Novel integrative methods for gene discovery associated with head and neck squamous cell carcinoma development.

OBJECTIVE To find head and neck squamous cell carcinoma (HNSCC) specific genetic changes using integrative genetics. DESIGN Genetic analysis. PATIENTS Three separate cohorts of patients with primary HNSCC were evaluated for expression-microarray of 33 000 genes (8 patients), quantitative real-time polymerase chain reaction (qRT-PCR) (36 patients), and quantitative DNA/qRT-PCR (12 patients). Controls with normal upper-aerodigestive mucosa were evaluated for expression microarray (6 patients) and qRT-PCR (7 patients). INTERVENTIONS We utilized (1) prior reports of DNA loss and gain HNSCC accompanied by comparative genomic hybridization high-definition array data of the entire human genome, (2) a genome-wide survey of cancer-specific DNA mutations from the consensus cancer coding sequence (13 023 genes), and (3) our RNA expression microarray data of 33 000 genes to define candidate oncogenes activated by amplification or candidate tumor suppressor genes inactivated by deletion. MAIN OUTCOME MEASURES Gene expression in tissue measured by quantitative reverse transcriptase PCR. Gene copy number was measured by quantitative PCR. RESULTS We found 20 genes that were in areas of demonstrated amplification or deletion overlapping with the somatic mutants from genome-wide screening of the consensus DNA cancer coding sequence reported by Sjöblom et al. Three were chosen for further study based on expression differences and proof of cancer causation from in silico study: RUNX1T1, RFC4, and DLEC1. From 12 patients with HNSCC, matched tumor DNA/RNA and leukocyte-derived DNA were studied. Six of 12 (50%) of the tumors demonstrated amplification of the RUNX1T1 locus (P = .01), and 4 of those 6 (67%) demonstrated upregulated transcription of this gene (P = .02). Five of 12 (42%) of the tumors demonstrated amplification of the RFC4 locus (P = .03), and 1 of those 5 (20%) demonstrated upregulated messenger RNA (mRNA) transcription of the gene (P = .60). Four of 12 of the tumors (33%) (P = .05) demonstrated deletion in the DLEC locus (consistent with previously published 3p22 loss of 40%), and 3 of those 4 (75%) demonstrated reductions in mRNA expression (P = .06). CONCLUSION With the advent of high-throughput techniques to study cancer genetics, novel comparisons of large data sets using integrative methods may elucidate genetic alterations in HNSCC cancer.