INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

BREATH STUDIES HAVE BEEN PERFORMED with a wide spectrum of ¹³C-labeled substances. The metabolism of virtually any organic substance in the human organism can be estimated with this technique. After a substance is labeled with ¹³C, the expiration of ¹³CO₂ correlates to the oxidation or turnover of the substance. Common experiments include estimation of the oxidation of exogenous glucose (20, 24, 28) or fatty acids (26, 31). One difficulty in these metabolic studies consists of the body's stores of CO₂, which are relatively large. It has been reported that the production of ¹³CO₂ is delayed by the CO₂ pools in various tissues before CO₂ is expired from the mouth (21, 27, 33). Studying the compartmental distribution of exogenous ¹³CO₂ is a prerequisite for the estimation of this retardation.

Other investigations using ¹³C labeling have focused on the physiology of the bicarbonate pools. The compartmental distribution of CO₂ in the body at rest and during exercise could be described using intravenous bolus injections of [¹³C]bicarbonate with subsequent measurement of breath enrichment (3, 37). In other applications, the metabolic rate has been estimated by the [¹³C]bicarbonate washout characteristics (2, 3, 6, 19, 37).

A consistent finding in all these studies is an incomplete occurrence of a given ¹³C dose in the expired air. Recovery of administered [¹³C]bicarbonate has been calculated to be 50-90% (1, 3, 22, 27, 32, 36). It is likely that this phenomenon affects studies of substrate oxidation in the same way. The physiological reasons for this irreversible "loss" of ¹³CO₂ are described as via urine, sweat, or urea (3, 16) and the transfer into bone (15, 16) or macromolecules (4, 14).

A first-pass effect of the venous blood through the lungs was supposed by Armon et al. (2), Barstow et al. (3), and Drury et al. (8) as an additional effect in the reduction of the administered ¹³C dose. Nevertheless, this effect could not be shown directly. A study with [¹³C]bicarbonate infusions administered to dogs could not show any effect of the passage of the substance through the lungs (7). In this study, a previously described breath-by-breath measurement of ¹³CO₂ in the expired air (30) is used to depict this first-pass effect. The high temporal resolution of a breath-by-breath system provides a direct observation of the washout curve instantaneously after injection of the [¹³C]bicarbonate. To estimate the exactness of the breath-by-breath method, compartmental parameters from the washout curves were calculated as described previously (3, 17).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Eight healthy men were examined. All subjects were free of metabolic or cardiovascular abnormalities, lean (9.2 ± 3.4% body fat by skinfold measurements), and nonsmokers (Table 1). The subjects granted written consent to participate in the study. Review by the Ethics Commission of the University of Tuebingen brought no objections to the performance of the study.

Experimental procedure. An intravenous bolus of 0.0125 g/kg body wt [¹³C]bicarbonate (NaH¹³CO₃; CIL, Andover, MA; purity 99% ¹³C) in an 8.4% water solution was administered in each test. The solutions were prepared and sterile filtered within 5 h before each test. To prevent evaporation of bicarbonate, gastight bottles with sealed plugs were used. The bolus injection was given into an antecubital vein of each subject's right arm with an automated volumetric infusion pump (model 591, Ivac) within 1.0 s. Gas analysis was started 5 min before the [¹³C]bicarbonate was administered. The measurement before bicarbonate administration served to record the baselines of the ¹³CO₂-to-¹²CO₂ ratio and spirometry. The moment of bolus injection was defined as time 0. Gas analysis ended 30 min after the bicarbonate injection. Seated rest and two different exercise intensities on a cycle ergometer (Excalibur, LODE) were tested. The bicarbonate washout was determined at a constant workload relative to 40 and 75% of each subject's maximal O₂ consumption (O_2 max). Exercise began 5 min before the bicarbonate injection. The subjects rested 2 days between the tests to allow for tracer washout.

Measurement of gas volumes. The total respired gas volume was determined using a light-weight low-impedance turbine flowmeter (Triple V transducer, Mijnhardt, Nijmwegen, The Netherlands) in BTPS conditions. The dead space of this flowmeter unit was 125 ml. The gas volume was integrated via software from the flow signal. The device has a linear flow-to-signal characteristic, which was checked before the tests by application of low to supraphysiological airflows with the calibration pump. The momentum of the turbine had no impact on the gas volume results. The flow/volume-measuring unit was calibrated before each individual test with a 3-liter calibration pump.

Respiratory gas fraction analysis. The respiratory gas fractions were determined using respiratory mass spectrometry (model AMIS 2000, Innovision, Odense, Denmark, with a Quadrupole QMA 20, Balzers, Balzers, Liechtenstein). The mass spectrometer was calibrated before each individual test by means of a two-point calibration with a zero offset with all valves closed. The calibration gas should contain 1% ¹³CO₂, 1% Ar, 5% ¹²CO₂, 16% O₂, and 77% N₂. The ¹³CO₂-to-¹²CO₂ ratio in the calibration gas (0.2) corresponds to the expected maximum values of the isotope ratio in the expired air during the tests. The actual concentration of each gas was validated gravimetrically with a relative accuracy of ±0.01%. We used these validated values in the calibration procedure. The recording frequency was 11.1 Hz, corresponding to a cycle duration of 90 ms. The respired gas fractions for O₂, ¹²CO₂, ¹³CO₂, N₂, and Ar were measured in each cycle. A delay of 0.28 s between the gas concentration signal and the signal for gas flow was taken into consideration for the breath-by-breath calculations.

View larger version (18K): [in this window] [in a new window]   Fig. 1.   13CO2-to-12CO2 ratio in the course of one breath cycle without previous administration of [13C]bicarbonate. Top: flow signal for inspiration (In) and expiration (Ex). With advancing duration of expiration, analyzed gas comes from deeper in the lung and will be less diluted than gas from the functional dead space (***). Middle: course of the CO2 concentration. Bottom: course of the 13CO2-to-12CO2 ratio. The average values of the plateau phase during the expiration were used as end tidal (*). The time delay between the flow and gas concentration signals (**) was taken into consideration for the breath-by-breath calculations.

Determination of the detection limit. To check the stability and reliability of the gas-measuring method, the detection limit (DL) was calculated as follows

(1)

Background concentration of ¹³CO₂. ¹³C can be found in various concentrations in the natural environment (13, 35). The source of daily nutrition determines the ¹³CO₂-to-¹²CO₂ ratio without additional ¹³C enrichment. This individual baseline is therefore subtracted from the measured isotope ratio to obtain the net changes in this value [delta over baseline (DOB)]. DOB is equivalent to the increase in the specific concentration of exogenous [¹³C]bicarbonate in the system (2, 3).

(2)

Recovery. Only a part of the exogenous amount of [¹³C]bicarbonate can be found in the expired air (2, 10, 15, 29). This is due to an unobserved loss of ¹³CO₂ from the central compartment (k_v1; Fig. 2). Equation 3 gives recovery as the relationship between the cumulative volume of the expired ¹³CO₂ (exV¹³CO₂) and the total dose of [¹³C]bicarbonate (D₀)

(3)

where exV¹³CO₂ is in liters and D₀ is in moles. The constant K (=0.0446) is for the conversion between moles and liters. As described in earlier studies, the values for recovery can be used for a correction of the calculations for ¹³CO₂ distribution (2, 9, 25).

View larger version (13K): [in this window] [in a new window]   Fig. 2.   Schematic illustration of a mammillary 2-compartment model for distribution of [13C]bicarbonate given in a dose D0. kij, Rate constant for transfer from compartment j to compartment i. Clearance of [13C]bicarbonate occurs only from the central compartment 1 (k01; not shown). Although some of this loss can be measured via the washout curves at the mouth (kB1), another amount remains unobserved (kL1). [Modified from Barstow et al. (3).]

(4)

Compartmental kinetics of bicarbonate. The distribution of ¹³CO₂ and [¹³C]bicarbonate can be described with a linear multicompartmental mammillary model (3, 17, 27). Figure 2 shows the two-compartment model that was used for bicarbonate distribution in this study. It is presumed that bicarbonate entry and irreversible loss of CO₂ are only via the central compartment 1. The steady-state quantity of CO₂ in compartment i is described by Q_i. The rate constant for the transfer from compartment j to compartment i is k_ij. For instance, k₂₁ is the rate constant for the transport from the central compartment 1 to the peripheral compartment 2; k_A1 is the transfer rate from compartment 1 to environmental air. The rate constant for unobserved loss of bicarbonate from compartment 1 is k_L1, which has been described as "nonrespiratory" in earlier studies (3). The total rate constant for the CO₂ flow from the central compartment (k₀₁) is k_B1 + k_L1. DOB stands directly for the ¹³CO₂ enrichment in the central compartment (3, 7, 15). Models with more than two compartments are built analogously (18). After administration of D₀ at time 0, the course of the concentration in compartment 1 is measured by the ¹³CO₂-to-¹²CO₂ ratio or DOB, respectively.

Washout curves. Bicarbonate and CO₂ are distributed quickly and homogenously after administration into the central compartment. The impulse of D₀ is answered by the washout course of DOB. With two compartments, the sum of n = 2 exponentials describes this washout curve (15, 19). An empirical model without additional constants was used for this study. This model has been evaluated as best fitting for bicarbonate washout kinetics (3) (Eq. 5)

(5)

The macroparameters of the exponential function, A_i and _i, are calculated from the course of the discrete data pairs by way of an iterative procedure. The parameter of each iteration was evaluated by using the Levenberg-Marquardt maximum neighborhood algorithm (12, 23).

AUC. With the use of the macroparameters, the course of the washout curves can be extrapolated to infinity. Subsequently, the total amount of expired ¹³CO₂ can be estimated from this extrapolated curve as the AUC (Eq. 6)

(6)

Identification of the microparameters. The steady-state constants k_ij and Q_i can be calculated from the washout curve parameters A_i and _i (3, 17). Compartment 2 does not have a substance leak (k₀₂ = 0); thus all microparameters can be explicitly determined for the two-compartment model used in this study.

(7)

(8)

First-pass effect. Some [¹³C]bicarbonate cannot be measured via the DOB washout curves after bolus injection. A first-pass effect is considered to be one of the factors for this unaccounted loss of [¹³C]bicarbonate. We progressively excluded the data at the beginning of the washout curves to make this effect apparent. The resulting two-exponential fitting was evaluated by an F test with every step (18). The best of these fits was used as the reference washout curve. This curve was extrapolated to time 0 and subtracted from the original measured washout curve. The area under the peak (FP_area) was taken as substance loss via the first-pass effect (hatched areas in Fig. 3). The first-pass effect as a percentage of the total D₀ was calculated as FP% analogously to Eq. 4 by using FP_area instead of AUC. D₀ was corrected by FP% in the compartmental calculations. Finally, the fraction of FP% in the tracer loss was calculated using the recovery values.

View larger version (21K): [in this window] [in a new window]   Fig. 3.   Washout curves after administration of a venous bolus of [13C]bicarbonate shown as specific 13CO2 enrichment [delta over baseline (DOB)] of the exhaled air breath-by-breath. Each data point represents the DOB of one single breath. The bicarbonate was administered at time 0 at rest and during exercise at 40 and 75% of maximal O2 consumption (O2 max). Bottom: total measurement duration. Top: same curves at shown at bottom for first 3 min. Hatched areas under the excessive peaks are interpreted as first-pass effect in the lungs. Dashed lines are the result of the best fit for the sum of 2 exponentials (see Eq. 6).

Statistics. Statistical calculations and fittings were performed using JMP (SAS Institute, Cary, NC) and KaleidaGraph (Abelbeck) software on a personal computer (Apple Macintosh).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The measuring characteristics for the mass spectrometry unit are listed in Table 2.

An excessive first peak in the isotope ratio could be observed instantly after [¹³C]bicarbonate administration in all subjects, at rest and during exercise. FP_area became smaller with increasing workload (Fig. 3) and was 2,373.5 ± 819.2 DOB at rest, 1,208.4 ± 281.3 DOB during moderate exercise (40% O_2 max), and 963.3 ± 301.1 DOB during intensive exercise (75% O_2 max). All means were significantly different (P < 0.001). However, the higher the workload, the higher the relative loss of [¹³C]bicarbonate was via this first peak. At rest, the first-pass peak accounted for the loss of 4.01 ± 1.1% of the total administered bicarbonate dose. During moderate and intensive exercise, significantly higher values were reached (Table 3).

With the use of Eq. 3, an average recovery of 61.4 ± 14.6% was calculated for resting conditions, but with a tendency to higher values during moderate (64.0 ± 8.9% at 40% O_2 max) and intensive exercise (71.0 ± 10.1% at 75% O_2 max). These recovery values show that the first-pass effect was responsible for 11.32 ± 6.1% of the unaccounted loss of bicarbonate at rest, 33.9 ± 9.4% during moderate exercise, and 58.2 ± 14.0% during intensive exercise. All means were significantly different (P < 0.001).

Beyond the first peak, the course of the washout curves could be fitted with the sum of two exponentials (Eq. 6). The fit resulted in a mean correlation coefficient of 0.997 ± 0.002 for all tested subjects. The use of a three-exponential model did not lead to a better fit at rest or with exercise (r = 0.980 ± 0.027 for all subjects). Table 4 shows the individual macroparameters for these curves.

The rate constants k_ij for the transfer of the substance between the compartments and the related quantities Q₁ and Q₂ are shown in Table 5. The CO₂ clearance could be derived from Eq. 8 (k₀₁Q₁) and is given in Table 6 compared with the spirometrically measured CO₂. There were no significant mean differences between CO₂ and compartmental CO₂ clearance at rest or during moderate exercise. During intensive exercise (75% O_2 max), the spirometric values for the CO₂ were significantly higher than the calculated values.

View this table: [in this window] [in a new window]   Table 6.   Average CO2 from spirometric measurements and from compartmental analysis

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The main finding of these experiments is the steep peak of the breath ¹³C enrichment directly after intravenous administration of [¹³C]bicarbonate. This early phase of the washout curves cannot be included satisfactorily into the established multiexponential distribution model for bicarbonate. We assume that this "nonfitting" segment corresponds to a first-pass effect of the ¹³C substance in the lungs. This results from a high bolus concentration of the substance through the pulmonary system for the first subsequent circulation cycle after administration.

The resulting transfer constants and quantities of the bicarbonate distribution are within the range of other studies (2, 3, 11, 15, 27, 36). In addition, the CO₂ clearance rate from the compartmental calculation does not differ from the values measured by our spirometric system. This indicates that the chosen breath-by-breath method shows the course of the ¹³C enrichment correctly. Admittedly, the fitting does not apply to a third exponential, as in the study of Barstow et al. (3). The shorter time span of our experiment could prevent the transfer to the third compartment from becoming evident.

A first-pass effect through the lungs is supposed as one factor for the unaccounted loss of the labeled substance in ¹³C studies (2, 3, 8). However, such an effect could not be shown directly in the past because of the low temporal resolution of the methods. Single-breath sample entrapment is very time consuming because of the need to dry and purify the individual gas samples before measurement.

The detection characteristics of ¹³CO₂ (Table 2) meet the requirements for the performed breath-by-breath application. This is especially true during the first minutes of the test procedure, when the first-pass effect occurs. The scattering of the ¹³CO₂-to-¹²CO₂ ratio increases slightly with advancing test duration and decreasing activity of the labeled substance in the system. This scattering is presumably caused by the biological variability and the increase in the signal-to-noise ratio. However, the results of the fitting procedure are not influenced by the scattering, as with a systematic error caused by the equipment and the experimental environment.

FP_area could be estimated and compared with the given total dose of the labeled substance. The result of ~4% of the total bicarbonate dose illustrates the low impact of the first-pass effect on isotope recovery at rest. Earlier studies report an isotope recovery of 50-90% (1-3, 5, 22, 36). Given an average value of 61%, the first-pass effect determines only about one-tenth of the total tracer loss at rest.

As reported in other studies (3, 34), bicarbonate recovery increased with exercise. This may be due to a shortening of slower CO₂ distribution processes to deeper compartments in favor of CO₂ expiration. At higher exercise intensity, an additional amount of CO₂ is exhaled as "nonmetabolic CO₂" via bicarbonate buffering. This contributes to a larger tracer recovery. Furthermore, the proton buffering by bicarbonate might be the cause of the higher CO₂ measured with the spirometer than with the dilution calculation during intensive exercise (Table 6). This indirectly confirms that tracer dilution is the result of only the "metabolic" portion of CO₂ from substrate oxidation (6).

A higher cardiac output during exercise with shortened lung transit time is assumed to negatively impact the first-pass effect. Indeed, as shown in Fig. 3, FP_area decreases with increasing work intensity. However, although the "unobserved loss" of tracer decreases in total, the relative influence of the first-pass effect increases markedly with exercise. With our recovery values taken into account, a mean of ~30% of the given bicarbonate was not exhaled during the intensive exercise. This means that the first-pass effect could be responsible for up to two-thirds, or even more, of the lacking bicarbonate under these conditions.

Interestingly, the application of an infusion (and not a bolus) of labeled bicarbonate failed to show a significant first-pass effect in another study (7). Downey et al. (7) found no difference in the pulmonary expiration between venous and arterial application of bicarbonate. Consequently, the higher substance concentration leads to the mentioned first-pass effect only due to a bolus administration. This is confirmed by a meta-analysis on 34 human bicarbonate studies, which showed statistically significant higher recovery values for infusion than for bolus administration (22). This difference might be partly caused by the described first-pass effect.

In conclusion, the illustrated first-pass effect must be considered particularly in kinetic impulse studies with venous bolus injection of [¹³C]bicarbonate. Because of the much slower metabolism and distribution, other ¹³C-labeled substances, such as [¹³C]glucose, will not be considerably affected, even if given intravenously. Studies of substrate oxidation, as with labeled glucose or amino acids, may not be affected by any first-pass effect, especially after "priming" of the bicarbonate pool (1).

Nevertheless, some studies have been performed that relate the influence of the bicarbonate distribution to other ¹³C experiments with regard to the oxidation of labeled substances (27). Our results suggest that bicarbonate characteristics cannot be applied directly for corrections in glucose studies, leaving a first-pass effect unconsidered. Especially during exercise, the first-pass effect leads to an overestimation of D₀ for the following washout calculations. With a first-pass loss of ~20%, for instance, the related dilution calculations for CO₂ and the size of the compartments (Q_n) will be overestimated by this 20%. Other pathways for losing ¹³CO₂ become rather insignificant with increasing exercise intensity. On the other hand, this allows for an initial correction of the tracer dose with the use of the absolute recovery values as performed in previous bicarbonate kinetic calculations (3). However, in resting conditions, only a small amount of the initial tracer dose should be corrected corresponding to the first pass (Table 3). It could be that the other part of the unobserved tracer loss acts as a function over time. Thus it is likely that the calculation of kinetic distribution parameters is more uncertain at rest than during exercise.