Positive end-expiratory pressure (PEEP) may impair extra-pulmonary organ function. Pathophysiologically, cardiac function, right atrial pressure, hepatic venous pressure, generation of intra-hepatic closing pressures due to diaphragmatic compression and intra-abdominal pressure may influence hepato-splanchnic blood flow [1–5]. In general, experimental and clinical studies on the effects of PEEP on hepato-splanchnic blood flow showed inconsistent results. Portal venous blood flow was observed to decrease parallel to a decrease in cardiac output starting with PEEP levels as low as 5–10 cmH2O . However, Kiefer and colleagues  demonstrated that an increase in PEEP itself did not have a consistent effect on splanchnic blood flow and metabolism when cardiac index (CI) is stable. However, the effects of PEEP on hepatic blood flow and function have been rarely studied in human beings, especially cardiac surgical patients, because liver catheterization techniques for the determination of absolute hepato-splanchnic blood flow are invasive, potentially harmful and cumbersome [6,7].
More recently, indocyanine green plasma disappearance rate (ICG-PDR) has been suggested to describe global hepatic blood flow and function . After injection into the circulation, indocyanine green (ICG) is exclusively eliminated from the blood by the liver. Interestingly, an increase in ICG-PDR with increasing PEEP has been described in an animal experimental setting . These authors speculated that their findings are explained by an increased backward pressure towards the hepatocytes. Furthermore, microcirculation within the splanchnic bed is clinically difficult to assess. For instance, gastric mucosal tonometry is used in the clinical setting for the assessment of microcirculation.
In general, application of PEEP is a common strategy, especially in patients after heart surgery and lung collapse, necessitating recruitment of lung tissue. In this randomized and controlled clinical study, we analysed whether a PEEP of 10 mbar after cardiac surgery may affect hepato-splanchnic blood flow.
With approval by our Local Ethics Committee and preoperative written consent, we prospectively studied 28 patients with elective coronary artery bypass surgery or valve replacement. Patients had no known liver disease as confirmed by a normal serum bilirubin prior to surgery (<18 μmol L−1). Contraindications for this study were known iodine allergy or thyrotoxicosis due to the fact that ICG contains iodine. Furthermore, we did not study patients with COPD or an intra-aortic balloon pump (IABP). Pre- and intra-operative management is standardised in our department. Patients received midazolam (7.5 mg) orally for pre-medication. Induction of anaesthesia was performed using midazolam (0.15 mg kg−1) and sufentanil (1 μg kg−1) and anaesthesia was maintained with volatile anaesthetics (sevoflurane in oxygen, 0.5 MAC) and sufentanil (0.5 μg kg−1 h−1). A non-depolarising muscle relaxant (pancuronium 0.1 mg kg−1) was used to facilitate intubation of the trachea. After induction of anaesthesia, a gastric 16-French catheter (Trip® NGS catheter; Tonometrics Division, Helsinki, Finland) was placed trans-orally for assessment of microcirculation within the splanchnic area (gastric mucosal PRCO2). For adequate interpretation, PCO2-gap (i.e. the difference between PRCO2 and end-tidal PCO2) was calculated. Since H2-inhibitors or proton pump inhibitors may influence tonometric measurements , such drugs were not administered on the morning of surgery or during the study period. As routine practice, all patients underwent urinary bladder catheterization and diuresis was monitored hourly. All patients underwent extended haemodynamic monitoring by a pulmonary artery catheter (7.5 French five-lumen pulmonary artery catheter, Edwards Swan Ganz®, CCO/SvO2, Model 744H 7.5 F; Baxter Healthcare Corporation, Irvine, CA, USA) and received a left atrial catheter by the surgeons. Body core temperature was continuously monitored by the pulmonary artery catheter. Heparin (350 units kg−1) was injected intravenously (i.v.) prior to cardiopulmonary bypass (CPB) to achieve an activated clotting time of >450 s. While maintaining a flow of 2.5 L min−1 m−2, mean arterial pressure (MAP) was kept between 50 and 60 mmHg. Body core temperature during CPB was kept at 32–33°C. During CPB, our standard is to keep haematocrit between 25% and 30%.
After surgery, all patients were transferred to the intensive care unit (ICU). Immediately after admission to the ICU, correct position of the gastric catheter was confirmed by X-ray. Postoperatively, all patients were sedated with propofol in an individual dose (100–300 mg h−1) and doses remained unchanged. Vasoactive drugs comprised norepinephrine and epinephrine, no other inotropes or vasopressors were used. Doses of vasoactive drugs were adjusted individually to keep MAP constant. An MAP of >70 mmHg was considered appropriate during the postoperative period. If administered, nitroglycerin dosage remained unchanged throughout the study period. No patient received phosphodiesterase inhibitors or nitroprusside. Postoperative fluid management was guided by central venous pressure (CVP) and left atrial pressure (LAP) which were kept between 8 and 12 mmHg. During transport to the ICU, all patients were mechanically ventilated with a PEEP of 5 mbar. After ICU admission, patients were mechanically ventilated in a pressure-controlled mode (BIPAP, Evita 4; Draeger, Germany) with a PEEP of 5 mbar. Driving pressure was set as to obtain a tidal volume of 6–8 mL kg−1 with an inspiratory to expiratory time ratio of 1:1. Inspiratory O2 fraction was adjusted individually to maintain normoxia. On admission to the ICU, patients were randomised by using sealed envelopes into two groups and baseline measurements were taken. Global haemodynamics and CI, ICG-PDR and gastric mucosal PCO2 were assessed on ICU admission (baseline, PEEP = 5 mbar), after 2h with either PEEP of 5 mbar (PEEP5 group) or 10 mbar (PEEP10 group) and again after 2 h of 5 mbar in both groups. During a PEEP of 10 mbar, inspiratory peak pressure (Pinsp) was adapted accordingly to maintain normocapnia.
For (triplicate) measurement of cardiac output, each bolus injection used normal saline (10 mL) at room temperature. The central venous injection port and pulmonary artery catheter were connected to the Vigilance® system (Software 6.3; Baxter Edwards Laboratories, Irvine, CA, USA) for determination of cardiac output by thermodilution. Bolus injections were made manually at end-expiration. Arterial and mixed venous blood gas samples were taken prior to bolus injections and immediately analysed for oxygen and carbon dioxide tensions, pH, haemoglobin content, haemoglobin oxygen saturation, lactate and glucose levels (Radiometer System 725®, Copenhagen, Denmark). In this system, O2 saturation is actually measured by spectrophotometry and not calculated from blood gases.
Pressure transducers were calibrated and zeroed against atmosphere at the mid-chest level. For the measurement of pressures, patients were in the horizontal position and not disconnected from the ventilator. Besides continuous monitoring of arterial and pulmonary artery blood pressures, haemodynamic monitoring included measurement of CVP, LAP, mean pulmonary artery pressure (MPAP) and pulmonary artery thermodilution cardiac output.
Liver function and blood flow was assessed non-invasively by transcutaneous measurement of ICG-PDR using a finger clip system (LiMon®; Pulsion Medical Systems AG, Munich, Germany). In this technique, ICG in the blood is recorded by a densitometric device and after backward extrapolation the decay curve of ICG blood concentration is expressed as percentage change over time (% min−1). For each measurement of ICG-PDR, 0.25 mg kg−1 ICG dissolved in 15 mL were injected as a bolus through the central venous line  and results were obtained within 400–600 s. Normal values for ICG clearance and the ICG-PDR are considered to be >700 mL min−1 m−2 and 18% min−1, respectively .
Results are expressed as mean ± standard deviation. Patients' characteristics were compared by a χ2 (Fisher's exact) test and a non-parametric U-test. Haemodynamic results in both groups were compared by a one-way analysis of variance (ANOVA) for repeated measurements and an all pair-wise comparison method (Student–Newman–Keuls). A power analysis (power = 0.8; change in ICG-PDR to be detected, 2.0% min−1; standard deviation of change, 2.0% min−1; α < 0.05) revealed a sample size of 13 individuals per group. For the statistical analysis, SigmaStat for Windows (version 1.0) was used. A P < 0.05 was considered as statistically significant.
We studied 14 patients in the PEEP10 group (13 male, 1 female; age 48–74, mean 63 ± 7 yr). As controls (PEEP5) we studied 14 other patients (8 male, 6 female; age 46–86, mean 68 ± 11 yr) in whom PEEP was 5 mbar throughout the study. All patients completed the study without adverse effects. Patients' characteristics were matched between both groups except for CPB and aortic cross-clamping time which were longer in the PEEP5 group. However, type of surgery and number of patients receiving epinephrine or norepinephrine on ICU admission were comparable (Table 1). Results on global cardio-circulatory and O2-transport variables are summarised in Table 2. There were no significant changes in heart rate (HR), arterial pressure, CVP and LAP. Furthermore, CI, systemic oxygen delivery and consumption were without significant differences within and between both groups. Significant differences were found for Pinsp which required adjustment in the PEEP10 group in order to maintain arterial PCO2. Body temperature significantly increased between each step within both groups, however, was not significantly different at each time point between both groups. Furthermore, epinephrine and norepinephrine doses remained unchanged throughout the study period and were not different between both groups. Unfortunately, one patient in each group could not be monitored by gastric tonometry for technical reasons. Results on metabolic and regional hepato-splanchnic variables are summarised in Table 3. No statistically significant differences were found for ICG-PDR, PCO2-gap and serum lactate. In detail, ICG-PDR in the PEEP10 group was 24.0 ± 6.9, 22.0 ± 7.9 and 25.5 ± 7.7% min−1 and 22.0 ± 7.5, 23.8 ± 8.4 and 21.4 ± 6.5% min−1 in the PEEP5 group (P = 0.05). The PCO2-gap throughout was slightly higher in the PEEP10 group (1.1 ± 0.9, 1.3 ± 0.7 and 1.3 ± 0.9 kPa) when compared to the PEEP5 group (0.8 ± 0.5, 0.9 ± 0.5 and 0.9 ± 0.5 kPa) (P = 0.05).
Although flow-dependent hepatic function (ICG-PDR) showed a tendency to decrease during a PEEP of 10 mbar for 2 h, hepato-splanchnic blood flow was not compromised in patients after elective cardiac surgery with maintained global haemodynamics. Furthermore, PCO2-gap as a marker of gastric mucosal perfusion remained unchanged. Although slightly higher during a PEEP of 10 mbar, these changes are of no clinical relevance, since a PCO2-gap <10 mmHg (1.4 kPa) is within the physiological range and pathologic values are above 20 mmHg (2.8 kPa) .
In general, positive pressure ventilation using PEEP may have significant influence on global and regional blood flow due to decrease in venous return and cardiac output caused by the increase in intrathoracic pressure. Furthermore, PEEP may reduce cardiac output and influence its distribution while especially hepato-splanchnic blood flow may be compromised [1–5]. However, studies indicate that as long as cardiac preload and thus output is maintained, hepato-splanchnic blood flow may be preserved in critically ill patients. Bruhn and colleagues  increased PEEP to 10, 15 and 20 cmH2O for 30 min each in ARDS patients. As result, PEEP had no effect on PCO2-gap (median baseline: 19 mmHg; PEEP10: 19 mmHg; PEEP15: 18 mmHg and PEEP20: 17 mmHg), respectively. In this study, median CI was actually high and also remained unchanged (baseline: 4.6 L min−1 m−2; PEEP10: 4.5 L min−1 m−2; PEEP15: 4.3 L min−1 m−2 and PEEP20: 4.7 L min−1 m−2). Thus, these authors concluded that a PEEP of 10–20 cmH2O does not affect gastric mucosal perfusion and is haemodynamically well tolerated in most patients with ARDS, including those receiving adrenergic drugs . Akinci and colleagues  started from an initial PEEP level of 5 cmH2O, and PEEP was titrated at 2 cmH2O increments while monitoring gastric mucosal pH. Again, incremental titration of PEEP based on improvement in oxygenation was found not to decrease gastric intramucosal perfusion when cardiac output is preserved. Furthermore, Claesson and colleagues  studied the effects of recruitment manoeuvres on gastric mucosal perfusion as assessed by Laser Doppler flowmetry. However, they observed a significant decrease in CI (P = 0.04) while gastric mucosal perfusion showed a trend towards gradual decrease (P = 0.05). In a study by Kiefer and colleagues  using liver venous catheterisation, mean PEEP was increased by 5 cmH2O from a clinically selected PEEP level (mean 8, range 6–11 cmH2O) up to (mean 13, range 10–14 cmH2O) followed by a return to baseline. Here, PEEP by itself did not have a consistent effect on splanchnic blood flow and metabolism when CI is stable. Furthermore, no deleterious effects on hepatic function were noted in 11 hepatic transplant patients after the administration of PEEP of 10 cmH2O during the postoperative period . In this study, hepatic function was evaluated by the invasive assessment of ICG-PDR by a fibre-optic arterial catheter. Unlike other studies in patients under controlled mechanical ventilation, this study was conducted in patients with a pressure-controlled mode. This ventilator setting allows spontaneous cycles, reducing intrathoracic pressures and hepatic haemodynamic alterations . For comparison, all patients in our study were controlled ventilated in the BIPAP mode without spontaneous breathing. Noteworthy, already lower levels of PEEP may exert negative regional effects. In human beings, a decrease in portal venous blood flow parallel to the decrease in CI starting with PEEP levels as low as 5–10 cmH2O was described . In our study, CI was always maintained, and our results are in line with previous studies in terms of gastric mucosal blood flow. However, since cardiac output was maintained in our study, an issue was whether PEEP has a specific effect on regional perfusion to the splanchnic bed independent of cardiac output, which however, was not confirmed. The fact that the use of PEEP did not cause any change in cardiac performance in our patients is most probably due to euvolaemia of the study patients. In summary, maintenance of adequate intravascular volume status is obviously of particular relevance for maintaining regional blood flow during ventilation using PEEP.
In general, ventilation with PEEP is also a common strategy in patients with hypoxemia or lung collapse after thoracic surgical procedures. In our study, we primarily decided to study the effects of a PEEP of 10 mbar because a PEEP up to this level is often applied in our institution in this scenario and can ethically be justified. Furthermore, a 2-h interval between each ICG-PDR and PRCO2 measurement was chosen to allow a stabilisation of haemodynamic status and a complete elimination of the ICG for which elimination half-live is about 4 min in patients without hepatic dysfunction . Joly and colleagues  and Lehmann and colleagues  reported that ICG-PDR may react in 1-h intervals after changes in treatment.
However, Matuschak and colleagues  found in a canine model an increase in ICG-PDR with increasing PEEP and these findings are in contrast to other studies which showed a decrease in hepatic blood flow and an increase in portal venous pressure parallel to the level of PEEP [7,21,22], Although remaining open, these authors speculated that their finding was probably due to an increased ‘backward pressure’ from the blood to the hepatocytes. As these authors, we used ICG-PDR for the assessment of flow-dependent liver function which is a marker of clearance. However, animal experimental data suggest that short-term changes in ICG-PDR do reflect more changes in liver blood flow (sum of portal venous and hepatic arterial flow) than hepato-cellular function (i.e. ICG biliary excretion) . In this study, ICG-PDR was unchanged as was splanchnic blood flow measured by a liver vein catheter while biliary ICG excretion significantly dropped during endotoxinaemia . However, for measurement of absolute hepato-splanchnic blood flow access to hepatic vessels (hepatic artery, portal and sub-hepatic veins) would be required. However, liver venous catheterisation is invasive, expensive and of potential risk, and we used ICG-PDR for the assessment of hepato-splanchnic blood flow. However, data on ICG-PDR in postoperative cardiac surgical patients are still rare , especially when using the non-invasive technique [25,26]. The transcutaneous system we used has been validated against an intravascular system with a pre-calibrated fibre-optic in critically ill patients receiving vaso-active drugs . Unfortunately, neither liver vein catheterisation nor ICG-PDR allows a differentiation of blood flow distribution within the splanchnic area.
Our study has several limitations. First, we used ICG-PDR without an invasive reference technique (i.e. liver vein catheterisation) which would allow determination of absolute hepato-splanchnic blood flow. In general, ICG-PDR as a surrogate for hepatic blood flow and function itself has methodological limitations . Attempts to calculate absolute hepatic blood flow from ICG-clearance without knowledge of ICG-extraction is inaccurate, since hepatic ICG-extraction cannot be reliably estimated. Uusaro and colleagues  could demonstrate that ICG extraction may vary considerably within and between different patient groups. However, since liver cell function may probably not change that fast, changes observed may be related to changes in liver blood flow and not cell function. Furthermore, for assessment of gastric mucosal perfusion, we used air tonometry which has been validated against the saline technique . However, changes in gastric tonometry may not accurately reflect changes in other regions of the splanchnic tract . Second, CPB and aortic cross-clamping time were longer in our control group. Interestingly, no significantly differences in ICG-PDR and PRCO2 between both study groups were found indicating this difference (e.g. in endotoxin release) was not a relevant factor. Third, we studied no group with zero end-expiratory pressure (ZEEP). However, in our patients after cardiac surgery we consider ZEEP inappropriate and not justifiable. Furthermore, we studied only one level of PEEP during a relatively short period, moreover in patients without negative alteration of their haemodynamic status or lung compliance. Finally, we studied relatively healthy patients, and findings would presumably have been more significant in sicker individuals with compromised lung or cardiac function and poor splanchnic perfusion.
In conclusion, a PEEP of 10 mbar over 2 h was not associated with a compromised flow-dependent liver function (ICG-PDR) or gastric mucosal blood flow in patients after cardiac surgery.
This study was undertaken using departmental funding.
1. Quist J, Pontoppidan H, Wilson RS, Lowenstein E, Laver MB. Hemodynamic responses to mechanical ventilation with PEEP. Anesthesiology
2. Fessler HE, Brower RG, Wise RA, Permutt S. Effects of positive endexpiratory pressure on the gradient for venous return. Am Rev Respir Dis
3. Winso O, Biber B, Gustavsson B et al
. Portal blood flow in man during graded positive end-expiratory pressure ventilation. Intens Care Med
4. Bonnet F, Richard C, Glaser P, Lafay M, Guesde R. Changes in hepatic flow induced by continuous positive pressure ventilation in critically ill patients. Crit Care Med
5. Berendes E, Lippert G, Loick HM, Brüssel T. Effects of positive end-expiratory pressure ventilation on splanchnic oxygenation in humans. J Cardiothorac Vasc Anesth
6. Kiefer P, Nunes S, Kosonen P et al
. Effect of positive end-expiratory pressure on splanchnic perfusion in acute lung injury. Intens Care Med
7. Aneman A, Eisenhofer G, Fandriks L et al
. Splanchnic circulation and regional sympathetic outflow during peri-operative PEEP ventilation in humans. Br J Anaesth
8. Sakka SG, Reinhart K, Meier-Hellmann A. Prognostic value of indocyanine green plasma disappearance rate in critically ill patients. Chest
9. Matuschak G, Pinsky MR. Effects of positive end-expiratory pressure on hepatic blood flow and performance. J Appl Physiol
10. Parviainen I, Ruokonen E, Takala J. Sodium nitroprusside after cardiac surgery
: systemic and splanchnic blood flow and oxygen transport. Acta Anaesthesiol Scand
11. Sakka SG, Koeck H, Meier-Hellmann A. Measurement of indocyanine green plasma disappearance rate by two different dosages. Intens Care Med
12. Kuntz HD, Schregel W. Indocyanine green: evaluation of liver function; application in intensive care medicine. In: Lewis FR, Pfeiffer UJ, eds. Practical Applications of Fiberoptics in Critical Care Monitoring
. Berlin: Springer, 1990: 57–62.
13. Kolkman JJ, Groeneveld AB, van der Berg FG et al
. Increased gastric PCO2
during exercise is indicative of gastric ischaemia: a tonometric study. Gut
14. Bruhn A, Hernandez G, Bugedo G, Castillo L. Effects of positive end-expiratory pressure on gastric mucosal perfusion in acute respiratory distress syndrome. Crit Care
15. Akinci IO, Cakar N, Mutlu GM et al
. Gastric intramucosal pH is stable during titration of positive end-expiratory pressure to improve oxygenation in acute respiratory distress syndrome. Crit Care
16. Claesson J, Lehtipalo S, Winso O. Do lung recruitment maneuvers decrease gastric mucosal perfusion? Intens Care Med
17. Krenn C, Krofft P, Schoefer B. Effects of positive end-expiratory pressure in hemodynamics and indocyanine green kinetics in patients after orthoptic liver transplantation. Crit Care Med
18. Leevy CM, Leevy CB, Howard MM. Indocyanine green and the liver. In: Davidson CS, ed. Problems in Liver Diseases
. New York: Georg-Thieme-Verlag, 1979: 42–52.
19. Joly LM, Monchi M, Cariou A et al
. Effects of dobutamine on gastric mucosal perfusion and hepatic metabolism in patients with septic shock. Am J Respir Crit Care Med
20. Lehmann C, Taymoorian K, Wauer H et al
. Effects of the stable prostacyclin analogue iloprost on the plasma disappearance rate of indocyanine green in human septic shock. Intens Care Med
21. Bredenberg CE, Paskanck A, Fromm D. Portal hemodynamics in dogs during mechanical ventilation with positive end-expiratory pressure. Surgery
22. Johnson E. Splanchnic hemodynamic response to passive hyperventilation. J Appl Physiol
23. Stehr A, Ploner F, Traeger K et al
. Plasma disappearance of indocyanine green: a marker for excretory liver function? Intens Care Med
24. Goedje O, Peyerl M, Seebauer T et al
. Reproducibility of double indicator dilution measurements of intrathoracic blood volume compartments, extravascular lung water, and liver function. Chest
25. Hofmann D, Thuemer O, Schelenz C, van Hout N, Sakka SG. Increasing cardiac output by fluid loading: effects on indocyanine green plasma disappearance rate and splanchnic microcirculation. Acta Anaesth Scand
26. Sakka SG, Hofmann D, Thuemer O et al
. Increasing cardiac output by epinephrine after cardiac surgery
: effects on indocyanine green plasma disappearance rate and splanchnic microcirculation. J Cardiothor Vasc Anesth
27. Sakka SG, Reinhart K, Meier-Hellmann A. Comparison between invasive vs. non-invasive measurement of indocyanine-green plasma disappearance rate in critically ill patients with mechanical ventilation and stable haemodynamics. Intens Care Med
28. Bonnet F, Richard C, Glaser P et al
. Changes in hepatic flow induced by continuous positive pressure ventilation in critically ill patients. Crit Care Med
29. Uusaro A, Ruokonen E, Takala J. Estimation of splanchnic blood flow by the Fick principle in man and problems in the use of indocyanine green. Cardiovasc Res
30. Heinonen PO, Jousela IT, Blomqvist KA et al
. Validation of air tonometric measurement of gastric regional concentrations of CO2
in critically ill septic patients. Intens Care Med
31. Parviainen I, Ruokonen E, Takala J. Dobutamine-induced dissociation between changes in splanchnic blood flow and gastric intramucosal pH after cardiac surgery
. Br J Anaesth
*This work has been presented at the Annual Meeting of the American Society of Anesthesiologists (ASA) in Atlanta (12–17 October 2005).