Association Between Breast Arterial Calcification on Mammography and Coronary Artery Disease: A Systematic Review and Meta-Analysis

Abstract

Background:

Breast arterial calcification (BAC), which may be detected during screening mammography, is hypothesized to be a noninvasive imaging marker that may enhance cardiovascular risk assessment.

Materials and Methods:

In this systematic review and meta-analysis, we sought to assess the association between BAC and coronary artery disease (CAD) by conducting a meta-analysis. We conducted a literature search of PubMed, Scopus, Cochrane library, ClinicalTrials.gov, and conference proceedings, from inception through December 24, 2019. The outcome of interest was the presence of CAD in patients with BAC. This was reported as crude and adjusted odds ratio (OR).

Results:

A total of 18 studies comprising 33,494 women (mean age of 60.8 ± 3.7 years, 25% with diabetes, 57% with hypertension, and 21% with history of tobacco smoking) were included in the current meta-analysis. The prevalence of BAC among study participants was 10%. There was a statistically significant association between BAC and CAD (unadjusted OR 2.14; 95% confidence interval [CI] 1.63–2.81, p < 0.001, I² = 76.5%). Moreover, adjusted estimates were available from 10 studies and BAC was an independent predictor of CAD (OR 2.39; 95% CI 1.68–3.41, p < 0.001, I² = 61.7%). In the meta-regression analysis, covariates included year of publication, age, hypertension, diabetes mellitus, and history of tobacco smoking. None of these study covariates explained the heterogeneity across studies.

Conclusions:

BAC detected as part of screening mammography is a promising noninvasive imaging marker that may enhance CAD risk prediction in women. The clinical value of BAC for cardiovascular risk stratification merits further evaluation in large prospective studies.

Introduction

Coronary artery disease (CAD) is the leading cause of mortality and morbidity among women worldwide.¹ Although women suffer similar cardiovascular morbidity and mortality as men, significantly fewer preventive therapies are employed in women and delays in the diagnosis of CAD remain major barriers to the optimal management of women with or at risk for cardiovascular disase.² These observations can be in part explained by an underestimation of atherosclerotic cardiovascular disease (ASCVD) risk in women using traditional risk factors or contemporary risk calculators provided by the American College of Cardiology (ACC) and American Heart Association (AHA).³ A large proportion of cardiovascular events occur in women whose 10-year estimated ASCVD risk is <7.5%.⁴ Therefore, improving ASCVD risk assessment in women remains an important unmet need.

Current cancer screening guidelines recommend consideration of mammography in women starting at the age of 40 years, and nearly two-thirds of U.S. women ≥40 years have undergone a mammography within the past 2 years.^5,6 Breast arterial calcification (BAC) can be detected on mammography and has emerged as a potentially useful noninvasive marker to enhance cardiovascular risk assessment.⁷ However, observational studies have yielded conflicting results regarding the association of BAC with CAD.^8–13 In this systematic review and meta-analysis, we aimed to clarify the association between BAC and CAD.

Materials and Methods

Search strategy

We conducted a literature search of PubMed, Cochrane library, Scopus, clinicaltrials.gov, and conference proceedings, from inception through December 31, 2019. Our search strategy is presented in Supplementary Table S1. We utilized the “related articles” function in PubMed to find relevant articles that were missed by the initial search. In addition, reference lists of included studies were hand searched to find additional relevant articles that were missed by the primary search. Our search and meta-analysis were conducted and reported according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses Protocols (PRISMA-P) Statement 2015.¹⁴ The protocol of this study was registered at the International prospective register of systematic reviews (PROSPERO) database.

Study selection

Titles and abstracts of the studies, which were retrieved by the initial search, were screened by two authors (S.R. and K.O). The full texts of relevant articles were reviewed to determine if they met the inclusion criteria. Any discrepancies were resolved by a third author (M.O).

Inclusion criteria

To be included in this analysis, the study had to (1) include patients in whom BAC was detected by screening mammogram and (2) report the association between BAC and CAD.

Outcomes and definitions

The outcome of interest was CAD. CAD was defined by either clinical diagnosis (e.g., history of myocardial infarction, unstable angina, and revascularization) or by imaging evidence of CAD (>50% stenosis of an epicardial coronary artery on invasive coronary angiography or coronary computed tomography angiography or abnormal results on stress testing).

Quality assessment

Two authors (S.R and K.O) assessed the quality of the included studies using the Modified Newcastle-Ottawa scale for assessment of observational studies (Supplementary Table S2).

Statistical analysis

Effect estimates were extracted from each study in the form of odds ratio (OR) and 95% confidence intervals (CIs). These measures of association were directly extracted from the article or calculated based on the available data presented in the results. Whenever available, adjusted ORs were extracted from the studies. The effect measures were pooled together using the random effect model to account for between-study variation. Heterogeneity between studies was explored by Cochran Q statistic (p < 0.05) and I-squared (I²) statistic. All statistical tests were two-sided and p-values <0.05 were considered significant. Sensitivity analyses were performed by removing one study at a time to assess the effect of each on the overall effect size (leave-one-out analysis). Funnel plots were examined for publication bias and potential sources of heterogeneity were investigated using meta-regression. We also conducted a sensitivity analysis, in which we only included studies that reported adjusted ORs to account for hidden confounders in the relationship between BAC and CAD. All statistical analyses were performed using Stata Statistical Software version 15.1 (StataCorp. 2017; StataCorp, LLC, College Station, TX).

Results

The initial database search retrieved 349 articles. After excluding duplicates, a total of 257 articles were screened for eligibility by reading the title and abstract of the study. A total of 39 articles were then screened using the predetermined inclusion criteria to assess eligibility. Funnel plots for the outcomes were examined for asymmetry (Supplementary Fig. S1). Details of the study selection process are reported following the PRISMA guidelines along with the PRISMA checklist (Fig. 1 and Supplementary Table S3).

FIG. 1.

Flowchart for the systematic review and meta-analysis as per the PRISMA. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analysis.

A total of 18 studies^{8–13,15–26} including 33,494 women (mean age of 60.8 ± 3.7 years, 25% with diabetes, 57% with hypertension, and 21% with history of tobacco smoking) met the inclusion and exclusion criteria. The prevalence of BAC among included participants was 10%. Detailed baseline characteristics of the included studies are shown in Table 1.

Table 1.

Baseline Characteristics of the Studies Included in the Systematic Review and Meta-Analysis

	N	BAC (%)	Age (mean)		HTN (%)		DM (%)		HLD (%)		CKD (%)		Smoking (%)		BMI (mean)
Author (year)Ref.			BAC	Control	BAC	Control	BAC	Control	BAC	Control	BAC	Control	BAC	Control	BAC	Control
Kataoka et al. (2006)11	1590	16	65.8	62.7	10.4	11.7	2.4	1.9	—	—	—	—	—	—	25.9	26.3
Newallo et al. (2015)12	204	20.6	52.5		74		33.3		62.3		5.4		45.6		32.9
Karm et al. (2015)18	198	—	72	61	80.2	87.8	39	28.4	78	82	—	—	—	—	—	—
Fiuza Ferreira et al. (2007)16	131	40	61		88		27		67		—	—	15
Henkin et al. (2003)10	319	41	64.3	59.9	83.2	65.6	—	—	83.2	82.4	—	—	12	28.4
Moshyedi et al. (1995)21	182	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—
Oliveira et al. (2009)22	80	41	—	—	—	—	—	—	—	—	—	—	—	—	—	—
Penugonda et al. (2010)23	94	61	68.9	63.2	72	73	21	23	53	54	9	11	16	38	28.1	27.9
Topal et al. (2007)26	123	40	68.85	52.34	82	70	33	23	—	—	—	—	6	13	—	—
Zgheib et al. (2010)13	172	57	72	60.4	70	52	49	36	66	63			23	30	—	—
Abou-Hassan et al. (2015)8	202	58	61.5	53.9	—	—	69	40	—	—	100	100	25	21	—	—
Iribarren et al. (2004)17	12761	3	—	—	—	—	—	—	—	—	—	—	—	—	—	—
Kemmeren et al. (1998)19	12239	—	60.1	57.5			6.4	3.5					14.2	28.8
Schnatz et al. (2011)25	1449	—	56.3		31		4.7		34.9		—	—	7.4		—	—
McLenachan et al. (2019)20	405	23	58		33		8		—	—	—	—	—	—	30
Crystal et al. (2000)15	865	—	65	54	52	25	13	5	32	25			9	30
Dale et al. (2008)9	819	10.5	—	—	—	—	—	—	—	—	—	—	—	—	—	—
Rotter et al. (2008)24	819	14	70	54	52	57	11	4	47	33	—	—	4	9	—	—

BAC, breast arterial calcification; BMI, body mass index; CKD, chronic kidney disease; DM, diabetes mellitus; HLD, hyperlipidemia; HTN, hypertension; N, total number of patients.

The unadjusted OR calculated from these 18 studies demonstrated a statistically significant association between BAC and CAD (OR 2.14; 95% CI 1.63–2.81, p < 0.001, I² = 76.5%) (Fig. 2). Moreover, adjusted estimates were available from 10 studies^{8,10–12,15,16,18,22,24,25} and BAC was an independent predictor of CAD even after covariate adjustment (OR 2.39; 95% CI 1.68–3.41, p < 0.001, I² = 61.7%) (Fig. 3 and Supplementary Table S4). A leave-one-out sensitivity analysis was performed by removing one study at a time. This analysis did not demonstrate a difference in the pooled ORs.

FIG. 2.

Forest plot comparing CAD in patients with versus without BAC. BAC, breast arterial calcification; CAD, coronary artery disease; OR, odds ratio.

FIG. 3.

Forest plot comparing CAD in patients with versus without BAC using adjusted estimates.

Owing to the heterogeneity detected in the meta-analysis, we performed meta-regression to investigate the contribution of study-level covariates, including year of publication, age, hypertension, diabetes mellitus, and history of tobacco smoking. Although we observed a trend suggesting that heterogeneity may be explained by specific covariates such as diabetes mellitus, tobacco smoking, and age, none of these covariates were found to be statistically significant in explaining the heterogeneity among the studies (Fig. 4 and Supplementary Table S5).

FIG. 4.

Meta-regression analysis for study-level covariates on the association of BAC and CAD.

Discussion

This meta-analysis assessed the association between BAC and CAD among 33,494 women from 18 observational studies. There are three key findings from this study. First, the overall prevalence of BAC in the participants was 10%. Importantly, there was a statistically significant association between BAC and CAD (OR 2.14; 95% CI 1.63–2.81). Moreover, in the meta-analysis of adjusted estimates, BAC continued to show a statistically significant association with CAD, even after controlling for additional possible confounders, further suggesting the potential independent clinical relevance of BAC.

Atherosclerotic risk is underestimated in many women using the traditional risk factor approach or contemporary risk calculators such as the ACC/AHA ASCVD Risk Calculator.³ There is increased recognition of the need to evaluate sex-specific risk factors in assessing risk for CAD in women.²⁷ It is also known that the prevalence of obstructive CAD is lower in women than men, yet women have increased morbidity and mortality from CAD.²⁸ This observation is likely due to underestimation of risk and suboptimal preventive measures, including underutilization of statin therapy and behavioral risk factor modification in women. Improved models to assess sex-specific risks are needed to address the current discrepancies in care. Identification of computed tomography-derived coronary artery calcium (CAC) is a well-validated guideline-directed modality to help refine risk in patients in whom risk uncertainty exists.²⁷ The presence of CAC, in women considered to be at otherwise low risk based on the Framingham risk score, was predictive of future coronary events in the Multi-Ethnic Study of Atherosclerosis.²⁹ Although computed tomography-derived CAC is a well-validated predictive marker of atherosclerotic risk, it is underutilized to refine risk estimates.³⁰ In contrast, nearly two-thirds of U.S. women ≥40 years have undergone a mammogram within the past 2 years.⁵ Given national guidelines for breast cancer screening with mammography, assessment of BAC to further enhance ASCVD risk assessment, if validated, would not require any additional expense or radiation exposure.

Although conflicting data concerning the strength of association between BAC and CAD exist, most studies to date have suffered from small sample sizes, their observational nature, and varying methods for detecting and defining CAD. The MINERVA (Multiethnic Study of Breast Arterial Calcium Gradation and Cardiovascular Risk) study, a 5-year prospective multiethnic observational cohort, hopes to address some of these shortcomings to investigate the ability of BAC to predict cardiovascular outcomes.³¹ Given the current absence of large well-constructed prospective studies, we undertook this systematic review and meta-analysis to synthesize the currently available data across similarly conducted studies to provide insight on the relation between BAC and CAD. Prior meta-analyses have suggested similar association between BAC and CAD; however, this analysis is different than the previous studies.^32,33 First, we conducted analysis for adjusted ORs reported by the primary studies to account for confounders, which was not done in the previous studies. Second, we used meta-regression techniques to explore explanation for heterogeneity in the included studies. After applying these advanced analytic techniques, the conclusion from this analysis support BAC detected as part of screening mammography as a promising noninvasive imaging marker that may enhance CAD risk prediction in women. The clinical value of BAC for cardiovascular risk stratification merits further evaluation in large prospective studies.

This analysis has limitations that need to be acknowledged. First, this was a study-level meta-analysis, which lacked individual patient level data. Second, the currently included studies were all observational in nature, which introduce many biases related to study design.

Conclusion

Based on currently available data, there is a statistically significant association between BAC and CAD. Whether the addition of BAC to standard quantitative risk assessment algorithms will enhance risk prediction remains unknown. Further evaluation of BAC in well-designed prospective studies will define the potential value of this noninvasive imaging marker in cardiovascular risk assessment.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

No funding was received for this article.

Supplementary Material

Supplementary Table S1

Supplementary Table S2

Supplementary Table S3

Supplementary Table S4

Supplementary Table S5

Supplementary Figure S1