We read with interest the article by Riener et al.1 reporting the potential usefulness of Golgi Phosphoprotein 2 (GOLPH2) as a new serum biomarker for diagnosing hepatocellular carcinoma (HCC). We would like to draw attention to similar studies on the use of serum GOLPH2 in the differentiation of HCC patients from controls with no liver disease and patients with cirrhosis. GOLPH2 was originally described as a resident Golgi type II transmembrane protein expressed primarily in epithelial cells of many human tissues.2 In the normal human liver, GOLPH2 expression was found in biliary epithelial cells, but it was barely detectable in hepatocytes. However, expression of GOLPH2 was found to be strongly up-regulated in hepatocytes from patients with viral and nonviral liver diseases.3 A recent study showed that GOLPH2 was detected in human serum, and it was found to be elevated in the serum and tissue in the woodchuck model of HCC in comparison with the nonneoplastic liver.4 Thus, GOLPH2 may be a potential marker for the early detection of HCC.
We tested the serum GOLPH2 levels in patients with HCC (50), patients with cirrhosis (50), and controls with no liver disease (50). Our results showed that GOLPH2 was detected in the sera of all patients, but the levels were higher in patients with HCC and cirrhosis in comparison with controls with no liver disease. Median serum GOLPH2 levels were 1.35 relative units (range, 0.52–12.99) in controls with no liver disease, 4.39 relative units (range, 0.52–26.78) in patients with cirrhosis, and 14.37 relative units (range, 1–57) in those with HCC (P < 0.0001 for HCC versus patients with cirrhosis and controls with no liver disease). Median serum alpha-fetoprotein (AFP) levels were 2.46 ng/mL (range, 1.15–7.68) in controls with no liver disease, 6.11 ng/mL (range, 0.28–625) in patients with cirrhosis, and 16.2 ng/mL (range, 0.89–352,000) in HCC patients (P < 0.0001 for HCC patients versus controls with no liver disease and patients with cirrhosis). When patients with cirrhosis were compared to those with early HCC (United Network for Organ Sharing modified tumor-node-metastasis stages 1 and 2), the median serum GOLPH2 levels were significantly higher in those with early HCC: 12.69 relative units (range, 1–49) versus 4.39 relative units (range, 0.52–26.78) in patients with cirrhosis (P < 0.0001). Median serum AFP levels were also significantly higher in patients with early HCC than in those with cirrhosis: 11.23 ng/mL (range, 1.48–137,000) versus 6.11 ng/mL (range, 0.28–625) in patients with cirrhosis (P < 0.001). However, when we evaluated the sensitivity and specificity of GOLPH2 in patients with early HCC versus those with cirrhosis, the results showed that GOLPH2 (sensitivity, 62%; specificity, 88%) was significantly better than AFP (sensitivity, 25%; specificity, 97%; P < 0.0001).
In summary, our data highlight three points. First, we have shown that GOLPH2 is a better marker than AFP for diagnosing early HCC, partially concurring with the findings from Riener et al.'s study. Second, the sample size is not large enough, so we could not establish the performance of GOLPH2 according to the etiology of the underlying liver disease. Third, before this marker can be applied clinically, multicenter cohorts of validation are still needed.