The visibility of cancer on earlier mammograms in a population-based screening programme.

The aim of this study was to examine how frequently the later-round screen-detected and interval breast cancers were visible in earlier screening mammograms by retrospective review and to compare their radiological and clinicopathological features with those diagnosed by primary screening. In a population-based mammography screening programme 63,731 women aged 50-59 years were invited and 56,158 examinations were carried out in the period 1987-1992 in the Tampere area in Finland. A total of 276 breast cancers were detected, of which 131 were diagnosed on later screening rounds or were interval cancers. A retrospective review of previous screening mammograms was carried out in 130 cases by the radiologist who diagnosed the breast cancer and thus knew the exact location of the tumour, no blinded review was carried out. 43 (33%) cancers were visible, 84 (65%) were not visible and 3 (2%) not included on the mammogram in a retrospective review. Later round screen-detected cancers were statistically significantly more often visible in earlier screening mammograms (43%) than interval cancers (19%) (P = 0.002). Tumours missed by screening mammography but which were visible on retrospective review were often histologically well-differentiated and were more often diagnosed in the subsequent screening round than by clinical diagnosis as interval cancers. If all retrospectively visible interval cancers had been diagnosed by screening 19% (10/54) of the interval cancers could have been avoided. If all retrospectively visible cancers had been diagnosed at the time of false-negative screening or assessment 65% (84/130) of all patients would have benefitted from an earlier diagnosis compared with the actual figure of 31% (41/130).