Clinicopathologic features associated with cytohistologic noncorrelation in the diagnosis of cervical dysplasia: a study of concurrently obtained samples.

Abstract

On the basis of the unique subset of our database comprising patients who received a Papanicolaou test and a cervical biopsy during the same clinic visit, clinicopathologic factors potentially associated with cytohistologic diagnostic noncorrelation in these concurrently collected samples is investigated. In the first analysis, a selected group of variables potentially associated with noncorrelation relative to the diagnosis of high-grade dysplasia were examined, whereas the second analysis was centered on the effect of varying levels of Papanicolaou test inflammation (below the Bethesda 2001 threshold, ie, partially obscuring inflammation) on noncorrelation regarding the overall diagnosis of dysplasia. For the latter, the overall density of neutrophilic infiltrate on each Papanicolaou test slide was graded in a blinded fashion on a 4-tiered scale (no significant amount, mild, moderate, and severe), followed by a comparison of correlating and noncorrelating cases at each tier. There was no overrepresentation of noncorrelating cases in severe inflammation group. Indeed, correlating and noncorrelating cases did not significantly differ at any level of inflammation. In the first analysis, correlating (n = 17) and noncorrelating (n = 17) cases did not significantly differ in patient age, number of biopsies obtained, endocervical curettage status, glandular involvement by high-grade dysplasia, or frequency of background grade 1 cervical intraepithelial neoplasia. Noncorrelating cases were more likely than correlating cases to be grade 2 cervical intraepithelial neoplasia rather than grade 3 or worse (82% vs 41%, respectively, P = .02), which is probably attributable to the absence of a basaloid proliferation in the most superficial layers of grade 2 lesions. Furthermore, noncorrelating cases showed a comparatively smaller percentage of submitted biopsies involved by high-grade dysplasia (52% vs 75%, respectively, P = .03), consistent with smaller extent of disease. These findings further illustrate that lesional factors are important potential contributors to the false-negative rate of the Papanicolaou test.