Re: Detection of hypermethylated genes in women with and without cervical neoplasia.

Feng et al. ( 1 ) examined whether changes in DNA methylation of 20 genes, selected on the basis of their role in cervical cancer, could be used as markers of cervical intraepithelial neoplasia (CIN) and invasive cervical cancer (ICC). The authors found varying frequencies of promoter hypermethylation in these 20 genes in 319 exfoliated cell samples and matched tissue biopsy specimens. For four of these genes (DAPK1, RARB, TWIST1, and CDH13), increasing frequency of hypermethylation was statistically signifi cantly associated with increasing severity of disease. The estimated specifi city of the three-gene panel (DAPK, RARB, and TWIST1) was 95%, which is higher than specifi cities re ported for cytology and human papillomavirus (HPV) testing ( 1 ) . Although the study was well designed and a molecular genetic test with high specifi city such as this one is needed, some of the fi ndings are inconsistent with previously reported results and the conclusions may therefore not be valid. We believe that the following points are important to consider when interpreting the results of Feng et al. First, the high frequency of hypermethylation in the SFN, HIC1, and APC genes observed in the samples of histologically normal cells and atypical squamous cells of undetermined signifi cance (ASCUS) is unusual. Although low frequency of methylation in a group consisting of ASCUS compared with CIN and ICC is not unexpected because of the diagnostic dilemma involved in classifying these cytologic changes into specifi c diagnostic categories, none of the previous studies reported such a high frequency of promoter hypermethylation for HIC1 and APC in normal cervical epithelium or other normal tissues ( 2 – 5 ) . In fact, in our series of normal and ASCUS ( N = 59) and ICC ( N = 82) specimens, we observed a different pattern of promoter hypermethylation in these two genes. That is, we saw promoter hypermethylation in 4% and 0% of the normal/ ASCUS group and 18.3% and 11% in ICC for HIC1 and APC genes, respectively. Similarly, the authors reported a higher frequency of promoter hypermethylation of SFN than has been seen in previous studies ( 6 , 7 ) . Misclassifi cation of methylation could occur due to amplifi cation of nonspecifi c targets in the genome, which ultimately makes the results on these three genes diffi cult to interpret. Whether these genes have utility in cervical cancer screening needs to be confi rmed in other studies. A number of previous studies have also shown a high frequency of promoter hypermethylation of DAPK, CDH1, RARB, and p16 in ICC ( 2 – 4 ) . However, the frequency of promoter hypermethylation of these genes in CIN lesions was not well studied. Thus, well-designed studies of CIN are required to determine whether inclusion of additional genes, such as HIC1, to the three-gene panel proposed by Feng et al. might improve its sensitivity in detecting high-risk lesions that progress. In the context of previously published studies and our series, it is premature to propose a panel of genes that would be the best for screening CIN3/CIS and ICC or for identifying high-risk CIN. Before such a test based on promoter hypermethylation can be proposed for screening cervical cancer, further studies are required on a larger series of specimens from different geographical populations, and these studies should carefully address the issue of false-positive rates.