Serum sHLA-G levels: A useful indicator in distinguishing colorectal cancer from benign colorectal diseases

Authors

  • Cheng-Bao Zhu,

    1. Department of Clinical Laboratory, Jinan Infectious Disease Hospital, Jinan, Shandong, People's Republic of China
    Search for more papers by this author
  • Chuan-Xin Wang,

    Corresponding author
    1. Department of Clinical Laboratory, Qilu Hospital, Shandong University, Jinan, Shandong, People's Republic of China
    • Department of Clinical Laboratory, Qilu Hospital, Shandong University, 107 Wenhua West Road, Jinan, Shandong 250012, People's Republic of China
    Search for more papers by this author
    • Tel.: +86-531-82169341, Fax: +86-531-86927544

  • Xin Zhang,

    1. Department of Clinical Laboratory, Qilu Hospital, Shandong University, Jinan, Shandong, People's Republic of China
    Search for more papers by this author
  • Jian Zhang,

    1. Department of Clinical Laboratory, Qilu Hospital, Shandong University, Jinan, Shandong, People's Republic of China
    Search for more papers by this author
  • Wei Li

    1. Department of Clinical Laboratory, Qilu Hospital, Shandong University, Jinan, Shandong, People's Republic of China
    Search for more papers by this author

Abstract

Soluble human leukocyte antigen-G (sHLA-G) has been reported in malignancies and is implicated in mediating immune surveillance of tumor. The aim of our study is to detect serum sHLA-G levels in colorectal cancer and to determine whether sHLA-G may be helpful in distinguishing colorectal cancer from benign colorectal diseases. Serum sHLA-G levels were determined using enzyme-linked immunosorbent assay. Receiver operating characteristic (ROC) curve was used to evaluate the feasibility of sHLA-G in differentiating colorectal cancer from benign colorectal diseases. Median sHLA-G concentrations were significantly higher in colorectal cancer compared to normal colorectum, hyperplastic polyp, inflammatory bowel disease and adenoma (all at p < 0.001, respectively). ROC curve for sHLA-G revealed an area under the curve of 84.2%, and when 88.6 U/mL was used as cutoff, a sensitivity of 72.2% and a specificity of 87.8% were achieved. Comparison of sHLA-G and carcinoembryogenic antigen ROC curves indicated that sHLA-G was superior to CEA in differentiating colorectal cancer from benign colorectal diseases (p < 0.001). ROC curves analysis of the combined sHLA-G and CEA showed a higher detection capacity (area under the ROC curve, 87.4%) than that of markers considered singly. These findings reveal that serum levels of sHLA-G are significantly increased in colorectal cancer which may serve as a potent mediator of immune escape in colorectal cancer, and sHLA-G may be a useful indicator in differentiating colorectal cancer from benign colorectal diseases.

Colorectal cancer is one of the most common malignancies and remains a major cause of cancer-related death in human. With regard to screening tests for the early detection of colorectal cancer, colonoscopy test offers high diagnostic accuracy, but it can cause high costs, inconvenience and the additional complications. Fecal occult blood testing (FOBT) is the most widely used screening technique for colorectal cancer. However, this test is prone to produce false results.1 Thus, the identification of noninvasive molecular biomarkers for detection of colorectal cancer has been a long-standing challenge.

Human leukocyte antigen-G (HLA-G) is a nonclassical major histocompatibility complex (MHC) class I with limited tissue distribution, low polymorphism and seven isoforms (HLA-G1 to -G7). It is primarily expressed in the trophoblast from early gestation placentas, where it has been shown to play a key role in maternal–fetal tolerance.2 However, the aberrant HLA-G expression has been identified in various types of tumors, including melanoma,3 glioma,4 breast cancer,5 renal cancer,6 lung cancer,7 cutaneous lymphoma,8 chronic lymphocytic leukemia9–11 and acute myeloid leukemia.12 This ectopic HLA-G may contribute to the impairment of immune responses against tumor cells.13 Recently, soluble HLA-G (soluble HLA-G1 and HLA-G5) in body fluids has also been detected in some of these tumor types.14 sHLA-G can interact with the CD8+ coreceptor to induce CD95/CD95 ligand-mediated apoptosis on T and natural killer (NK) cells and thereby exert its host's immune suppression function.15 Interestingly, the detection of sHLA-G may have clinical values in differentiating malignant versus benign conditions.16 However, until now, little research has been conducted on the expression of sHLA-G in colorectal cancer. In this study, we used enzyme-linked immunosorbent assay (ELISA) to detect the serum sHLA-G levels in patients with colorectal cancer to further evaluate the clinical utility of sHLA-G in distinguishing colorectal cancer from benign colorectal diseases.

Abbreviations:

AUC: area under the curve; CEA: carcinoembryogenic antigen; CI: confidence interval; HLA-G: human leukocyte antigen G; IBD: inflammatory bowel disease; IQR: interquartile range; MAb: monoclonal antibody; NK cell: natural killer cell; sHLA-G: soluble human leukocyte antigen-G

Material and Methods

Patients and samples

Our study was approved and monitored by the Ethics Committee of Qilu Hospital, Shandong University. A total of 398 subjects, before they underwent colonoscopy or colorectal cancer surgery at the Endoscopy Department of Qilu Hospital, Shandong University, were enrolled in our study. Based on standard clinical, radiological endoscopic and histologic criteria, subjects were classified into five mutually exclusive categories: normal colorectum (n = 60), hyperplastic polyp (n = 72), inflammatory bowel disease (IBD; n = 57), adenoma (n = 65) and colorectal cancer (n = 144). The distribution of gender and age in each group is shown in Table 1. All subjects excluded the other diseases, for example, allergic asthma, transplant rejection and HIV-infection. Moreover, all patients did not receive any type of therapy, such as radiation or chemotherapy. Blood was taken from 398 subjects and kept at room temperature for 20–30 min, then centrifuged (4,000 rpm) for 5 min, aliquoted and stored at −80°C for less than 3 months.

Table 1. Characteristics of patients studied (n = 398)
inline image

ELISA for sHLA-G

sHLA-G was quantified using a commercially available ELISA kit (Exbio, Prague, Czech Republic) according to the manufacturer's instructions. Briefly, calibrators and samples were incubated in microtitration wells precoated with monoclonal antibody Mouse Anti-HLA-G Monoclonal Antibody (MEM-G/9), which recognized the most abundant soluble isoforms (shedded sHLA-G1 and intron4-containing secreted HLA-G5). After 60-min incubation and washing, monoclonal anti-human beta2-microglobulin antibody labeled with horseradish peroxidase (HRP) was added to the wells and incubated for 60 min, which recognized the immobilized antibody-sHLA-G complex. Following another washing step, the substrate (H2O2 with tetramethylbenzidine) solution was added to react with the remaining HRP-conjugated antibody. After the addition of acidic stop solution, the absorbance of resulting yellow product was then measured at 450 nm using a microplate reader (BIO-RAD Model 550, USA). According to absorbance values proportional to sHLA-G concentrations of calibrators, a calibrators curve was constituted, and the sHLA-G concentrations of unknown samples were determined by this calibrators curve. Serum samples from study populations were mixed and performed in batches at the same time. The intra-assay and interassay coefficient variations of the human sHLA-G ELISA kit were 3.2–6.9% and 7.7–13.9%, respectively, and the limit of detection was 3 U/mL.

Carcinoembryogenic antigen assay

The carcinoembryogenic antigen (CEA) assay was done on serum using the chemiluminescent microparticle immunoassay technology. This was achieved with the Abbott ARCHITECT i2000 SR system (Abbott Diagnostics, Abbott Park, IL) and commercially available CEA immunoassay kits (Abbott Laboratories, North Chicago, IL). The limit of normality was considered to be 5 ng/mL.

Statistical analysis

Kolmogorov-Smirnov test was performed to determine the distribution of the samples of each group. Data were expressed as median [interquartile range (IQR)]. A nonparametric test (Kruskal-Wallis test) was applied to analyze differences between colorectal cancer and other groups. A p-value of less than 0.05 was considered as statistically significant. The area under the empirical receiver operating characteristic (ROC) curve was used to assess the detection capability of serum markers for colorectal cancer. All statistical analyses were performed with SPSS software versions 13.0 for windows (SPSS, Chicago, IL).

Results

Measurement of sHLA-G in serum samples

sHLA-G was detectable in all samples analyzed. The serum sHLA-G levels are presented for each sample population (Table 1), and the individual values on a log scale are represented in dot-plot graphs (Fig. 1a). No correlation between serum sHLA-G concentrations and gender or age of the subjects tested could be established from this study (data not shown).

Figure 1.

sHLA-G levels in serum samples collected from 398 individuals. (a) The individual values on a log scale are represented in dot-plot graphs. Cutoff value for sHLA-G is 88.6 U/mL (represented by line across each graph). (b) Serum sHLA-G levels on a log scale in box-and-whisker plots. Colorectal cancer statistically different from normal colorectum, hyperplastic polyp, IBD and adenoma (all at p < 0.001, respectively); hyperplastic polyp, IBD and adenoma statistically different from normal colorectum (all at p < 0.01, respectively).

As shown in Figure 1b, median sHLA-G concentrations were significantly higher in colorectal cancer compared to normal colorectum, hyperplastic polyp, IBD and adenoma (all at p < 0.001, respectively). Serum sHLA-G levels were statistically and significantly higher in patients with colorectal hyperplastic polyp, IBD and adenoma than in normal colorectum (all at p < 0.01, respectively). There were no significant difference among hyperplastic polyp, IBD and adenoma (all at p > 0.05, respectively). This finding prompted us to assess whether sHLA-G could offer the potential in distinguishing colorectal cancer from benign colorectal diseases.

ROC curve analyses

ROC curves were used to evaluate the performance of sHLA-G in discriminating colorectal cancer from benign colorectal diseases. The ROC curves used colorectal cancers (n = 144) as the end point for detection compared to normal, hyperplastic polyp, IBD and adenoma (total n =254). For the detection of colorectal cancer, the area under the curve (AUC) was 84.2% (95% CI: 79.7–88.6) for the sHLA-G assay (Fig. 2b). The sensitivity and specificity of sHLA-G were analyzed using variable cutoff values (Table 2). Sensitivity ranged from 29.2 to 100% and specificity varied from 31.1 to 100% for the different cutoff values. Specifically, when cutpoint (88.6 U/mL) was used as the cutoff value, a sensitivity of 72.2% (104/144) and a specificity of 87.8% (223/254) were achieved for the overall series. Lowering the cutpoint to 46.3 U/mL would improve the sensitivity for the detection of colorectal cancer to 84.7% but lower the specificity to 70.9%.

Figure 2.

ROC curves for the detection of colorectal cancer using CEA (a), sHLA-G (b) and combined CEA and sHLA-G (c). The area under the curve was 65.2% for CEA, 84.2% for sHLA-G and 87.4% for the combined CEA and sHLA-G. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Table 2. Sensitivity and specificity for sHLA-G (U/mL) and CEA (ng/mL) at specified cutoffs
inline image

We determined the levels of the established tumor marker CEA using a commercial immunoassay and compared the sensitivity and specificity of sHLA-G to that of the CEA. The AUC for the CEA assay was 65.2% (95% CI: 59.2–71.2; Fig. 2a), which was significantly smaller than that for sHLA-G (p < 0.001), indicating that sHLA-G was superior to CEA in differentiating colorectal cancer from benign colorectal diseases. Using the standard cutoff value (5 ng/mL) for CEA would generate a sensitivity of 44.4% (64/144) and a specificity of 76.0% (193/254) for the detection of colorectal cancer (Table 2).

Also, we tried to combine sHLA-G and CEA to improve the sensitivity for detection of colorectal cancer by allowing either test to be positive. ROC curves analysis showed that the AUC for the combined sHLA-G and CEA was 87.4% (95% CI: 83.5–91.2; Fig. 2c), which was statistically significant compared to the AUC of sHLA-G (p < 0.05) and the AUC of CEA (p < 0.001). We found there was significant correlation between sHLA-G and CEA (r = 0.210, p < 0.001).

Correlation between sHLA-G concentration and clinicopathological parameters of patients with colorectal cancer

We investigated the correlation between serum sHLA-G levels and clinicopathological parameters of patients with colorectal cancer, such as age, gender, TNM stage and histological grade (Table 3). We found no significant association between sHLA-G concentration and the clinicopathological parameters of age, gender, TNM stage and histological grade.

Table 3. Clinicopathological characteristics of patients with colorectal cancer
inline image

Discussion

HLA-G is a tumor-associated protein that is expressed by a variety of neoplastic diseases,13 which indicated that HLA-G might serve as a clinical marker for the diagnosis or prediction of the clinical outcomes of these diseases. Fukushima et al.17 studied HLA-G expression in 39 human colorectal cancers and 23 extraneoplastic colon tissue samples by reverse transcription polymerase chain reaction (RT-PCR). They found that the expression of HLA-G mRNA was significantly more frequent in colorectal cancer (34 of 39 cases) than in the extraneoplastic tissue. Ye et al.18 used immunohistochemistry method to investigate HLA-G expression in 201 patients with colorectal carcinomas. In their study, HLA-G protein expression was observed in 64.6% (130/201) of the primary-site colorectal carcinomas, but not in the normal colorectal tissues or benign adenomas. Moreover, HLA-G expression in the tumors was significantly correlated with the depth of invasion, histological grade, host immune response, lymph nodal metastasis and clinical stages of the disease. All these studies confirmed the aberrant expression of HLA-G protein in colorectal cancer tissues, which supported that HLA-G expression may represent a mechanism used by tumor cells to escape from immunosurveillance. However, little data on sHLA-G in serum of patients with colorectal cancer have been reported. Accordingly, serum sHLA-G levels in colorectal cancer and its potential clinical value for differentiating colorectal cancer and benign colorectal diseases become a subject of great interest.

In our study, we used the monoclonal antibody MEM-G/9 to detect the major sHLA-G (soluble HLA-G1 and HLA-G5), which is the most abundant soluble isoforms in serum,19 and we found that sHLA-G were significantly increased in patients suffering from colorectal cancer compared to normal individuals and patients with benign colorectal diseases, independent of age and gender. The sHLA-G expression may be linked to a specific regulatory process in the alternative splicing of the primary HLA-G transcript, which could favor selection of the soluble isoforms.20 The elevated sHLA-G in colorectal cancer may reinforce the host's immune suppression through inhibiting the functions of NK and CTL cells and thereby favor the survival of tumor cells.14 However, we did not find significant association between sHLA-G serum concentration and the stage or grade of colorectal cancer, which may be caused by the small numbers of patients. Because HLA-G protein is also present in some adult tissues,21 it is not surprising that it is detectable in the serum of normal individuals. Thus, the sHLA-G level in normal individuals may be considered as the physiological level. Also, we found that serum sHLA-G concentration in benign colorectal diseases was significantly higher than in normal colorectum, which confirmed that sHLA-G might be a key cytokine of immune and inflammatory responses.22, 23 Previous studies have found that HLA-G was upregulated in premalignant and malignant lesions, whose expression was mainly restricted to the malignant sites.24, 25 In this study, the increased serum sHLA-G levels in benign colorectal diseases and colorectal cancer was also revealed, suggesting that the elevated sHLA-G may serve to mediate immune responsiveness in benign colorectal diseases and colorectal cancer.

Based on the above studies, we further assessed the feasibility of using serum sHLA-G levels in differentiating colorectal cancer from benign colorectal diseases. Singer et al.16 have proposed that sHLA-G might serve as clinical marker for distinguishing malignant versus benign ascites. In their study, the level of sHLA-G was significantly higher in malignant ascites that in benign ascites. The area under the ROC curve was 95% in evaluating sHLA-G levels as the diagnostic tool to detect ovarian and breast cancer. Similarly, in our study, ROC curve for sHLA-G revealed an AUC of 84.2% for distinguishing colorectal cancer from benign colorectal diseases, which provides the additional evidence that sHLA-G is a useful indicator in differentiating colorectal cancer from benign colorectal diseases.

Over the years, numerous soluble tumor-associated markers in serum have been used for the detection of colorectal cancer, such as CEA, carbohydrate antigen 19-9 and lipid-associated sialic acid. However, according to the American Society of Clinical Oncology's statement, none of them can be recommended for screening and detection because of their low sensitivity, and their use should be limited to postsurgery surveillance.26–31 For example, CEA, the most widely used serum marker in colorectal cancer, offers a sensitivity of only 30–40% for early stage tumors,30–32 which has also been confirmed in our study with a sensitivity of 44.4% and a specificity of 76.0%. In our study, sHLA-G ELISA achieved a sensitivity of 72.2% and a specificity of 87.8% at a cutoff value of 88.6 U/mL for the detection of colorectal cancer. Both of sensitivity and specificity have been improved. Moreover, the AUC for sHLA-G showed higher detection capability than that for CEA (p < 0.001). These assays were based on ELISA methods, which were easily standardized and reproducible.

In our study, we do not find any associations between the levels of sHLA-G and all clinical pathological parameters in patients with colorectal cancer, which was inconsistent with the results of Ye et al.18 This discordant finding can be explained by the fact that HLA-G in colorectal cancer tissue specimens by immunohistochemistry is all HLA-G isoforms, whereas HLA-G in serum by ELISA is soluble isoforms (mainly soluble HLA-G1 and HLA-G5). Moreover, these results are based on the measurement of only a limited number of patients. Although these are our favorite explanations, it is clear that other possibilities should also be pointed out in further studies.

In conclusion, in our study, we quantitatively demonstrate that sHLA-G is significantly increased in the serum of patients with colorectal cancer which may serve as a potent mediator of immune escape in colorectal cancer, and the detection of sHLA-G in serum may provide a novel noninvasive approach in distinguishing colorectal cancer from benign colorectal diseases. In addition, we combined sHLA-G and CEA for the detection of colorectal cancer and found that this combination could improve the detection capability. Although higher efficiency would be desirable, further study should elucidate whether detection capability for colorectal cancer could be improved by combining the measurement of sHLA-G with other tumor-associated markers.

Acknowledgements

The authors thank Dr. Alan H. Wu (Department of Laboratory Medicine, University of California, San Francisco) for helpful suggestions and critically reviewing the manuscript, and the staff of the Endoscopy Department for their help in specimen collection.

Ancillary