Triaging abnormal cervical cancer screening tests using p16INK4a detection by ELISA on fresh cervical samples.

PROBLEM Cervical cancer screening strategies in the United States include cotesting (human papillomavirus (HPV) with cytology), primary HPV with genotyping and reflex cytology, and cytology alone. An ongoing challenge is the appropriate triage of patients to colposcopy to those at highest risk. We investigated whether incorporation of p16INK4a immunodetection by enzyme-linked immunosorbent assay (ELISA) on fresh cervical samples obtained at the time of screening could improve appropriate referral to colposcopy. METHOD OF STUDY A derivation group comprised of cervical swabs collected from subjects with high-grade dysplasia or cancer (positive control) and from subjects with negative screening history (negative control). Samples collected from colposcopy were used to evaluate the existing screening strategies individually and with incorporation of p16INK4a ELISA. Histology was used as the gold standard. RESULTS Among 163 subjects recruited, 138 were included. In the derivation group, mean p16INK4a level was 2.86 ng/mL (n = 31) and 0.58 ng/mL (n = 20) among positive and negative controls respectively (p = 0.002) with an area under the receiver operator characteristic curve of 0.79 (p < 0.001). Among colposcopy subjects, sensitivity/specificity for cotesting, primary HPV, and cytology were 94%/42%, 88%/45%, and 88%/49%, respectively. Incorporation of p16INK4a resulted in similar sensitivity and improved specificity (cotesting+p16 88%/58%, primary HPV+p16 88%/57%, cytology+p16 81%/62%; p = 0.23/p = 0.008) with decrease in colposcopy referrals by 15% to 22% (p = 0.01). CONCLUSIONS These results demonstrate the feasibility of quantifying p16INK4a by ELISA in fresh cervical samples, and its potential as an adjunct to existing screening strategies in the identification of high grade-dysplasia while reducing the number of colposcopic referrals.