The initial phase of surgery is a full laparotomy to determine operability. In the authors’ opinion, the role for an initial laparoscopy is limited except to exclude peritoneal disease. Such patients tend to have had previous major and/or multiple abdominal surgeries limiting a full laparoscopic assessment. It is our practice to use an incision that will give adequate access for assessment, whilst being fairly minimalist initially in case there are clear signs of inoperability. It is often possible to make use of an old incision site from previous surgery, but inevitably a variety of approaches are satisfactory and depend on surgeon preference and the availability of mechanical retractors. In our practice, an upper midline incision with a right or full transverse extension is used most commonly. Adhesions should be divided to assess the primary tumour site and a careful examination of all peritoneal surfaces carried out. Doubtful areas should be sampled for frozen section histopathological analysis, and samples should also be taken from the coeliac nodes as this may suggest a more conservative approach for metastatic disease. The liver should be fully mobilised to allow adequate examination of the tumour and uninvolved liver. It is usual to sling the portal triad structures individually and the inferior vena cava above and below the liver. In addition, one must decide whether to divide the right suprarenal vein or not, and this will depend on the position of the tumour in relation to the inferior vena cava (and whether the lower IVC clamp will be above or below this vein).

The use of intraoperative ultrasound can provide additional information about the relationship of the tumour to portal triad and hepatic venous structures and may detect small metastases not seen on the preoperative CT or MRI. Although much information will have been gained by preoperative radiology, careful dissection to examine the hepatic artery and portal vein in perihilar cholangiocarcinoma or the inferior vena cava in metastatic disease, intrahepatic cholangiocarcinoma or hepatocellular carcinoma is recommended. This can all be completed without commitment to any irreversible steps, although we have frequently observed that final decisions about the degree of vessel invasion are made after liver resection and on the back table. For example, in cases where only parts of segments 2 and 3 are to be reimplanted, close application of the tumour to the portal bifurcation may necessitate resection to the level of the segmental divisions of the left portal vein. In addition, involvement of the biliary tree by metastatic tumours can necessitate a cholangiocarcinoma style approach, with resection of the biliary tree to the segmental level in order to gain a margin of surgical clearance.

Major blood vessel involvement should not prevent successful surgical resection as there are many strategies for vessel repair and conduit formation. Often an adequate repair can be created by a simple suture technique or end-to-end anastomosis. We have observed that an effective angioplasty/venoplasty patch can be created using vein remnants from the excised portion of the liver. Alternatively, the saphenous vein can be used to replace the hepatic artery, or opened out sections can be sutured together to create a wider vessel to repair the portal vein or inferior vena cava. The internal jugular vein, internal iliac vein, or external iliac vein can be used to replace a section of portal vein without compromise as collateral channels open up. If a wide area of inferior vena cava must be excised, then it is our preference to use a prosthetic graft. Some experience with vascular conduits made from vessels retrieved along with donor organs has also been reported, but there is a theoretical risk of allograft rejection and stricture formation. It has been our practice to use a jejunal Roux loop for biliary diversion to reduce the chance of ischemic stricture formation following biliary re-anastomosis.

In summary, there are many surgical options that should be considered when deciding operability.

The liver needs to be completely separated from the posterior abdominal wall and any lumbar veins draining into the IVC between the diaphragm, and the right suprarenal (adrenal) vein must usually be ligated and divided so that the IVC can be encircled in slings above and below the liver. This sounds straightforward, but it can be fraught with difficulty as the liver is often congested due to chronic venous obstruction. The common bile duct is divided and ligated. The portal vein and hepatic artery should be mobilised so that they can be clamped individually, maximising lengths for subsequent re-anastomosis. The femoral (IVC) bypass is begun at this stage before vascular clamps are applied to the portal vein, the hepatic artery and the superior and inferior levels of the IVC to be excised. The liver is now ischemic, and it should be rapidly removed to the bench. The portal limb of the bypass is inserted and secured with a snugger technique and portal bypass begun. Once portal and systemic veno-venous bypass has been established, the patient should remain stable for several hours.

Once the liver has been removed, it must be flushed (down the portal vein for cooling then down the hepatic artery and biliary tree) with a suitable organ preservation solution and cooled to 0–4 °C as in liver transplantation. UW solution is the current “gold standard” for liver preservation and is our choice for ex situ work. The accepted length of times for perfusion and preservation of a liver on the back table prior to reimplantation is not established, but our experience suggests that ex situ dissection times of between 2 and 5 h are associated with good results. The dissection bowl is kept cool by sitting it on sterile crushed ice.

Hepatic parenchymal fracture techniques or ultrasonic dissection (cavitron ultrasonic surgical aspirator—CUSA) may be used without the fear of blood loss during the back table dissection phase, but great care must be taken to ligate or clip all visible vessels or ducts to avoid significant haemorrhage at reperfusion. It is our practice to use a tissue sealant such as a fibrin glue at the end of dissection. The most common reason for ex situ hepatic work will be extensive involvement of the IVC or hepatic veins by tumour. Although the major hepatic veins are quite thick walled near the IVC, more peripherally they are very friable and great care needs to be taken with the choice of suture material and technique. Consideration should be given to the use of extra-venous patches to reduce tension, as is done in live donor liver transplantation. The resected IVC may be then replaced with an autologous vein graft or prosthetic graft. Whilst replacement with autologous vein has advantages in terms of infection and thrombosis, it is not always technically feasible, particularly if there is a need to replace a long segment of IVC. In fact, we have found that the best results are obtained with IVC replacement by a ring-enforced PTFE graft. The most ideal size appears to be 20 mm in diameter and sutured with 3-0 PTFE sutures or 2-0 or 3-0 polypropylene at implantation. The hepatic vein(s) of the liver remnant needs to be anastomosed to the IVC graft, first cutting suitable-sized holes in the side of the vascular graft, and we have found that 4-0 PTFE is a good suture choice for major veins and 5-0 or 6-0 PTFE or polypropylene for smaller veins. Our preference is to carry out these anastomoses ex situ, but others have shown good results with IVC graft placement immediately after liver excision to reduce the need for veno-venous bypass (by creation if a temporary portocaval shunt) and then in situ hepatic vein anastomosis after completion of the ex situ resection phase. The use of a vein patch between the hepatic vein(s) and the graft may reduce tension and allow anastomosis with a lower risk of hepatic vein tears (as noted above).