The utility of radiographic assessment of the internal mammary arteries in chest wall irradiated patients

Purpose: The internal mammary arteries (IMA’s) are historically recognized to be protected against atherosclerosis. Whether chest wall-irradiation for breast cancer leads to significant IMA damage remains unclear. The utility of computed tomography (CT) and mammography to detect radiation-induced damage to the IMA’s and its branches is not known. The objective of this study is to assess the susceptibility of IMA’s to radiation-induced atherosclerosis, and the utility of CT scan and mammography in the assessment of IMA and its branches. Methods: A retrospective analysis of breast cancer patients who received chest wall-radiotherapy was performed. Patients with CT scans and/or mammograms ≥5 years post-radiotherapy were included. Baseline characteristics, coronary artery calcification (CAC), the presence of IMA damage assessed by CT scan, and IMA branch calcifications by mammography were recorded. Results: None of the 66 patients with CT scans post-radiotherapy revealed IMA atherosclerosis. There were 28 (42.4%) patients with CAC, of which four (14.3% of CAC subgroup or 6.1% of the total cohort) had calcifications on either side on mammogram (Chi-square test, p = 0.74). Out of the 222 patients with mammograms, 36 (16.2%) had IMA branch calcifications. Two hundred and ten patients received unilateral radiotherapy, and 27 (12.9%) of these patients had calcifications on the irradiated side, and 26 patients (12.4%) had calcifications on the contralateral side (OR = 1.0). Conclusion: IMA’s do not exhibit signs of radiation-induced atherosclerosis when evaluated by CT scan. In addition, there is no association between radiotherapy for breast cancer and the presence of IMA branch calcification on


Introduction
An expanded role for radiation therapy in the management of breast cancer has come at the cost of radiationinduced coronary artery disease (RICAD). Compared to non-irradiated patients, patients who underwent chest wallradiotherapy for breast cancer have a higher absolute risk of cardiac morbidity and death, with the increase in risk being proportional to the radiation dose [1]. Ionizing-radiation induces reactive oxygen species resulting in accelerated atherosclerosis by endothelial damage and the impaired ability to clear free radicals [2], which leads to a chronic inflammatory state [3]. Atherosclerotic lesions in RICAD are more proximal than those in non-irradiated vessels and the plaques are largely composed of fibrous tissue [4]. In left-sided breast cancer, the anterior and apical wall of the heart, and the internal mammary arteries (IMA) can be in the radiation field.
The frequency of IMA graft failure after coronary artery bypass grafting (CABG) is common (8.6%) [5], and the question of whether chest wall irradiation for the treatment of breast cancer leads to IMA damage remains unclear. Despite limited evidence, careful angiographic or radiographic evaluation and inspection of the IMA's before grafting has been recommended [6][7][8]. However, there is a paucity of data demonstrating radiographically that the IMA can be affected by radiation, or if it is inherently protected against atherosclerosis.
The present study aims to assess the utility of radiographic assessment of the IMA's and its branches by CT scan and mammography.

Setting and design
This is a retrospective cross-sectional analysis conducted at Saint Francis Hospital and Medical Center, a 617bed, urban tertiary care hospital in Hartford, CT. The hospital Institutional Review Board approved the study. After screening 1292 patients with breast cancer who received chest wall-radiotherapy at the single center within the study period, a total of 256 patients with a follow up imaging study at least 5 years or later from the date of initial radiation were included in the analysis. Inclusion was limited to breast cancer patients as opposed to other mediastinal cancers to preserve the homogeneity of radiation therapy protocols and dosing within the cohort.
All identifiable patients with breast cancer who received chest wall-radiotherapy from January 1st, 2003 to January 1st, 2014, were screened by chart review of electronic medical records. Patients with available chest CT scans and/or mammograms ≥5 years post-radiotherapy were included. Patients who had IMA grafting were excluded. Patients were included in either the CT analysis, mammogram analysis, or both, depending on test availability for a given patient (Fig. 1). Data abstraction was performed independently by two abstracters (VN and MA). For each patient, the baseline demographics, cardiovascular risk factors, laboratory data, concomitant chemotherapy, mastectomy, and radiation therapy characteristics were collected by chart review.

Image acquisition and analysis
Baseline (pre-radiation therapy) CT scan, and the latest CT scan (post-radiation therapy), along with the latest mammogram (post-radiation therapy) were independently interpreted by two cardiac radiologists (SB and AP). The readers were blinded to the laterality of breast cancer or radiation therapy. Any discrepancies were settled by joint review discussions and consensus.

Mammography
Surveillance mammograms were obtained with a Hologic scanner in craniocaudal and mediolateral-oblique projections. For mammograms, craniocaudal projections for each breast were evaluated for the presence of tram track vascular calcifications. The vascular calcifications in the breasts on mammograms are Monckeberg calcifications from medial arteriosclerosis [9]. The IMA's supply the medial or central aspect of the breast parenchyma [10]. Therefore, calcifications found medial to the midline of the breast parenchyma were considered to be within the IMA supply territory.

Computed tomography
For chest CT scans, eighteen CT angiograms (CTA and CCTA), twelve contrast-enhanced CT scans (CECT), thirty-three non-contrast CT scans and, three CT scans with and without contrast enhancement were included for analysis. Calcium scoring CT scans were not included as the entire extent of IMA's was not imaged by them. CT scans were obtained on Siemens Flash dynamic 256-detector CT and Siemens Definition 128-slice and Siemens Somatom 16-slice scanners. CECT and CT angiograms (CTA and CCTA) were obtained with an injection of 70 to 120 mL of Isovue 370 iodinated contrast. For CCTAs, electrocardiogram (ECG) gating was used.
Qualitative evaluation of the CT scans in trans-axial planes was performed for the presence of radiographicallydetected damage to the IMA's including the presence of IMA calcifications, patency, and occlusion/atresia. The presence of coronary calcifications was also recorded from the scans when assessable even on the non-gated studies and contrast-enhanced studies. IMA's were evaluated visually for calcifications. IMA patency was defined as a contrast column in the lumen as assessed by CTA or CECT. IMA occlusion or atresia was defined as non-visualization or an abrupt change in the vessel diameter at any site along its course in comparison to a proximal reference patent segment, as assessed by CTA or CECT. If streak or motion artifacts limited the evaluation, the CT was excluded from the analysis. Coronary calcifications if detected on visual interpretation were also recorded. No software calculations for calcium scoring were used. Patients with only CTA and CCTA had no pre-contrast imaging to accurately identify coronary calcifications. However, the presence or absence of calcifications was documented based on qualitative visual interpretation of these contrast-enhanced scans.

Radiation treatment planning
Radiation therapy was administered with 6, 10, or 18 megavolt photons to a standard field defined by the anatomic margins of the pre-operative breast: head of the clavicle, 2 cm below the inframammary fold, mid-sternum, and mid-axillary line. CT simulation was performed, and 3D planning was utilized. Standard fractionation with doses of 1.8-2 Gy per fraction was most common. Accelerated hypofractionation (2.66 Gy per fraction) was utilized in select cases. Tumor bed boost was used at the physician's discretion. Nodal radiation to axillary and supraclavicular lymph node basins was used when there was pathologic evidence of lymph node involvement. In the event of internal mammary lymph node involvement, a medial electron strip was utilized.

Statistical analysis
Continuous variables are expressed as mean ± SD and frequencies or percentages for categorical variables which were calculated for all patient characteristics. Three groups of patients were compared by the testing procedure they received: mammogram only, CT scan only, or both. The groups were compared using analysis of variance (ANOVA) for continuous variables (i.e., age, lab values) and chi-square tests for categorical variables (i.e., cardiac risk factors). All effects were considered significant at p-value less than 0.05. All the statistical analyses were performed with Statistical Analysis System (SAS) 9.4 software (Cary, NC, USA).

Baseline characteristics
The baseline demographic and clinical characteristics of the patients are summarized in Table 1. One hundred and ninety (74%) of them had follow up mammograms only, 34 (13%) had chest CT scans only, and 32 (12%) had both, mammograms and CT chest as follow up imaging modalities.
The mean follow-up duration of patients per individual imaging modality after initial radiation therapy is provided in Table 2. Breast cancer site and treatment modalities are listed in Table 3.
Variation in baseline characteristics, laboratory values, and incidence of co-morbidities across different imaging modalities is summarized in Supplementary Tables 1,2.
The mean and mode values of total radiation dose administered per patient were 6167 and 6440 cGy respectively. The mode values of tangent and boost doses received  per patient were 5040 and 1400 cGy respectively. Minor variations in radiation dose across different imaging subsets is outlined in Supplementary Table 3.
Disparities in the laterality of breast cancer and treatment details amongst the patients in different imaging groups are listed in Supplementary Table 4.
Significant proportion of the study group had dyslipidemia, which is a risk factor for atherosclerotic disease, that could be confounding the results. Excluding the 59.2% of patients with dyslipidemia and those with unknown dyslipidemia status, 19 patients with CT scans and 89 patients with mammograms at least 5 years after initial radiation therapy were analyzed independently as a subset of the total cohort.

Computed tomography
None of the 66 patients with CT scans, at least 5 years after initial radiation therapy, revealed IMA calcification or stenosis. One patient had known IMA atresia which was noted on a previous CT scan prior to receiving radiotherapy and was likely related to damage incurred during a surgical lymph node dissection.
Out of the patients that had post-radiation CT scans where the coronaries were assessed, 28 out of 66 (42.4%) had coronary artery calcifications (CAC) present (Fig. 2). Of those 28 patients with CAC visualized on follow up CT scans, only 9 patients had pre-radiation CT scans available for comparison. Out of these 9 patients, only 2 (22.2%) patients had no coronary calcification on pre-radiation imaging and therefore they were considered as new calcifications. Of the 28 patients who had CAC seen on follow up CT scans, only 4 (14.3%) had mammogram findings of calcification on either side. There was no statistically significant association found between CAC and breast calcification found on mammograms (p = 0.74) by Chi-square test.
None of the CT scans of the 19 patients without dyslipidemia revealed IMA calcification or stenosis. Ten of these patients had coronaries assessed on follow up CT scan and CAC was identified in 6 of them. Only 3 of the 6 patients had pre-radiation CT scans available for review and CAC was present in all three of them.

Mammography
A total of 222 patients had mammograms and 36 (16.2%) were found to have calcifications (Fig. 3) on either side. Two hundred and ten patients received radiation to one side only. Further analysis of the mammogram data for the presence of calcifications in relation to radiation therapy revealed that 27 out of 210 (12.9%) patients had calcifications on the side they received radiation therapy. Twenty-six of the 210 (12.4%) patients had calcifications despite not receiving ipsilateral radiation therapy. Therefore, compared to the non-irradiated side, the odds of developing calcification on the radiated side were similar (OR = 1.00). Among the 89 patients without dyslipidemia that had mammograms, 9 (10.1%) were found to have calcifications on either side. Six of the 89 (6.7%) patients had calcifications on the side they received radiation and an equal number of patients (6.7%) had calcifications despite not receiving ipsilateral radiation therapy.

Discussion
Ischemic heart disease rates increase with exposure to radiation therapy among breast cancer patients [1,11,12]. The effect of radiation on IMA remains uncertain given insufficient radiographic data demonstrating vascular damage. Several studies showed concerns that radiation might increase IMA fragility, leading to early graft failure [6,8,13,14]. On the other hand, IMA's have been historically recognized to have inherent protection against atherosclerosis compared to other arteries [15][16][17]. Using histologic analysis, recent studies reported no restrictions in the use of irradiated IMA's as grafts [18,19]. Other studies also used his-tomorphologic [19], and echocardiographic doppler investigations [20] to demonstrate that the IMA's are protected from radiation-induced damage [19]. Yet, a SCAI Expert Consensus Statement recommended using CCTA as an assessment tool for IMA's which could be affected by radiation therapy rendering it unfeasible as a graft [7].
This study showed that none of the patients with CT scans had new IMA calcifications, stenosis, or atresia at least 5 years after initial radiation therapy. The patient cohort represents a relatively intermediate-high risk cohort based on high rates of cardiac risk factors such as advanced age, hypertension, diabetes, dyslipidemia, and smoking history, despite them being of the female gender. As expected, based on this risk factor profile, CAC was seen in 42.4% of the patients. Yet, only 22% of the patients with coronary calcifications post-radiation were considered new calcifications. Given the small number of patients with preradiation CT scans, definitive conclusions as to the effect of radiation-induced atherosclerosis of the native coronary arteries in this cohort cannot be made. Nonetheless, the percentage of patients in this cohort with coronary calcification post-radiation is higher than what has been seen in prior studies with similar patient risk-profile [21]. The finding of new CAC on post-radiation CT scan compared to preradiation imaging was not observed in the small sub-set of patients without dyslipidemia, highlighting the influence of atherosclerotic risk factors, independent of radiotherapy on the study results. There was no statistically significant association found between CAC and breast calcifications found on mammograms. Finally, chest wall-irradiation was not associated with the development of vascular calcifications of the IMA branches when evaluated by mammography.
Study shortcomings such as inadequate radiation dose to cause IMA injury, insufficient follow-up duration, or less significant atherosclerotic risk profile of the patient cohort were considered as possible reasons for the absence of IMA damage. A cumulative dose of radiation >3000 cGy has been considered as the dose associated with a high risk for radiation-induced heart disease [22,23]. The mode value of the administered radiation dose in this study was 5040 cGy, and is higher compared to that reported in studies by Renner et al. [24] (3600 cGy), and Brown et al. [8] (3000-6000 cGy) which inferred that IMA's are susceptible to radiationinduced damage. In the studies reported by Gansera et al. [19] and Handa et al. [25] which concluded that there are no restrictions to the use of IMA conduits post-radiotherapy, the total dose reported was similar to the dose in the present study, i.e., 5000-6000 cGy. The mean duration from initial radiation exposure to IMA evaluation in this study was 10.5 years which is similar to that reported by Renner et al. [24] (11 years). The mean duration to the development of RICAD was reported as 9 years [26]. The prevalence of atherosclerotic risk factors in this study is comparable to that reported in a cohort of 25 patients by Brown et al. [8], which concluded that IMA graft patency in patients who received prior radiation therapy is less than expected for an arterial graft, and comparable to venous grafts. Compared to the patients in their study, the patient population in our study had a higher mean age in years (60 vs 56), a higher prevalence of hypertension (63% vs 52%), and diabetes mellitus (26% vs 16%). From this data, it can be inferred that the patients in this study were exposed to a high enough dose of radiation, had an adequate long-term follow-up, and found to have a significant risk factor profile to observe IMA damage from radiation if it were to occur.
Several reasons may explain why none of the IMA's showed atherosclerotic disease on CT scan, (a) CT scans are insensitive to detecting radiation-induced IMA damage as an imaging modality, (b) IMA's are inherently protected against radiation-induced atherosclerosis, (c) Standard chest wall radiation does not result in sufficient IMA dose to produce detectable injury, (d) Longer follow up duration may be required. Although the diameter of the IMA's is smaller than that of the coronary arteries [27], multidetector CT angiograms have been studied to be adequate for the anatomical evaluation of the IMA's [27,28]. The latest published SCCT Appropriate Use Criteria [29] recommends cardiac CT to localize bypass grafts and other retrosternal anatomy before re-operative cardiothoracic surgery, and with the evolution of improved scanner technology, CT scan as a means to noninvasively assess the IMA's is being pursued [30]. However, there remains a paucity of data in the published literature regarding the utility of CT scan to evaluate for IMA calcification and stenosis.
There is conflicting data in the current literature as to whether the IMA's are inherently protected from the effects of radiation therapy [6,13,14,19,20,24]. Histomorphological analysis performed by Gansera et al. [19] studied 133 irradiated IMA's and compared them with a control group of 133 non-irradiated IMA's. Histomorphologic investigations did not identify severe fibrosis or radiation-induced damage of the IMA grafts. Nasso et al. [20] compared two groups of patients undergoing elective CABG with the use of IMA. There was no significant difference in intraoperative IMA flow when assessed by transthoracic echocardiography doppler between irradiated and non-irradiated individuals. Similarly, a retrospective clinical review by Handa et al. [25] of 47 patients undergoing CABG after mediastinal radiation therapy observed good early postoperative results without IMA graft failure. On the other hand, some studies suggest a higher risk of IMA graft failure after CABG in patients with prior radiation therapy [6,8,13]. The failure of IMA as a graft may be related to increased vessel friability and/or extensive mediastinal fibrosis resulting in intraoperative injury rather than radiation induced atherosclerosis [14,18,24]. This could explain the poor outcomes after CABG despite the absence of stenosis/calcification in IMA's.
The prevalence of calcifications identified on mammograms in the present patient population was 16%. This is similar to the data published by Gunhan-Bilgen et al. [31], who observed a similar prevalence of breast calcification in post-radiation surveillance mammograms in breast cancer survivors. The prevalence of breast arterial calcification by mammogram in the general population ranges from 9 % to 17 % in women aged between 50-65 years and has been noted to be as high as up to 50% in older women [9]. As expected, the prevalence of calcifications on mammogram in the sub-set of patients without dyslipidemia is lower at 10% in comparison to the 16% observed in the entire cohort.
The clinical significance of calcifications seen on mammograms remains unclear. A statistically significant association between CAC and breast calcification was not observed. This is probably due to the small number of patients with CAC in our study. Breast artery calcification is known to be associated with the presence and even severity of CAC [32][33][34], but its diagnostic accuracy in predicting obstructive coronary artery disease is poor [33].
The precise anatomical site of breast calcification observed on mammograms can be difficult to localize. The compression of the breast tissue during the study itself can make it difficult to discern between the branches of IMA and other vasculature such as branches of the lateral mammary artery. There is also the possibility of macroscopic dystrophic/extravascular calcium deposition in the breast tissue which is not uncommon on screening mammograms [35]. Despite the calcifications seen on mammograms, no calcification of the IMA's was observed on the CT scans. The IMA branches may have different histological features compared to the IMA's, making them more susceptible to radiation-induced endothelial damage. In contrast to the IMA branches which run a more superficial course making them vulnerable to radiation, the IMA's run a relatively posterior course, perhaps shielded by the intercostal muscles and fascia.

Conclusions
This study provides a novel outlook in evaluating the radiographic assessment of IMA's with CT scans and mammograms in breast cancer patients exposed to chest wallradiation. To our knowledge, this is the first study to date to examine the utility of CT scans and mammograms in detecting IMA damage post-radiation therapy. Our study has some limitations. While this is a retrospective study conducted at a single center with small sample size, the strength of this analysis stems from the prospective, blinded examination of CT scans and mammograms by two independent radiologists. Our study included breast cancer patients only; this cohort was selected intentionally so that the effects of standardized radiation protocols could be compared. These findings, therefore, may not apply to other malignancies such as Hodgkin's Lymphoma that require a different dose and technique of radiation. The negative controls used were not healthy controls but rather the contralateral, non-irradiated sides. Although, this allows for equal baseline characteristics between the negative control and study sample, the absence of difference in calcifications between irradiated and non-irradiated sides in mammogram could have been potentially due to radiation scatter to the non-irradiated side. Our study however is hypothesisgenerating and future prospective studies with standardized imaging protocols (both chest CT and mammograms) should be entertained.

Ethics approval and consent to participate
Our study received the proper ethical oversight through our institutional IRB (SFH-18-110).