A DNA methylation pattern similar to normal tissue is associated with better prognosis in human cervical cancer.

Promoter hypermethylation has been recognized to play an important role in carcinogenesis. We analyzed the methylation status of 25 genes in 14 normal cervical tissue specimens and in 65 tissue specimens from cervical cancer patients using the MethyLight technique. Most of the analyzed genes have been shown to be methylated in various cancers. RB1 was never methylated in any analyzed cervical tissue. DNA methylation status of the remaining 24 genes in every tissue sample was subsequently analyzed using unsupervised agglomerative hierarchical cluster analysis to group specimens and CpG regions. We observed four clusters. All normal cervical tissue specimens were grouped together in one cluster. Neither grade, nor histological type nor age demonstrated any significant association with clusters formed. Interestingly, statistically significantly less patients whose tumor DNA methylation pattern clustered together with normal cervical tissue died within our observation period, as compared to those patients out of the three remaining clusters (P < 0.03) Cervical cancer patients, whose DNA methylation pattern clustered together with normal cervical tissue revealed a strong trend to better survival (P = 0.066) compared to patients grouped in the remaining cluster. This study shows for the first time that working solely with DNA methylation pattern a subgroup of cervical cancer patients can be defined that demonstrated strong similarity to non-neoplastic probands and had a better prognosis. Copyright 2004 Elsevier Ltd.