Application of proximity ligation assays to identify potential plasma biomarkers in oral cavity cancer patients: A case control study

Abstract

BACKGROUND:

Practical cancer biomarkers for oral cavity cancer are currently in limited use.

OBJECTIVE:

We aimed to investigate the differences in soluble E-cadherin between patients with oral cavity cancer and matched healthy participants via Proximity Ligation Assay (PLA).

METHODS:

Samples were taken from both patients diagnosed with oral cavity cancer, as well as non-cancerous participants. PLA was used to detect soluble E-cadherin and Cycle threshold (Ct) values derived from qPCR in order to calculate the number of starting amplicons.

RESULTS:

In total, 74 patients with oral cavity cancer and 55 matched non-cancerous participants were included for final analysis. The Ct value of E-cadherin was found to be lower in oral cavity cancer patients when compared with that of the matched non-cancerous participants (20.72 $\pm$ 0.39 versus 21.27 $\pm$ 0.45, $P<$ 0.001). Using a Ct value of 20.9 as a cut-off point, the sensitivity and specificity of discriminating patients with oral cavity cancer from the healthy controls was 63.5% and 87.3%, respectively.

CONCLUSION:

Plasma soluble E-cadherin levels were significantly higher in patients with oral cavity cancer when compared with those from the matched non-cancerous participants.

Keywords

Oral cavity cancer proximity ligation assay biomarker squamous cell carcinoma E-cadherin

1. Introduction

Amongst all cancers, oral cavity cancer incidence ranked 11th worldwide in 2013. It has been reported that 409,360 newly diagnosed cases of oral cavity cancer occurred globally and that 135,000 related deaths were found in 2013. In addition, oral cavity cancer caused 3,600,000 disability-adjusted life-years [1]. Oral cavity cancer has also ranked 5th amongst all causes of death from cancer in Taiwan. The annual death toll for oral cavity cancer in men has increased rapidly in recent years according to statistical data from Taiwan’s Ministry of Health and Welfare of the Executive Yuan [2]. Although a revolution of combined modalities for treatment of oral cavity cancer patients has improved the quality of life after diagnosis, the 5-year survival rate has remained stationary over the recent decade [3].

Epithelial-mesenchymal Transition (EMT) is a crucial pathway which develops metastasis and stem-like properties of cancer cells [4]. Suppression of the adhesion protein E-cadherin is closely associated with EMT. The reduced expression of E-cadherin during immunohistochemical examination has been linked to both a poorer prognosis and aggressive behavior in oral squamous cell carcinoma [5, 6]. On the contrary, the amounts of soluble E-cadherin identified by Enzyme-linked Immunosorbent Assay (ELISA) were meaningfully greater in patients with breast cancer [7], lung cancer [8], esophageal cancer [9], and colorectal cancer [10] when compared with healthy controls. However, few studies have investigated soluble E-cadherin in patients diagnosed with oral cavity cancer. Proximity Ligation Assay (PLA) was first reported by Fredriksson in 2002. PLA utilizes the simultaneous binding of two independent affinity reagents to their target protein to create a signal [11]. The assay offers both a better dynamic range and limit of identification when compared with traditional ELISA [12].

In the current study, we intended to examine the differences in soluble E-cadherin, which was detected by PLA, between patients with oral cavity cancer and matched healthy controls.

2. Materials and methods

This prospective protocol was carried out in Taichung Veterans General Hospital (TCVGH), a teaching hospital in Taiwan. This study was assessed and approved by the Institutional Ethical Committee of TCVGH. Patients with newly diagnosed oral cavity cancer from TCVGH beginning in July of 2015 were considered eligible for participation in the study. Those who refused surgical intervention, had a pathological report of non-squamous cell carcinoma, or were reluctant to join the study were excluded. Matched healthy controls without history of any type of cancer were recruited in clinic during the same period. Healthy participants were matched for gender, age, and personal habits. After a detailed explanation surrounding the complete protocol, all participants signed an informed consent form prior to enrollment. Demographic information and clinically relevant data were documented. Treatment protocol was conducted after discussion and through the consensus of a multidisciplinary team (the oral cavity cancer team of TCVGH). All patients were restaged in accordance to the American Joint Committee on Cancer (8th edition). All surgeries were completed by a single surgeon (S. A. Liu) and PLA was accomplished by the laboratory staff who were blind to the patients’ clinical data.

Peripheral blood from patients with oral cavity cancer was drawn before surgery and deposited in an EDTA-treated tube. Peripheral blood from the matched healthy controls was obtained during clinic visits. Peripheral blood was then centrifuged at 1000 $\times$ g for fifteen minutes, and the plasma layer transferred to a 1.5-ml microtube. The mononuclear cell layer of the sample was relocated to a clean 50 ml tube. The sample was washed twice with a balanced buffer solution, and then centrifuged at 150 $\times$ g for ten minutes. The samples were stored at $-$ 30 ${}^{\circ}$ C until use.

PLA was carried out mainly in accordance to protocol reported by Lundberg et al. [13]. Briefly, 1 $\mu$ l of each sample was mixed with a plasma dilution buffer at a 1:1 ratio, along with an internal control standard spike-in mix of 200 pm GFP, 20 pm PE, 40 pm APC, and a 40 fM of a sequence to assessthe quality of response [13]. Two $\mu$ l of each sample was then incubated at room temperature for twenty minutes, followed by adding 2 $\mu$ l of a proximity probe mixture comprising 1 X Probe Mix Diluent (RnD Systems, Minneapolis, MN, USA), 1% bovine serum albumin, plus 0.1% Triton X-100 (Olink Bioscience, Uppsala, Sweden). The mixture was then incubated overnight at four degrees Celsius. Afterward, 4 $\mu$ l of the probed mixtures was incubated at 37 ${}^{\circ}$ C for ten minutes with 96 $\mu$ l of a reaction mixture comprising 1 X T4 ligation buffer and 100 nm connector oligonucleotide, along with 0.006 units of T4 DNA ligase (Fermentas, Burlington, Ontario). The mixtures were then incubated at 65 ${}^{\circ}$ C for ten minutes for heat inactivation of the bound probes ligation. We subsequently added 1 units of uracil-DNA excision mix (Epicenter Madison, WI, USA) to the aforementioned mixtures, incubated at 37 ${}^{\circ}$ C for twenty minutes, followed by 70 ${}^{\circ}$ C for an additional ten minutes in order to digest the connector oligonucleotide for heat inactivation. Pre-amplification was then performed by blending 20 $\mu$ l of ligated mixtures and 5 $\mu$ l pooled Polymerase Chain Reaction (PCR) mix for seventeen cycles. This procedure involved a preliminary incubation at 95 ${}^{\circ}$ C for ten minutes, followed by two cycles for fifteen seconds at 95 ${}^{\circ}$ C, ten minutes at 46 ${}^{\circ}$ C, two minutes at 60 ${}^{\circ}$ C, and fifteen cycles for fifteen seconds at 95 ${}^{\circ}$ C, two minutes at 54 ${}^{\circ}$ C, and finally two minutes at 60 ${}^{\circ}$ C. The samples were lastly diluted 5-fold in a 1X Tris-EDTA buffer before real-time PCR.

Table 1
Descriptive and bivariate analysis between oral cavity cancer patients and non-cancerous participants

Variables	Total no. of patients (% in column)	No. of patients (%)		$P$ value
		Cancer ( $n=$ 74)	Non-cancerous ( $n=$ 55)
Age at presentation (year)		55.0 $\pm$ 10.0	54.1 $\pm$ 12.6	0.692
E-cadherin (Ct value)		20.72 $\pm$ 0.39	21.27 $\pm$ 0.45	$<$ 0.001
Gender (female/male)	36/93	9/65	8/47	0.895
Smoking
Yes	97 (75.2%)	59 (60.8%)	38 (39.2%)	0.239
No	32 (24.8%)	15 (46.9%)	17 (53.1%)
Pack-year		28.0 $\pm$ 25.0	17.9 $\pm$ 21.3	0.018
Alcohol
Heavy	38 (29.5%)	26 (68.4%)	12 (31.6%)	0.142
Social	52 (40.3%)	30 (57.7%)	22 (42.3%)
No	39 (30.2%)	18 (46.2%)	21 (53.8%)
Betel quid
Yes	92 (71.3%)	56 (60.9%)	36 (48.6%)	0.283
No	37 (28.7%)	18 (48.6%)	19 (51.4%)
Quid-year		306 $\pm$ 283	181 $\pm$ 159	0.004

The diluted DNA mixtures were incubated at 37 ${}^{\circ}$ C for thirty minutes in a combination which included 1.4 X Fast Universal Master Mix (Applied Biosystems Inc., Foster City, USA), dH ${}_{2}$ O and 0.05 units of uracil-DNA excision mix for the breakdown of primers which was added in the pre-amplification step to reduce interfering in the real-time quantitative PCR. The real-time PCR was carried out by ABI 9700 HT Fast (Applied BiosystemsInc., Foster City, USA). The reaction was started after blending 7 $\mu$ l of the pre-amplified product with 3 $\mu$ l of the qPCR mixture, (comprising 3 $\mu$ m primer, dH ${}_{2}$ O), plus 0.8 $\mu$ m TaqMan probe (Applied Biosystems Inc., Foster City, USA). The thermal cycler protocol was initiated at 95 ${}^{\circ}$ C for five minutes, followed by forty-five cycles at 95 ${}^{\circ}$ C for fifteen seconds, and finally at 60 ${}^{\circ}$ C for one minute.

We adapted Cycle threshold (Ct) values derived from qPCR to calculate the number of starting amplicons, or PLA units by computing 10 ${}^{(-0.301\times\text{Ct}+11.439)}$ as previously reported [14]. After computing PLA units, all data were transformed from linear to log10 scale to reduce non-normality in distribution.

2.1 Statistical analysis

Descriptive statistics were used to illustrate demographic data. Additionally, we applied the Student’s $t$ test for comparing continuous variables between cancer patients and matched healthy controls. Furthermore, we used the Chi-square test of Fisher’s exact test to investigate nominal or ordinal variables between cancer patients and the matched healthy control group. Moreover, a Receiver Operating Characteristic (ROC) curve was used to detect an appropriate cut-off point for the Ct value of E-cadherin at which participants could be separated into two subgroups. Finally, we applied the Kaplan-Meier method to analyze disease-specific survival. The statistical differences of disease-specific survival between the subgroups were investigated by the log-rank test. All statistical works were examined by SPSS for Windows, version 12.1 (SPSS, Chicago, IL). A $P$ value of less than 0.05 was regarded as statistically significant.

3. Results

From April 2015 until July 2018, a total of 129 participants were enrolled in the current study. Amongst them, 74 participants were identified as oral cavity cancer patients, while an additional 55 participants were matched as healthy participants who had come in to our clinic due to non-cancer related issues. The average age of the participants was 54.6 $\pm$ 11.2 years. With regards to gender, males accounted for 86.8% ( $n=$ 112) of all participants. In the oral cavity cancer patients, the most common primary site was the tongue, comprising more than half of the patients ( $n=$ 39, 52.7%), followed by the buccal mucosa ( $n=$ 18, 24.3%), gum ( $n=$ 8, 6.1%), and floor of mouth ( $n=$ 5, 6.8%). Sixteen patients (21.6%) were classified as stage I diseases, while 16 (21.6%), 15 (20.3%), and 27 (36.5%) patients were categorized as stage II, III, and IV diseases, respectively. Sixteen patients (21.6%) died and 15 patients (20.2%) developed local recurrence during the course of follow-up. Additionally, 10 patients (13.5%) experienced cervical lymph node recurrence, while no patients experienced distant metastasis. The average follow-up duration was 28.0 months ( $\pm$ 12.2 months).

Table 2
Descriptive and bivariate analysis in oral cavity cancer patients based on pathological stage

Variables	Total no. of patients (% in column)	No. of patients (%)		$P$ value
		Stage I & II ( $n=$ 32)	Stage III & IV ( $n=$ 42)
Age at presentation (year)		56.6 $\pm$ 10.6	53.7 $\pm$ 9.5	0.229
E-cadherin (Ct value)		20.81 $\pm$ 0.29	20.63 $\pm$ 0.43	0.036
Follow-up period (months)		29.1 $\pm$ 12.4	27.1 $\pm$ 12.1	0.491
Gender (female/male)	9/65	4/28	5/37	0.999 ${}^{{\dagger}}$
Smoking
Yes	59 (79.7%)	25 (42.4%)	34 (57.6%)	0.764
No	15 (20.3%)	7 (46.7%)	8 (53.3%)
Pack-year		29.3 $\pm$ 23.7	27.1 $\pm$ 26.2	0.706
Alcohol
Heavy	26 (35.1%)	11 (42.3%)	15 (57.7%)	0.990
Social	30 (40.5%)	13 (43.3%)	17 (56.7%)
No	18 (24.3%)	8 (44.4%)	10 (55.6%)
Betel quid
Yes	56 (75.7%)	26 (46.4%)	30 (53.6%)	0.483
No	18 (24.3%)	6 (33.3%)	12 (66.7%)
Quid-year		358 $\pm$ 286	267 $\pm$ 277	0.173
Perineural invasion
Yes	25 (33.8%)	5 (20.0%)	20 (80.0%)	0.019
No	49 (66.2%)	27 (55.1%)	22 (44.9%)
Angiolymphatic invasion
Yes	11 (14.9%)	1 (9.1%)	10 (90.9%)	0.013 ${}^{{\dagger}}$
No	63 (85.1%)	31 (49.2%)	32 (50.8%)
Extranodal extension
Yes	6 (8.1%)	0 (0%)	6 (100%)	0.033 ${}^{{\dagger}}$
No	68 (91.9%)	32 (47.1%)	36 (52.9%)
Local recurrence
Yes	15 (20.3%)	4 (26.7%)	11 (73.3%)	0.246
No	59 (79.7%)	28 (47.5%)	31 (52.5%)
Survival status
Alive	58 (78.4%)	30 (51.7%)	28 (48.3%)	0.012
Dead	16 (21.6%)	2 (12.5%)	14 (87.5%)

${}^{{\dagger}}$ Fisher’s exact test.

In bivariate analysis, there was no significant statistical difference in demographic data at presentation between the oral cavity cancer patients and non-cancerous participants. Nevertheless, the Ct value of E-cadherin was lower in oral cavity cancer patients when compared with that of non-cancerous participants (20.72 $\pm$ 0.39 versus 21.27 $\pm$ 0.45, $P<$ 0.001). Comprehensive statistics are presented in Table 1. When we divided oral cavity cancer patients according to pathological stage, there were no significant differences between these two groups in age, gender, personal habits, and local recurrence. Conversely, the proportions of peri-neural invasion, lymphovascular invasion, extranodal extension, and cancer-related death were higher in patients with late-stage disease when compared with those of early-stage disease. In addition, the Ct value of E-cadherin was lower in patients with late-stage disease when compared with that in patients with early-stage disease (20.63 $\pm$ 0.43 versus 20.81 $\pm$ 0.29, $P=$ 0.036). Complete information is presented in Table 2.

Figure 1.

Receiver operator characteristic curve for cut-off analysis of E-cadherin Ct value in patients with or without oral cavity cancer. The area under the curve is 0.834.

We used ROC curve analysis to recognize a suitable cut-off value for the Ct value of plasma E-cadherin at which the participants could be divided into two groups (Ct $\geqslant$ 20.9 and Ct $<$ 20.9). The curve was plotted based upon the sensitivity and specificity that the Ct value of E-cadherin could discriminate between participants with and without oral cavity cancer. The longitudinal axial indicates ‘sensitivity’, while the horizontal axial denotes ‘1-specificity’ (Fig. 1). The area under the ROC curve was 0.834 ( $P<$ 0.001). The sensitivity and specificity values were recorded and the cut-off value was nominated once the sensitivity and specificity values both reached acceptably high levels (Table 3). When using a plasma E-cadherin Ct value of 20.9 as the cut-off point, the sensitivity, specificity, positive predictive value, and negative predictive value for distinguishing the oral cavity patients from non-cancerous participants was 63.5%, 87.3%, 87.0%, and 64.0%, respectively. By using the Kaplan-Meier method for survival analysis, the 3-year disease-specific survival rate was comparable in patients with a plasma E-cadherin Ct value of $\geqslant$ 20.9 when compared with that of those with a plasma E-cadherin Ct value of $<$ 20.9 (91.2% vs. 82.6%, $P=$ 0.2307) (Fig. 2).

Table 3

Statistical parameters calculated for different cut-off values and their associations with oral cavity cancer

Cut-off value	20.80	20.85	20.90	20.95	21.00
Sensitivity (%)	51.4	54.1	63.5	67.6	74.3
Specificity (%)	90.9	90.9	87.3	83.6	67.3

Figure 2.

Disease-specific survival of oral cavity cancer patients based on the Ct value of plasma E-cadherin.

4. Discussion

This is the first study which has attempted to discover potential biomarkers for oral cavity cancer using PLA. Our study found that a significant difference existed between oral cavity cancer patients and non-cancerous participants in E-cadherin Ct value from peripheral blood. During conventional immunoassays, it is also crucial to wash out any additional secondary reporter antibodies. The sensitivity of the abovementioned assays can be compromised by the nonspecific combination of the secondary reporter antibody to the external part of the solid support [15]. Previous studies have indicated that PLA may be an encouraging approach for scalable and sensitive investigation of potential biomarkers [16, 17]. Signals from antibody cross-reactive events can be reduced through use of the dual-specific nucleic acid reporter system [17].

E-cadherin is a transmembrane glycoprotein and acts as a calcium-dependent intercellular linkage fragment in epithelial cells [16]. The extracellular portion of E-cadherin is trimmed by proteases to transform into a soluble ectodomain fragment, namely, soluble E-cadherin [19]. A previous study written in Spanish found that the level of soluble E-cadherin in serum was higher in head and neck cancer patients when compared with healthy controls [18]. Fredriksson et al. found that soluble E-cadherin was linked to Human Epidermal Growth factor receptor (HER)/Insulin-like Growth Factor-1 Receptor (IGF-1R). In addition, soluble E-cadherin was shown to initiate downstream Phosphatidylinositol 3-kinase (PI3K)/Akt/mammalian Target of Rapamycin (mTOR) and Mitogen-activated Protein Kinase (MAPK) pathway, which enhanced tumor progression, motility and invasion [17]. Furthermore, soluble E-cadherin can bind Killer cell Lectin-like Receptor G1 (KLRG1), which significantly inhibits not only antigen reactivity and cell proliferation of natural killer cells/T cells, but also the creation of TNF- $\alpha$ and other cytokines by effector immune cells in reaction to antigen stimulation. This effect suggests that soluble E-cadherin is associated with tumor immune escape [19]. These phenomena may explain why our study found higher soluble E-cadherin in patients with oral cavity cancer.

The current study also found that a plasma E-cadherin Ct value of 20.9 was a good cut-off value to discriminate oral cavity cancer patients from non-cancerous participants, with their specificity and sensitivity being 87.3% and 63.5%, respectively. Repetto et al. in their review article indicated that the accumulation of soluble E-cadherin in the blood may be regarded as a sensitive indicator of tumor-associated proteolysis and therefore can be a potential candidate for early cancer diagnosis [22]. A decreased expression of E-cadherin can modify cell to cell adhesion of epithelium and enhance EMT. The aforementioned down-regulation may be caused by decreased gene transcription, mRNA translation, or increased post-translational protein processing. Proteolytic cleavage of transmembrane E-cadherin is a post-translational procedure that decreases membrane E-cadherin levels, which in turns increases soluble E-cadherin [19].

The current work has indicated that the Ct value of soluble E-cadherin was different between early-stage and late-stage oral cavity cancer patients. Okugawa et al. in their report onpatients of colorectal carcinoma also showed that the concentration of soluble E-cadherin was greater in late-stage cancer patients [23]. Soluble E-cadherin was found to be correlated with prognosis of breast cancer [7], esophageal cancer [9], and colorectal cancer [23]. However, our study failed to find such correlations in oral cavity cancer patients. A possible explanation for this is that dissimilar cancer populations may have unique characteristics. In addition, the follow-up duration in the current study was short and survival benefit may appear after a longer surveillance period. Finally, different methods can yield disparate results. As previously noted above, there are important difference between ELISA and PLA [12].

There were some limitations which certainly existed in our study. The external validity of the results is restricted, as the present study was carried out at a single institute. In addition, data regarding the human papilloma virus, which is a significant prognosticator for oral cancer, was not documented. Moreover, immunohistochemical staining of surgical specimens was lacking and no comparison could be made between E-cadherin levels in tumor and peripheral blood. Finally, the follow-up duration was short and a longer observation may be necessary for determining real survival status.

5. Conclusions

Plasma soluble E-cadherin was found to be significantly higher in oral cavity patients when compared with that of non-cancerous participants. Further investigation involving both reliable and sensitive PLA remains warranted in order to discover novel potential candidate protein biomarkers for early diagnosis of oral cavity cancer.

Footnotes

Acknowledgments

The authors thank the Biostatistics Task Force of Taichung Veterans General Hospital for assisting with the statistical analysis. The study was supported by grants from the Ministry of Science and Technology, Taiwan, Republic of China (MOST 104-2314-B-075A-010-MY3).

References

Global Burden of Disease Cancer Collaboration Fitzmaurice

Dicker

et al., The Global Burden of Cancer 2013, JAMA Oncol 1 (2015), 505–527.

Hsu

S.H.

Wong

Y.K.

Wang

C.P.

et al., Survival analysis of patients with oral squamous cell carcinoma with simultaneous second primary tumors, Head Neck 35 (2013), 1801–1807.

Tsai

W.C.

Kung

P.T.

Wang

S.T.

Huang

K.W.

and Liu

S.A.

, Beneficial impact of multidisciplinary team management on the survival in different stages of oral cavity cancer patients: results of a nationwide cohort study in Taiwan, Oral Oncol 51 (2015), 105–111.

Yang

W.H.

Lan

H.Y.

Huang

C.H.

et al., RAC1 activation mediates Twist1-induced cancer cell migration, Nat Cell Biol 14 (2012), 366–374.

Luo

S.L.

Xie

Y.G.

J.H.

and Xu

, E-cadherin expression and prognosis of oral cancer: a meta-analysis, Tumour Biol 35 (2014), 5533–5537.

Foschini

M.P.

Cocchi

Morandi

et al., E-cadherin loss and delta Np73L expression in oral squamous cell carcinomas showing aggressive behavior, Head Neck 30 (2008), 1475–1482.

Liang

Sun

X.Y.

L.C.

and Fu

R.Z.

, Abnormal expression of serum soluble E-cadherin is correlated with clinicopathological features and prognosis of breast cancer, Med Sci Monit 20 (2014), 2776–2782.

Gogali

Charalabopoulos

Zampira

et al., Soluble adhesion molecules E-cadherin, intercellular adhesion molecule-1, and E-selectin as lung cancer biomarkers, Chest 138 (2010), 1173–1179.

Chung

Law

Kwong

D.L.

and Luk

J.M.

, Serum soluble E-cadherin is a potential prognostic marker in esophageal squamous cell carcinoma, Dis Esophagus 24 (2011), 49–55.

10.

Weiss

J.V.

Klein-Scory

Kübler

et al., Soluble E-cadherin as a serum biomarker candidate: elevated levels in patients with late-stage colorectal carcinoma and FAP, Int J Cancer 128 (2011), 1384–1392.

11.

Flanigon

Kamali-Moghaddam

Burbulis

et al., Multiplex protein detection with DNA readout via mass spectrometry, N Biotechnol 30 (2013), 153–158.

12.

Oliveira

F.M.S.

Mereiter

Lönn

et al., Detection of post-translational modifications using solid-phase proximity ligation assay, N Biotechnol 45 (2018), 51–59.

13.

Lundberg

Thorsen

S.B.

Assarsson

et al., Multiplexed homogeneous proximity ligation assays for high-throughput protein biomarker research in serological material, Mol Cell Proteomics 10 (2011), M110.004978.

14.

Chang

S.T.

Zahn

J.M.

Horecka

et al., Identification of a biomarker panel using a multiplex proximity ligation assay improves accuracy of pancreatic cancer diagnosis, J Transl Med 7 (2009), 105.

15.

Fredriksson

Dixon

Koong

A.C.

Mindrinos

and Davis

R.W.

, Multiplexed protein detection by proximity ligation for cancer biomarker validation, Nat Methods 4 (2007), 327–329.

16.

Juncker

Bergeron

Laforte

and Li

, Cross-reactivity in antibody microarrays and multiplexed sandwich assays: shedding light on the dark side of multiplexing, CurrOpin Chem Biol 18 (2014), 29–37.

17.

Fredriksson

Horecka

Brustugun

O.T.

et al., Multiplexed proximity ligation assays to profile putative plasma biomarkers relevant to pancreatic and ovarian cancer, Clin Chem 54 (2008), 582–589.

18.

Zhao

Sun

et al., Is E-cadherin immunoexpression a prognostic factor for head and neck squamous cell carcinoma (HNSCC)? A systematic review and meta-analysis, Oral Oncol 48 (2012), 761–767.

19.

Brouxhon

S.M.

Kyrkanides

Teng

et al., Soluble E-cadherin: a critical oncogene modulating receptor tyrosine kinases, MAPK and PI3K/Akt/mTOR signaling, Oncogene 33 (2014), 225–235.

20.

Al Kassam

Alvarez Marcos

Blanco

de Los Toyos

J.R.

and Llorente

J.L.

, Diagnostic value of E-cadherin, MMP-9, activated MMP-13 and anti-p53 antibodies in squamous cell carcinoma, Med Clin. (Barc) 129 (2007), 761–765. [Article in Spanish].

21.

Q.P.

Kuang

J.Y.

Yang

Q.K.

Bian

X.W.

and Yu

S.C.

, Beyond a tumor suppressor: soluble E-cadherin promotes the progression of cancer, Int J Cancer 138 (2016), 2804–2812.

22.

Repetto

De Paoli

De Re

Canzonieri

and Cannizzaro

, Levels of soluble E-cadherin in breast, gastric, and colorectal cancers, Biomed Res Int 2014 (2014), 408047.

23.

Okugawa

Toiyama

Inoue

et al., Clinical significance of serum soluble E-cadherin in colorectal carcinoma, J Surg Res 175 (2012), e67–73.