Lessons Learnt from Imaging Review of Interval Breast Cancers in a Single Center in the UK National Breast Screening Program

Abstract

AbstractFalse-negative mammograms are a part of any screening program and the National Breast Screening Program in the United Kingdom is no exception. Every year, 2.5 million women have screening mammograms in the United Kingdom. Around 8 per 1,000 women screened are diagnosed with a breast cancer and 3 per 1,000 women will present with an interval breast cancer following a negative screening mammogram and before their next mammogram.Robust quality standards have to be adhered to at every stage in the screening pathway in order to maintain the fine balance between detecting cancers early and avoiding unnecessary false-positive results. As part of this process, there is a mandatory requirement for screening units to review all breast cancers. We present a pictorial illustration of the lessons learnt from such a review in this essay. The cases described here are from one large breast screening unit in the North of England. In this unit, 30,000 to 40,000 women have a screening mammogram each year and these are all double read by human readers. All cases requiring recall and those where there is a disagreement between the first and second human reader go through a consensus/arbitration process involving a minimum of two human readers. Interval cancers are identified through a consistent process outlined by the screening program and are then subjected to a review by a minimum of two reviewers. In this process, the reviewers have access to the false-negative mammograms along with all priors that were available at the time of initial read. The reviewers make a decision on whether the initial mammogram is normal or abnormal without seeing the diagnostic mammogram with the interval cancer. They also categorize the density of the breast and describe the mammographic abnormality if there is any on a standardized interval cancer data collection form. Finally, they categorize the interval cancer into one of three types—“satisfactory,” “satisfactory with learning points,” or “unsatisfactory.”