Intensive Care Med (2001) 27: 1923±1930 DOI 10.1007/s00134-001-1141-1
Jyrki J. Tenhunen Ari Uusaro Esko Ruokonen
Received: 12 February 2001 Final revision received: 3 August 2001 Accepted: 21 September 2001 Published online: 8 November 2001 Springer-Verlag 2001
)
J. J. Tenhunen ( ) ´ A. Uusaro ´ E. Ruokonen Division of Intensive Care, Department of Anesthesiology and Intensive Care, Kuopio University Hospital, P. O.B. 1777, 70211 Kuopio, Finland E-mail:
[email protected] Phone: +3 58-17-17 34 30 Fax: +3 58-17-17 34 43
EXPERIM ENTAL
Gastric mucosal systemic partial pressure of carbon dioxide (PCO2) gradient in experimental endotoxin shock in swine ± comparison of two methods
Abstract Objectives: Clinically applicable methods for continuous monitoring of visceral perfusion/ metabolism do not exist. Gastric mucosal end-tidal partial pressure of carbon dioxide (PCO2) gradient has been used, but it has limitations, especially in patients with lung injury and increased dead space ventilation. We studied the agreement between gastric mucosal end-tidal (DPCO2gas) and gastric mucosal arterial PCO2 (D(t-a)PCO2) gradients, and especially the effect of dead space ventilation (Vd/Vt ratio) on the agreement. We hypothesized that DPCO2gas can be used as a semicontinuous indicator of mucosal arterial PCO2 gradient in sepsis. Design: A randomized, controlled animal experiment. Setting: National laboratory animal center. Interventions: Twelvehour infusion of endotoxin in landrace pigs. Measurements and results: We measured end-tidal PCO2 continuously, gastric mucosal PCO2 every 10 min (gas tonometry) and arterial PCO2 every 120 min. Carbon dioxide production and the Vd/Vt ratio were determined by indirect calorimetry. In the endotoxin group (n = 7) car-
Introduction Sepsis and septic shock are associated with gut mucosal hypoperfusion [1, 2], acidosis [3] and altered oxygen uti-
diac index increased and systemic vascular resistance decreased. Endotoxemia increased dead space ventilation by 27 % (p = 0.001). Both DPCO2gas and D(t-a)PCO2 increased significantly in the endotoxin group (p < 0.0001 and p = 0.049, respectively). Control animals remained stable throughout the experiment. When we compared DPCO2gas and D(t-a)PCO2 (BlandAltman analysis), the bias and precision were 0.9 and 0.9 kPa in the control group and 2.0 and 2.2 kPa in the endotoxin group, respectively. The disagreement between DPCO2gas and D(t-a)PCO2 increased as the Vd/Vt ratio increased. Conclusions: DPCO2gas is a clinically applicable method for continuous monitoring of visceral perfusion/ metabolism. Septic lung injury and increased dead space ventilation decrease the accuracy of the method, but this may not be clinically important. Keywords Endotoxin ´ Shock ´ Lipopolysaccharide ´ Dead space ventilation (Vd/Vt ratio) ´ Calorimetry ´ Tonometry ´ Splanchnic perfusion
lization [4, 5]. Gut hypoperfusion may increase mucosal permeability [6], which may activate inflammatory cascades causing organ failures. Prevention of gastric mucosal acidosis may improve the outcome of critically ill
1924
patients [7, 8]. Hence, early detection of perfusion defect/deranged metabolism of the intestine may be important. Conventional saline tonometry has several limitations [9, 10, 11] and clinically applicable methods for continuous monitoring of splanchnic perfusion/metabolism do not exist. We showed recently that gastric tonometric end-tidal PCO2 gradient (DPCO2gas) is a potential non-invasive tool for monitoring visceral perfusion/metabolism continuously in critically ill patients who are not septic [11]. We also estimated the dead space ventilation of our patients and found that increased dead space ventilation decreases the accuracy of this method. Sepsis, in addition to altering visceral perfusion/metabolism, causes lung injury. We designed a controlled, randomized, experimental study using a model of endotoxin shock in pigs. Our aim was to study the gastric mucosal endtidal PCO2 gradient in sepsis. Especially, we carefully evaluated the effect of increased dead space ventilation on this method.
Materials and methods The institutional animal use and care committee of the University of Kuopio approved the study protocol. Anesthesia Fourteen female Finnish landrace pigs (28±38 kg) were fasted for 48 h with free access to water before the experiment. The gastrointestinal tract was emptied with Colonsteril (Orion, Espoo, Finland) 2 days before the experiment. The animals were premedicated with atropine 0.05 mg/kg and azaperone 8 mg/kg intramuscularly. Thereafter an ear vein was cannulated for administration of sodium thiopentone 5±15 mg/kg to induce anesthesia. A tracheotomy was performed and the lung was ventilated using a volume-controlled mode (Servo 900E, Siemens Elema, Sweden) with 10±15 ml/kg tidal volume to achieve normocapnia. The arterial partial pressure of CO2 (PaCO2) was kept between 4.5±5.5 kPa (34±41 mmHg) and the minute volume (VE) of the ventilator was adjusted accordingly by changing the respiratory rate. The fraction of oxygen in the inspiratory gas (FIO2) was adjusted to keep the arterial partial pressure of oxygen above 13.3 kPa (100 mmHg). A positive end-expiratory pressure of 5 cmH2O was used throughout the experiment. Anesthesia was maintained with a continuous infusion of sodium thiopentone 5 mg/kg per h. Fentanyl was infused, 30 mg/kg per h during surgery and 5 mg/kg per h thereafter. We did not use neuromuscular blocking drugs. The operation table was covered with an antidecubitus gel mattress during the 24h experiment and the animals were in supine position. Animal preparation After induction of anesthesia, a tracheotomy was performed and the animals were connected to the ventilator (Servo 900C, Siemens Elema, Solna, Sweden). The right femoral artery was cannulated with a single lumen central venous catheter (Arrow, Arrow International, Reading, Pa., USA) for blood sampling and arterial pres-
sure measurement. A 7.5 Fr flow-directed pulmonary artery catheter (Arrow, Arrow International, Reading, Pa., USA) was inserted through the right jugular vein. We performed a midline laparotomy. The urinary bladder was drained. The descending aorta, celiac trunk, superior mesenteric artery, inferior mesenteric artery and portal vein were prepared and visualized. Precalibrated ultrasonic transit time flow probes (Transonic Systems, Ithaca, N. Y., USA) of appropriate size were applied around the vessels. The perivascular ultrasound flow probes were calibrated in vitro before each experiment. Signals from the flowmeters at 30 Hz (T206 and T106, Transonic) were recorded on a computer program for further analysis (Windaq 1.60, Dataq instruments, Akron, Ohio, USA). In vivo zero flow signals were recorded after each experiment. A gastric tonometer (Tonometrics, Datex/instrumentarium, Helsinki, Finland) was guided into the stomach orally. The laparotomy was closed in two layers and pleural drains were inserted through lateral incisions. Hemodynamic monitoring We measured systemic and pulmonary arterial, central venous (CVP) and pulmonary artery occlusion (PAOP) pressures (CS3, Datex-Ohmeda, Helsinki, Finland) and used automated data filtering (2min median) to collect continuous data (Clinisoft, Datex, Helsinki, Finland). Pressure transducers were zeroed to the level of the heart. Heart rate was measured from the ECG. Cardiac output was measured by a thermodilution technique (mean value of three measurements) using 5-ml injectates of room temperature saline. Core temperature was measured from the tip of the pulmonary artery catheter and kept constant at 38.5 0.5 C with a thermistor-controlled, heated operation table. Tonometry We measured gastric mucosal PCO2 with a semi-automatic gas analyzer (Tonocap, Datex-Ohmeda, Helsinki, Finland) every 10 min throughout the experiment. The device was calibrated every 2 months according to the guidelines of the manufacturer. End-tidal PCO2 (EtCO2) was recorded continuously (CS3, Datex-Ohmeda, Helsinki, Finland). The tonometric EtCO2 gradient (DPCO2gas) was calculated every 10 min. Tonometric arterial PCO2 gradient (D(t-a)PCO2) was calculated at baseline, after 30 min of endotoxin infusion and every 2nd h thereafter. We did not use H2-blockers [12, 13, 14]. Dead space measurement Whole body CO2 production (VCO2) was measured with continuous indirect calorimetry (Deltatrac, Datex, Helsinki, Finland). The fraction of dead space ventilation of total ventilation (Vd/Vt ratio) was calculated using the following equation [15]: Vd/Vt = 1 [(VCO2 * 0.115)/(VE * PaCO2)] where VCO2 is CO2 production (ml/min), VE is minute ventilation (l) and PaCO2 is arterial partial pressure of carbon dioxide (kPa). Fluid management Saline 0.9 % was infused 5 ml/kg per h throughout the experiment both for the endotoxic and control animals. Ringer's acetate and
1925
Table 1 Systemic and regional hemodynamics in endotoxin shock (n = 7) and in control group (n = 6). Probability values refer to non-parametric analysis of variance for repeated measurements 0h
(Friedman) (HR heart rate, CI cardiac index, PVR pulmonary vascular resistance, MAP mean arterial pressure, PAOP pulmonary arterial occlusion pressure)
2h
6h
10 h
12 h
p value, 0±6 h
p value, 0±12 h
96 (92±104) 69 (63±73)
118 (84±123) 72 (69±73)
134 (116±153) 66 (64±69)
145 (130±155) 70 (63±70)
0.03 0.67
0.004 0.65
83 (77±91) 102 (83±117)
123 (108±125) 103 (91±114)
229 (163±235) 102 (90±116)
157 (157±235) 97 (88±118)
0.004 0.15
< 0.0001 0.71
Systemic vascular resistance (dyn´s´cm5) Endotoxin 2336 (2231±2698) 2496 (2275±2710) 1816 (1599±2702) 1061 (693±1191) 726 (648±837) 0.008 Control 1967 (1905±2466) 1967 (1920±2531) 2065 (2004±2437) 1997 (1926±2427) 2105 (1771±2365) 0.53
< 0.0001 0.99
Splanchnic blood flow (ml/kg/min) Endotoxin 34 (27±37) 17 (15±21) Control 35 (31±39) 36 (32±39)
26 (24±34) 36 (35±39)
47 (41±58) 36 (31±42)
47 (40±50) 31 (30±37)
0.002 0.334
< 0.001 0.809
PVR (dyn´s´cm5) Endotoxin 274 (227±310) Control 242 (225±258)
700 (469±721) 226 (218±314)
373 (344±384) 249 (230±267)
367 (307±404) 254 (230±377)
< 0.0001 0.46
0.001 0.36
91 (79±112) 87 (83±89)
86 (70±90) 88 (84±91)
64 (62±67) 88 (85±88)
0.28 0.39
0.04 0.65
7 (6±9) 5 (4±6)
8 (7±9) 7 (6±7)
7 (7±7) 7 (5±8)
0.004 0.82
0.028 0.013
37 (26±41) 15 (14±15)
36 (34±37) 16 (16±16)
31 (30±33) 17 (16±18)
0.004 0.16
0.06 0.001
38.6 (38.4±38.8) 38.4 (38.2±38.7)
38.5 (38.4±38.6) 38.3 (38.0±38.4)
38.4 (38.4±38.5) 38.3 (38.0±38.4)
0.17 0.13
0.05 0.17
HR (min) Endotoxin 73 (70±74) Control 68 (64±78) CI (ml/kg per min) Endotoxin 102 (92±115) Control 102 (93±115)
933 (844±1008) 227 (202±278)
MAP (mmHg) Endotoxin 90 (89±91) 88 (79±89) Control 92(87±100) 88 (85±90) PAOP (mmHg) Endotoxin 5 (4±5) 6 (6±7) Control 6 (5±6) 6 (4±6) Mean pulmonary arterial pressure (mmHg) Endotoxin 14 (13±14) 33 (29±39) Control 15 (15±16) 14 (14±15) Core temperature (C0) Endotoxin 38.2 (37.9±38.4) Control 38.1 (37.7±38.4)
38.6 (38.4±38.9) 38.1 (37.9±38.5)
hydroxyethyl starch (Plasmafucin, Kabi-Pharmacia, Uppsala, Sweden) 1:1 were infused to keep PAOP at 5±7 mmHg and CVP at 3±5 mmHg. If the mean systemic arterial pressure dropped below 55 mmHg with a PAOP of 5±7 mmHg, 100±200 ml of fluid was infused to increase the PAOP to 8±9 mmHg. Glucose (50 %) was infused and the infusion rate was adjusted to keep blood glucose levels at 5±7 mmol/l. Experimental protocol The animals were allowed to stabilize for 8±10 h after the surgery. Thereafter, they were randomly allocated to one of two groups: group 1 (n = 7) with Escherichia coli endotoxin infusion (lipopolysaccharide B 0111:B4, Difco Laboratories, Detroit, Mich., USA) [16, 17] or group 2 (n = 7), which served as control group. We used sealed opaque envelopes for randomization. After randomization, we established the baseline measurements and took the blood samples. Arterial blood gas analysis was performed within 5 min (ABL-520, Radiometer, Copenhagen, Denmark). Infusion of E. coli endotoxin (20 mg/ml in 5 % glucose, group 1) was started at a rate of 1.0 mg/kg per h. After 2±4 h the infusion rate was increased stepwise to induce systemic hypotension. However, mean pulmonary arterial pressure was not allowed over 40 mmHg. We used this protocol, because our pilot study showed that if the infusion rate of endotoxin was increased too rapidly, death ensued due to low cardiac output with extreme pulmonary
hypertension. At the end of the experiment the animals were killed with intravenous magnesium sulfate while they were still anesthetized. One control animal was excluded from the study, because a pneumothorax developed during manipulation of the pleural drain. Two animals in the endotoxin group died at 6 and 10 h into the experiment due to cardiovascular collapse. Calculations and statistical analysis We estimated regional CO2 production by using an iterative procedure for regional CO2 content difference [18] and combining hepatic arterial and portal venous blood flows to represent total splanchnic blood flow. Splanchnic oxygen delivery (DsplO2) and splanchnic oxygen consumption (VsplO2) were calculated from arterial and venous oxygen contents and regional blood flow. Systemic vascular resistance (SVR) was calculated as: (mean arterial pressure (MAP)±CVP)/CO80 (dyn´s´cm5). Pulmonary vascular resistance was calculated as (mean pulmonary pressure±PAOP)/CO80 (dyn´s´cm5). The results are given as median (interquartile range) unless otherwise stated. We used Bland-Altman analysis [19] to study the agreement between DPCO2gas and D(t-a)PCO2. Bias refers to the mean difference between the two measurements and precision is 2 standard deviations of the bias. Non-parametric analysis of variance for repeated measurements (Friedman) was used for within group comparison. Regression analysis was used to study the correlation between the two vari-
1926
Table 2 Gas exchange in endotoxin shock (n = 7) and in control group (n = 6). Probability values refer to non-parametric analysis of variance for repeated measurements (Friedman) (VCO2 CO2
production, Vd /Vt ratio dead space to tidal volume ratio, DsplO2 splanchnic oxygen delivery, VsplO2 splanchnic oxygen consumption)
0h
2h
6h
10 h
12 h
p value, 0±6 h
p value, 0±12 h
3.7 (3.5±3.8) 3.6 (3.4±3.7)
4.4 (4.2±4.9) 3.6 (3.4±3.6)
4.3 (4.2±4.8) 3.6 (3.4±3.8)
5.3 (5.2±7.0) 3.7 (3.4±4.0)
5.3 (5.1±6.3) 3.6 (3.5±4.0)
0.02 0.91
0.001 0.35
0.5 (0.5±0.6) 0.5 (0.5±0.5)
0.7 (0.6±0.7) 0.5 (0.5±0.6)
0.7 (0.7±0.7) 0.6 (0.5±0.6)
0.7 (0.7±0.8) 0.5 (0.5±0.6)
0.7 (0.7±0.8) 0.6 (0.5±0.6)
0.003 0.28
0.001 0.2
PaO2/FIO2 ratio Endotoxin Control
474 (416±480) 442 (418±481)
406 (288±416) 417 (385±480)
307 (295±344) 424 (419±444)
132 (81±148) 423 (402±459)
297 (120±311) 425 (370±442)
0.013 0.17
0.002 0.33
DsplO2 (ml/kg per l) Endotoxin Control
4.0 (3.2±4.3) 4.3 (3.6±4.8)
2.2 (1.8±2.2) 4.1 (3.9±4.3)
2.8 (2.4±3.5) 3.9 (3.4±4.6)
3.7 (3.3±4.1) 3.6 (2.8±4.3)
3.8 (3.4±3.9) 3.0 (2.7±3.5)
0.003 0.457
0.012 0.154
VsplO2 (ml/kg per l) Endotoxin Control
1.3 (1.1±1.8) 1.6 (1.3±1.8)
1.5 (1.3±1.6) 1.6 (1.3±1.8)
1.3 (1.2±2.0) 1.4 (1.2±1.5)
1.5 (1.3±1.8) 1.2 (1.1±1.4)
1.4 (1.3±1.8) 1.1 (1.1±1.3)
0.543 0.086
0.661 0.05
Splanchnic CO2 production (ml/kg per min) Endotoxin 1.1 (1.0±1.6) 1.3 (1.2±1.5) Control 1.1 (0.9±1.4) 1.2 (0.9±1.4)
1.3 (1.3±2.0) 1.2 (1.0±1.3)
1.5 (1.3±1.8) 1.3 (1.1±1.5)
1.9 (1.3±2.0) 1.3 (1.1±1.5)
0.156 0.896
0.147 0.6
VCO2 (ml/kg per min) Endotoxin Control Vd/Vt ratio Endotoxin Control
ables. A p value less than 0.05 was chosen to indicate statistical significance.
Results Systemic and regional hemodynamics and gas exchange The primary response to endotoxin infusion was an increase in pulmonary vascular resistance (PVR) and a decrease in cardiac index (CI) (Table 1). Towards the end of the experiment CI increased while SVR markedly decreased in the endotoxic group. Whole body VCO2 and Vd/Vt ratio increased in endotoxic animals, whereas PaO2/FIO2 ratio decreased (Table 2). Animals in the control group remained stable. Splanchnic blood flow decreased during the hypodynamic shock but increased over the baseline during hyperdynamic circulation (Table 1). Splanchnic oxygen delivery decreased during the hypodynamic phase of endotoxemia but recovered to baseline level during the hyperdynamic shock. Regional oxygen consumption remained constant. Splanchnic CO2 production did not change (Table 2). Tonometry Median DPCO2gas was 2.5 (2.4±2.7) kPa at baseline in the endotoxin group and increased thereafter in a biphasic manner. An initial increase occurred during the hypodynamic phase of circulation. Later on, the hyper-
dynamic type of circulation with low SVR was associated with more pronounced increase in DPCO2gas (Fig. 1). The same pattern was apparent in D(t-a)PCO2 (Fig. 1). When we compared DPCO2gas and D(t-a)PCO2, Bland-Altman analysis showed a greater bias and poorer precision in endotoxemic than in the control animals. Bias and precision in the control animals were 0.9 kPa and 0.9 kPa, and 2.0 kPa and 2.2 kPa in the endotoxic animals, respectively (Fig. 2). The correlation between DPCO2gas and D(t-a)PCO2 is shown in Fig. 3.
Discussion We found a disagreement between gastric mucosal endtidal (DPCO2gas) and gastric mucosal arterial (D(ta)PCO2) PCO2 gradients in our experimental model of sepsis. More specifically, DPCO2gas systematically overestimated D(t-a)PCO2. Overestimation increased with acute lung injury and elevated dead space ventilation. However, DPCO2gas increased with the progression of sepsis. Hence, detection of changes in the gastric mucosal end-tidal PCO2 gradient may be useful. Furthermore, the disagreement between DPCO2gas and D(t-a)PCO2 may not be clinically important. Endotoxin shock model Similar experimental models of porcine endotoxin shock have been described [4, 14, 15]. In our study, en-
1927
Fig. 1 Upper panel Gastric mucosal end-tidal PCO2 gradient (DPCO2gas) in endotoxin shock (n = 7) (continuous thick black line) and in the control animals (n = 6) (continuous thin line). Median values and interquartile ranges are shown. Probability values refer to changes over time within a group (Friedman non-parametric test). Lower panel Tonometric arterial PCO2 gradient (D(t-a)PCO2) in endotoxin shock (gray column) and in the control animals (white column)
dotoxin induced hyperdynamic circulation with high cardiac index and low SVR, which are characteristic features of severe sepsis and shock. Endotoxin also induced acute lung injury, characterized by a low PaO2/ FIO2 ratio and an increased fraction of dead space ventilation. In addition, animals in the endotoxin group became hypermetabolic, as judged by increased whole body production of CO2. Finally, two animals died due to cardiovascular collapse while receiving endotoxin, indicating severity of shock. In contrast, animals in the control group remained stable in every respect. It is possible that surgical manipulation may have altered visceral perfusion. However, the aim of this study was to compare the agreement between two methods to estimate visceral perfusion/metabolism. We compared the agreement between two measurements performed at the same time points during the experiment. Therefore the potential changes in visceral perfusion induced by surgery do not invalidate the interpretation of our results.
Tonometry and dead space measurement We used gastrointestinal gas tonometry according to the guidelines of the manufacturer of the device. The need for H2-blockers with tonometry is controversial and we chose not to use them [12, 13, 14]. More importantly, this does not affect the agreement between the two measurements. Our device for indirect calorimetry is well validated [20, 21]. High FIO2, which was used in the endotoxic animals, does not affect the accuracy of the VCO2 measurement [20]. Limitations and potential clinical application The gastric mucosal end-tidal gradient overestimates D(t-a)PCO2 especially, if dead space ventilation is increased. Hence, false positive findings for abnormal mucosal perfusion or metabolism are possible. As an example, in our data the D(t-a)PCO2 remained low in two en-
1928
Fig. 2 The agreement between simultaneously measured DPCO2gas and D(t-a)PCO2 during endotoxemia (lower figure, black circles) and in control (upper figure, white circles) animals. Bias was 0.9 kPa (solid line) and precision 0.9 kPa (dashed lines) in control group. Bias and precision in endotoxemia were 2.0 kPa and 2.2 kPa, respectively
Fig. 3 Association between tonometric arterial (D(t-a)PCO2) and tonometric end-tidal PCO2 (DPCO2gas) gradients with regression line and 95 % confidence intervals (thin solid lines) in endotoxemic (lower panel, black circles) and control animals (upper panel, open circle). Line of identity is shown as a dashed line. The equation, r2 and p values refer to linear regression line
dotoxemic animals while the Vd/Vt ratio increased consistently. This can be seen in the Bland-Altman plot as a higher disagreement when the PCO2 gradient is low. On the other hand, true episodes of mucosal hypoperfusion/derangements in mucosal CO2 metabolism will not be left undetected. In addition, even though DPCO2gas is prone to error, the source of error is self-evident, easily recognized and the degree of the error estimated simply by measuring arterial blood PCO2. The clinical importance of the disagreement between DPCO2gas and D(t-a)PCO2 is difficult to judge. A recent experimental study suggested that mucosal hypoxia develops only after gastric mucosal PCO2 increases over 13 kPa [22].
Therefore, it is possible that the disagreement between DPCO2gas and D(t-a)PCO2, induced by septic lung injury, is not clinically important. In our study DPCO2gas increased with increasing severity of shock. Therefore, we would like to propose that increasing DPCO2gas is a warning sign of clinical deterioration in sepsis and acute lung injury, too. In this study we did not aim to investigate the pathophysiologic association/dissociation between gastric mucosal PCO2 and mucosal oxygenation or perfusion [23, 24, 25]. Neither did we address the question whether the stomach is the ideal site to monitor splanchnic perfusion/metabolism [24, 25]. In fact, in the present experiment an in-
1929
creasing mucosal systemic PCO2 gradient was associated with decreased splanchnic blood flow and regional oxygen supply only during hypodynamic shock. In addition, splanchnic CO2 production did not increase concomitantly with an increasing mucosal systemic PCO2 gradient. We also realize that the potential clinical benefits of gastric tonometry are not clear [8, 25, 26]. Nevertheless, mucosal PCO2, whether measured from stomach, small bowel [27], colon [28] or urinary bladder [29], reflects the local perfusion and/or metabolism. Also, the potential clinical benefit of continuous monitoring of splanchnic perfusion/metabolism has not been studied. To our knowledge this is the first study that systematically evaluates DPCO2gas in sepsis and acute lung injury. Mucosal end-tidal PCO2 gradient was used recently in patients with sepsis and a high gradient was considered to indicate visceral hypoperfusion [30]. However, these investigators did not evaluate the potential effect of lung injury (Vd/Vt ratio) on the measurement [30].
In our study, the Vd/Vt ratio increased early and then remained stable. This is not true in clinical practice. Also, theoretically, false negative findings for abnormal mucosal perfusion/metabolism are possible, if dead space ventilation decreases and mucosal perfusion/metabolism deteriorates simultaneously. Under such circumstances the gradient between mucosal and end-tidal PCO2 would not change. In conclusion, we found that semi-continuous measurement of the gastric mucosal end-tidal PCO2 gradient can be used to monitor mucosal perfusion/CO2 metabolism in sepsis with acute lung injury. The obvious limitations of this technique are easy to understand and to cope with. The effect of dead space ventilation on the measurements must be evaluated during a dynamic disease process. Acknowledgements This study has been supported by a grant from Kuopio University Hospital and in part by a grant from the Finnish Medical Foundation.
References 1. Neviere R, Chagnon JL, Vallet B, Lebleu N, Marechal X, Mathieu D, Wattel F, Dupuis B (1997) Dobutamine improves gastrointestinal mucosal blood flow in a porcine model of endotoxic shock. Crit Care Med 25: 1371±1377 2. Tugtekin IF, Radermacher P, Theisen M, Matejovic M, Stehr A, Ploner F, Matura K, Ince C, Georgieff M, Trager K (2001) Increased ileal-mucosal-arterial PCO2 gap is associated with impaired villus microcirculation in endotoxic pigs. Intensive Care Med 27: 757±766 3. Revelly JP, Ayuse T, Brienza N, Fessler HE, Robotham JL (1996) Endotoxic shock alters distribution of blood flow within the intestinal wall. Crit Care Med 24: 1345±1351 4. Crouser ED, Julian MW, Dorinsky PM (1999) Ileal VO(2)-O(2) alterations induced by endotoxin correlate with severity of mitochondrial injury. Am J Respir Crit Care Med 160: 1347±1353 5. Trager K, Radermacher P, Rieger KM, et al. (2000) Norepinephrine and N(G)-monomethyl-L-arginine in hyperdynamic septic shock in pigs: effects on intestinal oxygen exchange and energy balance. Crit Care Med 28: 2007±2014 6. Fink MP, Kaups KL, Wang H, Rothschild HR (1992) Ibuprofen improves survival but does not ameliorate increased gut mucosal permeability in endotoxic pigs. Arch Surg 127: 49±53
7. Mythen MG, Webb AR (1994) Intraoperative gut mucosal hypoperfusion is associated with increased post-operative complications and cost. Intensive Care Med 20: 99±104 8. Gutierrez G, Palizas F, Doglio G, et al. (1992) Gastric intramucosal pH as a therapeutic index of tissue oxygenation in critically ill patients. Lancet 339: 195±199 9. Takala J, Parviainen I, Siloaho M, Ruokonen E, Hämäläinen E (1994) Saline PCO2 is an important source of error in the assessment of gastric intramucosal pH. Crit Care Med 22: 1877±1879 10. Fiddian-Green RG (1995) Gastric intramucosal pH, tissue oxygenation and acid-base balance. Br J Anaesth 74: 591±606 11. Uusaro A, Lahtinen P, Parviainen I, Takala J (2000) Gastric mucosal end-tidal pCO2 difference as a continuous indicator of splanchnic perfusion. Br J Anaesth 85: 563±569 12. Parviainen I,Vaisanen O, Ruokonen E, Takala J (1996) Effect of nasogastric suction and ranitidine on the calculated gastric intramucosal pH. Intensive Care Med 22: 319±323 13. Calvet X, Baigorri F, Duarte M, et al. (1998) Effect of ranitidine on gastric intramucosal pH in critically ill patients. Intensive Care Med 24: 12±17
14. Bams JL, Kolkman JJ, Roukens MP, et al. (1998) Reliable gastric tonometry after coronary artery surgery: need for acid secretion suppression despite transient failure of acid secretion. Intensive Care Med 24: 1139±1143 15. Kiiski R, Takala J, Eissa NT (1991) Measurement of alveolar ventilation and changes in deadspace by indirect calorimetry during mechanical ventilation: a laboratory and clinical validation. Crit Care Med 19: 1303±1309 16. Santak B, Radermacher P, Adler J, et al. (1998) Effect of increased cardiac output on liver blood flow, oxygen exchange and metabolic rate during longterm endotoxin-induced shock in pigs. Br J Pharmacol 124: 1689±1697 17. Trager K, Radermacher P, Rieger KM, et al. (1999) Norepinephrine and nomega-monomethyl-L-arginine in porcine septic shock: effects on hepatic O2 exchange and energy balance. Am J Respir Crit Care Med 159: 1758±1765 18. Giovannini I, Chiarla C, Boldrini G, Castagneto M (1993) Calculation of venoarterial CO2 concentration difference. J Appl Physiol 74: 959±964 19. Bland JM, Altman DG (1986) Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1: 307±310 20. Takala J, Keinänen O, Väisänen P, Kari A (1989) Measurement of gas exchange in intensive care: laboratory and clinical validation of a new device. Crit Care Med 17: 1041±1047
1930
21. Phang PT, Rich T, Ronco J (1990) A validation and comparison study of two metabolic monitors. JPEN 14: 259±261 22. Schlichtig R, Bowles SA (1994) Distinguishing between aerobic and anaerobic appearance of dissolved CO2 in intestine during low flow. J Appl Physiol 76: 2443±2451 23. VanderMeer TJ, Wang H, Fink MP (1995) Endotoxemia causes ileal mucosal acidosis in the absence of mucosal hypoxia in a normodynamic porcine model of septic shock. Crit Care Med 23: 1217±1226
24. Kellum JA, Rico P, Garuba AK, Pinsky MR (2000) Accuracy of mucosal pH and mucosal-arterial carbon dioxide tension for detecting mesenteric hypoperfusion in acute canine endotoxemia. Crit Care Med 28: 462±466 25. Creteur J, De Backer D, Vincent JL (1999) Does gastric tonometry monitor splanchnic perfusion? Crit Care Med 27: 2480±2484 26. Gomersall CD, Joynt GM, Freebairn RC, Hung V, Buckley TA, Oh Te (2000) Resuscitation of critically ill patients based on the results of gastric tonometry: a prospective, randomized, controlled trial. Crit Care Med 28: 607±614 27. Walley KR, Friesen BP, Humer MF, Phang PT (1998) Small bowel tonometry is more accurate than gastric tonometry in detecting gut ischemia. J Appl Physiol 85: 1770±1777
28. Koga I, Stiernstrom H, Christiansson L, Wiklund L (1999) Intraperitoneal and sigmoid colon tonometry in porcine hypoperfusion and endotoxin shock models. Acta Anaesthesiol Scand 43: 702±707 29. Lang JD Jr, Evans DJ, deFigueiredo LP, Hays S, Mathru M, Kramer GC (1999) A novel approach to monitor tissue perfusion: bladder mucosal PCO2, PO2, and pHi during ischemia and reperfusion. J Crit Care 14: 93±98 30. Sakka SG, Meier-Hellmann A, Reinhart K (2000) Do fluid administration and reduction in norepinephrine dose improve global and splanchnic hemodynamics? Br J Anaesth 84: 758±762
