Oxygen uptake as related to work rate increment during cycle ergometer exercise

We postulated that the commonly observed constant linear relationship between $$\dot V_{{\text{O}}_{\text{2}} }$$ and work rate during cycle ergometry to exhaustion is fortuitous and not due to an unchanging cost of external work. Therefore we measured $$\dot V_{{\text{O}}_{\text{2}} }$$ continuously in 10 healthy men during such exercise while varying the rate of work incrementation and analyzed by linear regression techniques the relationship between $$\dot V_{{\text{O}}_{\text{2}} }$$ and work rate (Δ $$\dot V_{{\text{O}}_{\text{2}} }$$ / Δwr). After excluding the first and last portions of each test we found the mean ±SD of the δ $$\dot V_{{\text{O}}_{\text{2}} }$$ / Δwr in ml · min−1· W−1 to be 11.2±0.15, 10.2±0.16, and 8.8±0.15 for the 15, 30, and 60 W·min−1 tests, respectively, expressed as ml·J−1 the values were 0.187±0.0025, 0.170±0.0027 and 0.147±0.0025. The slopes of the lower halves of the 15 and 30 W·min−1 tests were 9.9±0.2 ml·min−1·W−1 similar to the values for aerobic work reported by others. However the upper halves of the 15, 30, and 60 W·min−1 tests demonstrated significant differences: 12.4±0.36 vs 10.5±0.31 vs 8.7±0.23 ml·min−1·W−1 respectively. We postulate that these systematic differences are due to two opposing influences: 1) the fraction of energy from anaerobic sources is larger in the brief 60 W·min−1 tests and 2) the increased energy requirement per W of heavy work is evident especially in the long 15 W·min−1 tests.


Introduction
There is good evidence to suggest that the 02 cost of aerobic cycle ergometry approximates 10.0 to 10.5 ml-min-1. W-1 (Gaesser and Brooks 1975 ;Wasserman and Whipp 1975;Spiro 1977) corresponding to a work efficiency of 27% (Pahud et al. 1980). Many investigators, using several-minute periods of constant cycle ergometry work over a wide range of intensities, have reported a linear relationship between the work rate and the resultant "steady-state" 02 uptake (12%) (Asmussen 1965;Astrand and Rodahl 1971;Cotes 1975). Others, after excluding the data obtained shortly after the onset of exercise, have found a similar linear relationship during cycle work to exhaustion, whether increment steps lasted a second or a minute or two (Nagle et al. 1971;Wasserman and Whipp 1975;Spiro 1977;Jones and Campbell 1982;Davis et al. 1982).
It is puzzling that a linear increase in work rate appears to elicit a strictly linear response of 12o2 because exercise at high work rates is not supported solely by atmospheric oxygen but is supplemented energetically by ATP generated by anaerobic metabolism (Keul et al. 1972). W e hypothesized that a careful analysis of V% responses to incremental work would reveal a nonlinear relationship and the pattern of the responses would depend on the magnitude of the incremental work rate. We analyzed the Vo2 response to exercise using breath-by-breath measurement of gas exchange in 10 healthy young men during cycle ergometry to exhaustion, using different work rate increment protocols.

Materials and methods
Offprint requests to: J. E. Hansen, Box 24, Harbor-UCLA Subjects. Ten healthy men volunteered for the study. Their Medical Center, Torrance, CA 90509, USA mean (_SD) age, height, and weight were 22+2.5 years, 177_+7.4 cm, and 83+4.9 kg, respectively. They were nonsmokers, they had no history or systemic disease and were not engaged in physical training or dietary programmes.
Protocol. Each subject performed 6 tests in random order on 3 different days. They involved work rate increments of 15, 30, and 60 W-min-~ ; duplicates of each l:est were performed by each subject. The daily tests were separated by 1 to 2 h, test days by 1 to 21 days. In each study, after 4 min of unloaded pedalling at 60 rev. min-1 on an electromagnetically braked cycle ergometer (Godart), work rate was increased every 1/2 s at 1 of the 3 rates under computer control (ramp pattern). The increments continued until the subject could no longer maintain pedalling frequency.
Data collection. The subject breathed through a mouthpiece attached to a turbine device (Alpha Technologies) which measured expired and inspired volume continuously. Respired gas was sampled from the mouthpiece at a rate of 60 ml. min-1 for continuous measurement of 02, CO2 and N2 by mass spectrometry (Perk• MGA 1100). After computer alignment of the gas concentration and volume signals for the transit delay and response time of the :mass spectrometer, 12o~ was computed breath-by-breath as previously described (Beaver et al. 1981). We defined the maximum Vo~ as the average V% during the last 10 s of exercise.
Data analysis. The difference between the breath-by-breath I2o 2 and the mean I5"% of the last 3 rain of unloaded pedalling was calculated and termed the AI)%. For each test, we plotted and analyzed the continuous relationshi p between either time or work rate on the abscissa and the AVo2 on the ordinate. For analysis of the A 12o~ versus work rate relationship (A l)o/Awr) which can also be expressed as Al2o~.J-I we laterally shifted the data for each test towards zero on the same graph by 45 s. This duration approximates the time constant for Vo~ increase . This lateral shift is 11.25 W for the 15 W-min -1 ramp, 22.5 W for the 30 W.min 1 ramp, and 45 W fur the 60 W-rain-~ ramp. (Please see Appendix for rationale and physiologic basis of this shift.) We used the least squares method of linear regression to analyze the average slope of the Al)%/Awr for each test, ex-cluding the first 100 s and last 15 s of the response for the following reasons: 1) The response of Vo2 to incremental exercise has been shown to approximate a first order system, so it is predictable that there will be an initial lag in A (/o2 after the onset of incremental work (Whippet al. 1981). As the time constant for 12% averages 40--45 s in healthy subjects, over 95% of this lag should be finished by 100 s (one time constant approximates 63% of the expected change while two approximate 95%). 2) In several tests the I5% reached a plateau 10 to 20 s before exercise ended.
Because the Al)o2/Awr for each subject was not strictly linear by inspection, we divided the response into two portions (lower and upper) and recalculated the AlYo2/Awr over the same time porti.ons of the tests, dividing the portions by the time at which Vo2 reached half of the distance between unloaded l;ro2 and maximum I~o2.
We used a paired t-test to compare slopes of the two halves of each exercise test and an analysis of variance with the Tukey test to compare slopes among the three exercise increments. We considered p < 0.05 significant.

General
We excluded 2 tests of the 60 tests from analyses (one each of the 15 W. min-~ increment tests of subjects 6 and 7) because of technical errors and based all calculations on the remaining 58 tests. The unloaded 1/o2 correlated positively with body weight. Among subjects, the maximum lZo2 did not differ significantly between work rate increments, tests performed on a given day, or during the course of the study. However, the maximum work rate achieved was always highest for the 60 W. min-1 work rate increment (Table 1).

Visual analysis
For each subject, l)'o~ (plotted against time): a) rose promptly and stabilized during unloaded pedalling, b) began rising in a curvilinear pattern for the first two minutes of incremental work, c) maintained a relatively linear pattern during the next portion of the test, and d) often deviated from this line in the latter portion of the test. 2. I)'o2 above that of unloaded pedalling (zX l)o~) vs. work

Fig.
rate at 3 rates of increasing work. Data are from the same tests as Fig. 1. Note that when the work rate increases at 60 W. rain-1, the maximum I)o 2 is reached at the highest absolute work rate ment rate were invariably steepest and for the 60 W. rain-1 were shallowest for the upper portions of the curves (see Fig. 3). This same systematic pattern was evident for each subject if the plots for the 3 work rate increments were shifted by 30 s or 60 s, representing shorter or longer time constants for Vo~ increase.

Computer analysis
The slopes of the A l2o2/Awr , analyzed by least squares, are presented in Table 2. Individually and as a group, the slopes were steepest during the 15 W-min -1 rate, intermediate during the 30 W.min -~ rate, and shallowest during the 60 W. min-i rate.
Because the AVe2 vs work rate relationship was not strictly linear at all rates of increase, the lower and upper halves of the curves were analyzed separately as shown in Table 3. Despite the similarity of the slopes at the 15 and 30 W. minrates for the lower halves, the slope for the upper half of the 15 W.min -~ rate exceeded the 30 W. min-1 rate. At both of these rate increments, the slopes for the upper halves were significantly greater than that for the lower halves. The slopes for the upper and lower halves of the 60 W. min-~ rate were not significantly different but they were both lower than the 15 and 30 W.min-~ incremental tests. To assess the possibility that our findings were dependent on exactly which portions of the relationship we excluded, ] and 2. Note that the shifted plots are virtually coincident at lower work rates but have definably different slopes at the higher work rates. The A12o 2 rises most steeply with the 15 W. min ~ test, These same visual relations are seen in the data from all 10 subjects Table 2. Average slope of 02 uptake vs work rate relationship (ml-min-1. W-1 ; divide by 60 to obtain ml. J-1) during cycle ergometer incremental work rate tests. Work rates were increased in ramp pattern at the rates shown we repeated the computer analyses after 1) excluding the first 135 s (rather than 100 s) and 2) excluding the last 60 s (rather than 15 s). These alternate procedures did not appreciably alter the calculated slopes nor the statistical results.

Discussion
We undertook this analysis to ascertain whether the apparent constancy of the slope of the l)'o2work rate relationship found by others during cycle ergometry was fortuitous or an invariantly observed phenomenon. For mild and moderate exercise, the constancy of the 02 cost of external work at 10.0 to 10.5 ml-min-l.W -1 seems well-established considering the reports of many investigators who used constant work rate protocols (Asmussen 1965; Astrand and Rodahl 1970;Nagle et al. 1971;Cotes 1975;Gaesser and Brooks 1975;Wasserman and Whipp 1975;Whipp et al. 1981) or slow or rapidly incremented exercise tests (Nagle et al. 197l;Wasserman and Whipp 1975;Spiro 1977;Jones and Campbell 1982;Davis et al. 1982). However, our study did not confirm the previously reported (Cotes 1975;Spiro 1977;Whippet al. 1981;Davis et al. 1982;Jones and Campbell 1982;Younes 1984)  heavy work is the temporary 02 sparing effect of the energy made available by anaerobic glycolytic mechanisms, i.e. the production of ATP accompanying the conversion of pyruvate to lactate (Di-Prampero 1981). DiPrampero suggested that lactate accumulation in young non-athletic subjects should spare a total of 50 ml of 02 per kg of body weight, equivalent to a volume of approximately 3-5 1 of 02 (02 deficit) in our subjects. In exhausting exercise of the durations used, the quantity of lactate accumulated and 02 deficits should be approximately equal in all tests of a given subject (Astrand and Rodahl 1970;Astrand et al. 1963;Karlsson et al. 1972). Considering the brief durations of the 60 W.min -1 tests (Table 1), these 3-5 1 of O2 would be temporarily "spared" during 2 to 3 min; whereas in 15 W.min -I tests, this ef-fect would be spread over approximately 8 to 10 min.
Several factors might be expected to increase 12o2 disproportionately during exhaustive work: concurrent metabolism of lactate to glucose; elevated body temperature and catecholamines; and disproportionate increases in ventilation and myocardial work. These will be briefly considered.
To the extent that lactate is metabolized to glucose or glycogen during exercise, there is an obligatory increase in 02 requirement without the accomplishment of external work (Krebs 1964;Katz 1986). This may be a dominant factor in the O2 cost of anaerobic work (Casaburi et al. 1987). The increase in body temperature in exercise of 5 to 20 min duration is likely to be less than 1.5 ~ (Saltin and Hermansen 1966). Although catecholamine levels rise markedly during heavy exercise (Hartley et al. 1972), their quantitative effects on 02 consumption during exercise, considering measurements made during rest (Sjostram et al. 1983), are likely to be small. Quantitative estimates of the energy cost of ventilation vary widely (Shephard 1966;Whipp and Pardy 1986), but all show an increase in energy cost per 1 of ventilation at higher minute ventilations. Shephard estimates that ventilatory energy costs increase from 1--3% at rest to 15% of the body's total energy requirements at a ventilation of 100 1. min-1. Extrapolating from the findings of Kitamura et al. (1972) and Nelson et al. (1974), who directly measured myocardial 1)%, heart rate, and blood pressure during several levels of mild to moderately heavy exercise in healthy young men, we estimate that myocardial V% increases from approximately 2.5% of the total 12o 2 at low levels of exercise to 3.5% to 4.0% of total V% at very heavy levels of exercise. In our studies, the maximum heart rates and maximum ventilations for each person differed by less than 10% for each of the work rate increments. Despite the uncertainties of the above estimates, the proportion of total energy expenditure attributable to all of these physiological factors would tend to be similar for each person for his six tests.
The low slopes of the early portion of the 60 W. min-1 tests may be partially due to the limited amount of data available for analysis (average of 69 s after excluding the first 100 s of each test). In contrast, the early portions of the 15 W.min -1 and 30 W.min -1 tests have slopes of 9.9 ml-min-l.W -1, values similar to those previously reported for incremental tests of "steady state" tests (Nagle et al. 1971;Whipp et al. 1981; Davis et al. 1982;Hughson and Inman 1986). The upper portions of the 30 W. min-1 tests have had AVo2/Awr slopes averaging 10.5 ml-min-1. W-1, also consistent with prior reports obtained in young men at this increment rate (Whipp et al. 1981;Davis et al. 1982). However, for the 15W-min -1 tests the upper portions of the A lkoJAWr were significantly higher at heavier work rates; for the 60 W.min-1 tests they were significantly lower. Previous investigators may not have discerned these differences because they utilized a narrower range of work rate increments or because data were not analyzed by computerized regression procedures.
We believe our findings may be explained by considering: a) the magnitude and duration of the 02 deficit accumulation, and b) probable differences in the efficiency of work at low and high work intensities. First, we postulate that the lower Alko2/Awr slope seen in the 60 W-min -1 tests resuits from the major energy contribution of anaerobic glycolysis (several 1 of oxygen deficit) over a very short period of time, approximately 2 to 3 min. Second, the finding that the A V%/Awr increases during the latter portions of the 15 W.min -1 strongly suggests a higher energy cost per W of heavy or very heavy work than per W of mild of mild or moderate work. During these slower tests, the temporary 12% sparing effect of the accumulating 02 deficit is, on average, spread over 8 to 10 min and thus influences the AV%/ Awr less than during rapid tests. Third, the nearlinearity usually found in the intermediate duration tests (30 W. min-1) is not due to an unchanging work efficiency, but can be explained by what appears to be a fortuitous balance between the higher energy cost per W of heavier work and the temporary 1/o2 sparing effect of the accumulating 02 deficit. Although this explanation must be considered conjectural at this point, recognition of the non-linearity of the Alk%/Awr relationship is a necessary first step towards establishing the mechanisms dictating the 12o 2 during heavy exercise.
where t is time after the work rate ramp begins, r is the time constant of the system, 12o~ is the slope of the steady-state 12o2 --work rate relation (ml.min -~.w 1) and wr.~ is the rate of increase in work rate (work rate slope). If the Vo~ response is related to work rate (wr) rather than time, 12o2 = Vo~ [wr-wrs r(1 -e -wr/ .... )] (2) note that for t> 2--3 r, (i.e., wr/wrs> 2--3 r) this reduces to Vo2 = 15"o2~ (wr-wr~ r) Thus it can be seen that, except for the initial few minutes of data, the Vo2-work rate relation for various ramp slopes should be parallel and displaced from each other by (wrsr') watts. The Vo2-work rate responses are plotted in Fig. 3 on axes where each curve was shifted by an amount (wrsT); deviations from linearity are thus more readily distinguished.