Cardiovascular Disease in Women with Breast Cancer – A Nationwide Registry Study

Background: There is increasing concern about cardiovascular disease (CVD) after breast cancer (BC). The aim of this study was to estimate the prevalence of different types of CVD in women diagnosed with BC compared to cancer-free controls as well as the incidence of CVD after BC diagnosis. Methods: We performed a cohort study based on data from national registries covering the entire Danish population. We followed 16,505 cancer-naïve BC patients diagnosed from 2003 to 2007 ve years before and up to 10 years after BC diagnosis compared to 165,042 cancer-free controls. Results: We found that 15.6% of BC patients were registered with at least one CVD diagnosis in hospital records before BC diagnosis. Overall, BC patients and controls were similar with regards to CVD comorbidity before BC diagnosis. After BC diagnosis, the incidence of all CVD diagnoses combined was signicantly higher in BC patients than controls up to approximately 6 years after the index date (BC diagnosis). The incidence of heart failure, thrombophlebitis/thrombosis and pulmonary heart disease including pulmonary embolism remained higher in BC patients than controls during the entire 10 year follow-up period. Furthermore, we found that the risk of heart failure and thrombophlebitis/thrombosis was higher after chemotherapy. Conclusions: Focus on CVD in BC patients is important to ensure optimum treatment with regards to BC as well as possible CVD. Strategies to minimise and manage the increased risk of heart failure, thrombophlebitis/thrombosis and pulmonary heart disease including pulmonary embolism in BC patients are especially important. on lipid metabolism [29–31], at the same time increases the risk of venous thrombosis and thromboembolism [32]. In our study, we did not nd a higher risk of CVD after hormonal therapy, but we see a higher incidence of thrombotic complications in BC patients compared to cancer-free controls.

Each person in the control group was assigned a random index date during the index year.

Outcomes and covariates
We de ned CVD as at least one primary or secondary CVD diagnosis in NPR, see Fig. 1. NPR holds information on all patients discharged from Danish hospitals since 1977 and on emergency department and outpatient visits since 1995 [16]. For each hospital contact, one primary and optional secondary diagnoses are registered according to the International Classi cation of Diseases (ICD). The primary diagnosis is the main reason for the hospital contact and secondary diagnoses identify other relevant diseases.
Furthermore, we identi ed BC patients with at least two CVD drug prescriptions (i.e. drugs related to CVD disease management) within the same year in NPRD, see Fig. 2. NPRD includes information on all primary care prescription drugs dispensed from community pharmacies in Denmark coded according to the Anatomical Therapeutic Chemical Classi cation System (ATC) [18]. We applied the criteria of two prescriptions to exclude persons who were apparently not in continuous treatment. It is not straightforward to identify CVD patients based on NPRD data as the drugs listed in Box 2 are not used exclusively for CVD disease management, and the NPRD does not contain the same information on diagnoses as NPR. However, NPRD data are still relevant to capture the large number of CVD patients who are not registered in NPR because their condition does not require hospitalization or outpatient visits to hospitals.
We used CCI as an aggregate measure of overall comorbidity [21] and diabetes at index date as covariates in multivariate Cox regression analyses. We calculated CCI using information from NPR on primary and secondary diagnoses up to ve years before the index date. We classi ed the study population into three groups with a CCI score equal to zero, 1, or 2 and above, respectively. We de ned diabetes at index date as at least one primary or secondary diabetes diagnosis in NPR (ICD-10: E10*-E11*) up to ve years prior to the index date or at least two insulin prescriptions or prescriptions for blood glucose lowering drugs in NPRD (ATC group A10A* and A10B*) within the same year during the 5-year period.
We identi ed persons in the BC group who had received different types of cancer treatments based on data from NPR. Surgery was de ned by a 'surgery procedure code' in combination with BC diagnosis. Chemotherapy was de ned by a BWHA procedure code, radiation therapy was de ned by a BWG procedure code, antibody therapy was de ned by a BOHJI procedure code, and hormonal therapy was de ned by a BWHC procedure code We identi ed persons in the BC and control group who died during follow-up (including date of death) based on data from the Civil Registration System. Data on age, region of residence and education were obtained from the Civil Registration System and education registers.
Statistical analysis CVD prevalence rates were calculated as number of persons with at least one CVD primary or secondary diagnosis in NPR up to ve years prior to the index date per 1,000 persons for all CVD diagnoses combined and speci c types of CVD. Furthermore, number of BC patients and cancer-free controls treated with CVD prescription drugs up to ve years prior to the index date per 1,000 persons were calculated based on NPRD data.
The cumulative incidence of all CVD diagnoses combined as well as speci c types of CVD were calculated up to 10 years after the index date (BC diagnosis) allowing for the presence of competing risk of death [22]. A competing risk is an event whose occurrence precludes another event of interest.
We performed multivariate Cox regression analyses to investigate the risk of developing CVD in BC patients following chemotherapy, radiation therapy, antibody therapy and hormonal therapy initiated within nine months after BC diagnosis. The risk of CVD was calculated based on either CVD diagnoses in NPR (model 1) or CVD prescription drugs in NPDR (model 2). Only BC patients without any CVD diagnosis in NPR ve years prior to the index date were included in model 1, and only BC patients without a history of CVD prescription drugs ve years prior to the index date were included in model 2. We included time dependent exposure dummies that classi ed the patients' time-to-event' as unexposed until treatment began to avoid immortal time bias [23]. BC patients who had received surgery, but not the cancer treatment in question (chemotherapy, radiation therapy, antibody therapy or hormonal therapy) were used as reference. Adjustments were made for differences between groups with regards to age, region of residence, education and cohort (year of BC diagnosis) as well as comorbidity. Persons who died during follow-up were censored at time of death.
The analyses were performed using Stata 14 (StataCorp, College Station, TX). The statistical signi cance of differences between groups were evaluated using two-sided t-tests with a 5% level of signi cance or 95% con dence intervals.

Results
The study population included 16,505 BC patients and 165,042 cancer-free controls, see Table 1. By matching design, BC patients and controls were identical with regards to age, region of residence, and education as well as gender (they were all women). In total, 156 per 1,000 BC patients (15.6%) were registered with at least one CVD diagnosis in hospital records up to ve years before the index date (BC diagnosis) compared to 151 per 1,000 cancer-free controls (15.1%), see Table 2. The most common type of CVD diagnosis was hypertension (7.3% of BC patients compared to 7.0% of controls) followed by ischemic heart disease/acute coronary syndrome (4.5% of BC patient compared to 4.7% of controls) and cerebrovascular disease (3.7% of BC patients compared to 3.6% of controls). The prevalence rates of CVD were not signi cantly different between BC patients and controls with the exception of atrial brillation (p = 0.0019), which was higher in BC patients (3.2% compared to 2.6% in controls). . Diuretics were the most common type of drug used by both BC patients and controls (20%) followed by agents acting on the renin-angiotensin system (14%) and beta blocking agents (11%). There was no signi cant differences between BC patients and controls (data not shown).
Up to approximately 6 years after the index date, the cumulative incidence of all CVD diagnoses combined was higher in BC patients than in cancer-free controls taking account of the presence of competing risk of death, see Fig. 3. Hereafter, the difference in cumulative CVD incidence between BC patients and controls was no longer statistically signi cant. After 10 years, 28% of both BC patients and controls (without any CVD diagnosis in NPR up to ve years before the index date) had at least one CVD diagnosis in NPR. The most common types of CVD diagnosis were hypertension with a cumulative incidence of 15% after 10 years followed by ischemic heart disease/acute coronary syndrome, cerebrovascular disease and atrial brillation (all with a cumulative incidence of 5% after 10 years).
During the rst 2 years after the index date, the cumulative incidence of most types of CVD was statistically signi cantly higher in BC patients than controls, see Fig. 4. However, only the cumulative incidence of heart failure, thrombophlebitis/thrombosis and pulmonary heart disease including pulmonary embolism remained higher in BC patients during the entire 10-year follow-up period. After 10 years, 2.7% of BC patients compared to 2.5% of controls had developed heart failure, 2.7% of BC patients compared to 1.5% of controls had developed thrombophlebitis/thrombosis, and 1.5% of BC patients compared to 1.0% of controls had developed pulmonary heart disease.
The risk of CVD was signi cantly elevated in BC patients after chemotherapy (HR > 1), no matter whether CVD was de ned according to CVD diagnoses registered in NPR (95% CI: 1.030-1.353), see Table 3 (model 1), or CVD drug prescriptions registered in NPRD (95% CI: 1.226-1.520), see Table 3 (model 2). Furthermore, the risk of CVD was signi cantly higher after radiation therapy (HR > 1), when CVD was de ned according to CVD drug prescriptions in NPDR (95% CI: 1.023-1.251), but not CVD diagnoses in NPR (95% CI: 0.887-1.117). There were no statistically signi cant results regarding the risk of CVD after hormonal or antibody therapy. Note) 95% con dence intervals in brackets. The following covariates were included in the models: Age, region of residence, education, cohort (year of diagnosis) and comorbidity at index date (CCI, diabetes or CVD (CVD at index date was de ned by either CVD drug prescriptions in NPDR (model 1) or CVD diagnoses in NPR (model 2)). Comorbidity at index date was associated with an increased risk of CVD during the follow-up period no matter whether comorbidity was measured by CCI, diabetes or CVD. * p < 0.05, ** p < 0.01, *** p < 0.001 When investigating the risk of different types of CVD after chemotherapy, we found that the risk of heart failure and thrombophlebitis/thrombosis was signi cantly elevated in BC patients who received chemotherapy compared to BC patients who did not, see Table 4. Furthermore, we found that the differences in risk of developing heart failure and thrombophlebitis/thrombosis increased during the 10-year follow-up period, see Fig. 5. Note) The following covariates were included in the model: Age, region of residence, education, cohort (year of diagnosis), distant metastasis when diagnose comorbidity at index date (CCI, diabetes or CVD (CVD at index date was de ned by CVD drug prescriptions in NPDR * p < 0.05, ** p < 0.01, *** p < 0.001

Discussion
We found that 15.6% of BC patients were registered with at least one CVD diagnosis in hospital records up to ve years before BC diagnosis, and a considerably higher percentage of patients had a history of CVD drug prescriptions (i.e. drugs related to CVD disease management). Overall, BC patients and cancer-free controls were similar with regards to CVD comorbidity at index date. The most common types of CVD diagnoses in both BC patients and controls were hypertension, ischemic heart disease/acute coronary syndrome, and cerebrovascular disease.
After BC diagnosis, the incidence of CVD was signi cantly higher in BC patients than controls for all CVD diagnoses combined up to approximately 6 years after the index date (BC diagnosis) taking account of the presence of competing risk of death. After 10 years, 28% of both BC patients and controls (without any CVD diagnosis in NPR ve years before the index date) had at least one CVD diagnosis in NPR. Only the cumulative incidence of heart failure, thrombophlebitis/thrombosis and pulmonary heart disease including pulmonary embolism remained higher in BC patients than controls during the entire 10- year follow-up period. Furthermore, the risk of heart failure and thrombophlebitis/thrombosis was signi cantly higher in BC patients who had received chemotherapy compared to BC patients who had not as well as cancer-free controls. Heart failure has been associated with chemotherapy in previous studies [24], and it is well established that cancer patients have increased risk of thrombotic complications including e.g. deep vein thrombosis, pulmonary embolism and arterial thrombosis [25].
Contrary to our results, Abel-Qadir et al. [11] found that the prevalence of different types of CVD was signi cantly higher in BC patients compared to cancerfree controls. A possible explanation for this discrepancy could be that Abel-Qadir et al. used different algorithms to determine the presence of CVD preceding the index date. Other possible explanations are differences in the de nition of the study population and matching criteria, as Abel-Qadir et al. included women diagnosed with early stage BC only, and controls were matched to BC patients according to gender and age, but not education as in our study.
[11] found a signi cantly higher incidence of CVD hospitalizations due to heart failure, cerebrovascular disease and arrhythmias in BC patients compared to cancer-free controls whereas we did not nd this for cerebrovascular disease or arrhythmias. In accordance with our results, Strongman et al.
[12] found a statistically signi cantly higher incidence of heart failure and venous thromboembolism in BC patients compared to cancer-free controls, but not signi cant differences in the incidence of arrhythmia, stroke or peripheral vascular disease. Furthermore, Strongman et al. found that the incidence of coronary artery disease was signi cantly lower in BC patients compared to controls.
Existing studies have shown an elevated risk of CVD in BC patients after anthracycline-based chemotherapy, radiation therapy and antibody therapy [11, [26][27][28]. Our study con rms an elevated risk of CVD after chemotherapy. Unfortunately, data available for our study did not allow distinction between different types of chemotherapy. We also saw a tendency towards a higher risk of CVD after radiation and antibody therapy, but these results were in general not statistically signi cant.
Hormonal agents like tamoxifen approved for BC treatment more than 30 years ago can have both bene cial and detrimental effects on the cardiovascular system [24]. Studies have shown that tamoxifen has a protective effect on lipid metabolism [29][30][31], but at the same time increases the risk of venous thrombosis and thromboembolism [32]. In our study, we did not nd a higher risk of CVD after hormonal therapy, but we see a higher incidence of thrombotic complications in BC patients compared to cancer-free controls.
The present study has several strengths. Firstly, it is a study at the population level based on a large real-world dataset with no selection bias or other potential issues related to more segmented and selective populations. Secondly, the study population was identi ed from CAR, which has high completeness and validity due to use of noti cations from different data sources and quality control [14]. Thirdly, we used prescription data from NPRD to identify CVD patients as a supplement to data from NPR, as the latter registry does not include patients who require primary care only. NPRD is the national registry of prescription drugs dispensed from community pharmacies and is considered both complete and valid from 1995 and onwards [18]. Fourthly, the study included a cancerfree control group, and the use of exact matching ensured that BC patients and controls were comparable with regards to age, region of residence, and education as well as gender (they were all women). We did not include CCI as a matching criteria to avoid overmatching, and we preferred exact matching to propensity score matching because exact matching ensures that groups are identical with regard to the matching criteria used. Finally, we allowed for the presence of competing risk of death when estimating the cumulative incidence function [22,33]. Since BC patients had a higher risk of dying during the 10- year follow-up period than controls, ignoring the presence of competing risk could result in substantial bias.
The study also has limitations. The study relies on diagnosis and procedure codes in NPR, which may contain errors or be incomplete, e.g. many doctors code the primary diagnosis only, even if the patient has other relevant diseases [34]. Furthermore, NPR does contain information on patients who require primary care only as mentioned above, and it is not straightforward to identify CVD patients based on NPRD data as prescription drugs used for CVD disease management may have other indications. Finally, there is a risk of confounding due to the observational nature of the study. The risk of confounding related to observable baseline characteristics was minimised by exact matching and a multivariable regression design, but BC patients and controls may differ on nonobservable lifestyle factors, that in uence the risk of CVD. However, we expect lifestyle factors to be similar in BC patients and controls because education is a proxy for lifestyle and was included as a matching criteria. Similar comorbidity at index date also indicate a comparable lifestyle between BC patients and controls.

Conclusions
In conclusion, we found that 15.6% of BC patients were registered with at least one CVD diagnosis in hospital records when diagnosed with BC. Overall, BC patients and controls were similar with regards to CVD comorbidity before BC diagnosis. After BC diagnosis, the incidence of all CVD diagnoses combined was signi cantly higher in BC patients than controls up to approximately 6 years after the index date. The incidence of heart failure, thrombophlebitis/thrombosis and pulmonary heart disease including pulmonary embolism remained higher in BC patients than controls during the entire 10 year follow-up period. Strategies to minimise and manage the increased risk of CVD in BC patients are important.
Neither informed consent from patients nor approval from an ethics committee was required by Danish law since the study was based on retrospective registry data only. Research Ethics Committees in Denmark assess only biomedical research.
The study was carried out in accordance with international ethical standards including the EU general data protection regulation. Data for statistical analyses were anonymized, and study results were reported so that it is not possible to identify individual persons.

Consent for publication
Not applicable.

Availability of data and materials
The data that support the ndings of this study are available from the Danish Health Data Authority and Statistics Denmark but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available.
Competing interests MJ, CK and MSJ report grants from P zer Denmark during the conduct of the study. TK and MA report personal fees from P zer Denmark during the conduct of the study. PBP and HK report personal fees from P zer Denmark during the conduct of the study and own shares in P zer Inc. outside submitted work.
MJ, CK and MSJ are employees of the National Center for Social Science Research (VIVE), which was a paid vendor to P zer Denmark on the project. VIVE is an independent research institute, which is under obligation by law to disseminate the results of its work to relevant public and private stakeholders and the public in general. MA and TK are medical doctors with expertise in breast cancer and cardiovascular disease, respectively. PBP and HK are employees of P zer Denmark.

Funding
The study was funded by P zer Denmark.
Authors' contribution MJ, CK, MSJ, PBP, HK, TK and MA were all involved in the conception and design of the study. CK and MSJ had the main responsibility for carrying out the statistical analyses, and MJ had the main responsibility for drafting the article. All authors were involved in the analysis and interpretation of data, revised the article critically and approved the nal version for publication.

Figure 1
Types of CVD included in the study (ICD-10 diagnoses) Figure 1 Types of CVD included in the study (ICD-10 diagnoses) Figure 2 CVD prescription drugs included in the study (ATC groups) Figure 2 CVD prescription drugs included in the study (ATC groups) Figure 3 Cumulative incidence of all CVD diagnoses combined, proportion with at least one CVD diagnosis Figure 3 Cumulative incidence of all CVD diagnoses combined, proportion with at least one CVD diagnosis Figure 4 Page 13/14 Cumulative incidence of speci c types of CVD diagnoses, proportion with CVD diagnosis Figure 4 Cumulative incidence of speci c types of CVD diagnoses, proportion with CVD diagnosis Figure 5