Abstract
BACKGROUND: Persistent infection %by with high risk human papillomavirus (hrHPV) types causes cervical cancer but most women who test positive are at very low risk of neoplasia. Strategies are needed which can retain high sensitivity of hrHPV testing but reduce the number of false-positives. We showed previously that a combination DNA methylation triage assay for HPV types 16, 18 and 31 and human gene EPB41L3 was useful to identify high grade cervical lesions. OBJECTIVE: Assess whether measurement of DNA methylation in HPV type 33 can improve the previous classifier. METHODS: A London colposcopy referral group of 1493 women of whom 556 (37%) had histologically-confirmed CIN (cervical intraepithelial neoplasia) 2 or 3 that included 114 HPV33 positive women with methylation measured for three L2 CpGs 5557, 5560 and 5566. Discrimination performance was assessed for the new classifier S5, built by adding HPV33 to the earlier classifier. RESULTS: HPV33 methylation measurement improved prediction among HPV33 positive women. Receiver operating characteristic analyses showed an area under the curve (AUC) for HPV33 methylation of 0.68 (95% CI 0.57-0.78). The earlier risk score was significantly improved by HPV33 methytlation (AUC = 0.82 vs 0.80; P < 0.001). For 90% sensitivity the specificity for CIN2/3 was 49% (95% CI 46-52%). CONCLUSIONS: Measurement of HPV33 DNA methylation contributes independent diagnostic information to EPB41L3 and HPV16, HPV18 and HPV31, and is superior to genotyping. Other HPV and human methylation target regions might be useful to further improve S5.