The impact of breast cancer radiation therapy exposure on the prevalence of breast arterial calcification

Abstract

Objectives

Mammographic breast arterial calcification (BAC) is an emerging imaging biomarker of cardiovascular disease (CVD) risk in women. The purpose of this study was to assess if breast radiation therapy (RT) exposure impacts the screening utility of this imaging biomarker.

Methods

This cross-sectional study included women ages 40-75 years who underwent index screening mammography between January 1, 2011 and December 31, 2012. Chart review was performed to extract data on the breast cancer RT exposure and CVD risk factors. Mammograms were reviewed for the presence of BAC. Multivariate logistic regression was used to examine breast RT exposure and BAC, with adjustment for age, body mass index, smoking status, hypertension, Type 2 diabetes, statin medication use, and antihypertensive medication use.

Results

Of the 1155 women included in this analysis, 222 (19.2%) had mammographic evidence of BAC, 122 (10.6%) had a history of RT exposure, and 39 (32%) women with RT exposure had mammographic BAC. Women with breast RT exposure had higher odds of BAC compared to women without (odds ratio: 2.18, 95% CI: 1.43, 3.28; P-value = .0008). However, this association became non-significant after multivariable adjustment, with the maximally adjusted model demonstrating an odds ratio of 1.52 (95% CI: 0.95, 2.40; P-value = .07).

Conclusions

Our findings suggest that breast RT exposure does not impact the prevalence of mammographic BAC. Therefore, it does not affect its utility as an imaging biomarker of CVD risk.

Advances in knowledge

This is the first observational study addressing the knowledge gap pertaining to the influence of breast RT exposure on BAC.

Introduction

The presence of breast arterial calcification (BAC) on screening mammograms is an emerging imaging biomarker of cardiovascular disease (CVD) risk, major adverse cardiovascular events (MACE), and CVD mortality.^1–6 The prevalence of BAC changes with age and BAC is present in 12%-30% of women who undergo screening mammography, including women with a personal history of breast cancer.^4–10 Most women with breast cancer undergo radiation therapy (RT) as part of their multi-modality treatment.¹¹ While RT improves breast cancer survival, it is complicated by radiation-induced vasculitis within the breast and adjacent organs. Radiation-induced heart disease (RIHD) is a known complication of breast cancer radiotherapy that may manifest as coronary artery calcifications (CAC) identified on CT scans.¹² Studies have demonstrated an association between CVD and progression of CAC on CT scans obtained at least 2 years after radiotherapy.^13–15 While the association between CAC and RT is well-defined, there is minimal data on the association between BAC and radiation-induced vasculitis within the mammary arteries.^16,17 The use of mammographic screening among women with a personal history of breast cancer treated by RT requires an understanding of the association between BAC and radiotherapy to allow for accurate incorporation of BAC into CVD risk assessment models.

CVD remains the leading cause of overall mortality among women accounting for 28% of female deaths compared to 3% of female deaths due to breast cancer in the United States in 2021.¹⁸ The annual CVD mortality rate is higher among women than men, and women have not experienced equivalent mortality reductions compared to men in the United States.^19,20 Current guidelines recommend the use of the atherosclerotic cardiovascular disease (ASCVD) risk estimator to assess the 10-year risk of ASCVD; however, the ASCVD risk estimator underestimates CVD risk in women, leading to delayed care and possible under-treatment.^3,5,21 There is a need for improved CVD risk assessment among women to decrease mortality, and studies suggests that BAC may enhance CVD risk stratification by identifying women at high risk who are deemed low risk by the ASCVD risk estimator.^22,23 Screening mammography is highly utilized by 76% of eligible women between the ages of 40 and 74 in the United States.²⁴ Due to improved screening and treatment, the number of breast cancer survivors who undergo screening mammography is increasing. BAC represents an opportunistic imaging biomarker of CVD risk that may be incorporated into CVD risk assessment for women above the age of 40, but the accurate use of BAC as a biomarker of CVD risk requires knowledge of the relationship between breast cancer radiotherapy and BAC. The goal of our study was to compare the prevalence of BAC on screening mammograms in women with and without a personal history of breast cancer RT.

Methodology

Study design

This is a cross-sectional study examining breast cancer RT exposure and prevalence of mammographic BAC.

Study population

This study was approved by our Institutional Review Board and informed consent was waived for this retrospective analysis. An electronic health record database query was used to identify women between the ages of 40 and 75 years who had a screening mammogram between January 1, 2011 and December 31, 2012, and who had data for clinical variables utilized by the 10-year ASCVD risk estimator collected within 1 year of the index screening mammogram. These variables included age, body mass index, smoking status (current smoker: yes/no), type 2 diabetes status (yes/no), hypertension (yes/no), lipids (HDL, LDL, triglycerides, total cholesterol), antihypertensive use (yes/no), and statin use (yes/no). This query identified 1226 women. A retrospective chart review for data extraction was completed for the variables of interest (history of breast cancer and RT). Women were excluded from this study if they had missing data for the variables of interests or known coronary artery disease (CAD). Women were also excluded if they had a breast RT exposure (and breast cancer diagnosis) within the 2 years preceding their index screening mammogram or any time after their index screening mammogram to ensure a minimum of 2 years between XRT exposure and the index mammogram used to assess for BAC. The final analytic cohort included 1155 women (see Figure 1).

Figure 1.

Study population. Flow diagram shows how the final analytic cohort was determined and reasons for exclusion.

Breast cancer and radiation therapy exposure

Electronic health record data extraction was completed to collect data on breast cancer diagnosis and treatment-associated variables, including breast cancer diagnosis date, breast cancer histology (ductal carcinoma, lobular carcinoma, etc.), breast cancer laterality (right, left, bilateral), RT (yes, no), and RT laterality (right, left, bilateral).

Breast arterial calcification

Standard full-field digital mammograms containing medio-lateral oblique and cranio-caudal views were reviewed for the presence of BAC (yes/no), severity of calcification (no calcification, intermediate/moderate, and severe), and laterality of calcification (left, right, bilateral). Mammograms were obtained with Selenia or Dimensions Hologic units (Hologic Incorporated; Bedford, MA, United States) and reviewed on Philips PACS v.3.6 (Philips Healthcare; Best, Netherlands). Blinded mammographic review was completed by 2 breast imaging radiologists with 20 and 22 years of experience (R.M.D.A. and J.M.A.S., respectively). Inter-rater agreement was assessed via Cohen’s Kappa using 30 randomly selected mammograms (via block randomization) that both radiologists reviewed.

Statistical analyses

All statistical analyses were completed using R Statistical Software (version 4.3.2; R Studio: Integrated Development Environment for R, Boston, MA, United States). All P-values were 2-sided (for applicable tests), with α < .05 being considered statistically significant. Distributions of continuous variables were assessed visually using density plots and statistically using Shapiro-Wilk normality tests to determine the appropriateness of parametric tests. The Mann-Whitney U test was utilized to examine between-group differences for continuous variables, and Pearson’s chi-square test was used to examine between-group differences for categorical variables.

The association between RT exposure and BAC was examined using multivariate logistic regression with 4 different regression models. Model 1 consisted of RT exposure (yes/no) as the independent variable and BAC (yes/no) as the dependent variable. The second model consisted of Model 1 with the addition of patient age as a covariate. Model 3 included the variables in Model 2 plus further adjustment for body mass index. Model 4 included breast RT exposure, age, and body mass index, and was further adjusted for known risk factors for vessel calcification, including hypertension status, Type 2 diabetes status, antihypertensive use, statin use, and smoking status.

Age was examined as a potential effect modifier of the association between breast RT exposure and BAC. A stratified analysis was completed by dichotomizing this population of women as ≤55 and >55 years of age, and an interaction was tested by including a cross-product between age and BAC in the regression model.

An additional analysis was performed examining if RT laterality was associated with BAC laterality using a Fisher’s exact test.

Results

Study population

Among the 1155 women included in this analysis, 222 (19.2%) had BAC, 138 had a history of breast cancer, and 122 had a history of breast RT exposure. Women with RT exposure were older, had lower body mass index, and had a higher prevalence of BAC (Table 1). Current smoker status, type 2 diabetes status, hypertension status, statin medication use, and antihypertensive medication use did not differ across breast RT exposure.

Table 1.

Patient characteristics across breast radiation therapy exposure status (N = 1155).

	Breast radiation therapy exposure N = 122	No breast radiation therapy exposure N = 1033	P-value
Age (years)a	61.5 ± 8.8	57.3 ± 9.0	<.0001b
Body mass index (kg/m2)a	27.0 ± 6.4	29.1 ± 7.3	.003b
Current smoker statusc			.78d
Yes	7 (6%)	66 (6%)
No	115 (94%)	967 (94%)
Type 2 diabetes statusc			.26d
Yes	10 (8%)	120 (12%)
No	112 (92%)	913 (88%)
Hypertension statusc			.63d
Yes	33 (27%)	259 (25%)
No	89 (73%)	774 (75%)
Antihypertensive medication usec			.29d
Yes	59 (48%)	448 (43%)
No	63 (52%)	585 (57%)
Statin medication usec			.53d
Yes	36 (30%)	277 (27%)
No	86 (70%)	756 (73%)
Breast arterial calcificationc			.0002d
Yes	39 (30%)	183 (18%)
No	83 (70%)	850 (82%)

^aRepresented as mean ± SD.

^b P-values were calculated using a Mann-Whitney U test. Two-sided; significance level was set at 0.05.

^cRepresented as N (%).

^d P-values were calculated using a Pearson’s chi-square test. Significance level was set at 0.05.

Breast radiation therapy exposure and breast arterial calcification

In this study, patients with breast cancer RT had a minimum of 2 years between the completion of breast RT and index mammogram used for BAC analysis. The median duration between breast RT treatment and the time of the screening mammogram was 7.0 years (range: 2-20 years). The crude logistic regression model demonstrated a statistically significant association between breast RT and mammographic BAC (odds ratio: 2.18, 95% CI: 1.43, 3.28; P-value = .0008; Table 2). However, the association was no longer statistically significant after adjustment for age (Model 2, P-value = .07), age and body mass index (Model 3, P-value = .08), or for age, BMI, and other CVD risk factors (Model 4, P-value = .07; Table 2). The association between breast RT and BAC did not appear to differ across age groups (P_interaction = .42).

Table 2.

Odds ratios and 95% confidence intervals for mammographic breast arterial calcification according to breast radiation therapy exposure status (N = 1155).

Model	Breast radiation therapy exposure N = 122	No breast radiation therapy exposure N = 1033	P-valuea
Model 1b	2.18 (1.43, 3.28)	1.00 (ref)	.0002
Model 2c	1.53 (0.97, 2.39)	1.00 (ref)	.07
Model 3d	1.50 (0.95, 2.35)	1.00 (ref)	.08
Model 4e	1.52 (0.95, 2.40)	1.00 (ref)	.07

^aBased on multivariate logistic regression analysis.

^bModel with radiation therapy exposure as the independent variable and breast arterial calcification as the dependent variable.

^cModel was adjusted for patient age (continuous, years).

^dModel was adjusted for patient age (continuous, years) and body mass index (continuous, kg/m²).

^eModel was adjusted for patient age (continuous, years), body mass index (continuous, kg/m²), type 2 diabetes status (yes/no), hypertension status (yes/no), antihypertensive medication use (yes/no), statin medication use (yes/no), and current smoker status (yes/no).

Breast radiation therapy laterality and breast arterial calcification laterality

Among 39 women with BAC with a history of RT exposure, breast RT laterality (left, right, bilateral) and BAC laterality (left, right, bilateral) were discordant in 30 (77%). BAC was present in the breast contralateral to RT exposure in 13 (33%) of patients, and bilateral BAC was present in 16 (41%) of patients with unilateral RT exposure and as shown in Figure 2. Furthermore, there was no statistical association between RT exposure laterality and BAC laterality (P-value = .75).

Figure 2.

Bilateral cranio-caudal views demonstrating bilateral BAC (arrows) in 71 y/o patient with history of left breast invasive lobular carcinoma treated by wide local excision and radiation therapy (scar marker indicated by star). Mammogram obtained 4 years after RT completion consisting of 50.4 Gy to the left breast and a 10 Gy boost to the tumour bed. BAC is more extensive in the untreated right breast than in the left breast that received RT. Abbreviations: BAC = breast arterial calcification; RT = radiation therapy.

Discussion

The American Heart Association (AHA) has issued a “call to action” to address sex-based disparities in CVD mortality and to identify female-specific risk factors to improve CVD risk assessment in women.²² BAC identified on screening mammography is associated with CVD risk, and universal reporting of BAC provides sex-targeted screening that improves CVD risk stratification and may allow for earlier preventive strategies with improved clinical outcomes.^1,25 Women with a personal history of RT for breast cancer treatment constituted 10.6% of women presenting for screening mammography in our cohort; therefore, knowing the association between BAC and breast cancer RT is highly relevant for accurate utilization of BAC as a biomarker of CVD risk. Radiation-induced vascular disease is a well-known complication of radiation treatment that may manifest as vascular calcifications on imaging exams.¹² Our study examined the relationship between breast cancer RT exposure and the prevalence of BAC; and our results showed that BAC was not associated with breast cancer RT after adjusting for age and other CVD risk factors. There was, furthermore, a lack of association between the laterality of BAC and the laterality of breast cancer RT exposure, supporting the non-significant relationship between BAC and RT. Bilateral BAC and contralateral BAC were present in 14% and 33% of patients with unilateral RT exposure, respectively (Figure 2). Our results suggest that breast RT exposure is not associated with mammographic BAC prevalence. If confirmed in larger studies, BAC does not require adjustment for RT exposure when utilized as an imaging biomarker for CVD risk assessment.

To our knowledge, only 2 other studies evaluated the association between RT exposure and BAC, and these were limited to women with RT exposure using the contralateral breast as a control. We are the first study to evaluate the prevalence of BAC in women with and without breast cancer RT. Soran et al. reported a lack of association between BAC and MACE in breast cancer patients with RT exposure to the left breast relative to the right breast but did not evaluate the prevalence of BAC independent of MACE.¹⁶ Naraparaju et al. did not find an association between RT exposure and BAC in the distribution of the internal mammary artery (IMA) (the medial breast on mammographic cranio-caudal views) relative to the radiation-naïve breast.¹⁷ They further explored the relationship between IMA calcification and RT exposure in a subset of 61 patients with available CT imaging and did not find evidence of internal mammary calcification associated with RT. The results of our study confirm prior results, and further expand those findings by additionally showing no significant difference in BAC among women with and without RT exposure after adjusting for age and other variables.

The lack of association between BAC and RT exposure differs from the known association between CAC and RT exposure in patients with breast cancer.^{13–15,26,27} The difference between BAC and CAC manifestation after radiotherapy is not entirely unexpected given the diverse aetiologies accounting for coronary artery versus mammary artery calcifications. CAC is due to calcium deposition within the intima of the coronary arteries resulting in CVD secondary to progressive intimal plaque, stenosis, and occlusion. BAC on the other hand, is the result of calcium deposition in the media of the mammary arteries, known as Monckeberg medial calcific sclerosis, resulting in circumferential medial wall thickening without vessel occlusion. The risk of CVD due to medial calcific sclerosis is due to vascular stiffness and decreased vessel compliance rather than vascular occlusion.^28,29 In addition, there are differences in the associations between CAC, BAC, and traditional CVD risk factors. For example, BAC is not significantly associated with hypertension (HTN), obesity, or dyslipidaemia and is inversely associated with smoking.^6,7 The variable pathophysiologic mechanisms underlying CAC and BAC may contribute to the different manifestations of arterial calcifications after RT exposure, and may furthermore inform the enhanced CVD risk stratification provided by BAC beyond the risk predicted by traditional CVD risk factors utilized in the 10-year ASCVD risk assessment.³⁰

There are several possible rationales for the incongruous association between RT and CAC versus RT and BAC. Ionizing radiation damages the endothelial cells of the vessel intima resulting in inflammation, fibrosis, and plaque formation.^12,31 Radiation-induced vasculitis therefore results in calcifications of intimal plaque explaining the high correlation between CAC and breast cancer RT. In contrast, BAC corresponds to calcifications of the medial layer of the mammary arteries consisting predominantly of smooth muscle cells that are less susceptible to radiation damage. A recent study evaluating the specific effects of radiation on blood vessels confirms radiation-induced injury to the endothelial cells but does not reveal evidence of radiation-induced damage to the smooth muscle cells within the vessel media.³¹ Furthermore, different vascular beds may display variable radio-sensitivity that is related in part to vessel size and surrounding tissue composition.¹² Some histologic and imaging studies have demonstrated a lack of atherosclerotic and radiation damage to the IMA, the major vessel supplying the medial breast.^17,32 This raises the possibility that the mammary arteries within the breast are less susceptible to radiation-induced vasculitis compared to the coronary arteries. Finally, while the difference in resolution between CT and digital mammography could limit the ability to detect BAC compared to CAC, the study by Naraparaju et al. showed that CT imaging in a subset of patients with RT exposure did not identify IMA calcifications in patients with CAC suggesting that resolution differences in imaging modality are not a major factor contributing to different manifestations of CAC and BAC in the setting of RT exposure.¹⁷ Of note, cardiac-sparing RT techniques such as treatment planning with 3D imaging and cardiac shielding have been developed to significantly reduce RIHD thereby making the association between CAC and RT less clinically concerning in the future.³³ As such, our goal was to focus on the relationship between BAC and RT to inform accurate use of BAC as an emerging biomarker of CVD risk that may improve CVD risk stratification in women.

Our study is limited by its retrospective design at a single institution that may introduce selection bias and impact reproducibility. Additionally, we used a 2-year minimum threshold (with a range of 2-20 years in this population) for the time between RT completion and index mammogram used for BAC assessment in keeping with prior studies showing vascular calcifications in this timeframe.¹⁶ However, this timeframe does not allow for the detection of BAC secondary to RT occurring within 2 years of RT completion. Lastly, our study population was limited to women ages 40-75 years, and therefore our results may not be generalizable to women >75 years. However, statin medication initiation for primary prevention of CVD is not recommended in adults 76 years or older, and so this limitation is not relevant to the use of BAC as a biomarker for CVD and the decision to start preventative pharmacotherapy.³⁴

Screening mammography is highly utilized, and due to improvements in breast cancer screening and treatment, the number of breast cancer survivors with RT history continues to increase. Considering efforts underway to incorporate universal reporting of BAC on screening mammography, it is important to understand the association between BAC and RT exposure. Our study does not show an association between BAC and RT exposure in women ages 40-75 years, and if confirmed in additional studies, universal reporting of BAC will not require adjustment for RT exposure to assess CVD risk in this population.

Conclusion

Our study is the first retrospective analysis to evaluate the prevalence of BAC on screening mammograms in women with and without breast cancer radiation treatment exposure. Our findings suggest that breast RT exposure is not associated with the prevalence of mammographic BAC. If confirmed in additional studies, RT exposure does not impact the utility of BAC as an imaging biomarker of CVD risk.

Funding

This study is based on original research which was funded by the Cardiovascular Medicine Research Award (sourced from the Dartmouth-Hitchcock Medical Center Heart and Vascular Center).

Conflicts of interest

The authors have no conflicts of interest or sources of funding to disclose for this research study.