INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Microcirculatory disturbances are a hallmark of sepsis and septic shock (1, 2). Decreased capillary flow and heterogeneity of microvascular perfusion have been implicated in the development of tissue hypoxia, cellular injury, and multiple organ failure. In addition, the paramount importance of gut mucosal perfusion for maintenance of the barrier function of the gastrointestinal tract has become clear. The splanchnic mucosal microvasculature is prone to injury because of its high oxygen requirements and the fact that splanchnic vasoconstriction is a common event in states of shock and related hemodynamic disturbances (3). Although not yet proved in humans, loss of gastrointestinal barrier function may allow the translocation of bacteria and bacterial toxins from the lumenal site into the systemic circulation, thereby initiating or perpetuating septic events, and it is these pathogenic sequelae for which the gut is considered to be the "motor" of septic multiorgan failure (4). In addition, the gut mucosal microcirculation is itself a target of endotoxin-induced events. In animal experiments, persistent intestinal hypoperfusion has been noted to occur in response to endotoxin challenge despite adequate volume resuscitation and normal parameters of systemic hemodynamics (7, 8). This led to speculation that the gastrointestinal tract might be the canary of the body, making early detection of inadequate tissue oxygen delivery feasible (9).

Augmentation of intestinal blood flow has been considered to be a reasonable therapeutic approach to gut mucosal hypoperfusion in sepsis. In the human splanchnic vasculature, dopexamine selectively binds to dopamine 1, dopamine 2, and ₂ adrenoreceptors, which in the absence of - and only moderate ₁-adrenergic receptor properties mediate vasodilatation (10, 11). In patients without sepsis, dopexamine was found to improve blood flow to the intestine (12). In the critically ill, however, the effect of dopexamine on intestinal perfusion still remains unclear and no direct measurements of mucosal oxygenation have been performed so far.

Against this background it is evident that diagnostic tools for monitoring splanchnic mucosal perfusion in critically ill patients and those at risk of developing multiorgan failure are highly desirable. Gastric tonometry was the first technique used to address this issue under clinical conditions (13). We now describe a new approach in this field, employing microlightguide reflectance spectrophotometry for direct assessment of microvascular hemoglobin oxygenation and its spatial distribution (absolute data) and the capillary hemoglobin concentration (relative data) in the mucosa of the upper gastrointestinal tract. We noted profound differences between healthy controls and patients in septic shock, the latter being characterized by overall lowered values of mucosal oxygenation and hemoglobin concentration, concomitant with apparent heterogeneity of regional oxygen delivery, and the appearance of severely hypoxic microareas. Short-term infusion of dopexamine was found to improve gastrointestinal mucosal oxygenation in patients with sepsis, although whole-body oxygen uptake did not change. Spectrophotometry is suggested as a new tool for assessment of regional oxygen delivery in the mucosa of the upper gastrointestinal tract in intensive care patients, and for testing its responsiveness to pharmacological interventions.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Population

A total of 15 patients fulfilling the sepsis criteria of Bone and coworkers (17) was investigated (8 men, 7 women; age, 54.9 ± 14.6 yr). Clinical data and underlying diseases are listed in Table 1. All patients suffered from septic shock, required mechanical ventilation owing to respiratory failure, and were recruited from the intensive care unit of the Department of Internal Medicine of the Justus-Liebig-University (Giessen, Germany). Investigations were carried out with arterial oxygenation being in the normal range. Anesthesia was performed by a continuous infusion of fentanyl and midazolam. The control group consisted of 11 healthy volunteers (7 men, 4 women; age, 46.3 ± 23.8 yr) not taking regular medication. Control subjects were sedated by intravenous bolus injection of 5 mg of midazolam. All patients and volunteers were fasted for 12 h and received 150 mg of ranitidine orally twice a day (controls) or 50 mg of ranitidine intravenously twice a day (patients) starting 24 h before investigations were performed. All patients were receiving vasopressors (norepinephrine [range, 0.14-1.14 µg/kg/min; mean, 0.46 ± 0.3] or epinephrine [range, 0.05-0.72 µg/kg/min; mean, 0.23 ± 0.21]). None of the patients received dopamine or dobutamine.

Monitoring of Hemodynamics and Oxygen Transport

Routine clinical monitoring of patients included the use of a thermodilution pulmonary artery catheter and a femoral or radial artery catheter. These were used for measurement of mean arterial pressure (a), central venous pressure (Pcv), and pulmonary artery occlusion pressure (Ppao). Heart rate was monitored continuously. Cardiac output (CO), referred to the body surface as cardiac index (CI), was measured by use of the thermodilution technique. Arterial and mixed venous blood samples were taken for measurement of oxygen and carbon dioxide tension (Pa_O2, Pv_O2, Pa_CO2, Pv_CO2), oxygen saturation (Sa_O2, Sv_O2) and pH (NOVA State Profile 5; Nova Biomedical Waltham, MA). Blood for measurement of hemoglobin concentration (Hb) and lactate was withdrawn from the arterial cannula. Arterial oxygen content (Ca_O2) and mixed venous oxygen content (Cv_O2) were calculated from the standard equations:

(1)

(2)

The oxygen delivery index (DI_O2) and oxygen uptake index (VI_O2) were calculated from the standard equations:

(3)

(4)

Systemic vascular resistance (Rsv) was calculated from

(5)

where 80 is the conversion factor for converting Wood units into dyn · s · cm⁵.

Gastric Tonometry

Gastric intramucosal pH (pHi) was measured using a gastric tonometer (Trip NGS-catheter; Tonometrics, Hopkinton, MA). Correct positioning of the balloon was confirmed by radiography. For each measurement of gastric intramucosal pH, the silicone balloon was filled with 2.5 ml of phosphate-buffered solution (NaH₂PO₄, 44 mmol; Na₂HPO₄, 6 mmol; adjusted to pH 7) as described previously (18, 19). After an equilibration time of 60 min the solution was sampled from the balloon for measurement of PCO₂, in parallel with an arterial blood sample for assessment of bicarbonate (ABL330; Radiometer, Copenhagen, Denmark). A steady state PCO₂ was calculated using the correction factor given for an equilibration time of 60 min (Tonometrics). Gastric pHi was calculated by applying a modification of the Henderson-Hasselbalch equation, as follows:

(6)

where bicarbonate is the bicarbonate concentration obtained from arterial blood, 0.03 represents the solubility of CO₂ in plasma at 37° C, and 6.1 is the pK of the HCO₃/CO₂ system in plasma at 37° C. In addition, the PCO₂ gradient (PCO₂), the pH gradient (pH), and the standardized pHi (pHi_s) were calculated as follows (20):

(7)

(8)

(9)

In contrast to pHi, these variables are assumed to be less influenced by respiratory or metabolic acid-base disturbances frequently found in critical illness.

Reflectance Spectrophotometry

The technique of reflectance spectrophotometry (Figure 1) for assessment of gastric mucosal blood flow characteristics was first described by Sato and colleagues (21). Spectrophotometric parameters were found to reflect gastroduodenal mucosal perfusion when compared with hydrogen gas clearance as the reference method (22, 23). In this study, an Erlangen microlightguide spectrophotometer (EMPHO II; Bodenseewerk Gerätetechnik, Überlingen, Germany), introduced by Frank and coworkers (24), was used for measurement of gastric mucosal microvascular hemoglobin oxygen saturation (Hbi_O2) and relative hemoglobin concentration (rel Hb_conc) as described before in animal experiments (22, 23). Briefly, light from a xenon high-pressure arc lamp was transmitted to the mucosal surface via a single highly flexible microlightguide with a diameter of 250 µm. Backscattered light was collected by six identical microlightguides (each 250 µm in diameter) arranged around the circumference of the illuminating lightguide. The illuminating and detecting microlightguides were encased in a flexible rubber tube. The aperture angle of the 250-µm lightguides used for the investigations was 60°, giving a numerical aperture of 0.5. The "visual field" of the lightguide bundle was ~ 0.09 mm², and the maximum tissue depth from which backscattered light was received was ~ 250 µm. The backscattered light was split into its spectral components by a rotating interference bandpass filter disk (502-628 nm). This filter disk allowed a sampling rate of 100 spectra/s. Owing to the small catchment volume together with a high recording frequency, reflectance spectrophotometry allowed assessment of spatial heterogeneity of perfused mucosal capillaries. The monochromated light was transmitted to a photomultiplier. The signal was then fed to an analog-to-digital converter and transferred to a personal computer. The recorded spectra were balanced against a standard white reference produced by a mirror and a dark reference. Absolute data for Hbi_O2 and relative data for Hb_conc in the mucosa were calculated by use of an algorithm described previously (24). For determination of hemoglobin oxygenation, spectra with known hemoglobin oxygenation were mixed by the computer, starting with the spectrum for fully oxygenated (two peaks) and fully deoxygenated (one peak) hemoglobin, and were fitted with the spectrum actually measured. By an iteration procedure and successive comparison the actual hemoglobin oxygenation was determined. The mucosal hemoglobin concentration representing capillary blood volume was evaluated from the signal intensity at the isobestic points of the hemoglobin spectral curve. Relative hemoglobin concentration was calculated by comparing all spectra with a standard derived from healthy controls and arbitrarily set at one.

View larger version (17K): [in this window] [in a new window]   Figure 1.   Erlangen microlightguide spectrophotometer (EMPHO II) and target areas for measurements in the upper intestinal tract. (A) light source; (B) rotating interference bandpass filter disk; (C ) photomultiplier; (D) personal computer; (E ) flexible microlightguides encased in a flexible rubber tube; (F ) cross-section through bundle of microlightguides, one illuminating lightguide surrounded by six detecting lightguides; (G) tissue.

The highly flexible microlightguides were introduced through the operation canal of a gastroscope and gently attached to the mucosa under visual control, preventing any local pressure. All measurements were performed in macroscopically normal mucosa. The proper placement of the probe was verified by an independent observer and ascertained for each measurement. If verification was not possible, the measured spectra were discarded. During the measurement procedure, the light source of the gastroscope was turned off. During ongoing measurements, results were blinded for the investigator. In each individual, spectrophotometric measurements were performed in five distinct areas of the upper intestinal tract (fundus, antrum, upper pylorus, lower pylorus, and duodenal bulb). In each area, 1,000 spectra were taken. For each patient, mean values of Hbi_O2 and rel Hb_conc were calculated by combining all 5,000 spectra. Furthermore, Hbi_O2 values from each group were depicted as histograms representing the heterogeneity of the intracapillary hemoglobin oxygenation.

Study Protocol

The study was approved by the institutional ethics committee, and informed consent was obtained from the closest relatives. Patients with a history of esophageal or gastric disease as well as patients with a history of former esophageal or gastric surgery were not included in the study. Serial observations of hemodynamics, gas exchange, and ventilatory parameters were obtained at intervals of 60 min. During the study period, the rate of infusion of any drug and the setting of the respirator were kept constant. Initial tonometric measurement of intramucosal PCO₂ was performed employing an equilibration time of 60 min. After drawing the solution from the tonometer balloon and measuring PCO₂ immediately from the cooled sample, a gastroscope was introduced and spectrophotometric measurement was performed. At the end of gastroscopy, all air was eliminated thoroughly from the gastric lumen. After finishing this first series of measurements, a continuous infusion of dopexamine was started at a dosage of 2 µg/kg/min. Patients with heart rate > 130 beats/min, history of coronary artery disease, or ventricular arrhythmias were excluded from dopexamine administration. In those patients receiving dopexamine, measurements were repeated 1 h after onset of infusion.

In controls, parameters of systemic hemodynamics an oxygen transport were not assessed, and no dopexamine was administered. Tonometric and spectrophotometric measurements were performed as described for patients.

Statistics

Data were analyzed using a personal computer and commercially available software (SPSS 6.1; SPSS Inc., Chicago, IL). All results are given as means ± SEM. Statistical analysis was performed by testing normal distribution first (Kolmogorov-Smirnov test), followed by a pairwise comparison of the pre- and postdopexamine data either with Student's t test for paired samples (in case of normal distribution) or with Wilcoxon's matched-pairs signed-ranks test. For comparison between control group and patient groups a one-way analysis of variance (ANOVA) was performed. For describing distribution types of Hbi_O2 histograms skewness and kurtosis were calculated. Distributions of histograms were considered to be significantly different when confidence intervals for these two parameters displayed no overlap. The null hypothesis was rejected for p < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient demographics and underlying diseases are shown in Table 1. Total mortality rate for the 15 patients was 66%. During the period of measurement, increased heart rate and cardiac index as well as decreased mean arterial pressure and systemic vascular resistance, indicating hyperdynamic sepsis, were noted throughout (Table 2). The administration of catecholamines to all patients to maintain arterial pressure, in companion with elevated lactate levels (4.7 ± 1.0 versus 1.6 ± 0.1 in healthy controls; p < 0.001), reflected septic shock. Intramucosal pH in septic patients under baseline conditions was markedly decreased as compared with controls (7.16 ± 0.02 versus 7.36 ± 0.01) (Table 3). Standardized pHi averaged 7.25 ± 0.01. Correspondingly, gastric intramucosal PCO₂ and the intramucosal-arterial PCO₂ gap were clearly elevated. The mean gastric mucosal hemoglobin oxygenation, averaging 70.3% ± 2.1 in controls, was highly significantly decreased to 51% ± 1.6 in the patients with sepsis (Table 4). Moreover, the distribution of Hbi_O2 data demonstrated a marked tailing of the histogram to severely hypoxic values, with 19% of data below 40% hemoglobin saturation (Figure 2). In contrast, all single values were above 40% in the controls, with a narrow dispersion of Hbi_O2 data. Similar to Hbi_O2, the relative mucosal hemoglobin concentration was significantly decreased in patients with sepsis as compared with the healthy controls (0.78 ± 0.05 versus 1.00 ± 0.04; Table 4), and the histogram of rel Hb_conc again displayed enhanced scattering to low values (data not given). Confidence intervals for skewness and kurtosis of both Hbi_O2 and rel Hb_conc histograms confirmed a significant difference in distribution between controls and patients with sepsis (Figure 2).

View larger version (20K): [in this window] [in a new window]   Figure 2.   Gastric mucosal HbO2 (HbiO2) histograms in normal controls (n = 11) (A) and septic patients (n = 15) (B). Data are given as mean ± SEM. In normal control subjects, values for skewness and kurtosis were 0.04 ± 0.05 and 0.76 ± 0.1, respectively. These values were significantly different from the corresponding data in the septic patients (0.823 ± 0.04 and 0.88 ± 0.09, respectively).

Dopexamine was administered to 10 of 15 patients with sepsis. The drug was well tolerated and no adverse effects were observed. It resulted in a significant increase in heart rate and CI, concurrent with a significant rise in DI_O2. In contrast, VI_O2 was not significantly altered in response to this vasoactive agent (Table 2). In 7 of 10 patients, gastric pHi increased under infusion of 2 µg/kg/min dopexamine (Figure 3). The mean value increased from 7.15 ± 0.03 to 7.18 ± 0.03 (p < 0.05). In parallel, arterial pH increased in response to dopexamine (from 7.30 ± 0.02 to 7.33 ± 0.02, p < 0.01), reflecting (statistically not significant) changes in systemic PCO₂ and bicarbonate. Consequently, variables excluding influences of systemic acid-base disturbances such as intramucosal-arterial PCO₂ gradient, pH gradient, or standardized pHi demonstrated no significant change during dopexamine application (Table 3). Dopexamine infusion caused a significant increase in the mean values of both Hbi_O2 and rel Hb_conc (Table 4). The impact on mucosal hemoglobin oxygenation was also reflected by the Hbi_O2 histogram, displaying a shift to the right of data as compared with the predopexamine baseline. The Hbi_O2 distribution was, however, still not fully normalized when compared with the healthy volunteers (Figure 4). Postdopexamine rel Hb_conc values were significantly increased as compared with baseline and even surpassed those of healthy controls (Table 4).

View larger version (16K): [in this window] [in a new window]   Figure 3.   Gastric intramucosal pH (pHi) in septic patients before and during administration of 2 µg/kg/min dopexamine (n = 10). Each line represents an individual patient; closed diamonds = mean ± SEM.

View larger version (20K): [in this window] [in a new window]   Figure 4.   Gastric mucosal HbO2 (HbiO2) histograms from septic patients (n = 10) before (A) and during infusion (B) of 2 µg/kg/min dopexamine. Data are given as mean ± SEM. Skewness and kurtosis before drug administration were 0.90 ± 0.05 and 0.64 ± 0.1, respectively. During dopexamine administration, skewness was 0.86 ± 0.05 and kurtosis was 0.7 ± 0.1.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The current use of microlightguide reflectance spectrophotometry for assessment of mucosal hemoglobin oxygenation and concentration in the upper gastrointestinal tract of septic patients demonstrated severe microcirculatory disturbances in this compartment as compared with healthy controls. In essence, mean values of both Hbi_O2 and rel Hb_conc were markedly lowered in sepsis despite high normal whole-body oxygen delivery, together with a tailing of the histograms to very low values, suggesting pronounced restriction and heterogeneity of mucosal perfusion. The defect in mucosal oxygenation responded favorably to short-term infusion of dopexamine, not reflected by significant changes in standardized pHi. Spectrophotometry is offered as a new technique for characterization of regional oxygen delivery in the mucosal layer of the upper gastrointestinal tract.

Reflectance spectrophotometry with the Erlangen microlightguide spectrophotometer (EMPHO II) is a powerful technique for assessment of hemoglobin oxygenation in a small catchment volume and has been evaluated under experimental conditions (24). We combined this technique with endoscopic access to the gastric and duodenal mucosa. This allowed attachment of the microlightguide to the mucosa under direct visual control to prevent major compression of the mucosa and thereby the microvessels. As maximum tissue penetration of the xenon lamp-derived light is estimated to reach approximately 250 µm, the information transported by the backscattered light is restricted to the microvessel-containing mucosal layer. Owing to the small catchment volume (approximately 90 × 90 × 250 µm), in combination with some movement of both lightguide and mucosal layer, recorded data do not represent a single point, but a small area. Together with the high analysis frequency (100 spectra/s), this allows spatial resolution for assessment of homo- versus heterogeneity of tissue oxygenation. In contrast to the absolute data represented by Hbi_O2 values of tissue hemoglobin concentration, measured at the isobestic points of the hemoglobin spectral curve, are relative, referring to an arbitrarily set standard. This is because the exact volume of tissue contributing to the measurements is unknown. Nevertheless, the data give an estimate of capillary blood volume that is independent of the value of oxygen saturation obtained from the spectral analysis.

As anticipated for well perfused mucosa under physiological conditions, a narrow, near-Gaussian distribution of capillary hemoglobin saturation was noted in the healthy volunteers, centered around a high mean value of ~ 70%. There was no major variability between the five different observation areas in gastric and duodenal mucosa. In contrast, the data on mucosal hemoglobin saturation were highly significantly decreased to a mean value of ~ 50% in the patients with hyperdynamic sepsis. As the microvascular hemoglobin oxygenation reflects the balance between regional oxygen delivery and oxygen uptake, decreased delivery or enhanced uptake or a combination of both might underlie this finding. For several reasons, a reduction in regional oxygen delivery owing to microvascular perfusion abnormalities is the most plausible explanation for the observed drop in hemoglobin oxygenation. First, there is no experimental evidence supporting a markedly enhanced mucosal oxygen consumption in sepsis; rather, this may be decreased (see below). Microvessel perfusion abnormalities, on the other hand, have been found to occur in the gastrointestinal mucosa in experimental models of endotoxemia and sepsis, even under conditions of normotension and conserved whole-body oxygen delivery (7, 8, 27). Second, the decrease in mean hemoglobin oxygenation in patients with sepsis was accompanied by a marked tailing in the histogram to extremely low, even nearly anoxic (< 10% saturation) values, strongly suggesting a pattern of severe heterogeneity of microvessel perfusion. Third, the decrease in mucosal hemoglobin oxygenation was accompanied by a reduction in the relative mucosal hemoglobin concentration, which is an independent variable. Although in cannot be decided whether reduced mucosal capillary density, as demonstrated experimentally by intravital microscopy in rats and dogs with normotensive endotoxemia (28, 30), or reduced average vessel diameter of a normal number of capillaries (8) underlies this finding, both interpretations correspond to an overall reduction in mucosal perfusion. And fourth, a prompt response to dopexamine, in the form of an increase in mean hemoglobin oxygenation and a shift to the right of the Hbi_O2 histogram were noted, as anticipated for a vasoactive drug targeting splanchnic vasodilatation. Notably, the decrease in mucosal hemoglobin oxygen saturation, including the appearance of microdomains with extremely low (< 30%) tissue oxygenation, was observed in regions with macroscopically normal mucosa, and it occurred in the presence of conserved and even high normal whole-body oxygen delivery. These findings, thus, again support the notion that macrohemodynamic variables are not suitable to reflect the oxygenation state of areas of interest, such as the gastrointestinal mucosa, under conditions of sepsis. Although effects of baseline medication (vasopressors, sedation) interfering with microvascular hemoglobin oxygen saturation and concentration are not completely ruled out in the present study, the most likely mechanism of the demonstrated abnormalities is septic microcirculatory maldistribution. While norepinephrine demonstrated a beneficial effect on splanchnic oxygenation in patients with sepsis (31), little to nothing is known about the effects of epinephrine or sedation on splanchnic perfusion in septic shock states.

Dopexamine has been recommended in the treatment of septic shock because it leads to a dopaminergic receptor- mediated vasodilatation (10, 11). Nevertheless, only a few studies have examined the effects of dopexamine on intestinal perfusion in sepsis. In rats with normotensive endotoxemia subjected to videomicroscopy of the exteriorized ileum, dopexamine was found to antagonize the lipopolysaccharide (LPS)- induced villous arteriolar vasoconstriction, thereby enhancing villous blood flow (32). In the current study, short-term dopexamine infusion clearly improved, but did not normalize, the mucosal oxygenation in patients with sepsis: mean Hbi_O2 and rel Hb_conc values increased, and the corresponding histograms were shifted to the right as compared with the preinfusion baseline. These data compare favorably with results from patients with systemic inflammatory response syndrome (SIRS), in whom dopexamine was noted to increase gastric mucosal pH, indocyanine green disappearance, and hepatic metabolism of lidocaine, all assumed to reflect enhancement of splanchnic blood flow (33, 34). In parallel with these microcirculatory changes, cardiac index, heart rate, and DI_O2 increased, although to a minor extent. Thus it remains open to discussion, whether the dopexamine-induced changes in mucosal oxygenation were promoted by a primary effect of the drug on the splanchnic vasculature, or were secondarily affected via the overall hemodynamic alterations. As the latter were comparably moderate, and changes in baseline catecholamines (epinephrine, norepinephrine) forwarded comparable macrochanges in the absence of significant microcirculatory changes (interventions undertaken outside the observation period; data not given), the authors assume predominance of the local dopexamine effect, but definite proof of this issue is still missing.

The responsiveness to dopexamine was much less obvious in the gastric mucosal acid-base state. As anticipated, decreased values of pHi and increased values of Pi_CO2 were noted in the patients with sepsis under baseline conditions. Dopexamine infusion caused a moderate improvement in both variables, but a detailed analysis of values more robust to systemic changes in the acid-base state (PCO₂, pH, pHi_s) suggested that the pHi alterations must be largely ascribed to systemic dopexamine effects (increase in bicarbonate and pHa, decrease in Pa_CO2). This finding thus supports an observation in patients undergoing cardiopulmonary bypass, in whom a dopexamine-induced increase in pHi was also ascribed to systemic acid-base changes in response to the vasoactive agent (35). Several reasons may underlie the discrepancy between the presently described dopexamine-induced improvement of mucosal oxygenation and lack of impact on the acid-base state in this compartment. First, the observation time might be too short for a "translation" of improved tissue oxygenation into a decay of tissue acidosis. Moreover, the tonometric technique itself requires an equilibration time that might not allow a timely assessment of increased intramucosal pH or allow an immediate feedback on intervention. Therefore further studies should employ air tonometry, a technique that requires much less time for PCO₂ equilibration between mucosa and tonometer. Second, the severity of tissue damage might be too severe to allow rapid recovery of aerobic metabolism in spite of improvement in regional oxygen delivery. This reasoning might explain why an increase in pHi in response to dopexamine was observed in patients with SIRS in the absence of shock (33, 34), but not in the currently investigated patients with sepsis, all suffering from shock and lactate formation. And third, the microcirculatory disturbances might have been acompanied or followed by derangements in cellular energy metabolism, nonresponsive to changes in the regional oxygen supply. Uncoupling of oxidative phosphorylation (as, e.g., described in response to LPS [36]), inhibition of mitochondrial respiration, and reduced availability of substrates have all been considered in this context (37). Further studies will be necessary to elucidate in more detail the acid-base alterations in the mucosa of the upper gastrointestinal tracts of patients with sepsis.

In conclusion, the currently described spectrophotometric technique allows the assessment of gastrointestinal mucosal microcirculatory disturbances under clinical conditions. In patients presenting a state of hyperdynamic septic shock, decreased oxygenation in this compartment in spite of high normal whole-body oxygen delivery was noted, together with apparent heterogeneity of perfusion and the appearance of severely hypoxic microdomains. This observation thus extends experimental findings of gastrointestinal intraorgan perfusion mismatch (mucosa versus muscularis) in endotoxemia (7), suggesting abnormalities of flow distribution within the mucosal layer itself. Such a maldistribution may underlie oxygen extraction defects in sepsis and must be assumed to compromise the integrity of the gastrointestinal mucosal barrier. The short-term effects of dopexamine, improving gastrointestinal mucosa oxygenation but not regional acid-base balance, require further clarification.