Quantitative promoter hypermethylation analysis of cancer-related genes in salivary gland carcinomas: Comparison with methylation-specific PCR technique and clinical significance

Abstract

Purpose: To compare the methylation status of tumor-associated genes by quantitative pyrosequencing and qualitative methylation-specific PCR (MSP) techniques and to correlate the results with clinicopathologic features and patients outcome to determine which method might have greater clinical utility. Experimental Design: The hypermethylation status of the retinoid acid receptor beta 2 (RAR beta 2), RAS association domain family 1A (RASSF1A), O-6-methylguanine-DNA methyltransferase (MGMT), and E-cadherin genes was analyzed in five salivary carcinoma cell lines and 69 human salivary gland carcinoma specimens by pyrosequencing and MSP techniques. The two datasets were compared by linear regression. Correlations between methods and with clinicopathologic characteristics were assessed by Pearson's chi(2) test or the two-tailed Fisher exact test, as applicable, using cutoff points determined from the regression curves and empirical fitting. We also investigated the effect of demethylating agents on methylated genes in cell lines to assess their effect on the expression of these genes. Results: Overall, regression analysis indicated high degrees of correlation of the two methods for measurement of methylation for the RARb2, RASSF1A, and MGMT genes (adjusted R-2 = 0.319, 0.835, and 0.178; P < 0.001, <0.001, and 0.0002, respectively) among the 69 tumors tested. However, the pyrosequencing technique yielded four more instances of methylation above background levels than MSP for RAR beta 2 and three more for RASSF1. Methylation of either RAR beta 2 and RASSF1A alone or both by pyrosequencing were correlated with tumor type (P = 0.027, 0.014, and 0.012, respectively). Methylation of RAR beta 2 alone and in combination with RASSF1A by pyrosequencing were also significantly correlated with tumor grade (P = 0.014 and 0.011, respectively) and 3-year survival (P = 0.002 and 0.004, respectively). The survival curves of patients who had hype rmethylation at both RAR beta 2 and RASSF1A were significantly lower than those of patients who had hypermethylation at neither or just for the RASSF1A (P = 0.008 and 0.007, respectively). 5-Azadeoxycytidine treatment of methylated cell lines led to the reactivation of RAR beta 2 expression in only one of the five cell lines. Conclusions: (a) Although the methylation status of RARb2, RASSF1A, and MGMT genes by both techniques were significantly correlated, pyrosequencing is generally more sensitive and its results correlate better with the clinical variables than those of MSP. (b) The methylation level of the RAR beta 2 and/or RASSF1A by pyrosequencing is significantly associated with aggressive tumor phenotypes and patients survival.