Evaluating clinical and prognostic implications of Glypican-3 in hepatocellular carcinoma

Hepatocellular carcinoma (HCC) is one of the most deadly cancers worldwide. In patients with HCC, histopathogical differentiation is an important indicator of prognosis; however, because determination of HCC differentiation is difficult, the recently described immunohistochemical (IHC) marker glypican3 (GPC3) might assist in HCC prognostication.The goal of our study was to investigate GPC3's IHC staining pattern and define the relationship between its expression and patients' clinicopathologic features and overall survival. We retrieved clinical parameters from 101 pathologically diagnosed HCC patients' medical records and classified these patients into 4 clinical score categories (0–3) based on increasing GPC3 staining intensity and the percentage of stained tumor cells in their resection and biopsy specimens. Histopathological samples were well, moderately, and poorly differentiated in 33, 22, and 12 patients, respectively, and the GPC3 expression rate was 63%, 86%, and 92%,respectively. The median overall survival was 49.9 months (confidence interval (CI): 35.3–64.6 months) for clinical scores 0–1 and 30.7 months (CI: 19.4–41.9 months) for clinical scores 2–3. This difference was not statistically significant (P = .06) but showed a strong trend. In conclusion, a greater GPC3 expression is associated with a worse HCC prognosis and may be a promising prognostic marker.


INTRODUCTION
Hepatocellular carcinoma (HCC) is the second most common cause of cancer-related mortality worldwide, and it is the fifth most common cancer in men (554,000 cases/year, 7.5% of total cancer cases) and the ninth in women (228,000 cases/year, 3.4% of total cancer cases). Each year, 745,000 people worldwide die of the disease [1]. Cirrhosis is the greatest single risk factor for HCC, and other risk factors include chronic hepatitis B or C infection, exposure to aflatoxin, alcohol abuse, fatty liver disease, and smoking. HCC is commonly diagnosed at advanced stages, by which time it is usually incurable, unless amenable for surgical intervention [2,3]. Patient prognosis is mainly dependent upon the size and number of tumor nodules, the presence or absence of portal venous invasion, and histopathological differentiation. However, in some cases, the distinction of tumor differentiation is challenging based on histologic grounds alone. Therefore, immunohistochemical (IHC) markers have been studied for prognostication in HCC.
GPC3 is an oncofetal protein that is expressed in the placenta and fetal liver, but not in normal hepatic parenchyma or nonmalignant liver tissue, and it is only occasionally and weakly expressed in preneoplastic lesions. However, many studies have shown significantly increased GPC3 expression in HCC [5,19,20]. The goal of the current study was to determine GPC3's staining pattern in HCC and to define the diagnostic utility of GPC3 in distinguishing early from advanced HCC. We also investigated the potential prognostic value of GPC3 by analyzing the survival rates of patients with low versus high GPC3 expression in HCC tumors and determining whether GPC3 expression was associated with the patients clinicopathologic parameters.

RESULTS
The detailed baseline demographic characteristics of the 101 HCC patients in our study are summarized in Table 1. The majority of patients (62.4%) were older than 60 years, with a mean age and standard deviation of 63.2 ± 11.8 years and a male-to-female ratio of 1.5:1. Risk factors for HCC were hepatitis (35.6%), alcohol consumption (64.4%). Twenty-two patients (21.8%) had extrahepatic disease (either lymph node involvement or distant metastasis). At the time of initial diagnosis, 62.4% of patients had an Eastern Cooperative Oncology Group performance status of zero, and the majority of patients had early-stage HCC according to different prognostic scoring and staging systems. Comparison between low clinical score 0-1 (N = 52) and high clinical score 2-3 (N = 49) GPC3 expression showed that there was no statistically significant difference between the two levels based on demographic characteristics, epidemiological parameters, HCC risk factors, clinicopathological characteristics, and baseline treatment modalities ( Figure 1).
The Cox proportional hazard models showed that the GPC3 clinical score tended to be a significant independent risk factor for HCC OS. Compared to patients with a GPC3 clinical score of 0-1, the adjusted HR for patients with a GPC3 clinical score of 2-3 was about 1.5 times higher (adjusted hazard ratio 1.57; 95% CI, 1.007-2.47; P = .047, Table 2).
Futhermore, when the OS of patients who had a clinical score of 0-1 (49.9 months [95% CI: 35.3-64.6 months]) was compared with that of those with a clinical score of 2-3 (30.7 months [95% CI: 19.4-41.9 months]), the difference was not statistically significant (P = 0.06)). However, comparing between varaiotion of the medican OS among patients with low and high clinical score of GPC3 based on the baseline treatment approaches showed that patients with clinical score 0-1 and treated with surgery have a significantly longer OS compared to patients with higher clinical score 2-3 (P = .02) ( Figure 2) Among the 26 patients with evaluable paired biopsy and resection samples, 15 had the same clinical score for both, even though most patients had a time gap of more than two months between the the acquision of the two samples. Six patients had a higher clinical score for the biopsy than the resection, and five patients had a higher clinical score for the resection than the biopsy ( Figure 3). The concordance of clinical scores 0 vs. 1 vs. 2 vs. 3 between biopsies and resections was 57.7% www.impactjournals.com/oncotarget

DISCUSSION
Our current study indicates that higher GPC3 expression level in HCC is a risk factor for shorter OS. GPC3 expression may also be associated with histopathologic differentiation grade and advanced clinicopathologic features, but these findings were not significant. Consistent with our findings, Yamauchi et al. found that levels of GPC3 expression in poorly differentiated tumor cells were higher than those in moderately and well differentiated tumor cells. GPC3 was expressed in 78% (14 of 18) of well-differentiated tumors, 83% (24 of 29) of moderately-differentiated tumors, and 100% (9 of 9) of poorly differentiated tumors, while the size of the HCC was not related to the level of GPC3 expression. In the same study, GPC3 expression was negative in sarcomatoid HCC, carcinoid tumors, and cholangiocarcinomas [43]. The reported sensitivity of GPC3 for HCC in the literature ranges from 75% to 100%, and in large-scale trials it ranges from 75% to 85% [26,43,47,[55][56][57].
Our study showed a large difference in OS between patients with a GPC3 clinical score of 0-1 and patients with GPC3 clinical score of 2-3. However, it wasn't statistically significant there is strong trend toward better OS. The median OS was 49.9 months (95% CI: 35.3-64.6 months) for GPC3 clinical score 0-1 and 30.7 months (95% CI: 19.4-41.9 months) for GPC3 clinical score 2-3 (P = 0.06) ( Figure 3). Furthermore, we found a correlation between GPC3 expression rate and vascular invasion, >50% tumor involvement in the liver, lymph nodes involvement, alcohol use, cigarette smoking, race, gender, age, or stage (CLIP,TNM,BCLC and CTP). However, due to small sample size in our single institution study, this limitation may account for the lack of statistical significance in some of the correlation analyzed .   Additionally, we compared the expression of GPC3 in 26 patients who had both biopsy and resection, the majority of samples were concordant even though most of them collected after a time gap of over 2 months between the procedures. Based on this result , our study indicated that biopsies may provide reliable information about GPC3 expression. However, due to the limitations of our study, including a small number of samples, a time gap between biopsy and resection, and the effects of therapy on tissue nature, this result needs to be confirmed by future well-designed trials.
GPC3 expression and staining is diagnostically important for fibrolamellar HCC, which differs clinically, histologically, molecularly, and prognostically from conventional HCC and which typically occurs in young patients without cirrhosis. Elevation of AFP is uncommon and immunohistochemistry for AFP is generally negative in fibrolamellar HCC [58][59][60][61][62]. In contrast, a few trials have shown that the disease expresses and stains with GPC3, although, unlike in conventional HCC and especially poorly differentiated HCC the expression rate is not significant [5,45].
Some studies have indicated that serum GPC3, in combination with AFP, improves diagnostic accuracy and sensitivity for early HCC [63][64][65]. AFP is the most commonly used serum marker for the diagnosis and detection of HCC. At a cutoff value of 20 ng/mL of serum, AFP shows a 60%-80% sensitivity in detecting tumors [66][67][68][69] that decreases to about 40% for the detection of tumors that are smaller than 3 cm [68] .Interestingly, GPC3 level is more frequently elevated than AFP level (88% versus 55%) in patients with liver cancer, and especially in those with HCC tumors < 3 cm (77% versus 43%) [70]. Thus, after independent validation, GPC3 immunoassays may be useful in diagnosing HCC, as GPC3 has been shown to have serological sensitivity and specificity of 53% and 95%, respectively [64].
Our study results suggest that the GPC3 expression rate could be a promising prognostic marker for HCC. However, large-scale, independent validation studies are warranted to confirm and further define the prognostic role and implications of GPC3 in this disease.

Study design and population
Our current investigation is part of an ongoing, hospital-based, case-control study that was approved by The University of Texas MD Anderson Cancer Center Institutional Review Board. Written informed consent was obtained from each study participant.
We searched our patients database and identified 101 patients with histologically confirmed HCC treated from March 1996 to September 2012. Seven patients hadbiopsy specimens only, 31 patients had resection specimens only, and 26 patients had both resection and biopsy specimens.The following clinical variables were recorded at the time of diagnosis and retrieved from the patients' medical records: patient demographics, HCC risk factors (including cirrhosis), tumor characteristics (such as histologic differentiation, vascular invasion, extrahepatic metastasis, size and number of tumor nodules, and the percentage of the liver occupied by tumor), HCC treatment regimens and modalities, and survival. We also collected information about disease stage using various HCC staging systems, including Barcelona Clinic Liver Cancer (BCLC), Cancer of the Liver Italian Program (CLIP); tumor-node-metastasis (TNM), and OKUDA.
We retrieved the patients' paraffin-embedded tissue specimens Unstained sections, and IHC was performed at Ventana Medical Systems, Inc. (VMSI, Tucson, AZ). IHC procedures for GPC3, including antigen recovery, antibody incubation, and antibody detection, were done in the college of American pathologist (CAP) and Cliical Laboratory Improvement Act (CLIA) accredited certified laboratory. Tissue sections were stained using anti-Glypican 3 mouse monoclonal primary antibody (clone GC33, Ventana Medical Systems, Inc. Tucson, AZ) on BenchMark ULTRA to detect membrane and cytoplasmic expression. Heat-induced epitope retrieval was used followed by incubation of the primary antibody for 32 min at 36°C. Immunodetection was accomplished with ultraView Universal DAB Detection Kit. Appropriate positive and negative controls were included for each stain.
Based on GPC3 staining intensity and the percentage of stained tumor cells, patients were classified into one of four clinical score categories (Figure 4) according to the criteria described in Table 3. During scoring, pathologists were blinded to specimen and clinical details.

Statistical methods
Stata software (Stata Corp, College Station, TX) was used for statistical analysis. Univariate analysis was done using the X 2 or Fisher exact test for categorical variables, and the Kruskal-Wallis test was used for continuous variables for all 101 patients and for comparison between low GPC3 expression (clinical score 0-1) and high GPC3 expression (clinical score 2-3). To identify independent prognostic factors for overall survival (OS), adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) were calculated using Cox proportional hazard models after adjusting for confounding factors such as age, sex, race, vascular invasion, lymph node involvement, distant metastasis, volume percentage of liver occupied by tumor, BCLC staging system, and HCC treatment modalities. The clinical score HR was adjusted for age, sex, race, vascular invasion, distant metastasis, lymph node involvement, the percentage of the liver occupied by the tumor(s), the Barcelona Clinic Liver Cancer (BCLC) www.impactjournals.com/oncotarget Membranous staining of any intensity in < 10% of tumor cells, and/or Cytoplasmic staining of any intensity in > 10% of tumor cells (note that strong cytoplasmic staining, if present, must be in < 50% of tumor cells) 2 Weak to moderate membrane staining in ≥ 10% of tumor cells (note that strong membrane staining, if present, must be in < 10% of tumor cells), and/or Cytoplasmic staining of any intensity in > 10% of tumor cells (note that strong cytoplasmic staining, if present, must be in < 50% of tumor cells) 3 Strong membrane staining in > 10% of tumor cells with or without cytoplasmic staining, or Strong cytoplasmic staining in ≥ 50% of tumor cells