Background and study aims: There is a common misconception that malignant ascites is equivalent to peritoneal carcinomatosis. It seems that malignancy-related ascites is a more appropriate description of malignant ascites, which is difficult to confirm. Carcinoembryonic antigen, a glycoprotein tumor marker shed by malignant cells, increases in a wide range of gastrointestinal malignancies. We carried out the current meta-analysis to determine carcinoembryonic antigen accuracy in the diagnosis of malignancy-related ascites.
Patients and methods: Pubmed/Medline and SCOPUS were searched using these search terms: malignan* AND ascites AND (CEA OR carcinoembryonic). The outcome of interest was carcino-embryonic antigen accuracy in the differentiation of malignancy-related ascites and nonmalignant ascites.
Results: Seven studies were included in this systematic review. Pooled diagnostic indices using random-effects model were as follows: sensitivity 43.1% [381-48.3]; specificity 95.5% [93-97.3]; LR+ (positive likelihood ratio) 7.33 [4.58-11.73]; LR- (negative likelihood ratio) 0.6 [0.54-0.68]; and DOR (diagnostic odds ratio) 12.93 [7.58-22].
Conclusions: Carcinoembryonic antigen of the ascitic fluid does not seem to be sensitive enough to diagnose malignancy-related ascites. However, due to high specificity, the positive predictive value of this marker is high and the higher the level of carcino-embryonic antigen, the more likely it is to be malignancy-related. Nevertheless, a negative test result cannot definitely rule out the malignancy.