Eur J Appl Physiol DOI 10.1007/s00421-014-2882-1
Original Article
Effect of acute normobaric hypoxia on the ventilatory threshold Carla A. Gallagher · Mark E. T. Willems · Mark P. Lewis · Stephen D. Myers
Received: 25 November 2013 / Accepted: 31 March 2014 © Springer-Verlag Berlin Heidelberg 2014
Abstract Purpose This study investigated the response of the ventilatory threshold (VT) to acute normobaric hypoxia and compared the agreement between software-based algorithms which use automatic detection to identify the VT. Results were used to examine whether the VT can be used as a physiological parameter to prescribe and monitor exercise intensity in hypoxic exercise training programs. Methods Fourteen untrained individuals (7 women, 7 men; age 22 ± 2 years, V˙ O2peak 46 ± 7 mL kg−1 min−1) completed five identical graded exercise tests (randomized order) on a cycle ergometer to measure VT at sea-level (SL) and in response to four normobaric hypoxic conditions (FIO2: 0.185, 0.165, 0.142, 0.125) equivalent to 1,000, 2,000, 3,000 and 4,000 m. Data were analyzed using a oneway analysis of variance (ANOVA) with repeated measures. Results The VT was similar across all conditions (SL = 1.98 ± 0.46, 1,000 m = 2.03 ± 0.61, 2,000 m = 2.27 ± 0.62, 3,000 m = 1.84 ± 0.50, 4,000 m = 2.29 ± 0.58 L min−1) for all algorithms used despite a reduction in arterial oxygen saturation at 3,000 (P ≤ 0.01) and 4,000 m (P ≤ 0.01) compared with SL values. Conclusion The VT appears to be a suitable physiological parameter for exercise prescription in normobaric hypoxia up to an altitude of 4,000 m. Communicated by Carsten Lundby. C. A. Gallagher · M. E. T. Willems · S. D. Myers (*) Department of Sport and Exercise Sciences, University of Chichester, College Lane, Chichester, West Sussex PO19 6PE, UK e-mail: [email protected] M. P. Lewis Loughborough University, Loughborough, Leicestershire, UK
Keywords Normobaric hypoxia · Exercise prescription · Cycle ergometry · Ventilatory threshold · Altitude Abbreviations ANOVA Analysis of variance FIO2 Ambient inspiratory oxygen fractions GXT Graded exercise test H+ Hydrogen ion HR Heart rate HRpeak Peak heart rate HRVT Heart rate at ventilatory threshold RCP Respiratory compensation point RH Relative humidity RPE Rating of perceived exertion SD Standard deviation SL Sea-level SPO2 Arterial oxygen saturation measured using infrared pulse oximetry tamb Ambient temperature V˙ CO2 Rate of carbon dioxide production V˙ Epeak Peak rate of ventilation V˙ O2 Oxygen consumption V˙ O2 Rate of oxygen consumption V˙ O2peak Peak rate of oxygen consumption V˙ O2VT Rate of oxygen consumption at ventilatory threshold ˙ peak Peak power output W VT Ventilatory threshold WVT Power at ventilatory threshold Introduction Exercise intensity is a key principle of training, and therefore an important tool for prescribing, assessing and
13
monitoring exercise training in all populations (Bompa and Haff 2009). Often, the peak rate of oxygen uptake (V˙ O2peak) is employed as a parameter to determine exercise intensity and predict performance. However, some authors have reported that nonlinear increases in physiological variables, such as blood lactate, heart rate and minute ventilation observed during progressive exercise tests represent “thresholds” that are better predictors of performance (Bosquet et al. 2002; Subudhi et al. 2006). A threshold indicates the upper limit of sensitivity, the point at which further increases in intensity have no typical effect (Kent 2007). Moreover, using V˙ O2peak to equate exercise intensity may lead to different physiological responses, for example, 60 % V˙ O2peak may be above the “threshold” for one individual and below for another which would result in differing training stimuli (Davis 1985). Numerous studies testify to the sensitivity of the ventilatory threshold (VT) to endurance training (Carter et al. 1999; Jones and Carter 2000; Weltman et al. 1992) and a rightward shift of the VT to a higher power output or running speed is characteristic of a successful training program (Jones and Carter 2000; Wells and Pate 1988). Therefore, the VT is a measurement that is sensitive to change and therefore allows for the effective monitoring of training adaptations. The VT can also be reliably detected using submaximal exercise tests (Wasserman and McIlroy 1964) and therefore is an ideal measure for populations in which exhaustive exercise may be contraindicated (e.g. sedentary, elderly, obese). Despite the advantages of VT assessments, there have been relatively few measurements in response to hypoxia. Although a hypoxia-induced fall in aerobic capacity is incurred at altitude, it has been demonstrated that the same cardiovascular benefits are received with lower exercise intensity (Cerretelli 1980; Friedmann et al. 2005). Consequently, exercising/training in hypoxia seems a viable method of optimizing training such that certain individuals (e.g. sedentary, elderly, obese) receive maximal metabolic and cardiovascular benefit whilst minimizing their risk of injury through a reduction in exercise intensity (Haufe et al. 2008). Moreover, combining exercise training with hypoxia may lead to greater losses of body mass than exercise training at SL (Quintero et al. 2010). Evidence suggests that hypoxia causes a further increase in lactate above values observed at sea level (SL) (Hughson et al. 1995; Mazzeo et al. 1991) and the accumulation of H+ associated with lactate ion accumulation in the blood is well correlated during exercise at SL (Subudhi et al. 2006). Thus, it would be expected that the VT would be reduced further during exercise in hypoxic environments compared with SL measurements. Previous research has shown that reduced inspired O2 concentrations (FIO2 0.12–0.14) during graded cycle tests can reduce the VT
13
Eur J Appl Physiol
by up to 33 % (Fukuoka et al. 2003; Hughson et al. 1995; Ozcelik and Kelestimur 2004). Subudhi et al. (2006) supported these findings by reporting reduced power output at the VT (41 %) at 4,300 m compared with SL, as well as reporting an overall reduction in the VT. Running speed, O2 uptake (V˙ O2) and heart rate at the VT have also been shown to decrease in response to acute hypoxia during treadmill exercise (Friedmann et al. 2004). In contrast, at a lower equivalent altitude of 1,500 m there was no reported difference in the VT compared with a SL control condition (Mateika and Duffin 1994). From the studies discussed above, we hypothesize that a reduction in VT may occur at a particular threshold altitude. If this is the case and a threshold altitude can be identified, then the VT could be used determine and prescribe exercise intensity within hypoxic training programs, thus further exploration in this area is warranted. Since there are little data available in the literature on the effects of hypoxia on the VT, the primary aim of this study was to investigate the effect of acute normobaric hypoxia of varying inspired O2 concentrations (FIO2 0.209–0.125; SL4,000 m equivalent altitude) on the VT determined from a graded cycle exercise test. Several techniques based on visual inspection of respiratory gas exchange graphical plots have been described to determine the VT (Santos and Giannella-Neto 2004). However, the results are often subjective (Bosquet et al. 2002; Yeh et al. 1983). Therefore, automatic computerized algorithms have been designed, which may provide more accurate results. To date, these algorithms are yet to be compared during the analysis of data collected in hypoxia. Thus, a secondary aim of this study was to examine the agreement between algorithms used to identify the VT using the computerized software program Winbreak 3.7® (Ekkekakis et al. 2008).
Methods Participants Fourteen healthy volunteers (7 women, 7 men; mean ± SD: age 22 ± 2 years; height 175.8 ± 10.3 cm; body mass 72.2 ± 14.3 kg; BMI 23.2 ± 2.8 kg/m2) participated in the study. All participants completed a medical history questionnaire, stating no contraindications to exhaustive exercise. All were SL residents and had not been above 1,500 m in the 6 months preceding the study. After receiving both written and oral information on the experimental protocol and procedures, participants gave their written informed consent. The research conformed to the guidelines laid down in the Declaration of Helsinki (2008) and was approved by the University of Chichester Research Ethics Committee.
Eur J Appl Physiol
Experimental procedure Each participant visited the laboratory on six occasions, including one familiarization session, with each visit separated by a minimum of 48 h. To ensure standardized testing, all participants were instructed not to partake in any vigorous physical training or consume alcohol in the 24 h prior to each session and to avoid the consumption of caffeinated beverages, and food for a 2-h period prior to all sessions (Noonan and Dean 2000). All exercise tests were performed in an environmental chamber (TISS series 201003-1, TIS Services, UK), where normobaric hypoxia was achieved via a molecular sieve. The ambient temperature (tamb) and relative humidity (RH) were controlled for in all sessions (tamb 21.3 ± 1.7 °C; RH 50 ± 2 %). In all tests completed, the ambient inspiratory O2 fractions (FIO2) were 0.209, 0.185, 0.165, 0.142 and 0.125, corresponding to SL, 1,000, 2,000, 3,000 and 4,000 m, respectively. To keep participants uninformed of each condition, when performing at SL (FIO2 0.209), the compressor connected to the environmental chamber was running at all times to create a ‘sham hypoxia’ environment (Netzer et al. 2008). To monitor participant’s perception of altitude exposure, following each visit they were asked to report which condition they thought they had been exposed, none of the participants were able to identify the correct condition on every occasion, and only 40 % of participants correctly identified the SL condition. All conditions were randomized to each participant using a 5 × 5 Latin Square (Keppel 1983) and administered in a single-blind manner. The VT was determined using a graded exercise test (GXT) conducted on an electromagnetically braked cycle ergometer (Lode, Excalibur Sport, Cranlea and Co, Bourneville, UK). The GXT began at 50 W, which was subsequently increased by 20 W for women and 25 W for men every minute thereafter (Amann et al. 2006). Participants selected a cadence between 70 and 90 revolutions per minute and were asked to maintain this throughout each GXT. During the GXT, heart rate (HR), V˙ E, V˙ O2 and FIO2 were recorded continuously using breath-by-breath analysis (Cosmed K4b2, Cosmed srl, Rome, Italy). Arterial O2 saturation [(SPO2) accuracy ± 2 %; Model 3800, Datex-Ohmedia Division, Instrumentarium Corp, Finland] was monitored continuously during the GXT and recorded every minute, with readings blinded to participants. Participants also gave their rating of perceived exertion [RPE (Borg 1982)] in the last 10 s of every stage (Shephard et al. 1992). Computerized determination of the VT The VT was identified using a computerized program (WinBreak 3.7, Epistemic Mindworks, USA) that incorporates a number of algorithms commonly used to determine
this parameter (Ekkekakis et al. 2008). Data preparation involved three working steps according to the methods of Ekkekakis et al. (2008). Firstly, non-physiological (e.g. negative) values were removed from each data set. Secondly, data were averaged every 20 s (Gaskill et al. 2001). Lastly, lower and upper boundaries for the VT calculations were set. The lower boundary was always set after the first minute (minute 1). The upper boundary was set either at the end of the test or at the respiratory compensation point (RCP), if one was found. The RCP was determined using a modified version of the method of Beaver et al. (1986) in which the RCP is identified as follows: “The V˙ E vs V˙ CO2 data are divided into two linear segments, the intersection of the two segments is the RCP if the change in the slope between them is greater than a preselected amount (15 % of initial slope)” (p. 2023). Beaver et al. (1986) noted that a RCP does not always occur. Therefore, the data sets were first examined for a significant departure from linearity. In the present study, this was the case for 2 out of 70 cases, in this instance the end of the GXT was used as the upper boundary for the VT calculations. If data showed a significant departure, the RCP was set at the point of the largest slope difference between the two segments, as using a fixed slope difference (e.g. 15 %) occasionally resulted in untenable results (e.g. points below 50 % V˙ O2peak). After application of preliminary data processing steps, the VT was estimated using the following five algorithms as used by Ekkekakis et al. (2008): Algorithm 1. The “breakpoint” algorithm (Jones and Molitoris 1984) Algorithm 2. The “brute force” algorithm (Orr et al. 1982) Algorithm 3. The “V-slope” method (Beaver et al. 1986) Algorithm 4. The “Dmax” method (Cheng et al. 1992) Algorithm 5. The “simplified V-slope” method (Dickstein et al. 1990; Sue et al. 1988). Statistical analysis Statistical analyses were computed by the statistical software package PASW 18.0 (SPSS Inc., Chicago, USA). All data were normally distributed in accordance with the Kolmogorov–Smirnov test (P ≥ 0.05). To compare changes between tests, data were analyzed using a one-way analysis of variance (ANOVA) with repeated measures and planned comparisons comparing hypoxic measurements against SL were carried out using Bonferroni corrected t-tests. Effect sizes for ANOVAs were calculated using the omega squared (Ω2) method, and can be interpreted as small (0.15) (Cohen
13
1992). Effect size for t test comparisons were calculated by converting a t value into an r value (Field 2005; Rosenthal et al. 2000), and can be interpreted as small (0.10), medium (0.30) and large (0.50) (Cohen 1988). Sample size was calculated a priori using G*Power software (release 3.1.2: Kiel, Germany) which resulted in a total of nine participants requiring for the study for a power of 0.80. Post-hoc power analyses resulted in a power of 0.99 for a sample size of 14, with a P value of 0.025 and effect size of 0.63. All values are presented as mean ± 1SD and for statistical analyses an alpha of P ≤ 0.05 was considered significant unless otherwise stated.
Results Indeterminate cases During analysis of the VT, some data files did not fit the criteria for the algorithms used; these files were therefore termed indeterminate cases. In the discussion, the algorithms referred to will be those that produced the least indeterminate cases, which were algorithms 1, 2 and 4. Algorithms 3 and 5 are not presented due to a large number of missing data points. All VT calculation methods, because of the restrictions they place on the viability of the solutions, resulted in some indeterminate cases. Specifically, algorithms 1, 2 and 4 led to one, Algorithm 3 led to four, and Algorithm 5 to six indeterminate cases in the identification of the VT at SL, respectively. Further indeterminate cases (number of cases) were also visible at 1,000 m (9), 2,000 m (12), 3,000 m (11) and 4,000 m (13). Algorithm 1 elicited the least indeterminate cases (3 cases) whilst Algorithm 5 elicited the most (35 cases). Indeterminate cases were not altitude dependent. Any algorithm that demonstrated indeterminate cases equal to, or more than 10 % of the data set were eliminated from statistical analyses (Algorithms 3 and 5). In the case where less than 10 % of the values were missing, values were then estimated using the column means (Table 1).
Eur J Appl Physiol Table 1 Estimates of the ventilatory threshold, expressed as oxygen cost (V˙ O2), percentage peak rate of V˙ O2 (% V˙ O2peak), power and heart rate Algorithm SL
1
2
4
V˙ O2VT (L min−1) 1.98 ± 0.46 2.15 ± 0.63 1.95 ± 0.47
HRVT (b min−1) WVT % V˙ O2peak
128 ± 11 127 ± 36
130 ± 20 127 ± 49
128 ± 15 119 ± 33
60 ± 8
65 ± 10
59 ± 9
1,000 m V˙ O2VT (L min ) 2.03 ± 0.61 2.16 ± 0.70 2.05 ± 0.61 HRVT (b min−1) 127 ± 16 127 ± 16 127 ± 15 132 ± 40 132 ± 41 130 ± 34 WVT 61 ± 12 65 ± 13 61 ± 11 % V˙ O2peak −1
2,000 m V˙ O2VT (L min−1) 2.27 ± 0.62 2.47 ± 0.74 2.29 ± 0.55 HRVT (b min−1) 139 ± 20 137 ± 24 136 ± 18 139 ± 50 140 ± 65 148 ± 41 WVT 69 ± 14 74 ± 18 69 ± 13 % V˙ O2peak 3,000 m V˙ O2VT (L min−1) 1.84 ± 0.50 1.87 ± 0.61 1.80 ± 0.50 135 ± 21 130 ± 17 130 ± 20 HRVT (b min−1) 133 ± 40 129 ± 32 115 ± 23 WVT 57 ± 16 56 ± 14 55 ± 15 % V˙ O2peak 4,000 m V˙ O2VT (L min−1) 2.29 ± 0.58 2.36 ± 0.63 2.25 ± 0.54 141 ± 16* 139 ± 15 139 ± 12 HRVT (b min−1) 150 ± 35 142 ± 33 129 ± 27 WVT 67 ± 14 69 ± 13 66 ± 14 % V˙ O2peak Algorithms: 1. Jones and Molitoris (1984), 2. Orr et al. (1982), 4. Cheng et al. (1992) * Significantly higher than SL, P ≤ 0.01
was higher than SL at 4,000 m (P = 0.012, r = 0.63). Neither the VT expressed as % V˙ O2peak, nor WVT differed in hypoxia compared to SL (Table 1). Algorithm 2:Orr et al. (1982) In hypoxia, there was an overall significant change in V˙ O2VT (Ω2 = 0.63) and VT as % V˙ O2peak (Ω2 = 0.76). Neither the HRVT nor WVT differed in hypoxia compared to SL (Table 1).
Ventilatory threshold Algorithm 4:Cheng et al. (1992) The VT determined by each algorithm is presented in Table 1, the VT is expressed as V˙ O2 (V˙ O2VT), heart rate (HRVT), and power (WVT). The VT is also expressed relative to V˙ O2peak (% V˙ O2peak) at the corresponding altitude.
In hypoxia, there was an overall significant change in V˙ O2VT (P = 0.008, Ω2 = 0.68), VT as % V˙ O2peak (P = 0.014, Ω2 = 0.72) and WVT (P = 0.028, Ω2 = 0.67). The HRVT did not differ in hypoxia compared to SL (Table 1).
Algorithm 1: Jones and Molitoris (1984) Agreement between algorithm determinations of the VT In hypoxia, there was an overall significant change in V˙ O2VT (P = 0.025, Ω2 = 0.62). A significant change was also observed for HRVT (P = 0.031, Ω2 = 0.66), which
13
During tests in all conditions (SL, 1,000, 2,000, 3,000 and 4,000 m), V˙ O2VT values determined by each algorithm
Eur J Appl Physiol
were not significantly different from each other (Fig. 1). No significant differences between algorithms were identified for HRVT (Fig. 2) and WVT (Fig. 3) in any of the conditions. Cardiorespiratory responses Peak cardiorespiratory responses are shown in Table 2. In ˙ peak (P ≤ 0.001, hypoxia, there was an overall decline in W Ω2 = 0.40) and V˙ O2peak (P = 0.014, Ω2 = 0.36) which was significantly lower than SL at 4,000 m for both measures ˙ peak: P = 0.001, r = 0.77). (V˙ O2peak: P = 0.006, r = 0.67; W In contrast, neither HRpeak nor V˙ Epeak differed in hypoxia compared to SL. The SPO2 was reduced overall (P ≤ 0.001, Ω2 = 0.98) and was significantly lower than SL at 3,000 (P ≤ 0.001, r = 0.80) and 4,000 m (P ≤ 0.001, r = 0.91).
Fig. 1 The oxygen cost (V˙ O2) at ventilatory threshold identified by each algorithm in each hypoxic condition (1) Jones and Molitoris (1984), (2) Orr et al. (1982), (4) Cheng et al. (1992)
Discussion The primary aim of the present study was to examine the effect of varying inspired O2 concentrations (hypoxia) on the VT response. The main finding was that there was no significant change in V˙ O2VT with hypoxia. However, a significant change in VT relative to V˙ O2peak using algorithms 2 and 4, and a significant reduction in V˙ O2peak and Wpeak were identified. The secondary aim was to compare the V˙ O2VT, HRVT and WVT identified by the different algorithms (1, 2 and 4), and these were found to be in agreement. Despite the advantages of VT assessments, in particular for individuals for whom maximal exercise is contraindicated, there have been relatively few measurements of VT in response to acute hypoxia. This despite exercise in hypoxia offering cardiovascular benefits (Cerretelli 1980; Friedmann et al. 2005) and possibly enhancing weight loss in the obese (Quintero et al. 2010). Using the algorithms which produced the least indeterminate cases [1, 2 and 4 (Cheng et al. 1992; Jones and Molitoris 1984; Orr et al. 1982)] a significant overall change in V˙ O2VT was identified in hypoxia compared with SL, however post hoc analyses failed to distinguish the direction of change which differs from previous research which reported a reduction (Fukuoka et al. 2003; Hughson et al. 1995; Ozcelik and Kelestimur 2004). It can therefore be concluded that the differences observed were due to individual variation and there was no change in the VT with hypoxia which can be supported by research conducted at an equivalent of 1,500 m (Mateika and Duffin 1994). A reduction in VT with hypoxic exposure is often attributed to increased sympathetic activity and glycolytic flux at a given work rate, resulting in an accelerated efflux of lactate and H+, which subsequently increases the ventilatory drive (Subudhi et al. 2006). Consequently, it can be concluded that such an acute bout of hypoxic exposure may not increase sympathetic
Fig. 2 The heart rate at ventilatory threshold identified by each algorithm in each hypoxic condition (1) Jones and Molitoris (1984), (2) Orr et al. (1982), (4) Cheng et al. (1992)
Fig. 3 The power at ventilatory threshold identified by each algorithm in each hypoxic condition (1) Jones and Molitoris (1984), (2) Orr et al. (1982), (4) Cheng et al. (1992)
drive and thus reduce the VT via an increase in ventilatory drive. Moreover, other research suggests that the changes in lactate concentration observed during exercise may not be directly related to FIO2 (Mateika and Duffin 1994) and could therefore explain why a reduction in VT is observed
13
Eur J Appl Physiol
Table 2 Peak cardiorespiratory responses
SL ˙ peak (W) W V˙ O2peak (L min−1)
1,000 m
2,000 m
3,000 m
4,000 m
243 ± 58
239 ± 57
236 ± 55
229 ± 57
216 ± 43*
3.39 ± 0.99
3.22 ± 0.98
3.27 ± 0.91
3.09 ± 0.96
2.91 ± 0.72*
46.34 ± 7.00 44.89 ± 9.78 44.93 ± 8.34 42.57 ± 10.24 40.45 ± 7.41* V˙ O2peak (mL kg−1 min−1) V˙ Epeak (L min−1) 133.90 ± 41.17 128.10 ± 34.55 132.21 ± 32.52 132.63 ± 33.08 133.44 ± 27.78 * Significantly lower than sealevel (SL), P ≤ 0.05
HRpeak (b min−1) SPO2 (%)
180 ± 11
178 ± 11
178 ± 9
95 ± 5
95 ± 2
92 ± 2
below a FIO2 of 0.14 (3,000 m) but not above. Another possible explanation is that lactate concentration, due to an increase in glycolytic flux, is only elevated in a more severe hypoxic condition (i.e. at or above 4,000 m). According to others (Overgaard et al. 2012) arterial lactate concentrations in severe hypoxia (FIO2 0.10) during exercise is 10 times greater of that observed in normoxia with the same, however at moderate altitudes (FIO2 0.15) similar capillary lactate concentrations between hypoxic and normoxic conditions have been reported during cycling time trial performance workload (Amann et al. 2006). Therefore, from this evidence a reduction in VT would not be expected below 3,000 m. As such, it is also recognized that a lack of lactate and pH measurements is a limitation of the present study and future studies should if feasible look to include these measures. Previous research has demonstrated that WVT is a more accurate measure of physical performance (e.g. time trial performance) than V˙ O2peak (Amann et al. 2006; Lucia et al. 2004; Subudhi et al. 2006). Research has previously demonstrated that WVT is reduced with hypoxic exposure by up to 41 % (Subudhi et al. 2006). In the present study algorithms 1 and 2 did not confirm such findings suggesting that acute hypoxic exposure does not reduce WVT. However, algorithm 4 produced an overall change in WVT but similar to V˙ O2VT the direction of change could not be identified. In summary, it can be concluded that acute hypoxia had no effect on the VT determined via a GXT. Those studies, which reported a significant reduction in V˙ O2VT, were performed at or above 4,000 m (Fukuoka et al. 2003; Ozcelik and Kelestimur 2004; Subudhi et al. 2006), which may explain the disparity between results. However, it is difficult to clarify whether arterial O2 saturations were similar between studies as previous research did not report these data. Moreover, differences in acclimatization status/ exposure duration (Subudhi et al. vs. current study; 24 h vs. ˙ peak 354 vs. 243 W) and the
Original Article
Effect of acute normobaric hypoxia on the ventilatory threshold Carla A. Gallagher · Mark E. T. Willems · Mark P. Lewis · Stephen D. Myers
Received: 25 November 2013 / Accepted: 31 March 2014 © Springer-Verlag Berlin Heidelberg 2014
Abstract Purpose This study investigated the response of the ventilatory threshold (VT) to acute normobaric hypoxia and compared the agreement between software-based algorithms which use automatic detection to identify the VT. Results were used to examine whether the VT can be used as a physiological parameter to prescribe and monitor exercise intensity in hypoxic exercise training programs. Methods Fourteen untrained individuals (7 women, 7 men; age 22 ± 2 years, V˙ O2peak 46 ± 7 mL kg−1 min−1) completed five identical graded exercise tests (randomized order) on a cycle ergometer to measure VT at sea-level (SL) and in response to four normobaric hypoxic conditions (FIO2: 0.185, 0.165, 0.142, 0.125) equivalent to 1,000, 2,000, 3,000 and 4,000 m. Data were analyzed using a oneway analysis of variance (ANOVA) with repeated measures. Results The VT was similar across all conditions (SL = 1.98 ± 0.46, 1,000 m = 2.03 ± 0.61, 2,000 m = 2.27 ± 0.62, 3,000 m = 1.84 ± 0.50, 4,000 m = 2.29 ± 0.58 L min−1) for all algorithms used despite a reduction in arterial oxygen saturation at 3,000 (P ≤ 0.01) and 4,000 m (P ≤ 0.01) compared with SL values. Conclusion The VT appears to be a suitable physiological parameter for exercise prescription in normobaric hypoxia up to an altitude of 4,000 m. Communicated by Carsten Lundby. C. A. Gallagher · M. E. T. Willems · S. D. Myers (*) Department of Sport and Exercise Sciences, University of Chichester, College Lane, Chichester, West Sussex PO19 6PE, UK e-mail: [email protected] M. P. Lewis Loughborough University, Loughborough, Leicestershire, UK
Keywords Normobaric hypoxia · Exercise prescription · Cycle ergometry · Ventilatory threshold · Altitude Abbreviations ANOVA Analysis of variance FIO2 Ambient inspiratory oxygen fractions GXT Graded exercise test H+ Hydrogen ion HR Heart rate HRpeak Peak heart rate HRVT Heart rate at ventilatory threshold RCP Respiratory compensation point RH Relative humidity RPE Rating of perceived exertion SD Standard deviation SL Sea-level SPO2 Arterial oxygen saturation measured using infrared pulse oximetry tamb Ambient temperature V˙ CO2 Rate of carbon dioxide production V˙ Epeak Peak rate of ventilation V˙ O2 Oxygen consumption V˙ O2 Rate of oxygen consumption V˙ O2peak Peak rate of oxygen consumption V˙ O2VT Rate of oxygen consumption at ventilatory threshold ˙ peak Peak power output W VT Ventilatory threshold WVT Power at ventilatory threshold Introduction Exercise intensity is a key principle of training, and therefore an important tool for prescribing, assessing and
13
monitoring exercise training in all populations (Bompa and Haff 2009). Often, the peak rate of oxygen uptake (V˙ O2peak) is employed as a parameter to determine exercise intensity and predict performance. However, some authors have reported that nonlinear increases in physiological variables, such as blood lactate, heart rate and minute ventilation observed during progressive exercise tests represent “thresholds” that are better predictors of performance (Bosquet et al. 2002; Subudhi et al. 2006). A threshold indicates the upper limit of sensitivity, the point at which further increases in intensity have no typical effect (Kent 2007). Moreover, using V˙ O2peak to equate exercise intensity may lead to different physiological responses, for example, 60 % V˙ O2peak may be above the “threshold” for one individual and below for another which would result in differing training stimuli (Davis 1985). Numerous studies testify to the sensitivity of the ventilatory threshold (VT) to endurance training (Carter et al. 1999; Jones and Carter 2000; Weltman et al. 1992) and a rightward shift of the VT to a higher power output or running speed is characteristic of a successful training program (Jones and Carter 2000; Wells and Pate 1988). Therefore, the VT is a measurement that is sensitive to change and therefore allows for the effective monitoring of training adaptations. The VT can also be reliably detected using submaximal exercise tests (Wasserman and McIlroy 1964) and therefore is an ideal measure for populations in which exhaustive exercise may be contraindicated (e.g. sedentary, elderly, obese). Despite the advantages of VT assessments, there have been relatively few measurements in response to hypoxia. Although a hypoxia-induced fall in aerobic capacity is incurred at altitude, it has been demonstrated that the same cardiovascular benefits are received with lower exercise intensity (Cerretelli 1980; Friedmann et al. 2005). Consequently, exercising/training in hypoxia seems a viable method of optimizing training such that certain individuals (e.g. sedentary, elderly, obese) receive maximal metabolic and cardiovascular benefit whilst minimizing their risk of injury through a reduction in exercise intensity (Haufe et al. 2008). Moreover, combining exercise training with hypoxia may lead to greater losses of body mass than exercise training at SL (Quintero et al. 2010). Evidence suggests that hypoxia causes a further increase in lactate above values observed at sea level (SL) (Hughson et al. 1995; Mazzeo et al. 1991) and the accumulation of H+ associated with lactate ion accumulation in the blood is well correlated during exercise at SL (Subudhi et al. 2006). Thus, it would be expected that the VT would be reduced further during exercise in hypoxic environments compared with SL measurements. Previous research has shown that reduced inspired O2 concentrations (FIO2 0.12–0.14) during graded cycle tests can reduce the VT
13
Eur J Appl Physiol
by up to 33 % (Fukuoka et al. 2003; Hughson et al. 1995; Ozcelik and Kelestimur 2004). Subudhi et al. (2006) supported these findings by reporting reduced power output at the VT (41 %) at 4,300 m compared with SL, as well as reporting an overall reduction in the VT. Running speed, O2 uptake (V˙ O2) and heart rate at the VT have also been shown to decrease in response to acute hypoxia during treadmill exercise (Friedmann et al. 2004). In contrast, at a lower equivalent altitude of 1,500 m there was no reported difference in the VT compared with a SL control condition (Mateika and Duffin 1994). From the studies discussed above, we hypothesize that a reduction in VT may occur at a particular threshold altitude. If this is the case and a threshold altitude can be identified, then the VT could be used determine and prescribe exercise intensity within hypoxic training programs, thus further exploration in this area is warranted. Since there are little data available in the literature on the effects of hypoxia on the VT, the primary aim of this study was to investigate the effect of acute normobaric hypoxia of varying inspired O2 concentrations (FIO2 0.209–0.125; SL4,000 m equivalent altitude) on the VT determined from a graded cycle exercise test. Several techniques based on visual inspection of respiratory gas exchange graphical plots have been described to determine the VT (Santos and Giannella-Neto 2004). However, the results are often subjective (Bosquet et al. 2002; Yeh et al. 1983). Therefore, automatic computerized algorithms have been designed, which may provide more accurate results. To date, these algorithms are yet to be compared during the analysis of data collected in hypoxia. Thus, a secondary aim of this study was to examine the agreement between algorithms used to identify the VT using the computerized software program Winbreak 3.7® (Ekkekakis et al. 2008).
Methods Participants Fourteen healthy volunteers (7 women, 7 men; mean ± SD: age 22 ± 2 years; height 175.8 ± 10.3 cm; body mass 72.2 ± 14.3 kg; BMI 23.2 ± 2.8 kg/m2) participated in the study. All participants completed a medical history questionnaire, stating no contraindications to exhaustive exercise. All were SL residents and had not been above 1,500 m in the 6 months preceding the study. After receiving both written and oral information on the experimental protocol and procedures, participants gave their written informed consent. The research conformed to the guidelines laid down in the Declaration of Helsinki (2008) and was approved by the University of Chichester Research Ethics Committee.
Eur J Appl Physiol
Experimental procedure Each participant visited the laboratory on six occasions, including one familiarization session, with each visit separated by a minimum of 48 h. To ensure standardized testing, all participants were instructed not to partake in any vigorous physical training or consume alcohol in the 24 h prior to each session and to avoid the consumption of caffeinated beverages, and food for a 2-h period prior to all sessions (Noonan and Dean 2000). All exercise tests were performed in an environmental chamber (TISS series 201003-1, TIS Services, UK), where normobaric hypoxia was achieved via a molecular sieve. The ambient temperature (tamb) and relative humidity (RH) were controlled for in all sessions (tamb 21.3 ± 1.7 °C; RH 50 ± 2 %). In all tests completed, the ambient inspiratory O2 fractions (FIO2) were 0.209, 0.185, 0.165, 0.142 and 0.125, corresponding to SL, 1,000, 2,000, 3,000 and 4,000 m, respectively. To keep participants uninformed of each condition, when performing at SL (FIO2 0.209), the compressor connected to the environmental chamber was running at all times to create a ‘sham hypoxia’ environment (Netzer et al. 2008). To monitor participant’s perception of altitude exposure, following each visit they were asked to report which condition they thought they had been exposed, none of the participants were able to identify the correct condition on every occasion, and only 40 % of participants correctly identified the SL condition. All conditions were randomized to each participant using a 5 × 5 Latin Square (Keppel 1983) and administered in a single-blind manner. The VT was determined using a graded exercise test (GXT) conducted on an electromagnetically braked cycle ergometer (Lode, Excalibur Sport, Cranlea and Co, Bourneville, UK). The GXT began at 50 W, which was subsequently increased by 20 W for women and 25 W for men every minute thereafter (Amann et al. 2006). Participants selected a cadence between 70 and 90 revolutions per minute and were asked to maintain this throughout each GXT. During the GXT, heart rate (HR), V˙ E, V˙ O2 and FIO2 were recorded continuously using breath-by-breath analysis (Cosmed K4b2, Cosmed srl, Rome, Italy). Arterial O2 saturation [(SPO2) accuracy ± 2 %; Model 3800, Datex-Ohmedia Division, Instrumentarium Corp, Finland] was monitored continuously during the GXT and recorded every minute, with readings blinded to participants. Participants also gave their rating of perceived exertion [RPE (Borg 1982)] in the last 10 s of every stage (Shephard et al. 1992). Computerized determination of the VT The VT was identified using a computerized program (WinBreak 3.7, Epistemic Mindworks, USA) that incorporates a number of algorithms commonly used to determine
this parameter (Ekkekakis et al. 2008). Data preparation involved three working steps according to the methods of Ekkekakis et al. (2008). Firstly, non-physiological (e.g. negative) values were removed from each data set. Secondly, data were averaged every 20 s (Gaskill et al. 2001). Lastly, lower and upper boundaries for the VT calculations were set. The lower boundary was always set after the first minute (minute 1). The upper boundary was set either at the end of the test or at the respiratory compensation point (RCP), if one was found. The RCP was determined using a modified version of the method of Beaver et al. (1986) in which the RCP is identified as follows: “The V˙ E vs V˙ CO2 data are divided into two linear segments, the intersection of the two segments is the RCP if the change in the slope between them is greater than a preselected amount (15 % of initial slope)” (p. 2023). Beaver et al. (1986) noted that a RCP does not always occur. Therefore, the data sets were first examined for a significant departure from linearity. In the present study, this was the case for 2 out of 70 cases, in this instance the end of the GXT was used as the upper boundary for the VT calculations. If data showed a significant departure, the RCP was set at the point of the largest slope difference between the two segments, as using a fixed slope difference (e.g. 15 %) occasionally resulted in untenable results (e.g. points below 50 % V˙ O2peak). After application of preliminary data processing steps, the VT was estimated using the following five algorithms as used by Ekkekakis et al. (2008): Algorithm 1. The “breakpoint” algorithm (Jones and Molitoris 1984) Algorithm 2. The “brute force” algorithm (Orr et al. 1982) Algorithm 3. The “V-slope” method (Beaver et al. 1986) Algorithm 4. The “Dmax” method (Cheng et al. 1992) Algorithm 5. The “simplified V-slope” method (Dickstein et al. 1990; Sue et al. 1988). Statistical analysis Statistical analyses were computed by the statistical software package PASW 18.0 (SPSS Inc., Chicago, USA). All data were normally distributed in accordance with the Kolmogorov–Smirnov test (P ≥ 0.05). To compare changes between tests, data were analyzed using a one-way analysis of variance (ANOVA) with repeated measures and planned comparisons comparing hypoxic measurements against SL were carried out using Bonferroni corrected t-tests. Effect sizes for ANOVAs were calculated using the omega squared (Ω2) method, and can be interpreted as small (0.15) (Cohen
13
1992). Effect size for t test comparisons were calculated by converting a t value into an r value (Field 2005; Rosenthal et al. 2000), and can be interpreted as small (0.10), medium (0.30) and large (0.50) (Cohen 1988). Sample size was calculated a priori using G*Power software (release 3.1.2: Kiel, Germany) which resulted in a total of nine participants requiring for the study for a power of 0.80. Post-hoc power analyses resulted in a power of 0.99 for a sample size of 14, with a P value of 0.025 and effect size of 0.63. All values are presented as mean ± 1SD and for statistical analyses an alpha of P ≤ 0.05 was considered significant unless otherwise stated.
Results Indeterminate cases During analysis of the VT, some data files did not fit the criteria for the algorithms used; these files were therefore termed indeterminate cases. In the discussion, the algorithms referred to will be those that produced the least indeterminate cases, which were algorithms 1, 2 and 4. Algorithms 3 and 5 are not presented due to a large number of missing data points. All VT calculation methods, because of the restrictions they place on the viability of the solutions, resulted in some indeterminate cases. Specifically, algorithms 1, 2 and 4 led to one, Algorithm 3 led to four, and Algorithm 5 to six indeterminate cases in the identification of the VT at SL, respectively. Further indeterminate cases (number of cases) were also visible at 1,000 m (9), 2,000 m (12), 3,000 m (11) and 4,000 m (13). Algorithm 1 elicited the least indeterminate cases (3 cases) whilst Algorithm 5 elicited the most (35 cases). Indeterminate cases were not altitude dependent. Any algorithm that demonstrated indeterminate cases equal to, or more than 10 % of the data set were eliminated from statistical analyses (Algorithms 3 and 5). In the case where less than 10 % of the values were missing, values were then estimated using the column means (Table 1).
Eur J Appl Physiol Table 1 Estimates of the ventilatory threshold, expressed as oxygen cost (V˙ O2), percentage peak rate of V˙ O2 (% V˙ O2peak), power and heart rate Algorithm SL
1
2
4
V˙ O2VT (L min−1) 1.98 ± 0.46 2.15 ± 0.63 1.95 ± 0.47
HRVT (b min−1) WVT % V˙ O2peak
128 ± 11 127 ± 36
130 ± 20 127 ± 49
128 ± 15 119 ± 33
60 ± 8
65 ± 10
59 ± 9
1,000 m V˙ O2VT (L min ) 2.03 ± 0.61 2.16 ± 0.70 2.05 ± 0.61 HRVT (b min−1) 127 ± 16 127 ± 16 127 ± 15 132 ± 40 132 ± 41 130 ± 34 WVT 61 ± 12 65 ± 13 61 ± 11 % V˙ O2peak −1
2,000 m V˙ O2VT (L min−1) 2.27 ± 0.62 2.47 ± 0.74 2.29 ± 0.55 HRVT (b min−1) 139 ± 20 137 ± 24 136 ± 18 139 ± 50 140 ± 65 148 ± 41 WVT 69 ± 14 74 ± 18 69 ± 13 % V˙ O2peak 3,000 m V˙ O2VT (L min−1) 1.84 ± 0.50 1.87 ± 0.61 1.80 ± 0.50 135 ± 21 130 ± 17 130 ± 20 HRVT (b min−1) 133 ± 40 129 ± 32 115 ± 23 WVT 57 ± 16 56 ± 14 55 ± 15 % V˙ O2peak 4,000 m V˙ O2VT (L min−1) 2.29 ± 0.58 2.36 ± 0.63 2.25 ± 0.54 141 ± 16* 139 ± 15 139 ± 12 HRVT (b min−1) 150 ± 35 142 ± 33 129 ± 27 WVT 67 ± 14 69 ± 13 66 ± 14 % V˙ O2peak Algorithms: 1. Jones and Molitoris (1984), 2. Orr et al. (1982), 4. Cheng et al. (1992) * Significantly higher than SL, P ≤ 0.01
was higher than SL at 4,000 m (P = 0.012, r = 0.63). Neither the VT expressed as % V˙ O2peak, nor WVT differed in hypoxia compared to SL (Table 1). Algorithm 2:Orr et al. (1982) In hypoxia, there was an overall significant change in V˙ O2VT (Ω2 = 0.63) and VT as % V˙ O2peak (Ω2 = 0.76). Neither the HRVT nor WVT differed in hypoxia compared to SL (Table 1).
Ventilatory threshold Algorithm 4:Cheng et al. (1992) The VT determined by each algorithm is presented in Table 1, the VT is expressed as V˙ O2 (V˙ O2VT), heart rate (HRVT), and power (WVT). The VT is also expressed relative to V˙ O2peak (% V˙ O2peak) at the corresponding altitude.
In hypoxia, there was an overall significant change in V˙ O2VT (P = 0.008, Ω2 = 0.68), VT as % V˙ O2peak (P = 0.014, Ω2 = 0.72) and WVT (P = 0.028, Ω2 = 0.67). The HRVT did not differ in hypoxia compared to SL (Table 1).
Algorithm 1: Jones and Molitoris (1984) Agreement between algorithm determinations of the VT In hypoxia, there was an overall significant change in V˙ O2VT (P = 0.025, Ω2 = 0.62). A significant change was also observed for HRVT (P = 0.031, Ω2 = 0.66), which
13
During tests in all conditions (SL, 1,000, 2,000, 3,000 and 4,000 m), V˙ O2VT values determined by each algorithm
Eur J Appl Physiol
were not significantly different from each other (Fig. 1). No significant differences between algorithms were identified for HRVT (Fig. 2) and WVT (Fig. 3) in any of the conditions. Cardiorespiratory responses Peak cardiorespiratory responses are shown in Table 2. In ˙ peak (P ≤ 0.001, hypoxia, there was an overall decline in W Ω2 = 0.40) and V˙ O2peak (P = 0.014, Ω2 = 0.36) which was significantly lower than SL at 4,000 m for both measures ˙ peak: P = 0.001, r = 0.77). (V˙ O2peak: P = 0.006, r = 0.67; W In contrast, neither HRpeak nor V˙ Epeak differed in hypoxia compared to SL. The SPO2 was reduced overall (P ≤ 0.001, Ω2 = 0.98) and was significantly lower than SL at 3,000 (P ≤ 0.001, r = 0.80) and 4,000 m (P ≤ 0.001, r = 0.91).
Fig. 1 The oxygen cost (V˙ O2) at ventilatory threshold identified by each algorithm in each hypoxic condition (1) Jones and Molitoris (1984), (2) Orr et al. (1982), (4) Cheng et al. (1992)
Discussion The primary aim of the present study was to examine the effect of varying inspired O2 concentrations (hypoxia) on the VT response. The main finding was that there was no significant change in V˙ O2VT with hypoxia. However, a significant change in VT relative to V˙ O2peak using algorithms 2 and 4, and a significant reduction in V˙ O2peak and Wpeak were identified. The secondary aim was to compare the V˙ O2VT, HRVT and WVT identified by the different algorithms (1, 2 and 4), and these were found to be in agreement. Despite the advantages of VT assessments, in particular for individuals for whom maximal exercise is contraindicated, there have been relatively few measurements of VT in response to acute hypoxia. This despite exercise in hypoxia offering cardiovascular benefits (Cerretelli 1980; Friedmann et al. 2005) and possibly enhancing weight loss in the obese (Quintero et al. 2010). Using the algorithms which produced the least indeterminate cases [1, 2 and 4 (Cheng et al. 1992; Jones and Molitoris 1984; Orr et al. 1982)] a significant overall change in V˙ O2VT was identified in hypoxia compared with SL, however post hoc analyses failed to distinguish the direction of change which differs from previous research which reported a reduction (Fukuoka et al. 2003; Hughson et al. 1995; Ozcelik and Kelestimur 2004). It can therefore be concluded that the differences observed were due to individual variation and there was no change in the VT with hypoxia which can be supported by research conducted at an equivalent of 1,500 m (Mateika and Duffin 1994). A reduction in VT with hypoxic exposure is often attributed to increased sympathetic activity and glycolytic flux at a given work rate, resulting in an accelerated efflux of lactate and H+, which subsequently increases the ventilatory drive (Subudhi et al. 2006). Consequently, it can be concluded that such an acute bout of hypoxic exposure may not increase sympathetic
Fig. 2 The heart rate at ventilatory threshold identified by each algorithm in each hypoxic condition (1) Jones and Molitoris (1984), (2) Orr et al. (1982), (4) Cheng et al. (1992)
Fig. 3 The power at ventilatory threshold identified by each algorithm in each hypoxic condition (1) Jones and Molitoris (1984), (2) Orr et al. (1982), (4) Cheng et al. (1992)
drive and thus reduce the VT via an increase in ventilatory drive. Moreover, other research suggests that the changes in lactate concentration observed during exercise may not be directly related to FIO2 (Mateika and Duffin 1994) and could therefore explain why a reduction in VT is observed
13
Eur J Appl Physiol
Table 2 Peak cardiorespiratory responses
SL ˙ peak (W) W V˙ O2peak (L min−1)
1,000 m
2,000 m
3,000 m
4,000 m
243 ± 58
239 ± 57
236 ± 55
229 ± 57
216 ± 43*
3.39 ± 0.99
3.22 ± 0.98
3.27 ± 0.91
3.09 ± 0.96
2.91 ± 0.72*
46.34 ± 7.00 44.89 ± 9.78 44.93 ± 8.34 42.57 ± 10.24 40.45 ± 7.41* V˙ O2peak (mL kg−1 min−1) V˙ Epeak (L min−1) 133.90 ± 41.17 128.10 ± 34.55 132.21 ± 32.52 132.63 ± 33.08 133.44 ± 27.78 * Significantly lower than sealevel (SL), P ≤ 0.05
HRpeak (b min−1) SPO2 (%)
180 ± 11
178 ± 11
178 ± 9
95 ± 5
95 ± 2
92 ± 2
below a FIO2 of 0.14 (3,000 m) but not above. Another possible explanation is that lactate concentration, due to an increase in glycolytic flux, is only elevated in a more severe hypoxic condition (i.e. at or above 4,000 m). According to others (Overgaard et al. 2012) arterial lactate concentrations in severe hypoxia (FIO2 0.10) during exercise is 10 times greater of that observed in normoxia with the same, however at moderate altitudes (FIO2 0.15) similar capillary lactate concentrations between hypoxic and normoxic conditions have been reported during cycling time trial performance workload (Amann et al. 2006). Therefore, from this evidence a reduction in VT would not be expected below 3,000 m. As such, it is also recognized that a lack of lactate and pH measurements is a limitation of the present study and future studies should if feasible look to include these measures. Previous research has demonstrated that WVT is a more accurate measure of physical performance (e.g. time trial performance) than V˙ O2peak (Amann et al. 2006; Lucia et al. 2004; Subudhi et al. 2006). Research has previously demonstrated that WVT is reduced with hypoxic exposure by up to 41 % (Subudhi et al. 2006). In the present study algorithms 1 and 2 did not confirm such findings suggesting that acute hypoxic exposure does not reduce WVT. However, algorithm 4 produced an overall change in WVT but similar to V˙ O2VT the direction of change could not be identified. In summary, it can be concluded that acute hypoxia had no effect on the VT determined via a GXT. Those studies, which reported a significant reduction in V˙ O2VT, were performed at or above 4,000 m (Fukuoka et al. 2003; Ozcelik and Kelestimur 2004; Subudhi et al. 2006), which may explain the disparity between results. However, it is difficult to clarify whether arterial O2 saturations were similar between studies as previous research did not report these data. Moreover, differences in acclimatization status/ exposure duration (Subudhi et al. vs. current study; 24 h vs. ˙ peak 354 vs. 243 W) and the
