Evaluation of ascitic soluble human leukocyte antigen-G for distinguishing malignant ascites from benign ascites

Abstract

The overexpression of soluble human leukocyte antigen-G is associated with malignant tumours. The purpose of our study was to detect soluble human leukocyte antigen-G concentrations in ascites and to evaluate the value of ascitic soluble human leukocyte antigen-G for the diagnosis of malignant ascites. Enzyme-linked immunosorbent assay was used to detect soluble human leukocyte antigen-G levels in 64 patients with malignant ascites and 30 patients with benign ascites. Receiver operating characteristic curves were used to evaluate the diagnostic efficacy of ascitic soluble human leukocyte antigen-G for the detection of malignant ascites. Ascitic soluble human leukocyte antigen-G levels were significantly higher in the malignant ascites group than in the benign ascites group (20.718 ± 3.215 versus 12.467 ± 3.678 µg/L, t = 7.425, p < 0.001). The area under the receiver operating characteristic curve for ascitic soluble human leukocyte antigen-G was 0.957 (95% confidence interval, 0.872–0.992). At a cut-off value of 19.60 µg/L, the sensitivity and specificity of ascitic soluble human leukocyte antigen-G were 87.5% (95% confidence interval, 71.0%–96.5%) and 100% (95% confidence interval, 88.4%–100%), respectively. With respect to area under the receiver operating characteristic curve, sensitivity and specificity, ascitic carcinoembryonic antigen (0.810, 68.75% and 83.33%, respectively) and carbohydrate antigen 19-9 (0.710, 65.63% and 70%, respectively) significantly differed (all p < 0.05). In malignant ascites that were cytology-negative and biopsy-positive, the rate of positivity for ascitic soluble human leukocyte antigen-G was 75%, which was higher than the corresponding rates for ascitic carcinoembryonic antigen (31.25%) and carbohydrate antigen 19-9 (6.25%; both p < 0.05). In conclusion, The detection of ascitic soluble human leukocyte antigen-G exhibited good performance for diagnosing malignant ascites, and particularly those that were cytology-negative and biopsy-positive.

Keywords

Ascitic soluble human leukocyte antigen-G malignant ascites benign ascites distinguish evaluation

Introduction

The differential diagnosis of ascites is a difficult problem for clinicians and laboratory examinations. It not only has a great value in defining the aetiology and evaluating the prognosis of a disease but also is directly related to the determination of therapy regimens. Consequently, it is essential to define the nature of ascites at an early stage. Currently, the most widely used clinical method for the differential diagnosis of ascites consists of cytological examination, peritoneal biopsy and the combined detection of ascitic tumour markers, but each method has its shortcomings, including expense, technical requirements and lack of sensitivity or specificity. Therefore, a high sensitivity and specificity biomarker that can distinguish malignant from benign ascites may have great significance for the early diagnosis of disease.

Human leukocyte antigen-G (HLA-G) is a non-classical major histocompatibility (MHC) class I antigen. Studies have shown that the HLA-G may play a direct immunosuppressive role by binding to inhibitory receptors expressed on the immune cells and mediating immune tolerance through the induction of regulatory T cells and ‘trogocytosis’, such that tumour cells can escape host immune surveillance and further grow in the body.¹ Clinically detected soluble human leukocyte antigen-G (sHLA-G) is partly composed of sHLA-G1 and mainly composed of HLA-G5.² Accumulating evidence has demonstrated that sHLA-G is overexpressed in tissues or plasma of many types of malignant tumours, including lung cancer,³ colorectal cancer (CRC),⁴ gastric cancer (GC),⁵ oesophageal squamous cell carcinoma (ESCC),⁵ neuroblastoma,⁶ cervical cancer,⁷ haematological malignancies,⁸ ovarian cancer and breast cancer.⁹ Research has also shown that sHLA-G levels are associated with disease stage, and higher levels of sHLA-G are usually observed among non-small-cell lung cancer (NSCLC) patients at a later stage of the disease or who have poor survival.³ These studies suggest that sHLA-G may be used as a potential biomarker to identify malignant and benign tumours. Therefore, we hypothesised that a characteristic concentration of sHLA-G may be detected in malignant ascites caused by various types of tumours. In this study, we measured the level of sHLA-G in ascites using an enzyme-linked immunosorbent assay (ELISA). We then evaluated the potential value of ascitic sHLA-G for diagnosing malignant ascites by performing analyses of receiver operating characteristic (ROC) curves and comparisons with tumour-associated markers. Finally, we further calculated the positive rate of ascitic sHLA-G in malignant ascites that were cytology-negative and biopsy-positive.

Materials and methods

Patients

In this study, 94 patients with ascites admitted to the First Affiliated Hospital of Xi’an Medical University from January 2016 to December 2016 were divided into a malignant ascites group (64 cases) and a benign ascites group (30 cases) according to their final diagnoses. The participants included 50 males and 44 females with an average age of 56.04 ± 14.53 years. The malignant ascites group included 18 primary liver cancers, 15 lung cancers, 10 colon cancers, 8 GCs, 6 pancreatic cancers, 2 cervical cancers, 1 ampullary carcinoma, 1 renal cancer, 1 ovarian cancer, 1 breast invasive ductal carcinoma and 1 non-Hodgkin’s lymphoma. The benign ascites group included 20 patients with liver cirrhosis and 10 patients with tuberculous peritonitis. Diagnosis of malignant ascites was dependent on cytology-positive, endoscopic biopsy pathology–positive or surgical pathology confirmed status, except in cases with other causes of ascites. Benign ascites were diagnosed by medical history, clinical manifestation, physical examination, imaging examination and the corresponding laboratory examination. No subjects were administered treatment prior to the first ascites collection, and the malignant ascites group did not receive systemic or local chemotherapy 3–4 weeks prior to collection of ascites. Informed consent was obtained from all individual participants included in our study, which was approved by the ethics committee of the First Affiliated Hospital of Xi’an Medical University.

Ascites samples

A 20-mL sample of ascites was collected from each patient by intraperitoneal puncture, and 10 mL was centrifuged at 3000×g for 10 min at 4°C. The supernatant was divided into aliquots and frozen at −80°C. The remaining ascites material was transferred rapidly to the laboratory for measurements of alpha-fetoprotein (AFP), carbohydrate antigen 125 (CA125), carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 (CA19-9).

Methods

ELISA for ascitic sHLA-G

sHLA-G was detected using the human sHLA-G ELISA kit (AMEKO, Shanghai, China) following the manufacturer’s instructions. Calibrators and ascites samples were first added to the microtiter plate wells precoated with mouse anti-HLA-G monoclonal antibody, which could recognize the soluble isoforms (sHLA-G1 and HLA-G5). After incubation for 60 min and washing, monoclonal antihuman beta2-microglobulin antibody labelled with horseradish peroxidase (HRP) was added to the wells for 60 min, allowing formation of the antibody–sHLA-G complex. Following another washing step, the remaining HRP conjugate was reacted by addition of the substrate solution. After sulphuric acid stop solution was added, a microplate reader was used to measure the absorbance of the yellow product at 450 nm (Bio-Rad Model 550, Hercules, CA, USA). The linear regression equation for the standard curve was constructed according to the concentrations of the calibrators and the corresponding absorbance values, and the sHLA-G levels of ascites samples were calculated using this calibrator curve. The intra-assay variation was 4.5%, and the inter-assay variation was 8.8%. The sensitivity of detection was 0.6 µg/L.

Ascitic AFP, CA125, CEA and CA19-9 assay

Ascitic AFP, CA125, CEA and CA19-9 were measured using chemiluminescence. Kits for the quantitative determination of the four indices were purchased from Roche Diagnostics Co., Ltd (Shanghai, China). The quantitative determination of ascitic AFP, CA125, CEA and CA19-9 was performed by a specialist in the laboratory of our hospital.

Statistical analysis

Data were expressed as mean ± standard deviation (SD) or median (range) depending on their distribution types as assessed using the Kolmogorov–Smirnov test. An independent sample T-test was used to compare sHLA-G levels between the two groups. Mann–Whitney U-tests were used to compare AFP, CA125, CEA and CA19-9 levels between the two groups. Rates were compared using chi-square tests. The area under the ROC curves (AUC) was compared by a Z-test using MedCalc statistical software version 15.2.2 (MedCalc Software bvba, Ostend, Belgium). A value of p < 0.05 was considered statistically significant. The statistical analysis was performed using SPSS software version 18.0 for Windows (SPSS, Chicago, IL, USA).

Results

Levels of ascitic sHLA-G and other tumour-associated markers in the two groups

The levels of sHLA-G and other tumour-associated markers are presented in Table 1. No significant differences in age and gender were found between the two groups (both p > 0.05). Ascitic sHLA-G levels were significantly higher in the malignant ascites group than in the benign ascites group (20.718 ± 3.215 versus 12.467 ± 3.678 µg/L, t = 7.425, p < 0.001).

Table 1.

Ascitic sHLA-G and other tumour-associated markers in the two groups (n = 94).

	Benign ascites group (n = 30)	Malignant ascites group (n = 64)	p
Gender
Male/female	16/14	34/30	0.985
Age (years)
Mean ± SD	54.467 ± 7.978	58.438 ± 8.120	0.069
Ascitic sHLA-G (µg/L)
Mean ± SD	12.467 ± 3.678	20.718 ± 3.215	0.000
Ascitic AFP (ng/mL)
Median (range)	1.175 (0.140–86.210)	1.910 (0.014–1280.48)	0.063
Ascitic CA125 (U/mL)
Median (range)	771.665 (119.02–4581.32)	1251.60 (238.95–5135.00)	0.082
Ascitic CEA (ng/mL)
Median (range)	0.700 (0.130–4.940)	11.045 (0.190–1304.960)	0.000
Ascitic CA19-9 (U/mL)
Median (range)	3.180 (0.800–26.240)	7.035 (0.700–1327.850)	0.005

SD: standard deviation; sHLA-G: soluble human leukocyte antigen-G; AFP: alpha-fetoprotein; CA125: carbohydrate antigen 125; CEA: carcinoembryonic antigen; CA19-9: carbohydrate antigen 19-9.

In addition, the median ascitic CEA level in the malignant ascites group was 11.045 ng/mL (range 0.190–1304.960 ng/mL), which was significantly higher than the corresponding level of 0.700 ng/mL in the benign ascites group (range 0.130–4.940 ng/mL; Z = −4.160, p = 0.000). The median ascitic CA19-9 level in the malignant ascites group was 7.035 U/mL (range 0.700–1327.850 U/mL), which was also significantly higher than the corresponding level of 3.180 U/mL in the benign ascites group (range 0.800–26.240 U/mL; Z = −2.817, p = 0.005). However, levels of ascitic AFP and CA125 were not significantly different between the two groups (both p > 0.05).

No correlations among levels of ascitic sHLA-G, CEA and CA19-9 were found in the malignant ascites group (p > 0.05).

ROC curve analysis

ROC curves were used to evaluate the diagnostic values of ascitic sHLA-G, CEA and CA19-9 levels in the differential diagnosis of benign and malignant ascites (Figure 1). Our analysis showed that the AUC values were 0.957 (95% confidence interval (CI), 0.872–0.992) for ascitic sHLA-G, 0.810 (95% CI, 0.690–0.899) for ascitic CEA and 0.710 (95% CI, 0.580–0.819) for ascitic CA19-9 (Table 2). A Z-test was used to compare the AUC of each marker. The results showed that the AUC of ascitic sHLA-G was significantly greater than the AUCs of ascitic CEA (Z = 2.368, p = 0.018) and ascitic CA19-9 (Z = 3.431, p = 0.001).

Figure 1.

ROC curves for the detection of malignant ascites by ascitic sHLA-G, CEA and CA19-9. The AUC of ascitic sHLA-G: 0.957(95% CI: 0.872–0.992). The AUC of ascitic CEA: 0.810(95% CI: 0.690–0.899). The AUC of ascitic CA19-9: 0.710 (95% CI: 0.580–0.819).

Table 2.

ROC curve analysis results for ascitic sHLA-G, CEA and CA19-9.

Marker	AUC	SE	Cut-off value	95% CI
Ascitic sHLA-G	0.957	0.023	19.60 µg/L	0.872–0.992
Ascitic CEA	0.810*	0.055	2.28 ng/mL	0.690–0.899
Ascitic CA19-9	0.710**	0.067	7.82 U/mL	0.580–0.819

AUC: area under the receiver operating characteristic curve; SE: standard error; CI: confidence interval; sHLA-G: soluble human leukocyte antigen-G; CEA: carcinoembryonic antigen; CA19-9: carbohydrate antigen 19-9.

p < 0.05, **p < 0.01, versus ascitic sHLA-G.

At a cut-off value of 19.60 µg/L, the sensitivity and specificity of ascitic sHLA-G were 87.5% (95% CI, 71.0%–96.5%) and 100% (95% CI, 88.4%–100%), respectively. The performances of ascitic sHLA-G, CEA, CA19-9 and cytological examination for the differential diagnosis of benign and malignant ascites are shown in Table 3. The differences in the sensitivity and specificity of ascitic sHLA-G and CEA were statistically significant (χ² = 3.961 and χ² = 5.10; p < 0.05 and p < 0.05, respectively). When the sensitivity and specificity of ascitic sHLA-G were compared with those of CA19-9, the differences were also statistically significant (χ² = 4.267 and χ² = 6.10; p < 0.05 and p < 0.05). The difference in the sensitivity of ascitic sHLA-G and cytological examination was also statistically significant (χ² = 6.05, p = 0.014), and the specificity of both was 100%.

Table 3.

The performance of ascitic sHLA-G, CEA, CA19-9 and cytological examination in predicting malignant ascites.

Marker	Sensitivity (%)	Specificity (%)	Accuracy (%)	PPV (%)	NPV (%)
Ascitic sHLA-G	87.5 (56/64)	100 (30/30)	91.49 (86/94)	100	88.24
Ascitic CEA	68.75 (44/64)*	83.33 (25/30)*	73.40 (69/94)	81.48	71.43
Ascitic CA19-9	65.63 (42/64)*	70 (21/30)*	67.02 (63/94)	72.41	66.67
Ascites cytology	50 (32/64)*	100 (30/30)	65.96 (62/94)	100	65.22

PPV: positive predictive value; NPV: negative predictive value; sHLA-G: soluble human leukocyte antigen-G; CEA: carcinoembryonic antigen; CA19-9: carbohydrate antigen 19-9.

p < 0.05, versus ascitic sHLA-G.

The relationships between ascitic sHLA-G level and age, gender and tumour type for patients with malignant ascites

In the malignant ascites group, differences in age and gender were not associated with significant differences in ascitic sHLA-G level (p > 0.05; Table 4).

Table 4.

The relationships between the ascitic sHLA-G level and gender, age and tumour type for patients with malignant ascites.

	N	Ascitic sHLA-G (µg/L), mean ± SD	t/F	p
Gender
Male	34	16.856 ± 4.660	t = −0.554	0.582
Female	30	17.726 ± 4.737
Age (years)
<60	28	21.064 ± 3.215	t = 0.565	0.576
≥60	36	20.414 ± 3.282
Tumour types
Primary liver cancers	18	20.095 ± 1.286
Lung cancers	15	23.180 ± 1.718
Colon cancers	10	17.586 ± 4.687
Gastric cancers	8	18.373 ± 4.633
Pancreatic cancers	6	21.420 ± 1.693	F = 2.661	0.039
Cervical cancers	2	21.955 ± 2.864
Ampullary carcinoma	1	25.340 ± 0.000
Renal cancer	1	23.880 ± 0.000
Ovarian cancer	1	21.372 ± 0.000
Non-Hodgkin’s lymphoma	1	21.791 ± 0.000
Breast cancer	1	20.346 ± 0.000

sHLA-G: soluble human leukocyte antigen-G; SD: standard deviation.

The positive rate of ascitic sHLA-G in malignant ascites that were cytology-negative and biopsy-positive

The positive rate of ascitic sHLA-G was 75% in malignant ascites that were cytology-negative and biopsy-positive, which was significantly different than the rates of ascitic CEA (31.25%, p = 0.001) and ascitic CA19-9 (6.25%, p = 0.000; Table 5).

Table 5.

The positive rate of ascitic sHLA-G, CEA and CA19-9 in malignant ascites that were cytology-negative and biopsy-positive.

Marker	Malignant ascites that were cytology-negative and biopsy-positive (n = 32)		Positive rate (%)
Marker	(+)	(−)	Positive rate (%)
Ascitic sHLA-G	24	8	75.00
Ascitic CEA	10	22	31.25**
Ascitic CA19-9	2	30	6.25***

sHLA-G: soluble human leukocyte antigen-G; CEA: carcinoembryonic antigen; CA19-9: carbohydrate antigen 19-9.

p < 0.01, ***p < 0.001, versus ascitic sHLA-G.

Discussion

The overexpression of sHLA-G in malignant lesions and increased sHLA-G levels in the periphery have been detected for various solid tumours. In a prior study, HLA-G expression was observed in 70.7% of breast cancer lesions, and plasma sHLA-G levels were significantly higher in the breast cancer group than in the control group. The AUC for predicting breast cancer was 0.953. In addition, plasma sHLA-G levels were significantly associated with increased frequencies of CD4+ CD25+ FoxP3+ Treg cells in breast cancer patients.¹⁰ Research has also shown that serum sHLA-G levels were significantly higher in patients with CRC than in patients with hyperplastic polyps, adenoma or inflammatory bowel disease or in healthy subjects. In particular, the AUC of serum sHLA-G for predicting CRC was 0.842; at a cut-off value of 88.6 U/mL, the sensitivity and specificity of serum sHLA-G were 72.2% and 87.8%, respectively.⁴ Furthermore, significantly higher plasma sHLA-G levels have been observed in CRC, GC, ESCC and NSCLC patients than in healthy controls; at a cut-off value of 49 U/mL, the highest sensitivities for sHLA-G were 94%, 85%, 91%, and 51%, respectively; the corresponding specificities were all 100%, and the AUCs for sHLA-G were 0.97, 0.91, 0.98 and 0.80, respectively.⁵ However, sHLA-G levels in malignant ascites caused by various solid tumours have rarely been reported. The research by Singer et al. showed that sHLA-G levels in malignant ascites caused by ovarian and breast cancer were significantly higher than in benign ascites. In particular, at a cut-off of 13 ng/mL, the AUC, sensitivity and specificity for ascitic sHLA-G were 0.95, 78% and 100%, respectively.⁹ In contrast, Sipak-Szmigiel et al.¹¹ showed that serum and ascitic sHLA-G concentrations were not statistically significantly different among ovarian serous cysts, ovarian endometriosis and ovarian cancers but that the median levels of ascitic sHLA-G were higher than the corresponding serum levels. Our study found that levels of ascitic sHLA-G were significantly higher in malignant ascites induced by various solid tumour sources than in benign controls. The ROC curve for ascitic sHLA-G revealed an AUC of 0.957. At a cut-off value of 19.60 µg/L, the sensitivity and specificity of ascitic sHLA-G were 87.5% (95% CI, 71.0%–96.5%) and 100% (95% CI, 88.4%–100%), respectively. These results suggested that ascitic sHLA-G exhibited good performance for diagnosing malignant ascites.

How does the ascitic sHLA-G compare with other tumour-associated markers in predicting the diagnosis of malignant ascites? The most common cause of malignant ascites is local malignant neoplasms, particularly gastrointestinal tumours and gynaecological tumours, for which the related tumour markers are AFP, CA125, CEA and CA19-9. This study showed that sHLA-G was superior to CEA in the differential diagnosis of benign and malignant colorectal disease.⁴ sHLA-G also has better diagnostic performance than squamous cell carcinoma antigen (SCC-Ag) and CEA for CRC, GC, ESCC and NSCLC.⁵ Our study showed that ascitic CEA and CA19-9 levels were significantly higher in the malignant ascites group than in the benign ascites group but that there was no significant between-group difference in ascitic AFP and CA125 levels. The AUC for the detection of malignant ascites was 0.810 for ascitic CEA and 0.710 for ascitic CA19-9, both AUC values were lower than that of ascitic sHLA-G. In addition, the sensitivity and specificity of ascitic CEA were 68.75% and 83.33%, and those of CA19-9 were 65.63% and 70%, respectively, which were also lower than those of ascitic sHLA-G. Taken together, our results indicate that ascitic sHLA-G is superior to ascitic CEA and CA19-9 for the differential diagnosis of benign and malignant ascites.

In a prior study, HLA-G expression in ESCC lesions was significantly associated with histological tumour grade, degree of invasion, lymph node metastasis, immune response and disease stage.¹² A study of 201 patients with CRC produced similar results.¹³ In addition, an investigation found that HLA-G expression in hepatocellular carcinoma was significantly correlated with disease stage and age.¹⁴ In contrast, another study showed no significant correlations between HLA-G expression in primary bladder transitional cell carcinoma (TCC) lesions and gender, age or disease stage. Furthermore, no difference in plasma sHLA-G levels was found between TCC patients and normal controls.¹⁵ Another study also did not reveal correlations between serum sHLA-G levels in cervical cancer patients and clinicopathological characteristics.¹⁶ Our results showed that in patients with malignant ascites, ascitic sHLA-G levels were unrelated to age and gender and were also not correlated with ascitic CEA and CA19-9 levels, suggesting that ascitic sHLA-G may be used as an independent indicator for the early diagnosis of malignant ascites. Because the sample size for each tumour type was small, we could not determine whether the ascitic sHLA-G levels in malignant ascites induced by different types of tumours were different.

Ascites cytology is the gold standard for malignant diagnosis in ascitic fluid, but the sensitivity is too low. This study showed that the sensitivity was only 50%. Therefore, diagnosing malignant ascites that are cytology-negative is a problem for clinicians. A previous study demonstrated that tumour markers were helpful in distinguishing malignant ascites from ascites that were cytology-negative, with sensitivity and specificity of 86% and 97%, respectively, in the case of detection using a combination of ascitic tumour markers.¹⁷ Our study showed that the rate of positivity for sHLA-G was 75% in the 32 cases with malignant ascites that were cytology-negative; this rate was higher than the corresponding rates for ascitic CEA and CA19-9. These data suggested that ascitic sHLA-G would be helpful in screening for cytology-negative malignant ascites and can be used as an important adjuvant diagnostic tool for ascites cytology. Studies by Singer et al. showed that sHLA-G could be detected in almost all malignant ascitic fluids from ovarian cancer and breast cancer, but the positive rate of HLA-G expression was only 61% and 25% in tissues of ovarian cancer and breast cancer, respectively. In addition, immunoreactivity for HLA-G was not detected by immunohistochemistry in certain cancer tissues and ascitic cell pellets, but sHLA-G levels were found to be elevated in the corresponding ascites supernatants. Tumour cells in ascites likely could secrete sHLA-G, leading to elevated sHLA-G levels in ascites.⁹

Real-time polymerase chain reaction (RT-PCR) analysis of free tumour cells has been widely used in malignant tumour micrometastasis detection. In one study, RT-PCR was used to detect CEA messenger RNA (mRNA) in peritoneal lavage fluid from 156 patients with GC undergoing staging laparoscopy; the sensitivity of the PCR was 79%, compared to 61% for cytology of peritoneal disease detected at laparoscopy, and the positive rate of the PCR was still 47% in cytology-negative cases. In patients with laparoscopy-negative results, the positive rate of cytology was 7%, whereas that of PCR was 24%. This result illustrates that RT-PCR was more sensitive for detecting subclinical peritoneal disease than cytology. In addition, further follow-up showed that the positive rate of PCR was related to increased recurrence and poor prognosis of the disease.¹⁸ In another study, RT-PCR-positive and cytology-negative patients with pancreatic cancer who underwent R0 resection had earlier recurrence of peritoneal and overall disease than RT-PCR-negative patients, and no correlation with the state of the lymph nodes was found.¹⁹ Similar results were found in CRC.²⁰ Moreover, quantitative RT-PCR results (CK20 and/or CEA levels) were an independent prognostic factor in GC patients with R0 resection.²¹ These studies suggest that we can perform RT-PCR analysis of sHLA-G in cytology-negative ascites to further explore and evaluate the diagnostic value of ascitic sHLA-G.

In summary, we quantitatively demonstrate that ascitic sHLA-G is significantly increased in malignant ascites caused by various types of solid tumours. In addition, ascitic sHLA-G exhibited a better performance in detecting malignant ascites than ascitic CEA and CA19-9. The rate of positivity for ascitic sHLA-G was also high in malignant ascites that were cytology-negative; thus, ascitic sHLA-G may be helpful in screening for cytology-negative malignant ascites.

Footnotes

Acknowledgements

The authors thank the following members of doctors of Department of Gastroenterology and Oncology, the First Affiliated Hospital of Xi’an Medical University, for providing ascites samples for this study: Bin Li, Rong Yan, Jie-Fei Song and Qiu-Mei Tian. Conceived and designed the experiments: Chun-Yan Niu. Performed the experiments, analysed the data and wrote the article: Juan Sun. Contributed ascite samples: Yan-Xiang Chang. Chun-Yan Niu: This article has not been published and is not under consideration for publication elsewhere. All authors have read and approved the final version of the article.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Ethical approval

Informed consent was obtained from all individual participants included in our study, which was approved by the ethics committee of the First Affiliated Hospital of Xi’an Medical University.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Yan

WH.

Human leukocyte antigen-G in cancer: are they clinically relevant?

Cancer Lett 2011; 311: 123–130.

Pistoia

Morandi

Wang

et al . Soluble HLA-G: are they clinically relevant? Semin Cancer Biol 2007; 17: 469–479.

Lin

Zhu

Chen

et al . Clinical relevance and functional implications for human leucocyte antigen-G expression in non-small-cell lung cancer. J Cell Mol Med 2010; 14: 2318–2329.

Zhu

Wang

Zhang

et al . Serum sHLA-G levels: a useful indicator in distinguishing colorectal cancer from benign colorectal diseases. Int J Cancer 2011; 128: 617–622.

Cao

Yie

Liu

et al . Plasma soluble HLA-G is a potential biomarker for diagnosis of colorectal, gastric, esophageal and lung cancer. Tissue Antigens 2011; 78: 120–128.

Morandi

Levreri

Bocca

et al . Human neuroblastoma cells trigger an immunosuppressive program in monocytes by stimulating soluble HLA-G release. Cancer Res 2007; 67: 6433–6441.

Zheng

Wang

Zhang

et al . Up-regulation of HLA-G expression in cervical premalignant and malignant lesions. Tissue Antigens 2011; 77: 218–224.

Sebti

Le Maux

Gros

et al . Expression of functional soluble human leucocyte antigen-G molecules in lymphoproliferative disorders. Br J Haematol 2007; 138: 202–212.

Singer

Rebmann

Chen

et al . HLA-G is a potential tumor marker in malignant ascites. Clin Cancer Res 2003; 9: 4460–4464.

10.

Chen

Lin

Shen

et al . Upregulation of human leukocyte antigen-G expression and its clinical significance in ductal breast cancer. Hum Immunol 2010; 71: 892–898.

11.

Sipak-Szmigiel

Włodarski

Ronin-Walknowska

et al . Serum and peritoneal fluid concentrations of soluble human leukocyte antigen, tumor necrosis factor alpha and interleukin 10 in patients with selected ovarian pathologies. J Ovarian Res 2017; 10: 25.

12.

Yie

Yang

et al . Expression of HLA-G is associated with prognosis in esophageal squamous cell carcinoma. Am J Clin Pathol 2007; 128: 1002–1009.

13.

Yang

et al . Human leukocyte antigen G expression: as a significant prognostic indicator for patients with colorectal cancer. Mod Pathol 2007; 20: 375–383.

14.

Lin

Chen

Zhu

et al . Aberrant human leucocyte antigen-G expression and its clinical relevance in hepatocellular carcinoma. J Cell Mol Med 2010; 14: 2162–2171.

15.

Gan

Huang

Zhang

et al . Tumor-specific upregulation of human leukocyte antigen-G expression in bladder transitional cell carcinoma. Hum Immunol 2010; 71: 899–904.

16.

Samuels

Ferns

Meijer

et al . High levels of soluble MICA are significantly related to increased disease-free and disease-specific survival in patients with cervical adenocarcinoma. Tissue Antigens 2015; 85: 476–483.

17.

Liu

Kong

Dou

et al . Evaluation of tumor markers for the differential diagnosis of benign and malignant ascites. Ann Hepatol 2014; 13: 357–363.

18.

Wong

Kelly

Mittra

et al . RT-PCR increases detection of submicroscopic peritoneal metastases in gastric cancer and has prognostic significance. J Gastrointest Surg 2012; 16: 889–896.

19.

Kelly

Wong

Gladdy

et al . Prognostic impact of RT-PCR-based detection of peritoneal micrometastases in patients with pancreatic cancer undergoing curative resection. Ann Surg Oncol 2009; 16: 3333–3339.

20.

Lloyd

Mclver

Stephenson

et al . Identification of early-stage colorectal cancer patients at risk of relapse post-resection by immunobead reverse transcription-PCR analysis of peritoneal lavage fluid for malignant cells. Clin Cancer Res 2006; 12: 417–423.

21.

Katsuragi

Yashiro

Sawada

et al . Prognostic impact of PCR-based identification of isolated tumour cells in the peritoneal lavage fluid of gastric cancer patients who underwent a curative R0 resection. Br J Cancer 2007; 97: 550–556.