Molecular Characterisation Of Cell Carcinoma In Genome Biology Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Prediction of future metastasis and molecular characterization of head and neck squamous-cell carcinoma based on transcriptome and genome analysis by microarrays


Head and neck squamous-cell carcinoma is the most common cancer of the head and neck and can involve any of the mucous producing surfaces in the upper parts of the respiratory or digestive systems including the mouth, the pharynx, and the larynx. This leads to problems with speech and swallowing (Johnson and Jacobson, 2007, p.23). Of all cancers, head and neck squamous cell carcinoma is the sixth most common cancer in the world. (Yang et al, 2010)

According to the authors of this study alcohol intake, smoking and infection with human papillloma virus are among the causes of head and neck squmous cell carcinoma. Other known causes are lack of dental hygiene, chewing of tobacco or bethel nuts and exposure to radiation or compounds of nickel. (Lucente and Gady, 2004, p.305). Squamous cell carcinoma begins in the keratin synthesing cells in the layer of skin adjacent to the basal layer, the stratum spinosum. It often manifests itself as a scaly red growth. (Nicpon-Marieb and Hoehn 2007, p.166) The age range of patients entered into this study is 35-82 with 58.1 being the average age. An Irish study of patients with the disease who were less than 40 years old showed that just over half of these patients were younger than 35 (Toner and O'Regan 2009) this shows that age is not a factor in the development of the disease.

The spread of a tumour involves many steps. Firstly, a group of cancer cells colonize surrounding tissue, enter the bloodstream, and transfer from arteries and veins to capillaries. The cells then migrate into the surrounding tissues of the capillaries and grow into large secondary tumours. (Zhang et al, 2009). Untreatable metastasis is the main cause of death from cancer, metastases usually occurs in lymph nodes and organs separate to the organ where the original tumour is located. A tumour with a larger diameter than 0.25mm will not survive due to lack of oxygen and nutrients therefore must generate its own blood vessels (Harney, 2004, p.107).

The aim of this study is to find genes in tumours from patients who have previously been treated for head and neck squamous cell carcinoma. The purpose of finding these genes is to determine if the tumour will return and spread to another location in the future. The knowledge of the genes present could prove useful to healthcare professionals as patients with the genes that are prone to metastasis could be closely monitored and treatment could be formulated depending on the patients' prognosis.

Nineteen of the patients entered into the study were subsequently eliminated as they had previously been infected with human papilloma virus. These patients were excluded because the tumours of patients infected with human papilloma virus had differences in their RNA and DNA so therefore needed to be examined in another study. Human papilloma virus is a separate disease causing factor to the other known causes of head and neck squamous cell carcinoma mentioned above (Ragin et al, 2006)


This study can't be viewed in isolation as it is a follow on from previous studies. A previous study used reverse transcription, polymerase chain reaction and polyacrylamide gel electrophoresis in a process called differential display to identify 820 RNA molecules that were implicated in both tumours and non tumour tissue. Approximately 10% of these molecules were different in tumours that spread to tumours that didn't spread. (Carles et al, 2006) This showed that there are RNA molecules present in tumours that could indicate if the tumour was likely to metastasise. Affymetrix arrays discovered 164 RNA molecules the quantities of these molecules were different for tumours that developed metastasis to tumours that didn't develop metastasis. Affymetrix is a company that produces nucleotide microarrays to measure RNA molecules. The function of microarrays is to measure the amounts of thousands of genes at the same time using a glass or nylon membrane that has oligonucleotides printed on it. cDNA or RNA is bound to the microarray after being labelled with a radioactive or fluorescent probe. Complementary sequences that have been printed on the microarray will bind to cDNA. Quantity of RNA is known by measuring fluorescence. (Meyers, 2008 p. 9-10) The microarrays produced by Affymetrix are known as GeneChipsâ„¢. Each GeneChipâ„¢ represents a gene from a tumour or other biological sample. The GeneChipâ„¢ contains probes that will either match or won't match the test sample. This technique is used to determine the exact RNA sequences in a biological sample (Kumari et al, 2007).

Another earlier study examined gene signatures in breast cancer that could predict tumour spreading and identified three groups of genes involved in metastasis namely "metastasis initiation genes", "metastasis progression genes" and "metastasis virulence genes". "Metastasis initiation genes" cause cells from a primary cancer to be released into the circulation, "metastasis progression genes" cause development of primary tumour and also cause the cancer to spread to other tissues and "metastasis virulence genes" are involved in colonization of cancer cells in another location. There are 54 genes involved in breast cancer spreading to the lungs but only 18 of these are "metastasis progression genes" (Nguyen and Massague, 2007) The identification of these genes shows that there are various steps required in the spread and colonization of a tumour to another anatomical location.

This study is an advancement on previous studies mentioned as it is a larger study and uses bacterial artificial chromosomes comparative genomic hybridisation. 94 DNA samples were examined with array comparative genomic hybridisation microarrays. This is a sensitive method used to detect abnormalities in chromosomes and can detect multiple abnormalities in a genome. This is done by comparing the genome of a patient against the genome of a healthy individual. Both genomes are marked with separate fluorochromes and bound to target DNA. The change in fluorochrome colour after it is bound to DNA shows if chromosome material has been altered. (Lennon et al, 2008) QRT-PCR (Real time quantitative reverse transcription polymerase chain reaction) was used in this study to analyse 182 tumour samples for 59 genes. 31 of the genes were analysed for prediction of metastasis. This technique can be used in place of Northern blotting as it is a faster method of transcribing RNA to DNA and requires lesser quantities of RNA whereas a Northern Blot requires much larger quantities of RNA. (Maquat et al, 2008, p.104)

In this study 142 tumours were analysed and as a result four genes were found that signalled the return and spread of the carcinoma in the future. These genes are PSMD10, HSD17B12, FLOT2 and KRT17. PSMD10 is also known as gankyrin it is an oncoprotein found in the liver that causes the breakdown of p53 and retinoblastoma protein these are proteins that inhibit tumour formation so when they are broken down cell death is inhibited in tumours. (Umemura et al, 2008) Research was conducted on neuroblastoma, a childhood tumour. The research showed that patients who had a recurrence of the disease had higher levels of PSMD10 than patients who didn't experience a recurrence of the disease. (Schramm et al, 2005) Another study showed that PSMD10, was present at higher than normal levels in tumours from oesophageal squamous cell carcinomas. (Ortiz et al, 2008)

FLOT2 is a gene that codes for the protein flotillin 2. It is a protein isolated from caveolae, caveolae are types of lipid rafts involved in cell signalling. FLOT2 was previously found to have been associated with upregulation of mRNA and protein levels in cancerous melanomas. (Zhang et al, 2006) Another study has shown that p53 regulates the transcription of FLOT2 (Sasaki et al, 2008) This finding may explain why FLOT2 is a metastatic predictor in this study as p53 is expressed at abnormal levels in cancer cells. HSD17B genes [Hydroxysteroid{beta}Dehydrogenases] code for enzymes involved in metabolising steroids. HSD17B12 plays a part in the formation of estradiol a form of oestrogen. It has also been found to play a part in the lengthening of fatty acids. An experiment on mice showed that HSD17B12 is necessary for organ and nerve development in embryos. (Rantakari et al, 2010) In another study a mutation in the HSD17B12 gene was found to be associated with breast cancer because estradiol is a form of oestrogen and exposure to high levels of oestrogen has been found to be linked to the development of breast cancer. This is because breast tissue is one of the tissues where oestrogen is broken down. (Pluorde et al, 2009) In a separate study levels of KRT17 were found to be increased to 30 times greater than expected levels in breast cancer tumours. (Sharp et al, 2008) KRT17 codes for the protein keratin 17 found in skin and nails, mutations in the KRT17 gene have been linked to a rare skin disorder known as pachyonychia congenita. (Gruber et al, 2009)

Summary and Discussion

186 patients were entered into the study, 142 patients were chosen to be statistically analysed as they required a follow up within the 36 month timeframe of the study and were not infected with human papilloma virus. Unsupervised hierarchical classification was performed, this involved grouping samples into clusters, 449 genes were used to form 4 clusters. Clustering is a process where information is classified and sections of information are found without the structures of the sections being known, this process is used to identify groups of information that are possibly related. (Pham and Sobh 2008) A dendrogram was generated to show the arrangement of the clusters, this dendrogram was based on the stage of the tumour, (TNM staging system, see table*), the location of the tumour, whether the tumour has differentiated and the ratio of cells that have metastised to cells that havn't metastasised.

Fisher's exact test is used to measure statistical significance for tables with small values (Glantz 2005 pg. 158) This test found that the clusters 1, 2 and 3 were linked to the levels of tumour cell development whereas cluster 4 wasn't linked to tumour cell development, the area of the carcinoma or whether the carcinoma had spread from its primary location. The 449 genes were then divided into 6 groups (a-f) using unsupervised cluster analysis. Unsupervised analysis is concerned only with the data being analysed and doesn't take previous information about data into account. (Walker and Rapley et al, 2008, p.277)

The 6 gene groups were studied to analyse if the pathways of the genes were altered in any way, this analysis was done using hypergeometric tests. The 6 gene groups were then divided into their most closely related gene onthology. Gene onthology involves identifying proteins that are involved in several processes, instead of researching proteins that are involved in individual processes it then standardises the information across many databases. (Ashburner et al, 2000) Genes present in sample groups 1-4 were then analysed to identify their expression in clusters a-f, cluster a is associated with cell movement and it was found that it was over expressed in sample group C3. Cluster f was over expressed in sample group C4, cluster f had genes encoding muscle proteins this explains why it isn't involved in the formation of tumour cells.

Supervised analysis was used to identify three sets of genes that were present in two of the groups C1-C3, 50 of the genes involved in the unsupervised analysis were among the 835 genes identified in the supervised analysis. QRT-PCR was performed on seven genes and showed that gene expression varied from badly to well differentiated tumours. These seven genes were chosen because they showed a varied expression between in their tumour division status. Affymetrix and QRT-PCR gave very similar results.

Cox univariate and multivariate analysis was performed on Affymetrix and array comparative genomic hybridisation results from 81 patients and 31 of the Affymetrix results on RNA were further analysed. The results of the array comparative genomic hybridisation used to analyse DNA didn't have the power to predict tumour differentiation and spreading.

RNA molecules from 134 tumours were analysed by QRT-PCR, using Cox analysis on a training group of 55 samples the four genes to predict metastasis were identified. Another group of 79 samples independent to the study were analysed as a control to show the accuracy of the four genes in prediction of metastasis, this group was 74% accurate in predicting metastasis and 78% accurate in predicting non-metastasis, overall the four gene method of predicting metastasis was 77% accurate. Cox proportional hazard univariate analysis was used, this showed that the four gene model was largely correlated with foretelling the return and spread of a carcinoma as the p value obtained was 0.0003 showing that this method is statistically significant in foretelling cancer spreading.

Multivariate analysis showed that the four gene prediction had the lowest p value ie. was the most statistically significant method of cancer prognosis than other clinical methods. This indicated that the identification of the four genes is a better method of prognosing head and neck squamous cell carcinoma than methods such as detecting the presence of tumour cells in the lymph nodes, tumour size, location, the degree of change of normal cells or age. Kaplan-Meier curves were drawn, these curves are used to show the effects of various disease causing elements on survival, this curve can give data on survival even if patients are analysed for different lengths of time. (Joy et al, 2005 p.200) The curve was drawn using selected groups from the four gene model, these groups were (i) test groups (ii) the entire population (iii) patients with nodal status N0/N1 and N2 (iv) patients who were ≤ stage IV (from TNM staging system see table). These curves proved that the four gene model worked well in predicting metastasis in comparison with other clinical factors, presently used.

Cox univariate analysis was used to identify genes that give potential to primary tumours to metastasise. 614 genes were identified that were expressed diffierently in metastasised tumours to non metastasised tumours. 22 genes were chosen that were also expressed differently. These 22 genes were analysed using QRT-PCR. These genes were analysed as a control for the microarray analysis of the 614 genes analysed. The functions encoded by the 614 genes are thought to be necessary for metastasis. eg. cell movement and processing of RNA. Head and neck squamous cell carcinoma is related to changes in chromosomes, aCGH was used to analyse 74 head and neck squamous cell carcinomas. Chromosome changes that were found in previous studies were noticed. But other changes that weren't noticed previously were discovered. This shows that metastasis is regulated at the RNA level and at the gene level.

In this study four genes were identified that could signify the return and metastasis of a head and neck squamous cell carcinoma. The authors of the study state that screening for the 4 genes named is a better method for giving a prognosis of the disease than methods used previously. A test for these mentioned genes could be carried out clinically using quantitative PCR. While the four gene model is statistically significant in predicting cancer spreading it can't be used in isolation to prognose cancer as the other factors aforementioned used to in cancer prognosis are also statistically significant.

According to the authors of this study, a study on a larger group of patients is now taking place, the importance of a larger study can be demonstrated by reviewing a previous study undertaken to identify genes causing tumour generation and metastasis in hypopharyngeal cancer, four normal and thirty-four tumour samples were analysed, microarray analysis was used to identify genes that were expressed differently between tumours and normal tissue. 164 genes were identified as having the potential to cause a tumour to metastasise. (Cromer et al, 2004) If a larger study was done using more samples including a variety of head and neck squamous cell carcinomas as well as hypopharyngeal tumours, it is possible that the number of genes found might have been more limited and the results may have been of more value.


This method of predicting metastasis is not yet suitable for clinical decision making. Before this method is ready for clinical decision making multiple tests must be carried out to avoid bias, larger databases are needed for the development of assays, independent groups must be tested to verify the results obtained and clinical trials must be performed. When all of these factors are taken into consideration the cost of molecular profiling remains as a great problem, the diagnosis could cost more than the treatment if this method of prognosis came into effect. (Ioannidis 2007)

If this method of prognosis is used in the future, it will very useful in predicting the outcome for patients with head and neck squamous cell carcinoma. Perhaps in the future this study could be repeated for other cancers and possibly a large study could be conducted to identify a gene model that predicts metastasis in all cancers. As described above the metastasis predictors KRT17, HSD17B12, FLOT2 and PSMD10 have been overexpressed in other cancers besides head and neck squamous cell carcinoma. In the future it may be possible to collaborate all of the studies conducted on cancers and to generate a database of metastasis causing genes. This database could then be used by clinicians and scientists in the prognosis for a cancer patient.

Herceptin is a drug used in the treatment of breast cancer, it competes with the oncogene HER2 which is often up regulated in breast cacner (Nguygen and Massague, 2007) This gives hope for the development of antagonistic therapies to PSMD10, HSD17B12, FLOT2 and KRT17 for the treatment of head and neck squamous cell carcinomas and possibly other cancers where these genes are over expressed. Further studies will be needed to determine if the genes found in this study are involved in the same signalling mechanisms as those found in other studies.