Detection of human papillomavirus deoxyribonucleic acid in exfoliated cervicovaginal cells as a predictor of cervical neoplasia in a high-risk population.

Specific types of human papillomavirus are currently implicated as etiologic agents of precancerous and cancerous lesions of the cervix. We have previously described the use of cervicovaginal lavage and molecular hybridization to detect human papillomavirus infections of the cervix. We report here the predictive value of this method of human papillomavirus detection to identify women with biopsy proved dysplastic and cancerous lesions of the cervix. One hundred ninety-one women from a city hospital colposcopy clinic underwent concurrent Papanicolaou smear, cervicovaginal lavage, and coloposcopically directed cervical biopsy. Human papillomavirus deoxyribonucleic acid was detected in 114 (59.7%) of these women. Of the positive results, human papillomavirus type 16 accounted for 23.7%, human papillomavirus type 18 for 10.5%, human papillomavirus type six or 11 for 6.2%, related human papillomavirus types for 52.6%, and 7.0% contained more than one type. The distribution of human papillomavirus types was similar in both women younger than 40 years of age and in older women. Eighty-nine of 128 (69.5%) women less than 40 years old with cervical lesions had positive findings of human papillomavirus, and 18 of 29 (62.1%) older women with cervical lesions had positive findings of human papillomavirus. Detection of human papillomavirus types 16 and 18 identified only 35 of 157 (22.3%) women with cervical lesions. The sensitivity of detecting all types of human papillomavirus as a predictor of a biopsy proved lesion (68.0%) was comparable with the sensitivity of cytologic examination alone (74.0%). However, human papillomavirus detection combined with the Papanicolaou smear provided an increased overall sensitivity of 89.3% (p less than 0.01). In fact, women either positive for human papillomavirus or having abnormal cytologic findings were 11.8 times more likely to have a biopsy proved cervical lesion than human papillomavirus-negative women with negative cytologic results (95% confidence interval for odds ratio: 5.3 to 26.6). We conclude that the sensitivity of cytologic examination plus human papillomavirus detection is superior to the use of either cytologic studies or human papillomavirus detection alone in identifying patients with cervical lesions.