Relative Importance of Flow versus  Pressure in Splanchnic Perfusion during Cardiopulmonary Bypass in Rabbits

The study was performed in an authorized animal care laboratory according to the French health authority and was approved by the University research committee. Ten New Zealand white rabbits of both sex (weighting 3.0–3.5 kg) were given midazolam 0.5 mg/kg body weight administered intramuscularly as premedication and were then anesthetized with ketamine HCl 75 mg/kg body weight administered intramuscularly (Imalgene; Rhone-Mérieux, Lyon, France). Adequate depth of anesthesia was ensured before any surgical procedures by the absence of pedal and palpebral reflexes. The marginal ear vein was cannulated (20-gauge cannula) for the administration of fluids (hetastarch, 3 ml · kg⁻¹· h⁻¹) and drugs. The central ear artery was also cannulated (20-gauge cannula) and connected to a pressure transducer for measurement of systemic arterial pressure during surgery. All catheters were flushed with heparinized saline 10 IU/ml to prevent clotting during the experiment. A tracheotomy was performed (endotracheal tube size, 3.5 mm OD), and the lungs were ventilated mechanically using a neonatal apparatus (Dräger, Cicero, Lübeck, Germany) with 100% oxygen. Tidal volume was set at 15 ml/kg, and respiratory frequency was set at 35 breaths/min. Ventilation was adjusted to keep the partial pressure of carbon dioxide and p  H in the physiologic range. End-tidal carbon dioxide concentrations were measured by integrated capnography. Body temperature was recorded using an electrical thermometer (Baxter, San Diego, CA) inserted orally into the esophagus, and temperature was maintained at 38.5–39.5°C before CPB by an electrical heating exchanger. Limb lead II of the electrocardiogram was continuously monitored by means of subcutaneous needle electrodes. Continuous anesthesia was maintained with ketamine 25 mg · kg⁻¹· h⁻¹and midazolam 0.125 mg · kg⁻¹· h⁻¹during all procedures. The left carotid artery was cannulated with a 20-gauge cannula and connected to a fluid-filled pressure transducer (Medex Medical SARL, Nantes, France). This catheter was used to take blood samples during the experiment and to measure mean nonpulsatile blood pressure during CPB.

A laparotomy was performed by a midline incision, and hemostasis was adjusted by diathermy. Systemic heparinization was performed (300 IU/kg body weight). The femoral arteries were then ligated for the CPB period, and the distal abdominal aorta was retrogradely cannulated with an 8-French-gauge arterial perfusion cannula (DLP; Medtronic, Grand Rapids, MI) approximately 10 mm above the femoral bifurcation. The aortic cannula was flushed with methylene blue after the end of the measurements to check the integrity and normality of the mesenteric arteries. This technique has been previously described. 9After checking the pulsatility of flow in the arterial cannula, a sternotomy was performed. The right atrium was cannulated with a 16-French-gauge venous perfusion cannula (Stockert, Munich, Germany) via  a purse string suture. Both aortic and venous cannulae were connected to the CPB circuit.

This technique of CPB involved the use of a small neonatal oxygenator (Microsafe; Polystan, Ballerup, Denmark) with a venous reservoir and heat exchanger, and connected with 7-mm tubing. Temperature was maintained by a thermostatic water pump adjusted to 38°C connected to the heat exchanger. Blood-gas concentrations were checked regularly, and 0.42% (wt/vol) sodium bicarbonate (10-ml bolus) was injected to maintain base excess > −5 mM and p  H > 7.30 to avoid acidosis from femoral occlusion. Circuit priming consisted of 6.5% (w/v) hetastarch 300 ml, 0.42% sodium bicarbonate 20 ml, calcium chloride 250 mg, heparin 1,000 IU, and 100 ml fresh whole rabbit blood. One dose of antibiotic (cefazoline 1 g) was added to mimic clinical practice and prevent bacterial contamination. This priming fluid has been found to achieve a circulating hemoglobin concentration of 6–10 g/dl. 10Bypass was initiated at 100 ml · kg⁻¹· min⁻¹using a calibrated roller pump (Hospal BSM, Lyon, France) and then adjusted according to the experimental protocol. Oxygenation was ensured by continuous oxygen infusion (100 ml · kg⁻¹· min⁻¹) through the oxygenator, and gas flow was adjusted to keep blood-gas parameters (arterial oxygen and carbon dioxide partial pressures, p  H) in the physiologic range and particularly arterial carbon dioxide partial pressure between 4.7–6.0 kPa. After establishment of CPB, pulmonary ventilation was discontinued. The left ventricle was vented by a 16-gauge cannula inserted in the apex of the left ventricle to drain the blood in the venous circuit of the bypass. Pleural and mediastinal blood was scavenged and reinfused into the venous line.

Baseline measurements (S0) were made before CPB. After steady state, CPB was instituted, and four sets of measurements were performed in random order, chosen in a permutation table: S1 = low pressure/low flow; S2 = low pressure/high flow; S3 = high pressure/low flow; and S4 = high pressure/high flow.

Each set of measurements included direct laser Doppler velocimetry (Perimed, Jarfalla, Sweden) in four different areas: stomach, ileum, jejunum, and liver with simultaneous measurements of two sites. After calibration with a Perimed motility standard kit, one laser Doppler probe (Perimed P 424) was gently placed directly on the intestinal wall with a flexible holder and moved from site to site (with each site being marked by a thin superficial suture). Another similar probe, but including a small plastic suction disc allowing a more efficient and nontraumatic contact, was placed on the surface of the liver. These probes were connected to a 632-nm helium–neon laser tube and measured flow velocity to 1-mm depth. 11 

High-flow CPB was defined as a pump flow of 100 ml · kg⁻¹· min⁻¹, and low-flow CPB was defined as 50 ml · kg⁻¹· min⁻¹. High pressure was defined as a mean arterial pressure of 90 mmHg, and low pressure was defined as 40 mmHg. At the end of the preparation, pump flow was turned down to 50 ml/min and then gradually increased by 50 ml/min every minute up to 500 ml/min. Continuous measurement of arterial pressure and LDF on gastric and ileum area were stored on computer to assess the relationship between tissue flow, perfusion pressure, and CPB flow.

Drugs were administered as a bolus in the arterial line of the CPB, close to the animal to avoid biomaterial adsorption. Small amounts of noradrenaline (0.2-mg bolus) as a vasoconstrictor, or nicergoline (1-mg bolus) as an α¹-antagonist vasodilator, were used as necessary to obtain low or high arterial pressure, whatever the CPB flow. We chose these two drugs to mimic our clinical practice in human CPB, when maintenance of pump flow is the primary goal. These drugs are characterized by a short half-life and a minimal action on venous return. Steady state was considered as a period with a stable LDF signal and adequate flow and arterial pressure levels.

Laser Doppler flow data were collected using a multichannel monitor and two probes during steady-state periods of 10 min. Samples were computerized and stored by specially designed software (Perisoft, Jarfalla, Sweden). Curves were retrospectively analyzed, and mean values were calculated automatically by the software for each period.

Laser Doppler velocimetry flow data were expressed in perfusion units.

Initial sample size was calculated from previous results validating the model to obtain α and β risk > 50% and a power > 50%, which are acceptable for this type of model. Values were expressed as mean ± SD. Statistical analysis and graphs were performed using Systat version 7.0 for Windows, 1997 SPSS Inc statistical software, and Prism version 2.0 for Windows 1995 (Graphpad, Inc., San Diego, CA) statistical and graph software. Comparisons were made by two-way analysis of variance for randomized block design, followed by post hoc  Scheffé test on the matrix of pairwise comparison probabilities. A control of the time effect was introduced in the analysis of covariance model as a “sequence order” cofactor, and the vasopressive drugs bolus effect was introduced as a “drug cofactor.” Differences in means between groups were tested by Bonferroni-adjusted t  test. Correlations were tested by parametric test (Pearson) or nonparametric test (Spearman). P < 0.05 was considered significant.

All 10 animals successfully completed the experimental protocol. Descriptive data are summarized in table 1. LDF measurements were not standardized to increase the robustness of the statistics. LDF during high-flow CPB steps were significantly (P < 0.05) greater than during corresponding low-flow CPB, except for liver. Splanchnic LDF values in the low-versus  high-flow groups expressed as perfusion unit were (mean ± SD): stomach, 94 ± 66 versus  137 ± 75; jejunum, 118 ± 78 versus  172 ± 75; ileum, 95 ± 72 versus  146 ± 83; and liver, 79 ± 72 versus  108 ± 118. No significant difference of LDF was observed between the high- and low-pressure groups, whatever the flow, except for liver (mean ± SD): stomach, 115 ± 64 versus  117 ± 83; jejunum, 141 ± 80 versus  148 ± 83; ileum, 127 ± 87 versus  114 ± 76; and liver, 114 ± 88 versus  73 ± 70. Two-way analysis of variance between all splanchnic LDF data, including flow and pressure factors, showed only CPB flow as a significant factor (F ratio = 3.93–6.40). No significant interaction was found between flow and pressure and between flow and sequence order. Statistics and differences of means between groups are summarized in table 2. A significant decrease of LDF was observed between high and low CPB flow for gastric (−31%), jejunum (−31%), and ileum (−34%) tissue. No significant difference was shown between low- and high-pressure levels in the different tissue beds studied. The observed difference for liver LDF was not significant for either pressure or flow factors (−27%).

Central esophageal temperature was kept constant during CPB. It remained mildly hypothermic from 38.9°C (normal rabbit temperature, approximately 39°C) before CPB to 35.8–36.5°C during CPB. No significant difference of temperature between sequences was observed.

Analysis of variance was performed between all LDF measurements after coding for drugs (no drug, vasodilator or vasopressor bolus) and pump flow level. This analysis of variance introducing vasoactive agents as a cofactor was not significant (F = 1.02–1.8) and did not alter the main results concerning arterial pressure level in a clinical point of view.

Descriptive data of arterial pressure and ileum LDF computerized at the end of the experimental design during progressive stepwise increase of CPB flow are represented in figure 1. They showed an acceptable variability of the model and a linear relationship between ileal LDF measurements and CPB flow. Linear regression between gastric and ileum LDF and CPB blood flow or arterial pressure are presented in figure 2. Dispersion was more important for the pressure data. A significant correlation (P < 0.001) using a parametric or nonparametric test has been shown between CPB blood flow and gastric LDF (R = 0.69; ρ= 0.70; R²= 0.48), ileum LDF (R = 0.74; ρ= 0.80; R²= 0.55), and arterial pressure (R = 0.79; ρ= 0.84; R²=0.62). The correlation between pressure and gastric LDF (R = 0.48; ρ= 0.51; R²= 0.23) or ileum LDF (R = 0.47; ρ= 0.56; R²=0.22) was also significant, but the fraction of total variance explained by the variation of pressure expressed by R²was low.

The main result from this study is that low splanchnic tissue perfusion during CPB is primarily dependent on CPB output (blood flow) rather than on perfusion pressure. This result may seem to conflict with an early study using microspheres to measure flow, 4which did not find such a decrease in the splanchnic area, or partially with more recent study that also used microspheres. 3This difference is an example of the variation between macrocirculatory and microcirculatory studies. Microsphere techniques are validated to measure volume blood flow 12; laser Doppler velocimetry is sensitive to erythrocyte (RBC) velocity (V^RBC) and number of RBCs passing through the sample volume per unit time (N^RBC). The product of V^RBCand N^RBCexpressed as perfusion unit is indicative of the tissue perfusion and seems more correlated to capillary partial pressure of oxygen than volume flow. 13A similar discrepancy between visceral perfusion and tissue perfusion has been described when the measurements are made at a tissue level with different techniques such as laser Doppler or gastric tonometry. 14LDF measurements were validated in various situations such as in vitro  gut segment preparation, 11assessment of rectal mucosal perfusion in neonates, 15and gastric perfusion measurement during CPB. 16–18These measurements were reproducible. 19In an experimental model using pigs, Tao et al.  20showed that CPB increased the superior mesenteric blood flow, whereas ileal mucosal blood flow assessed by laser Doppler decreased consistently with the occurrence of a gut acidosis, confirming a tissue level injury and the special interest of LDF monitoring.

An experimental crossover study 21similar to ours has been performed comparing high or low perfusion pressure during CPB with or without dopamine. However, in this study, the high-pressure state was obtained by increasing pump output. Another study, the most similar to the present one, used the microspheres technique in a three-step trial 3comparing low flow with or without phenylephrine to normal flow. Despite the different design, the results are comparable to ours, except that the high-flow/low-pressure state was not obtained. This situation is important in clinical practice and occurs often, mostly at the onset of CPB. It is also difficult to draw conclusions about the effect of low flow or phenylephrine alone. Unlike LDF measurements, the radioactive method does not permit continuous measurement of tissue perfusion and cannot be used to investigate splanchnic autoregulation.

The local autoregulatory control of splanchnic tissue perfusion assumes a constant tissue perfusion inside an arterial pressure range or volume blood flow. 22The mechanisms involve a release of vasoactive mediators, capillary recruitment, and autonomic feedback. This study is not designed to argue for a precise mechanism. Nevertheless, the linear relationship between arterial blood flow mechanically induced by CPB and tissue laser Doppler velocimetry could be interpreted as a disturbance of autoregulation, with a complete dependency between tissue perfusion and CPB flow. This requires confirmation with comparable human studies. Autoregulation of splanchnic perfusion includes a myogenic response 11,23such as renal autoregulation, occurring < 5 min after a decrease in arterial pressure. Therefore, the present trial design with a 10-min step to be appropriate. Furthermore, some of the initial physiologic studies of oxygen uptake and blood flow regulation have used very high levels of blood flow 24to obtain a steady state of autoregulation. In conclusion, splanchnic autoregulation at normal CPB pump flow level has never been assessed. The relation between flow and resistance can be described by the hemodynamic version of Ohm’s law: Q = (P^a− P^v)/R, where Q is the mesenteric blood flow, P is the hydrostatic pressure in the artery and vein, and R is the vascular resistance. 22Because venous pressure stays constant (and low) during regular CPB, splanchnic blood flow can be linearly correlated to CPB output. The Doppler technique could be a valuable method to test regional vascular resistance during CPB and vasodilators. This point has been discussed 12in relation to the discrepancy between LDF and other blood flow measurements, especially during use of vasoactive drugs. Nevertheless, cardiac outputs of animals in this study 12could be considered to be very low, and blood volume uncontrolled, implying an acceptable conclusion only in the low-flow/low-volume state. This is not the case during CPB, where these parameters are controlled: volume flow by the priming solution and CPB pump output, venous portal pressure by the right atrial cannula that vented all caval blood, and RBC number constant after reinfusion.

The absence of a significant relationship between hepatic LDF and CPB output or arterial pressure could be explained by the physiologic balance in perfusion between arterial and venous hepatic flow. Unfortunately, venous portal blood flow could not be measured in this study. No differences in hepatic perfusion were found between pulsatile and nonpulsatile CPB 25if blood flow is maintained in the high range, but total blood flow seems better preserved by high CPB flow. Hepatic disturbance in inflammatory disease may be more related to endothelial permeability and macrophage activation than hypoperfusion. 26 

Mild hypothermia does not seem to be a major factor in splanchnic hypoperfusion 27,28but can reduce metabolic demand so that perturbations in pressure-dependent flow could be tolerated without ischemia. Surface temperature can influence tissue perfusion, but this phenomena is probably attenuated by the direct infusion of CPB thermostat-temperature-controlled blood.

Hemodilution by the pump prime in small animal models has been accepted in other studies 9; hematocrit is kept constant throughout because all intravenous and extravascular blood is reinfused. Moreover, gastric mucosal acidosis is not worsened by hemodilution. 29 

One limitation in transposing our results to the medical practice is the presence of atherosclerosis, which seems to be more relevant than age. 1Non-CPB animal models are necessary to confirm the CPB action on autoregulation.

Ketamine used to maintain anesthesia does not act on intestinal motility. 30Animals were allowed to eat and drink freely until the beginning of experiments to prevent stress and intestinal mucosal injury. This does not occur in human clinical guidelines. Furthermore, potential effects from laparotomy on visceral vascular resistance could not be excluded.

In this study using small animals, Doppler signals were approximately 1 mm deep from the surface, including global mucosal and muscularis tissue flow measurements. A comprehensive approach of flow redistribution between mucosa and muscularis or tissue heterogeneity was not permitted in our model. In situ  experiments with needle probes to differentiate the two intestinal layers is not possible because of bleeding during CPB. Caution must therefore be exercised in extrapolating these results from the present animal model to the human clinical situation.

Splanchnic perfusion is altered during CPB and could not be assumed by monitoring only arterial pressure. Flow is more important than pressure in maintaining splanchnic perfusion. During CPB, splanchnic autoregulation seems to be altered in this animal model, and a linear relation between tissue perfusion assessed by LDF and CPB blood flow was found. Nevertheless, differences in behavior between hepatic, gastric, and ileal tissue perfusion have been demonstrated.

The authors acknowledge G. Timour, J. P. Viale, and G. Annat from the University Cl Bernard for their methodologic help. A special acknowledgment is due to Florence Arnal, laboratory technician, for her contributive work and organization. We also thank the L. Pradel hospital perfusionist team for their judicious technical assistance.