INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

QUANTITATIVE GAS-EXCHANGE analysis remains one of the more difficult and challenging physiological measurements. Conventional instrumentation is confined to the measurements of oxygen consumption (O₂) and carbon dioxide production (CO₂) and performs poorly in high-oxygen environments (2, 5, 6, 9, 13, 16, 19). These limitations restrict its applicability. Additionally, indirect calorimetry only reveals a small part of the metabolic process. Some advances in the study of metabolism necessitate the ability to quantify other expired gases (i.e., ¹³CO₂) to explore pathways in organ function (8, 10, 14, 15, 17).

Cumbersome and generally limited to its comparison with other techniques, the Douglas bag technique has demonstrated the potential for greater accuracy than conventional systems and offers the opportunity to study the entire expired gas volume of innumerable species (3, 4, 12). We sought to devise an instrument capable of 1) improving the accuracy of conventional gas-exchange measurements over a broader range of clinical conditions, 2) providing the flexibility to study other aspects of metabolism related to the measurement of ¹³CO₂, and 3) using Douglas bag methodology to provide a system of automated gas-exchange analysis (SAGE).

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Two 60-liter vinyl nondiffusing gas-collection bags (no. 22619, Warren E. Collins, Braintree, MA) are suspended from a valve manifold and mounted within a 66 cm × 66 cm × 108 cm (width × depth × height) movable enclosure (Fig. 1). The gas-collection bags are modified with the addition of four gas ports near the corners of each bag. A pump (model VP0660, Medo USA, Hanover Park, IL) with a nominal flow rate of 25 l/min is used in conjunction with each Douglas bag's supplemental ports to implement the recirculation manifold (Fig. 2). This manifold mixes gases within 11 s, ensuring concentration uniformity throughout the bag before mass spectrometer sampling. To prevent condensation, the enclosure is heated to ~41°C. Circulation fans ensure uniform temperature distribution inside the enclosure. The gas-collection manifold is formed from 25-mm-ID polyvinyl chloride plumbing stock and associated fittings (elbows, T fittings, threaded adaptors). Large-bore (25-mm) solenoid valves (model NGB7B, Alcan Valves, Itasca, IL), labeled V₁-V₅, direct collected gas into the bags or to an evacuation fan. Valve V₂ acts as a bypass valve, venting the subject's exhalation to room air when there is no collection activity. A quadrupole mass spectrometer (RAMS-200, Marquette Medical Systems, Milwaukee, WI) programmed to measure O₂, N₂, ¹²CO₂, ¹³CO₂, Ar, and SF₆ is used for gas analysis. It is configured to discriminate between atomic mass units 44 (¹²CO₂) and 45 (¹³CO₂) with sufficient resolution and sensitivity to perform ratio measurements. Small valves, V₉-V₁₁ (Clippard EVO-3, Cincinnati, OH) route gas samples to the mass spectrometer from either the inspired gas or the Douglas bag contents. The gas is analyzed for 4 s, providing the average of 200 measurements for use by the system's equations (see Table 1). This device is portable and requires only 15 min of warm-up time to achieve its specified accuracy. Physical measurements, analog-to-digital conversions, valve actuation and sequencing, and communications with the RAMS-200 mass spectrometer are under control of a printer port computer interface (LPTek, Westbury, NY) and a standard personal computer.

View larger version (33K): [in this window] [in a new window]   Fig. 1.   Schematic of system of automatic gas-exchange (SAGE). V1, V3, V4, and V5 are valves in the path of collected gas; V2 is the bypass valve; V6-V8 are SF6 manifold valves; V9 and V10 are mass spectrometer sampling line valves; V11 is inspired gas sample line; V12 and V13 are mixing manifold valves. V and P denote vacuum and pressure sides of recirculation pump, respectively. T1, T2, and T3 are thermistors; mass spec, mass spectrometer; vac fan, evacuation fan; SF6, sulfur hexafluoride.

View larger version (24K): [in this window] [in a new window]   Fig. 2.   Schematic of convection manifold. Rapid mixing is achieved by recirculating the collected gas with high-output pumps. The pump can also be used to aide in bag evacuation (exhaust). There is one system for each collection bag. Table 2 outlines the error and precision against predicted measurements made with either the Deltatrac, MediGraphic, or the SAGE using the methanol-burning lung of Miodownik et al. (7).

An indicator-dilution gas system employing sulfur hexafluoride (SF₆) is used to determine the volume of gases collected in each bag. SF₆ (99.96%), from a supply tank located in the enclosure, pressurizes a steel dosing tank of known volume (1,590 ml) via valve V₈. The dosing tank is packed with copper wool to ensure rapid temperature equilibration. An absolute pressure transducer (142PC30A, Honeywell Microswitch Div., Freeport, IL) is used to determine the dosing tank pressure. During operation, the dosing tank is pressurized to ~1.5 atm as measured by the absolute pressure transducer. Valve V₆ or V₇ routes the pressurized SF₆ to either bag 1 or bag 2 via a centrally located port. By measuring the absolute ambient and tank pressures, and the tank and Douglas bag temperatures, it is possible to deliver a highly accurate volume of SF₆ into the Douglas bags. A typical SF₆ dose volume is ~750 ml. Cylinders containing 4.5 kg of SF₆, which liquefies at >2,068 kPa, provide sufficient indicator gas for ~900 measurements. The dilution of SF₆ in the Douglas bag is measured after dosing and again after collection to calculate the residual and total volumes, respectively.

The functional operations of the system are priming, evacuation, dosing, gas collection, analysis, and evacuation. These are interleaved between the two Douglas bags to permit continuous gas-exchange measurement. The total time required to perform one cycle of all functions is ~115 s/bag. The system is, therefore, programmed for a nominal 2-min collection interval with gas volumes reported in milliliters per minute. Under software control, alterations of the collection intervals can be made. The interconnection between the instrument and the subject consists of a face mask or a mouthpiece and nose clip and a set of nonrebreathing valves coupled to the front panel connector with a length of ventilator or pulmonary function tubing. A continuous flow hood could also be used. The exhalation port of a mechanical ventilator can be accessed to provide expired gas for collection.

Validation

Part 1. The system was validated by using a modified version of the quantitative funnel burner lung model of Miodownik et al. (7). The modification substituted a water jacket for the open pool of water used for cooling. This validation technique is rigorous and generates precise, adjustable O₂ and CO₂ while simultaneously producing saturated water vapor gas. Consequently, it closely reflects the clinical environment encountered during measurement on spontaneously breathing and mechanically ventilated patients. In addition, the accuracy and precision of the SAGE total volume accumulation measurement between 7 and 25 liters were tested by using standard water displacement.

Part 2. The accuracy of the SAGE was compared with that of the SensorMedics Deltatrac (Yorba Linda, CA) and the MedGraphics CPX (St. Paul, MN) gas-exchange instruments. These devices are representative of mixing-box and breath-by-breath techniques, respectively. Following manufacturer-specified calibration, these instrument were subjected to the validation protocol outlined in Part 1.

Part 3. The validation technique was expanded to test the ability of the SAGE to quantify small differences in ¹³CO₂ concentration in volumes of gas collected over time. This process simulated the collection of ¹³CO₂-enriched gas that results from the metabolism of ¹³C-enriched hydrocarbons. The abundance ratios of ¹³C to ¹²C (¹³C/¹²C; 1.0581 atom%) in both the batch of methanol burned and a tank of 100% CO₂ (1.10532 atom%) were verified by mass ratio spectrometry (Metabolic Solutions, Nashua, NH) and provided, respectively, reference and measurement values for the simultaneous validation of ¹³CO₂ production and ¹³C/¹²C enrichment. Methanol infused at 20 ml/h was burned for 2 h in the validation lung model to establish a baseline ¹³CO₂ production. The lung model was ventilated at a nominal 10 l/min with room air. After this baseline collection, the flame was extinguished, and the tanked CO₂ (abundance ratios of ¹³C/¹²C 1.0581 atom%) was introduced into the outlet of the lung model at a comparable rate (~185 ml/min) and collected by the SAGE for 2 h. The atom% difference between the ¹³CO₂ produced by burning methanol and the ¹³CO₂ of the tank provided a predetermined enrichment level. This was calculated to produce ~6 ml of enriched ¹³CO₂ over the collection interval. This quantity of enriched ¹³CO₂ is comparable to the expected recovery observed in ¹³C-labeled studies (14).

Statistics

The atom% difference between the two sources of ¹³CO₂ was evaluated by t-test. The results were expressed as means ± SD. The difference between the ¹³CO₂ atom% in tanked gas at each time collection and the mean of the ¹³CO₂ atom% of the reference methanol burn was calculated to generate a ¹³CO₂ enrichment volume for each bag collection of tank gas. The 2-h collection was calculated as the sum of ¹³CO₂ enrichment volume collected over 2 h. This value was compared with the calculated control from the known standard volumes and atom% of the test gas mixtures.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The results of O₂ and CO₂ at each set of experimental conditions for the SAGE, the SensorMedics Deltatrac, and the MedGraphics CPX are shown in Table 2. All data are expressed as mean error% with precision given in parentheses. The SAGE measured expired gas volume with a mean error of 1.1% (0.9%). The average O₂ error of the SAGE over FI_O2 of 0.21-0.80 was 2.4% (3.7). The average CO₂ error over this range was 0.8% (1.5). The SAGE O₂ error% was significantly lower than the other units at FI_O2 > 0.40. The SensorMedics Deltatrac average O₂ and CO₂ error was 11.4% (3.7) and 2.1% (0.9), respectively, over the range of FI_O2 studied. The MedGraphics CPX average O₂ and CO₂ error was 2.6% (4.5) and 1.7% (2.7), respectively, over a narrower FI_O2 range. The CPX was not tested at FI_O2 > 0.60 because of manufacturer assertion of lack of accuracy at that range. The Deltatrac and the CPX devices both exhibited increasing O₂ errors as FI_O2 was elevated. There was no difference in CO₂ accuracy between instruments. CO₂ readings for all devices were largely unaffected by FI_O2.

The ¹³CO₂/¹²CO₂ atom% measurements for the burning methanol and for the tanked CO₂, respectively, were 9.6 × 10³ ± 2.0 × 10⁴ and 1.0 × 10² ± 1.7 × 10⁴. The differential atom% between the two ¹³CO₂ gases was measured at 4.8 × 10⁴ with an error of +4.2% compared with the expected differential. The 2-h collection of ¹³CO₂ enrichment volume was 6.4 ml with an error of +3.4% compared with the predicted 2-h collection.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

As an indirect calorimeter, the average O₂ error of the SAGE was significantly less than the respective error for the Deltatrac and the CPX at FI_O2 > 0.40. At lower FI_O2, all three systems exhibited errors that were not significantly different from each other. The measurement of a varying gas flow is difficult and prone to multiple sources of error. The increasing inaccuracy of the Deltatrac and the CPX can be traced to the limitations of their flow-measuring devices. Because there is no correction of pneumotachometer flow signal for FI_O2, temperature, and CO₂ concentration, it is important to calibrate the instrument under conditions close to those prevailing during testing. Both instruments are initially and only calibrated at room air without correction for ventilation of gases at higher FI_O2. The expired volume of water vapor must also be carefully accounted for (22). Additional sources of error include the measurement of dynamically changing gas concentrations and their proper alignment with the flow signal over an acceptable frequency response range. Some of these systems can yield O₂ errors approaching 50-100 ml/min at elevated FI_O2 (1, 18).

The literature continues to compare the accuracy of modern gas-exchange techniques to the Douglas bag method (2-6, 9, 12, 13, 16, 19). The gas-collection bag contains the true mixed expired volume of gas expressed during the collection interval, independent of the flow pattern of the gas entering the bag, its collection site, or the activity of the subject. This is a major and obvious advantage noted by Douglas (4). The major factors that have made this technique cumbersome are ensuring thorough mixture of collected gas, determining the volume collected, withdrawing gas samples for analysis, and emptying the bags in preparation for their next use. To determine volume, the Douglas bag is often displaced into a spirometer (e.g., Tissot or gas meter) with corrections made for water vapor content (20). The literature refers to some of the problems in managing leaks, contamination with unwanted gases during bag manipulation, and other measurement errors during hyperoxic states (3). These functions have, traditionally, been done by hand. Although multiple bags can be collected and labeled for subsequent analysis, continuous measurement is unrealistic.

Our implementation of the SAGE addresses a number of the error sources associated with conventional gas exchange techniques and with manual Douglas bag methods. They are avoided by employing the two-bag automated collection system shown schematically in Fig. 1. By automatically interleaving, filling, analyzing, and emptying the Douglas bags contained within our instrument, a virtual infinite set of empty bags is made available, thus allowing continuous gas exchange measurement. The gas circulation system for each bag described in MATERIALS AND METHODS ensures rapid gas concentration equilibration before gas measurement. The SF₆ indicator gas dilution permits automation of the volumetric measurement and eliminates the need for the error-prone measurement of flow to determine volume. In the context of a 2-min collection cycle, SAGE measurements are performed on static volumes having stable gas concentrations at each stage of the collection process, allowing accurate gas concentration measurements. Advantage can be taken of this static situation to significantly reduce errors by averaging several hundred individual physical determinations.

The mass spectrometer automatically normalizes the sum of all of the gas species selected for measurement to 1.00. Since water vapor is not selected as a measured gas, its volume is disregarded as a component of normalization. Consequently, when the SF₆ indicator is diluted to determine bag volume, the component representing water vapor is conveniently ignored. Through automation, opportunities for leaks and contamination are minimized, and human error is eliminated. Our results indicate that the SAGE O₂ error increases to 5.8% at an FI_O2 of 0.8. This is a result of the N₂ concentration measurement accuracy required by the Haldane transformation as the FI_O2 is increased and the accuracy limitation of the RAMS-200 mass spectrometer (21). However, the SAGE is not particularly limited or intimately tied to the RAMS-200 mass spectrometer. Indeed, we fully anticipate making use of advances in mass spectrometry and gas analysis instrumentation as they become available. The indicator dosing system is not constrained to SF₆ gas. The instrument is usable with virtually any indicator gas that is accurately determined by the gas measuring system.

The SAGE is restricted by the maximum gas volume that can accumulate during the collection interval (~60 liters). Although this system cannot respond to instantaneous changes in gas exchange, the 2-min cycle time for each bag collection is still short enough to allow observation of many dynamic events. This is comparable to the operation of both mixing-box and breath-by-breath instruments. The large volume associated with mixing chambers, i.e., Deltatrac, can reduce the resolution of the minute-by-minute measurements and yield substantially slower moving averages (1). Some breath-by-breath instruments, i.e., CPX, provide a moving average of eight or more breaths, obscuring instantaneous changes.

Having a true mixed expirate permits the expansion of metabolic gas exchange to include species other than O₂ and CO₂. It offers the intriguing possibility of investigating aspects of metabolism via gas exchange in addition to via indirect calorimetry. The ¹³CO₂ validation test demonstrates the ability of the SAGE to measure the ¹³CO₂ component of CO₂ with sufficient accuracy to identify changes in the ¹³C/¹²C enrichment ratio and the excreted volume of enriched ¹³CO₂. The ability to discriminate between small differences in atom% of ¹³CO₂ allows the determination of enrichment volume. This system permits analysis of expressions of metabolic activity that are traceable to the conversion of ¹³C-labeled substrate to ¹³CO₂ such as gastrointestinal absorption; fat, carbohydrate, and protein metabolism; and liver function (8, 10, 14, 15, 17). After administration of the labeled metabolite, the excretion of expired ¹³CO₂ is quantitated by collecting numerous aliquots for later gas study. This process, time consuming and fraught with errors, marginally approximates the excretion curve. Conventional ¹³C-labeled breath tests often approximate subjects' CO₂ production on the basis of either body weight (9 mmol · kg¹ · h¹) or body surface area (5 mmol · m² · min¹). Such crude estimates of CO₂ are a significant source of calculation error (14). The aliquot system is costly and time consuming, and it fails to provide a complete portrait of excretion. Without the assumptions inherent in the conventional sampled-aliquot technique, the SAGE can offer a more accurate and inexpensive alternative for ¹³CO₂-substrate breath tests. This application could provide sequential collections of exhaled gas in a quantitative time course.

In conclusion, we have developed a new automated gas-exchange measurement instrument based on the Douglas bag method. The SAGE can reduce errors at high FI_O2. We have demonstrated the ability to quantify small volumes of enriched ¹³CO₂. The ability to perform continuous, unattended, quantitative analysis of expired gas could permit the study of the dynamics of metabolism over a multitude of conditions within a markedly reduced time frame.