Chapter 46B. Perspective on Liver Surgery

Chapter 45 expertly describes the preoperative evaluation, patient selection, integration of chemotherapy, and results of resection for colorectal cancer (CRC) metastatic to the liver. Below are comments about the technique of resection.

Most liver resections can be done with a right subcostal incision extended vertically in the midline to the xyphoid process. Patients prefer this incision to the rooftop or “Chevron” incision that rarely is needed. Inspection and palpation should exclude extrahepatic disease and assess intrahepatic disease. At this point, intraoperative ultrasound (IOUS) should be performed to determine the size and location of all tumor nodules. IOUS can help plan the resection to maximize the chance of obtaining a negative margin. With no contraindication to resection, the appropriate resection can be performed. Resections depend on the segments that need to be removed. A diagram of Couinaud's segments is in Fig. 46B-1.

The nomenclature for resections depends on the segments removed as summarized in Table 46B-1. Right hepatectomy is the most common major liver resection and it is described in detail below.

Right Hepatectomy

The right side of the liver is mobilized by dividing the right triangular ligament and the anterior and posterior leaflets of the right coronary ligament. The inferior vena cava (IVC) is exposed and small branches from the liver to the IVC are ligated and divided to facilitate exposure and avoid tearing. At this point, attention is focused on the porta hepatis for the hepatoduodenal dissection.

The gallbladder is removed and the bile duct is traced to the liver. The hepatic artery is identified and the right hepatic artery (RHA) isolated, usually anterior to the hepatic duct but occasionally posterior to the hepatic duct. The RHA is ligated and divided, carefully protecting the main and left hepatic arteries. The right hepatic duct is divided and the hepatic duct can be lifted anteriorly to the left to expose the portal vein (Fig. 46B-2). The junction of left and right portal vein (RPV) is exposed cautiously and a vessel loop is placed around the RPV. It is preferable to ligate and divide the RPV with sutures or a vascular stapler that ligates and cuts. At this point, inflow control is achieved and the liver will quickly demarcate such that the right liver will discolor. It now is reasonable to maximize outflow control.

An alternative method to achieve inflow control is to isolate and ligate the right portal pedicle intrahepatically. This can be done by incising the capsule in segments four and six and placing a finger around the pedicle in the liver. IOUS can monitor this technique. A vascular stapler can be applied to control the inflow.

The liver is rotated anteriorly and to the left. Small veins emptying directly from the liver into the IVC are ligated and divided. This might require suture ligation of the veins at the IVC as these branches are quite short. The surgeon should work from caudad to cephalad to secure these branches; usually there are about four pairs. A broad fibrous band of tissue extends from the IVC to the liver just caudad to the right hepatic vein (RHV). This tissue has small veins in it or next to it. There can be a larger accessory RHV in or next to this tissue. This band needs to be divided to expose the RHV (Fig. 46B-3).

Careful sharp and blunt dissection usually can isolate the RHV outside the liver such that it can be divided with clamps or a vascular stapler. The vascular stapler with 2.5-mm staples works well; clamps and Prolene sutures occasionally slip or tear producing significant bleeding. With inflow and outflow control, the parenchyma is transected to obtain a 1-cm margin. If the central venous pressure (CVP) is kept low, then backbleeding will be minimized. To minimize the risk of air embolism when the CVP is low during transection, the patient should be kept in a Trendelenberg position. There are several methods for parenchymal transection. The standard technique is “finger fracture” that requires fracturing the parenchyma with the index finger and thumb to define vessels and bile ducts. The vessels and ducts are ligated or clipped and the liver is expeditiously transected (Fig. 46B-4). Variations of the finger fracture technique use a Kelly clamp, closed scissors, or a metal suction device. The key is to use a method that works well for the surgical team such that one surgeon can do the fracturing while another surgeon secures the vessels. During this maneuver, bleeding can be minimized by occluding inflow at the porta hepatis as a Pringle maneuver with clamping for 10–15 minutes with 2–5 minutes of unclamping. This maneuver can be repeated until the liver has been divided. A normal liver without steatohepatitis should tolerate 45 minutes of normothermic ischemia.

Other devices have been designed to divide the parenchyma. These include the ultrasonic dissector, the water jet dissector, the saline-linked radiofrequency (RF) dissecting sealer, the bipolar vessel sealing device, and the harmonic scalpel. Each device has its proponents. Some use these devices to meticulously and slowly divide the complete parenchyma of the liver, suturing, tying, or clipping vessels or ducts. Others use these devices to divide a portion of the parenchyma; when they are deep in the parenchyma and when the margin is not an issue, these surgeons switch to a finger fracture technique to finish the dissection. To expedite deep parenchymal division, many will use vascular staplers. Vascular staplers control vessels efficiently and they can be used for parenchymal transection. To perform transection of the right liver can require over 15 vascular stapler cartridges. The parenchymal transection can be performed in less than 10 minutes without the need for a Pringle maneuver. The staplers can rip vessels if not guided appropriately. The firm metal blade of the stapler should be introduced carefully; there should be no force required to introduce the stapler. The obvious argument against staplers is cost. If the stapler companies would reduce the costs of the staplers, many surgeons would use multiple loads of the vascular stapler to divide the liver. If two experienced surgeons use finger fracture, the cost of transection is small and the resection can be performed with little blood loss. If a liver surgeon performs finger fracture with an inexperienced resident, the time for transection and the blood loss can be considerable. The RF resection device (such as the Habib) ablates liver tissue that then can be cut with a knife or a bipolar cautery device. The problem with these devices are twofold: (1) cost—one RF resection device can cost as much as 30 or more stapler cartridges; and (2) vessels less than 5 mm in diameter can be ablated but larger vessels can bleed significantly from the holes of the electrodes. The RF resection devices are very effective for peripheral nonanatomic wedge resections and left lateral liver resections. Theoretically, the radiofrequency ablation (RFA) can treat microscopic tumor cells at the margin. The first laparoscopic liver resection with a RF resection device was done at the Brigham and Women's Hospital (BWH),1 but now most laparoscopic liver resections at the BWH are done with the staplers or a variation of the finger fracture method.

The above operation can be performed with a laparoscope either as a complete laparoscopic procedure or as a hand-assisted laparoscopic liver resection.2,3 Multiple methods of parenchymal transection can be performed laparoscopically. The vascular staplers work well with the use of a handport. A reason to use staplers to transect the liver during open liver resection is to become familiar with this technique for laparoscopic resection. The intra-abdominal pressure with pneumoperitoneum can tamponade some of the smaller venous bleeding. This nuisance bleeding can be controlled with the Argon Beam Coagulator, topical sealants, or with sutures as necessary. The Argon Beam Coagulator should be used with caution as a venous embolus with Argon does not dissipate as quickly as carbon dioxide. Draining the right upper quadrant usually is done for several days if there is concern about a possible bile leak.

Patients should be discharged in 4–5 days. Blood transfusions should be avoided, if possible, as the survival rate decreases after liver resection if transfusion is used.

Left Hepatectomy

Left hepatectomy essentially is the mirror image to right hepatectomy. A hilar dissection controls inflow (Fig. 46B-5). If the caudate lobe needs to be removed, it can be freed from the IVC facilitating isolation of the left hepatic vein (LHV). A stapler can be used to divide it. Alternatively, the parenchyma can be divided and the LHV can be secured during the parenchymal dissection (Fig. 46B-6).

Administering chemotherapy directly into the hepatic artery via a catheter introduced through the femoral artery has been practiced for over 30 years. Toward the end of the 1970s, a totally self-contained implantable pump was developed that allowed continuous arterial infusion of chemotherapy via a catheter placed in the gastroduodenal artery at the junction of the hepatic artery. The rationale for HAI chemotherapy takes advantage of the observation that CRC metastases in the liver derive the majority of their blood supply from the hepatic artery, not the portal vein.4 Further, the drug most commonly used for HAI, fluorodeoxyuridine (FUDR), has over 95% uptake in the liver on first pass as compared to 5-fluorouracil (5-FU) that has less than 50% uptake on first pass.5 This major difference allows HAI with FUDR at 100–400 times the concentration of systemic 5-FU.

Randomized studies evaluating HAI with FUDR versus systemic 5-FU conducted in the 1980s showed response rates of 30–50% with HAI FUDR but did not conclusively demonstrate a survival benefit with this technique.5-10 Thus, the appropriate role for this treatment continues to be debated. The literature is clear that the frequency and severity of complications decrease with increasing experience of the surgeon. These complications include pump malfunction, pocket problems, catheter occlusion or dislodgement, and arterial complications (hemorrhage, thrombosis, or perfusion problems).10 Currently, the indications for HAI chemotherapy are twofold: (1) for the treatment of unresectable liver-only metastases from CRC; and (2) as an adjunct after the resection of CRC metastatic to the liver. Both indications are controversial. With no definitive randomized study showing a clear benefit to HAI chemotherapy, a multi-institutional trial of HAI with FUDR versus systemic 5-FU-based chemotherapy was conducted in the 1990s. The HAI chemotherapy group had about a 3-month median survival time advantage over the systemic therapy group.11 However, this trial was criticized by many medical oncologists because (1) the survival time advantage was considered too short to argue for pump implantation; and (2) newer systemic agents provide longer survival times than the systemic regimen used in this trial. A single-institution randomized trial showed a survival advantage at 2 years with HAI chemotherapy after liver resection versus no HAI chemotherapy.12 This trial was criticized because there was not a significant long-term survival advantage in the HAI group. At the moment, HAI chemotherapy is not widely used principally because most oncologists are not comfortable managing pump chemotherapy and they feel systemic chemotherapy is adequate.

The technique of pump implantation requires insertion of a catheter into the gastroduodenal artery with the tip close to the junction with the hepatic artery. The pump is placed in the subcutaneous space on the abdominal wall. The gallbladder is removed to obviate chemical cholecystitis. Potential arterial branches from the hepatic artery to the stomach or duodenum are interrupted to avoid chemical duodenitis or gastritis.

RFA is a treatment that destroys tumors in situ by protein denaturation and thermal coagulation. It has been used to treat tumors since the early 1900. Beer used RF coagulation through a cystoscope to treat bladder tumors in 1908. Cushing used RFA to treat intracranial tumors in 1926. In 1990, two groups independently used ultrasound to guide RFA treatment of liver tumors. RFA uses alternating current with a frequency between 10 kHz and 900 MHz. Tissue adjacent to the electrode dies from the heat generated in the tissue by RF energy; the electrode itself does not become hot but it delivers RF energy that causes mechanical friction in the tissue that leads to thermal ablation.13

Resection is the best treatment for CRC metastatic to the liver. RFA is used as an alternative to resection when (1) comorbid disease precludes resection, or (2) anatomical considerations preclude resection. Examples of the latter indication principally are twofold. First, if a discrete number of metastases are located in different segments such that resection of the lesions would result in inadequate residual liver volume, then RFA would be reasonable. Second, if several lesions in one-half of the liver can be removed with an ipsilateral hepatectomy but a lesion deep in the contralateral half of the liver cannot be removed, an ipsilateral hepatectomy with RFA of the lesion in the contralateral liver would be appropriate.

RFA can be performed percutaneously in an imaging suite with computed tomography (CT) scan or ultrasound guidance, or in the operating room either open or laparoscopically with ultrasound guidance. Complication rates associated with RFA in one large review and one single-institution study were 7 and 2.4%, respectively.14,15 The reported procedure-related mortality rate is 0.5% (10 deaths in 1931 patients).14 Complications are more frequent in those undergoing open RFA versus percutaneous RFA. Complications include abscess, biloma, bile duct injury, pleural effusion, and pain.16,17

Some claim that RFA for solitary CRC metastases is associated with infrequent local recurrences. The M.D. Anderson group recently examined this issue. With a median follow-up time of 31 months, they found a 37% local recurrence rate for RFA treatment of solitary metastases as compared to a 5% local recurrence for resection (p = .0001).18 The subset with smaller tumors—less than 3 cm in diameter—had a 31% local recurrence rate with RFA compared to 3% with resection. Clearly, RFA of CRC metastases to the liver is not equivalent to resection; RFA should be considered investigational or—at best—an inferior, second-choice treatment.

Cho and Fong appropriately state that imaging modalities should differentiate hemangiomas, focal nodular hyperplasia (FNH), adenomas, and cancers. When it is difficult to confirm the diagnosis of a benign (or malignant) tumor,19–23 a biopsy can be considered; however, as the authors state, liver biopsies of suspected tumors notoriously are inaccurate, and complications such as bleeding and spread of tumor can occur. The inadequate accuracy and bleeding complications of liver biopsy particularly become relevant when attempting to distinguish the atypical FNH that does not need resection from the adenoma that does.

Focal Nodular Hyperplasia

FNH is the second most common solid tumor or the most common nonvascular benign tumor of the liver. It possibly comes from a polyclonal hyperplastic response to locally altered blood flow.24 Resection infrequently is necessary to diagnose or treat pain. Pain usually is from a source other than the FNH; thus, a search for other causes is mandatory. If pain is from the FNH, the FNH usually is large but a small FNH can cause pain.

Hepatocellular Adenomas

Hepatocellular adenomas (HCAs) typically occur in young females who are taking oral contraceptive pills (OCPs). The initial management is to stop the OCP which should shrink the tumor. This action both confirms diagnosis and treats the tumor. In general, all adenomas greater than 5 cm or symptomatic should be resected. But this approach generates questions that cannot be answered adequately because the natural history of adenomas is not well known. These questions include:

If an adenoma shrinks to less than 5 cm with cessation of the OCP, can it safely be observed? Probably, but adenomas less than 5 cm can bleed or transform into hepatocellular carcinomas (HCCs).25

Will small adenomas cause trouble during pregnancy? Probably not. Dokmak et al reported 11 patients followed through pregnancy with no problems.26 But adenomas less than 5 cm can bleed and transform,25 and pregnancy is associated with a significant increase in estrogen that stimulates the growth of adenomas. It probably is reasonable to monitor a pregnant patient with a small adenoma by using ultrasound of the liver during pregnancy.

Does adenomatosis of the liver (defined by most as >10 adenomas) demand a liver transplant? The literature certainly is not clear about this issue. The Mayo Clinic used a definition of adenomatosis to be more than three tumors and they recommended observing adenomas less than 3 cm in diameter and resecting adenomas greater than 5 cm.27 Others have reported selective transplantation for adenomatosis.28

Can small adenomas be treated by percutaneous RFA? The literature does not have adequate information to answer this question.

Regenerative Nodular Hyperplasia

These tumors do not need treatment but they are becoming an increasing clinical problem because they are associated with chemotherapy given before resection of CRC metastatic to the liver. Wicherts et al found regenerative nodular hyperplasia in 22% who had six cycles of preoperative chemotherapy containing oxaliplatin. Distinguishing these tumors from other tumors becomes an issue, and these tumors are associated with an increase in postoperative hepatic morbidity (50 vs 29%).29

Hepatocellular Carcinoma

The best treatment of HCC is resection but the cirrhosis generally associated with HCC makes resection difficult. As Cho and Fong note, it is difficult to use preoperative information to determine whether or not a minimally cirrhotic liver can tolerate resection. In selected cases, preoperative portal vein embolization (PVE) might demonstrate whether or not a chronically damaged liver has the ability to grow. If 4–6 weeks after PVE the future liver remnant does not grow, then liver resection probably would not be tolerated. At this point, this strategy is speculation at best.

After resection, RFA is a second choice for treatment of HCC. Ideally, HCC should be resected with a healthy negative margin, but this approach can be difficult from a bleeding standpoint as well as from a liver function standpoint. To minimize bleeding and theoretically optimize tumor control as well as liver function, it is reasonable to use a RF resection device to ablate a plane of liver tissue at the edge of the tumor. Theoretically, this should kill microscopic tumor at the margin, and minimize bleeding during resection.

Chapter 43 expertly discusses hepatic abscesses and cystic diseases of the liver. Given the advances in laparoscopy, the management of cystic diseases of the liver particularly has evolved. Patients with large liver cysts who have symptoms typically have pain. A search for other causes of pain is reasonable as most liver cysts do not cause pain. If pain appears to be due to a large liver cyst, then a magnetic resonance imaging (MRI) with magnetic resonance cholangiopancreatography (MRCP) is reasonable specifically to examine the cyst lining and any possible communication with the biliary tree. If the lining does not have nodularity and there is no obvious communication with the biliary tree, then laparoscopy with aspiration of the cyst contents followed by generous laparoscopic resection of protruding cyst wall can provide significant tissue for frozen section analysis. While awaiting frozen section analysis, the remainder of the cyst lining can be examined and biopsied if there is any concerning tissue. If frozen section reveals a simple cyst, then the remaining lining can be cauterized to minimize recurrence of the cyst. If frozen section reveals cystadenoma, then the remaining epithelium should be removed. If it technically is not possible to remove the epithelium laparoscopically, then open resection should be considered. An alternative is to cauterize laparoscopically the remaining epithelium in the cyst. This should kill any residual cystadenoma which should prevent cystadenocarcinoma from occurring. We have used this approach selectively although there are no long-term large series demonstrating the efficacy of this strategy. If frozen section reveals cystadenocarcinoma, then formal resection to obtain negative margins—either by open or laparoscopic technique—should be performed.