Analysis of interval breast carcinomas in a randomized screening trial in Stockholm

Abstract

In the interval between screening examinations, some cases of breast cancer are invariably detected clinically in patients whose mammogram was considered to be normal at the earlier screening. During the first interval in the Stockholm study, 60 interval cancers were detected, giving a rate of 1.8 cases/1000 examinations/24 months. About half of these interval cases (31/60) were true interval tumours in that no sign of them could be found on the first mammogram; the other half, non-true, were possible to trace on the first mammogram. It is mainly women under 50 who feature in the interval group, above all in the sub-group of true interval cancers (p<0.05). The incidence of interval cancer rises, as expected, with the length of the interval (Fig. 1). In the final six months of the 2-year interval the incidence of interval cancers had risen to 88 per cent of the cancers detected in the control group in the same period of time. The cumulative incidence of interval cancers supports the hypothesis that the distribution of sojourn time in the interval 0–2 years is approximately rectangular. This means that shortening the interval by one-half would halve the number of interval cases. If mammography becomes a wide spread screening method for early detection of breast cancer, the number of non-true interval cancers could be a feed back on the effectiveness of the screening. An analysis of the malignancy of the true interval cancers, based on tumour size, axillary status, tumour stage, and estrogen receptor content, shows that these are the more malignant tumours with an unfavourable prognosis, while the control group is intermediate in this respect and the tumours detected at screening have the most favourable prognosis.