Mammographic breast density: How it affects performance indicators in screening programmes?

Abstract

ObjectivesTo investigate how breast density affects screening performance indicators in a digital mammography context. MethodsWe assessed the effect of breast density over the screen-detected and interval cancers rates, false-positives, specificity, sensitivity, recall rate, positive predictive value of recall (PPV-1), and PPV of invasive tests (PPV-2). Radiologists classified breast density using the BIRADS System. We used generalized estimating equations to account for within-woman correlation by means of the robust Huber-White variance estimator. ResultsWe included 177,164 women aged 50–69 years who underwent 499,251 digital mammograms from 2004 to 2015 in Spain. According to the fibroglandular tissue percentage, 24.7% of mammograms were classified as BI-RADS 1 (<25% glandular), 54.7% as BI-RADS 2 (25–50% glandular), 14.0% as BI-RADS 3 (51–75% glandular) and 6.6% as BI-RADS 4 (>75% glandular). Overall, women with BI-RADS 3 had the highest screen-detected cancer rate (5.9 per 1000) and BI-RADS 4 the highest interval cancer rate (2.4 per 1000). Sensitivity decreased from 89.2% in women with BI-RADS 1 to 67.9% in BI-RADS 4. Both PPV-1 and PPV-2 decreased from 10.4% to 5.7% and from 49.8% to 32.4% in women with BI-RADS 1 and BI-RADS 4, respectively. Women aged 60–69 years with BI-RADS 4 had the lowest sensitivity (54.9%) and the highest interval cancer rate (3.8 per 1000). ConclusionsPerformance screening measures are negatively affected by breast density falling to a lower sensitivity and PPV, and higher interval cancer rate as breast density increases. Particularly women aged 60–69 years with >75% glandular breasts had the worst results and therefore may be candidates for screening using other technologies.