Goal-Directed Therapy in Liver Surgery

A working knowledge of liver anatomy facilitates anesthetic planning. Multiple naming schemes have been proposed over the years for liver subdivisions, but a consensus definition was recently agreed upon [4]. The first-order division of the organ into right and left hemilivers follows the branching of the portal vein (Fig. 1A). The term lobe is no longer recommended as it was based on the surface anatomy of the falciform ligament and not the important vasculature. The right hemiliver is larger than the left and constitutes approximately two-thirds of the organ volume. The second-order division of the liver into sections is defined by boundaries following the three hepatic veins (right, middle, and left) [5]. The third-order division splits each section into superior and inferior parts called segments.

Viewing the right hand as a closed fist with the thumb tucked inside provides a convenient three-dimensional liver model (Fig. 1B) [6]. To remember the segments, simply start with the thumb (caudate) as segment 1 and count upwards while moving clockwise around the hand. A selection of anatomic liver resections and recommendations for hemodynamic monitors are listed in Table 1. Bleeding risk varies with numerous conditions including surgeon experience, presence of cirrhosis, number of segments resected, tumor size, and vascular proximity or invasion. Overall, the anesthesiologist should expect greater perioperative fluid shifts [7], metabolic acidosis [8, 9], and coagulopathy with the larger and right-sided resections [10].

Prevention, mitigation, and resuscitation of operative bleeding are a shared responsibility between the surgeon and anesthesiologist during liver resection. Despite advances in surgical technique and instruments [11], blood loss remains high and regularly exceeds 500 mL during major resections (> 2 segments). Large volume blood loss is associated with increased transfusions, morbidity, and mortality [12].

Both the inflow and outflow vessels of the liver contribute to bleeding (Fig. 1C). Surgical maneuvers are the most common means of inflow control [13]. The “Pringle maneuver” involves clamping the hepatoduodenal ligament that contains the portal vein and hepatic artery. This technique is limited to intervals of about 10 min to minimize liver ischemia [11]. The Pringle maneuver may reduce bleeding but is not always required in modern liver surgery [14]. “Selective inflow control” by clamping the left or right portal vein pedicles is also possible and reduces ischemia to the hepatic remnant. Pharmacologic inflow modulation with splanchnic vasoconstrictors such as vasopressin or octreotide is a current area of investigation. These drugs are commonly used in liver transplantation to reduce portal venous blood flow [15, 16], but there is little experience using them for liver resection [17, 18].

Inflow control does not eliminate reflux of blood from the hepatic veins, which remain a major source of hemorrhage. Many strategies have been investigated to address this problem. Surgical techniques such as a “hanging maneuver” [19], whereby the liver is lifted by a tape passed between the IVC and liver, can achieve outflow control. In more extreme or complex circumstances, infrahepatic IVC clamping or total vascular occlusion are occasionally deployed to reduce retrograde hemorrhage [13]. Unfortunately, IVC clamping increases the risk of thromboembolic events [20, 21]. A non-surgical strategy for outflow control is generally preferred. Consequently, the main technique is maintenance of a low CVP (typically below 5 mmHg) via volume restriction and/or vasodilation by the anesthesiologist.

In many situations, CVP is a poor marker of left ventricular preload and has low specificity to guide fluid administration [22]. However, CVP accurately reflects the hydrostatic pressure in the suprahepatic IVC and hepatic veins. This pressure is transmitted to the central venules and liver sinusoids (Fig. 1C), resulting in hepatic congestion and a greater tendency to bleed with surgical transection and manipulation [23]. While it is not always necessary to directly measure the CVP during liver resection [24], this physiologic concept guides restrictive fluid management regardless of hemodynamic monitors used.

Crystalloids are the standard fluids used to maintain and replace intravascular volume during surgery. Crystalloids are divided into “balanced” and “unbalanced” solutions in reference to buffer and electrolyte composition. Normal (0.9%) saline (“unbalanced”) contains no additional buffer or electrolytes and induces a hyperchloremic metabolic acidosis when administered in large volumes. Normal saline use is associated with increased morbidity in acute care [31], critical care [32], and surgical settings [33]. Thus, perioperative crystalloid use has largely shifted to favor “balanced” solutions such as Lactated Ringer’s (LR) and Plasma-Lyte.

Lactated Ringer’s and the very similar Hartmann’s solution contain about 130 mEq L⁻¹ of sodium and 3 mEq L⁻¹ of calcium. While large volumes can theoretically cause mild hyponatremia, this is generally preferred over the potential metabolic acidosis caused by normal saline. The lactate from LR, which undergoes hepatic metabolism to bicarbonate, may accumulate after liver resection due to decreased metabolic capacity of the liver remnant. The most common alternative to LR is Plasma-Lyte, a balanced crystalloid that contains significant amounts of acetate and gluconate anions [34]. Plasma-Lyte does not contain lactate or calcium. When patients receive Plasma-Lyte instead of LR/Hartmann’s during liver resection, lower lactate levels are observed [35, 36]. However, the clinical significance of lactate elevation in the LR group is unclear. Plasma-Lyte is also reported to yield less hyperchloremia and lower peak prothrombin time compared with LR when used during major liver resection [35, 36]. However, these perturbations have not shown any association with patient-centered outcomes.

Plasma-Lyte theoretically yields a stronger alkalinizing (pH-raising) effect due to its greater strong ion difference than LR and the presence of acetate, which is rapidly metabolized to bicarbonate independent of liver function [34, 37]. However, the clinical impact on base excess and pH appears minimal [35, 38]. In critically ill burn patients given large volumes of Plasma-Lyte, a transient accumulation of plasma gluconate occurs in synchrony with a depletion of ionized calcium [38]. These findings suggest that calcium undergoes chelation by gluconate. Ionized calcium should be carefully monitored when large volumes of Plasma-Lyte are administered [36], and calcium repletion may be required during or after liver resection. Gluconate undergoes partial hepatic metabolism and is mostly excreted by the kidneys [37]. Therefore, the chelation effect should diminish over time after liver surgery.

In conclusion, while balanced crystalloids are likely safer than normal saline as first-line fluids during liver resection, there is insufficient evidence to recommend a specific balanced crystalloid in this setting. Anesthesiologists should be aware of the differing biochemical profiles of LR and Plasma-Lyte during liver surgery.

Colloids are commonly utilized as an alternative to crystalloids to maintain oncotic pressure and expand intravascular volume. There are no large prospective studies guiding colloid administration during liver resection. Studies do exist in liver transplantation, but the quality of evidence is low and results may not be generalizable to liver resection [39].

Synthetic colloids such as hydroxyexthyl starch (HES) were used in the recent past as an alternative to human albumin. Numerous studies have investigated HES in critically ill and septic patients. These studies have shown more acute kidney injury and higher mortality in patients receiving HES [40]. Trials of HES in major abdominal surgery show no clinical benefit [41], and the drug may impair coagulation during liver resection [42•]. After integrating all available data, the US Food and Drug Administration issued a Boxed Warning that HES increases risk of acute kidney injury, bleeding, and death [43]. Consequently, HES should not be used during liver resection.

Albumin does not carry the same risks as HES. Albumin is commonly administered to patients with decompensated cirrhosis [44‐47] and safety in liver resection is extrapolated from extensive use in this population as well as decades of clinical experience. However, albumin is not always used during liver resection because of cost and a lack of evidence for improved outcomes [48]. Nevertheless, albumin remains a useful option for resuscitation during major liver resections [45]. Rapid bleeding, especially in fluid-restricted patients, precipitates hypotension and elevates risk of acute kidney injury if not quickly treated [49, 50]. Milliliter for milliliter, albumin infusion restores stroke volume and cardiac output faster and for a longer duration as compared to crystalloids [51]. Given the clear associations between allogeneic blood transfusion and postoperative complications after liver resection [52•, 53•], it is reasonable to manage rapid, self-limited hemorrhage with albumin boluses when hemoglobin levels are adequate.

While balanced crystalloid and judicious albumin boluses should be the first-line therapies for intravascular volume replacement in liver surgery, major resections are often accompanied by large volume blood loss and coagulopathy. Transfusion of blood components such as packed red blood cells, fresh frozen plasma, and platelet concentrates is often indicated. A complete review of intraoperative transfusion management is outside the scope of this article, but typical transfusion triggers apply. Notably though, coagulopathy is often more severe and persistent in liver resection than expected from dilution alone. This is due to perturbations of hepatic blood flow (particularly if Pringle clamping is performed) and reduced synthetic function of the liver remnant. Consequently, higher fresh frozen plasma to red blood cell transfusion ratios may be required as compared to other scenarios such as trauma. Caution is advised with the use of potent procoagulants such as prothrombin complex concentrates given the hypercoagulable risk in patients undergoing liver resection [54•]. Even when transfusing during active hemorrhage, the anesthesiologist should keep in mind the principles of restrictive fluid therapy for liver resection to avoid hepatic congestion (see Table 2).

Acute normovolemic hemodilution is an autologous transfusion technique that may be useful for reducing allogeneic blood transfusion during liver resection. To perform this technique, the patient’s starting hemoglobin is used to calculate a safe amount of blood that can be removed. Phlebotomy is typically performed through a large-bore central venous catheter. Typical volumes of whole blood removed are 500–2000 mL [55]. The volume deficit is replaced with a mixture of crystalloid and colloid. The blood is stored in bags containing an anticoagulant, and reinfused if needed to replace surgical blood loss or at the end of the procedure. Acute normovolemic hemodilution is reported to reduce allogeneic blood transfusion rates during major liver resection and may help lower the CVP [56]. This technique may be useful during liver resection as some surgeons remain hesitant to use autologous cell salvage (“cell saver”) during cancer surgery out of concern for disseminating malignant cells. Whether or not intraoperative cell salvage contributes to metastatic disease remains controversial [57•, 58].

Postoperative management varies and evidence-based protocols are not well established after liver surgery. One large trial did establish that restrictive fluid management over the first 24 h following major abdominal surgery is associated with acute kidney injury and does not improve survival [59]. However, less than 10% of the cohort underwent hepatobiliary surgery. Another major trial tested cardiac output monitoring versus usual care during and 6 h after major gastrointestinal surgery [60]. While there was a trend towards fewer complications and lower mortality in the group receiving cardiac output monitoring, the generalizability of this trial to liver resection is limited due to mixed surgery types and a protocol that routinely used inotropes. Overall, the evidence for specific fluid management protocols in the postoperative setting is weak.

Low CVP technique has been a mainstay of anesthesia management in liver resection for at least 25 years [61, 62]. Lowering CVP reduces the pressure gradient for retrograde venous bleeding (Fig. 1C). Additionally, low CVP facilitates outflow of blood from the hepatic veins, decreasing the total blood volume and pressure within the liver. Central venous pressure should be thought of as a correlate of liver congestion and not intravascular volume status. Multiple studies and subsequent meta-analyses have shown that maintaining a low intraoperative CVP during hepatectomy (most commonly < 5 mmHg) reduces blood loss and transfusions [21]. However, data on improved postoperative morbidity and mortality is conflicting [21, 63, 64], and the evidence base is noted to have significant risk of bias [65•, 66•].

Low CVP targets can be achieved by multiple methods, including restrictive intravenous fluid administration (with or without diuretics) [62, 64], hypoventilation to reduce intrathoracic pressures (which are transmitted to CVP), changes in body position (both Trendelenburg and reverse Trendelenburg) [67], vasodilators such as nitroglycerin or milrinone [68, 69], and anesthetics (volatiles, opioids, and epidural infusions) [21]. Multiple techniques can be combined, and the optimal approach is not known. The advantages of a low CVP must be balanced against the risks of systemic hypotension. Low CVP targets are associated with lower mean arterial pressure, which can put end organs at risk of hypoperfusion. Despite this risk, multiple studies do not show an increased incidence of acute kidney injury when targeting a low CVP [62, 64, 65•].

Given the risks of central venous catheters and recent trends showing decreased utilization by anesthesiologists [70], many patients undergo liver surgery without central access especially if it is only indicated for monitoring. Strategies known to achieve a low CVP can be implemented in most patients without direct measurement. There has been increasing interest in alternative measures and surrogates for CVP. One proposed strategy is transduction of peripheral venous pressures from large upper extremity veins [71]. This is accomplished by placing an 18- or 20-g peripheral intravenous catheter in a large antecubital vein. Several studies show excellent correlation between peripheral venous pressure and CVP, with a bias of < 2 mmHg, suggesting that this measurement may be used in place of CVP for patients undergoing liver surgery [72, 73]. Care must be taken to avoid specific scenarios that can decouple the correlation, such as noninvasive blood pressure cuffs and tucked or bent arms. No studies have yet shown that peripheral venous pressure guidance has any impact on blood loss or transfusions in liver surgery. Thus, while likely a viable strategy to reduce unnecessary central venous catheterization, more research is required for validation.

Over the last 15 years, minimally invasive liver resection (MILR) has rapidly grown in popularity and case volume around the world [88]. Initial work focused on minor and left sided resections. As surgical experience and skill has grown, it is now increasingly feasible to perform major, right sided, and even more complex resections. The rise in popularity of MILR is due to recognition of improved postoperative recovery (less pain, shorter hospital stays), potentially decreased blood loss in some cases, and equivalent oncologic outcomes [3]. Various techniques exist including total laparoscopic and laparoscopic hand-assisted. Robotic liver surgery is also growing in popularity. Minimally invasive liver resection may be preferred in patients with underlying cirrhosis due to a decreased risk of postoperative liver decompensation [89].

Anesthesiologists face additional challenges in fluid management and hemodynamic monitoring during MILR as compared with open surgery. Like all laparoscopic procedures, pneumoperitoneum during MILR has a significant impact on the cardiovascular system. Abdominal insufflation with carbon dioxide determines the intraabdominal pressure (IAP, typically 10–15 mmHg for MILR) [90]. Insufflation causes an initial increase followed by a decrease in venous return to the heart. The use of reverse Trendelenburg position is common in MILR and further impairs venous return. Under these conditions, cardiac output and renal perfusion may be compromised.

Pneumoperitoneum also directly perturbs the CVP and decreases its utility as a marker of intravascular volume status. The CVP upon insufflation is affected by airway pressure due to the coupling of intrathoracic pressure and IAP [91]. CVP increases more with insufflation at higher airway pressures. Despite this, surgical bleeding tends to decrease because the relevant driving force is the difference between IAP and CVP (Fig. 1) [90]. Conceptually, pneumoperitoneum helps to tamponade venous bleeding. The best operative conditions may be obtained with high insufflation pressure, low CVP, and low airway pressure. However, at extremely low CVP, the risk of carbon dioxide embolism increases [92]. Inflow control maneuvers analogous to the Pringle are still possible [14, 93]. Due to the restricted field and magnification, even low-volume bleeding can impair surgical visualization and increase risk during MILR. Optimizing the surgical and anesthetic conditions to reduce bleeding increases the likelihood of operative success.

During MILR, maintenance of mean arterial pressure with fluid restriction and vasoconstrictors is recommended. Goal-directed therapy is difficult due to limitations of SVV and PPV monitoring in this setting. Increased IAP, low tidal volumes, and possibly reverse Trendelenburg position all impact the relationship between SVV/PPV and intravascular volume status [75, 94]. Despite the limitations, SVV monitoring is reported to improve outcomes in MILR compared to CVP monitoring, with benefits including reduced blood loss and reduced conversion to open surgery. In one randomized trial of patients undergoing MILR, targeting an SVV > 12% was found to reduce blood loss and conversion to open surgery as compared to CVP monitoring [95]. These findings further reduce the incentive for central venous catheter placement in routine MILR.

Despite limitations in dynamic measures of volume responsiveness, arterial line monitoring remains routine for all major MILRs. In particular, resections involving segments 1, 4a, 7, and 8 or a right posterior sectionectomy (6–7) are technically complex for the minimally invasive surgeon with higher expected blood loss [89]. Additionally, variable compression of parenchyma and the IVC may result in sudden drops in venous return during retraction of the liver or manipulation with laparoscopic tools. Aside from a vigilant eye on the surgeons’ screen and robust bidirectional communication, the arterial line remains an important continuous monitor to diagnosis instability that can be relieved by relaxing the liver to a neutral position.

Surgical emergencies such as massive hemorrhage, venous gas embolus, or thromboembolism should trigger volume resuscitation. These crises take priority over elegant restrictive fluid management. In these situations, the anesthesiologist must rapidly shift their thinking from restrictive fluid therapy to resuscitation according to the principles of GDT to maintain vital organ perfusion. The team must be aware of the risks of “starting behind” in volume resuscitation particularly when fluid restriction is the primary strategy. Conceptually, fluid-restricted patients are positioned near a steep drop-off on the Frank-Starling curve with minimal bleeding. Should the IVC require urgent clamping for repair, intracardiac filling may be all but lost in a fluid restricted state. Judicious administration of volume to restore cardiac output is vital, and overcorrection may impair control of venous bleeding. A fluid restrictive state also increases risk for entraining venous air; appropriate vigilance is indicated. It is important to proceed with balanced blood component transfusion in situations of high-volume blood loss, with monitoring of labs including arterial blood gas, with electrolytes and considering viscoelastic coagulation testing. In cases of MILR, prompt conversion to an open procedure should also be considered, to reduce the burden of pneumoperitoneum and to obtain expeditious surgical control with awareness that bleeding may increase upon opening the abdomen.