Background/aims: Prognosis of gastric cancer with concomitant liver metastasis is poor. Gastrectomy and chemotherapy had been reported to be beneficial to this group of patients. Whether all the patients can benefit from that treatment modality and whether the clinical characteristics can give some information about survival have not been evaluated.
Methodology: Eighty-three gastric cancer patients with concomitant liver metastases who had received treatment in the past 10 years were retrospectively studied with special reference to different extent of liver metastases. Clinical characteristics (peritoneal tumor dissemination, tumor markers, clinicopathological factors), treatment modalities (gastrectomy or chemotherapy) and survival were analyzed for their possible relationship with the extent of liver metastases.
Results: There were 33 patients with liver metastases limited to one lobe and 50 patients had metastases in both lobes. The clinicopathological characteristics show no difference between uni- and bilobar liver metastases except higher percentage of concomitant peritoneal dissemination in patients with bilobar disease. More unilobar involvement patients underwent gastrectomy (70% vs. 48%, p=0.018). Higher frequency of abnormal CEA level in bilobar metastases (38% vs. 60%, p=0.045). Median survival of unilobar metastasis is 7.8 months and 4.3 months for the bilobar involvement (p=0.001). Gastrectomy might prolong the survival in patients with unilobar metastasis (p=0.005), but not in patients with bilobar diseases (p=0.074). Chemotherapy could prolong the survival in patients without gastrectomy, but not the survival after gastrectomy.
Conclusions: The clinicopathological characteristics and tumor markers have no role in predicting the survival among patients with liver metastases. Both gastrectomy and chemotherapy for the patients without gastrectomy could prolong survival, but chemotherapy could not prolong the survival after gastrectomy, the benefit of resection and chemotherapy are probably caused by selection bias.