Meta-analysis of prognostic factors for overall survival in patients with resected hilar cholangiocarcinoma

Background Hilar cholangiocarcinoma is staged using the AJCC staging system. Numerous other prognostically important histopathological and demographic characteristics have been reported. The objective of this meta-analysis was to assess statistically the effect of postresectional tumour characteristics on overall survival of patients undergoing attempted radical curative resection for hilar cholangiocarcinoma. Methods Relevant studies were identified by searching the Ovid MEDLINE and PubMed databases. The search was limited to studies published between 2009 and 2017. Papers referring to intrahepatic or distal cholangiocarcinoma were excluded from review. Data extraction used standard Parmar modifications to determine pooled univariable hazard ratios (HRs). Results Twenty-four articles, containing 4599 patients, were assessed quantitatively. In pooled analyses, age (HR 1·16, 95 per cent c.i. 1·04 to 1·28), T category (HR 1·49, 1·30 to 1·70), lymph node involvement (HR 1·78, 1·65 to 1·93), microvascular invasion (HR 1·49, 1·34 to 1·68), perineural invasion (HR 1·54, 1·40 to 1·68) and tumour differentiation (HR 1·54, 1·38 to 1·72) were significant prognostic factors, with low heterogeneity. Portal vein resection (HR 1·54, 1·15 to 1·70) and resection margin status (HR 1·77, 1·57 to 1·99) had significant effects, but with high heterogeneity. Sex, tumour size and preoperative carbohydrate antigen 19-9 levels did not have a statistically significant effect on postoperative prognosis. Conclusion Several tumour biological variables not included in the seventh edition of the AJCC classification affect overall survival. These require incorporation into prognostic models to ensure a personalized approach to prognostication and treatment.