Edula et al.1 in their retrospective study of 172 patients concluded that serum concentrations of cancer antigen-125 (CA-125) are elevated in cirrhotic patients with ascites. This relationship has been well known for several decades,2–6 as well as the relationship of CA-125 with pleural and pericardial effusions, among many other non-malignant conditions,4,5,7 because CA-125 is expressed in the coelomic epithelium.8 In fact, ascitic fluid concentrations of CA-125 are even higher than in serum.2
Edula et al.1 said that cirrhotic patients without ascites had normal mean CA-125 concentrations, and define this observation as “a new finding”. However, these observations are not new. More than 25 years ago our group published a prospective and comprehensive study of 159 patients with liver diseases, including 85 cirrhotics with and without ascites.3 In this article, cited by Edula et al. as their reference 14, we found the same results as Edula et al.1 reported for ascitic and non-ascitic patients.1 Moreover, our group published other articles focused exclusively on patients without ascites.9,10 Like Edula et al.,1 we found that elevations of CA-125 concentrations in patients without ascites were uncommon, but also that these elevations were related to the degree of liver dysfunction.
Therefore, CA-125 increases markedly (even 100 times above the upper normal level) in patients with ascites, an increase that we previously found to be highly proportional to the amount of ascites as measured semiquantitatively in 5 degrees, and it decreases rapidly with the diminution of ascites. In fact, CA-125 proved to be a reliable marker of ascites (sensitivity 98.4%, specificity 95.9%, positive predictive value 93.8%, negative predictive value 98.9%, efficiency 96.9%).3
Although to a considerably lower degree, and much more infrequently, we also previously found increased CA-125 concentrations in some patients who did not have ascites, as evaluated ultrasonographically, as well as significant correlations between CA-125 and some liver function markers, such as albumin and prothrombin.9 Our findings suggested that, apart from the determinant role of ascites, liver dysfunction itself is also responsible for moderate increases in the serum concentrations of CA-125, probably due to a poor metabolization of this glycoprotein.9,10
On the other hand, the absence of statistical significance regarding portal hypertension (p = 0.1) reported by Edula et al.1 is surprising, considering that the differences in the CA-125 concentrations that they found were substantial (414 vs. 256 U/mL) and that portal hypertension is tightly related to ascites. The authors evaluated portal hypertension by a history of esophageal varices. Besides the limited sensitivity of this method to detect portal hypertension, perhaps the number of patients studied was too small (Type II error). Also, the parametric tests used by the authors for statistical calculations (t-test, ANOVA) might have been inappropriate, as we found that the distribution of CA-125 was markedly non-Gaussian, and the illustrations of Edula et al.1 suggest the same.
In our series, we found marked differences in CA-125 concentrations in patients with or without portal hypertension, as evaluated separately by three different methods: echography, esophageal varices and splenomegaly (p < 0.0001 for each).3 Even portal hypertension was associated with higher CA-125 concentrations in patients without ascites (p = 0.049).9
Declarations
Conflict of interest
The author has no conflict of interest related to this publication.