Epidural analgesia using epidural local anaesthetics is widely used for abdominal surgery due to superior postoperative pain control . On the other hand, sympathomimetics have been shown to modify oedema formation during inflammatory reactions . β-adrenergic stimulation causes decreased [3-5], whereas α-adrenergic stimulation causes increased pro-inflammatory mediator levels . Therefore, it is difficult to predict the local effect of epidural local anaesthetic block on oedema formation in gut and mesenteric tissues.
Tissue oedema increases the distance of oxygen and nutrient transport and the swelling may cause strangulation at the anastomotic suture sites, eventually leading to anastomotic leakage, a feared complication in bowel surgery . Thus, if epidural local anaesthetics cause increased oedema formation in the gut wall due to decreased sympathomimetic stimulation, an increased risk of anastomotic leakage may ensue.
In abdominal surgery, epidural local anaesthetics and accelerated mobilization has been shown to increase body weight during the first postoperative week compared to epidural morphine analgesia . The rapid changes were probably due to differences in perioperative water sequestration, which must be remobilized postoperatively , causing an increased strain on the cardiopulmonary and renal systems. These effects might balance some of the positive effects of epidural analgesia with local anaesthetics  and may necessitate an adjustment of the perioperative care of abdominal surgery patients.
The present study was performed to test the hypothesis that epidural analgesia with local anaesthetics causes increased oedema formation in gut and mesentery both in normal and anastomotic tissues.
Materials and methods
The National Animal Research Supervisor, Copenhagen, Denmark, approved the protocol. No muscle relaxation was used. The anaesthetized animals were sacrificed at the end of the study by an intravenous (i.v.) pentobarbital injection.
Thirty female pigs (Dansk Landrace, Påskehøjgaard, Århus, Denmark), weighing 45-55 kg, were randomly allocated to three groups: epidural morphine (Sygehusapotekerne, Copenhagen, Denmark) (n = 10), epidural bupivacaine 0.5% (Sygehusapotekerne, Copenhagen, Denmark) (n = 10) or i.v. fentanyl (Jansen-Cilag, Beerse, Belgium) (n = 10). One animal in both the epidural bupivacaine and the i.v. groups were excluded due to surgical complications.
In the unpremedicated animals, anaesthesia was induced using intramuscular (i.m.) ketamine, 8-10 mg kg−1 (Ketalar, Parke-Davis, Solna, Sweden) and diazepam, 0.25-0.35 mg kg−1 (Stesolid, Dumex-Alpharma, Copenhagen, Denmark). Tracheal intubation was performed under intermittent doses of i.v. thiopentone, 1-5 mg kg−1 (Sygehusapotekerne, Copenhagen, Denmark) and anaesthesia was maintained with isoflurane (Forene, Baxter, Deerfield, IL, USA) 0.5-2.5% and i.v. midazolam (Dormicum, Roche, Basel, Switzerland), 5 mg h−1, until the end of the experiments. To resemble the clinical situation, the isoflurane anaesthesia was regulated according to blood pressure (BP), heart rate (HR) and general signs such as movement, pupil size and shivering. The animals were ventilated with a Servo B ventilator (Siemens, Solna, Sweden) at supranormal partial pressure of oxygen (PaO2) and normal partial pressure of carbon dioxide (PaCO2).
All the animals received an 18-G epidural catheter (Braun, Melsungen, Germany), which was introduced between the tenth and eleventh thoracic vertebrae using a loss of resistance method. The epidural catheter placement was controlled by X-ray using 0.3 mL iohexol (Omnipaque, Nycomed Imaging, Oslo, Norway) injection.
The surgeon was blinded to which i.v. infusion, epidural bolus and epidural infusion the animal received. The i.v. infusion was fentanyl (7.5 μg kg−1 h−1) or isotonic saline. The epidural bolus was morphine (4 mg every 6 h) or isotonic saline. The epidural infusion was bupivacaine (5 mg mL−1) or isotonic saline, given at a rate of 10 mL h−1 after two initial boluses at 10 min intervals of 4 mL bupivacaine or isotonic saline. Heat was conserved by placing the pig on a heat mattress, covering the animal with blankets, and by covering the gut with a plastic barrier. The temperature was controlled using the central temperature measured by the pulmonary artery catheter.
Following a midline abdominal incision, the lymphatic intestinal trunk was located from the right side and cannulated using a 0.5 mm inner diameter, 0.6 mm outer diameter clear vinyl tube (Dural Plastics and Engineering, Auburn, Australia). The animals were then rested for 1 h. Time zero was immediately after the 1-h rest.
At time zero, a 1 cm small bowel resection and an adjacent, 1 cm by 1 cm, mesenteric resection was performed 20 cm from the caecum using diathermy and a two-layered anastomosis was made with two by eight sutures (Polysorb 3-0, Sherwood, Davies-Geck, USA) by the same surgeon. The procedure was repeated obtaining a total of six primary anastomoses with corresponding primary gut and mesenterial samples.
The anastomoses were then re-resected in the reverse order obtaining six post-anastomotic bowel and six mesenteric samples as illustrated in Figure 1. The samples taken at 20 and 120 cm from the caecum were not included in the present study.
All animals received a 10 mL kg−1 bolus of isotonic saline during the positioning of the epidural catheter followed by a continuous i.v. infusion of 15 mL kg−1 h−1 isotonic saline.
The relative water content of the gut and mesenterial samples was expressed as the ratio between wet and dry weights of the samples. The dry weights were obtained by heating the tissue at 70°C for three weeks. Lymph flow was measured as g min−1, every half hour. Urine output was measured hourly via a bladder catheter.
The animals were monitored for central haemodynamic changes using a cardiac output monitor (Vigilance, Baxter, Irvine, CA, USA) and a pulmonary artery catheter (7.5-F, Swan-Ganz, CCO/SvO2 catheter, Edward Lifesciences, Baxter, Irvine, CA, USA) placed via the right jugular vein. HR and arterial BP was measured using a monitor (Cardiomed, Oslo, Norway) via a catheter (Avanti Cordis, Johnson & Johnson, Miami, FL, USA) in the right carotid artery. Blood for erythrocyte volume fraction, haemoglobin and plasma protein concentrations were sampled.
The study was powered so a difference of 1.33 standard deviations (SD) was detected in 80% of cases using α = 0.05 (unpaired t-test). Based on this power analysis we needed 10 animals in each group. Results are given as mean ± standard error of mean (SEM). All statistical analyses were performed in SPSS, version 10. Primary end-points were analysed using repeated measures analysis of variance (ANOVA) for time and group effects and for time treatment interactions. If repeated measures ANOVA revealed statistical significance, post hoc testing was performed using univariate ANOVA and Bonferonni protected t-tests. To avoid mass significance, secondary end-points were analysed by the less sensitive univariate ANOVA at baseline, 3 h (pre-anastomosis surgery) and 6 h (post-anastomosis surgery). P-values <5% were considered statistically significant.
Gut water content (Fig. 2) changes were small, the mean for the groups ranging from +4% (bupivacaine group) to −3% (fentanyl group) (P = 0.11). There was no significant increase in oedema formation induced by surgery, no significant difference in water content induced by the anaesthetic techniques over time and no significant changes in the surgical response induced by the different anaesthetic techniques.
Mesenterial water content (Fig. 3), showed no significant increase prior to anastomosis. In the post-anastomotic samples a highly significantly greater water content was seen compared to the pre-anastomotic samples (P < 0.01). On average the increase was 146%, 123% and 109% [(mean of all post-anastomotic samples - mean of all pre-anastomotic samples)/mean of all pre-anastomotic samples] in the epidural bupivacaine, i.v. fentanyl and epidural morphine groups, respectively. The increase took place during the first 2 h after the anastomoses were performed. Using ANOVA for repeated measures, a significant time treatment interaction was revealed in the post-anastomotic samples (P < 0.05). As can be seen in Figure 3, bupivacaine showed the highest water content of the three groups in the post-anastomotic samples.
Lymph flow (mL h−1) (Fig. 4) did not change significantly during the course of the experiments. The mean ranged from 4.17 (±0.7) (bupivacaine) to 3.56 (±0.4) (fentanyl) (P = 0.8).
Urine output (mL 6 h−1) is shown in Table 1. The mean ranged from 639 (±91) (bupivacaine) to 921 (±144) (morphine) for the three groups (P = 0.17).
Haemoglobin and extravascular fluid increased (P < 0.01), whereas albumin decreased (P < 0.01) during the study. No group effect or time group interaction were seen for any of these three variables.
The haemodynamic values were recorded every 30 min. The values at 0, 3 and 6 h are presented in Table 2. Using a test for group differences at these three time points, no statistically significant differences were detected.
The principal findings of the present study were that bowel surgery in pigs caused no swelling in normal or peri-anastomotic gut wall during the first 5 h after small intestine anastomoses. Similarly, mesenteric oedema was not seen prior to gut resection. On the contrary, highly significant oedema developed in the peri-anastomotic mesentery and the use of epidural bupivacaine might enhance this oedema formation after surgery.
The animals were randomized and the surgeon was blinded to which regimen the animal received. The latter necessitated identical fluid regimens and for the anaesthetist to regulate the level of general anaesthesia using BP and HR as guides. This may have limited the systemic differences induced by the different anaesthetic techniques as illustrated by the haemodynamic results showing no statistical differences. Similarly, this entailed a theoretically relative fluid overload of the fentanyl group although this could not be detected in pulmonary capillary wedge pressure or central venous pressure (CVP) measurements. The fluid therapy regimen was not according to every day practice. Often, patients receiving epidural anaesthesia with local anaesthetics receive more fluid than the other regimens  although recently this practice has been questioned . In the clinical setting, this greater fluid load during epidural analgesia may increase oedema formation. All three groups received a relatively high dose of fluid compared to that recommended for human beings. This was first due to the pigs having relatively greater body volumes and weights than human beings and second, it was imperative that none of the groups were hypovolaemic during surgery. Pulmonary capillary wedge pressures and CVPs showed that none of the groups seemed to have been overhydrated during surgery.
To limit non-surgical effects, the animals were non-premedicated and an identical induction method was used. Heat conservation was carried out, controlling the central temperature of the animals. This may have masked some of the effects of the differences of the alternative anaesthetic techniques, since epidural anaesthesia with local anaesthetics normally would cause greater vasodilatation with increased temperature loss. The protocol concerning external heat conservation was similar to the modern theatre logistics of anaesthesia. Thus, a non-anaesthetic effect of vasodilatation due to heat conservation may have resulted.
The effects of epidural local anaesthetic block on local tissue oedema formation are complex and not fully understood. The balance between transcapillary fluid flux (Jv) and lymphatic drainage (JL) from a tissue governs the interstitial fluid volume (IFV) . Mathematically this can be expressed as:
The factors governing the transcapillary fluid flux are as described by Starling, expressed as:
where CFC is the capillary filtration coefficient (the product of hydraulic conductivity and area of microvascular exchange), P and COP are the hydrostatic and oncotic pressures, respectively. σ is the capillary reflection coefficient for plasma proteins.
The present protocol did not try to look into the individual possible changes in the Starling equation, but addressed the global local effects. A measure of the hydrostatic pressures (capillary and interstitial) as well as the interstitial oncotic pressure would imply major local manipulation. This was unwanted in the present protocol.
Theoretically, opposing effects on the Starling forces may result from an epidural block with local anaesthetics with effects on the sympathetic nervous system. Local vasodilatation and recruitment (increased CFC), a predominant precapillary vasodilatation (increased capillary hydrostatic pressure)  as well as a direct anti-inflammatory effect on σ  (indirectly limiting the ΔCOP effect of albumin extravasation), would all tend to increase oedema formation. On the contrary, the relatively lower sympathetic tone due to the epidural block would decrease oedema formation tendency through increased lymphatic drainage both indirectly through increased gut motility  and directly .
The results show that surgery itself causes no oedema formation in the gut wall during the first 5 h after small intestinal anastomosis and there were no detrimental effects of the different anaesthetic techniques. Furthermore, Adolph and co-workers  have shown that thoracic epidural block in the rat inhibited hypotension induced impairment of capillary perfusion in the muscular layer of the gut. Likewise, epidural block increases intestinal blood flow . These results, together with our lack of early oedema tendency in the gut wall after gut resection, seem to suggest that epidural block is not theoretically detrimental with reference to anastomotic leakage, due to local oedema formation or decreased blood flow. Recently, a meta-analysis (562 patients)  found no significant increase in anastomotic leakage after postoperative epidural local anaesthetic or epidural local anaesthetic-opioid mixtures, although a possible Type 2 error was discussed.
Increased mesenterial oedema was seen, highly significantly due to surgery, but also slightly increased by the epidural block, primarily during the first 2 h. Surgical stress is well known to cause systemic fluid retention by increased antidiuretic and aldosterone hormone levels . Nevertheless, epidural local anaesthetic block caused an increased weight gain due to fluid shifts compared to a regimen using epidural morphine . Both the increased mesenteric water content and the tendency for increased systemic water retention in the epidural bupivacaine group in the present study, suggest that the anti-inflammatory effects of stress may at least overrule the effects of increased antidiuretic and aldosterone hormone levels in non-blocked patients. A recent randomized, multicentre trial  recommended restricted fluid therapy for colorectal surgery due to reduced major complications. Thus, perioperative adrenergic support might be preferential to fluid loading to limit BP decrease due to epidural local anaesthetic block, thus limiting the possible clinical implications of fluid overload . Our study tends to support the view that this may be more important when epidural block with local anaesthetics is used.
In conclusion, the principal findings of the present study are that bowel surgery in pigs caused no swelling in normal or peri-anastomotic gut wall during the first 5 h of small intestinal anastomosis. Similarly, mesenteric oedema was not seen prior to gut resection. On the contrary, a highly significant oedema developed in the peri-anastomotic mesentery and this oedema was increased by the use of epidural bupivacaine.
L. Vestergård and P. Wara are to be thanked for technical and surgical support. The study was performed at Institute for Experimental Clinical Research, Skejby, Aarhus University Hospital. The study has received instrumental support from Baxter, Denmark and economic support from Aarhus University, AP Møller and Mrs Castile Mc-Kinney Møller Fond for the Support of Medical Research and the Research foundation of Oberstinde Kirsten Jensa La Cour.
1. Liu S, Carpenter RL, Neal JM. Epidural anesthesia and analgesia: their role in postoperative outcome. Anesthesiology
2. Rippe B, Grega GJ. Effects of isoprenaline and cooling on histamine induced changes of capillary permeability in the rat hindquarter vascular bed. Acta Physiol Scand
3. Marciniak DL, Dobbins DE, Maciejko JJ, Scott JB, Haddy FJ, Grega GJ. Antagonism of histamine edema formation by catecholamines. Am J Physiol
4. Minnear FL, DeMichele MA, Leonhardt S, Andersen TT, Teitler M. Isoproterenol antagonizes endothelial permeability induced by thrombin and thrombin receptor peptide. J Appl Physiol
5. Tighe D, Moss R, Bennett D. Cell surface adrenergic receptor stimulation modifies the endothelial response to SIRS. Systemic inflammatory response syndrome. New Horiz
6. Steinhelper ME, Fisher RA, Revtyak GE, Hanahan DJ, Olson MS. Beta 2-adrenergic agonist regulation of immune aggregate- and platelet-activating factor-stimulated hepatic metabolism. J Biol Chem
7. Rullier E, Laurent C, Garrelon JL, Michel P, Saric J, Parneix M. Risk factors for anastomotic leakage after resection of rectal cancer. Br J Surg
8. Gaarden M, Jespersen TW, Hessov I, Hansen HV, Rodt SÅ. Weight-increase due to water retention after thoracal epidural blockade is significantly greater than that seen using lumbar epidural morphine as the sole supplement to balanced anaesthesia in colonic surgery. Acta Anaesthesiol Scand Suppl
9. Randall HT. Fluid, electrolyte, and acid-base balance. Surg Clin North Am
10. Molnar R, Pian-Smith MCM. Spinal, epidural, and caudal anesthesia. In: Hurford WE, Bailin MT, Davison JK, Haspel KL, Rosow C, eds. Clinical Anaesthesia Procedures of the Masachusetts General Hospital,
5th edn. Philadelphia, USA: Lippincott Williams & Wilkins, 1998: 242-263.
11. Brown DL. Spinal, epidural, and caudal anesthesia. In: Miller RD, ed. Anesthesia,
5th edn. Philadelphia: Churchill Livingstone Inc., 2000: 1491-1521.
12. Aukland K, Reed RK. Interstitial-lymphatic mechanisms in the control of extracellular fluid volume. Physiol Rev
13. Mellander S, Johansson B. Control of resistance, exchange, and capacitance functions in the peripheral circulation. Pharmacol Rev
14. Holte K, Kehlet H. Prevention of postoperative ileus. Minerva Anesthesiol
15. Benoit JN, Zawieja DC. Gastrointestinal lymphatics. In: Johnson LR, ed. Physiology of the Gastrointestinal Tract,
3rd edn. New York: Raven Press, 1994: 1669-1691.
16. Adolphs J, Schmidt DK, Mousa SA, et al.
Thoracic epidural anesthesia attenuates hemorrhage-induced impairment of intestinal perfusion in rats. Anesthesiology
17. Johansson K, Ahn H, Lindhagen J, Tryselius U. Effect of epidural anaesthesia on intestinal blood flow. Br J Surg
18. Holte K, Kehlet H. Epidural analgesia and risk of anastomotic leakage. Reg Anesth Pain Med
19. Kehlet H. Surgical stress: the role of pain and analgesia. Br J Anaesth
20. Brandstrup B, Tonnesen H, Beier-Holgersen R, et al.
Effects of intravenous fluid restriction on postoperative complications: Comparison of two perioperative fluid regimens - a randomized assessor-blinded multicenter trial. Ann Surg
21. Holte K, Sharrock NE, Kehlet H. Pathophysiology and clinical implications of perioperative fluid excess. Br J Anaesth