Although recent studies have suggested that p16INK4a may be a useful surrogate biomarker of cervical neoplasia, Ki-67 and human papillomavirus testing have also been shown to be useful in detecting neoplasia. To help delineate the utility of p16INK4a, biopsy samples (n = 569: negative, 133; reactive, 75; atypical, 39; low grade, 76; moderate, 80; and severe intraepithelial neoplasia, 113; also, squamous cell carcinoma, 46; adenocarcinoma, 7) were analyzed by immunohistochemistry for expression of p16INK4a and Ki-67 (n = 432), as well as by in situ hybridization for human papillomavirus Type 16 (n = 219). Testing for high-risk human papillomavirus types by polymerase chain reaction and HybridCapture2 was performed on concurrent cervical swab specimens. Recuts of the original blocks were reexamined (n = 198). Endometrial biopsies (n = 10) were also analyzed for p16INK4a expression. Degree of p16INK4a and Ki-67 expression correlated with degree of cervical neoplasia (P < .001) and with presence of high-risk human papillomavirus types (P < .001). There was no relationship between p16INK4a overexpression and inflammation or hormonal status. Ki-67 expression correlated with inflammation (P = 0.003) and was expressed in more reactive and atypical lesions than p16INK4a (P = 0.008). Probes for human papillomavirus 16 stained 54% of cervical neoplastic lesions; the degree of staining correlated significantly with degree of neoplasia (P < .001) and p16INK4a staining (P < .001). Interobserver reproducibility was substantial for p16INK4a and Ki-67 interpretation (weighted κ: 0.74 and 0.70, respectively). Expression of p16INK4a was observed in all endometrial biopsies. Compared with Ki-67 expression and detection of high-risk human papillomavirus, p16INK4a was less likely to be positive in samples from women with negative, reactive, and atypical biopsies. Although expression of p16INK4a in endometrial epithelium may be problematic in terms of screening, the potential of p16INK4a as a screening test warrants investigation.
