Screening for Breast Cancer: The Role of Supplemental Tests and Breast Density Information

Abstract

Mammography screening is the gold standard for timely detection of breast cancer, but it has less than perfect accuracy. This has led to increased consideration of supplemental screening tests such as ultrasound and Magnetic Resonance Imaging (MRI) in recent medical practice, particularly for women with strong risk factors including those with dense breasts. More than 35 million (about 45%) US women aged 40 or more have dense breasts. Breast density is not only a strong risk factor for developing breast cancer, but it also impairs mammography performance. The net result is that millions of women with dense breasts face disproportionate risk of death from breast cancer. Medical community, however, remains divided and mostly indecisive about when to use supplemental screening, and how its use should be adjusted with breast density. We formulate the optimal breast cancer screening problem using a discrete-time partially observable Markov decision process model and scrutinize the screening recommendations of this optimization model in a detailed simulation model. We find that screening intensity increases with age, breast density, and other risk factors, but use of supplemental tests is not always preferred since accumulated disutility can outweigh their benefits. For the objective of maximizing quality-adjusted life expectancy (QALE), supplemental testing is most valued for high-risk patients that carry multiple significant risk factors (e.g., family history, previous biopsy, extremely dense breast), while mammography-only testing is generally sufficiently effective for lower risk patients. Increased breast density is associated with more frequent use of mammography, and when it falls short for high-risk patients, with more frequent use of supplemental tests. For high-risk patients, supplemental ultrasound is used more often than MRI when maximizing QALE and results in significantly fewer false positives, while MRI results in higher cancer detection rates and shorter time to detection. For high-risk patients with extremely dense breasts, optimally using supplemental tests compared to optimally using mammography-only screening detects about 0.9% more cancer cases at the in situ stage, shortens time until detection by 5.7%, and reduces false-positive outcomes by 6.2%. For the objective of maximizing QALE, we also identify thresholds on supplemental test parameters (i.e., specificity, sensitivity, disutility) at which the dominant supplemental screening modality switches from one test to the other, which would help guide the medical community when making screening choices under various parameter settings.