Mammographic breast density and arterial calcification as predictors of incident cardiovascular events
R Nandurkar, C Soh, V Goel, S Vasanthakumar, N Sultana, E Aldous, H Tu, N NerlekarAbstract
Background
Breast arterial calcification (BAC) and breast density assessed on routine mammography have been associated with cardiovascular risk. Whether their combined assessment provides incremental prognostic information beyond conventional cardiovascular risk models remains uncertain.
Purpose
To evaluate the independent and joint associations of BAC and low breast density with incident major adverse cardiovascular events (MACE), and to assess their incremental prognostic value beyond established cardiovascular risk models.
Methods
We conducted a retrospective cohort study of women aged >40 years undergoing screening mammography between 2005 and 2020. BAC was quantified using a validated deep-learning software and classified as present or absent. Breast density was visually assessed using Breast Imaging Reporting and Data System (BI-RADS) criteria and categorised as low versus high density. The primary outcome was incident MACE, defined as a composite of death, ischaemic heart disease, stroke or heart failure identified from linked medical records. Cox proportional hazards models were adjusted for age, traditional cardiovascular risk factors, and preventative pharmacotherapy. Model discrimination was assessed using Harrell’s c-statistic, and net reclassification improvement was calculated to evaluate incremental prognostic value.
Results
Among 22,285 women (mean age 55 ± 12 years), 5,413 (24%) had BAC and 8,024 (36%) had low breast density. Over a mean follow-up of 4.1 ± 3.2 years, 1,964 MACE occurred. Both low breast density (hazard ratio [HR] 1.67, 95%CI 1.53-1.84; p<0.001) and BAC (HR 2.49, 95%CI 2.28-2.73; p<0.001) were independently associated with MACE. Women with concurrent BAC and low breast density had the highest risk (HR 3.70, 95%CI 3.25-4.21; p<0.001). Addition of BAC and breast density to a baseline cardiovascular risk model improved discrimination (c-statistic 0.67 to 0.71) and yielded a 12% net reclassification improvement.
Conclusion
Both BAC and low breast density independently predict incident cardiovascular events in women undergoing screening mammography. Combined assessment modestly improves risk discrimination beyond conventional risk models. Prospective studies are warranted to determine whether integration of mammographic parameters into cardiovascular risk assessment improves clinical outcomes.Kaplan-Meier Curve