The first one had the highest number of characteristic differences between benign conditions and CCA. Malignant mass of bile duct, of course, causes some degree of biliary obstruction, and sometimes liver parenchymal injury, resulting in a higher level of all types of liver enzymes. Common hepatobiliary-tumor markers (i.e., CEA, CA19-9, AFP), size of the tumor, a simple inflammatory marker, NLR, predicted the chance of malignant mass lesion very well. These findings were quite straight forward. Our previous study reported that CA19-9 level was associated with the level of invasiveness of bile duct tumors and predicted prognosis of the patients very well [3], since non-specific inflammation accounts for most mass-forming benign conditions, and patients with diabetes mellitus, who are immunocompromised, are prone to all types of infection, whether bacterial or parasitic. Diabetes mellitus was thus found to be associated with benign conditions. We unexpectedly found that eosinophil count was unable to differentiate benign lesions from CCA. We had hypothesized that a common parasitic inflammatory CCA-mimicking lesion, a Fasciolar liver abscess, might be associated with higher levels of eosinophil count. Actually, CCA in our region is also related to another kind of parasitic infection, Opisthorchis viverrini, therefore an elevated eosinophil count could not be used as a marker of benign conditions [14, 15]. There were no differences in the rate of abdominal pain and weight loss between benign conditions and CCA, after undergoing hepatic resection. An explanation for this was that the patients with symptomatic benign conditions would eventually undergo hepatic resection, regardless of their imaging findings. When considering hepatic resection for asymptomatic benign-suspected conditions, especially for the diabetes mellitus patients, the risk of postoperative liver failure and benefit of resection of benign lesions should be carefully weighed. We suggest that preoperative diagnosis should be as accurate as possible before hepatic resection. Owing to the simplicity of the percutaneous biopsy of mass- forming lesions, the young patients with small (<5 cm) mass with normal tumor markers level, those with high risk of postoperative liver failure, should undergo either extensively workup to rule out benign conditions or closely follow up, rather than hepatectomy. On the other hand, older patients with larger masses and rising tumor markers should be treated as having malignancies, until proven otherwise [16,17].

Bile duct tumor, including cholangiocarcinoma, can be manifested as an unexplained biliary dilatation or cystic lesion [10, 15]. We found that NLR and CA19- 9 level predicted the chance of malignancy very well. This finding is consistent with our previous report that NLR was a likely candidate for predicting malignancy, lymph node involvement, and survival of bile duct tumor, especially in patients with micro-papillary lesions, which are difficult to differentiate from benign conditions [18]. Having only a few markers able to differentiate benign from malignant biliary dilatation, we thus suggest that unexplained biliary dilatation should be extensively searched for the actual diagnosis, especially in a region that has a high incidence of CCA. Cholangioscopy, either per-orally or intraoperatively, is the best way to determine the actual cause of biliary dilatation. Eventually, if the chance of CCA was not fully eliminated after the patient being fully investigated, hepatic resection is an acceptable option [7, 9].

We found that neoplastic cysts tended to be of larger size, with more abdominal pain. The reason for this is the slow-growing nature of non-neoplastic cysts, which would distend the liver capsule gradually, compared with a more rapid enlargement of a neoplastic cyst. This is in line with the findings from a previous study that there was a correlation between lesion size and symptom intensity [11]. The small number of patients with benign cystic lesions in our series may account for a higher level of CA19-9 and CEA in neoplastic cyst, which was not statistically significant. Benign cystic lesions rarely mimic neoplastic cysts, eventually leading to therapeutic misadventures with undergoing hepatic resection, because these lesions are usually clearly described on cross-sectional imaging. Treatment for hepatic cystic lesions is different between neoplastic and non-neoplastic cysts. Neoplastic cysts, including cystic-IPNB and hepatic mucinous cystic neoplasm (hMCN), require formal hepatic resection [10], whereas non-neoplastic cysts require de-roofing only when the patient has symptoms. The breaking of neoplastic cysts, either by cyst de-roofing or aspiration, may lead to dissemination of the tumor cells, resulting in worsening of the patient survival. Selection of the operative procedure for the patient with cystic lesions usually relies mainly on imaging, because percutaneous biopsy and cholangioscopy are unable to be performed in this setting. We therefore recommend that standard hepatic resection, rather than de-roofing, should be performed if the chance of neoplastic cyst is not fully eliminated.

To the best of our knowledge this is the first and the largest study demonstrating the differences between benign conditions and CCA. The stratification according to main imaging finding is practical for the real-life situation. Moreover, we followed up the patients long enough to assure long-term outcomes of benign conditions. We intended to conduct this study over the period when our center performed all hepatic resection by open surgery, which allowed the surgeon to re-evaluate the lesions intraoperatively before performing hepatic resection. This simulated a real-life situation in most low to middle income countries. However, this study had some limitations. First, the retrospective nature from a single center may introduce some selective biases. Second, we were unable to obtain actual pathological diagnoses in the patients with non-specific inflammation. Lastly, during the study period, there was no uniform treatment protocol at our center, so the decisions regarding operative procedure were subjective. Moreover, most recent imaging technologies and interventions, allowing surgeons to make more accurate diagnoses, were not available at that time. In conclusion, this study provides a view of benign conditions mimicking CCA. We found that the proportion of hepatic resection for these conditions was quite low, only 3.78%. There were many characteristic differences used to differentiate benign mass-forming conditions from CCA. We recommend to perform biopsy in young patients with small (<5 cm) mass, especially those with normal liver functions and tumor markers. However, only a few markers were able to differentiate benign from CCA in the setting of cystic lesions and biliary dilatation. We recommend that standard hepatic resection be performed in every case, regardless of the type of imaging features, if the lesion has a chance of being CCA.