Ascitic fluid "humoral tests of malignancy".

B. Runyon

IF: 17.298

Hepatology

Abstract:The recent paper by Jungst et al. (1) is another in a series of papers that propose a new ascitic fluid “humoral test of malignancy” (2, 3). The basic premise of these publications is that the ascitic fluid cytology is insensitive in detecting cancer as the cause of ascites formation. In general, cytologic analysis is said to detect only about two-thirds of patients with malignant ascites, as if all patients with malignant ascites should have peritoneal carcinomatosis (2, 3). However, in my experience, cytology has greater than 90% sensitivity in detecting peritoneal carcinomatosis, but is negative in patients with massive liver metastases and resulting portal hypertension-related ascites (Runyon, B. A. and Hoefs, J . C., Hepatology 1985; 5:1000, Abstract). The problem is that a significant fraction of patients that are included in the series of patients with intraabdominal tumor have only massive liver metastases without peritoneal carcinomatosis. Patients who have only massive liver metastases do not have positive ascitic fluid cytologies. It is my impression that investigators have assumed that the vast majority of cancer patients who develop ascites have peritoneal carcinomatosis as the cause of ascites formation. In fact, ascites caused by massive liver metastases is not rare. With careful investigation, massive liver metastases have been found to cause ascites in 26 to 33% of patients who have cancer and ascites; in general, patients have either peritoneal carcinomatosis or massive liver metastases, but not both (4; Runyon, B. A. and Hoefs, J. C. , Hepatology 1985; 51000, Abstract). It is the inclusion of patients with massive liver metastases in series of malignant ascites that leads to the 67% sensitivity value of cytology. Cytology is much more sensitive when it is applied to the appropriate subgroup of malignant ascites patients-the patients with peritoneal carcinomatosis. Ascitic fluid caused by peritoneal carcinomatosis is markedly different than fluid caused by massive liver metastases. The former has a high protein concentration and low serum-ascites albumin concentration gradient (unless the patient also has cirrhosis), and the latter has a low protein concentration and a high gradient (5). The ascitic fluid of patients with massive liver metastases is so similar to that of cirrhotic ascites that it is possible that some patients with massive liver metastases might have been erroneously classified into the cirrhotic ascites category of some studies. Fibronectin concentration (mean f S.D.) in ascitic fluid caused by massive liver metastases-related ascites is 55 f 21 pg per ml compared to 123 & 45 pg per ml for peritoneal carcinomatosis; only 20% of massive liver metastases-related ascites specimens have a fibronectin concentration greater than 75 p g per ml (6) , which is the proposed lower limit cut-off for malignancy (3). To analyze malignant ascites without subgrouping patients is to ignore the heterogeneity that is present in the series. Another problem with 2 of the 3 papers on “humoral tests for malignancy” (1, 2) is that they did not include a control group of noncirrhotic, nonmalignant ascites samples. In my study, the fibronectin concentration of “miscellaneous” samples (pancreatitis, heart failure, renal failure, etc.) overlapped essentially completely with that of the malignant ascites samples (including both subgroups), rendering the test useless in discriminating between miscellaneous and malignant samples (6). In my study of ascitic fluid cholesterol and triglycerides (analyzed by Technicon Autoanalyzer), the overlap between groups was also excessive (see Figures 1 and 2). Although the cholesterol concentration of cirrhotic ascites samples (18 & 13 mg per dl) was significantly (p < 0.001) lower than that of peritoneal carcinomatosis samples (83 f 37 mg per dl) and significantly (p < 0.02) lower than massive liver metastases samples (43 k 23 mg per dl), there was no difference between the cholesterol of miscellaneous samples (63 & 30 mg per dl) and that of either subgroup (or the aggregate) of malignant samples. There were no differences between any groups regarding ascitic fluid triglyceride concentration in my study. Tests that cannot discriminate between malignant samples and miscellaneous samples are of limited value, in my opinion. It is unrealistic to expect any one ascites test to detect all patients with malignant ascites just as it was unrealistic to expect the plasma fibronectin test to detect all patients with cancer (7). What we need are better definitions of subgroups of malignant ascites patients and stratification of data into subgroups. What we do not need are more ascitic fluid “humoral tests of malignancy.”