Intramucosal–arterial PCO₂ gap fails to reflect intestinal dysoxia in hypoxic hypoxia

Tonometry is one of the few clinical tools available for the monitoring of tissue oxygenation [1]. Decreases in gastrointestinal intramucosal pH (pH_i) have usually been considered as indicators of dysoxia [2,3,4,5]; that is, as heralds of insufficient O₂ to meet tissue demands. Recently, the intramucosal–arterial PCO₂ gradient (ΔPCO₂) has been claimed to be a better marker of gastrointestinal mucosal state of oxygenation [6]. PCO₂ can increase in intestinal lumen by two mechanisms [7]. One is by bicarbonate buffering of protons from the breakdown of high-energy phosphates and metabolic acids generated anaerobically, such as lactate, in which case increased PCO₂ would represent tissue dysoxia. Alternatively, in an aerobic state, it might be the result of hypoperfusion and decreased washout. In this latter case, oxygen metabolism could be preserved if the flow were adequate.

Grum et al. [2] found that pH_i and intestinal oxygen uptake (VO₂) were correlated in ischemia, in hypoxemia, and in a combination of both. However, in hypoxemic experiments, neither VO₂ nor pH_i fell. Hence, the value of tonometry in hypoxemia remains uncertain. If a rise in ΔPCO₂ reflected only a decrease in blood flow, this gradient might not be altered in hypoxemia, in which cardiac output (CO) is usually maintained and even increased. Our hypothesis was that ΔPCO₂ would not be modified in hypoxic hypoxia (HH) with preserved blood flow.

Hemoglobin concentration, and arterial, mixed venous, and mesenteric venous blood gases and oxygen saturations in basal conditions, and during IH and HH and in the sham group are shown in Table 1.

Systemic and intestinal supply dependence was induced in both the IH and HH groups. There were no significant changes in systemic and intestinal DO₂ and VO₂ in the sham group (Figure 1). In the IH group, supply dependence appeared with critical decreases in CO and superior mesenteric artery blood flow (0.104 ± 0.024 versus 0.048 ± 0.006 l/min per kg, and 0.664 ± 0.227 versus 0.258 ± 0.082 l/min per kg, respectively; P < 0.0001). In the HH group it was due to a progressive decrease in arterial oxygenation. CO and intestinal blood flow were maintained (0.446 ± 0.085 versus 0.431 ± 0.140 ml/min per kg, respectively; not significant), owing to the administration of normal saline (median 630 ml; range 20–1310 ml). Arterial and intra-mucosal pH fell significantly in the IH and HH groups. In the HH group it was primarily related to systemic respiratory and metabolic acidosis (Tables 1 and 2), because ΔPCO₂ did not increase. In addition, systemic and intestinal venoarterial PCO₂ gradients were not modified. CO₂ content differences also did not change (Table 2 and Figure 2). In contrast, in the IH group, ΔPCO₂, systemic and intestinal venoarterial PCO₂, and CO₂ content gradients increased significantly (Table 2 and Figure 2). In the sham group, CO₂ gradients and pH_i remained unchanged.

Increased mucosal intestinal PCO₂ is used as a tool to detect delivery can no tissue dysoxia, the condition in which O₂ longer sustain O₂ uptake [11]. A great body of literature supports the role of intestinal PCO₂ as an early marker of dysoxia and regional hypoperfusion. Early studies considered pH_i as the reference parameter. Recently, some investigators have claimed that intramucosal PCO₂, the variable actually measured by the tonometer, and ΔPCO₂ could more adequately reflect mucosal oxygenation [6,12]. pH_i is a calculated variable, from the Henderson–Hasselbach equation, with the assumption that arterial bicarbonate is representative of intra-mucosal bicarbonate. In a steady state, both values might be similar. However, in rapidly changing physiological situations, differences between arterial and mucosal CO₂ might arise owing to slow CO₂ equilibrium kinetics [13]. Therefore, pH_i values calculated from tonometry might differ from those directly measured with tissue electrodes [14]. Moreover, acid–base states could influence in the absence of pH_i altered mucosal oxygenation. As a result, the acid–base status of arterial blood will be reflected in both mucosal pH_i and PCO₂ [6,14]. In our experiments, pH_i fell progressively during hydrochloric acid-induced lung injury and decreased DO₂, reflecting ongoing systemic respiratory and metabolic acidosis. However, ΔPCO₂ remained unchanged.

Another issue that has been discussed extensively is the relative impact on mucosal PCO₂ of anaerobic production of CO₂ in comparison with decreased washout of aerobically generated CO₂ during low flow states. Many investigators [15,16,17] have ascribed increased PCO₂ found in shock states to continuing aerobic CO₂ production with decreased elimination; that is, to 'respiratory acidosis'. However, Schlichtig and Bowles [7] showed evidence supporting the role of intramu-cosal PCO₂ as a marker of dysoxia in extreme hypoperfusion, when VO₂ falls. In a dog model of cardiac tamponade, they demonstrated that mucosal PCO₂ could rise because of anaerobic CO₂ production below the critical DO₂. These conclusions were drawn by using the Dill nomogram, which can theoretically detect anaerobic CO₂ production from a comparison of the measured (%HbO_2,v) and calculated (%HbO_2,v^Dill) venous oxyhemoglobin, within a given venous PCO₂ value. Because venous PCO₂ is considered to be representative of tissue PCO₂, Schlichtig and Bowles made the calculation with its intestinal equivalent, intramucosal PCO₂. If %HbO_2,v^Dill is lower than the measured %HbO_2,v, anaerobic production of CO₂ might be assumed. Similar values would represent aerobic CO₂ generation. Notwithstanding the original contribution of Schlichtig and Bowles [7] to the analysis of these topics, the use of low flow to produce critical oxygen delivery and falling VO₂ has been signaled as a potential confounding factor [18].

We studied these issues in a model of HH with preserved flow, because it allows a clear discrimination between hypoxia and hypoperfusion. There have been attempts to analyze pH_i behavior in HH, but critical intestinal DO₂ was not attained, and intestinal VO₂ and pH_i remained unchanged [2]. Our model consisted of an acute lung injury produced by endotracheal instillation of hydrochloric acid that rapidly generated severe hypoxemia, shown by the decrease in arterial PO₂ and pH. The acid also enhanced microvascular permeability [19,20,21], demonstrated by increased requirements of saline solution to maintain intestinal blood flow and by the increase in hemoglobin levels. However, other mechanisms could be acting to preserve blood flow, such as tachycardia and enhanced left ventricular contractility [22]. Deep arterial hypoxemia caused significant reductions in systemic and intestinal DO₂, but systemic and intestinal blood flow were preserved and hemoglobin concentration increased. Despite the increase in systemic and intestinal oxygen extraction, systemic and intestinal VO₂ values decreased, and dependence of O₂ uptake on transport ensued. Dependence of oxygen consumption on transport during HH has been described by Cain et al. in a classical study [23], and it has been considered an indicator of anaerobic metabolism. Additional evidence of tissue dysoxia was the appearance of metabolic acidosis. Cain [24] also showed that there is a correlation between pH and lactate/pyruvate relationship in HH.

Another potential confounding factor that could affect arterial and intestinal PCO₂ and their differences is the shift of the CO₂ dissociation curve. As Jakob et al. [25] have shown, there can be a lack of correlation of CO₂ contents and PCO₂, and, consequently, of their differences. Many determinants of the shifts of the CO₂ dissociation curve, such as changes in pH, in hemoglobin concentrations, and especially in oxygen saturations (the Haldane effect), were present. To discard a possible increase in venoarterial and intramucosal–arterial CO₂ contents without changes in PCO₂ differences in the HH group, we calculated CO₂ content differences. There were no increases in venoarterial and intramucosal–arterial CO₂ content differences during the period of supply dependence, as there were no changes in both PCO₂ differences. Shifts of the CO₂ dissociation curve therefore do not seem to influence our results.

Our model of HH is useful for discriminating the effects of hypoxia and low blood flow, because this last factor was kept constant throughout the experiment. ΔPCO₂ remained stable, although there were signs of anaerobic metabolism. Systemic and venoarterial PCO₂ differences also remained unchanged. Conversely, during supply dependence of VO₂ induced by hemorrhage, ΔPCO₂ and systemic and intestinal venoarterial PCO₂ differences widened, as well as the respective ΔPCO₂ content differences. Moreover, these parameters increased before any change in VO₂, as we have described previously [26]. These results suggest that, at least in our experiments, tissue perfusion is a key determinant of increased ΔPCO₂.

Nevière et al. [27] tested a similar hypothesis in pigs. They compared the effects of diminished blood flow with diminished inspired fraction of oxygen. In IH, ΔPCO₂ increased to 60 mmHg. In HH, ΔPCO₂ increased to 30 mmHg only in the last step of hypoxemia, although mucosal blood flow measured by laser Doppler flowmetry was preserved. The authors concluded that elevated intramucosal PCO₂ indicated local CO₂ generation. However, in the two previous stages of reduced F_iO₂ there was supply dependence, and ΔPCO₂ remained unchanged. In our HH model, ΔPCO₂ was also stable. The differences between our data and those of Nevière et al. [27] could be ascribed to distinct microvascular features of the experimental subjects (pigs and sheep) or to different degrees of hypoxemia. In addition, as Nevière et al. pointed out, some degree of decrease in gut mucosal blood flow and heterogeneity might have been present, because only global microvascular blood flow changes can be assessed by laser Doppler flowmetry. Nevertheless, both studies show that ΔPCO₂ could fail to reflect tissue dysoxia at some time during HH. In results similar to ours, Vallet et al. [28] showed that perfusion is a major determinant of venoarterial PCO₂ difference during critical IH or HH in isolated hindlimb. This gradient increases in ischemia and is preserved in hypoxia.

Venoarterial and intramucosal–arterial PCO₂ gradients are the result of interactions of changes in aerobic and anaerobic CO₂ production, the CO₂ dissociation curve, and blood flow to tissues. During oxygen supply dependence induced by hemorrhage, opposite changes in aerobic and anaerobic CO₂ production are present: aerobic CO₂ production decreases as a consequence of depressed aerobic metabolism, but anaerobic CO₂ production starts because of bicarbonate buffering of protons from fixed acids. Total CO₂ production might not increase, but O₂ consumption falls, so there is an increase in respiratory quotient [29,30]. This increase in VCO₂ relative to VO₂ might generate tissue and venous hypercarbia only in low flow states, in which there is diminished CO₂ removal. Other situations in which intramucosal acidosis could arise with preserved tissue perfusion are reperfusion injury [31] and cytopathic hypoxia generated by endotoxemia [32], with cellular damage and metabolic abnormalities as underlying mechanisms. However, impaired villous microcirculation has been advocated as the causal phenomenon in the latter [33].

To our knowledge, this is the first study showing that ΔPCO₂ fails to mirror intestinal tissue dysoxia. Our findings also suggest that blood flow might be the main determinant of ΔPCO₂. Tonometry seems to be a useful method for monitoring perfusion, with rather limited value in detecting anaerobic metabolism when blood flow is preserved. Additional studies in other models of hypoxic and anemic hypoxia are needed to confirm our findings and to resolve discrepancies between studies.