Diagnostic accuracy evaluation fo screening tests for detecting high grade cervical leasions in Flanders

Abstract

Cervical cancer is the second most common cancer among women worldwide and about 54% of the disease is caused by the most common high-risk human papillomavirus (HR-HPV) infection, HPV16. Screening methods such as cytology and HR-HPV testing present some important shortcomings and require additional testing to distinguish transforming from productive HPV infections. It is apparent that future screening should combine multiple diagnostic tests to identify women at risk for cervical cancer and reduce unnecessary follow-up of transient HPV infections. It emerged from this thesis that via the aid of a logistic regression model and the AUC as discriminatory tool of classification accuracy, markers combinations such as the viral load and the amount of HPV E2 genes per cell on the one hand, and the viral integration state and cytological outcome on the other hand could enhance the sensitivity of cervical cancer screening programmes in Flanders. More specifically, it resulted that the combinati