Purpose/Objective: Our group has had a long-time interest in identifying prognostic variables, and developing novel therapeutic strategies for nasopharyngeal carcinoma (NPC). In an effort to achieve these objectives using a genome-wide approach, we undertook micro-array expression profiling studies using primary human NPC biopsy specimens. Materials/Methods: Freshly obtained nasopharyngeal biopsies were flash frozen in liquid nitrogen upon clinical suspicion for NPC. The histologic diagnosis was confirmed by one head & neck pathologist, and scored for proportion of the sample occupied by cancer. Micro-array analyses were conducted on 19 patients with NPC, 11 samples of lymphoid hyperplasia (LH), and 6 normal. Amongst these 36 samples, six sets were paired, derived from the same patients (3 normal & 3 NPC, plus 3 LH & 3 NPC). The median age of the NPC population was 54 years, with initial stage ranging from I to IV, including one patient with metastatic disease. All NPC patients (except the M1) were treated with curative intent, with radiotherapy alone (4), or combined with chemotherapy (5-FU + Cisplatin) for 14 patients. Total radiotherapy (RT) dose to the primary site ranged from 66–70 Gy, at 2 Gy per day, using either conformal or IMRT techniques. Total RNA was extracted using Trizol, purified using the Qiagen RNeasy kit, and then hybridized onto the Affymetrix U133A oligonucleotide microarray chip. Hybridization intensity was normalized using Affymetrix Microarray Suite, and then normalized and log-transformed to calculate the difference between normal and cancer tissues. Probe sets differently expressed were then selected after the unsupervised split clustering analysis. Probe sets with a 2-fold change in expression between non-recurrent and recurrent samples were also identified using Significance Analysis of Microarrays (SAM). Identified genes from the micro-array analysis were confirmed by both QRT-PCR and immunohistochemistry, using published primer sequences, and commercially available antibodies respectively. Results: Comparison of the expression of 18,400 genes in 19 NPC and 6 normal biopsies, using unsupervised clustering, revealed a clear separation between these two categories of specimens. Significance Analysis of Microarrays (SAM) derived from cancer and normal samples discovered 1089 differentially regulated genes, with 411 up- and 678 being down regulated. Pathway Assist analyses revealed multiply dysregulated cascades, with NFkB2 up-regulation appearing to play a central role, resulting in increased resistance to apoptosis, and changes in integrin and WNT/b-catenin signaling. Both pro and anti-apoptotic signals were evident, demonstrated by over-expression of p53, Bax, Bim, along with anti-apoptotic Bcl-2, Bfl-1 and survivin transcripts. With a median follow up time of 2.3 years, 5 of 18 patients experienced a relapse. Sub-analysis of the initial cancer specimens derived from patients with or without relapse identified 108 differentially expressed genes, some already described to be associated with poor outcome in other malignancies. Both QRT-PCR and immunohistochemistry confirmed the over-expression of NFkB2, bcl-2, and survivin on the apoptotic pathway. On the WNT/b-catenin signaling cascade, under-expression of WIF1, and over-expression of TCF3 and LEF1 were likewise confirmed, as were over-expression of integrin B1, integrin B3, and integrin av on the integrin pathway. Conclusions: Our data provide novel insight into the mechanisms leading to the development and progression of NPC. These data may impact on the development of novel therapeutic strategies for NPC therapy, and the design of diagnostic tools to identify patients at risk of recurrence with this disease.