Oral [ 13C]bicarbonate measurement of CO2 stores and dynamics in children and adults

Oral [13C]bicarbonate measurement of CO2 stores and dynamics in children and adults. J. Appl. Physiol. 69(5): 1754-1760, 1990.-During exercise, less addi- tional COn is stored per kilogram body weight in children than in adults, suggesting that children have a smaller capacity to store metabolically produced CO*. To examine this, tracer doses of [13C]bicarbonate were administered orally to 10 children (8-12 yr) and 12 adults (25-40 yr) at rest. Washout of 13C02 in breath was analyzed to estimate recovery of tracer, mean resi- dence time (MRT), and size of CO2 stores. COZ production (%o~) was also measured breath by breath using gas exchange techniques. Recovery did not differ significantly between chil- dren [73 t 13% (SD)] and adults (71 t 9%). MRT was shorter in children (42 t 7 min) compared with adults (66 * 15 min, P < 0.001). VCO* per kilogram was higher in the children (5.4 t 0.9 mlgmin-l l kg-l) compared with adults (3.1 t 0.5, P


IN HUMANS AND OTHER MAMMALS,
COz regulation is characterized by large tissue COa stores, high PCO~ in the circulating blood relative to other animals, and control of blood COa concentration within a very narrow range. The stores of COa increase transiently with increasing metabolic rate (e.g., during exercise, Ref. 22) or under pathological conditions (e.g., respiratory failure). There is reason to believe that there are growth-related changes in the way CO:! is stored. First, both blood PCO~ and bicarbonate concentrations are lower in infants compared with young adults (4). Second, for a given increase in CO:! production (during exercise) children increase ventilation more than adults (7). Finally, although children, like adults, increase their COa stores in response to low-intensity exercise, the increase of stored CO, in children (per kg body wt) is only one-half that observed in adults (22). These observations suggest that regulation of COz dynamics and stores are different in children compared with adults.
We hypothesized that COa stores in children will be smaller than in adults. To test this hypothesis, it was necessary to develop methodologies specifically feasible for children. The washout of '*CO2 in the exhaled breath after intravenous injection of labeled bicarbonate was originally used in adult human and animal studies to characterize COz dynamics and pool sizes (15,20,23).
Breath measurements can be used because the ratio of labeled to unlabeled COn in the breath is virtually the same as in the venous blood (9). But radioactive tracers are clearly unwarranted in studies of healthy children. More recently, intravenous ['3C]bicarbonate (13C is a stable isotope representing -1% of carbon in the environment) has been used in both adults and infants (13,24), but oral administration of the labeled bicarbonate would be far more acceptable than intravenous for most children. Thus we designed this study to test the hypothesis that CO2 dynamics and stores in children are different from those in adults. We used the simultaneous measurement of COs production by gas exchange and the washout of 13C02 in the exhaled breath after a bolus oral dose of labeled bicarbonate. Groups of adults and children were studied under resting conditions. METHODS Subjects. Ten prepubertal children (6 boys, 4 girls) and 12 adults (11  The key variables are 1) the mean several reasons. First, it is about the longest time before residence time (MRT), which indicates the average time the onset of a stimulus such as our study that most spent by a labeled COn molecule in the whole system healthy children will tolerate fasting. Second, the gastric after oral administration, 2) the rate of COa production, emptying time in children is an exponential decay funcand 3) the steady-state mass of unlabeled COz in which tion with a half time of -1 h (19); therefore by 4 h only the tracer is distributed.
The area under the washout 6% of gastric contents will remain. curve (AUC) and area under the moment curve [AUMC, Within 30 min of ingestion, a 5-ml solution of 2 mg/ the moment curve is (DOBtime) as a function of time] kg NaH13C03 (99.0 atom% 13C, MSD Isotopes, Canada) were obtained by finding the sum of the areas obtained in water was prepared and sealed. Two baseline samples from two parts of the washout curve, the initial (t = 0 to of exhaled gas were collected to determine the subject's -60 min) and the tail. Trapezoidal fitting was used to natural enrichment of CO, with 13C. Each subject in-calculate the area of the initial part of the curve. Accurate gested the labeled bicarbonate rapidly and completely. fits of the tail were obtained in all subjects using a single Additional breath samples were taken after the ingestion exponential equation, and the area under the tail could at 2, 5, 10, 15, 20, 30, 45, 60, 90, 120, 140, 160, and 180 then be calculated analytically. min. Exhaled breath was collected in a balloon, trans- The following computations were made ferred to a 60-ml syringe, and sealed. Pulmonary gas exchange was measured breath by breath (as described MRT = AUMC/AUC (1) mass of CO2 = MRT&oz (2) below) for lo-min intervals every 0.5 h during the testing period.
Pulmonary gas exchange measurement.
The subjects breathed through a low-impedance turbine volume transducer and a breathing valve with a combined dead space of 90 ml. Paz and PCO~ 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 O2 uptake (VO,, STPD), C-O2 elimination (VCO~, STPD), and expired ventilation (VE, BTPS) were calculated on-line with each breath as previously The validity of the method of estimating MRT depends on the assumption that the system is linear and stationary, that there are no Con-bicarbonate traps within the exchanging system, and that label measured in the breath and any other unrecovered label is eliminated exclusively from the same "central pool." For estimating mass of COa we also assume that the mean time for label to reach the central pool is a negligible fraction of the MRT and that the equivalent source constraint (8) can be applied to entry of endogenous COz into the central pool.  venous studies, we also tested two-and three-exponential fits to the washout data. The general model was where Ai and Xi are the macroparameters of the model, t is time after oral administration, and n = 2, 3, or 4. In several cases we also fit a polynomial exponential model DOB(t) = (Al + A2t) *exit + A3-e'zt with the constraint Al + A3 = 0. The need for a polynomial exponential model was often revealed by two Xi in a three-exponential fit being close to each other, with both of their corresponding Ai of opposite sign and large in magnitude. A polynomial exponential model would be expected, for example, when gut absorption is a firstorder process and its rate constant matches one of the Xi (eigenvalues) of the system.
For each candidate mathematical expression, the best fit was found by the weighted least-square program BMDP3R (14) using weights inversely proportional to the square root of DOB, as determined by a previous analysis of residuals. The choice of the best-fitting model among these was made by eye and by appropriate comparisons among the fits using the F test (3, 16). The curve-fitting analysis was performed in five of the adults and five of the children. Our goal was to determine whether a single model could be used to accurately characterize the washout data in all subjects. Since performing the analysis first in five adults and five children, randomly chosen, demonstrated clearly that a single model did not emerge (see RESULTS), additional analysis of the remaining subjects was not performed.
NormaZization to body mass. When an attempt is made to study metabolic responses in children compared with adults, scaling the parameters to body size is of critical importance (6). To determine how a particular response differs between children and adults, it is necessary to minimize the effects of size alone on the response in question. Body weight is an accurate and easily obtained index of body size. Differences in metabolic parameters observed after normalizing to body weight imply the existence of fundamental processes of metabolism that are independent of body size and may be related to growth and development.
Statistical analysis. In addition to the tests outlined above, standard techniques of independent t tests, correlation, and linear regression were used. Results are presented as means t SD.

RESULTS
Characteristics and modeling of washout curves (Fig.   1). The shapes of the washout curves were different from those that have been observed after intravenous administration of tracer. In the latter, the peak DOB occurs virtually instantaneously, and, as noted, an equation consisting of the sum of three exponentials accurately fits the washout data. In the curves after oral administration (Fig. l), the peak DOB did not occur for several minutes. The peak DOB occurred at variable times with a mean of 11 t 4 min for the children and 12 t 7 min for the adults (NS). By 60 min an exponential decay of DOB followed in virtually all subjects. One consequence of the relatively long and variable initial (most likely, absorption) phase in the oral studies is to confound attempts to gain compartmental information. In all cases where fitting was performed, at least one of three expressions fit the washout data well. The twoexponential model produced a good fit in only one child and one adult. In three of five children tested and in four of five adults, the three-exponential model produced good fits. But the polynomial model produced equally good fits in all five of the children and in three adults. As expected, AUC and AUMC calculated from the well-fit models were virtually the same as those derived from the trapezoidal and exponential extrapolation techniques described above.
Mean residence time and recovery (Fig. 2, A and B).
The MRT of resting children (41.6 t 7.2 min) was significantly smaller than in resting adults (66.5 t 14.6 min). No difference was observed in the recovery between the two groups (children 73 t 13% and adults 71 t 9%).
To assess whether the recovery was dependent on the metabolic rate, a linear regression was performed using recovery as a function of the VCO~ normalized to body weight (VcoJkg). No significant correlation was found.
As expected (because the MRT was smaller in children), there was a significant correlation between MRT and body weight (r = 0.67, P C 0.001). However, no correlation was found between body weight and MRT within each group as can be seen in Fig. 3. Similarly, because children had higher metabolic rate per kilogram, there was a significant negative correlation between MRT and normalized metabolic rate (i.e., VcoJkg; r = -0.72, P = 0.0001); however, no correlation was found within each group.
Gas exchange and tracer estimates of CO2 production ( Fig. 4, A   There was a significant correlation (r = 0.67, P < 0.001). This was expected, because MRT was shorter in children compared with adults (see Fig. 2). However, as can be seen, no correlation between MRT and body weight was observed within each group. the uncorrected tracer estimate of COa production and the gas exchange measurement of VCO~. The regression equation for COZ production (tracer) vs. ho2 (gas exchange) in milliliters per minute was DISCUSSION There were marked differences between the children and the adults in COa storage dynamics measured by tracer dilution.
The MRT (Fig. 2) was significantly smaller in children, most likely reflecting their higher metabolic rate. Endogenous COa production and release to the atmosphere was faster in children; hence the average time spent by a COa molecule when introduced to the gastrointestinal tract was shorter. The shorter MRTs in children were observed despite the fact that the time at which peak DOB occurred was the same in children and adults. This latter observation argues against the possibility that growth-related differences in tracer estimate = 1.095 x \~co~ (gas exchange) + 65.2; r = 0.86, P c 0.0001 This regression was different from the line of identity; the magn itude of this discrepancy depends on the tracer recovery. When each uncorrected COZ production measurement was multiplied by the mean recovery of 72%, the regression line of the data was not significantly different from unity (Fig. 4B). The more rapid metabolic rate of resting children compared with adults can be explained by several factors. In homeotherms, a significant portion of resting metabolic rate is determined by the maintenance of body temperature. Children with a larger surface area-to-body mass ratio than adults, and therefore possibly a greater  4. CO2 production calculated from 13C02 washout data after administration of tracer (see text) as a function of ho2 measured directly breath by breath. A: estimation of CO, production from washout data not corrected for recovery (i.e., recovery assumed to be 100%). Note that true CO2 production is overestimated by tracer calculations. B: tracer-derived estimation of COP after correcting for recovery. Each value was multiplied by average recovery rate for all subjects (72%). Note close proximity of estimated CO2 values to line of identity.  5. CO2 production and stores normalized to body weight in children compared with adults. A: mean COz production was significantly higher in children compared with adults. This was not accompanied by an increase in CO, stores. B: there was no difference in normalized CO, stores between 2 groups. heat loss per kilogram than adults, might metabolize more quickly even at rest. In addition, we observed that children under resting conditions were more active (fidgeting, moving in place) than were most adults studied.
The present study also suggests that the differences in MRT observed between adults and children are not explained by body size or metabolic rate alone. First, there was no correlation between MRT and bodv mass within each group (Fig. 3). Second, there was no correlation between the MRT and the normalized metabolic rate (VcoJkg).
The 13C02 washout was used to estimate the metabolic rate. As noted in Fig. 4, there was a highly significant correlation between the uncorrected C$-production measured by tracer methodology and the VCO~ measured by gas exchange techniques. This correlation was observed despite the fact that the range of metabolic rates was quite small. We found that the tracer-derived values of CO2 production were invariably higher reflecting the incomplete recovery of the 13C tracer. The magnitude of unrecovered tracer determines the error in estimation of CO2 production because the latter is calculated assuming complete recovery of the tracer dose.
There was no difference in recovery between the adults and children (Fig. 2B), and we found no correlation between metabolic rate and recovery. The apparent ageindependent nature of the recovery and the high correlation between CO2 production estimated by tracer and the ho2 suggests that orally administered [13C]bicarbonate can, under the proper conditions, serve as a noninvasive measurement of metabolic rate. For example, if we used the empirically derived mean recovery of 72% as a correction factor, then the regression slope of the corrected tracer-derived CO2 production was indistinguishable from identity (Fig. 4B), and the error in estimating resting VCO~ would have been only 3 t 14% [error calculated from the mean (observed -predicted)/ observed].
There has been controversy surrounding the fate of the unaccounted for 13C. Several possibilities exist including the recycling or fixation of CO2 in intermediary metabolism (5,12), incorporation of tracer into metabolically inactive bone carbonates (18), or the incorporation of 13C into urea (15). In addition, it was shown with intravenous studies that radioactive CO2 was eliminated at a very high rate during the initial few seconds after injection during the first passage through the lungs (firstpass phenomenon) (10). In oral studies, eructation could also result in some tracer loss. But despite these pathways of irretrievable 13C loss, the mean recovery of 73% in children and 71% in adults compares favorably with the recovery obtained in intravenous studies in resting adults and babies (1,13,24).
To exclude methodological factors that might have contributed to the differences observed between children and adults, we further analyzed the data within the children. As noted, there was no significant difference between the children who fasted overnight and those who fasted for 4 h in time of peak DOB, MRT, and ho2. Similarly, no gender-related differences in CO2 dynamics and storage were observed.
In children the estimated resting CO2 stores per kilogram was slightly higher than that in adults, but the difference was not statistically significant. This observation does not substantiate our original hypothesis that CO2 stores in children are smaller than in adults. In interpreting these data, it must be recognized that the size of the CO2 stores depends on the metabolic rate, which was significantly (74%) higher (per kg) in children than in adults (Fig. 5. A and B ). In studies performed in adults using intravenous 13C and 14C, increases in metabolic rate (consequent to low-intensity exercise) did result in apparent increases of the size of rapidly exchanging COn stores (1,21). But in this study the difference in absolute metabolic rate between adults and children was small, and the magnitude of possible increases in CO2 stores due to increases in metabolic rate could easily be within the error imposed by intrasubject variability. Thus the mechanism of children's smaller CO2 storage capacity during exercise cannot be explained by smaller COZ stores under resting conditions. Rather, the mechanism must be related to COn transport dynamics associated specifically with the hemodynamic and metabolic response to increased metabolic rate.
The interpretation of the steady-state mass of COa measured by tracer-dilution techniques must be made with caution (8). Because the mass of COa is estimated by the product of MRT and VCO~, the mass measured by the classic "noncompartmental" calculation is not necessarily the total system mass. First, if tracer-estimated VCO~ is used without correction for recovery, the true ho2 will be overestimated.
This was not a problem in our study because we used VCO~ as measured by gas exchange. Second, MRT may be underestimated if any unrecovered label is directly eliminated from or trapped within kinetically slower ("deep") pools that are separate from the central pool. Finally, in systems like COabicarbonate in which several compartments exist, the site or sites of endogenous COn production relative to the site of tracer absorption may not be known (1). Because our calculation of total mass assumes that the site of endogenous COZ production is the same as the site of tracer absorption, the total mass estimated by tracer dilution may be in error. It was therefore encouraging to find that the values we obtained by tracer dilution (203 ml COJkg in adults and 222 ml COJkg in children) were surprisingly close to those obtained by Farhi and Rahn (11) using rebreathing techniques. In their study, the "total labile CO2 store" (i.e., excluding bone) was 237 ml COJkg body wt.
As noted, the sum of three exponentials is remarkably accurate in describing the washout curve of 13C02 or 14C02 after intravenous injection of tracer, as has been shown in studies done in rats, cats, dogs, human adults, and babies. Based on these data, a three-compartment model is currently used to explain the functional movement of CO2 throughout its exchangeable pools in the body (1,13,20). But the characteristics of the washout curve after oral administration were fundamentally different from intravenous studies. As noted above, the peak DOB in the oral studies occurred several minutes after administration, whereas the peak DOB occurs virtually instantaneously in intravenous experiments. The delayed peak represents the phase of absorption of tracer from the gastrointestinal tract into the circulating blood, and the difference between the MRTs obtained from oral and intravenous studies represents the mean absorption time of tracer. The difference in MRT [based on comparison of intravenous studies done in resting adults (1) with the oral data reported here] appears to be on the order of 7 min, suggesting that the mean time for absorption is only -10% of the MRT.
We reasoned that if the process of tracer absorption was similar in all subjects, then a generalized mathematical expression for oral tracer studies could be found and used to determine structural information about intercompartmental rate constants, pool sizes, and absorption rates (e.g., by deconvolution techniques). Although we were unable to find a single equation, data from individual subjects could be well fit by the two-exponential, three-exponential, or polynomial exponential model and this may be useful empirically for estimating AUC and AUMC in future studies. The variability of the absorption phase was reflected in part by the wide range of times at which the peak DOB appeared in the individual subjects. Obtaining compartmental information about COB-bicarbonate from oral data may require more detailed analysis of [ 13C] bicarbonate absorption kinetics.
In summary, we found that oral administration of [ 13C]bicarbonate can be used to obtain information about CO2 dynamics in adults and children. COa production can be estimated with reasonable accuracy; thus the 13C02 breath test may play a role in the noninvasive measurement of COa production in situations where devices for measuring gas exchange are not available or are too cumbersome to use. Finally, our data indicate that resting CO2 stores in children and adults are similar and that growth-related differences in storage capacity known to occur during exercise must be related to as yet undiscovered dynamic responses of the organism to increased metabolism.