Journal of Cancer Research and Therapeutics

: 2020  |  Volume : 16  |  Issue : 3  |  Page : 410--424

Genetic and proteomic biomarkers of head-and-neck cancer: A systematic review

David Kasradze1, Gintaras Juodzbalys1, Zygimantas Guobis2, Albinas Gervickas1, Marco Cicciù3,  
1 Department of Maxillofacial Surgery, Lithuanian University of Health Sciences, Kaunas, Lithuania
2 Department of Dental and Oral Diseases, Lithuanian University of Health Sciences, Kaunas, Lithuania
3 Department of Biomedical and Dental Sciences, School of Dentistry, University of Messina, Messina, Italy

Correspondence Address:
Marco Cicciù
Department of Biomedical and Dental Sciences, School of Dentistry, University of Messina, Policlinico G. Martino, Via Consolare Valeria, 98100 Messina


Development of human genetic and proteomic research has increased the interest in alternative head-and-neck cancer (HNC) detection methods. The aim of this article, the second of two-part series, was to review the scientific literature about novel HNC genetic and proteomic biomarkers. A comprehensive review of the current literature was conducted according to the Preferred Reporting Item for Systematic Review and Meta-analyses guidelines by accessing the NCBI PubMed database. Authors conducted the search of articles in English language published from 2004 to 2015. A total of 50 relevant studies were included in the review. Thirty of them concerned proteomic and twenty genetic alterations in HNC. The present systematic review discovered 242 genes and 44 proteins associated with HNC. Due to inconsistent and sparse results, novel biomarkers cannot be firmly established. Prognostic capacity of genetic markers was not evaluated. Proteins (14-3-3γ, extracellular matrix metalloproteinase inducer, and PA28γ) were described as most valuable for prognostic observation of HNC. A strict methodological protocol for molecular studies must be established.

How to cite this article:
Kasradze D, Juodzbalys G, Guobis Z, Gervickas A, Cicciù M. Genetic and proteomic biomarkers of head-and-neck cancer: A systematic review.J Can Res Ther 2020;16:410-424

How to cite this URL:
Kasradze D, Juodzbalys G, Guobis Z, Gervickas A, Cicciù M. Genetic and proteomic biomarkers of head-and-neck cancer: A systematic review. J Can Res Ther [serial online] 2020 [cited 2020 Aug 3 ];16:410-424
Available from:

Full Text


Currently, the main methods of cancer diagnosis are histopathological tests of biopsy samples and blood tests. Although it is the golden standard, in the last 10–15 years, the interest in more advanced diagnostic methods has considerably increased.[1] With the development of understanding of human genome, the appliance of genetic and molecular alterations as diagnostic and prognostic tools is increasing.

Molecular detection tools can be divided into nucleic acid-based and protein-based markers. Nucleic acid-based alterations occur due to preceding epigenetic processes or existing genetic mutations, amplifications, and polymorphisms. These mechanisms lead to aberrant expressions of genes. Unlike nucleic acid-based techniques, protein-based early detection tools detect posttranscriptional and posttranslational changes that may take place as a result of tumorigenesis. The epigenetic involvement in tumorigenesis of head-and-neck cancer (HNC) was discussed in the first part of this review.[2],[3],[4],[5],[6],[7],[8],[9],[10],[11],[12],[13],[14]

In this article, authors will discuss genetic and molecular pathways of carcinogenesis, which can serve as useful prognostic and diagnostic HNC markers.


This study aimed to overview the most recent literature on diagnostic and prognostic possibilities of genetic and proteomic biomarkers of HNC. To make it easier for readers, we presented the following important aspects: genetic alterations and proteomic changes in human HNC.

 Materials and Methods

Protocol and registration

The methods of the analysis and inclusion criteria were specified in advance and documented in a protocol. The review was registered in PROSPERO, an international prospective register of systematic reviews. The protocol can be assessed at: ID=CRD42015026821.

Registration number: CRD42015026821.

The reporting of this systematic analysis adhered to the Preferred Reporting Item for Systematic Review and Meta-analyses (PRISMA) Statement (Moher et al., 2009).

Types of publications

The review included studies on humans published in the English language. Letters, editorials, literature reviews, PhD theses, and abstracts were excluded.

Types of studies

The review included all human prospective and retrospective follow-up studies, clinical trials, cohort studies, case–control studies, and case series studies published between January 1, 2004, and August 31, 2015, on various genetic and proteomic alterations occurring in HNC tissues and various evaluation methods to ensure the sensitivity of the systematic review.

Information sources

The search strategy incorporated the examination of electronic databases, supplemented by hand searches. A search was conducted on the National Library of Medicine database (MEDLINE) through its online site (PubMed). The references of each relevant study were screened to discover additional relevant publications and to improve the sensitivity of the search.


The PubMed resource database was explored through advanced searches. The keywords and search inquiries used during the primary stage were as follows: “Head and neck cancer gene expression” and “Head and neck cancer protein biomarkers.” Additional manually selected articles were included regarding the eligibility criteria. A flow diagram of selected studies according to PRISMA guidelines is summarized in [Figure 1].{Figure 1}

Selection of studies

The resulting articles were independently subjected to clear inclusion and exclusion criteria by three reviewers as follows: reviewers compared decisions and resolved differences through discussion, consulting a third party when consensus could not be reached. The third party was an experienced senior reviewer. At the title and abstract stage, two reviewers accepted the citations that appeared to meet inclusion criteria and send them on to full-text review, with a third reviewer assessing only those citations and abstracts that the first reviewer deemed ineligible. For the stage of reviewing of full-text articles, a complete independent dual review was undertaken.

Inclusion and exclusion criteria

The applied inclusion criteria for the studies were as follows:

English languageIn vivo/in vitro studiesHuman studies of HNC (upper respiratory system, oral cavity, esophageal, thyroid, and nasopharyngeal cancers).

The following types of articles were excluded as follows:

Studies of testing medication and/or new treatment methodologies-Studies of cancer in locations other than mentionedStudies not relevant to our selected diagnostic methodsAnimal studiesLiterature reviews.

Sequential search strategy

Following the initial literature search, all article titles were screened to eliminate irrelevant publications, review articles, case reports, and animal studies. Next, studies were excluded based on data obtained from screening the abstracts. The final stage of screening involved reading the full texts to confirm each study's eligibility based on the inclusion and exclusion criteria.

Data extraction

The studies, which satisfied the inclusion criteria, were categorized into gene expression assessment and protein expression assessment. Data were independently extracted from reports in the form of variables according to the aim and themes of present review as listed onward.

Data items

Data were collected from the included articles and arranged in the following fields:

“Year” – Reveals the year of publication“Sample origin” – Describes the number of particularly investigated samples in the study and its origin (e.g., BS – blood sample, saliva sample, and tumor tissue)“Subjects” – Describes the number of patients examined“Control” – Describes the quantity of control samples that were used in a trial“Result” – Indicates the parameters that were coherent with alterations of particular biomarkers in prognostic studies“P value” – Indicates the chosen value of significance by authors for particular biomarker or biomarker panel.

Assessment of methodological quality

The quality of all included studies was assessed during the data extraction process. The quality appraisal involved evaluating the methodological elements that might influence the outcomes of each study. The Cochrane Collaboration's 2-part tool for assessing the risk of bias (Higgins and Green, 2011) was used to assess bias across the studies and identify articles with intrinsic methodological and design flaws.[52] Risk of bias (e.g., lack of information or selective reports on variables of interest) was assessed on study level. The risks were indicated as lack of precise information of interest in each individual study that can blind the reader from particular information about examined samples, particularly “Sequence generation,” “Selective reporting,” “Allocation concealment,” and “Outcome data completeness.”

Synthesis of results

Relevant data of interest according to variables stated previously were collected and organized in three tables that were divided according to the molecular type of assessment and type of study. The tables include results according to the evaluation of qualitative and quantitative alterations of biomarkers.

Statistical analysis

No meta-analyses could be performed due to the heterogeneity between the studies (different study designs, control groups, and observation periods).


Study selection

The initial search provided a total of 23,364 articles. After activation of search filters, 327 available articles were identified. Independent screening of the abstracts resulted in the selection of thirty publications for possible inclusion. The inclusion and exclusion criteria were applied to the 263 full-text articles. In addition, nine manually obtained publications regarding eligibility were identified. Finally, 48 articles that met predefined criteria were included in the systematic review [Figure 1].

Exclusion of studies

The reasons for excluding studies after full-text assessment were as follows: during preliminary exclusion, 74 articles were excluded concerning irrelevancy of article titles and abstracts. Finally, articles that did not meet the inclusion and exclusion criteria were filtered as follows: Not HNC trials (n = 77) and testing of medication efficacy and/or treatment methods (n = 147). Additional manual selection according to eligibility was performed (n = 9). The data were obtained from 8436 samples.

Quality assessment

Summarized quality assessment of all included studies revealed a unclear risk of bias (for one or more key domains) for majority of studies (Kondoh et al., 2007; Ziober et al., 2006; Jacques et al., 2013; Belbin et al., 2005; Roepman et al., 2005; Nguyen et al., 2007; Zhu et al., 2012; Tsai et al., 2011; Canova et al., 2009; Poeta et al., 2007; Akdi et al., 2010; Hirano et al., 2009; Howell et al., 2009; Ku et al., 2005; Bertonha et al., 2014; Zhi et al., 2014; Korostoff et al., 2011; Brailo et al., 2012; Sato et al., 2010; AbdulWahab et al., 2011; SahebJamee et al., 2008; Wu et al., 2013; Gurzu et al., 2012; Liang et al., 2009; Kong et al., 2015; Araujo et al., 2008; Chung et al., 2011; Filho et al., 2009; Ge et al., 2009; Gourin et al., 2007; Monteiro et al., 2014; Goulioumis et al., 2009; Aquino et al., 2013; Zhu et al., 2013; Cao et al., 2011; Mega et al., 2005; Nankivell et al., 2013; Seibold et al., 2013; Ong et al., 2013; Snietura et al., 2012; Patel et al., 2008; Franzmann et al., 2007; and Li et al., 2015). Four studies (Zhao et al., 2008; Kainuma et al., 2006; Roepman et al., 2006; and Cho et al., 2012) were classified as a high risk of bias (for one or more key domains). Only two studies were classified as a low risk of bias (Fenner et al., 2005 and Benchekroun et al., 2010) [Table 1].{Table 1}

Study characteristics

The included studies were further divided into three groups according to the studied molecular level. The division provided a better understanding of effect on cancerogenesis of separate molecular levels and contributed to the sensitivity of the review.

Genetic alterations in head-and-neck cancer

A total of twenty studies with 9421 biological samples were analyzed. Finally, 242 were suggested as significantly altered in HNC. The individual results of them are summarized in [Table 2].{Table 2}

Gene expression alterations occur mainly due to DNA-based changes which are classified as follows: point mutations, gene amplification, fusion, deletion, insertion, or nucleotide polymorphisms. Altered gene expression patterns can be identified before the cancer phenotype has manifested. Twelve of the reviewed studies concerned differences of expression patterns in HNC tissues.[3],[4],[5],[6],[7],[8],[9],[10],[11] The rest eight were investigating the association of predetermined single-gene alterations (single-nucleotide polymorphism [SNP], mutations) with HNC.[12],[13],[14],[15],[16],[17],[18],[19]

cDNA microarrays provide the information on whole transcription activity in a particular sample. This test method allows investigating the genes of a large part of the genome and detecting their unusual and new functions.[2] DNAs in this group of studies were extracted from histologically embedded HNC tumor tissues. Microarrays were separately provided up to 234 aberrantly expressed genes comparing HNC and normal tissues.[8] To approach the most reliable genes, statistical verification was performed in all but two studies.[4],[10] Under- or over-expressed genes are involved in the formation of extracellular matrix (MMP1, MMP3, MMP10, HABP2, LAMC2, MGP, PRELP, DPT, and MFAP4), immune response (interleukin-8 (IL-8), IP-10, and CRIP1), epithelial/epidermal formation (KRT4, KRT13, EPSTI1, C1orfl0, and TG3), secretory functions (FST, PTHLH, TGFBI, USP18, and SLIT3), apoptosis (CLU), transcription factors (TP53, HOXA9, RUNX1, and CCNE1), metabolic pathways (TPO, CaMKIINalpha, CA2, and DPYSL3), and cell-to-cell integrity (SNAI2, MYH2, and LOXL2), as well as genes with undefined function in cancerogenesis.

Authors combined the most reliable genes to establish gene panels. A 11-gene panel (MGP, NR2F2, SLIT3, KRT1, KRT13, TG3, FXYD6, CXCL10, LAM2, FST, and IGJ) distinguished oral squamous cell carcinoma (OSCC) from leukoplakia at 97.8% accuracy. Another set of seven genes (DPT, FXYD6, C1orfl0, TG3, PRELP, FOSB, and IGJ) with an >95% accuracy differentiated mild dysplasia and hyperplasia from higher grade dysplasia and OSCC.[5] The selected 25-gene molecular predictor could classify tumor and normal samples with a 100% accuracy. In addition, this molecular predictor was also accurate (96%) on cross-validation to predict nonoral tumors.[6] The combined 13 genes discriminated thyroid adenomas, oncocytic variants of follicular thyroid tumors, and papillary thyroid carcinomas (P < 0.05).[7]

Four studies applied gene expression patterns in order to distinguish different tumor/node/metastasis (TNM) stage samples.[9],[10],[11],[18] Two separate gene analyses were conducted on samples of nine patients with Stage III or IV OSCC in order to investigate significantly expressed genes in different tumor stages (normal, OSCC, and metastatic tissue samples). After reducing data sets, 140 genes that consistently increased and 94 genes that consistently decreased in expression during tumor progression were identified.[8] Indicated genes, such as glycoprotein tenascin and osteopontin, can inhibit macrophage function and enhance survival of SCC metastases. Arachidonate 5-lipoxygenase is a main enzyme-producing eicosanoid, a substance which play a role in tumor progression and metastasis. Enhanced expression of Stanniocalcin-1 in head-and-neck SCC (HNSCC) is associated with poorer prognosis and aggressiveness of tumor. Other altered genes that were associated with invasiveness of oral SCC were laminin 5, moesin, and erzin. These genes are interfaced with keratinocyte adhesion and membrane-cytoskeletal linking. A variety of genes decreased in expression during tumor progression. Genes involved in keratinization-keratins 4, 13, and 15 as well as occludin, structure-determining transmembrane protein, tumor suppressor genes-BRUSH-1, MXI1, and oxidative stress-response 1 were observed to lose a degree of expression.[8] Similar study concerning differentiation between separate stages of cancer identified 102 predictor genes that have a predictive accuracy for N0 status of 100%, for N+ status of 77%, and overall of 86%. Most of the genes are epithelial marker genes, encoding extracellular matrix components, involved in cell adhesion, having a role in maintaining tissue integrity, cell death, cell growth, and maintenance, and encoding hydrolyzing activities.[10] The later study of the same author accomplished gene expression array on 19 matched primary tumor and metastatic lymph node samples. The only gene that significantly elevated in lymph node samples was metastasis-associated gene 1 (MTA1) (P < 0.05).[9] Another study revealed 85 differently expressed genes comparing OSCC with neck metastases. To minimize the predictive error, 8 genes were selected – DCTC, IL-15, THBD, GSDML, SH3GL3, PTHLH, RP5 = 1022P6, and C9orf46 for further investigation. The gene panel accurately predicted the metastases in 12 out of 13 samples (P < 0.04). IL-15 and PTHLH genes showed the best P values: P < 0.00711 and P < 0.022891, respectively. The IL-15 plays an important role in cell proliferation, apoptosis, invasion, and metastasis. PTHLH encodes the protein of the parathyroid hormone family.[11] The expression of 6 tetraspanin family genes (CD9, CD63, CD81, CD82, CD151, and NAG-2) in 73 cases of gingival SCC was analyzed. Results revealed that CD9+ACTB (P = 0.013) and CD9+CD82 (P = 0.013) correlated with cervical lymph node metastasis and CD151+GAPDH (P = 0.024) with death outcome. Tetraspanins affect cell adhesion, signal transduction, cell proliferation, cell movement, cell differentiation, activation of immunocytes, cell fertilization, and viral infection.[18]

A SNP is a DNA sequence variation occurring when a single nucleotide (A, T, C, or G) in the genome differs from the normally expected nucleotide. These SNPs are known to underlie differences in our susceptibility to diseases. SNPs need to be determined only once and are easy to determine, making them interesting biomarkers. Polymorphisms occurring in genes involved in DNA repair, cell cycle, and molecular pathways may alter the genomic stability. Results described several genes that are significantly associated with different locations of HNC. One of the SNPs in insulin-like growth factor receptor 1IGF1R (rs2229765, Glu1043Glu) was significantly associated with papillary thyroid carcinoma (P < 0.05). IGF1R is highly overexpressed in the most malignant tissues and plays a central role in cell cycle progression and transformation of the cells.[12] SNP of the excision repair cross complementing group 5 gene (rs2296147) was less frequent among cases than among controls (P = 0.025) and associated with a significantly decreased risk of SCC (P = 0.006).[13] ERCC5 is a DNA repair gene which plays role in the initiation of carcinogenesis and its deficiency leads to DNA repair defects, genomic instability, and failure of gene transcription modulation. Cyclin D1 (CCND1) plays a critical role in the G1 to S phase transition of the cell cycle. Thus, CCND1 is a commonly observed gene in human carcinomas and frequently an overexpression of CCND1 has been reported as a potential biomarker.[13] Significant differences were shown between the oral cancer and control groups in the distribution of the genotypes (P = 0.0014) and allelic frequency (P = 0.0027) in the CCND1 rs9344 genotype. Individuals who carried at least one G allele (GG or AG) had a 0.64-fold decreased risk of developing oral cancer compared to those who carried the AA wild-type genotype (95% confidence interval [CI]: 0.50–0.81). CNND1 A870G was significantly different between the oral cancer patients and the controls who had a smoking habit (P = 0.0006).[14] In another case–control study, 115 SNPs from 62 a priori-selected genes were studied in relation to upper aerodigestive tract (UADT) cancer. ERCC1 (rs3212961) (odds ratio [OR]: 0.45; 95% CI: 0.23–0.90) and ERCC4 (rs1799801) (OR: 0.83; 95% CI: 0.70–0.98) reduced cancer risk, and CDKN1A (rs2395655) and GASC1 genes were associated with an increased risk of UADT cancer (P = 0.045 and P = 0.008, respectively). Three SNPs in the MDM2 gene, involved in cell cycle control, were associated with UADT cancer. The rare variant allele of CYP2A6 (rs28399433) was associated with a reduced risk for UADT cancer (P = 0.01), whereas the rare variant allele of CYP2C8 (rs1934951) was found to be associated with an increased risk of UADT (P = 0.02).[15] There was a statistically significant difference in the distribution of the vascular endothelial growth factor (VEGF) gene C/T polymorphism between normal controls and oral cancer patients (P < 0.001). VEGF is a potent inducer of endothelial cell growth and its levels are elevated in several tumor types. VEGF, a cytokine that plays an important role in the neovascularization, is an angiogenic factor.[19] Haplotype analysis revealed that combination of certain WDR3 variants, such as haplotype CAT, increases the risk of thyroid cancer (P < 0.063).[17] WDR3 is involved in cell cycle progression and signal transduction, among other cellular processes. Single of the reviewed studies concerned the effect of tumor gene mutations to development of HNC. Tumor protein p53 is involved in apoptosis and cell cycle control. Its contribution to various cancers has been described earlier. Mutation of p53-encoding gene (TP53) was observed in 53.3% of HNSCC patients. This particular study was carried out only to provide results concerning the prognostic values of genetic biomarkers. Mutation status of TP53 was associated with decreased overall survival (P = 0.009) and disruptive mutations (P < 0.001), leading to hypothesis of TP53 gene being a prognostic factor of HNSCC.[16]

Proteomic changes of head-and-neck cancer

A total of thirty studies, in which samples were analyzed, described 52 significant protein biomarkers. The individual results of them are summarized in [Table 3]. Included studies concerned the HNCs of defined localizations: upper respiratory tract, esophagus, oral cavity, thyroid gland, and nasopharynx. Studies involved evaluated the aberrant expressions of candidate protein biomarkers and quantitative yield of them in specimens. Overall, 3608 HNC samples (tumor tissue, blood, and saliva) were examined. The follow-up for prognostic studies (n = 13) ranged from 15 days to 8 years.{Table 3}

Genetic and epigenetic alterations lead to dysregulation of various proteins. The proteins are essential for normal cell mechanisms and signaling. Aberrant expressions of candidate proteins affect cell division, differentiation, immune recognition, tissue invasion, and metastasis. Specific patterns of protein expression or individual proteins have been established as biomarkers for cancer diagnostics and prognostics.[20],[21],[22],[23],[24],[25],[26],[27],[28],[29],[30],[31],[32],[33],[34],[35]

Two of the reviewed studies used mass spectroscopy to analyze proteome-wide alterations in HNC.[36],[49] Protein peak detection was used to analyze the protein mass differences between HNC and control BSs. The analysis classified patients with HNSCC with 82% sensitivity and 76% specificity. Subgroup analysis correctly classified 83% of the oral cavity tumors, 81% of oropharyngeal tumors, and 88% of laryngeal tumors.[36] In a proteome-wide array comparing normal oral tissue with HNSCC, 127 proteins were identified only in HNSCC tissues.[49] A vast number of identified proteins were involved in cell migration, signaling, and proteolysis. Among them, the most abundant was vimentin, protein involved in epithelial–mesenchymal transition. In well-differentiated tumor tissues, proteins of interest included those involved in Notch signaling pathway, function of cytoskeleton (Delta 4, Delta 1). In moderately differentiated tumor tissues, proteins of interest included those involved in Wilms' tumor-associated protein, esophageal cancer-related gene coding leucine-zipper motif, unusual cadherins, and desmosomal proteins for epithelial cells. Interesting proteins detected only the PD group which include the potential oncoprotein AF1q (17) and numerous proteins involved in cell cycle control and fatty acid metabolism and membrane trafficking.

Twenty-three of the reviewed studies accomplished proteomic identification on a priori literature-derived cancer-associated biomarkers. Studies identified 44 HNC-specific protein markers. Described proteins are involved in cell proliferation (STAT3, Repp86, SPARC, EGFR, CD151, CD9, VEGF, and SIRT2), cell cycle control (p53, PCNA, Ki-67, E2F1, Cyclin D1, and PTEN), transcriptional repression (GLI3, Enhancer of zeste homolog 2 (EZH2), ER-ß, SNAI1, and ZEB2), differentiation (extracellular matrix metalloproteinase inducer[EMMPRIN], LAMC2, and TWIST1), cell adhesion (E-CAD, CD44, and delta 1), inflammatory response (Annexin A1, COX-2), binding of signaling proteins (14-3-3 γ), regulation of actin cytoskeleton (MENA), processing of tumor-associated proteins (TACE), and epithelial–mesenchymal transition (Vimentin). Twenty studies used immunohistochemical evaluation of samples. For quantitative results, two studies conducted enzyme-linked immunosorbent assay (ELISA) and one study conducted automated quantitative protein expression analysis (automated quantitative analysis [AQUA]).[32],[33],[34],[35]

Results indicated the correlation of protein expression with regional metastasis, distant metastasis, clinical stage,[27],[38] histopathological grading,[36],[37],[38],[39],[40],[41],[42] size,[43] and localization.[29],[36] Protein markers could classify premalignant, benign, and malignant samples of oral cavity, salivary gland, and thyroid tumors with specificity up to 91.5%.[28],[29],[44] Several of the identified HNC-specific proteomic markers have well-described involvement in tumorigenesis. 14-3-3γ is a member of 14-3-3 proteins, a family of highly conserved phosphoserine/threonine-binding proteins that regulates diverse cellular processes including cell cycle progression, apoptosis, transcriptional regulation, and cell proliferation by functioning as chaperones and adaptor.[27] Mena (mammalian Ena) proteins play a crucial role in cell motility by antagonizing actin filament capping and contribute to the promotion of metastasis.[28] Glioma-associated oncogene family zinc finger GLI1 generally function as a transcriptional activator and mediates the Hedgehog signaling pathway.[32] Repp86 (restrictedly expressed proliferation-associated protein) is a cell cycle-associated protein, and CD44 antigen is a cell surface glycoprotein involved in cell–cell interactions, cell adhesion, and cell migration.[45],[46],[47],[48],[49],[50] EZH2 is the catalytic subunit of polycomb repressive complex 2, a highly conserved histone methyltransferase that methylates lysine 27 of histone H3 (H3-K27). H3-K27 methylation is commonly associated with DNA methylation and silencing of genes which are responsible for differentiation in organisms, in that way showing the possible epigenetic-genetic-proteomic pattern of cancerogenesis.[42] EMMPRIN transmembrane protein is discovered by its capacity of inducing the expression of matrix metalloproteinases. EMMPRIN contributes to cell adhesion modulation, tumor growth, invasion, and angiogenesis.[30]

Prognostic capacities of protein biomarkers were discussed as well. Specific HNC proteins correlated with overall survival, progression, and metastasis-free, disease-free, and cancer-specific survival.[43],[44],[45],[51] Previously described 14-3-3γ, EMMPRIN, and proteasome activator PA28γ have shown the most reliable prognostic values.

Cytokine-level changes in saliva

A total of five studies concerned the protein concentration alterations in HNC patients. Studies were overviewed in order to test the diagnostic capacities of saliva as there is a clinical need for more reliable, accessible, and cost-effective diagnostic sampling in HNC. Saliva is one of the reflections of human health and reserve system resources, which reach the oral cavity by different ways. Steroids, amines, and peptides, such as melatonin, insulin, and leptin, reach the saliva by passive diffusion. Immunoglobulins (e.g., IgA) and enzymes, such as amylase and lysozyme, may be synthesized in salivary glands. The quantity of these endogenous materials may be measured quantitatively and their effect may be evaluated. The changes in salivary composition may be used as markers of hormonal, immunological, toxicological, and infectious diseases.[22] One of the methods of early oral cancer diagnostics is the use of biomarker technology. If the phenotype of certain genes is expressed, primary tumor mutations may be found in human fluids, for example, blood or saliva, before the tumor processes occur in the tissues.[23] Saliva provides a noninvasive possibility to observe the changes of disease at the level of emergence, progress, recurrence, and treatment.[24] However, the use of saliva is criticized as a diagnostic matrix because the quantity of its information is not sufficient and it has considerable daily or seasonal fluctuations – these are among the disadvantages that render salivary analysis unreliable.[25] Therefore, supernatant or strictly standardized saliva tests are used.

Changes in the immune, inflammatory, angiogenic, and metabolic systems were detected in saliva and blood as signs of pathology for people who had OSCC. A hypothesis was raised that the activity of cytokines (pro-inflammatory, pro-angiogenetic, and immunoregulators) is caused by SCC and may be controlled by its pathogenic activity.[22] Many molecules (IL-1α, IL-6, IL-8, tumor necrosis factor-alpha (TNF-α), KEGF-a, etc.) are treated as biomarkers, whose quantitative change occurs with OSCC or precancerous forms of this tumor.[26]

Previous studies have determined that ILs-1, 6, and 8 produced by OSCC disrupt the normal physiological mechanism, stimulate the growth and invasion of irregular cells, inhibit the immune system, and prevent the inhibition of the tumor. Chen has determined considerably higher concentration of proteins IL-6 and IL-8 and VEGF, which are found in OSCC tumor cell culture. TNF-α can also act as a potential endogenous mutagen that causes direct damage to DNA by inducing reactive oxygen forms.[22] IL-1 β acts as a carcinogenesis activator by strengthening the activity of chemical carcinogens, which results in reproduction of mutated cells and accumulation of genetic damage.[23] Having evaluated the effect of these biomarkers, studies were carried out, results of which are presented in [Table 4]. Biomarkers were analyzed by measuring the protein level using a human ELISA kit.{Table 4}

Having compared the values of cytokine concentration in saliva among control, risk, and diseased HNC groups, the obtained data showed that the increase of cytokine level corresponds to the tendency of disease occurrence. Cytokine concentrations were considerably higher for patients with endophytic HNC. It confirms that these elements have a correlation with the severity of disease. In all overviewed studies, the increase of IL-6 concentration in the saliva was considerable (P ≤ 0.05) in comparison to healthy persons, patients with leukoplakia, smokers, and alcohol consumers. In researches 1, 4, and 5, the gap of IL-8 is distinct between healthy people and patients, but only the data of researches 1 (P ≤ 0.05) and 4 (P ≤ 0.001) are statistically significant. TNF-α had a statistically significant change in research 4 (P ≤ 0.05). KEGF had a statistically significant change in research 1 (P ≤ 0.05). IL1-α did not have a statistically significant change in any of the researches. In research 2, significant differences in the saliva of leukoplakia patients and the control group were not observed (P ≥ 0.05). In research 2, concentration of IL-1 β in saliva was significantly higher for the oral cancer patients than leukoplakia patients and the control group (P ≤ 0.05).

Although the above biomarkers showed tendencies in monitoring the development and growth of HNC and could be a cost-effective and minimally invasive means in diagnosing HNC, they cannot be a diagnostic tool for quality and accurate detection of HNC due to relatively low sensitivity and specificity.[23]

The increase of the quantity of anti-inflammatory cytokines in the saliva of HNC patients indicates that they are important in the development and growth of this pathology. Increased concentration of cytokines in the saliva of HNC patients occurs due to their activity in the initial phase of response to local inflammation and activation of lymphocytes, which is not specific and cannot be considered to be a diagnostic marker. This is because the increase of these markers may be also observed with frequent inflammatory oral diseases (e.g., gingivitis and periodontitis).[24] Several studies were carried out, which showed a significant increase of IL-6 and IL-8 concentration in the case of periodontitis in comparison with the control group (P < 0.05). Therefore, due to other inflammatory conditions, saliva biomarkers cannot be used for cancer diagnostics because they are not specific. They could be compared to indicators such as erythrocyte sedimentation rate or C-reactive protein.


The purpose of this review was to systematically overview published studies concerning novel genetic and proteomic biomarkers for detection and prognosis of HNC. All genetic studies were methodologically based on cDNA microarray and polymerase chain reaction and used fresh frozen or histologically embedded tumor tissue samples. In spite of heterogeneity observed in used particular array platforms and sample size, studies identified statistically significant 234 gene biomarkers with altered expression. In addition, 8 SNPs and 1 gene mutation were significantly associated with HNC. However, no accurate gene biomarkers could be established yet, as none of the studies used statistical measurements for specificity and sensitivity evaluation. Identified genes were involved in the formation of extracellular matrix (MMP1, MMP3, MMP10, HABP2, LAMC2, MGP, PRELP, DPT, and MFAP4), immune response (IL-8, IP-10, and CRIP1), epithelial/epidermal formation (KRT4, KRT13, EPSTI1, C1orfl0, and TG3), secretory functions (FST, PTHLH, TGFBI, USP18, and SLIT3), apoptosis (CLU), transcription factors (TP53, HOXA9, RUNX1, and CCNE1), metabolic pathways (TPO, CaMKIINalpha, CA2, and DPYSL3), and cell-to-cell integrity (SNAI2, MYH2, and LOXL2). Majority of the HNC-associated genes were with undefined function in cancerogenesis. In addition, gene biomarker alterations were specific to HNC types, localization, and were associated with TNM stage. Only one study conducted proper prognostic evaluation and identified the association of TP53 mutation with the overall survival of patients.[50],[51],[52]

The studies published within author-selected time range did not concern much about genes that have been already clinically applied. Besides study including-criteria focused on diagnostic and possible prognostic biomarkers. Authors excluded 147 studies (55.9%) concerning predictive biomarkers, oncological treatment outcome or post-oncological quality of life evaluation during eligibility stage of screening as not relevant to the chosen topic. Authors assumed that these studies would not provide diagnostic information of a biomarker, rather evaluate treatment method applicability. In addition excluded studies mainly concerned on already clinically applied biomarkers. These factors had an impact on the fact that moderate amount of specific biomarkers were discovered and scientifically approved HNC biomarkers were only sparsely mentioned. On the other hand, this review encompassed potential biomarkers accordingly with related scientific published data. A structure of this review, as it is a two-part review series, may be another factor for paradoxical results. Some of the expected HNC biomarkers were mentioned in the first part (NOTCH1, NOTCH3, Ki67, Cyclin D1, p16, VEGF, E-Cadherin, and CD44) as they were assorted as epigenetic biomarkers.[53] Combining the results of both parts, it is evident that more genetic mechanisms are involved in HNC genesis such as cell cycle regulation and DNA repair. In addition, this review did not concern solely on one type of neoplasm, instead choosing a wider specter of oncological disease types and localizations. This circumstance resulted in broad amount of functionally and clinically different potential biomarkers useful for several medical fields.

Proteomic studies were conducted on whole proteome in two studies using mass spectroscopy, whereas particular protein studies were based on immunohistochemistry in twenty studies, ELISA in two, and AQUA in one. Proteomic changes identify proteins that have inadequate expression in diverse cellular processes including cell cycle progression, apoptosis, transcriptional regulation, cell proliferation, cell differentiation, tumor cell invasion, cell communication, cell adhesion, and cancer cell growth. Proteomic biomarkers specifically distinguish HNC from normal oral mucosa. Besides, individual markers correlated with different primary head-and-neck sites, tumor differentiation level, and type of malignancy.

Four studies evaluated the prognostic capacities of protein markers. Specific HNC proteins correlated with overall survival, progression, and metastasis-free, disease-free, and cancer-specific survival. Proteins, namely 14-3-3γ, EMMPRIN, and PA28γ, have shown the most reliable prognostic values. However, none of the novel biomarkers were identified in separate studies, which leaves a big risk of detection and publication bias.

Molecular changes of protein concentrations are associated with HNC characteristics. Elevation of the pro-inflammatory proangiogenic cytokines such as IL-1a, IL-6, IL-8, TNF-a, and VEGF-a as potentially important angiogenic factors in tongue squamous cell carcinoma, OSCC, and nasopharyngeal carcinoma in salivary and BSs correlates with increased metastasis and poor prognosis, development of HNC, leukoplakia, and regulation of a wide spectrum of biological processes including cell proliferation, differentiation, apoptosis, lipid metabolism, and coagulation.[30]

The main shortcoming of salivary biomarkers is that concentrations of salivary cytokines might be the result of a lesion with epithelial discontinuity and surrounding inflammation, not directly related to cancer. Moreover, data indicate that altered cytokine production and responsiveness in oral cancer takes place primarily in the oral cavity and does not reflect on serum cytokine concentrations. However, with proper differential assays, salivary biomarkers have the potential to serve as noninvasive, widely available screening tools that do not rely on the localization of a lesion for diagnosis. This advantage over other detection methods gives salivary biomarker screening the potential to identify patients with premalignant lesions. Collection is inexpensive and can be performed in any setting without the assistance of a health-care practitioner. Studies show that immunological and biological markers may be associated with the pathogenesis of oral premalignant and malignant tumors. However no specificity to HNC can yet be established.

The complexity of oncological processes is perceived with more trials conducted. This increases the necessity to inspect more accurate and related molecular pathways. The review did not discover common genetic-proteomic biomarkers with equally significant specificity and sensitivity. Despite this, mentioned genes and proteins are involved in common pathways that minister to understand the basic head-and-neck cancerogenic network. The mentioned strict screening protocol and characteristics of reviewed trials influenced on the lack of common biomarkers. Nevertheless, combining the results of both parts of the review, it is evident that there is an intermolecular pathway. Significant promoter methylation expression changes of CDKN2A, CCND2, and CDH1 genes match with the overexpression of respective genes and proteins such as P16, cyclin D1-D2, and E-Cadherin. Moreover, methylation of histone H3K9 is considered to suppress the expression of CDH1 and p16 genes, whereas downregulation of miRNA miR-320 negatively regulates NRP1 expression which promotes proliferation, survival, and migration/invasion of cancer cells and endothelial cells in various tumor types as a co-receptor with diverse ligands and receptors, including the VEGF that has been upregulated in oral cancer specimens. These scarce accordance implicate about intermolecular head-and-neck cancerogenic network. To obtain streamlined results, multilevel molecular assays must be conducted. None of the reviews investigated whole epigenetic-genetic-proteomic relation of chosen biomarkers. Nonetheless, identified potential biomarkers of different molecular levels are involved in same functions: Cell cycle control, cell proliferation, cell adhesion, cell differentiation, cell-to-cell integrity, immune response, cell signaling, and metabolic pathways.


The study can be limited due to the same reasons as the first part of this two-series article. Most of the studies revealed unclear risk of bias; four studies were classified as a high risk of bias (for one or more key domains) and only one study was at a low risk of bias. Only nine studies provided detailed data on randomization methods and only four described allocation concealment. These limitations increase the probability of detection and selection bias. Randomization is essential for minimizing selection bias; however practically, it is constrained due to the low number of specific HNC patients. It is complicated to obtain randomized patient groups and maintain optimal number of patients at the same time. Thirteen studies reported on double or partial blinding. Researchers blinding from the clinicopathological parameters and outcome evaluations is the way to minimize detection bias. The reliability of certain studies is decreased because of the low number of the studied patients. The number of patients in separate studies varies from 7 to 1511 in genetic and from 12 to 666 in proteomic studies. Besides, a number of studies did not use healthy specimens to obtain specific results, instead established mean expression values. In many studies, descriptions of sample collecting and processing reports were scarce and incomplete, which leads to increased risk of reporting bias. In addition, studies that evaluated the diagnostic potentials of biomarkers did not report on specificity and/or sensitivity.

Practical use of the overviewed genetic and proteomic test methods has so far been an auxiliary means, due to insufficient reliability, specificity, and sensitivity of biomarkers; test results cannot be used for final diagnosis or prescription of initial treatment yet.


The present systematic review discovered 242 genes and 44 proteins associated with HNC. Due to inconsistent and sparse results, novel diagnostic biomarkers cannot be firmly established. True capacities of molecular diagnostic methods are not yet utilized. In proteomic studies, three reliable prognostic proteins were described (14-3-3γ, EMMPRIN, and PA28γ), whereas no prognostic evaluation of gene expression patterns was conducted. Diversities in study design (control group, array platform, and sample origin) and unsuitable risk levels of selection, detection, and reporting bias are the main shortcomings. Strict protocol for molecular studies must be established. Prognostic values of separate biomarkers are ambiguous. No established standards for molecular assay of HNC were found in order to elude the paradoxical results and discrepancies in separate trials.

Financial support and sponsorship


Conflicts of interest

There are no conflicts of interest.


1Liu J, Duan Y. Saliva: A potential media for disease diagnostics and monitoring. Oral Oncol 2012;48:569-77.
2Viet CT, Schmidt BL. Understanding oral cancer in the genome era. Head Neck 2010;32:1246-68.
3Zhao M, Begum S, Ha PK, Westra W, Califano J. Downregulation of RAD17 in head and neck cancer. Head Neck 2008;30:35-42.
4Kainuma K, Katsuno S, Hashimoto S, Oguchi T, Suzuki N, Asamura K,et al. Differences in the expression of genes between normal tissue and squamous cell carcinomas of head and neck using cancer-related gene cDNA microarray. Acta Otolaryngol 2006;126:967-74.
5Kondoh N, Ohkura S, Arai M, Hada A, Ishikawa T, Yamazaki Y,et al. Gene expression signatures that can discriminate oral leukoplakia subtypes and squamous cell carcinoma. Oral Oncol 2007;43:455-62.
6Ziober AF, Patel KR, Alawi F, Gimotty P, Weber RS, Feldman MM,et al. Identification of a gene signature for rapid screening of oral squamous cell carcinoma. Clin Cancer Res 2006;12:5960-71.
7Jacques C, Guillotin D, Fontaine JF, Franc B, Mirebeau-Prunier D, Fleury A,et al. DNA microarray and miRNA analyses reinforce the classification of follicular thyroid tumors. J Clin Endocrinol Metab 2013;98:E981-9.
8Belbin TJ, Singh B, Smith RV, Socci ND, Wreesmann VB, Sanchez-Carbayo M,et al. Molecular profiling of tumor progression in head and neck cancer. Arch Otolaryngol Head Neck Surg 2005;131:10-8.
9Roepman P, de Jager A, Groot Koerkamp MJ, Kummer JA, Slootweg PJ, Holstege FC,et al. Maintenance of head and neck tumor gene expression profiles upon lymph node metastasis. Cancer Res 2006;66:11110-4.
10Roepman P, Wessels LF, Kettelarij N, Kemmeren P, Miles AJ, Lijnzaad P,et al. An expression profile for diagnosis of lymph node metastases from primary head and neck squamous cell carcinomas. Nat Genet 2005;37:182-6.
11Nguyen ST, Hasegawa S, Tsuda H, Tomioka H, Ushijima M, Noda M,et al. Identification of a predictive gene expression signature of cervical lymph node metastasis in oral squamous cell carcinoma. Cancer Sci 2007;98:740-6.
12Cho SH, Kim SK, Kwon E, Park HJ, Kwon KH, Chung JH,et al. Polymorphism of IGF1R is associated with papillary thyroid carcinoma in a Korean population. J Interferon Cytokine Res 2012;32:401-6.
13Zhu ML, Shi TY, Hu HC, He J, Wang M, Jin L,et al. Polymorphisms in the ERCC5 gene and risk of esophageal squamous cell carcinoma (ESCC) in Eastern Chinese populations. PLoS One 2012;7:e41500.
14Tsai MH, Tsai CW, Tsou YA, Hua CH, Hsu CF, Bau DT,et al. Significant association of cyclin D1 single nucleotide polymorphisms with oral cancer in Taiwan. Anticancer Res 2011;31:227-31.
15Canova C, Hashibe M, Simonato L, Nelis M, Metspalu A, Lagiou P,et al. Genetic associations of 115 polymorphisms with cancers of the upper aerodigestive tract across 10 European countries: The ARCAGE project. Cancer Res 2009;69:2956-65.
16Poeta ML, Manola J, Goldwasser MA, Forastiere A, Benoit N, Califano JA,et al. TP53 mutations and survival in squamous-cell carcinoma of the head and neck. N Engl J Med 2007;357:2552-61.
17Akdi A, Giménez EM, García-Quispes W, Pastor S, Castell J, Biarnés J,et al. WDR3 gene haplotype is associated with thyroid cancer risk in a Spanish population. Thyroid 2010;20:803-9.
18Hirano C, Nagata M, Noman AA, Kitamura N, Ohnishi M, Ohyama T,et al. Tetraspanin gene expression levels as potential biomarkers for malignancy of gingival squamous cell carcinoma. Int J Cancer 2009;124:2911-6.
19Ku KT, Wan L, Peng HC, Tsai MH, Tsai CH, Tsai FJ,et al. Vascular endothelial growth factor gene-460 C/T polymorphism is a biomarker for oral cancer. Oral Oncol 2005;41:497-502.
20Bertonha FB, Barros Filho Mde C, Kuasne H, Dos Reis PP, da Costa Prando E, Muñoz JJ,et al. PHF21B as a candidate tumor suppressor gene in head and neck squamous cell carcinomas. Mol Oncol 2015;9:450-62.
21Zhi X, Lamperska K, Golusinski P, Schork NJ, Luczewski L, Kolenda T,et al. Gene expression analysis of head and neck squamous cell carcinoma survival and recurrence. Oncotarget 2015;6:547-55.
22Korostoff A, Reder L, Masood R, Sinha UK. The role of salivary cytokine biomarkers in tongue cancer invasion and mortality. Oral Oncol 2011;47:282-7.
23Brailo V, Vucicevic-Boras V, Lukac J, Biocina-Lukenda D, Zilic-Alajbeg I, Milenovic A,et al. Salivary and serum interleukin 1 beta, interleukin 6 and tumor necrosis factor alpha in patients with leukoplakia and oral cancer. Med Oral Patol Oral Cir Bucal 2012;17:e10-5.
24Sato J, Goto J, Murata T, Kitamori S, Yamazaki Y, Satoh A,et al. Changes in saliva interleukin-6 levels in patients with oral squamous cell carcinoma. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2010;110:330-6.
25Abdul-Wahab R. Hamad, Shanaz M. Gaphor, Maissa T. Shawagfeh, Al-Talabani Nazar G. Study of serum and salivary levels of proinflammatory cytokines, potential biomarkers in the diagnosis of oral squamous cell carcinoma. Acad J Cancer Res 2011;4:47-55.
26SahebJamee M, Eslami M, AtarbashiMoghadam F, Sarafnejad A. Salivary concentration of TNFalpha, IL1 alpha, IL6, and IL8 in oral squamous cell carcinoma. Med Oral Patol Oral Cir Bucal 2008;13:E292-5.
27Wu Z, Weng D, Li G. Quantitative proteome analysis of overexpressed cripto-1 tumor cell reveals 14-3-3γ as a novel biomarker in nasopharyngeal carcinoma. J Proteomics 2013;83:26-36.
28Gurzu S, Krause M, Ember I, Azamfirei L, Gobel G, Feher K,et al. Mena, a new available marker in tumors of salivary glands? Eur J Histochem 2012;56:e8.
29Liang HS, Zhong YH, Luo ZJ, Huang Y, Lin HD, Luo M,et al. Comparative analysis of protein expression in differentiated thyroid tumours: A multicentre study. J Int Med Res 2009;37:927-38.
30Monteiro LS, Delgado ML, Ricardo S, Garcez F, do Amaral B, Pacheco JJ,et al. EMMPRIN expression in oral squamous cell carcinomas: Correlation with tumor proliferation and patient survival. Biomed Res Int 2014;2014:905680.
31de Araújo VC, Furuse C, Cury PR, Altemani A, de Araújo NS. STAT3 expression in salivary gland tumours. Oral Oncol 2008;44:439-45.
32Chung CH, Dignam JJ, Hammond ME, Klimowicz AC, Petrillo SK, Magliocco A,et al. Glioma-associated oncogene family zinc finger 1 expression and metastasis in patients with head and neck squamous cell carcinoma treated with radiation therapy (RTOG 9003). J Clin Oncol 2011;29:1326-34.
33Fenner M, Wehrhan F, Jehle M, Amann K, Radespiel-Tröger M, Grabenbauer G,et al. Restricted-expressed proliferation-associated protein (Repp86) expression in squamous cell carcinoma of the oral cavity. Strahlenther Onkol 2005;181:755-61.
34Goulart Filho JA, Nonaka CF, da Costa Miguel MC, de Almeida Freitas R, Galvão HC. Immunoexpression of cyclooxygenase-2 and p53 in oral squamous cell carcinoma. Am J Otolaryngol 2009;30:89-94.
35Ge L, Baskic D, Basse P, Vujanovic L, Unlu S, Yoneyama T,et al. Sheddase activity of tumor necrosis factor-alpha converting enzyme is increased and prognostically valuable in head and neck cancer. Cancer Epidemiol Biomarkers Prev 2009;18:2913-22.
36Gourin CG, Moretz WH 3rd, Weinberger PM, Xia ZS, Liu Z, Terris DJ,et al. Serum protein profile analysis following definitive treatment in patients with head and neck squamous cell carcinoma. Arch Otolaryngol Head Neck Surg 2007;133:1125-30.
37Kong YH, Syed Zanaruddin SN, Lau SH, Ramanathan A, Kallarakkal TG, Vincent-Chong VK,et al. Co-expression of TWIST1 and ZEB2 in oral squamous cell carcinoma is associated with poor survival. PLoS One 2015;10:e0134045.
38Goulioumis AK, Fuxe J, Varakis J, Repanti M, Goumas P, Papadaki H,et al. Estrogen receptor-beta expression in human laryngeal carcinoma: Correlation with the expression of epithelial-mesenchymal transition specific biomarkers. Oncol Rep 2009;22:1063-8.
39Taoudi Benchekroun M, Saintigny P, Thomas SM, El-Naggar AK, Papadimitrakopoulou V, Ren H,et al. Epidermal growth factor receptor expression and gene copy number in the risk of oral cancer. Cancer Prev Res (Phila) 2010;3:800-9.
40Aquino G, Sabatino R, Cantile M, Aversa C, Ionna F, Botti G,et al. Expression analysis of SPARC/osteonectin in oral squamous cell carcinoma patients: From saliva to surgical specimen. Biomed Res Int 2013;2013:736438.
41Zhu DW, Yang X, Yang CZ, Ma J, Liu Y, Yan M,et al. Annexin A1 down-regulation in oral squamous cell carcinoma correlates to pathological differentiation grade. Oral Oncol 2013;49:542-50.
42Cao W, Younis RH, Li J, Chen H, Xia R, Mao L,et al. EZH2 promotes malignant phenotypes and is a predictor of oral cancer development in patients with oral leukoplakia. Cancer Prev Res (Phila) 2011;4:1816-24.
43Mega S, Miyamoto M, Ebihara Y, Takahashi R, Hase R, Li L,et al. Cyclin D1, E2F1 expression levels are associated with characteristics and prognosis of esophageal squamous cell carcinoma. Dis Esophagus 2005;18:109-13.
44Nankivell P, Williams H, McConkey C, Webster K, High A, MacLennan K,et al. Tetraspanins CD9 and CD151, epidermal growth factor receptor and cyclooxygenase-2 expression predict malignant progression in oral epithelial dysplasia. Br J Cancer 2013;109:2864-74.
45Seibold ND, Schild SE, Bruchhage KL, Gebhard MP, Noack F, Rades D,et al. Prognostic impact of VEGF and FLT-1 receptor expression in patients with locally advanced squamous cell carcinoma of the head and neck. Strahlenther Onkol 2013;189:639-46.
46Ong CA, Shapiro J, Nason KS, Davison JM, Liu X, Ross-Innes C,et al. Three-gene immunohistochemical panel adds to clinical staging algorithms to predict prognosis for patients with esophageal adenocarcinoma. J Clin Oncol 2013;31:1576-82.
47Snietura M, Jaworska M, Mlynarczyk-Liszka J, Goraj-Zajac A, Piglowski W, Lange D,et al. PTEN as a prognostic and predictive marker in postoperative radiotherapy for squamous cell cancer of the head and neck. PLoS One 2012;7:e33396.
48Howell VM, Gill A, Clarkson A, Nelson AE, Dunne R, Delbridge LW,et al. Accuracy of combined protein gene product 9.5 and parafibromin markers for immunohistochemical diagnosis of parathyroid carcinoma. J Clin Endocrinol Metab 2009;94:434-41.
49Patel V, Hood BL, Molinolo AA, Lee NH, Conrads TP, Braisted JC,et al. Proteomic analysis of laser-captured paraffin-embedded tissues: A molecular portrait of head and neck cancer progression. Clin Cancer Res 2008;14:1002-14.
50Franzmann EJ, Reategui EP, Pedroso F. Soluble CD44 is a potential marker for the early detection of head and neck cancer. Cancer Epidemiol Biomarkers Prev 2007;16:1348-55.
51Li J, Feng X, Sun C, Zeng X, Xie L, Xu H,et al. Associations between proteasomal activator PA28γ and outcome of oral squamous cell carcinoma: Evidence from cohort studies and functional analyses. EBioMedicine 2015;2:851-8.
52Higgins JP, Green S. Cochrane Handbook for Systematic Reviews of Interventions. Wiley; 2008. p. 187-241.
53Juodzbalys G, Kasradze D, Cicciù M, Sudeikis A, Banys L, Galindo-Moreno P,et al. Modern molecular biomarkers of head and neck cancer. Part I. Epigenetic diagnostics and prognostics: Systematic review. Cancer Biomark 2016;17:487-502.