Background and aims Golgi protein 73 (GP73) as a potential serum marker for hepatocellular carcinoma (HCC) has not been validated in large cohort studies. Furthermore, its significance in the assessment of tumour recurrence after HCC resection remains unknown. The aim of this study was to determine the value of serum GP73 in the diagnosis of HCC.
Methods Serum GP73 and alpha-fetoprotein (AFP) were compared in a total of 4217 human subjects in this multicentre study, including 1690 healthy adults, 337 hepatitis B virus (HBV) carriers, 512 patients with cirrhosis, 789 patients with HCC, 61 patients with other malignant liver lesions, 206 patients with benign liver lesions and 622 patients with 14 different kinds of non-liver cancers. The main outcome measures were the specificity and sensitivity of GP73 in patients at risk for the development of HCC.
Results Using 8.5 relative units as a cut-off value, the sensitivity and specificity of serum GP73 for HCC were 74.6% (95% CI 71.5% to 77.6%) and 97.4% (95% CI 96.8 to 98.3%), compared with 58.2% (95% CI 55.2% to 62.1%) and 85.3% (95% CI 83.4% to 88.1%) for AFP (p<0.001) using 35 ng/ml as a cut-off value. The GP73 level was significantly increased in patients with HCC compared with healthy controls (14.7 vs 1.2, p<0.001). Although GP73 levels in HBV carriers (2.9) and patients with cirrhosis (4.7) were somewhat elevated, they were much lower than that in patients with HCC (p<0.001). GP73 decreased following surgical resection of HCC lesions and increased with tumour recurrence. Fourteen types of non-liver cancers were analysed; all the benign and other malignant liver lesions had moderate elevations of GP73, albeit at a much lower level than in HCC.
Conclusions GP73 is an accurate serum marker for the detection of HCC and its recurrence after surgery, with higher sensitivity and specificity than AFP. Clinical implementation of serum GP73 measurement as a standard test for HCC is recommended.
- Liver cancer
- tumour marker
- liver resection
- hepatitis B virus
- liver cirrhosis
Statistics from Altmetric.com
Significance of this study
What is already known?
AFP has been the only standard serum marker for the detection of HCC for the last 40 years, even though its sensitivity of 39–65% is unsatisfactory.
Golgi protein 73 (GP73) is a potential serum marker for HCC but has not been validated in large cohort studies.
Whether GP73 is a better serum biomarker than AFP is controversial.
What are the new findings?
Combined measurement of GP73 and AFP increased the sensitivity for HCC to 89.2% (95% CI 86.7% to 91.5%), with a specificity of 85.2% (95% CI 83.4% to 86.4%).
GP73 decreased following surgical resection of HCC lesions and increased with tumour recurrence.
Non-liver cancers in 14 organs (colon, rectum, stomach, oesophagus, pancreas, lung, thyroid, kidney, bladder, prostate, breast, ovary, cervix and endometrium) were analysed; all the benign and other malignant liver lesions had moderate elevations of GP73, albeit at a much lower level than in HCC.
How might it impact on clinical practice in the foreseeable future?
Combined measurement of AFP and GP73 shows promise to improve the detection and treatment of HCC.
Hepatocellular carcinoma (HCC) is strongly associated with either chronic hepatitis B virus (HBV) or hepatitis C virus (HCV) infection,1–3 and is the fifth most common cancer and the third leading cause of cancer death worldwide.1–4 In the USA, the incidence of HCC is increasing faster than that of any other cancer.5 Notably, HCC is a leading cause of cancer death in China. In China, 7.18% of the entire population are HBV carriers (93 million, two-thirds of the world's total number of carriers); Chinese HCC patients account for >55% of new HCC cases worldwide (Global Cancer Facts & Figures 2007, American Cancer Society).4 6 A 10-year survey (1990–2001) conducted in China indicates that HCC ranks first among chronic diseases for the social cost and burden in the WHO ‘disability-adjusted life year (DALY)’ list.7 The 5-year survival rate of all HCC is <5%, placing it among the cancers with the worst prognosis.2 5 8 The prognosis can be significantly improved by early diagnosis and treatment.1
The use of serological markers in patients at the highest risk for developing HCC can thus decrease the cancer-related mortality and reduce medical costs. Alpha-fetoprotein (AFP) has been the only standard serum marker for the detection of HCC for the last 40 years, even though its sensitivity of 39–65% is not very satisfactory.4 5 9 Identification of better early diagnostic biomarkers is crucial.5 Recent studies have identified Golgi protein 73 (GP73; also named Golgi phosphoprotein 2 (GOLPH2)), as a potential novel HCC serum marker.10 11 GP73 is a 400 amino acid, 73 kDa transmembrane glycoprotein that normally resides within the cis-Golgi complex. Its mRNA was first identified in a search for upregulated hepatic genes in a patient with syncytial giant cell hepatitis.11 Although upregulated GP73 was initially identified in hepatic viral infections with unknown function, it can be excreted after modification.12–14 Subsequent studies showed that the GP73 serum level is elevated in diverse viral and non-viral liver diseases, including hepatitis, cirrhosis and HCC, and also in non-liver malignances.9–11 15–27 Of significance is that serum GP73 is dramatically elevated in patients with HCC, and the sensitivity and specificity of GP73 for HCC might be superior to those of AFP.9 11 19 20 23 24 However, the potential value of GP73 as a better serum biomarker than AFP is controversial.22 25 26 One possible explanation for continued controversy is the lack of larger observational studies. There is a consensus that comprehensive studies of large cohorts with a broad spectrum of disease are needed to validate GP73 as an HCC serum marker.9 24 25
In the present study, we compared GP73 and AFP in >4200 serum samples from healthy subjects and patients with various benign and malignant liver diseases, non-liver cancers, plus samples from patients with HCC collected before and after surgery, from multiple centres in China and the USA.
Materials and methods
In this multicentre study, a total of 4217 subjects were enrolled from November 2007 to May 2010, including 169 individuals at the University of Chicago (UOC) in the USA. The study protocol was approved by the Peking Union Medical College Hospital (PUMCH) Review Board and the UOC Institutional Review Board (IRB), and written informed consent was obtained from the subjects. The diagnosis of HCC was made based on guidelines from the Chinese Society of Hepatology and Chinese Society of Infectious Diseases, Chinese Medical Association.28 The subjects included healthy controls (n=1690), HBV carriers (n=337), patients with cirrhosis (n=512) including HBV positive (n=320), HCV positive (n=90) and cirrhosis of other origin (n=102), patients with HCC (n=789), and patients with other malignant liver lesions (n=61), benign liver lesions (n=206) and 14 different non-hepatic cancers (n=622). The age (mean ± SD) and gender (male:female) of each group are healthy control 49±15, 1072:618; HBV carriers 49±10, 209:128; cirrhosis 55±11, 345:167; HCC 56±11, 524:265; other malignant liver lesions 55±12, 39:22; benign liver lesions 47±8, 119:87; and non-liver cancers 50±12, 299:323. Preoperative and postoperative (postoperation day (POD) 3, 7 and 14) GP73 levels were assessed in 94 patients undergoing HCC resection, and in those with other liver-related malignant lesions (n=26) and benign liver lesions (n=126).
The serum samples of the subjects were obtained from multiple institutions including PUMCH, Cancer Institute and Hospital of PUMC, Chinese Academy of Medical Sciences, Qidong Liver Cancer Institute and the UOC. The healthy subjects with normal liver biochemistry, no history of liver disease or alcohol abuse, and no viral hepatitis were enrolled from the Peking Union Medical Health Physical Examination Center of PUMCH.
Those patients suspected of harbouring a HCC were enrolled in the study and a serum sample was obtained. For those patients in the various control groups, serum was obtained from IRB-approved serum banks or from patients in whom the various stages of liver disease had been established and HCC was ruled out with standard diagnostic imaging. Only the patients that were enrolled following resection were sampled prospectively.
The diagnosis of HCC was made either by histopathology or, if not available, by two imaging tests (ultrasound, CT, MRI or angiography) showing an arterial enhancing lesion with the significance of a HBV infection background in the Chinese population. HBV infection status was based on hepatitis B surface antigen (HBsAg). HCV infection status was based on serum HCV antibody and HCV RNA determination. Diagnosis of cirrhosis was based on liver histology or clinical, laboratory and imaging data. The symptoms and signs can include ascites, anorexia, portal hypertension, hypersplenism, oesophageal varices and encephalopathy; abnormal laboratory tests supporting the diagnosis are decreased serum albumin and clotting factors (with an elevated international normalised ratio), and elevation of serum liver enzymes. A liver biopsy may be obtained to confirm the diagnosis. All other liver-related lesions were confirmed by histopathological examinations. Fourteen different non-liver cancer samples were taken from preoperative patients who had no history of viral hepatitis, other liver diseases or liver metastasis.
GP73 and AFP analysis
Blood specimens were drawn prior to initiation of treatment for HCC. The samples were centrifuged at 6000 rpm for 5 min and the sera were stored at −80°C until testing. The normal and non-HCC samples were collected and stored in the same way as the HCC samples. AFP was tested using commercially available immunoassays utilising enhanced chemiluminescence at the PUMCH Clinical Diagnostic Laboratories.
GP73 was detected by immunoblotting17 19 and was performed at the PUMCH Liver Surgery Laboratory and the UOC Transplant Center Immunology Laboratory with the same methodology and under quality control by a well trained postdoctoral scientist and three technicians. Equal volumes of sera from the above patients (1 μl/lane) were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) on 4–12% Novex Bis-Tris precasting gels (Invitrogen, Carlsbad, California, USA). For normalisation, each gel also included a lane containing 1 μl of serum from a healthy subject. The proteins were transferred onto nitrocellulose membranes with a 7 min iBlot Dry Blotting System (Invitrogen). After blocking, the blots were then incubated with goat anti-GP73 polyclonal antibody (Santa Cruz Biotechnology, Santa Cruz, California, USA). The blots were washed and incubated with horseradish peroxidase-conjugated mouse antigoat secondary antibody. After washing, the blots were then developed using the ECL Plus chemiluminescent detection system (Thermo, Rockford, Illinois, USA). Densitometric analysis of the immunoblots was performed to quantify the amounts of GP73 protein. GP73-specific signals were quantified from x-ray films using a scanner with BandScan 4.30 densitometry software, and expressed as integrated intensity units relative to the GP73 signal detected in a control serum standard. The results were analysed by physicians in a blinded manner.
All values were reported as median (range, 25th and 75th percentiles) unless otherwise stated. Group differences in GP73 values were tested using the Mann–Whitney test. To determine the optimal cut-off value for GP73 in the diagnosis of HCC, the receiver operating characteristic (ROC) curves were constructed using all possible cut-offs for each assay. The areas under the ROC (AUROC) curves were constructed, and comparisons were tested using the U test. A two-tailed p value of <0.05 was considered to be of statistical significance. All analyses were performed using SPSS 12.0 (SPSS, Cary, North Carolina, USA).
Serum GP73 is significantly increased in patients with HCC
There were 1690 healthy subjects, comprising 618 females and 1072 males with an average age of 49±15 years. The median GP73 value in the healthy subjects was 1.2 (0.9–1.7) relative units (RU) (figure 1A). In contrast, the median value of GP73 in 789 patients with HCC was 14.7 (8.4–29.4) RU which is significantly higher than that of healthy subjects and patients with cirrhosis (4.7 (2.6–8.0) RU) (p<0.001). The GP73 value of healthy subjects has no correlation with age and gender (data not shown).
GP73 has higher sensitivity and specificity than AFP in diagnosis of HCC
A ROC curve was plotted to define the optimal cut-off values, and to identify the sensitivity and specificity of serum GP73 and AFP levels in differentiating patients with HCC versus all other conditions. The AUROC for GP73 was 0.94 (95% CI 0.93 to 0.96), with a sensitivity of 74.6% (95% CI 71.5% to 77.6%) and specificity of 97.4% (95% CI 96.8% to 98.3%), and an optimal cut-off value of 8.5 RU; the AUROC for AFP was 0.79 (95% CI 0.76 to 0.81), with a sensitivity of 58.2% (95% CI 55.2% to 62.1%) and specificity of 85.3% (95% CI 83.4% to 88.1%), and a cut-off of 35 ng/ml (figure 1B). GP73 had a better AUROC compared with AFP (p<0.001), indicating both a higher sensitivity and specificity of GP73 compared with AFP in the diagnosis of HCC (figure 1B). Using the optimal cut-offs derived from the ROC curve, GP73 was positive in 218 of 326 (67%) patients with an AFP <35 ng/ml. In contrast, AFP was positive in 92 of 200 (46%) patients with a GP73 <8.5 RU. Combined measurement of GP73 and AFP increased the sensitivity for HCC to 89.2% (95% CI 86.7% to 91.5%), with a specificity of 85.2% (95% CI 83.4% to 86.4%), and the AUROC was 0.96 (95% CI 0.94 to 0.97) (figure 1B). The positive likelihood ratio and negative likelihood ratio of GP73 and AFP in the diagnosis of HCC are 13.41 and 0.27, and 6.20 and 0.46, respectively. The positive and negative predictive values of GP73 and AFP in the diagnosis of HCC are 82.7% and 91.2%, and 68.8% and 85.9%, respectively.
Elevated GP73 in patients with hepatitis B virus-related liver diseases
GP73 values for healthy controls, HBV carriers, patients with cirrhosis and patients with HCC are shown in figure 2. The values in patients with HCC and in healthy controls, HBV carriers or HBV carriers with cirrhosis were noted to be significantly different, p<0.001 (higher in the HCC group).
Variations of GP73 values in patients with cirrhosis with different aetiologies in different races
There were no differences in GP73 values between patients with HBV-cirrhosis (4.2 (2.4–6.6) RU) and HCV-cirrhosis (3.7 (2.2–6.4) RU). GP73 values of cirrhosis with other causes (alcoholic, non-alcoholic steatohepatitis, biliary-origin cirrhosis, cryptogenic and other) (5.2 (2.5–7.2) RU) are slightly higher, but not statistically, than those of either HBV-cirrhosis or HCV-cirrhosis (figure 2B). There was no difference in Caucasian and African-American patients in the western subgroup, although the sample sizes for these subgroups are not large enough at present to make definitive comments.
GP73 levels and HCC tumour characteristics
The relationships between GP73 levels and tumour size or differentiation status or liver function were investigated. For comparison, AFP values of different tumour size or differentiation status or liver function were also analysed. GP73 values were not correlated with tumour size, histological differentiation or Child–Pugh class (table 1). There are no correlation between the levels of GP73 and albumin (data not shown). In contrast, the AFP value of patients with small HCCs (≤3 cm) was significantly less than that of other HCCs (≥5 cm, >3 and <5 cm, and diffuse HCC) (p<0.001). GP73 was positive (≥8.5 RU) in 30 of 34 (88.4%) AFP-negative (<35 ng/ml) patients with small HCCs, while AFP was positive in 3 of 7 (42.9%) GP73-negative patients with small HCCs.
GP73 is moderately elevated in various non-HCC hepatic tumours and extrahepatic malignancies
Moderate elevations were seen in other lesions related to the liver such as metastatic liver cancers (3.2 (1.5–7.9) RU, n=24), intrahepatic cholangiocellular carcinoma (5.0 (2.1 to 9.5) RU, n=14), cholangiocarcinoma (7.4 (3.4–12.1) RU, n=17) and gallbladder cancer (4.1 (2.4 to 8.0) RU, n=6), benign hepatic lesions including haemangioma (2.0 (1.2 to 4.1) RU, n=95) and others (including hepatic cyst, focal nodular hyperplasia and hepatic cystadenoma) (3.0 (1.4 to 8.3) RU, n=111); all of these levels were significantly lower than those found in HCC (figure 3A). Fourteen different types of cancers unrelated to the liver, including colon cancer (4.6 (3.4–6.9) RU, n=53), cancer of the rectum (3.5 (2.5–5.7) RU, n=55), stomach cancer (4.5 (2.4–6.3) RU, n=59), oesophageal cancer (3.9 (2.2–5.5) RU, n=33), pancreatic cancer (3.5 (2.6–6.2) RU, n=32), lung cancer (4.3 (2.6–6.6) RU, n=85), thyroid cancer (3.6 (1.9–5.7) RU, n=47), renal cancer (2.2 (1.2–3.9) RU, n=30), bladder cancer (2.2 (1.4–3.9) RU, n=34), prostate cancer (2.8 (1.6–4.0) RU, n=42), breast cancer (2.2 (1.3-4.1) RU, n=35), ovarian cancer (2.9 (1.9–7.4) RU, n=45), cervical cancer (2.6 (1.7–4.1) RU, n=32) and endometrial cancer (3.1 (2.3–4.4) RU, n=40), also had elevated serum GP73, albeit at a much lower level than those found in HCC (figure 3B). While the size of any individual cancer group is small and thus definitive statements cannot be made at this time, none of the values with the 95% CI approached the levels seen in HCC. The level of GP73 was statistically greater in HCC than in all the other tumours noted above.
GP73 values following hepatectomy and HCC recurrence
We monitored serum GP73 values in patients with HCC, benign liver lesions and other malignant liver-related lesions before and after hepatectomy from November 2007 to July 2009. GP73 levels in patients with HCC decreased following surgical resection of tumours; the preresection values, and POD 3, POD 7 and POD 14 were 14.0 (3.9–17.9), 10.1 (3.5–15.1), 5.3 (1.9–9.2) and 3.3 (2.1–5.4) RU (p<0.001 at POD 14), respectively (figure 4A). In contrast, GP73 values in patients with either benign or other malignant liver tumours were much lower than those in patients with HCC before surgery and did not change significantly after hepatectomy (figure 4A). In the last 21 months, 94 patients were followed-up postoperatively with routine test of GP73 as well as other imaging examinations. Of the 30 patients with tumour recurrence, currently (follow-up ranging from 3 months to 18 months after surgery) 22 were observed to have increased GP73 levels (12.1 (9.6–14.8) RU), which were significantly higher (p<0.001) those at POD 14 (3.3 (2.1–5.4) RU) (figure 4B).
Since HCC is among the cancers with the worst prognosis, early diagnosis and treatment are the keys for effective treatment of patients with HCC. The use of serological markers in patients at the highest risk for developing HCC may thus decrease HCC mortality and reduce medical costs. However, AFP has been the only standard serum marker for the detection of HCC for >40 years, even though its sensitivity of 39–65% is not satisfactory. Although recent studies have identified serum GP73 as a potential biomarker for HCC,11 19 the small sample sizes in those studies precluded the establishment of this putative tumour marker as a standard clinical test for HCC. Furthermore, whether GP73 is a better serum biomarker than AFP is controversial and its significance in the assessment of tumour recurrence after HCC surgery remains unknown. In this study, we have demonstrated in the largest cohort of patients reported to date from several premier medical centres in China and the USA that GP73 is a valuable tumour marker for HCC, with sensitivity and specificity as well as likelihood ratio and predictive values superior to those of AFP, independent of the age and gender of the subjects. The combined measurement of GP73 and AFP can further increase the sensitivity for the detection of HCC.
HBV is the primary cause of HCC, therefore it is important to demonstrate that the serum levels of GP73 identified in patients with various stages of hepatitis B are quantitatively different and statistically distinct from those levels identified in patients with HCC. We have found that the elevation of serum GP73 is modest in virus carriers, moderate in patients with cirrhosis and dramatic in patients with HCC. Therefore, serum GP73 can be used to monitor disease progression from HBV infection to cirrhosis to HCC. Non-HBV-, non-HCV-related cirrhosis is associated with slightly higher GP73 levels than HBV-cirrhosis and HCV-cirrhosis. The underlying cause of elevation in these diseases and why the levels are different in different diseases in the same clinical disease state is subject to further investigation. Even though this study is not a prospective study, given that HCC almost always arises in patients with cirrhosis, the samples from patients with chronic liver disease are representative of those patients in whom there is a concern that HCC will develop. Moreover, the GP73 values of Asian patients with cirrhosis appear to be lower than those of western patients. This observation has to be confirmed in further studies since the sample size of western patients is small compared with the number of Asian patients in this study.
Our study is the first published study that demonstrates that surgical resection of the tumour results in diminished serum GP73 levels and that tumour recurrence correlates with the recurrence of elevated GP73 in the blood. Reappearance of serum GP73 indicates the existence of tumour lesions and thus may serve as an indicator for the recurrence of HCC. Similarly, we also detected dynamic changes in AFP throughout the course of tumour removal and recurrence. The role of intrahepatic metastasis of the original tumour versus the development of de novo tumours, which also expresses GP73, is not able to be determined in this study. Unexpectedly, serum levels of GP73 in patients with HCC were not consistently affected by the tumour sizes and the status of tumour differentiation. We speculate that this may reflect the potential origin of HCC from cancer stem cells. This feature may nevertheless be beneficial for the earlier detection of HCC among the population at risk if this observation is found in further investigations with larger numbers of patients with HCC. Other than documenting the recurrence of disease following resection, we did not perform serial examinations of GP73 and thus cannot at the current time determine if GP73 is valuable in assessing progression (increased size, greater number or distant disease) of HCC.
We have found that both liver benign tumours and non-HCC liver malignant lesions had elevated serum GP73, although the magnitude is much smaller than that in HCC. Serum GP73 can therefore be a useful tool in determining the nature (benign vs HCC) of hepatic tumours. Furthermore, 14 common non-liver cancers also had moderate elevation of serum GP73, none of which, however, reached the level identified for HCC cases. Serum levels of GP73 diagnostic for HCC thus seemed not to be a pan-cancer marker.
In summary, GP73 is a valuable serum marker that can aid in the diagnosis of HCC, and may also be used in the surveillance of HCC recurrence in postoperative management. In combination, measurement of AFP and GP73 has the promise to further improve the detection and treatment of one of the most common malignancies worldwide. Therefore, we recommend the installation of serum GP73 measurement as a standard blood test for HCC. High-throughput quantitative serum assays for GP73 should be developed for clinical application and the surveillance of the population at risk.
We thank Ms Jinchun Zhang of PUMC for her generous participation.
Funding This work was supported by the China Medical Board in New York (CMB) (06-837), the National Natural Science Foundation of China (30788004) and the National Basic Research Program of China (973 Program) (grants 2009CB522203).
Competing interests None.
Patient consent Obtained.
Ethics approval This study was conducted with the approval of the Peking Union Medical College Hospital Review Board and the UOC Institutional Review Board.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.