Introduction
There is accumulating evidence that radiotherapy (RT) for breast cancer (BC) can lead to subsequent ischemic heart disease (IHD) [
1‐
4]. The incidence of BC is increasing but prognosis has improved substantially due to earlier detection by screening mammography and to more effective adjuvant therapies, including RT [
2,
5,
6]. As more patients become survivors, the balance between benefit and harm, including potential long-term side effects, of BC treatments is important.
The incidental cardiac radiation doses are generally higher in left-sided RT compared to right-sided RT, and several studies show an increased risk of IHD after RT in women with left-sided compared to right-sided BC [
1,
7,
8]. However, studies in more recently treated patients show conflicting results, and since RT-induced IHD is generally considered to be a late event, the follow-up may be too short draw firm conclusions [
9‐
12]. A higher incidence of coronary stenosis in the left anterior descending artery (LAD) has been reported after RT of left-sided compared to right-sided BC, and some studies show a relationship between LAD radiation dose and the risk of developing coronary stenosis [
13‐
17]. Increasing awareness of long-term cardiac toxicity and the development of new radiation techniques have contributed to reduced cardiac radiation dose exposure over recent decades. The anterior part of the heart, including the LAD, may still receive considerable doses in left-sided RT [
15,
18‐
23].
Retrospective studies often rely on the validity of diagnosis coding in patient charts and in population-based registers. Since symptoms from the chest are common and multifactorial, there is a risk that misdiagnosis or overdiagnosis can influence the results. In this study, we examined the risk of coronary stenosis in women who received adjuvant RT for BC, and who were subsequently referred to coronary angiography and percutaneous coronary intervention (PCI).
Methods
Study population
The study population was created by selecting all women diagnosed with BC between 1992 and 2012 in three of Sweden’s six health care regions: Uppsala-Örebro, Stockholm, and the northern region. Since the 1970s, regional breast cancer registries have collected information regarding histopathological data and BC treatments in Swedish women with BC, and from 2008, all BC patients have been registered in the National Quality Registry for Breast Cancer [
24]. Data on women with BC were then linked to the Swedish Coronary Angiography and Angioplasty Registry (SCAAR) to identify women with BC who subsequently underwent coronary angiography. The SCAAR is a part of the nationwide Swedish cardiac register SWEDEHEART and contains information on all patients who are referred to angiography and angioplasty. The register comprises information on baseline characteristics and detailed descriptions of angiographic findings and coronary interventions [
25].
Women with a pathological coronary finding prior to or within 180 days from BC diagnosis based on registrations in the SCAAR or other SWEDEHEART registers were excluded. Women who had undergone coronary angiography with normal findings or only atheromatosis prior to or within 180 days from BC diagnosis were, however, included. Women with bilateral BC, with unknown BC laterality, with previously irradiated in situ BC, and with metastatic disease at BC diagnosis were excluded. A flowchart for study inclusion is shown in Figure S‑1.
Statistical methods
Two types of time-to-event analyses were performed: time to a coronary finding and time to a PCI. Date of inclusion was defined as 180 days from BC diagnosis, and the women were followed to the date of first coronary finding/PCI in the SCAAR, death, migration, or end of follow-up (15 January 2021), whichever came first. Women with a normal coronary finding in SCAAR at the start of follow-up were excluded from the time-to-coronary-finding analysis.
To assess the cumulative incidence of coronary findings, competing risk analyses were performed censoring for death and end of follow-up, and considering the various types of coronary findings as competing risks. Findings were classified as follows: left main coronary artery (LMCA) disease, one-vessel disease (LMCA excluded), two-vessel disease (LMCA excluded), three-vessel disease (LMCA excluded), coronary atheromatosis, or normal angiographic findings. Results were presented using stacked cumulative incidences of findings presented separately for women diagnosed with right- and left-sided BC. The analyses were performed for the whole cohort and stratified by RT.
To estimate the risk of a PCI, Cox proportional hazard regression analysis was performed by comparing women who had received left-sided RT to women who had undergone right-sided RT. The analyses were performed separately for the LMCA, the LAD (proximal, mid, and distal LAD), the right coronary artery (RCA) (proximal, mid, and distal RCA), and the left circumflex artery (LCx). The proximal, mid, and distal LAD, and proximal, mid, and distal RCA were also analyzed separately. Coronary artery segments not included in the LAD, RCA, or LCx were grouped together and defined as “other locations.” The analyses were stratified for RT, type of surgery, and lymph node status recorded in the BC registries. Analyses were also performed separately on the risk of a PCI in the LAD in women treated between 1992 and 2001, and in women treated between 2002 and 2012. Separate analyses on the risk of a PCI in the RCA, the LCx, or other location were performed for women receiving RT, stratified for the presence or absence of a simultaneous PCI in the LAD.
The risk of a PCI in the LAD and RCA, comparing women receiving RT to women not receiving RT for right-sided BC and left-sided BC, respectively, was also assessed by Cox regression analysis. These analyses were adjusted for type of surgery, chemotherapy, and endocrine therapy, and age was used as the timescale.
All analyses were performed using the statistical software R [
26].
Discussion
The main finding of the present study was an increased risk of a PCI in the LAD in women receiving left-sided RT compared to right-sided RT for BC (HR of 1.47, 95% CI 1.21–1.77). The risk of a PCI increased with more advanced nodal disease and was most pronounced in the distal LAD (HR 2.43, 95% CI 1.33–4.41). An increase in the cumulative incidence of coronary findings in women receiving left-sided RT compared to right-sided RT was apparent at 10 years from BC diagnosis and increased with longer follow-up, with an absolute increase in risk of 0.4% at 10 years of follow-up, and of 2.6% at 25 years of follow-up.
To our knowledge, this is the largest study to evaluate the risk of coronary findings and PCI in patients who undergo coronary angiography after previous adjuvant RT for BC. The registrations in the SCAAR give information on the exact localization of a PCI, and also indicate that these findings were of clinical significance to the patients.
Many previous studies on radiation-induced cardiac disease rely on diagnoses coded in patient charts and in population-based registers, and the accuracy of these registrations may affect the validity of the results [
1,
7,
27]. Diagnosing IHD and heart failure can be challenging in clinical practice, as symptoms associated with cardiac diseases are common and multifactorial, and the risk of both over- and misdiagnosis must be considered [
28,
29]. A strength with the present study is that all coronary findings and interventions were confirmed by coronary angiography and classified by cardiologists in the SCAAR. The register has been previously validated, with high coverage for patients undergoing coronary angiography and PCI [
25].
Several weaknesses of the study need to be addressed. Information on individual radiation doses and fields were not available. However, most women likely received a dose of 2 Gray (Gy) to 50 Gy over 5 weeks, according to regional and national BC treatment guidelines throughout most of the study period [
30]. Most women with axillary lymph node metastases also received RT to the axilla, the supraclavicular fossa, and in some cases to the internal mammary nodes (IMN). Deep inspiration breath hold (DIBH) techniques are shown to lower the radiation doses to heart and LAD [
20,
21] and the risk of coronary events may be lower if DIBH techniques are used. The women in the present study were treated before these techniques were implemented in Sweden. The comparison of left-sided RT to right-sided RT is widely acknowledged in studies on cardiac side effects of RT [
1,
7,
11]. These risk estimates may, however, underestimate the total risk of coronary disease, since cardiac doses after right-sided RT can also be considerable, especially if the IMN are included in the radiation target [
20].
There are some recently published studies reporting coronary events after RT, registered by coronary angiography or coronary computed tomography angiography (CCTA) [
9,
15,
17]. In a study by Tagami et al. [
17], 94 patients treated with RT and subsequently diagnosed with coronary stenosis on CCTA were identified. In women treated with left-sided RT, a statistically significantly increased risk of coronary disease was seen in the LAD and the mean LAD radiation dose correlated with the risk of coronary disease. Based on these results, the authors suggested that ∼3 Gy may be a reasonable constraint for LAD doses at RT planning [
17]. Since individual radiation doses were not available in the present study, we could not estimate risk of coronary stenosis per Gy. On the other hand, the present study consisted of a considerably larger cohort of women than that of Tagami et al. [
17]. In a previous study by our group based partly on the same BC cohort as the present study, radiation dosimetry was studied in 182 women diagnosed with a coronary stenosis subsequent to RT. For these women, the mean median dose to the LAD was 10.9 Gy in left-sided RT [
15]. In the study by Tagami et al., women receiving left-sided RT had a higher incidence of coronary findings in RCA compared to women receiving right-sided RT [
17]. Individual differences in anatomy were discussed as one explanation for this finding, or the abscopal effect on tissues beyond the area of treatment. The analyses in the present study show a similar pattern, with an increased risk of PCI in the RCA in left-sided RT. However, when the analyses were stratified for a simultaneous PCI in the LAD, this increase in risk was no longer evident. These findings may be due to less clinically relevant coronary stenoses in the RCA that are detected and treated as a result of a symptomatic stenosis in the LAD.
In a study by Milo et al., a large cohort of Danish women treated with RT for BC between 1999 and 2016 was studied [
9]. An increased incidence rate ratio for a cardiac event in left- versus right-sided RT of 1.44 (95% CI 1.8–2.4) was seen in women treated before computed tomography (CT)-based RT planning was introduced in Denmark (1999–2007), findings that are in line with the results of the present study. In women treated with CT-based RT between 2008 and 2016, however, no increased risk of cardiac events was observed in left-sided vs right-sided RT. In contrast to the study by Milo et al., we could also show the specific location of the coronary stenosis, and by this show an even higher risk of PCI in parts of the LAD, which is a strength of the present study.
The results of this study stress the importance of implementing and further developing RT techniques that lower the cardiac doses to reduce the risk of long-term side effects that impair health and quality of life for BC survivors.