alexa
Reach Us +44-1647-403003
Comprehensive Study of Oral Squamous Cell Carcinoma Patients Using Blood Samples and Gene Expression Profiles | OMICS International
ISSN: 1948-5956
Journal of Cancer Science & Therapy

Like us on:

Make the best use of Scientific Research and information from our 700+ peer reviewed, Open Access Journals that operates with the help of 50,000+ Editorial Board Members and esteemed reviewers and 1000+ Scientific associations in Medical, Clinical, Pharmaceutical, Engineering, Technology and Management Fields.
Meet Inspiring Speakers and Experts at our 3000+ Global Conferenceseries Events with over 600+ Conferences, 1200+ Symposiums and 1200+ Workshops on Medical, Pharma, Engineering, Science, Technology and Business

Comprehensive Study of Oral Squamous Cell Carcinoma Patients Using Blood Samples and Gene Expression Profiles

Nobuo Kondoh1*, Eiji Takayama1, Masako Kamiya1, Harumi Kawaki1, Masayuki Motohashi2, Yasunori Muramatsu2, Michio Shikimori2, Kenji Mitsudo3and Iwai Tohnai3
1Department of Oral Biochemistry, Asahi University School of Dentistry, Mizuho-shi, Hozumi 1851Gifu 501-0296, Japan
2Department of Oral Surgery, Asahi University School of Dentistry, Mizuho-shi, Hozumi 1851Gifu 501-0296, Japan
3Department of Oral and Maxillofacial Surgery, Yokohama City University Graduate School of Medicine, Yokohama-shi, Kanagawa 236-0004, Japan
Corresponding Author : Nobuo Kondoh
Department of Oral Biochemistry
Asahi University School of Dentistry
Mizuho-shi, Hozumi 1851
Gifu 501-0296, Japan
Tel: 81 58 1416
Fax: 81 58 329 1417
E-mail: [email protected]
Received November 01, 2012; Accepted November 24, 2012; Published November 26, 2012
Citation: Kondoh N, Takayama E, Kamiya M, Kawaki H, Motohashi M, et al. (2012) Comprehensive Study of Oral Squamous Cell Carcinoma Patients Using Blood Samples and Gene Expression Profiles. J Cancer Sci Ther S18:001. doi: 10.4172/1948-5956.S18-001
Copyright: © 2012 Kondoh N, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Related article at
DownloadPubmed DownloadScholar Google

Visit for more related articles at Journal of Cancer Science & Therapy

Abstract

Oral squamous cell carcinoma (OSCC) is an aggressive malignancy which shows a variable degree of malignant behavior. To identify molecular signatures and establish a new diagnostic model for oral malignancies, we have identified marker genes representing pre-malignant and malignant phenotypes of oral mucosal lesions. The expression of marker genes was examined by quantitative reverse transcription-PCR. Then, we created discriminatory predictor models using Fisher’s linear discriminant analysis and leave-one-out cross validation. These models were applicable for the diagnoses of pre-malignant leukoplakias (LPs), and of invasion status for advanced OSCCs.

The clinical course of various cancers is also influenced by host immune response. Our preliminary data using flow cytometric analysis demonstrate that the percentage of CD4+CD57+ T cells in peripheral blood lymphocyte was higher in the high grade OSCCs than that in the low grade ones. Furthermore, lipopolysaccharides (LPS)-induced ex-vivo production of Interferon (IFN)-γ from peripheral blood cells was highest in stage I patients and gradually decreased during the course of OSCC progression up to stage III. These decreased levels in the early stages were inversely correlated with tumor size.

In this review, we propose that the usage of the immunological status of OSCC patients combined with the molecular signatures of tumor tissues could provide valuable indices for diagnosis of oral malignancies.

Keywords
Oral squamous cell carcinoma (OSCC); Leukoplakia (LP); Interferon (IFN)-© CD4+CD57+ T cells
Introduction
Leukoplakias (LPs) are white lesions that include hyperplasias and dysplasias of the oral mucosa, and often undergo malignant transformation to oral squamous cell carcinoma (OSCC) [1]. The histological features associated with LP, and with OSCC, have been described previously [2]. The dysplasias are classified as mild, moderate or severe, based on histopathological findings, and these designations are thought to be the sequential phases of oral carcinogenesis. However, we sometimes experience discordance between the pathological diagnosis of these lesions and the corresponding prognosis in an individual patient. It has been suggested that the detection of clonal genetic changes, such as loss of heterozygosity or microsatellite instability, in both primary OSCC and pre-malignant lesions could be a more informative method of monitoring these cancer patients [3]. The staging and grading of OSCCs are still, for the most part, dependent on traditional clinicopathological observations [3]. In line with our approach [4], there are several attempts for the identification of molecular biomarkers associated with specific phenotypes of head and neck squamous cell carcinomas [5-7]. Although, these approaches are effective to observe the wide variety of genetic alterations in oral malignancy, in order to accurately diagnose cancer subtypes, supervised learning methods are the most suitable [8-11]. Thus, we attempted to generate a multigene classifier for the diagnosis of pre-malignantto- malignant transition. Using cDNA microarray and QRT-PCR techniques, a comprehensive gene expression profile was generated and compared among OSCC and LP tissues. We subsequently defined a list of 27 marker genes that are either significantly elevated or downregulated in OSCCs, compared with LPs. Among these genes, predictor gene sets for OSCC-LP classification were determined by Fisher’s linear discriminant analysis (LDA) and validated by the leaveone- out cross validation [12].
On the other hand, OSCC is an aggressive malignancy which shows a variable degree of malignant behavior. The biological characteristics of this cancer are not yet well understood. We also attempt to generate diagnostic classifiers for OSCCs of an advanced stage [13]. Metastasis is a major cause of local relapse and death after definitive therapy in patients with these tumors. There are several reported approaches for the identification of molecular biomarkers and for predicting patients that are at high risk of recurrence [9,14-18]. Among the conventional staging and grading systems used for OSCC tissues, we focused upon the Yamamoto-Kohama’s (YK’s) mode of invasion [19], since the criteria involved can be largely correlated with prognosis, particularly in the case of lymph node metastases [20,21]. However, pre-operative clinicopathological estimations of the invasion status of these lesions are often inaccurate, because of the difficulty in biopsy sampling of the deepest portion of the invasive front, which is closely related to lymph node metastasis. It has been suggested that the genetic backgrounds associated with malignant phenotypes are equally responsible for the bulk of the resulting primary tumor [22]. Hence, using primary OSCCs, we have attempted to generate molecular classifiers that can predict the YK’s mode of invasion using Fisher’s LDA, and have evaluated the diagnostic significance of our findings. Our present results suggest that these differentially expressed genes can provide valuable prognostic tools for LP or OSCC patients.
Although there are several attempts to molecularly diagnose advanced tumors, the predictive values for metastases based on the gene expression profiles of various cancer cell types can vary from 75% to 100% [9,15,17]. It has been postulated that several discrete steps comprise the biological cascade leading to tumor metastasis, including the evasion of growth suppression, invasiveness, motility, detachment, angiogenesis, the contribution of tumor-associated macrophages, lymphangiogenesis, vascular adhesion, homing, and resistance to the innate immune response [23]. These should be defined not only by the phenotype of the cancer cells themselves, but also by the conditions of the host microenvironment [23]. We had previously reported that a specific subset of T-lymphocyte, blood cytokine levels and cytokine-producing capability of peripheral blood cells (PBCs) are well correlated with the progression of gastric cancer [24] and hepatocellular carcinoma [25]. Therefore, we attempt to diagnose the malignancy by evaluating systemic sign based on immunological status of OSCC patients. Hence, we attempt to evaluate general conditions of OSCC patients.
Corroborating information from both local tissues and patient’s conditions could be meaningful to accomplish quite accurate diagnosis of oral malignancy.
Discrimination of Oral LP Subtypes and OSCC Using Gene Expression Signatures
The identification of potential marker genes between LP and OSCC
To identify potential marker genes that are differentially expressed between LP and OSCC, cDNA microarray analyses were performed using RNA mixtures of 5 OSCC and of 5 LPs [12]. Among the 16,600 target cDNAs on the chip arrays, 63 genes were highly expressed (3-fold or more) in the OSCC mixture, compared with the LP mixture. In addition, 55 genes were preferentially expressed (3-fold or more) in the LP mixture [12]. To validate the identified marker gene candidates, QRT-PCR analysis using several OSCC and LP tissues was performed. We focused finally on 27 differentially expressed marker genes, among which 15 (denoted as LP) were overexpressed in LPs compared to OSCCs, whilst 12 genes were upregulated in OSCCs and in some moderate- to severe-dysplasias (denoted as SC) (Table 1).
Supervised classification based on Fisher’s LDA
The goal of this study is to establish a clear distinction between LP and OSCC, based upon a molecular classification. In order to identify marker gene sets that could discriminate between OSCCs and LPs, a supervised classification approach using LDA was performed [12]. The expression of these 27 marker genes was analyzed among 27 OSCCs and 19 LPs, including hyperplasias and dysplasias. This approach involves parameter (gene) selection by the use of a stepwise increment and genetic algorithm. When the Fisher’s ratio was employed as a score, a model with 11 parameters was selected as the best model. The stability of this model was examined by the leave-one-out cross validation (loo) method. The LDA score for each sample is given as the following linear discrimination function:
Score= - 0.231 (LP1) + 0.223 (LP4) - 0.0537 (LP28) - 0.0734 (LP21) - 0 .892 (LP12) - 0.0617 (LP29) - 0.282 (LP8) + 0.0122 (SC1) + 0.0669 (SC13) - 0.0684 (SC43) - 0.0366 (SC5).
The score was obtained when the levels of 11 genes were substituted into the equation. As shown in figure 1, the scores for OSCCs become plus, while that for LPs became minus. However, the absolute values of scores have no meaning. With an exception of moderately-differentiated dysplasia, Mo dys 33, all sample sets were correctly discriminated by the 11 marker genes selected by the Fisher’s ratio. The optimal prediction accuracy with this set of 11 genes was 97.8% (loo).
Discrimination of Non-aggressive and Aggressive Primary OSCC Using Gene Expression Signatures
The identification of marker genes for non-aggressive and aggressive OSCCs
To identify potential marker genes that are differentially expressed between non-aggressive and aggressive OSCCs, cDNA microarray analysis was performed using RNA mixtures from metastasis-negative and less invasive SCCs and also from relatively aggressive OSCCs that have a developing lymph node and/or local metastasis [13]. Among the 16,600 target cDNAs on the chip arrays used, 46 genes were found to be highly expressed in the aggressive OSCCs compared with the non-aggressive ones, and a further 37 genes showed the opposite pattern.
The expression levels of all these marker gene candidates were verified by QRT-PCR analysis using 64 OSCC tissues. Among the 83 marker gene candidates, we selected, 53 genes showed markedly different expression levels that could be associated with an YK’s mode of invasion transition, T classification and/or lymphnode metastasis (p<0.06). Of these, 29 were found to be markedly down-regulated (Table 2) and 24 were observed to be up-regulated (Table 3), concomitantly with the acquisition of an invasive phenotype. Some of the marker genes demonstrated differential expression along with T grades and/or metastasis, however, the number was restrictive [13].
Supervised classifications to establish predictor models for YK’s mode of invasion
It has been reported that YK’s mode of invasion is largely correlated with the incidence of lymph node metastasis [20,21]. Histological criteria of the YK’s mode of invasion are defined as the following 5 grades: grade 1, well defined borderline; 2, cords, less marked borderline; 3, groups of cells, no distinct borderline; 4C, diffuse invasion with cord-like type; and 4D, with diffuse type invasion [19]. Since, we had isolated a large number of candidate marker genes for the invasion status of an OSCC; we attempted to establish a molecular classification along with the YK’s mode of invasion using biopsied OSCC samples. Using the same strategies as already mentioned [13], total RNAs were isolated from primary OSCCs of both node-positive and –negative patients, and marker gene selection was performed between them. We isolated 53 marker genes characteristic for YK’s mode of invasion.
As we have reported [12,13], a supervised classification approach based on LDA fitted with a step-wise increment method was performed on this same panel of marker genes in the 64 patient samples. Then, we created four discriminatory predictor models (from LDA-1 to -4) based on from 16 to 25 gene signatures (Tables 2 and 3), which could best distinguish the five established grades of YK’s mode of invasion. The stability of this model was examined using the leave-one-out cross validation (loo) method and then compared with that of the stepwise increment method. The LDA score for each sample is given by the following linear discrimination functions:
LDA-1 (for YK-1 vs. -2, -3, -4C and -4D); Score = 0.394866 + 0.229884 (HOP) + 0.211169 (CKM) + 0.324503 (CDA) - 0.620754 (CSRP2) - 0.338919 (C1S) + 0.004993 (ODC1) + 0.136036 (TNFSF10) +0.05849 (TAGLN) - 0.104157 (NK4) + 0.15035 (HLA-DBP1) - 0.164207 (HLA-DMB) + 0.359343 (GRCC10) + 0.159684 (GLG1) - 0.097888 (C4.4A) + 0.078056 (KLK7) - 0.161743 (S100A12) + 0.031256 (SULT2B1) - 0.131336 (TGM3).
LDA-2 (for YK-1, -2 vs. -3, -4C and -4D); Score = 0.986068 - 0.010085 (CLSP) - 0.03203 (RNASE7) + 0.088897 (LOC14450) - 0.250882 (SLPI) + 0.03128 (ZNF185) + 0.182597 (CKM) + 0.224274 (CDA) - 0.360236 (D4S234E) + 0.398142 (TNFSF10) - 0.136753 (CBR3) + 0.213823 (MYL9) + 0.045013 (MMP11) - 0.222319 (HLA-DMB) + 0.116148 (ACTN1) - 0.124201 (HSPC159) - 0.267171 (FLJ11036) - 0.538498 (AFG3L2) + 0.221296 (TM7SF2).
LDA-3 (for YK-1, -2, -3 vs. -4C and -4D); Score = - 0.328603 + 0.128613 (CCL19) - 0.00548 (CLSP) + 0.345872 (FHL1) - 0.39984 (SLPI) - 0.023614 (APM2) - 0.052443 (AQP3) - 0.040878 (CSTB) + 0.24299 (CSRP2) - 0.141004 (D4S234E) - 0.00906 (ODC1) + 0.211764 (TNFSF10) + 0.046316 (TAGLN) + 0.157812 (PLAU) - 0.038082 (NK4) - 0.165494 (HLA-DMB) + 0.337967 (LGALSI) - 0.369404 (GLG1) - 0.090304 (COL1A1) - 0.372974 (FLJ11036) + 0.217867 (AFG3L2) + 0.19053 (KLK7) + 0.010609 (MPN) + 0.092418 (SULT2B) - 0.138387 (WFDC12) + 0.098447 (TGM3).
LDA-4 (for YK-1, -2, -3, -4C vs. -4D); Score = -0.771245 + 0.071189 (LOC144501) - 0.401477 (D4S234E) - 0.066541 (CBR3) + 0.080463 (PLAU) - 0.14224 (MMP2) - 0.045855 (MMP11) + 0.481743 (LGALSI) - 0.106972 (HIST1H2BK) - 0.412974 (GLG1) + 0.087268 (C1QG) + 0.026049 (COL6A1) - 0.513483 (FLJ11036) + 0.019158 (MPN) + 0.0567 (SULT2B1) - 0.259138 (WFDC12) + 0.209169 (TGM3).
The gene sets for each LDA equation are summarized in tables 2 and 3. The optimal prediction accuracies examined by the leave-one-out cross validation of the predictor gene sets were 93.8% (LDA-1), 95.3% (LDA-2), 92.2% (LDA-3) and 93.8% (LDA-4), respectively. As a validation test, these 4 LDA models were then applied to data from 13 independent primary OSCCs. The prediction fidelity to the pathological observations of these tumors was 77% (LDA-1), 85% (LDA-2), 77% (LDA-3) and 100% (LDA-4) (Table 4). Among these 13 samples, five demonstrated discrepancies between the pathological and molecular diagnoses. However, four cases remained within one rank (SCC126, 134, 125 and 132), whereas only one case (SCC129) demonstrated fluctuation on three ranks. Interestingly, most of the discrepancies (80%) could be attributed to an over-estimation by molecular diagnosis (Table 4).
Diagnostic Significance of Cytokine Production and T cell Subsets in the Peripheral Blood Cells from OSCC Patients
As our next approach, we attempt to evaluate general conditions of OSCC patients. In the previous study, we have reported that CD4+CD57+ T, a subset of CD4+ (conventional helper) T, cells are increased in PBCs of tumor patients including hepatocellular carcinoma and gastric cancer [24,25]. Thus, we performed fluorescence activated cell sorting (FACS) analysis to see CD4+CD57+ T cells in PBCs of OSCC patients. We also focused on immunological status of OSCC patients. To this end, we examined LPS-Induced cytokine production and/or T lymphocyte subsets in peripheral blood cells from them. In this assay, blood sample was stimulated by lipopolysaccharides (LPS), resulting in the activation of Th1 cells which could release IFN-γ. After 48 h of the stimulation, the production of IFN-γ was assayed by ELISA. As shown in figure 2A, our preliminary data demonstrated that LPS-induced IFN-γ-producing capability of PBCs from OSCC patients was higher in stage I and decreased in a step-wise manner up to stage III during the course of tumor progression. In contrast, the IFN-γ production seemed to be increased in stage IV, compared to that in stage III patients. This decreased levels in the early stages were inversely correlated with tumor size (Figure 2B), while the levels regain in the last stage seemed to be associated with lymph node metastasis (Figure 2C). However, to reach this conclusion, we have to examine additional samples (manuscript in preparation). As shown in figure 3, the ratio of CD4+ CD57+ T cells against CD4+ (conventional helper) T cells was gradually increased in a step-wise manner up to stage III during the course of tumor progression.
Discussion
In order to identify marker gene candidates, we first screened differential gene expression between OSCCs and LPs. We identified 27 marker genes representing the differential expression between LPs and OSCCs. To further identify marker gene sets and establish appropriate algorism that can sufficiently discriminate between OSCCs and LPs, a supervised classification approach based on LDA was performed. After intensive parameter selection and cross validation, we reached an optimal prediction with a set of 11 genes. According to our classification, however, a moderately differentiated displasia sample, Mo dys 33, was insistently classified as an OSCC. Interestingly, Mo dys 33 had been clinically diagnosed and treated as an OSCC, because of its cancerous macroscopic appearances and history of multiple OSCC [12]. Therefore, a molecular diagnosis based on these 11 genes may, in part, predict clinical features or genetical background of the patient rather than histological grades.
After intensive parameter selection and cross validation, we reached optimal predictions for the YK’s mode of invasion with four sets of marker genes [13]. As a validation test, a data set of 13 independent primary OSCCs were applied to the 4 LDA models and our results demonstrated that the prediction fidelity of these models with the pathological observations was higher than 77%. Among the inconsistencies found between the molecular and pathological diagnoses, four out of five remained within one rank up or down and could be attributed to an over-estimation, rather than under-estimation, by molecular diagnosis. In general, the pathological diagnosis complies with the highest grade among limited numbers of tissue sections. In contrast, however, OSCC tissues are molecularly diagnosed using homogenously extracted RNAs from a certain volume of tissue sample. Thus, there may be an oversight during the pathological inspection, but not in the molecular-based inspection. In fact, case OSCC130 listed in Table 4 had first been diagnosed as grade YK-2 tumors, which are two ranks below its molecular diagnosis. However, this case was pathologically re-evaluated and agreement was found with the molecular diagnosis (i.e. an YK-4C grading was assigned). Gene expression may thus obediently reflect the cellular potency of a tumor whereas the pathological appearance of a lesion may demonstrate cell behavior which is more or less modified by its tissue microenvironment.
There are several attempts to molecularly diagnose advanced tumors; However, in line with our results, the predictive values for metastases based on the gene expression profiles of various cancer cell types can vary from 75% to 100% [9,15,17]. There are several discrete steps that comprise the biological cascade leading to metastasis, including, for example, invasion, cell homing and evasion from the innate immune system. Thus, the clinical prognosis may be affected not only by the phenotype of cancer cells, the so called “seed”, but also by the conditions of the host microenvironment, the “soil” [23]. There should be an interaction between cancer cells and host immunity during epithelial-mesenchymal transition (EMT). During EMT, Snail is a major transcription factor involved in cancer metastasis partly by inducing multiple immunosuppression and immunoresistant mechanisms [26]. Our preliminary data demonstrated that LPS-induced ex-vivo production of IFN-γ from peripheral blood cells (PBC) of OSCC patients is inversely correlated with the tumor progression. In addition, the ratio of CD4+CD57+ T cells was gradually increased along with the tumor progression.
Our results strongly suggest that corroborating information from both local tissues and patient’s immunological indicators should be essential to accomplish quite accurate diagnosis of oral malignancy.
References


























Tables and Figures at a glance

Table icon Table icon Table icon Table icon
Table 1 Table 2 Table 3 Table 4

 

Figures at a glance

Figure Figure Figure
Figure 1 Figure 2 Figure 3
Select your language of interest to view the total content in your interested language
Post your comment

Share This Article

Relevant Topics

Recommended Conferences

Article Usage

  • Total views: 12096
  • [From(publication date):
    June-2013 - Dec 11, 2019]
  • Breakdown by view type
  • HTML page views : 8293
  • PDF downloads : 3803
Top