Radiotherapy is commonly used to treat oral squamous cell carcinoma (OSCC), but its therapeutic effects are unpredictable. To determine which genes correlate with radiation resistance in oral cancer, the authors evaluated radiation sensitivity using a standard colony formation assay with a gene microarray system for seven OSCC cell lines. They found significant associations between dozens of gene-expression levels and radiation resistance of OSCC cell lines. Following analysis of the different radiosensitive cancer cell lines, the friend leukaemia insertion (Fli)-1 gene was selected as a prediction marker gene for OSCC radiotherapy resistance. Fli-1 expression was associated with radiation resistance in OSCC patients. These data help to predict the effects radiation therapy has on OSCC, in turn contributing to the development of alternative radiation therapies.
Radiation therapy is acceptable treatment for patients with oral squamous cell carcinoma (OSCC). Preoperative radiation therapy is often effective in reducing the area of tumour infiltration. It allows improvements in overall survival rates and maintains oral morphology and function. The reaction to radiation of each patient’s carcinoma is variable. To choose the most appropriate therapy and avoid side effects, a method for predicting radiation therapeutic effectiveness must be established. Intrinsic radiation resistance depends on the stage and volume of the disease and the molecular pathogenic mechanisms of the individual cells that comprise the tumour’s global phenotype.
Gene expression profiling of the cDNA microarray of human cancers is a way of efficiently accessing genetic information on tumour phenotypes . Expression genomics may be a way of further understanding the biological pathways involved in cancer and help to predict disease outcome and treatment response. Many genetic alterations are associated with the development and progression of oral cancer . Some of these genetic changes predict radiation therapy outcomes. The difference in the effectiveness of radiation therapy in patients has been associated with variations in cancer cell gene expression profiles .
The objective of the current study was to determine potential radiation response predictors for oral cancer using cDNA microarray analysis. The results will contribute to the understanding and function of genes that are linked to clinical outcomes in the treatment of oral cancer.
Materials and methods
Cell lines and cell culture
The authors evaluated seven OSCC cell lines (HSC2, HSC3, HSC4, SCC25, SCC66, SCC111, Ca9-22). HSC2, HSC3, HSC4, Ca9-22 were obtained from RIKEN cell bank (RIKEN, Tsukuba, Japan) and SCC25, SCC66 and SCC111 were kindly provided by Dr. David T.W. Wong (Dental Research Institution of UCLA). The tumours from which these lines were derived did not undergo any treatment prior to establishment. HSC3, HSC4, SCC25, SCC66, SCC111 originate from the tongue, Ca9-22 has gingival origin, HSC2 originates from the mucosa of the mouth. All cell lines were maintained in DMEM/F12 medium (Gibco-BRL, Gaithersburg, MD, USA) and supplemented with 10% heat-inactivated foetal bovine serum (Sigma, St. Louis, MO, USA) with 50 units/ml penicillin and streptomycin at 37 °C in 95% air/5% CO 2 .
Clonogenic survival assay
Eight hundred cells were plated in 35-mm tissue culture dishes in triplicate and irradiated 48 h later with various single radiation doses using a linear accelerator (MBR.1520R, Hitachi Medico, Tokyo, Japan) at a source with a target distance of 50 cm. The medium was changed 24 h after radiation and incubated for approximately 10 days, allowing colonies to form. The colonies were fixed, stained with a solution of crystal violet and counted. The survival fractions were calculated as a ratio of plating efficiencies for treated and untreated cells. These experiments were repeated a minimum of three times. The linear quadratic model was used to fit the cell survival data to the following equation: In[survival fraction (SF)] = − αD − βD 2, where D represents radiation dose (Gy), and α and β are fit to the individual data. The parameters α of and β were calculated for each curve.
Isolation of total RNA
Total RNA OSCC cell lines were prepared by lysing cells with ISOGEN (Wako, Osaka, Japan) according to the manufacturer’s instructions. In the final step, RNA was dissolved with 11 μl RNase-free water. RNA quality was analysed on a 1% agarose gel and the concentration was determined using a RNA/DNA spectrophotometer (Rio-Rad, Tokyo, Japan).
Applied Biosystems Human Genome Survey Arrays were used to analyse the transcriptional profiles of nine OSCC cell line RNA samples. The 60-mer oligo probes were synthesized using standard phosphoramidite chemistry and solid-phase synthesis, and quality controlled using mass spectrometry. The probes were deposited and covalently bound onto a derivatized nylon substrate (2.5 in. × 3 in.) that are backed by glass slides by contacting spotting with a feature diameter of 180 μm and space of >45 μm between each feature. A 24-mer oligo internal control probe (ICP) was co-spotted at every feature with 60-mer gene expression probe on the microarray.
Digoxigenin-UTP labelled cRNA was generated and linearly amplified from 1 μg total RNA using Applied Biosystems Chemiluminescent RT-IVT Labelling Kit v 2.0 according to the manufacturer’s protocol. Array hybridization (two arrays per sample), chemiluminescence detection, image acquisition and analysis were performed using Applied Biosystems Chemiluminescence Detection Kit and Applied Biosystems 1700 Chemiluminescent Microarray Analyzer (Rio-Rad, Tokyo, Japan), in accordance with the manufacturer’s protocol.
Identification of genes associated with radiation sensitivity
Gene spring 7 software (Silicon Genetics) was used to extract Assay Signal as well as Assay Signal to Noise ratio values from the microarray images. Bad spots flagged by the software were removed from the analysis. The Assay Signal of the remaining set of 99.5% genes was log transformed and quantile-normalized across arrays. To select differentially expressed genes, the remaining set of 99.5% genes was further filtered by Standard Expression Array System Signal to Noise threshold (S/N greater than 3 in at least one sample).
Twenty-six OSCC specimens were analysed from patients who underwent preoperative external radiation therapy before surgery after having obtained informed consent. The age of the 26 patients (19 males, 7 females) ranged from 34 to 82 years, with an average age of 61.5 years. Tumour extent was classified according to the TNM system by UICC . There were 6 cases of stage II, 8 cases of stage III, and 12 cases of stage IV. Radiation was performed five times a week with 2 Gy. The mean total dose was 32.9 Gy (26–40 Gy). All patients underwent surgery 2 weeks after radiation treatment.
Tumour response assessment
The pathologic response in the primary tumours was scored as previously reported : 0, no evidence of treatment effect; 1+, treatment effect involving up to one-third of the gross tumour mass; 2+, effect involving one-third to two-thirds of the gross tumour mass; 3+, treatment effect in more than two-thirds of the gross tumour mass; 4+, treatment effect of the entire tumour with no viable carcinoma identified. The pathologic response was scored by two authors (SS and HH) who were blind to each patient’s clinical response to radiation therapy.
Specimens were fixed in 10% formalin and embedded in paraffin. Serial sections with a 4 μm thickness from each specimen were made for immunohistochemistry. Tissue sections were deparaffinized and dehydrated in graded alcohols. The sections were treated for 30 min with absolute methanol, including 0.3% H 2 O 2 , to inhibit endogenous peroxidase activity, then incubated in trypsin (Difco, MI, USA) and 1.5% normal horse serum diluted 1:75 in Tris-buffered saline (145 mM NaCl, 20 mM Tris, pH 7.6) for 10 min. The specimens were incubated in primary antibody (anti-friend leukaemia insertion (Fli)-1 polyclonal antibody; Santa Cruz Biotech, CA, USA) at 4 °C for 16 h. Bound antibody was detected using the Envision system (Dako, Japan). Diaminobenzidine (1 mg/ml) in the presence of 0.03% hydrogen peroxidase was used to visualize any bound peroxidase. Sections were counterstained with hematoxylin.
Cases were considered positive if more than 80% of the cancer cells manifested cytoplasmic positivity. At least 10 high-powered tumour fields were evaluated.
All calculations were performed using the statistical computer program JMP IN (SAS Institute, Inc., NC, USA). The Pearson’s correlation coefficient analysis was performed on all data in order to determine the overall correlation coefficient between radiation sensitivity and Fli-1 expression. Statistical analysis of the correlation between Fli-1 expression and radiation therapy response was performed by means of the two-sided Fisher’s exact test. The significance level was set at P < 0.05 for each analysis.
Clonogenic survival of seven OSCC cell lines after radiation
To determine radiosensitivity, a clonogenic survival assay was performed on the seven human OSCC cell lines. Ca9-22 and SCC25 were the most radiosensitive, while SCC111 and HSC2 did not respond to radiation ( Fig. 1 ). Cell survival parameters of these survival curves are shown in Table 1 . The seven SCC cell lines exhibited a wide variety of radiation responses; for example, D 10 ranged from 1.7 Gy for Ca9-22 to 10.3 Gy for SCC111.