Emily E. King


March 2010

Document Type


Degree Name



Dept. of Public Health and Preventive Medicine


Oregon Health & Science University


Introduction: Cervical intraepithelial neoplasia (CIN) diagnostic methods are suboptimal due to limitations in reproducibility and accuracy, which affects clinical management and understanding of risk factors for CIN. These limitations are especially relevant for the histologic diagnosis of moderate dysplasia (CIN 2). We hypothesize that molecular markers, such as p16 and Ki-67, may improve diagnostic reproducibility and accuracy and also lead to a better understanding of cervical cancer epidemiology. Methods: A randomly selected retrospective cohort of 300 women with cervical dysplasia diagnosed by colposcopic biopsy and at least five years of clinical follow-up at Kaiser-Permanente Northwest in Portland, Oregon was obtained. Two experienced gynecologic pathologists (A & B) independently reviewed histologic sections of the colposcopic biopsy material while blinded to each other’s assessments and long-term clinical outcome. Diagnoses made using routine H&E stained histologic slides were compared to diagnoses made using 1) H&E plus p16 and 2) H&E plus p16 and Ki-67. Cohen’s kappa statistic was used to quantify diagnostic reproducibility. Measures of test accuracy were determined by comparing colposcopic biopsy diagnoses to a consensus outcome from 5 years of clinical followup. Multivariate logistic regression was used to model the relationship between risk factors and reviewer's CIN diagnoses. Results: The reproducibility of CIN 2 diagnosis was significantly improved by using H&E plus p16 (κ=0.4783) compared to H&E only (κ=0.4041, p<0.05). The reproducibility of CIN 2 diagnosis using H&E plus p16 and Ki-67 was significantly improved compared to H&E only (κ=0.5204, p<0.05), but not compared to H&E plus p16 (κ=0.4783, p>0.05). H&E plus p16 significantly improved sensitivity and negative predictive value compared to H&E only for both reviewers. Unadjusted odds ratio estimates for CIN 2+ if exposed to high (>$45,000 annual) family income was not significant (OR 0.892, p >0.05) using H&E only as diagnostic method but was significant when using H&E plus p16 (OR 0.585, p <0.05). Multivariate logistic regression showed a similar trend in magnitude of ORs in this pilot study. Conclusions: This pilot study suggests that misclassification in diagnosis of CIN may have significant ramifications for epidemiologic research. For example, p16 contributed to significant changes in the estimated risk of CIN 2+ for women with low family income, which would have been missed by conventional diagnostic methods. It demonstrated that improvements in diagnostic precision and accuracy gained with markers like p16 may provide a "new gold standard" with which to evaluate strength of associations.




School of Medicine



To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.