heart rate (%HRmax) was significantly different (P<0.05) in the low cardiorespiratory group between the VT (80.8±4.2 %HRmax; 153.0±13.5 bpm) and the HRVT RMSSD (77.7±5.4 %HRmax; 146.8 ± 9.8 bpm), and between the VT and the HRVT SDNN (84.1±4.7 %HRmax; 159.2±12.3 bpm). In the high cardiorespiratory fi tness group, there were no diff erences (P>0.05) in heart rate between the VT (78.2±7.2 %HRmax; 147.8±17.7 bpm) and HRVT RMSSD (77.7±5.4 %HRmax; 139.9±10.0 bpm), or between the VT and HRVT SDNN (80.3±6.9 %HRmax; 151.1±11.4 bpm).
Differences were found in HRVT RMSSD and HRVT SDNN in both groups for % VO 2 (P<0.01). No diff erences were detected between the low and high cardiorespiratory fi tness groups regarding power output at VT (P=0.21), HRVT RMSSD (P=0.44), or HRVT SDNN (P=0. 26). No differences in %HRmax at VT, HRVT RMSSD , or HRVT SDNN were identifi ed between groups (P=0.36, 0.21, 0.18, respectively). Th e RPE at HRVT RMSSD and HRVT SDNN were signifi cantly (VT) and heart rate variability thresholds (HRVT RMSSD and HRVT SDNN ). Th e aforementioned analyzes were performed in SPSS version 20.0 software for Windows. Additionally, Cohen's eff ect size (ES) was used to interpret the magnitude of diff erence between the high and low VO 2 max groups considering the RMSSD and SDNN values at each percentage (from 10 to 100 % of maximum workload). Th e threshold values for Cohen's ES statistics were 0.20 -0.50 (small), 0.50 -0.80 (moderate), and >0.80 (large) 21 .
Considering the HRV analyzes assuming the percentage of maximum workload from the progressive test, the Cohen's ES of the pairwise comparison was large at 70% (ES: 0.94) and 80% (ES: 1.10), and moderate at 60% (ES: 0.52) and 100% (ES: 0.68) for RMSSD. Regarding the SDNN values, the ES was small from 10 to 100 % of maximum workload. Signifi cant diff erences (P<0.05) and a large Cohen's ES were observed when comparing the low and high cardiorespiratory fi tness for relative workloads with VT (ES: 1.50), HRVT RMSSD (ES: 1.22), and HRVT SDNN (ES: 1.32). Th e Cohen's ES between VT and HRVT RMSSD within groups was 0.69 (moderate) and 0.81 (large) for the low and high cardiorespiratory fi tness groups, respectively (see TABLE 2). When the data from both groups were pooled and power output was expressed relative to body mass (W•kg -1 ), the ES in the comparison between VT (2.09±0.58) and HRVT RMSSD   Comparisons between low (n=9) and high (n=10) cardiorespiratory fitness groups for VO 2 (ml•kg −1 •min −1 ), relative workload (W•kg -1 ) and percentage of Wmax (%) corresponding to ventilatory (VT) and heart rate variability thresholds (HRVT RMSSD and HRVT SDNN ). Pearson product-moment correlation coeffi cients between VO 2 (ml•kg −1 •min −1 ) values at ventilatory (VT) and heart rate variability thresholds (HRVT RMSSD and HRVT SDNN ).

Discussion
Th e main result of the study was that the diff erences between VT, HRVT RMSSD , and HRVT SDNN were greater in the high cardiorespiratory fi tness group than in the low cardiorespiratory fi tness group. Th us, this fi nding indicates that the use of the threshold as an aerobic capacity assessment tool (fi rst threshold) should be viewed with some caution, especially in individuals with high cardiorespiratory fi tness. Th e rationale between VT and RR oscillations measured through vagal HRV indices is evidenced as being synchronous with respiratory sinus arrhythmia, mediated by the vagus nerve 22,23 . However, the greatest diff erences were observed in individuals who presented higher VO 2 max values, although this was not noted by other authors 11,23 . We consider that there is a certain mismatch in physiological responses in HRV and respiratory/ventilatory dynamics in individuals with high aerobic fi tness during a progressive test.
Abnormal behavior concerning ventilatory responses in progressive tests have been reported with VT determination 24 , as well as in one subject in the present study from the low cardiorespiratory fi tness group. Th is fact demonstrates that VT should be analyzed carefully, requiring trained staff and gas exchange analysis equipment, which generates high cost. On the other hand, the HRVT using the visual inspection technique is an easy tool and less costly than VT, besides presenting high reproducibility 25 . Although the correspondence between metabolic transition thresholds and HRVT has been documented previously 11,26 , we found that it may present some disagreement in well-conditioned individuals who demonstrate higher rates of VO 2 max , possibly due to the more pronounced physiological mechanisms underlying vagal autonomic activity being more pronounced in well-trained individuals 27 . However, in the low cardiorespiratory fi tness group, the correspondence between HRVT measured using the RMSSD parameter and VT were similar, as reported by Sales et al. 26 comparing HRVT through vagal-mediated indices and VT 26 .
Considering the diff erences between HRV indices through progressive exercise testing, the stabilization point using sympathetic or overall parameters (i.e., SDNN) occurs later than vagal-mediated indices 5,25 . Similarly, the VO 2 at HRVT RMSSD was lower than VO 2 at HRVT SDNN in both groups. However, no diff erence was found between VT and HRVT SDNN in the low or high cardiorespiratory fitness groups when expressed in VO 2 . Th is diff erence is probably related to the physiological signifi cance of HRV indices which refl ect the sympathetic and parasympathetic drives of cardiac autonomic control during exercise 5,15 . Th e RMSSD is considered as a parasympathetic parameter related to the autonomic control perspective, which leads to a faster decay rate during progressive exercise testing, while SDNN is influenced by both sympathetic and parasympathetic pathways 1, 5 and generally achieves asymptotic values after VT 5 . Similar values were found to our results when the percentage VO 2 max at VT and HRVT (~64%) were analyzed according to parasympathetic parameters 22 . Th e HRV by time domain has proven to be a valid way to examine the vagal withdrawal in progressive tests 5,11,22,28 , and high reliability was observed when ascertaining HRVT using the vagal parameter through visual inspection 25 , as performed in the present study.
While the VO 2 (ml•kg −1 •min −1 ) at VT, HRVT RMSSD , and HRVT SDNN was higher in the high cardiorespiratory fi tness group, the percentage (%) VO 2 values were similar between the thresholds analyzed and the groups studied. Differences were found between groups for relative to body mass power output corresponding to VT and HRVT SDNN , but no diff erence was observed for HRVT RMSSD . In addition, although HRVT SDNN displayed better correlations in the comparison with VT than HRVT RMSSD , the use of the vagally mediated RMSSD throughout the progressive test appears to discriminate cardiorespiratory fi tness levels (FIGURE 1, panel A) considering percentages of workloads near to the VT or HRVT (60-80% of VO 2 max). Conversely, SDNN did not diff er between groups (FIGURE 1, panel B) at any workload. Th ese results confi rm the assumption that vagal indices are higher in individuals with good cardiorespiratory fitness levels during progressive exercise testing 15 . Th is could be due to a physiological concept, the higher 'parasympathetic reserve' present in the high cardiorespiratory fi tness group, as previously observed 5,6,15 . In this sense, the use of a vagal HRV parameter analyzed by RMSSD was sensitive to discriminate physical fi tness for initial workloads comparing the low and high cardiorespiratory fi tness groups. Th us, our results suggest, according to ES, the use of a vagal HRV parameter as a method able to discriminate by cardiac vagal modulation withdrawal throughout progressive exercise and its relationship with cardiorespiratory physical fi tness, as highlighted by other studies 14,15 .
A greater relationship between VT and HRVT RMSSD than the overall (SDNN) HRV parameter as assessed by HRVT SDNN was expected, as observed previously 5,11,22 . However, the present study demonstrated a signifi cant correlation only between the VO 2 at VT and HRVT SDNN (r=0.77) for the low cardiorespiratory fi tness group and a weaker non-signifi cant correlation (r=0.39) for the high cardiorespiratory fi tness group. Th e Pearson's correlation presented lower values between the VO 2 at VT and HRVT RMSSD in the low (r=0.52) and high (r=0.38) cardiorespiratory fitness groups. Th is fi nding confi rms the hypothesis that cardiorespiratory fi tness level is an important factor to consider in HRVT evaluation and its physiologic mechanisms linked to the VT or metabolic transition should be revised. Diff erent results were found to relate the VT with HRVT, using vagal parameters, (r=0.69-0.97) by Cottin 22 in welltrained subjects and (r=0.89) by Karapetian 11 in non-athletes. Th is disagreement could be due to the diff erent methods applied to determine HRVT or the profi le of the individuals studied.
A strict methodological criterion was applied to determine VT and HRVT in the present study to minimize potential errors by using the visual inspection technique. In this way, physical fitness appears to be the main factor regarding HRVT determination, considering the results of individuals with high cardiorespiratory fitness. It is not possible to state precisely what physiologically led to such a result, whether there is a mechanistic link between cardiac autonomic response and VT during the incremental test. A larger relationship was found between the VO 2 at VT with a vagal HRV parameter (SD1, an equivalent parameter of RMSSD) by Tulppo 5 and overall HRV parameter (SDNN) by Yamamoto 29 at the moment that the RR interval variability decreases progressively toward stability (~60% of Wmax). To summarize, although HRVT has been validated previously with both LT and VT1 1,22,26 , the present study demonstrated that only HRVT SDNN was correlated with VT in the low cardiorespiratory fitness group, therefore caution is recommended, especially with individuals with high cardiorespiratory fitness levels.
The increase in heart rate during exercise can be partly explained by the vagal withdrawal. The HR at HRVTRMSSD preceded the HR at VT in the low cardiorespiratory fitness group, as observed by Shibata 30 . In spite of this, in the present study the HR at HRVT RMSSD was not different from that observed in the VT and HRVT SDNN . This disagreement may indicate that the physiological mechanisms may be partly distinct and that methodological aspects of VT, LT, and HRVT determination could contribute to mismatched results. The main reason for this is the higher variability of HRV during exercise observed in the high cardiorespiratory fitness group (see FIGURE 1C and 1D as an example), influenced by larger vagal tone 15 . Additionally, one advantage of HRVT through visual inspection is related to the fact that VT evaluation is an expensive method and LT is an invasive procedure, whereas HRVT through visual inspection using an HR monitor with RR interval recording (every 2 min per stage), besides being a non-invasive method, is easily applicable and highly reproducible 25 .
The limitations in the present study must be pointed out. The determination of VT and HRVT by visual inspection is open to some subjectivity, despite presenting higher reliability than HRVT measured by mathematical models 25 . To reduce this subjectivity, methodological criteria are applied using two, or in some cases three evaluators, which can lower the practicality of this method. In addition, the split of only two groups, the small sample size, and recruitment of only males makes it difficult to generalize our results to different populations. In this way, additional research is warranted to confirm our hypothesis and findings with a larger population using clustering analysis.
In conclusion, the main finding of this study was that cardiorespiratory physical fitness may affect HRVT evaluation and that the HRVT SDNN was closer to VT in the low cardiorespiratory fitness group than the high cardiorespiratory fitness group. However, the use of vagal modulation assessed by HRV, for which we used RMSSD, is more appropriate than SDNN to observe possible differences regarding cardiorespiratory fitness during progressive exercise testing. Thus, the use of HRVT as a tool to assess metabolic transition should be used with caution.