Monitoring Gastric Mucosal Carbon Dioxide Pressure Using Gas Tonometry: In Vitro and in Vivo Validation Studies

We used a metal container filled with 150 ml distilled water as an equilibration chamber to perform in vitro tonometry. The gas source was provided by a mixture of nitrogen and carbon dioxide, introduced into the chamber at a nominal flow rate of 5 l/min. We adjusted the gas mixture to maintain the P^CO2of the bath (Pbath^CO2) at approximately 40 mmHg or 80 mmHg, as desired. The container was placed in a larger water bath maintained at a constant temperature of 37 [degree sign] Celsius.

A first standard tonometry nasogastric tube equipped with a semipermeable Silastic balloon (TRIP, NGS catheter; Tonometrics, Helsinki, Finland) was placed in the bath and connected to an automated gas analyzer (Tonocap; Datex, Helsinki, Finland). This system pumps room air into the balloon and aspirates the sample after a predefined equilibration time of 10, 15, 30, or 60 min. The P^CO2in the aspirated air is measured by an infrared sensor. With each measurement cycle, 8 ml air is inflated into the balloon and aspirated after a set time. The first 2 ml of the sample is discarded, because this corresponds to the dead space of the system (including catheter, connecting tube, and tube in the Tonocap analyzer). The P^CO2is measured in the final 6 ml of aspirate. Thirty seconds later the balloon is reinflated with the same 8-ml sample and a new measurement is obtained. The entire sequence of inflation, aspiration, P^CO2measurement and reinflation is fully automated.

A second standard tonometry nasogastric tube was plunged into the bath, but this time it was filled with saline. It was prepared according to the manufacturer's instructions and filled with 2.5 ml saline solution. After a predetermined equilibration time, 1 ml of dead space volume was aspirated from the catheter and discarded. The remaining saline solution was then aspirated and analyzed for P^COsub 2 (ABL 500; Radiometer, Copenhagen, Denmark).

We first studied the effect of changes in equilibration times on the P^CO2measurements obtained by the saline and the gas techniques. We stabilized the Pbath^CO2at about 40 mmHg and altered the equilibration times as follows: for each measurement technique, three gas P^CO2measures were obtained with an equilibration time of 10 min, followed by six measures with an equilibration time of 15 min, two with an equilibration time of 30 min, and two with an equilibration time of 60 min. The equilibration times were then progressively decreased in a similar manner: two measures with an equilibration time of 30 min, six with an equilibration time of 15 min, and three with an equilibration time of 10 min. The entire procedure lasted 8 h. The next day the process was repeated for a Pbath sub CO²of 80 mmHg using the same tonometry catheters. The gas measurements were fully automated, with the operator needing to alter only the equilibration time of the gas device. The saline tonometry measurements were performed with the tonometry catheter in place, with only emptying and refilling the balloon with saline solution. Thus 24 measures for each level of Pbath^CO2and for each measurement technique were recorded. Every 30 min, the Pbath^CO2was checked by analysis of a sample of bath water (Radiometer ABL 500 gas analyzer), and this value was used as the reference for the saline and gas P^CO2measurements obtained during the preceding 30 min. These in vitro manipulations were performed five times using a new pair of catheters for each manipulation.

The biases of the automated gas analyzer and the saline tonometer were defined as the mean of the differences between Pbath^CO2and Pg^CO2, and between Pbath^CO2and Ps^CO2, respectively. The precision is the standard deviation of the bias. Data are expressed as mean +/- SD.

Using an identical in vitro system, we compared the response times of gas and saline tonometry during acute changes in Pbath^COsub 2. After ensuring that the Pbath^CO2was stable at about 40 mmHg, we abruptly increased PbathCO²to approximately 80 mmHg by altering the gas mixture. During the next 5 min, the Pbath^CO2was determined every minute by analyzing a direct sample of bath water (Radiometer ABL 500 gas analyzer). We used the shortest equilibration time of the gas analyzer that was provided by the manufacturer (10 min) and the same equilibration time for the saline tonometer. We repeated these measurements after an abrupt decrease of the Pbath^CO2from 80 to 40 mmHg. We did the entire procedure five times. The response time was defined as the time needed to reach 95% of the change in P^CO2, for an abrupt change in Pbath^CO2.

For this in vitro part of the study, we first used uncorrected raw P^CO2measures obtained by the saline tonometry. Then we corrected these P^CO2measures by multiplying the raw P^CO2values by the correction factors provided by the catheter manufacturer, which depend on the length of time the saline solution remains in the tonometer balloon (correction factor 1.62 for equilibration time of 10 min; 1.44 for 15 min; 1.24 for 30 min; 1.13 for 60 min). We also calculated our own in vitro saline tonometer correction factors for each equilibration time by dividing the mean Pbath^CO2value by the corresponding mean Ps^CO2value obtained at the corresponding equilibration time. We finally compared our correction factors to the manufacturer's values.

After obtaining approval of the ethics committee of Erasme University Hospital, two nasogastric tonometry catheters (Tonometrics) were inserted together in seven sedated, mechanically ventilated, and hemodynamically stable critically ill patients during a study period of 6 h. Three patients had septic shock due to cerebral abscess (n = 1) or bronchopneumonia (n = 2), two patients had acute respiratory distress syndrome due to peritonitis (n = 1) and pulmonary graft rejection (n = 1), and two patients had intracerebral bleeding. There were five men and two women aged 42-80 yr (mean, 62 +/- 13 yr). All patients were mechanically ventilated, sedated (with midazolam and morphine), and paralyzed with pancuronium. All patients were invasively monitored with a pulmonary artery catheter (Swan Ganz catheter 93A-131-7F; Baxter Healthcare, Irvine, CA) and an arterial catheter. Every 30 min throughout the study, thermodilution cardiac output was measured (computer 9520 A, Baxter Healthcare) by successive injections of at least five boluses of 10 ml cold (< 8 [degree sign] Celsius) 5% dextrose in water, via a closed system (CO-set system, Baxter Healthcare). Each patient had been treated with H²receptor blockers (150 mg ranitidine given intravenously each day), with no enteral feeding for at least the previous 6 h. In each patient, the position of the two nasogastric catheters was checked radiographically.

The first catheter, dedicated to saline tonometry, was prepared according to the manufacturer's instructions and filled with 2.5 ml saline solution. After an equilibration time of 30 min, 1 ml dead space volume was aspirated from the catheter and discarded. The remaining saline solution was aspirated and analyzed for P^CO2(Radiometer ABL 500 gas analyzer). The adjusted saline P^CO2(Ps^CO2) was obtained by multiplying the measured P^CO2by the correction factor obtained in our in vitro study for an equilibration time of 30 min.

The second catheter, dedicated to gas tonometry, was connected to an automated gas analyzer (Tonocap, Datex). After balloon filling with air in a closed circuit (see previous), the gas P^COsub 2 (Pg^CO2) was automatically measured by infrared spectroscopy after pre-established equilibration times of 10, 15, or 30 min.

In each patient, 24 Pg^CO2measures were performed successively, in the following order: six measures with an equilibration time of 10 min, four measures with an equilibration time of 15 min, four with an equilibration time of 30 min, four with an equilibration time of 15 min, an six with an equilibrationd time of 10 min. The Pg^CO2values obtained during a period of 30 min were compared with the simultaneous Ps^CO2value obtained after the same 30 min of equilibration.

The effects of acute changes in Pa^CO2on Pg^COsub 2 were also studied in one sedated, paralyzed, and mechanically ventilated patient with acute respiratory failure as part of the study of respiratory mechanics under a FI^O2of 100%. The gastric mucosal P^CO2was measured by saline tonometry (Ps^CO2) with an equilibration time of 30 min and by gas tonometry (Pg^COsub 2) with an equilibration time of 10 min, and the arterial pressure of carbon dioxide (Pa^CO2) was measured every 10 min using a blood gas analyzer (Radiometer ABL 500).

Statistical evaluation included linear regression and a Bland and Altman analysis. [4] 

The Pbath^CO2remained stable during all the studies at 40.2 +/- 0.7 mmHg (lower level) and 76.4 +/- 0.8 mmHg (higher level), respectively. There was a close correlation between Pbath^CO2and Pg^CO2values for each level of Pbath^CO2and for each gas equilibration time. For the lower level of Pbath^CO2and for an equilibration time of 10 min, the bias and the precision of the gas tonometer were -0.3 mmHg and 0.7 mmHg, respectively. Table 1shows the bias and precision for the two Pbath^CO2levels and different equilibration times. Increasing the equilibration time of the gas analyzer did not improve the precision or decrease the bias of the Pg^CO2measurements.

(Table 2) shows the in vitro bias and precision of saline tonometry using raw uncorrected P^CO2values in the same conditions. For the lower level of Pbath^CO2and for an equilibration time of 10 min, the bias and the precision of the saline tonometer were 11.2 mmHg and 1.4 mmHg, respectively. The bias was greater for saline than for gas tonometry, especially for short equilibration times (10 and 15 min). Table 2also shows the in vitro bias and precision of saline tonometry in the same conditions, but after correction of the saline P^CO2measurements by the correcting factor provided by the catheter's manufacturer. Our own saline tonometer correction factors, extrapolated from the in vitro study, are shown and compared with the manufacturer's data in Table 3.

Acute changes in Pbath^CO2were achieved in less than 3 min. In this situation, Pg^CO2obtained after an equilibration time of 10 min was closer to the final steady-state Pbath sub CO²than the Ps^CO2but neither reached 95% of the change in P^CO2(Table 4). Reaching this value required two periods of 10-min equilibration time with both techniques.

All patients were hemodynamically stable and had no cardiac output changes greater than 5% during the study. There was a close correlation between Pg^CO2and Ps^CO2for each gas equilibration time, with the following regression coefficients: equilibration time of 10 min, r²= 0.986 (Figure 1); 15 min, r sup 2 = 0.967; 30 min, r²= 0.946. A Bland and Altman analysis was performed for each gas tonometry equilibration time. The bias (+/- 2 SD) was 0.1 +/- 6.8 mmHg for an equilibration time of 10 min (Figure 2), -2.1 +/- 11.4 mmHg for an equilibration time of 15 min, and -5.1 +/- 11.4 mmHg for an equilibration time of 30 min.

When each patient was included only once, analyzing only the Pg^CO2value obtained at the end of the equilibration time for the Ps^CO2, the correlation coefficients were similar (r²= 0.992, data not shown).

In the patient in whom hypercapnia developed acutely, there was a close correlation between Pa^CO2, Ps^CO2, and Pg sub CO²(Figure 3).

First our study demonstrates first that in vitro P^CO2measurements obtained by gas tonometry are reliable. Second, due to the rapid carbon dioxide diffusion in gas, these measurements do not require an equilibration time of more than 10 min. In fact, the accuracy of gas tonometry was not greater with prolonged equilibration times. The great bias between the in vitro Pbath^CO2value and the saline tonometry P^CO2measurements, especially for short equilibration times (10 and 15 min), is due to the incomplete equilibration of saline solution with environmental P^CO2after these short equilibration times. Factors have been developed in in vitro and in vivo studies to correct for this incomplete equilibration if the measurement time, as applied in many studies, is much shorter (+/- 30 min) than the complete equilibration time. In vitro saline tonometry biases remained greater than those of gas tonometry, even after correction of raw saline P^CO2values by the manufacturer's correction factors. This difference may be due to the correction factor recommended by the manufacturer. This measurement bias for short equilibration times may be clinically compromising if we consider that a normal P^CO2gap value, as defined by the difference between gastric mucosal P^CO2and arterial P^CO2, is less than 6 mmHg. We derived our own scale of correction factors from the in vitro study and obtained values that were different from those of the manufacturer, especially for short equilibration times. The differences between our correction factors and the manufacturer's values could be due to use of different blood gas analyzers. Therefore, use of such a correction factor can amplify the P^CO2measurement error. Equilibration times of less than 30 min, which need the greatest correction factors, therefore cannot be recommended for saline tonometry. The in vivo study shows that P^CO2measurements by gas tonometry correlate well with those obtained by saline tonometry and again do not require an equilibration time of more than 10 min. In a recent study in pigs, Knichwitz et al. [5]described a fiberoptic P sub CO²sensor that can determine intramucosal gut P^CO2in a precise and reliable manner. Their in vitro study indicated that the response time of their P^CO2sensor was identical to our gas device.

The major advantage of gas tonometry is that balloon filling with air rather than saline allows a fully automated procedure that avoids saline manipulations, which are time consuming and potential sources of error. The proper interpretation of Pg^CO2requires the recognition of two important principles. First, Pg^CO2is directly influenced by Pa^CO2, so that changes in ventilatory status can alter Pg^CO2in the absence of any change in regional blood flow. This observation was made in 1959 when a group of Hungarian physicians proposed the measurement of gastric P^CO2in lieu of arterial P^CO2during mechanical ventilation. [6]Similar observations were reported more recently in animals. [7]We had the opportunity to document an excellent correlation between Pa^CO2and Ps^CO2or Pg^CO2in one patient in whom transient hypercapnia developed (Figure 3). Thus one should always use the P^CO2gap (the Pg^CO2- Pa^CO2difference) when assessing gastric oxygenation. This is not an additional constraint, because pHi measurement should be considered in relation to arterial pH to avoid the influence of metabolic acidosis. [8]The use of capnometry in patients with a stable dead space ratio may be very valuable, especially during anesthesia.

Second, an increase in Pg^CO2(like a reduction in pHi) does not necessarily reflect gut hypoxia. According to the Fick equation, a reduction in blood flow is associated with an increase in the venoarterial gradient in carbon dioxide content and thus in P^COsub 2. However, the decrease in oxygen delivery below a critical value is associated with a much greater increase in the venoarterial P^COsub 2 gradient because the P^CO2generated by the acidotic cells accumulates locally. [9,10]Hence, an increased PgCO²-PaCO²gradient should be interpreted as reduced splanchnic perfusion, which does not necessarily indicate splanchnic hypoxia.

The clinical importance of early identification of organ ischemia to prevent the development of multiple organ dysfunction and its associated high mortality rate has been well recognized. [11-13]Monitoring systemic oxygen transport lacks the sensitivity to detect regional, or even global, tissue underperfusion. Gut monitoring has three attractive features. First, splanchnic blood flow is reduced early during even minor cardiovascular alterations in an attempt to preserve blood supply to more vital organs, namely the heart and the brain. Second, the tip of the gut villus may be particularly susceptible to a reduction in blood flow, given the local countercurrent mechanism supplying oxygen. [14]Third, the gut is easily accessible to regional monitoring by inserting a nasogastric or rectal probe.

Our study did not evaluate the clinical utility of Pg^COsub 2 monitoring. Several clinical studies have indicated that a low pHi, and especially the persistence of a low pHi, are associated with a greater incidence of organ dysfunction and increased mortality rates in critically ill patients. [11,15-21]Whether gastric tonometry can influence the individual management of an acutely ill patient has not been fully determined. A multicenter study [22]has suggested that pHi-guided therapy may improve outcome, but this was true only in patients with normal pHi on admission, which emphasizes the importance of the early detection of splanchnic hypoperfusion. Another study [23]indicated that gastric tonometry may help to identify failure to wean from mechanical ventilation. Nevertheless, several groups of investigators have stressed that the interpretation of these measurements in individual patients is sometimes difficult. [24-26] 

Gas tonometry thus represents an alternative reliable technique for bedside monitoring of gastric perfusion. Clinical studies can now evaluate the clinical utility of gas tonometry. We hope that improved monitoring of tissue perfusion will help to identify those patients who would benefit from increased oxygen delivery, facilitating a more tailored approach to therapy, rather than maintaining supranormal oxygen delivery levels in all "at risk" patients, an approach that has failed to improve survival rates. [27-29]