Hepatic metastases from carcinomas of unknown primary site.

AIM Hepatic metastases are often present at diagnosis of carcinoma of unknown primary site (CUP). The objective of this study was to describe the diagnostic and therapeutic strategies used. METHODS One hundred and eighteen patients were treated at the Cancer Center of Montpellier from 1993 to 2002 for CUP initially metastatic to the liver. Initial characteristics, diagnostic tests, chemotherapies and outcome were retrospectively recorded. RESULTS The most frequent histological types observed were adenocarcinoma, undifferentiated, neuroendocrine and squamous-cell carcinomas. Hepatic metastases revealed the cancer in 66 patients and were isolated in 32 patients. Other metastatic sites involved lymph nodes, lung and bone. Pretreatment computed tomography scans of the chest, abdomen and pelvis evaluation were available for more than 80% of patients. Colonoscopy, gastroscopy and bronchoscopy were performed in 58, 56 and 26% of patients respectively. One hundred and seven patients had received at least a front-line of chemotherapy. Seventy-four patients had received platin salt-based chemotherapy, 67 in front-line treatment and 10 in second line. In first-line chemotherapy, overall response rates with or without platin were 19.4 and 20% respectively. One hundred and two of 111 deaths were due to disease progress and seven toxic-related deaths occurred. The median survival was 6.6 months, and 7.8 and 4.6 months in the with or without platin groups respectively (P=0,35). The median survival for treated patients was 7 months. Multivariate analysis identified two prognostic factors: serum lacto-dehydrogenase level and performance status. CONCLUSIONS According to this study, pretreatment evaluations, which were very extensive in some patients, were insufficient to identify the primary site of liver metastases. Because of the poor prognostic, chemotherapy, in absence of clinically demonstrated benefit, has to be reserved for patients with better prognosis. Prospective trials are needed to determine whether use or not of cisplatin should be proposed for standard protocols.