Evaluation of an ELISA for p16INK4a as a screening test for cervical cancer

The low sensitivity of cytology and low specificity of human papillomavirus testing prompts searching for more accurate cervical cancer screening strategies. Our goal was to evaluate an ELISA-based test for p16(INK4a).

1,781 women undergoing routine screening provided cervical specimens for p16(INK4a) ELISA (original and enhanced versions of a prototype), liquid-based cytology, and Hybrid Capture II (hc2) testing. All women with a positive result and a random sample of those with negative results on all tests were referred for histologic diagnosis. Cervical intraepithelial neoplasia grade >or=3 (>or=CIN3) was the main outcome. The original analysis included all >or=CIN3 outcomes (n = 28). The a posteriori analysis was used to represent clinically relevant results with >or=CIN3 as outcomes only when detected after a positive screening test (n = 27).

Participants had a median age of 23 years. The prevalence of high-risk human papillomavirus DNA was 30.6%. In a posteriori analyses, the sensitivity and specificity for p16(INK4a) ELISA (>or=8 pg/mL cut-point), cytology, and hc2 were 50.9%, 58.1%, and 100.0%, respectively, and 90.4%, 89.3%, and 69.2%, respectively. Referral to colposcopy of women with positive results for hc2 and p16(INK4a) (enhanced ELISA, >or=6 pg/mL cut-point) had a sensitivity of 91.8% (95% confidence interval, 79.1-100.0%) and specificity of 86.0% (95% confidence interval, 82.0-89.0%). Results of the original analyses had similar specificity but substantially lower sensitivity due to the strong influence of the single CIN3 case with completely negative screening results.

An enhanced version of this prototypic p16(INK4a) ELISA showed promise in screening, particularly when combined with hc2.