NA confirms the presence of EGFRvIII. C. Real-time amplification plot showing EGFRvIII-positive OSCC patient and the positive control. doi:10.1371/journal.pone.0031723.g003 Four samples were excluded from the analysis 
either due to failed amplification of the reference genes or having Ct values greater than 38 for one or both of these reference genes. All other samples successfully amplified both reference genes. Our real time PCR assay revealed that only one patient was positive for EGFRvIII mRNA expression. Retesting of this 532-91-2 supplier sample confirmed EGFRvIII positivity. In addition, both direct cDNA sequencing and conventional RTPCR were performed on this sample, however both methods failed outright. This failure may be attributed to formalin fixationinduced degradation and modification of DNA/RNA and further highlights the limitations of conventional methods when using FFPE tissue. We also measured total EGFR protein levels for all samples by quantitative fluorescent immunohistochemistry using the HistoRx AQUAH platform . EGFR AQUA scores, representing the concentration of EGFR protein, showed a range of expression from 426 to 1696 in normal oral cavity squamous epithelium with a median score of 1151 and a standard deviation of 6406. We used the median EGFR AQUA score plus one standard deviation as our definition for EGFR over-expression. Using this definition, we found that tumors from 22 of 50 patients over-expressed EGFR. Comparison of EGFR AQUA scores with EGFRvIII expression showed that the tumor sample with the highest EGFR AQUA score was the EGFRvIII-positive case identified by our real-time RT-PCR assay. In order to address tumor heterogeneity, an additional FFPE tumour ” block was randomly selected from 22 patients in the cohort and was tested for EGFRvIII expression. The additional samples included a tumor block from the patient that had tested positive for EGFRvIII. EGFRvIII transcript was not present in any of these additional tumor samples. Furthermore, AQUAnalysisH of the EGFRvIII-negative sample obtained from the single EGFRvIIIpositive patient showed this sample to have significantly lower wild-type EGFR protein expression. Taken together, these results suggest that EGFRvIII expression in OSCC is a rare event and most likely to be present in tumors which express very high levels of EGFR protein. Discussion We have developed a highly sensitive and specific real-time RTPCR assay for EGFRvIII detection. We validated our assay in a cohort of glioblastoma patients and compared its efficiency to conventional PCR and direct sequencing. Furthermore, in light of the conflicting reports regarding the ” frequency of EGFRvIII in HNSCC, we investigated the frequency of EGFRvIII in OSCC, the most prevalent form of HNSCC, using our novel technique. Despite the highly sensitive nature of our assay, we only detected a single EGFRvIII-positive patient in our OSCC cohort of 50 patients. This discrepancy between our novel real-time RTPCR assay results and the reported frequency of EGFRvIII in HNSCC may be due to differing EGFRvIII mutation frequencies in specific HNSCC subsites or selection bias inherent in early phase clinical trials. Patient eligibility for such trials was based on recurrent or metastatic disease status or failure of first-line therapy. Since EGFRvIII-positive tumors are expected to be less responsive to conventional therapies than EGFRvIII-negative tumors, it would be expected that EGFRvIII-positive tumors would be over-represent