Head and neck squamous cell carcinoma (HNSCC) is the sixth most common cancer in the world, and is highly mortal due to tumor recurrence and metastasis. Invasion of tumor cells into lymph nodes is one of the characters of regional metastasis and DNA hypermethylation is characterized in all types of human cancers. We therefore assessed whether epigenetic modification of DNA methylation is correlated to HNSCC metastasis, and evaluated the potential clinical implications of DNA methylation-based biomarker. Methylation array (Illumina infinium beadsarray) was applied to profile aberrant methylated CpG loci in HNSCC with lymph node invasiveness. Fourteen HNSCC tumors and two normal oral tissues were included. ≤ value, the indicator of methylation status of each tested CpG locus, was derived with normalization. A cluster of 111 CpG loci was selected based on the difference between the mean ≤ value of HNSCC tumors with and without regional metastasis (Δβmetz minus non-metz ≥ 0.2 or ≤ −0.2). Hierarchical diagram revealed distinct methylation signature was observed between tumors with or without regional metastasis. Ingenuity pathway analysis showed that the major diseases associated were neurological disorder and endocrine system disorder; and the top canonical pathway was Huntington's disease signaling. Of the top aberrant hypermethylated alleles, p16 and brain-derived neurotrophic factor (BDNF) were selected for further investigation on clinical implication. The methylation status of p16 (n=107) and BDNF (n=88) in HNSCC tumors was assayed by methylation-specific PCR. The primer sequences of both alleles were adopted from published report. Statistic analysis (t-test, χ2 or Fisher's exact, logistic regress, Kaplan-Meier survival, and Cox regression) was performed using SPSS. Hypermethylation of p16 or BDNF were not associated with stage, metastasis at diagnosis, and differentiation. Hypermethylation of p16 was significantly associated with age (p=0.008, t-test) and lymph node invasion (p=0.021, χ2-test), in contrast to distant metastasis of follow-up (p=0.061, χ2-test). However, hypermethylation of BDNF was associated with distant metastasis of follow-up (odds ratio: 4.96, 95% CI: 1.28-18.87, p=0.002), and combination with p16 improved the prediction (odds ratio: 10.31, 95% CI: 1.68-62.5, p=0.012). Hypermethylation of BDNF was associated with poor survival of patients (odds ratio: 2.80, 95% CI: 1.17-6.71, p=0.021), and hypermethylation of both alleles increased the risk (odds ratio: 3.81, 95% CI: 1.71-8.49, p=0.001). Together, our data suggest hypermethylation of p16 and BDNF may be potential biomarker associate with poor prognosis of HNSCC. The results also show the association of p16 and BDNF in gain of metastatic properties, which warrants further investigation.

Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 4038. doi:1538-7445.AM2012-4038