INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The study that is the subject of this article was prompted by a series of clinical and experimental observations by our group, establishing that increases in tissue PCO₂ were characteristic of critical low-flow states of circulatory shock and cardiac resuscitation. We initially detected disproportional increases in mixed venous blood PCO₂ during the low-flow states of cardiac resuscitation, with decreases rather than increases in arterial PCO₂ (1, 2). An even greater increase in PCO₂ of great cardiac vein blood was then observed (3, 4). We subsequently demonstrated that the PCO₂ of the heart, the liver parenchyma, the kidney, and the cerebral cortex were increased during the low-flow state of circulatory shock and cardiac resuscitation (5). We subsequently confirmed that gastric tissue PCO₂ (PgCO₂) increased during hemorrhagic, anaphylactic, and septic shock states (10). Gastric tonometry emerged at the same time as a quantitative indicator of the severity of perfusion failure and, more specifically, of the capability to extract and use oxygen (13, 14). The initial focus was on gastric intramural pH (pHi), determined by using an intragastric saline-filled balloon serving as a gastric tonometer (15, 16). In its clinical application, pHi predicted short-term outcome of patients who presented with perfusion failure (17). Gastric tonometry has come into active clinical use for detection of visceral hypoperfusion (21). Our own studies subsequently confirmed that, ideally, the PCO₂ of the saline in the tonometric balloon approximates the PCO₂ of the stomach wall (10). It was for this reason that we maintained focus on the direct measurement of PCO₂ in the stomach wall, recognizing that such, in part, obviated errors due to variable acid secretion by the stomach and assumptions relating to tissue bicarbonate. We subsequently found that the esophageal wall also served as an appropriate site for tissue PCO₂ measurements for quantitation of severity of hemorrhagic shock (22, 23). Accordingly, we advanced the hypothesis that hypercarbia, as a universal phenomenon of critically reduced tissue perfusion, would also involve the very proximal gastrointestinal tract, namely in the sublingual mucosa.

Preliminary trials confirmed that measurements of PCO₂ under the tongue would provide a quantitative indication for diagnosis and quantitation of severity of circulatory shock, like that of gastric and esophageal PCO₂. We therefore sought systemic comparisons of sublingual PCO₂ (PslCO₂) with PgCO₂ and with conventional hemodynamic and metabolic quantitators of severity and prognosis of circulatory shock states in an established rodent model (10). Additional comparisons were made with arterial (aortic) pressure, end-tidal PCO₂ (EtPCO₂), cardiac index, and arterial blood lactate. Sublingual PCO₂ was measured in settings of hypovolemic shock induced by bleeding and septic shock after intravenous injection of live staphylococci.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The protocol was reviewed and approved by our Institutional Animal Care and Use Committee (University of Southern California School of Medicine, Los Angeles, CA). All animals received humane care in compliance with the Guide for the Care and Use of Laboratory Animals, formulated by National Research Council and published by National Academy Press (1996).

Hemorrhagic Shock

Eighteen Sprague-Dawley rats, weighing between 450 and 550 g, were investigated. The animals were fasted overnight except for free access to water. Anesthesia followed intraperitoneal injection of pentobarbital sodium (45 mg/kg). The animals were placed on a surgical board in the supine position. The trachea was surgically exposed at a site 2 cm caudal to the larynx. A 14-gauge cannula (Abbocath-T; Abbott Hospital, North Chicago, IL) was then advanced into the trachea for a distance of 1 cm. End-tidal PCO₂ was continuously monitored with a sidestream infrared CO₂ analyzer (End-Tid IL 200; Instrument Laboratory, Lexington, MA). For measurement of aortic pressure, an 18-gauge polyethylene catheter (PE50; Intramedic, New York, NY) was advanced into the thoracic aorta from the surgically exposed right carotid artery. For the measurement of right atrial pressure, an additional catheter was advanced into the right atrium through the left jugular vein. For blood shedding, reinfusion of blood, and arterial and venous blood sampling, two additional catheters were inserted, one into the abdominal aorta and the second catheter into the inferior vena cava, through a surgically exposed left femoral artery and vein. For the measurement of cardiac output, a thermocouple microprobe (9030-12-D-34; Columbus Instruments, Columbus, OH) was advanced into the thoracic aorta through the surgically exposed right femoral artery. Blood temperature was monitored with this sensor and maintained between 36.5 and 37.5° C with an infrared heating lamp. The aortic catheter was connected to the barrel of a standard 20-ml plastic syringe that served as a reservoir for shed blood. A conventional lead II electrocardiogram was continuously monitored.

For the measurement of PslCO₂, we used a miniature carbon dioxide electrode (MI-720 CO₂ electrode; Microelectrodes, Londonderry, NH). The sensor was calibrated in a water-filled tonometer maintained at 37° C. A mixture of nitrogen and either 5% or 20% CO₂ gas (Air Liquide, Etiwanda, CA) was bubbled through the water. The sensor was advanced so as to lodge it between the tongue and the left sublingual mucosa, and the mouth was closed.

Gastric PCO₂ was simultaneously measured in 10 of the animals: five unbled control and five hemorrhaged animals. For the measurement of PgCO₂, an ion-sensitive field-effect transistor (ISFET) sensor (CO-1035; Nihon Kohden, Tokyo, Japan) was used. For placement of the PgCO₂ sensor, a midline surgical incision was made in the upper abdomen and the stomach was exposed. The sensor was imbedded into the submucosa of the anterior wall of the stomach to a depth of 5 mm as previously described (11). The abdomen was then closed in one layer.

The animals were randomized to serve as hemorrhagic shock or sham controls by the sealed envelope method. Our model was previously described (10, 11, 22). In brief, blood from the left femoral artery was allowed to flow into the reservoir, which was pressurized. In this instance, it was pressurized to 100 mm Hg for 10 min, 80 mm Hg for 20 min, 70 mm Hg for 20 min, and within the range of 50 to 60 mm Hg for the ensuing 70 min with adjustments to prevent reinfusion of shed blood. Reservoir pressure was controlled with a pressure regulator (model 10; Fairchild, Winston-Salem, NC) incorporated into a mercury manometer system, which allowed for relative precise manual adjustment of the pressure. After 2 h, the pressure in the reservoir was increased to 150 mm Hg, such that shed blood was reinfused over an interval of 10 min. The animals were then observed for an additional 50 min and then euthanized by intravenous injection of pentobarbital. An autopsy was routinely performed on all animals with gross inspection of thoracic and abdominal organs such as to identify adverse effects of the interventions. In control animals, the procedures were identical except that the position of a stopcock between the left femoral artery and the reservoir precluded blood flow into the reservoir.

Septic Shock

Septic shock was induced in five animals following intravenous injection of live organisms. The organisms were supplied as Staphylococcus aureus (25923, FDA strain Seattle; Difco Laboratories, Detroit, MI) as freeze-dried pellets and stored at 4° C. The pellets were exposed to room temperature for 1 h and then inoculated into trypticase soy broth (4311768; Becton Dickinson Microbiology Systems, Cockeysville, MD). They were incubated for 5 h at 37° C. The inoculum was then streaked onto medium plates of trypticase soy agar and incubated at 37° C for 18 h. Approximately 2 h before infusion, bacteria were washed from the plates with sterile saline to produce a suspension. Sterile tubes containing the suspension were then centrifuged. The bacterial "button" was then resuspended in sterile saline. The suspension was then concentrated or diluted to achieve an optical density of 1.0, which corresponded to a concentration of 6 × 10⁹ organisms/ml at 600 nm on a spectrophotometer. The individual concentration was based on colony counting in each instance.

After baseline measurements were completed, five animals were randomized and received 5.7 (± 0.2) × 10⁹ S. aureus organisms/kg in 3 ml of saline at a rate of 0.05 ml/min over an interval of 1 h. The animals were continuously monitored for 12 h. Aortic pressure, EtPCO₂, and PslCO₂ were continuously measured. Cardiac output was measured at hourly intervals. Blood gases and arterial blood lactate concentration were measured at 2-h intervals.

Procedures

Dynamic data were continuously recorded with the aid of a PC-compatible 486 computer using data acquisition hardware/software (DATAQ Instruments, Akron, OH). Cardiac output was measured by an adaptation of the thermodilution technique in which a bolus of 200 µl of saline at a temperature of 15° C was injected into the right atrium and blood temperature was measured in the proximal thoracic aorta. Cardiac index was computed as the cardiac output per kilogram body weight with the adaptation of commercially available software (National Instruments, Austin, TX). Cardiac output measurements were obtained before hemorrhage and at defined intervals after start of hemorrhage as shown in Figure 1. Aortic and right atrial blood pH, PCO₂, and PO₂ were measured in a 0.5-ml blood sample, using a Stat Profile Ultra analyzer (NOVA Biomedical Co., Waltham, MA).

View larger version (82K): [in this window] [in a new window]   Figure 1.   Mean aortic pressure (MAP), end-tidal PCO2 (EtPCO2), and cardiac index (CI) before, during, and after hemorrhage in 10 animals compared with 8 controls. Values represent means ± standard deviation. p < 0.05, *p < 0.01 versus baseline. The numbers in parentheses by each data point indicate if one or more animals succumbed.

After withdrawal of blood for laboratory measurements, an equal amount of blood from an anesthetized donor rat was injected into the left femoral vein as previously described (10).

Measurements were reported as mean ± standard deviation (SD). Baseline measurements between groups were compared by ANOVA. Time-based measurements within groups were compared by repeated ANOVA measurements. A probability value of less than 0.05 was considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Baseline measurement for both hemorrhagic and septic animals and their controls were within the physiological ranges previously reported (10). These are summarized in Tables 1 and 2.

Hemorrhagic Shock

During the 120-min interval of bleeding, mean arterial pressure (MAP) decreased from an average of 138 to 49 mm Hg and cardiac index (CI) decreased from 290 to 77 ml/kg/min. EtPCO₂ decreased from 35 mm Hg to approximately 13 mm Hg (Figure 1). Sublingual PCO₂ increased from 45 to 125 mm Hg. Gastric PCO₂ increased from 46 to 87 mm Hg. As in earlier studies, there was a striking increase in blood lactate (LAC) from 0.9 to 15.8 mmol/L during the same interval. The oxygen saturation of mixed venous blood decreased from 75 to 30% (Table 1). Following reinfusion, PslCO₂ and PgCO₂ returned to near-baseline level over the ensuing 30 min, but there was a delay in the reversal of arterial blood lactate (Figure 2). There were no significant changes in control animals during the same intervals.

View larger version (82K): [in this window] [in a new window]   Figure 2.   Comparison of sublingual PCO2 (PslCO2), gastric PCO2 (PgCO2), and arterial lactate (LAC) before, during, and after reversal of hemorrhage. All values represent means ± standard deviation. p < 0.05, *p < 0.01 versus baseline. The numbers in parentheses indicate the same as in Figure 1.

Individual values of PslCO₂ and other parameters were then correlated for every 10-min interval from baseline to 180 min. The individual data points which relate PslCO₂ and PgCO₂ are shown in Figure 3. There were significant correlations (p < 0.001) between PslCO₂ and CI (r = 0.74), EtPCO₂ (r = 0.45), LAC (r = 0.59), and PgCO₂ (r = 0.71).

View larger version (18K): [in this window] [in a new window]   Figure 3.   Correlation between sublingual PCO2 (PslCO2) and gastric PCO2 (PgCO2).

Septic Shock

Mean arterial pressure decreased from an average of 146 to 110 mm Hg over 12 h. End-tidal PCO₂ decreased from 40 to 17 mm Hg and cardiac index decreased from 291 to 164 ml/kg/ min (Figure 4). Sublingual PCO₂ increased from 45 to 69 mm Hg with this less severe insult. Arterial blood lactate increased from 1 to 8 mmol/L during the same interval (Figure 5). No significant changes were observed in control animals.

View larger version (43K): [in this window] [in a new window]   Figure 4.   Mean aortic pressure (MAP), end-tidal carbon PCO2 (EtPCO2), and cardiac index (CI) during sepsis in five animals compared with three controls. All values represent means ± standard deviation. p < 0.05, *p < 0.01 versus baseline. The numbers in parentheses indicate the same as in Figure 1.

View larger version (35K): [in this window] [in a new window]   Figure 5.   Comparison of sublingual PCO2 (PslCO2) and arterial lactate (LAC) during sepsis. Values represent means ± standard deviation. The stated values represent means ± standard deviation. p < 0.05, *p < 0.01 versus baseline. The numbers in parentheses indicate the same as in Figure 1.

Highly significant correlations (p < 0.001) were demonstrated between PslCO₂ and CI (r = 0.78), EtPCO₂ (r = 0.71), and LAC (r = 0.69).

The sensor was recalibrated immediately after completion of each study. Change in the slope of the two-point calibration was less than 2%. The maximal baseline drift over the 3-h interval was 10%.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A marked increase in PslCO₂ was demonstrated in animals during hemorrhagic shock. The increases in PslCO₂ were closely related to the decreases in cardiac index, mean aortic pressure, and EtPCO₂, and to increases in arterial lactate that followed onset of hemorrhage. Sublingual PCO₂ increased within 15 min after the start of hemorrhage. Increases in PslCO₂ in this model were more prominent than those of PgCO₂ but relative changes were comparable. These differences may reflect, at least in part, the location of PCO₂ sensors. They were implanted within the wall of the stomach for PgCO₂ but on the surface of the sublingual mucosa for PslCO₂ measurement.

In settings of septic shock following infusion of S. aureus (24), increases in PslCO₂ were also closely associated with decreases in cardiac index, EtPCO₂, and arterial lactate. In a study of critically ill human patients, including patients with cardiogenic shock, septic shock, and hemorrhagic shock, there also was a highly significant correlation between PslCO₂ and arterial blood lactate (r = 0.83, p < 0.001) (25).

Present evidence indicates that the source of increased tissue PCO₂ is intracellular buffering of excess hydrogen ions by bicarbonate (8). The excess hydrogen ions are in turn traced to anaerobic generation of excess of lactic acid and degeneration of high-energy phosphate compounds during tissue hypoxia caused by the perfusion failure of circulatory shock. In earlier studies during hemorrhagic shock, gastric blood flow markedly decreased in close parallel with increases in tissue carbon dioxide tension (22, 26).

The diagnosis of the severity of circulatory shock is clinically related to reductions in arterial pressure. However, such diagnoses lose reliability because of adrenergically primed arterial and arteriolar vasoconstriction during the early stages of circulatory shock. The resulting increases in arterial resistance compensate for decreases in cardiac output and thereby delay a fall in arterial pressure (27). Cardiac output serves as a more reliable measurement of severity of hemorrhagic shock, but it is typically more invasive as currently measured and does not apply to septic shock. Measurement of end-tidal carbon dioxide is an attractive, new, noninvasive option. It represents an indirect measurement of pulmonary blood flow and therefore cardiac output, but only in settings in which cardiac output is profoundly reduced (28). It requires more elaborate instrumentation. Gastric tonometry, a useful option, also requires more elaborate and costly procedures together with the use of drugs that will block gastric acid secretion (20, 22).

Since 1964, our group has used arterial blood lactate for measuring the severity of shock (31). It has had two limitations. Measurements are made on either mixed venous or arterial blood, requiring blood sampling from a central venous or arterial site (33). The clinical utility of the measurement to guide management is constrained by the substantial delay in the clearance of lactate after reversal of perfusion failure. This typically requires four or more hours. When liver function is impaired, even greater delays are encountered (20, 32). This contrasts with measurements of tissue PCO₂, including PslCO₂, in which there is a prompt reversal within minutes.

We conclude that PslCO₂ promises to serve as a useful measurement for triage, diagnosis, and estimation of severity of circulatory shock. It has the added potential for use in continuous measurements, so as to allow real-time monitoring. Finally, it is a noninvasive, technically simple, and potentially low-cost option for triage, monitoring, and guiding management of patients in critical care, anesthesia, emergency, and mass casualty settings.