The pathological and radiological features of screen-detected breast cancers diagnosed following arbitration of discordant double reading opinions

Abstract

To compare the mammographic background pattern, the mammographic and the pathological features of screen-detected cancers diagnosed following arbitration of discordant double reading opinions with screen-detected cancers diagnosed following concordant double reading. Between April 2002 and December 2003, 249 patients were diagnosed with screen-detected malignancies following concordant double reading. In the same period 38 patients were diagnosed with screen-detected malignancies after their mammograms had undergone arbitration prior to recall. Mammograms of both groups of patients were reviewed retrospectively and the mammographic features documented. Histological data for both groups were also compared. Cancers detected following arbitration were more likely to manifest as parenchymal distortions (44 versus 8%, p<0.001) and less likely to manifest as spiculate masses (19 versus 42%, p=0.014). Arbitration cancers were less likely to be detected in fatty breasts (4 versus 29%, p=0.01). Arbitration cancers were smaller (p=0.045). Lobular cancers were commoner in the arbitration group, although this was of borderline significance (19 versus 8%, p=0.057) There was no significant difference in patient age, tumour grade or lymph node stage between the two groups. Cancers detected following arbitration are smaller and more likely to manifest as a parenchymal distortion compared with cancers detected by both readers. Arbitration cancers have broadly similar prognostic features to cancers detected by concordant double reading. It is estimated that approximately 11% more cancers are detected as a result of double reading with arbitration compared with single reading alone, after taking into consideration second reader bias.