Diagnostic accuracy of colposcopy in relation to human papillomavirus genotypes and multiple infection.

OBJECTIVE: The aim of this study is to evaluate the diagnostic accuracy of colposcopy for cervical intraepithelial neoplasia grade 3 or worse (CIN3+) in relation to the detection of human papillomavirus (HPV) type 16 and multiple HPV infection. METHODS: A cohort study of 2526 subjects attending a colposcopic service because of cytological abnormalities. HPV genotypes were identified using the INNO-LIPA genotyping system. RESULTS: The final colposcopic/pathological diagnoses were as follows: 1282 (50.8%) negative, 709 (28.1%) CIN1, 169 (6.7%) CIN2, 318 (12.6%) CIN3 and 48 (1.9%) invasive cervical cancer, respectively. Among women with ASCUS/LSIL, assuming any colposcopic abnormality as a cut-off, there were no significant differences in the sensitivities (83.8%, 95% CI=76-89.6 as compared to 84.1%, 95% CI=73.2-91.1, p=0.9) and ROC curves (0.61, 95% CI=0.58-0.65 as compared to 0.59, 95% CI=0.54-0.64, p=0.5) in the detection of CIN3+ lesions between subjects with single and multiple high-risk infection, and between subjects infected by HPV16 (83.1%, 95% CI=73.7-89.7, ROC=0.59, 95% CI=0.54-063) or other high-risk HPVs (84.7%, 95% CI=75.6-90.8, ROC=0.62, 95% CI=0.58-0.66, p=0.8 and p=0.6 compared to HPV16). After correction for confounders, the odds ratios of CIN3+ associated with any abnormal colposcopic findings were 2.47 (95%CI=1.44-4.23, p=0.001) among HPV16 positive, 3.34 (95% CI=2.16-5.42, p<0.001) among other high-risk HPVs and 1.3 (95% CI=0.72-2.48, p=0.36) among subjects with negative/low-risk HPVs. CONCLUSION: In routine clinical practice, multiple infection or HPV16 positivity did not affect colposcopic accuracy in the diagnosis of CIN3+ lesions. The sensitivity of colposcopy was poor among subjects who were uninfected or infected by low-risk HPV genotypes.