Translation gene expression profiling into the identification of high frequency of methylation markers in cervical neoplasia: novel panel of 3 genes of the CpG island methylator phenotype(CIMP) for high-grade squamous intraepithelial (HSIL) or worse.

A101

We have developed a systematic and stepwise genome-scale screening system to discover a new panel of genes enable to represent the high frequency of promoter methylation in the cervical cancer and its precursor lesions including carcinoma in situ (CIS) and high-grade squamous intraepithelial(HSIL). Considering that aberrant promoter methylation-mediated silencing of tumor suppressor genes has significant tissue specificity, we firstly investigated a subset of genes relatively decreased in gene expression in cervical cancer(CC) and biopsy specimens of precursor lesions (CC in 8, CIS in 2, and HSIL in 5) compared with normal-appearing biopsies in 7 using 35K microarray. We secondly analyzed expression profiles to identify genes epigenetically reactivated in four CC derived cell lines after treatment with methylation inhibiting agent, 5-aza-2'-deoxycytidine. We obtained 53 genes by cross-comparison two gene lists and reduced to 28 through two additional steps, in silico search for whether CpG islands exist or not and following examination for methylation status of each gene by PCR after digestion with restriction-enzymes, HpaII and MspI, in 4 CC cell lines. To determine the degree of methylation frequency of 28 filtered genes in CC or its precursor cells present at exfoliated cell samples, each gene was examined with DNA of smears from patients with CC in 1, CIS in 2, HSIL in 3, LSIL in 3, and normal in 3. We selected a panel of 10 genes for further clinical validations because they showed the high frequency of methylation in HSIL or worse while very low in LSIL or less. We then evaluated the panel as a CpG islands methylator phenotype (CIMP) with 134 cytological smears diagnosed as normal in 70, atypical squamous cell of undetermined significance (ASCUS) in 5, LSIL in 26, HSIL in 11, CIS in 10 and CC in 12. The frequency of methylation was overall increased with increasing grade of histological abnormality. Five out of ten showed a statistical significance (p < 0.0001) of the trend in the proportion of samples while rest of them displayed less significant rather monotonous pattern. We evaluated its performance in detecting HSIL or worse with independent sets of clinical samples of 131 exfoliated cells from a similar proportion of different grade of cervical lesions. Three out of five genes showed statistically a good correlation between two independent evaluation. Therefore, we determined those three genes as the best panel of CIMP for HSIL or worse. The sensitivity and specificity of the panel of three genes in detecting HSIL or worse in combined data were estimated as 72.4% [95% confidence interval (CI) = 59.1% to 83.3%] and 89.9% (95% CI = 84.9% to 93.6%), respectively. Methylation analysis of the panel of three genes discovered through genome-wide systematic identification and stepwise filtering approach in this study may be a potential diagnostic tool to identify precancerous lesions at high risk to progression.