Research article
Assessing the usefulness of a novel MRI-based breast density estimation algorithm in a cohort of women at high genetic risk of breast cancer: the UK MARIBS study
-
* Corresponding author: Deborah J Thompson deborah.thompson@phpc.cam.ac.uk
1 Cancer Research UK Genetic Epidemiology Unit, Department of Public Health and Primary Care, University of Cambridge, Worts Causeway, Cambridge, CB1 8RN, UK
2 Institute of Cancer Research and Royal Marsden NHS Foundation Trust, Downs Road, Sutton, Surrey, SM2 5PT, UK
3 Gynaecological Cancer Research Centre, UCL EGA Institute for Women's Health, University College London, Gower Street, London, WC1E 6BT, UK
4 Department of Radiology, University of Cambridge, Addenbrookes Hospital NHS Foundation Trust, Hills Road, Cambridge, CB2 0QQ, UK
5 Breast Screening Unit, Nightingale Centre & Genesis Prevention Centre, Wythenshawe Hospital, Southmoor Road, Manchester M23 9LT, UK
6 Academic Unit of Medical Genetics and Regional Genetics Service, St Mary's Hospital, Hathersage Road, Manchester, M13 0JH, UK
7 See Additional data file 1 for full authorship list
Breast Cancer Research 2009, 11:R80 doi:10.1186/bcr2447
Published: 11 November 2009Abstract
Introduction
Mammographic breast density is one of the strongest known risk factors for breast cancer. We present a novel technique for estimating breast density based on 3D T1-weighted Magnetic Resonance Imaging (MRI) and evaluate its performance, including for breast cancer risk prediction, relative to two standard mammographic density-estimation methods.
Methods
The analyses were based on MRI (n = 655) and mammography (n = 607) images obtained in the course of the UK multicentre magnetic resonance imaging breast screening (MARIBS) study of asymptomatic women aged 31 to 49 years who were at high genetic risk of breast cancer. The MRI percent and absolute dense volumes were estimated using our novel algorithm (MRIBview) while mammographic percent and absolute dense area were estimated using the Cumulus thresholding algorithm and also using a 21-point Visual Assessment scale for one medio-lateral oblique image per woman. We assessed the relationships of the MRI and mammographic measures to one another, to standard anthropometric and hormonal factors, to BRCA1/2 genetic status, and to breast cancer risk (60 cases) using linear and Poisson regression.
Results
MRI percent dense volume is well correlated with mammographic percent dense area (R = 0.76) but overall gives estimates 8.1 percentage points lower (P < 0.0001). Both show strong associations with established anthropometric and hormonal factors. Mammographic percent dense area, and to a lesser extent MRI percent dense volume were lower in BRCA1 carriers (P = 0.001, P = 0.010 respectively) but there was no association with BRCA2 carrier status. The study was underpowered to detect expected associations between percent density and breast cancer, but women with absolute MRI dense volume in the upper half of the distribution had double the risk of those in the lower half (P = 0.009).
Conclusions
The MRIBview estimates of volumetric breast density are highly correlated with mammographic dense area but are not equivalent measures; the MRI absolute dense volume shows potential as a predictor of breast cancer risk that merits further investigation.