Surgical resection of cholangiocarcinoma is represented by anatomic hepatic resection. Advanced-stage neoplasms must often be treated by extended hepate-ctomy, with extension of the resection to the extrahepatic biliary tract, vascular hilar structures, vena cava and diaphragm. Usually, this type of neoplasm develops in a non-cirrhotic liver, which allows the surgeon to perform extended resection without the need of portal-vein embolisation. The mortality and morbidity rates in these cases vary from 3-9% and 30-40%, respectively [4,12-15].
The surgical approach has to be tailored in consideration of the gross type of neoplasm, the presence of multifocal lesions (perilesional satellitosis and distant metastases), vascular, biliary and serosal involvement and lymph-nodes metastases. However, even after aggressive resections the prognosis remains unsatisfactory, with a 5-year survival rate of 21-42% [11,15-19].
The intraductal growth (IG) type of ICC shows intraductal and/or granular growth and is sometimes associated with carcinoma that spreads over the superficial mucosa or with intraductal tumour thrombus. This type of biliary epithelial neoplasia is frequently associated with gastrointestinal metaplasia and overproduction of mucin and mucobilia, i.e. biliary intraductal growth mucin-pro-ducing cholangiocarcinoma . In these patients, in order to obtain correct preoperative staging of the tumour Sakamoto  recommended preoperative percutaneous biliary drainage and percutaneous cholangioscopy with biopsy in order to accurately assess the extent of the tumour into the intrahepatic segmental duct. This type of staging can be also achieved non invasively through the endoscopic route with peroral cholangioscopy.
After anatomic hepatic resection, the biliary duct margins should be evaluated by frozen sectioning. If neoplastic infiltration of the proximal biliary margin is diagnosed, the resection must be extended to the biliary confluence (Fig. 1). If the distal intrapancreatic margin is positive, pancreaticoduodenectomy is indicated.
Because of the high percentage of satellite nodules in mass-forming-type (MF) neoplasms, from 26 to 58% of patients treated surgically [1,2,7,10,21], an accurate intraoperative sonographic study is mandatory, both near the main lesion and of all the liver in order to identify metastatic nodules and verify the plane and radicality of resection. All suspicious nodules should be verified with intraoperative frozen sectioning. Anatomic resection with a section parenchymal margin at least 5-10 mm (R0) is adequate for MF neoplasms located peripherally (Fig. 2).
In cases of MF lesions either with biliary duct infiltration or located centrally near the biliary confluence, frozen section is always indicated to assess the margins of the biliary resection; if the results are positive, the resection has to be extended.
The presence of intrahepatic metastases represents a negative independent factor that is always associated with very low survival similar to those of palliative treatment: Madariaga  reported a 3-year survival that is nil and no survivors after 14 months; Isa  found a median survival of 19 months; in the series of Nakagawa , none of the patients with multiple lesions and extended lymph-node involvement reached a survival of 3 years, although Uenishi  published a 3-year survival rate in such cases of 6%.
In the light of these results, surgical indications for multifocal lesions are the subject of debate. We believe that multifocal lesions with unilobar satellite nodules or located in the same segmental area of the main lesion can be resected if the programmed operation is of low risk and no other negative prognostic factors, such as nodal metastases, are present.
Multiple bilobar lesions or lymph-node metastases contraindicate surgical resection, since R0 resection is not possible. In these patients with lymph-node and intrahepatic metastases, the prognosis is as poor as that of non-resected patients, and adjuvant or neoadjuvant therapies must be considered.
Periductal-infiltrating-type (PI) neoplasms that spread along Glisson's sheath through lymphatic vessels, with a low incidence of spreading in the portal system, frequently show nodal involvement as well as perineural and vascular invasion. In these cases, the surgical approach is the same as for MF neoplasms and necessitates anatomic hepatic resection always in association with biliary-duct sampling and extrahepatic biliary resection, if the margins are positive. Lymph-node dissection is also indicated with the modalities described below.
Ultimately, mixed forms (MF+PI) of neoplasms are characterised by considerable biological aggressiveness, with vascular and lymph-node infiltration in 80% of the cases, intrahepatic metastases in 46%  and a worse prognosis than either MF or PI neoplasms. The 5-year survival rate is between 0 and 7% [16,22]. These poor results are the consequences of early intrahepatic and extra-hepatic diffusion of the neoplasm, which makes it difficult to achieve R0 resection (Fig. 3).
In these cases, the surgical approach is the same as for the PI type, i.e. anatomic hepatic resection and biliary-duct sampling with likely extrahepatic biliary resection. It must be kept in mind that patients with advanced PI or mixed form (MF+PI) of cholangiocarcinoma have not benefited from extended surgery (hepatic and biliary resection, extended lymphadenectomy) in terms of improved surgical outcome compared to conventional resection (hepatic resection alone), with increased rates of mortality and morbidity. The surgical indications are controversial in these cases and must be limited to patients without other negative prognostic factors (intrahepatic metastases and positive lymph nodes).
The size of the neoplasm does not represent a prognostic factor that limits surgical indication, even if small tumours show a better prognosis than large ones. This suggests that if radical removal of the tumour is technically feasible and can be accomplished safely, it should be performed independently of the size of the intrahepatic tumour .
Vascular involvement is present in 27-85% of patients  and represents a negative prognostic factor. Casavilla , in a series of 39 patients, found that those with vascular infiltration did not reach 5-year survival. However, Inoue  reported 5-year survival rates of 60 and 22% in patients without and with vascular infiltration, respectively; survival in the latter group was significantly better than in non-resected patients. Therefore vascular invasion, even if it is a negative prognostic factor, does not represent an absolute contraindication to surgery.
Was this article helpful?