RESUMEN
Numerous molecular biomarkers have been suggested for early detection of cervical cancer, but their usefulness in routinely collected exfoliated cells remains uncertain. We used quantitative reverse transcription-PCR to evaluate expression of 40 candidate genes as markers for high-grade cervical intraepithelial neoplasia (CIN) in exfoliated cervical cells collected at the time of colposcopy. Samples from the 93 women with CIN3 or cancer were compared with those from 186 women without disease matched (1:2) for age, race, and high-risk human papillomavirus status. Normalized threshold cycles (C(t)) for each gene were analyzed by receiver operating characteristics to determine their diagnostic performance in a split sample validation approach. Six markers were confirmed by an area under the curve >0.6 in both sample sets: claudin 1 (0.75), minichromosome maintenance deficient 5 (0.71) and 7 (0.64), cell division cycle 6 homologue (0.71), antigen identified by monoclonal antibody Ki-67 (0.66), and SHC SH2-domain binding protein 1 (0.61). The sensitivity for individual markers was relatively low and a combination of five genes to a panel resulted in 60% sensitivity with 76% specificity, not positively increasing this performance. Although the results did not indicate superiority of RNA markers for cervical cancer screening, their performance in detecting disease in women referred for colposcopy suggests that the genes and pathways they highlight could be useful in alternative detection formats or in combination with other screening indicators.