Quantitative assessment of DNA methylation for the detection of cervical neoplasia in liquid-based cytology specimens

Abstract

Previously, we have shown that methylation-specific PCR (MSP) analysis of a key panel of genes may be useful as an ancillary tool for diagnosing squamous cell carcinomas (SCC) and high-grade squamous intraepithelial lesions (HSIL) in cervical scrapings. Because quantitative MSP (QMSP) is more suitable as a screening tool than conventional MSP, we investigated the diagnostic role of QMSP for the detection of SCC and HSIL in cervical scrapings. A quantitative multiplex-MSP approach was used to examine promoter methylation of five genes (APC, HIN-1, RAR-beta, RASSF1A, and Twist) in biopsy-confirmed SCC (n = 63), HSIL (n = 45), low-grade SIL (LSIL, n = 26), and negative (n = 28) liquid-based cytology samples. For four genes (HIN-1, RAR-beta, RASSF1A, and Twist), the methylation levels among four groups were significantly different (p < 0.001 for each). Methylation levels of HIN-1, RAR-beta, RASSF1A, and Twist were increased in HSIL and SCC samples, compared with either negative or LSIL samples. However, methylation levels were not significantly different between SCC and HSIL, with the exception of RASSF1A. Receiver-operating characteristic analysis demonstrated that HIN-1, RAR-beta, RASSF1A, and Twist had the ability to distinguish HSIL/SCC from LSIL/negative samples. The two-gene combination (RASSF1A/Twist) showed the best performance in distinguishing HSIL/SCC from LSIL/negative samples. The estimated specificity of this two-gene panel for detecting HSIL/SCC was 90.7%, and its sensitivity was 74.1%. These results suggest that quantitative detection of aberrant DNA methylation in cervical scrapings may be a promising high-throughput approach for the diagnosis of HSIL/SCC.