Changes in breath 13COJ12C02 consequent to exercise and hypoxia

BARSTOW,THOMASJ.,DAN AND Changes in breath 13C02/12C02 consequent to exercise and hypoxia. J. Appl. Physiol. 66(2): 936- 942, 1989.-Because the natural enrichment of carbohydrate with 13C is greater than that of lipid, we hypothesized that the natural enrichment of exhaled CO2 with 13C (EN) could be used to gauge endogenous substrate utilization in exercising human subjects. To test this, EN and the respiratory exchange ratio (R) which equals the respiratory quotient (RQ) in the steady state, were measured simultaneously in seven subjects. Rest and exercise protocols, performed under conditions of room air (sea level) and hypoxic (inspired O2 fraction = 0.15) breathing, were chosen to cause a variety of patterns of oxidative substrate utilization. Work rates were performed both below and above the subject’s lactate threshold (LT). Work above the LT was expected to cause the greatest increase in EN reflecting greater utilization of glucose. There was significant intersubject (P < 0.05) but not intrasubject variability in resting EN. By 40 min of exercise, EN increased significantly (P < 0.05) over resting values in all exercise protocols during both room air and hy- poxia conditions. In the room air studies, we found no difference in EN during the below-LT work, even though there were significant increases in O2 uptake (VO,). In contrast, above-LT work resulted in significantly greater increases in EN by 20 and 40 min of exercise (P C 0.05). Contrary to our expectations, we observed no separate effect by hypoxia on the EN during exercise. Both EN and R tended to increase from rest to exercise, but during exercise there was no overall correlation between R and the EN. EN reflects changes in endogenous substrate utili- zation over relatively long periods of time such as at rest, but delays in the appearance of 13C02 at the mouth due to dilution in body CO2 pools, and possibly isotopic fractionation, preclude the usefulness of EN as an indicator of endogenous fuel mix during short-term exercise.

consequent to exercise and hypoxia. J. Appl. Physiol. 66(2): [936][937][938][939][940][941][942]1989.-Because the natural enrichment of carbohydrate with 13C is greater than that of lipid, we hypothesized that the natural enrichment of exhaled CO2 with 13C (EN) could be used to gauge endogenous substrate utilization in exercising human subjects. To test this, EN and the respiratory exchange ratio (R) which equals the respiratory quotient (RQ) in the steady state, were measured simultaneously in seven subjects. Rest and exercise protocols, performed under conditions of room air (sea level) and hypoxic (inspired O2 fraction = 0.15) breathing, were chosen to cause a variety of patterns of oxidative substrate utilization.
Work rates were performed both below and above the subject's lactate threshold (LT). Work above the LT was expected to cause the greatest increase in EN reflecting greater utilization of glucose. There was significant intersubject (P < 0.05) but not intrasubject variability in resting EN. By 40 min of exercise, EN increased significantly (P < 0.05) over resting values in all exercise protocols during both room air and hypoxia conditions. In the room air studies, we found no difference in EN during the below-LT work, even though there were significant increases in O2 uptake (VO,). In contrast, above-LT work resulted in significantly greater increases in EN by 20 and 40 min of exercise (P C 0.05). Contrary to our expectations, we observed no separate effect by hypoxia on the EN during exercise. Both EN and R tended to increase from rest to exercise, but during exercise there was no overall correlation between R and the EN. EN reflects changes in endogenous substrate utilization over relatively long periods of time such as at rest, but delays in the appearance of 13C02 at the mouth due to dilution in body CO2 pools, and possibly isotopic fractionation, preclude the usefulness of EN as an indicator of endogenous fuel mix during short-term exercise.
natural enrichment; carbon isotopes; respiratory exchange ratio; fuel mix A STABLE ISOTOPE of carbon, 13C, occurs naturally in both organic and inorganic substances at a variety of abundances relative to 12C. The abundance of 13C is higher in glucose than in lipid (5, 6). Enrichment of exhaled CO2 with 13C02 (EN) accurately reflects the pattern of enrichment of venous and arterial blood bicarbonate (8,20). Thus, one might predict that changes in the relative contribution of carbohydrate to lipid oxi-dation could be detected by changes in the ratio of 13C/ 12C in exhaled CO2. In human subjects during resting conditions the EN has been shown to reflect the relative rates of oxidation of both endogenous (20) and exogenous (14, 18) substrate. Furthermore, Pirnay et al. (19) and Krzentowski et al. (13) used changes in EN to calculate the oxidation of orally administered (exogenous) carbohydrate given during exercise. We therefore wondered if changes in EN could be used in endogenous substrate oxidation during exercise. In fact, we hypothesized that EN may more accurately reflect substrate oxidation than the respiratory exchange ratio [R, the ratio of CO2 elimination (VCO~) to 02 uptake (VO,)]. During unsteady states, such as for the first several minutes of exercise, changing CO2 stores preclude R from equaling the tissue respiratory quotient [RQ, which reflects the proportion of carbohydrate to lipid oxidation (9, 12, IS)]. Under these conditions EN, which may be insensitive to transient hyper-or hypoventilation, could potentially give a better indication of the oxidative fuel mixture. Because 13C tracers are increasingly used in studies of metabolism during exercise, it is also important to document the magnitude and patterns of exercise-induced changes in EN-Preliminary reports suggest that the naturally occurring 13C/12C of breath CO2 does change during exercise (16,20,26), but they are inconclusive as to the time course of change in EN during exercise, and how these changes might reflect differences in the oxidative fuel mix. To test whether changes in EN could be used to gauge endogenous substrate utilization during exercise, we examined '"CO2/'"CO2 in normal subjects under conditions known from previous studies to represent different mixes of substrate utilization, namely, rest and exercise of varying work rates under conditions of room air breathing and hypoxia (4, 23). Gas exchange was measured continually throughout the exercise, simultaneously with periodic determinations of EN. Thus, any changes in the R would be readily apparent and could be compared to changes in EN.  Fig. 1). At the low work rate [WR-1; 50% of the subject's lactate (or anaerobic) threshold (LT)], we expected to find little or no increase in EN because there may be little change in whole body substrate utilization at low work rates. The second work rate (WR-2) was selected so as to be below the subject's LT during normoxie exercise, but above it during hypoxic breathing.

METHODS
We have previously demonstrated that under conditions identical with WR-2, breathing FIN, of 0.15 can itself increase blood glucose turnover (4). Therefore, we predicted for WR-2 an additional increase in '"COJ'"CO2 under the hypoxic compared with the room air condition. Finally, a heavy work rate (WR-3) was chosen to be above the subject's room air LT. We predicted that EN would be further elevated at this high work rate where glucose turnover is increased (4). Subjects. The subject population consisted of seven male volunteers, ranging in age from 21 to 37 years, who were in good health and without any previous history of chronic respiratory or other disease. Each subject's age, weight, and room air and hypoxic gas breathing VOzrnax and LT values are presented in Table 1. Informed consent was obtained from each subject before entry into the study.
ProtocoZ. Each subject participated in six separate exercise sessions, each in a separate day. The first session consisted of two ramp cycle ergometer tests to fatigue, the first one under normoxic conditions and the second while breathing hypoxic gas. At least 30 min of rest was allowed between ramp tests. The order of these tests was chosen for two specific reasons: 1) incremental tests to fatigue under room air conditions demonstrate good reproducibility regarding the parameters of interest . WO 2 max9 LT, efficiency of work, etc) when separated by 30 min or more (24); 2) we did not, however, know what potential interactive effect previous exposure to hypoxia would have on subsequent exercise tolerance during normoxia. Thus the room air test was administered before the hypoxic one in all subjects. The LT for each inspired oxygen tension was determined from pulmonary gas exchange responses as previously described (24) ( Table 1). From the LT and Vozrnax the work rates were selected for the constant exercise tests described above.
Each of the five constant work rate tests was conducted, 1 wk apart, in the morning, with the subject having fasted for lo-12 h. Diet was otherwise not controlled, permitting us to evaluate day-to-day variability in resting E N. Base-line samples of exhaled gas were collected for determination of EN as described below. When the study was performed during hypoxia, the subject breathed 0.15 FIN, for 40 min before exercise. For the last 5 min of the resting base-line period and throughout 40 min of exercise, pulmonary gas exchange and heart rate were measured breath-by-breath.
Base-line samples (60 ml) of exhaled gas for determination of 13C02 enrichment were collected at rest, at 20 and 40 min of rest during hypoxia, and at 5, 10, 20, and 40 min of exercise under each condition.
Measurement ofpulmonary gas exchange. The subjects breathed through a low-impedance turbine volume transducer and breathing valve, with a combined deadspace of 170 ml. Oxygen and carbon dioxide tensions were determined by mass spectrometry from a sample drawn continuously from the mouthpiece at 1 ml/s. The inspired and expired volume and gas fraction signals underwent analog-to-digital conversion, from which oxygen uptake (VO,, STPD), carbon dioxide elimination (VCO~, STPD), and minute expired ventilation (VE, BTPS) were calculated on-line, breath-by-breath, as previously described (3). Heart rate was obtained beat-by-beat using a modified lead V5 electrocardiogram.
Determination of 13C/12C of CO2. 13C/12C ratios in CO2 were determined and reported relative to the 13C!/12C of the PDB (BelemniteZZa americana) standard (1.1237% 13C). The results are expressed in 613C as The 613C values of the CO2 samples were determined with a Nier 60" double-collecting mass spectrometer as modified by McKinney et al. (17). An alternative unit sometimes used to report '"C02/'"C02 is the atom per cent excess (APE) where 1 "per mil" (6) is equal to 1.123 x 1O-3 APE (26).
Data analysis. The effects of work intensity and hypoxia on enrichment of CO2 during exercise were tested using analysis of variance (one-and two-way with repeated measures). Differences between means were BREATH '"CO,/'"CO, IN EXERCISE AND HYPOXIA tested with Duncan's multiple range test. Because of unequal variances among the subjects for resting natural enrichment values, Friedman's nonparametric analysis of variance was used to evaluate differences between the subjects. Student's paired t test was used to compare the value of natural enrichment at 5 min into exercise with the preexercise base line. Significant difference was declared at P < 0.05. Regression analysis was used to determine any association between the respiratory exchange ratio R and the natural enrichment EN.

RESULTS
Rest. There were no significant differences in VOW or R among the rest periods preceding the three room air exercise bouts (mean \jo2 of 0.422,0.400,0.398 liter/min for rest preceding WR-1, WR-2, and WR-3, respectively; corresponding mean R of 0.86, 0.83, and 0.87); likewise, 40 min of breathing hypoxic gas produced no significant changes in either variable (mean v02 of 0.430 and 0.428 l/min for WR-1 and WR-2, respectively, with corresponding R value of 0.85 and 0.82).
Typical natural enrichment values for CO2 during rest and the five constant work rate protocols are shown for one subject in Fig. 2. As can be seen, there is day-to-day variability in the base-line EN for this subject, and the magnitude of this variability differed among the seven subjects. Hypoxia did not significantly change the EN from base-line values at either 20 or 40 min of rest (for rest preceding WR-1: mean base-line EN = -21.5,20 min = -21.5,40 min = -21.6; for rest preceding WR-2: baseline EN = -21.7, 20 min = -21.8, 40 min = -21.9). Table 2 details the intra-and intersubject variability in resting EN in the overnight fasted state determined from the initial base-line values obtained before exercise or exposure to hypoxia. There was a significant subjectto-subject difference for the 5-day average of the resting base line natural enrichment (P < 0.05). The coefficients of variation determined for each subject from the resting values for the five protocols ranged from 2.0 to l3.1%, whereas that determined from the means for each of the -   seven subjects was 5.2%. There was no significant relationship between resting respiratory exchange ratio and resting EN (not shown).
Exercise. The average responses of TO, and R over minutes 5-20 of exercise are summarized in Table 3. VOW increased with each work rate. Hypoxia had no effect on exercise v02, either at work rate 1 or 2. As shown in Table 3, the selection of work rates produced the desired metabolic rates (Fig. 1). Both of the ~oZ responses for WR-1 were below the respective LTs, the VO, response to WR-2 was below LT on room air but exceeded the LT on hypoxia, whereas WR-3 resulted in a v02 which was clearly above LT. In addition, the average R for the three work conditions below the LT were indistinguishable as shown in Table 3, whereas work performed above the LT resulted in significantly higher R values (P < 0.05).
The effect of work intensity during normoxia on group mean values for EN is shown in Fig. 3A, plotted as the change in EN from resting values. EN fell significantly 5 min into exercise (P < 0.05, paired t test) for all three work rates, and then rose, becoming significantly greater than resting values at 20 and 40 min of exercise for each protocol (P < 0.05). There was no difference in the changes in EN between the two below LT work rates at any time during exercise. However, the changes in EN at 20 min (As = 1.79%0) and 40 min (As = 2.71%0) for the above LT work rate were significantly greater (P C 0.05) than those observed at the two below LT protocols (work rate 1: A6 = 0.66%0 and 1.16%~~; work rate 2; A6 = 0.93%0 and l.33%0 for 20 and 40 min, respectively).
The effect of work rate relative to the lactate threshold on changes in breath 13C02 during air breathing can be seen clearly in Fig. 4, where the change in P3C02 is   3. Effect of work intensity on breath '"C02/'"C02. Values (ME of mean difference) represent change from rest (A613C02). Symbols as in Fig. 2. t Significantly different from resting values (P < 0.05). * Significantly different from below LT values (P c 0.05). A: normoxic response. All 3 protocols resulted in an increase in EN above resting values at 20 and 40 min. There was no difference between the 2 below LT work rates at any time; WR-3 (above LT) resulted in significantly higher EN values at 20 and 40 min. B: normoxic vs. hypoxic responses at WR-1. There were no differences between responses at any time. C: normoxic vs. hypoxic responses for WR-2. Hypoxia depressed the EN below rest at both 5 and 10 min, but neither value was significantly different from the corresponding normoxic value.
plotted as a function of metabolic rate (VOW) at the end of exercise. Although there was a substantial increase in metabolic rate between work rates 1 and 2, there was no concomitant increase in P3C02. However, when a further increase in metabolic rate resulted in a VOW which exceeded the normoxic lactate threshold ( Fig.  2. Values are means t SE for 602 and SE of mean difference for A613C02. All end exercise values were significantly greater than resting. Despite a large increase in $70~ from WR-1 to WR-2, there was no additional increase in end exercise A613C02. Only when exercise exceeded LT did end exercise A613C02 rise further.
at the end of exercise was significantly greater than for either below LT work rate.
As shown in Fig. 3B, the pattern of change in EN during hypoxic exercise at the lower work rate  was identical with that seen during normoxia, except that EN was not depressed 5 min into exercise. At the intermediate work rate (WR-2), which was below LT for normoxia but above LT under hypoxic conditions, the fall in EN during hypoxia was significant at both 5 and 10 min of exercise compared with rest (P < 0.05), but was not significantly different from the normoxic response (Fig. 3C). By 40 min, EN had risen significantly above rest (P c 0.05) and was indistinguishable from the normoxic value. Thus, hypoxic exercise performed above the hypoxic LT did not result in a significantly higher EN at the end of exercise compared with below LT exercise, in contrast to the normoxic above LT condition (WR-3). Although the changes in EN during hypoxic exercise were similar to those found during normoxia, the responses were more variable, as denoted by the larger standard error bars for the hypoxic data in Fig. 3, B and C.
The time course of change in EN during exercise (Figs. 2 and 3) did not parallel the change seen in the R (Fig.  5). For clarity, only the R during normoxic exercise is shown; the hypoxic responses showed the same trend over time, and the mean values are given in Table 3. Although both the R and EN signals decreased initially with the onset of exercise, the R rose more quickly, so that a peak value was obtained by 5-10 min; after this the R fell slowly over the remainder of exercise. In contrast, EN continued to rise from 5-10 min through 40 min of exercise (Fig. 3). However, the inverse responses of R and EN with time during exercise were generally not related to each other, as demonstrated in Fig. 6, where group mean responses for A613C02 are plotted against the corresponding changes in R from base line at 5, 10, 20, and 40 min of exercise for all five exercise protocols. The exception was WR-3, where changes in EN from rest  Table 3). Note downward trend from 10 min to the end of exercise (40 min). of changes in respiratory exchange ratio (AR). Symbols as in Fig. 2. Note overall lack of relationship between these 2 potential indicators of oxidative fuel mix. For WR-3, however, 13C02 was significantly negatively correlated (r = 0.93, P < 0.05) with R.

DISCUSSION
This study demonstrated that natural enrichment of breath CO2 with 13C02 during exercise was consistently higher than that during rest (Figs. 2 and 3). However, our d .ata sh .ow ' that the changes in EN of breath CO2 seen with exerci .se unde r normoxic conditions are n .ot monotonically related to the intensity of work being performed (Fig. 4). Increases in work below the LT, even when accompanied by a substantial increase in metabolic rate, did not result in a concomitantly higher EN. In contrast, increases in work which required a metabolic rate that was above the LT resulted in a significantly greater increase in '"COJ'"CO2 in the expired gas. It also appears that hypoxia, even when it transforms below LT work into above LT work, does not have any significant additional effect on the pattern of change in EN during 40 min of exercise.
The resting background EN of exhaled CO2 was signif-icantly different from subject to subject, suggesting that it should be considered when interpreting stable isotope changes under various metabolic conditions. Because resting RQ was not significantly different, this variability was likely caused by variability in the '"COJ'"CO2 of the foodstuffs chosen ad libitum by the subjects.  (2l), which would therefore prevent direct association of a particular absolute EN with a specific carbohydrate-to-lipid oxidation ratio, as can be done for the RQ. Both R and EN were elevated above resting levels during exercise, with work above the normoxic LT (WR-3) producing the greatest increase in both variables (Figs. 3 to 5). This is consistent with an apparent increase in carbohydrate utilization occurring above the LT (4, 23).
However, several factors may influence the relationship between either R or EN and the oxidative substrate (i.e., the true RQ). If the various body depots of glycogen (liver and muscle) and triglycerides (intramuscular and adipose) differ in their EN values, then conceivably any changes in the tissue source of oxidative substrate with exercise could result in a dissociation between changes in EN and changes in the true oxidative substrate mixture. Thus, as pointed out above for the rest condition, during exercise it may not be possible to associate a specific breath EN with a specific oxidative fuel mix. We are unaware of any information regarding the variability in EN for carbohydrate or triglycerides in various tissue. R will reflect substrate oxidation (RQ) under conditions where CO2 stores are likely to be constant (e.g., rest and steady state exercise below the LT). However, during conditions when CO2 stores are not constant, such as with hyperventilation, at the onset of exercise, and in response to changing lactic acid concentrations seen for work above the LT, R will not reflect the oxidative fuel mix. This can be seen in Fig. 5, where the R for WR-3 on room air is above 1.0 for approximately the first 20 min of exercise.
Both R and EN showed a transient fall early in exercise, followed by a slower rise; the time course for the changes in R, however, was much faster than that of EN (Figs. 3 and 5). This more rapid response of R relative to EN when the oxidative fuel mix is manipulated during exercise can also be seen for endogenous substrate utilization (16,20) and after an oral glucose load (19). For neither R nor EN is the drop early in exercise reported here likely a reflection of transient increased lipid oxidation. Both R and EN are ratios of two processes, where the numerators for each [total CO2 elimination (VCO,) for R, 13COa elimination for EN] increase more slowly during exercise than the respective processes in the denominator (VOW for R, VCO~ for EN). The consequence of this discrepancy in rates of increase between numerator and denominator leads to an obligatory transient fall in the ratio for both R and EN. For R, the discrepancy between VOW and VCO~ kinetics is due to the relatively greater storage capacity for COz relative to 02 (2). Regarding EN, the attenuation in the rate of rise in 13C02 excretion relative to that of total COa elimination (VCO~) appears not to be the result of isotopic fractionation across the lungs (8) (11). boz, and thus the total CO2 in these pools, appear to approach a new steady state more rapidly after the onset of exercise (within 4-6 min) than does the enrichment of those pools with 13C02. In summary, VOW adjusts slightly more rapidly from VCO~ in the transition from rest to exercise, whereas 13C02 adjusts much more slowly. Therefore, changes in EN of breath CO2 will be less likely to reflect the underlying oxidative fuel mix during short duration exercise (up to 40 min) than will the R.
Hypoxia did not alter EN at rest, consistent with our previous finding of no change in peripheral blood glucose turnover (4). In that study it was also shown that a similar alteration of inspired oxygen concentration at a constant work intensity spanning the lactate threshold produced a marked increase in glucose turnover. Under similar conditions here (WR-2), R, but not EN, was elevated during hypoxic compared to normoxic work. This suggests that the attenuation in rise in breath EN with hypoxia may have obscured a true underlying increase in 13C02 production from carbohydrate oxidation. Hypoxia is known to profoundly affect the circulation (7, lo), which may alter the dynamics of bicarbonate flux in the body, and thus alter the appearance of 13C02 at the lungs. In summary, EN of expired CO2 during rest in postabsorptive, healthy subjects can show large intra-and intersubject variability. Both EN and R increase during exercise, and above LT exercise elicits a significantly greater increase in both variables than does below LT work, but the time course of change in EN is delayed relative to that of R for both below and above LT exercise. Finally, hypoxia (FIN, = 0.15) does not significantly alter the pattern of change in EN at rest or during 40 min of exercise. EN may be a useful, noninvasive means of assessing endogenous fuel mix, so long as a steady state for 13C02 in the body's bicarbonate pools 1 1 1 l 1 has been achievea.