Incident screening cancers detected with a second mammographic view: pathological and radiological features

Previous studies and epidemiological data from the UK National Health Service Breast Screening Programme (NHSBSP) have indicated significantly increased sensitivity for cancer detection with two-view rather than one-view mammographic screening. The radiological and pathological features of these extra cancers have not been previously reported in detail. We have studied all screen-detected cancers found as incident cases in the South West London Breast Screening Service between 1994-1997 on the second round of screening. To assess the effect of two-view versus one-view mammography on cancer detection, these cases were mixed with controls in a 1:2 ratio in nine test sets and each set read independently by three film readers. They initially read the oblique view, then the craniocaudal views, and recorded abnormalities on the films and likelihood of recall. Radiological and histological data were recorded for each case. Using two views, 8.9% (P < 0.05) more invasive cancers were detected. The sensitivity increase was highest for invasive cancers less than 10 mm (11%) and cancers of low grade (11.9%). These sensitivity increases may underestimate the increase in 'real life' because of over-recalling of normal mammograms, particularly with one view, under study conditions. The most significant radiological feature of invasive cancers was an irregular mass, which, seen on one view had a positive predictive value of 82.2% and 89.9% with two views. The craniocaudal view was helpful, firstly, because some cancers were not visible on the oblique view only. Secondly, benign appearing round masses and asymmetric densities seen with the oblique view only were resolved as more suspicious irregular masses with both views, leading to recall. In conclusion, there are cancers that cannot be adequately visualized on the oblique view alone. These are most commonly the small invasive cancers, which are of the greatest prognostic significance in breast cancer screening.