Portal de Periódicos da CAPES

Venovenous bypass in orthotopic liver transplantation: Time for a rethink?

2005; Lippincott Williams & Wilkins; Volume: 11; Issue: 7 Linguagem: Inglês

10.1002/lt.20482

1527-6473

K. Ratna Reddy, Susan Mallett, Tim Peachey,

Organ Donation and Transplantation

The successful use of venovenous bypass (VVB) in orthotopic liver transplantation (OLT) was first reported by Shaw in 1984.1 Subsequently the technique was generally adopted as a de facto standard, with the vast majority of transplant centers in the United States and Europe using VVB routinely during OLT. It was felt that the use of VVB was associated with improved hemodynamic stability and reduced blood loss, and that it allowed less experienced surgeons more time during the anhepatic phase. Despite these perceived advantages, many clinicians have questioned the reported benefits of VVB in OLT. There are also increasing concerns about the morbidity and even mortality that can be associated with its use. This is reflected in the fact that in the mid 1990s, many centers in the United States were slowly moving to a more selective use of VVB, with less than 50% using it routinely. However there is still enormous variability between centers, both in the frequency of use and the indications for bypass.2, 3 In 2003, a survey of the transplant centers in the United Kingdom found that 2 centers used VVB routinely, 3 used it selectively (10-30% patients), and 2 centers used it rarely, if ever. This difference is despite 5 of the centers routinely using caval preservation techniques (unpublished data). The marked differences between transplant centers both in terms of the use of and indications for VVB is somewhat perplexing and has stimulated this review. A literature search for venovenous bypass/liver transplantation, renal failure/liver transplantation, reperfusion syndrome/liver transplantation was performed using Medline, PubMed, Embase, Cinahl, [email protected] These searches revealed 84 studies, of which 34 were relevant to the topic. Further references were obtained from these papers and by hand search of relevant transplant journals. In total 47 studies were identified, of which only 2 were randomized controlled trials. Only 1 study showed improved outcomes with the use of bypass; however, this was not a randomized controlled trial.1 The need for a bypass circuit during OLT was first recognised by Moore in 1960.4 Early trials involving humans used passive shunts during OLT to bypass the flow from the inferior vena cava (IVC) to the superior vena cava via the internal jugular vein. However this was associated with a high incidence of pulmonary embolism resulting in the death of 3 patients out of the first four transplants performed.5 In 1979, Calne et al introduced partial cardiopulmonary bypass to attenuate the hemodynamic instability resulting from occlusion of the vena cava. They used femorofemoral bypass with a pump-oxygenator, which collected blood from the IVC and returned it to the femoral artery. Although this provided the desired cardiovascular stability, the need for systemic heparinization resulted in uncontrollable bleeding.6 Later modifications to avoid the use of heparin by removing the oxygenator from the circuit led to marked desaturation of renal arterial blood, and the technique was subsequently abandoned.7, 8 The early trials in Pittsburgh were also complicated by massive blood loss following the use of heparin.9 In 1983, Griffith et al. introduced a VVB system that used a centrifugal force pump and heparin bonded tubing,10 and in 1984, Shaw et al. described the successful use of VVB in clinical liver transplantation without the need for systemic heparinization.1 Abbreviations: VVB, venovenous bypass; OLT, orthotopic liver transplantation; IVC, inferior vena cava; RPP,- renal perfusion pressure; CO, cardiac output; SVR, systemic vascular resistance; MAP, mean arterial pressure. VVB involves the extracorporeal circulation of blood from the venous system below the caval clamps (inferior mesenteric and femoral veins) and return to the central veins (axillary or internal jugular veins). The pressure in the IVC during cross-clamping is usually in the range of 30-50 mm Hg11 facilitating drainage of blood along the bypass tubing to a centrifugal, nonocclusive pump. Siphoning is further aided by the slight negative pressure created by the pump head. Blood is then pumped into the return cannulae. Shunt flows described in the literature range from 1.5-5.0 L/min.1, 12, 13 Scherel et al. have shown that shunt flows less than 3.0 L/min failed to normalize caval pressure gradient and renal perfusion pressure (RPP), whereas flows greater than 5.0 L/min did not substantially further improve hemodynamic state.14 Double-limb bypass (portal and femoral vein) remains standard; however, an increasing number of centers use only single-limb (caval) bypass.15, 16 Classically, vascular access for bypass is established by inserting outflow cannulae into the portal and femoral veins and inflow cannula into the axillary vein following surgical cutdown. However, this had a high incidence of associated complications (seromas, lymphoceles, wound infection, and nerve injuries). A percutaneous technique was first described in 1994 by Oken et al.17 Overall this technique has been shown to be quicker and easy to perform, and by avoiding open dissection especially of the axilla, it reduces complications such as seromas.18, 19 Studies have demonstrated better shunt flows and hemodynamics with the percutaneous technique.18, 20 During the anhepatic phase of liver transplantation, the suprahepatic and infrahepatic parts of the IVC and the portal vein are cross-clamped. Clamping of the IVC results in the loss of venous return from the lower part of the body, and this causes a sudden and profound decrease in central venous and pulmonary capillary wedge pressures and cardiac output (CO). The degree of cardiovascular instability depends on the extent of collateral circulation and the underlying cardiovascular reserve. The natural collateral veins (azygous, epidural, and superficial abdominal veins) are limited in their ability to increase flow. Cross-clamping of IVC typically results in a 50% reduction of cardiac index, but mean arterial pressure (MAP) is generally relatively well maintained because of the compensatory increase in systemic vascular resistance (SVR) and moderate increase in heart rate21; however, there is wide variability in these responses.22-24 The majority of patients will tolerate IVC clamping for a period of about one hour.25 Beyond this period, decompensation becomes increasingly apparent with a fall in mixed venous oxygen saturation and MAP. Patients with cirrhosis generally tolerate portal vein clamping well due to the presence of enlarged portosystemic collaterals around the gastro-oesophageal and splenic areas.26 This allows for significant improvement in venous return during caval clamping and limits the fall in CO. Clamping of the portal vein leads to a marked rise in hydrostatic pressure in the portal venous beds causing a reduction in splanchnic perfusion, exacerbation of bowel edema, and bleeding due to the increase in portal venous pressure. Renal perfusion pressure (RPP) is reduced by as much as 50% due to a combination of the increase in renal venous pressure due to renal venous outflow obstruction and decreased MAP.27 Venous bypass has been shown to maintain adequate venous return and cardiac filling pressures during the anhepatic phase, thereby maintaining hemodynamic performance.1, 28 However, VVB ameliorates but does not fully prevent the fall in CO or rise in SVR. Studies have demonstrated that CO decreases by 25% and SVR increases by 50%, with little or no change in cardiac filling pressures even with the use of VVB.29, 12 In addition, the use of VVB does not guarantee adequate perfusion of abdominal organs and lower limbs, as venous return is not maintained at prebypass levels. There is only partial decompression of venous circulation below the level of the clamps, as evident from increased IVC pressures (24-38 mm Hg), even when bypass is used.27 A reduction in CO (of greater than 50%) during the anhepatic phase has been shown to be associated with an increased perioperative morbidity and mortality.30 Hemodynamic instability following test clamping of IVC is the most common indication for initiating VVB in many centers, although the criteria used vary among centers. Veroli et al. suggested that hypotension ( 50%) during a 5-minute test period of hepatic vascular occlusion can be used to identify the group of patients who require VVB.30 Schwarz et al. retrospectively analyzed 182 patients undergoing OLT and found no increase in perioperative morbidity and mortality in patients in whom the decrease in CO was greater than 50%, compared with patients with less pronounced reduction in CO.24 The need for fluids and inotropes was similar in both groups, but it is of note that while there was no significant difference in MAP between the groups, mixed venous oxygen saturation was significantly lower in the group with low CO. In addition Wall et al. showed that despite a reduction in CO by 50%, a decrease in systemic blood pressure by 21%, and an increase in SVR, the mortality rates were comparable to other reported series where VVB was used routinely.28 These results question the practice of basing the indication for VVB during the anhepatic phase solely on a reduction in CO of >50% after trial clamping of IVC. It is the overall hemodynamic stability that is important, but both the way that it is defined (e.g., fall in CO and/or fall in MAP) and the way that is managed (fluid and/or inotropes) varies among centers, making comparison of the published literature difficult. Adequate maintenance of preload is essential prior to application of the cross-clamps to be able to tolerate the hemodynamic instability during the anhepatic period. Administration of fluids and blood products to compensate for the marked fall in preload during the anhepatic phase (unless carefully titrated) may lead to fluid overload and pulmonary edema following reperfusion of the transplanted liver, particularly in patients with poor cardiac reserve. It has been estimated that in patients for whom VVB is not used, up to an additional 3,000 mL of fluid would be required to maintain adequate filling pressures.31 An alternative approach is to maintain the central venous pressure (CVP) at low values during the dissection phase to reduce blood loss and then use vasoconstrictors during the anhepatic phase. Presence of pulmonary hypertension, impaired ventricular function from previous myocardial infarction, ischemic heart disease, and cardiomyopathy are some of the cardiac conditions where VVB is indicated.32-34 In patients with pulmonary hypertension, excessive fluid loading to compensate for hemodynamic changes occurring during anhepatic phase may result in acute right ventricular dysfunction, which may exacerbate hypotension following reperfusion of the graft. Patients with cardiomyopathy have impaired left ventricular function and consequently a limited ability to generate an adequate CO in the face of the acute increase in SVR normally seen during the anhepatic phase. The compensatory mechanisms (increase in SVR and heart rate) along with arterial hypotension during anhepatic phase may exacerbate underlying ischemia in patients with coronary artery disease. Hence, VVB may benefit this group of patients who tolerate the anhepatic period poorly. However, there are no studies to demonstrate any benefit of using VVB in such patients. Most centers use VVB in patients with impaired renal function with a view to preventing further damage to the kidneys during the anhepatic phase and to reduce the need for postoperative renal support. Patients undergoing OLT are at high risk of developing postoperative renal dysfunction. Factors that have been implicated in the development of renal insufficiency include preexisting renal dysfunction, acute changes in intraoperative hemodynamics, and pharmacological agents influencing renal function (e.g., immunosuppressants, amino glycoside, antibiotics.), and increased intraoperative transfusion of blood products.35 VVB has been postulated to reduce the incidence of acute renal failure in patients undergoing OLT by decompressing the venous circulation and maintaining the MAP, thus improving RPP. It has been demonstrated that urine output during the anhepatic period is a linear function of RPP (r = 0.57), but when expressed as a percentage, the decrease in urine output related to the decrease in RPP is less strongly correlated (r = 0.40).23 The initial study by the Pittsburgh group showed that in patients for whom VVB was not used, the mean maximum rise in serum creatinine was 3.0 mg/dL within 3 days after transplantation, compared to 1.5 mg/dL in patients for whom bypass was used. In addition, 6 out 57 patients in the nonbypass group required dialysis in the first postoperative week, compared to none in the bypass group.1 However, subsequent studies have failed to reproduce this effect.13 In 1996, Johnson et al. retrospectively analyzed 2 groups of patients (group 1 [112], OLT with VVB; group 2 [38], OLT without VVB) undergoing OLT. They found no significant differences between the 2 groups in terms of postoperative serum creatinine levels and the need for hemodialysis.36 Grande et al. prospectively randomized 77 patients undergoing OLT into those procedures performed with VVB or without VVB.29 A significant decrease in inulin clearance and increase in urinary β2- microglobulin and N-acetyl-β-D glucosaminidase (NAG) excretion (markers of glomerular filtration rate and tubular damage, respectively) were found during the anhepatic phase in patients undergoing OLT without VVB compared to patients with VVB (Table 1). This study demonstrated that patients with VVB showed a lower degree of renal impairment during the anhepatic phase than those without VVB; however, this effect could not be demonstrated in the subsequent perioperative phases, at 24 hours or 7 days after surgery. In addition, there was no difference between the 2 groups with regard to the need for hemodialysis during the first postoperative week. Low MAP at the time of induction of anesthesia was found to be the only independent risk factor for early postoperative severe renal failure. Patients with advanced chronic liver disease have some degree of renal impairment. This is thought to be due to the activation of endogenous vasoconstrictors that are released in response to a decrease in systemic blood pressure seen in these patients, causing a decrease in renal blood flow. Furthermore, induction of anesthesia is associated with some degree of hypotension and reduction in RPP. Hence, it is perhaps not surprising that a low MAP preoperatively is associated with renal failure in the perioperative period. There is always some degree of renal dysfunction following liver transplantation regardless of the use of VVB, as demonstrated in this study. What is also evident from this study is that VVB does not ensure adequate RPP. Subsequent studies have confirmed that renal function markedly deteriorates during OLT, and use of VVB does not seem to prevent this.37 More recently, Cabezuelo et al. have compared the effect of standard surgical technique vs. piggyback technique on the development of postoperative acute renal failure. They found standard surgical technique to be an independent risk factor for the development of postoperative renal failure. In addition, the use of VVB, the presence of postreperfusion syndrome (PRS), and the transfusion of fresh frozen plasma and cryoprecipitate (>20 units) were found to be additional risk factors that contributed to postoperative renal failure. The use of piggyback technique was shown to reduce the probability of acute renal failure after liver transplantation.38 However, there are no studies that specifically investigate the beneficial effect or otherwise of VVB in patients with impaired renal function. Approximately 75 % of patients with fulminant hepatic failure have cerebral edema.39 During OLT several factors can compromise cerebral perfusion, leading to neurological sequelae. The lack of an adequate collateral venous circulation in these patients can profoundly compromise systemic blood pressure and, hence, cerebral blood flow. Infusion of large fluid volumes to compensate for the cardiovascular instability may result in fluid overload and compromised cerebral venous drainage that may worsen cerebral edema. In addition, acid metabolites including carbon dioxide released following reperfusion of the grafted liver results in cerebral vasodilatation, further increasing intracranial pressure. VVB has been shown to limit these hemodynamic changes. It is postulated that routine use of VVB is associated with a reduction in neurological sequelae due to cerebral edema. Prager et al. retrospectively reviewed records of 10 patients with fulminant liver failure undergoing OLT without VVB and found no evidence of neurological sequelae in the postoperative period.40 They also demonstrated that with adequate maintenance of systemic blood pressure using vasoconstrictors and the use of cerebral protective measures (thiopental), VVB was unnecessary. VVB has been reported as being associated with a systemic inflammatory response.41 The increased capillary leak associated with a systemic inflammatory response may further worsen cerebral oedema. A more profound rise in PaCO2 and hydrogen ion concentration has been shown to occur following reperfusion of the graft in patients undergoing OLT with the use of bypass, compared with those patients for whom a piggyback technique was used.42, 43 This has been attributed to better organ perfusion, particularly of splanchnic tissues, during the anhepatic phase with the piggyback technique. In addition, use of the piggyback technique in this group of patients has been shown to maintain hemodynamic stability and RPP.44 The rationale for using VVB in this situation is to reduce portal venous pressure and mesenteric bed congestion associated with venous clamping. The elevated venous pressure during this phase is mitigated to a certain extent by the reduction in CO and splanchnic vasoconstriction. In addition, extensive venous collaterals in cirrhotic patients help to decompress the portal system during the period of venous occlusion. However, large varices, particularly those in the retroheptic area, can be associated with severe bleeding and make the hepatectomy extremely difficult. In this situation, early devascularization of the liver with clamping of the caval and portal veins is usually essential, and early institution of VVB may be helpful in these circumstances. The number of patients undergoing OLT without the need for blood transfusion is increasing every year. However, despite refinements in surgical techniques, use of antifibrinolytic agents and coagulation monitoring, a certain obligatory blood loss is almost inevitable during many OLT procedures. Factors that contribute to bleeding during recipient hepatectomy include the presence of portal hypertension, intrinsic coagulopathy associated with severe liver disease, and previous abdominal surgery. Initial studies demonstrated a significant reduction in blood product requirements (18.9 vs. 32.7 units) in the bypass group,1 and this was attributed to venous decompression by the bypass system enabling hemostasis. It must be noted that patients in this study were compared with historical cohorts, when hemostatic techniques and coagulation management were still in the experimental stages. In 1986, Wall et al. found that in patients for whom bypass was not used, the mean blood requirement during OLT was 13.5 units.28 Subsequent studies have failed to demonstrate significant differences with regard to transfusion requirement in either group.36, 15 More recently, Fan et al. have shown that use of VVB is associated with an increase in transfusion of red cells (15 vs. 6 units, P < 0.001).45 This is thought to be due to increased fibrinolysis,46 hemolysis,47 and platelet adhesion48 to the bypass tubing. An absolute contraindication to the use of VVB is acute Budd-Chiari syndrome, as the use of VVB in such patients is associated with poor shunt flows and increased risk of pulmonary embolization.1 The use of VVB is not without risks, and serious fatal adverse effects have been reported. A North American survey of 50 major liver transplant centers reported a complication rate of 10-30%, with 1 death from pulmonary embolism.3 Complications can be divided into those associated with use of extracorporeal circuit and those related to vascular access. These include hypothermia and air and thrombotic pulmonary embolism. Hypothermia during OLT is not uncommon, and several factors contribute to heat loss. The use of an extracorporeal circuit that does not contain a heat exchanger potentiates heat loss at a rate of (0.75°C per hour) during the anhepatic phase of the procedure.49 Hypothermia not only has deleterious effects on myocardial function but also can exacerbate bleeding by its effect on coagulation. Most centers now routinely incorporate a heparin-bonded heat exchanger into the circuit to prevent hypothermia. Cases of fatal pulmonary embolism have resulted from thrombus either forming in the extracorporeal circuit or being translocated from the IVC to the right atrium.3, 50 Blood clotting in the bypass system may occur if the flow rate is very low (<1 L/min).1 These include lymphoceles, hematoma formation, air embolism, major vascular injury, nerve injury, and vessel thrombosis. The reported complication rate with standard techniques is about 10%,3 the most common complication being lymphocele formation. The reported incidence of lymphocele formation with this technique ranges from 15.6-18.6%.51, 52 Brachial plexus injury has been reported in patients with surgical cutdown of axillary veins (2.5%).53 Complications following percutaneous techniques are rare, but when they do occur they can be life threatening. Budd et al. retrospectively reviewed 312 patients who underwent OLT with VVB using a percutaneous technique using large bore cannulae (18-20F). They reported 4 cases of serious morbidity (hemothorax; incidence,1.28%) and 1 death (0.32%) directly related to placement of percutaneous cannula.54 The authors are also aware of other unpublished reports of deaths occurring as a direct result of vascular trauma following percutaneous central venous cannulation for establishing VVB. Air embolism is a potential problem during decannulation of the vessels, but it has also occurred in the bypass circuit.55, 56 It has been suggested that use of ultrasound guidance to facilitate the placement of the cannulae would reduce complications.54 Transient hemodynamic instability as a result of systemic vasodilation and some degree of myocardial depression is well recognized following reperfusion of the grafted liver. Where this effect is profound it is called "post reperfusion syndrome," a term that was first described by Aggarwal and colleagues in 1987.57 They reported a 30% incidence of PRS in patients undergoing OLT with the use of bypass. In contrast, Estrin's group reported an incidence of 3.7% in their series where VVB was not used.23 However, the definitions of PRS and other conditions were not the same, and it is not possible to make a direct comparison. Estrin suggested that the lower incidence of PRS could be explained by the increased volume of fluid that is given to patients who are not supported by VVB, making them more tolerant of vasodilation at reperfusion. In addition, it is possible that patients who are not bypassed have higher levels of endogenous vasoconstrictors as a result of the physiological responses to caval cross-clamping. The causes of PRS are multiple, but they include metabolic changes and release of vasoactive substances from the reperfused liver.58-62 Exposure to any extracorporeal circulation leads to activation of inflammatory cytokines and other vasoactive substances that will tend to increase vasodilatation and hypotension.63, 64 There are, therefore, a number of theoretical reasons why the use of VVB might be associated with a higher incidence of PRS. Other disadvantages that have been described with the use of VVB include increased operative time and warm ischemia time.65 However, this paper compared data from non-VVB patients with data from a group of historical controls who were bypassed. The experience of the surgical team may not have been directly comparable, perhaps contributing to longer operative times in the earlier cohort. When VVB was used routinely in Pittsburgh, Shaw et al. found an improved 30-day survival in the bypass group (91%) compared to (73%) in the nonbypass group. The reported benefits at 90 days failed to reach significant levels, with 73.2% surviving in the bypass group vs. 68.3% in the nonbypass group, and at 6 months there was no difference in the mortality rates between the 2 groups.1 It should be noted that these patients were compared with previous cohorts (1982-1984), when the use of VVB, surgical experience with OLT, and advanced anesthetic management and monitoring of coagulation was still in its early stages. However, it is of note that other centers reported comparable results without the use of VVB.28 In 1996, Johnson et al. retrospectively analyzed records of 150 patients for whom VVB was used on a selective basis and of whom only 62 underwent OLT without VVB and 74 underwent OLT with VVB. The 30-day mortality rates were similar between the two groups (13.2% vs. 15.2%, P = not significant). The mortality rates were similar at 6 months (21.1% vs. 26.8%) and 1 year (68.6% vs. 71.7%). VVB was deemed to be indicated in this study, if hemodynamic instability developed following test cross-clamping of suprahepatic IVC for a period of 2 minutes, despite maximal fluid resuscitation.36 Chari et al. retrospectively analysed bypass practice in 547 patients in Duke and Toronto University Hospitals between 2 different periods (1986-1992 and 1994-1996). VVB was used routinely between the period of 1986-1992 and selectively (most common indications being intraoperative hemodynamic instability with a MAP of less than 60 mm Hg following test clamping of IVC, and portal vein and major bleeding during hepatectomy between 1994-1996. The 1-year actual survival rate of 215 patients having transplantation with routine use of VVB was 71.9%, vs. 89.7% in the 332 patients during the period of selective use. The rates of retransplantation were 9.7% and 3.9%, respectively, and a reduction in any type of postoperative complications was by a third. This was attributed mainly to increasing experience and use of sophisticated surgical and anesthetic techniques in these centers.3 VVB is associated with unique morbidities and is not essential to the success of OLT; hence, alternative techniques have been developed. The piggyback technique is an attractive alternative that preserves the recipient IVC and allows continuation of venous return to the right heart during the anhepatic phase. Blood flow is still impaired to varying degrees as a result of caval narrowing from the application of side clamps. Since its introduction in 1968 by Calne,66 this technique has undergone several technical modifications. The piggyback technique is routinely used in reduced-size graft and pediatric living-related liver transplantation. A number of centers are successfully using caval preservation techniques without VVB or passive portocaval shunts for the majority of primary transplant procedures.67 Although this technique is widely used among various centers in United States and Europe in adult OLT, there is wide variability in its use in UK transplant centers. In addition, it has made only a marginal impact on use of VVB, with some centers continuing to bypass all patients, even when caval preservation techniques are used. In reviewing the literature, it is evident that there is still controversy concerning the use of VVB. While there is a trend to reduce its use to carefully selected indications, many centers continue to use it routinely and others not at all. Strangely, when there is such diversity of opinion, hard evidence in the form of adequately powered, randomized, controlled trials is notable for its paucity. Of the few studies that actually meet this requirement, none have demonstrated a benefit of VVB over simple occlusion techniques. Much of the literature is based on historical comparisons of routine vs. later selective use, and this unavoidably introduces an element of bias, as outcome will tend to improve with increasing experience and expertise. Although there are many theoretical benefits of VVB, the data supporting them are soft (Table 2). The move to selective use indicates that more centers have recognized that with adequate experience of the surgical and anesthetic team, it is not of routine benefit. Moreover, the use of VVB is not without risk. There clearly are circumstances where VVB will be of benefit in particular situations and to particular patients; however, it is difficult to tease out any hard and fast rules. The range of indications for VVB and the relatively small number of patients who might demonstrably be shown to have an improved outcome with bypass would make the construction of randomized controlled trials extremely difficult. To be adequately powered, multicentred studies might be considered, but this would introduce further confounding variables that would limit their value. For the foreseeable future, the decision as to whether or not to use VVB will depend to a large degree on institutional culture, personal experiences and preferences and professional judgement.