Diagnostic precision and accuracy in interpretation of specimens from cancer screening programs

Few areas in modern medicine provide a better example of evidence-based data than the anatomic pathologists' classification of fully developed malignancies. Beginning with the Papanicolaou (Pap) smear, morphologic tools also were applied to specimens obtained in cancer screening programs directed at large, asymptomatic patient populations. The Papanicolaou test was quickly responsible for a reduction in the incidence of invasive squamous cell carcinoma of the cervix, proving the concept that screening could interdict the development of advanced malignancy. Other screening programs followed the Papanicolaou test initiative, producing a revolutionary change in the specimens submitted to pathologists. Cancer screening generates specimens containing morphologic deviations from normal that are thought to put currently healthy patients at risk of future malignancy. However, translating morphologic findings in such samples into risk estimates raises a number of statistical and ethical problems. When diagnostic thresholds are set to favor specificity, unwanted false-negative results accrue. Conversely, aiming at sensitivity over specificity is associated with biologically false-positive results, which are likewise undesirable. Pathologists interpreting these specimens find themselves facing a screening paradox. These screening programs attempt to discover the very earliest changes of neoplastic transformation, but these same earliest changes are inherently the most difficult to identify with precision and accuracy. This paper discusses these challenges.