Fax: (011) 39 0815257447
Squamous cell carcinoma antigen-immunoglobulin M complexes as novel biomarkers for hepatocellular carcinoma
Article first published online: 10 MAY 2005
Copyright © 2005 American Cancer Society
Volume 103, Issue 12, pages 2558–2565, 15 June 2005
How to Cite
Beneduce, L., Castaldi, F., Marino, M., Quarta, S., Ruvoletto, M., Benvegnù, L., Calabrese, F., Gatta, A., Pontisso, P. and Fassina, G. (2005), Squamous cell carcinoma antigen-immunoglobulin M complexes as novel biomarkers for hepatocellular carcinoma. Cancer, 103: 2558–2565. doi: 10.1002/cncr.21106
- Issue published online: 2 JUN 2005
- Article first published online: 10 MAY 2005
- Manuscript Accepted: 9 FEB 2005
- Manuscript Revised: 22 NOV 2004
- Manuscript Received: 30 AUG 2004
- Ministero dell'Istruzione dell'Università e della Ricerca. Grant Number: 11467
- hepatocellular carcinoma;
- squamous cell carcinoma antigen;
- immunoglobulin M immune complexes;
Early detection of hepatocellular carcinoma (HCC), one of the most common and deadly tumors worldwide, still is difficult due to the lack of adequate biomarkers that show high sensitivity and specificity. The authors recently demonstrated that squamous cell carcinoma antigen (SCCA) variants were overexpressed remarkably in all surgically resected HCCs.
For the current study, the authors assessed the presence of SCCA, as a free form and complexed with immunoglobulins, in serum from patients with HCC, cirrhosis, and chronic hepatitis and from healthy control participants and compared SCCA measurement with the measurement of α-fetoprotein (AFP) levels.
Circulating immune complexes (ICs) composed by SCCA and immunoglobulin M (IgM) IC (SCCA-IgM IC) were undetectable (< 120 arbitrary units [AU]/mL) in serum from a healthy control population (0 of 73 controls); however, 35 of 50 patients with HCC (70%) were reactive for SCCA-IgM IC independent of etiology (mean ± standard deviation [SD], 2568.5 ± 6797.3 AU/mL). No correlation was found with AFP levels, which were elevated significantly in only 21 of 50 patients with HCC (42%). By using an AFP cut-off value of 20 ng/mL, 96% of patients with HCC were positive for at least 1 marker. Among cirrhotic patients, the presence of circulating SCCA-IgM IC was displayed in 13 of 50 patients (26%), but at lower levels compared with the patients who had HCC (mean ± SD, 147.5 ± 348.3 AU/mL; P < 0.01; Student t test), whereas 9 of 50 patients with chronic hepatitis (18%) were reactive (mean ± SD, 39.5 ± 89.7 AU/mL). No significant presence of free SCCA, free anti-SCCA variants IgG or IgM, or SCCA-IgG IC was found.
The study results indicated that SCCA-IgM ICs represent novel serologic biomarkers, which, alone or in combination with AFP, can increase the sensitivity for diagnosing HCC significantly. Cancer 2005. © 2005 American Cancer Society.
The increased diffusion of the etiologic agents responsible for the vast majority of hepatocellular carcinomas (HCCs), such as hepatitis B virus (HBV), hepatitis C virus (HCV), alcohol abuse, and the poor prognosis for patients with primary HCC, require the development of adequate screening programs for patients at risk, including chronic carriers of HBV and HCV.1 The most widely used serologic marker to detect HCC is α-fetoprotein (AFP), which is elevated (> 20 ng/mL) in a wide number of patients with HCC (30–60%) but with low specificity (70–80%), because a considerable number of patients with chronic liver disease may have AFP levels in the 20–200 ng/mL range.2 In addition, AFP serum levels in patients with cirrhosis and in patients with HCC often overlap, and higher AFP cut-off values (> 100 ng/mL) have been used to increase specificity, but this reduced sensitivity to extremely low values (5–15%).3 Given the high heterogeneity of HCC,4 it has been found that other biomarkers are overexpressed in the liver and/or serum from patients.
Elevated levels of des-γ-carboxy prothrombin (DCP) have been found in 35–53% of patients with HCC5 as a result of an acquired defect in the posttranslational carboxylation of the prothrombin precursor in neoplastic cells.6 Although some investigators have reported the usefulness of this marker compared with AFP,7 others found no improvement over AFP determination, recommending a combination of both assays to improve sensitivity and specificity.8
Glypican-3 (GPC3) messenger RNA overexpression has been found in the liver in 75% of HCCs but in only 3.2% of normal livers.9 Immunohistochemistry results have confirmed the occurrence of GPC3 protein in 72% of patients10; and, using an enzyme-linked immunosorbent assay (ELISA), circulating GPC3 protein has been detected in 40–53% of patients with HCC.10, 11 Other markers proposed for HCC surveillance, including lens culinaris agglutinin reactive species of AFP termed AFP-L3, p53 autoantibodies, carbohydrate-deficient-transferrin, HBV-encoded X antigen, and α-L-fucosidase, lack adequate specificity to support a diagnostic value.12
Overexpression of squamous cell carcinoma antigen (SCCA) variants (SCCA-1, SCCA-2, and SCCA-PD) has been reported recently in all surgically resected HCC specimens but in none of the normal control normal livers, as detected by immunohistochemistry.13 SCCA is a serine protease inhibitor that is found physiologically in the spinous and granular layers of normal squamous epithelium, but it is expressed typically by neoplastic cells of epithelial origin.14 Both SCCA isoforms SCCA1 and SCCA2 protect neoplastic cells from apoptotic death induced by several kinds of stimuli, and in vivo experiments have demonstrated that SCCA1 can promote tumor growth.15
For the current study, we developed ELISA assays first to assess the presence of SCCA variants as free protein and/or as circulating immune complexes (ICs) in patient serum and, then, to determine the usefulness of circulating SCCA variants in terms of sensitivity and specificity for HCC detection compared with AFP determination.
MATERIALS AND METHODS
Serum samples from 160 patients with different spectrum of liver disease and from 73 healthy donors were analyzed. Patients with liver disease included 60 patients with HCC (mean age ± standard deviation [SD], 64 ± 14 years; male:female ratio, 2:1), and the etiology was HCV in 80% of patients, HBV in 18% of patients, and both HBV/HCV in 2% of patients. The diagnosis of HCC was based on the presence of hepatic focal lesion > 2 cm, detected by liver ultrasound (US) and confirmed by US-assisted fine-needle biopsy, computed tomography, and magnetic resonance imaging, when indicated. Blood samples from patients with HCC were taken prior to initiation of any HCC treatment. Samples from a group of 50 patients with cirrhosis (mean age ± SD, 51 ± 9 years; male:female ratio, 2:1) also were included: 86% of those patients were positive for HCV, and 14% were positive for HBV. All patients in this group underwent regular liver US screening to exclude the occurrence of hepatic nodules. The third group included 50 patients with histologically confirmed chronic HCV (mean age ± SD, 44 ± 12 years; mean ± SD alanine aminotransferase level, 96.19 ± 81.06 U/L; male:female ratio, 1:1).
Serum levels of anti-SCCA immunoglobulin G (IgG) and IgM were measured using two distinct enzyme immunoassay. Briefly, 96-well ELISA plates were coated with 0.5 μg of recombinant SCCA in 100 μL bicarbonate buffer, pH 9.6, at 4 °C overnight and then blocked for 2 hours with 1% bovine serum albumin (BSA) in phosphate buffered saline, pH 7.2 (PBS). One hundred microliters of serially diluted samples in PBS containing 1% BSA and 0.05% Tween 20 (T) were incubated for 1 hour at room temperature. The presence of anti-SCCA IgG and IgM was detected using peroxidase conjugated goat antihuman IgG (Sigma Aldrich, Italy) and peroxidase-conjugated goat antihuman IgM (Sigma Aldrich), respectively, with 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS) (Sigma Aldrich) and hydrogen peroxide as substrate.
SCCA and IC assays
Ninety-six-well ELISA plates were coated with 1 μg of rabbit anti-human SCCA antibody (Xeptagen, Italy) in 100 μL PBS, pH 7.2, at 4 °C overnight and then blocked for 1 hour with 1% BSA in PBS. After blocking, 100 μL of serially diluted samples in PBS containing 1% BSA and 0.05% T were incubated for 1 hour at room temperature. SCCA was detected using biotinylated rabbit antihuman SCCA antibody (Xeptagen) and horseradish peroxidase-conjugated avidin. ABTS (Sigma Aldrich) and hydrogen peroxide were used as substrate. To quantify the SCCA concentration in serum, a calibration curve of purified SCCA in PBS containing 1% BSA and 0.05% T was performed in parallel (detection limit, 0.2 ng/mL). ELISA assays to detect SCCA ICs were performed using similar coating and blocking steps. Serially diluted standards or serum samples in PBS containing 1% BSA and 0.05% T were incubated for 1 hour at room temperature. The SCCA-IgM IC and the SCCA-IgG IC were revealed using peroxidase conjugated antihuman IgG and peroxidase-conjugated antihuman IgM, respectively, and were developed with ABTS and hydrogen peroxide as substrate. For the determination of SCCA-IgM IC, optimal conditions were standardized by using SCCA-IgM IC purified by gel filtration, and the amount of SCCA-IgM IC was expressed in arbitrary units (AU) using a calibration curve of the reference standard purified by gel filtration. The assay, which is called Hepa-IC, displayed intraassay and interassay coefficients of variation of 5.5% and 6.2%, respectively. For quantification of SCCA-IgM IC levels, samples with values > 250 AU/mL (matching the upper limit of the standard curve) were diluted further and remeasured.
Serum AFP levels were determined in parallel on each sample using Beckman Coulter Access reagents for AFP on Access® I analyzer (Beckman Coulter, Fullerton, CA).
One hundred microliters of pooled HCC sera were analyzed using a gel filtration column BioSep SEC S-4000 (Phenomenex, Macclesfield Cheshire, UK) on a SpectraSystem high-performance liquid chromatograph (Thermo Finnegan, Los Angeles, CA). The elution was carried out in PBS at a flow rate of 0.5 mL per minute, and sample absorbance was monitored at 280 nm. Fractions were collected every 30 seconds, and immunoreactivity was tested by ELISA, as described above. Before running the samples, a calibration run was carried out according to the manufacturer's instructions.
Gel-filtration fractions containing SCCA-IgM IC and SCCA, as assessed by ELISA, were pooled; concentrated by precipitation with polyethylene glycol 6000 or 50% trichloroacetic acid, respectively; then resuspended in reducing Laemmli buffer. The protein samples and 100 ng of human IgM (Sigma Aldrich) were separated by 12% sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred to a nitrocellulose membrane (BioRad, Palm Springs, CA). The membrane was blocked in 5% BSA in PBS containing 0.05% T for 1 hour at room temperature. Blots were then incubated with peroxidase-conjugated goat antihuman IgM (Sigma Aldrich) at 1:1000 dilution or, alternatively, with 10 μg/mL rabbit antihuman SCCA antibody (Hepa-Ab; Xeptagen) and peroxidase-conjugated goat antirabbit IgG (Sigma Aldrich) at 1:5000 dilution. Protein bands were revealed utilizing enhanced chemiluminescence (ECL) reagents (WestPico ECL kit; Pierce, IL) and a ChemiDoc instrumentation (BioRad).
Liver specimens, which were obtained by percutaneous liver biopsy for diagnostic purposes, were available from 22 patients with chronic hepatitis, 46 patients with cirrhosis, and in 40 patients with HCC. Control liver biopsies were obtained from 20 patients who underwent cholecystectomy or liver biopsy for the staging of mediastinal or laterocervical Hodgkin disease. Formalin fixed and paraffin embedded sections were analyzed for the presence of SCCA variants, as described previously,13 by using an oligoclonal rabbit antibody raised against SCCA variants (Hepa-Ab; Xeptagen). The percentage of stained cells in each specimen was scored on a scale from 0 to 3, in which 0 denoted negative staining, a score of 1 indicated 1–30% positive hepatocytes, a score of 2 indicated 31–50% positive hepatocytes, and a score of 3 indicated > 50% positive hepatocytes. In all specimens, SCCA semiquantitative immunoreactivity was evaluated independently by two pathologists who were experts in the field. Intraobserver and interobserver differences were < 5%, and specimens with discordant results were reevaluated simultaneously by 2 observers.
The SCCA-IgM IC and AFP values were reported as means ± SD and medians with confidence intervals around the median of 95%. Statistically significant differences between the groups were determined by using the Student t tests and the Mann–Whitney U test. All analyses were performed using Analyse-It® Software (Analyse-It Software, Leeds, UK).
To characterize circulating SCCA variants, serum samples from 10 HCC patients were grouped and subjected to gel-filtration analysis. Fractions collected from the column were tested for the presence of free SCCA, SCCA-IgG IC, SCCA-IgM IC, IgG anti-SCCA, and IgM anti-SCCA by using the dedicated ELISA assays (see Materials and Methods, above). Although no significant presence of either free anti-SCCA variants IgG or IgM or SCCA-IgG IC was found, a strong reactivity was observed instead by using the SCCA-IgM complex ELISA assay (Hepa-IC) in the fraction eluting at high molecular weight (> 2,000,000 Da), as shown in Figure 1A. Peaks showing high reactivity by Hepa-IC were collected and subjected to reducing SDS-PAGE and Western blot analysis.
Figure 1B shows that the collected sample contained components migrating at the expected molecular weight for reduced IgM that were stained by antihuman IgM antibodies. On the same sample, a band corresponding to the expected molecular weight of SCCA (49 kDa) was stained by the anti-SCCA variant antibody (Hepa-Ab), thus confirming the presence of both SCCA and IgM in this high-molecular-weight fraction. Circulating uncomplexed SCCA, eluting by gel filtration after much higher volumes, also was detected by ELISA, and the corresponding fractions were stained by anti-SCCA antibodies on Western blots. Preliminary investigations by ELISA on serum samples from patients with HCC indicated that free SCCA was detectable with low sensitivity and low specificity compared with samples from the healthy control population (data not shown). The situation differed remarkably when SCCA-IgM ICs were evaluated. By expressing SCCA-IgM IC concentration in AUs, using a gel filtration-purified fraction of SCCA-IgM IC as a reference standard, the vast majority of HCC samples (35 of 50 samples; 70% of sensitivity) were strongly reactive (mean ± SD, 2568.5 ± 6797.3 AU/mL), whereas all 73 healthy control samples were negative (cut-off value, 120 AU/mL; specificity, 100%; positive predictive value [PPV], 100%; negative predictive value [NPV], 83%).
No correlation with HCC etiology was found. In cirrhotic patients, SCCA-IgM ICs were detectable in 13 of 50 serum samples (26%; mean ± SD, 147.5 ± 348.3 AU/mL), yielding a specificity of 74% (PPV, 73%; NPV, 71%). In patients with chronic hepatitis C, SCCS-IgM ICs were detectable in 9 of 50 serum samples (18%; mean ± SD, 39.5 ± 89.7 AU/mL) (Fig. 2B), yielding a specificity of 82% (PPV, 80%; NPV, 73%). The same samples were tested in parallel for AFP content by ELISA, but no correlation was found with AFP levels, which were elevated significantly (> 20 ng/mL) in only 21 of 50 patients with HCC (42%) (Fig. 2A).
Concentrations of circulating SCCA-IgM IC paralleled the extent of SCCA overexpression detected by immunohistochemistry in liver specimens. SCCA was detectable in 50% of liver samples from patients with chronic hepatitis, and 65% of their samples had a staining score of 1 (< 30% positive hepatocytes). The percentage of SCCA reactivity increased to 75% in patients with cirrhosis, and 50% of their liver specimens had a staining score of 2 (30–50% positive hepatocytes); whereas, in patients with HCC, 93% of liver samples were reactive, and 70% of their samples had a staining score of 3 (> 50% positive hepatocytes) (Fig. 3).
The detection of both AFP and SCCA-IgM IC significantly improved the ability to discriminate HCC from cirrhosis and chronic hepatitis. The diagnostic accuracy determined as the area under the receiver operating characteristic (ROC) curve was higher (0.741) for SCCA-IgM IC than the accuracy calculated for AFP (0.647) (Fig. 4A). A similar improvement was observed in differentiating HCC from chronic hepatitis, in which SCCA-IgM IC showed higher accuracy (0.797) than AFP (0.702). When using an AFP cut-off value of 20 ng/mL and an SCCA-IgM IC cut-off value of 120 AU/mL, 48 of 50 samples (96%) were positive for at least 1 marker without compromising the detection specificity of the control group (Fig. 4B).
One of the main features of a good marker for HCC is its ability to distinguish between HCC and cirrhotic liver. In our study, the distribution of serum levels of SCCA-IgM IC was significantly higher in the HCC group compared with the cirrhosis group (P < 0.0001; Mann–Whitney U test). Determination of the ROC curve indicated that, by using a SCCA-IgM IC value of 1600 AU/mL, HCC could be distinguished from cirrhosis with 100% specificity (Fig. 4A, top). Similar discrimination specificity was obtained by using an AFP value of 110 ng/mL, as determined by the ROC curve (Fig. 4A, top). Although, at these high cut-off values, HCC detection sensitivity was only slightly higher by measuring SCCA-IgM IC (20%) than by measuring AFP (18%), the combined detection of biomarkers increased the diagnostic accuracy for identifying hepatic carcinoma, reaching a sensitivity of 36% (Fig. 4B). ROC curves also were plotted to identify cut-off values that achieved discrimination between HCC and chronic hepatitis with 100% specificity (Fig. 4A, bottom). The cut-off levels for SCCA-IgM IC and AFP were 340 AU/mL and 40 ng/mL, respectively. SCCA-IgM IC levels were > 340 AU/mL in 22 of 50 patients with HCC (44%). AFP was elevated in only 15 of 50 patients with HCC (30%), and the combined detection of biomarkers led to better sensitivity (66%).
Detection of HCC at an early stage may reduce mortality significantly. This particular malignancy develops in > 90% of patients who are affected by cirrhosis or chronic hepatitis, and mass screening may be justified, because 1) the at-risk population can be identified easily, 2) tumor ablation or resection at an early stage can increase survival, and 3) HCC tends to grow slowly and to stay confined to the liver. However, massive screening should be justified only when sensitive and specific diagnostic procedures are available. Currently, US and AFP monitoring are the only reasonable screening strategy to detect HCC, but they are not effective enough to justify massive screening programs. In the current study, we have demonstrated that SCCA variants, which are overexpressed remarkably in liver tumor tissues, are detectable in serum from patients with HCC coupled to IgMs to form circulating ICs. Compared with other biomarkers, such as AFP, DCP and GPC3, which are detected in 6–40% of patients,3, 5–8, 10–12 SCCA-IgM IC levels measured by ELISA are elevated significantly in a greater proportion of patients (70%); whereas, in the healthy control population, the levels are below the limit of detection. The concentration of circulating SCCA-IgM IC at different stages of liver disease reflects the extent of SCCA overexpression detected by immunohistochemistry in liver specimens. SCCA-IgM IC does not overlap with AFP and, thus, offers the possibility to increase the sensitivity for detecting HCC remarkably without compromising specificity.
Preliminary results indicate that SCCA overexpression may be an early event of carcinogenesis in the liver, because immunohistochemical analysis in dysplastic nodules that are deemed HCC putative morphologic precursors indicates high SCCA expression, whereas benign hepatic nodules do not show SCCA expression.17 SCCA is a serine protease inhibitor that is found physiologically in the spinous and granular layers of normal squamous epithelium, but it is expressed typically by neoplastic cells of epithelial origin, and it has been used as serologic marker of cervical squamous cell carcinoma.18 Therefore, SCCA overexpression also may characterize other types of epithelial tumors. However, it should be noted that, in our investigation, we did not find significantly elevated levels of free SCCA in patients with HCC but, instead, found high levels of SCCA-IgM ICs.
Recent analyses of human systemic immune response against malignancy have led to the identification of a number of tumor-associated antigens, suggesting that at least some malignancies are immunogenic and that autoantibodies are elicited at the perturbed state of apoptosis.19 The occurrence of biomarkers associated with Igs to form circulating ICs has been reported for a few other biomarkers, such as carcinoembryonic antigen20 and TA9021 for colon carcinoma and MUC-122 or p5323 for breast carcinoma. IgG ICs were found mainly in these malignancies, but always accompanied by significant amounts of the free biomarker. In patients with HCC, free SCCA variants were not elevated significantly compared with healthy controls or compared with free or SCCA-complexed anti-SCCA IgG.
In addition, free anti-SCCA IgMs were not elevated significantly. It is possible to speculate that, in patients with HCC, initial low-dose exposure to SCCA variants may induce a weak IgM immune response without activation of IgG antibody-mediated immunity. To shed light on this aspect, studies are ongoing to evaluate the role of SCCA-IgM IC in tumor behavior, assessing whether or not the IgM autoantibody response is likely to persist after patients receive adequate tumor treatment.
The authors thank Dr. Massimo Gion for critical reading of the article.
- 1Epidemiology of hepatocellular carcinoma. Gastroenterol Clin North Am. 1987; 16: 545–551..
- 2Tumor markers. In: BurtisCA, AshwoodA, editors. Tietz textbook of clinical chemistry, 3rd ed. Philadelphia: Saunders, 1999: 722–749., .
- 6Measurement of immunoreactive prothrombin precursor and vitamin-K dependent gamma-carboxylation in human hepatocellular carcinoma tissues: decreased carboxylation of prothrombin precursor as a cause of des-gamma-carboxyprothrombin synthesis. Tumor Biol. 1990; 11: 319–326., , , , .
- 15Suppression of a squamous cell carcinoma (SCC)-related serpin, SCC Antigen, inhibits tumor growth with increased intratumoral infiltration of killer cells. Cancer Res. 2001; 61: 176–180., , , et al.