Performance of the partial CO2 rebreathing technique under different hemodynamic and ventilation/perfusion matching conditions

Abstract

The partial CO2 rebreathing technique has been demonstrated to accurately measure the effective pulmonary capillary blood flow (PCBF) in different clinical situations. Usually, PCBF is calculated from changes in CO2 elimination (VCO2) and end-tidal partial pressure of CO2 (PetCO2 ), which can be obtained noninvasively. In this study, we investigated the performance of the partial CO2 rebreathing technique under different conditions of ventilation/perfusion matching and hemodynamic states. In addition, we investigated whether the determination of arterial blood gases combined with mathematical modeling of gas exchange can improve the performance of this method. Prospective, controlled animal laboratory study. Experimental research facility of a university hospital. Sixteen female sheep weighing 45-55 kg. Cardiac output and ventilation/perfusion matching were manipulated during three phases: phase I, variation in cardiac output to achieve normal, hyperdynamic and hypodynamic states; phase II, increase of alveolar deadspace and variation in cardiac output; phase III, lung injury and increased alveolar deadspace. Partial CO2 rebreathing maneuvers were performed to obtain variations in VCO2 and PetCO2 between a nonrebreathing (NR) and a rebreathing (R) period. PCBF was measured by the rebreathing method as PCBF = -DeltaVCO2/f(Pc'CO2 (R), Pc'CO2(NR), Hb), where f is the CO2 dissociation curve in blood, Pc'CO2 is the end-capillary partial pressure of CO2, Delta is the variation between NR and R periods, and Hb is hemoglobin concentration. Pc'CO2 was estimated from PetCO2 according to two algorithms. In the so-called "noninvasive algorithm," Pc'CO2 = PetCO2, with PetCO2(NR) and PetCO2(R) being determined as the mean PetCO2 value of the last 60 secs preceding rebreathing and within 15-30 secs of rebreathing, respectively. In the "semi-invasive algorithm," Pc'CO2(NR) was estimated as the PaCO2, and Pc'CO2(R) was estimated as follows: First, a monoexponential function was fitted to PetCO2 values during rebreathing and the asymptote represented PetCO2(R). Second, the Pc'CO2(R) to PetCO2(R) difference was calculated by means of a bicompartmental, tidal model of gas exchange, which showed that such differences decrease with the degree of rebreathing. PCBF values obtained with both algorithms were compared with thermodilution cardiac output minus intrapulmonary shunt flow. Bias and precision calculations with the noninvasive algorithm in phases I, II, and III were, respectively, -1.0 +/- 1.9, -2.1 +/- 2.6, and -2.4 +/- 1.2 L/min. The semi-invasive algorithm had an overall better performance in the phases investigated: -1.2 +/- 1.9, -0.6 +/- 2.0, and -0.2 +/- 3.0 L/min, respectively. The noninvasive algorithm showed a slight tendency to overestimate lower reference PCBF values and, importantly, to underestimate higher PCBF values in all three phases (r = -.66, p<.0001; r = -.75, p<.001; r = -.60, p<.0001, respectively). A similar figure was observed with the semi-invasive algorithm in phase I (r = -.47, p<.01) but not in phases II and III (r = -.1, p=.54; r =.62, p<.001, respectively). Although PCBF is systematically underestimated during hyperdynamic cardiac output states and high alveolar deadspaces, the performance of the partial CO2 rebreathing technique can be improved by means of arterial blood gas sampling and an algorithm that takes in account the effects of nonequilibration of PetCO2 during rebreathing and the variation of Pc'CO2 to PetCO2 differences from the nonrebreathing to the rebreathing period. Such an algorithm may prove useful under moderately increased alveolar deadspace and normal to hypodynamic cardiac output states.