Interval cancers in the National Health Service Breast Screening Programme.

Abstract

The National Health Service (NHS) Breast Screening Programme defines objectives with acceptable and, where possible, achievable standards. This paper examines the problem of interval cancers. Slopes due to shrinkage of primary breast cancers in response to primary systemic therapy are used as surrogates for growth slopes. From the distribution of these slopes the proportion of tumours which might reach particular sizes in any time interval may be estimated. For specific differences in size between screen and clinical detection levels the number of interval cancers which could appear in 1, 2 and 3 years after screening is estimated. The justification for these predictions is evaluated. The literature in relation to growth is reviewed; we add three growth slopes from our own measurements. The range of volume doubling times and volume halving times is the same whereas the mean value derived from the literature is about 90 days in contrast to only 26 days for inferred growth rates. Because of the difficulties inherent in the measurement of growth, it is concluded that the literature defines the left half of a distribution; the right hand side is completed by shrinkage data. The observed rates for interval cancers from different sources vary widely; there is an underlying assumption that interval cancers are an index of failure. These calculations suggest that there is an irreducible minimum of interval cancers, which will depend upon the screening interval and the size at which tumours are detected. The interval cancers will contain more poorly differentiated tumours. The anticipated achievable standards appear to be over optimistic.