The influence of breast density on the sensitivity and specificity of ultrasound and mammography in breast cancer diagnosis

Abstract

Objective. Th e aim of this study was to analyse the sensitivity and specificity of ultrasound and mammography according to breast densityand determine which of these diagnostic imagings is a more accuratetest for diagnosis of breast cancer. Patients and methods. By meansof a cross-sectional study, ultrasound and mammographic examinationsof 148 women with breast disease symptoms were analysed.All women underwent surgery and all lesions were examined by histologicalexamination which revealed the presence of 63 breast cancers,and 85 benign lesions. Histological examination was used as the “goldstandard”. In relation to breast density, the women were separated intotwo groups, group A: women with “fatty breast” (ACR BI-RADS densitycategories 1 and 2) and group B: women with “dense breast”(categories3 and 4). Ultrasound and mammographic fi ndings were classifi ed onthe BI-RADS categorical scale of 1-5. For statistical data processing, thelogistic regression analysis and the McNemar chi-square test for pairedproportions was used. Th e diff erences on the level of p<0.05 were consideredstatistically signifi cant. Results. In the group of women with breastdensity categories 1 and 2 the diff erence in the sensitivities (p=1) as wellas in the specifi cities (p=0.11) of the two imaging tests was not statisticallysignifi cant. In the group of women with breast density categories3 and 4 the ultrasound sensitivity was signifi cantly higher than themammographic sensitivity (p=0.03) without a statistically signifi cantdiff erence in specifi city (p=0.26). Sensitivity of mammography was(linearly – ex; linearity exists between breast density and the logarithmof odds for a positive result) associated with breast density (likelihoodratio χ2 =15.99, p =0.0001). Th e odds ratio for (the probability of –ex) a positive mammographic result was 0.25 (95% CI, 0.11-0.58). Th esensitivity of ultrasound and specifi city of each test were not (linearly- ex) associated with breast density. Conclusion. Breast density hada signifi cant infl uence on the sensitivity of mammography but noton specifi city. Th is is very important because a certain percentage ofwomen, not only under 40 but also aft er 40, have heterogenous andextremely dense breasts (density categories 3 and 4). In these women,ultrasound is a more accurate imaging test than mammography, whilein the women with fatty breasts (density categories 1 and 2) these imagingtests are almost equally accurate in breast cancer diagnosis.