Abstract
Nasopharyngeal carcinoma (NPC) is a human malignant tumor with a high incidence and a poor prognosis in Southern China and South-eastern Asia. In this study, we comprehensively analyzed the gene expression profiles in 24 samples of primary differentiated-type nonkeratining NPC (DNK-NPC) tissues, 24 samples of normal nasopharyngeal tissues and 4 DNK-NPC cell lines using cDNA microarray technology and bioinformatics methods. We found expression level of some genes was wildly alerted in the DNK-NPC samples. In addition, our hierarchical clustering analysis revealed 2 distinctive subtypes of gene expression patterns in DNK-NPC tissue samples. The discriminator genes were identified using a signal-to-noise (S(2)N) algorithm by permuting of the data set 10,000 times. To further characterize the clinical relevance of the tumor subtypes, we evaluated a surrogate marker, CCND2, differentially expressed between the 2 tumor subgroups by using immunohistochemistry in an independent set of 137 DNK-NPC samples. CCND2 was highly expressed in the subgroups with "aggressive" features and was associated with T classification (p = 0.006) and clinical stage (p = 0.013). Patients with high level of CCND2 expression had poorer overall survival than those with low level (p = 0.034). Our results suggest that DNK-NPC can be classified into 2 subtypes based on gene expression patterns, which can be used in determining prognosis and treatment of the tumor.