Long term adjuvant endocrine therapy and risk of cardiovascular disease in female breast cancer survivors: systematic review

Abstract Objective To investigate the effect of endocrine therapies on a wide range of specific clinical cardiovascular disease outcomes in women with a history of non-metastatic breast cancer. Design Systematic review and meta-analysis of randomised controlled trials and observational studies. Data sources Medline and Embase up until June 2018. Eligibility criteria for selecting studies Studies were included if they investigated the risk of a specific cardiovascular disease outcome associated with use of either tamoxifen or an aromatase inhibitor, or compared the two treatments, in women with a history of non-metastatic breast cancer. Appraisal and data extraction Relevant studies were originally identified and results extracted by one researcher, with a full replication of the study identification process by a combination of two other researchers. The Cochrane Collaboration’s tool for assessing risk of bias was used to assess risk of bias in randomised controlled trials, and this tool was adapted to assess risk of bias in observational studies. Results 26 studies were identified, with results for seven specific cardiovascular disease outcomes (venous thromboembolism, myocardial infarction, stroke, angina, heart failure, arrhythmia, and peripheral vascular disease). Results suggested an increased risk of venous thromboembolism in tamoxifen users compared with both non-users and aromatase inhibitor users. Results were also consistent with a higher risk of the vascular diseases myocardial infarction and angina in aromatase inhibitor users compared with tamoxifen users, but there was also a suggestion that this may be partly driven by a protective effect of tamoxifen on these outcomes. Data were limited, and evidence was generally inconsistent for all other cardiovascular disease outcomes. Conclusion This review has collated substantial randomised controlled trial and observational evidence on the effect of endocrine therapies on several specific cardiovascular disease outcomes including venous thromboembolism and myocardial infarction, progressing knowledge. Although the choice of aromatase inhibitor or tamoxifen will primarily be based on the effectiveness against the recurrence of breast cancer, this review shows that the individual patient’s risk of venous or arterial vascular disease should be an important secondary consideration. Systematic review registration Prospero CRD42017065944.


Ryden
2016 RCTs with long-term (at least 5 years) follow-up data of AI compared to tamoxifen or placebo with either efficacy (DFS and OS)  Appendix 2 -Systematic review search strategy Medline Breast Cancer MeSH terms breast neoplasms or carcinoma, ductal, breast or carcinoma, lobular or inflammatory breast neoplasms or unilateral breast neoplasms or triple negative breast neoplasms Keywords breast cancer or breast neoplasm* or breast tumour or breast adenocarcinoma or breast carcinogenesis or breast carcinoma or breast sarcoma Endocrine Therapy MeSH terms tamoxifen or aromatase inhibitors Keywords tamoxifen or aromatase inhibitor* or anastrazole or exemestane or letrozole or endocrine therapy Cardiovascular Disease MeSH terms cardiovascular diseases or heart diseases or cardiotoxicity or coronary artery disease or cardiomyopathies or heart arrest or heart failure or heart failure, diastolic or heart failure, systolic or heart valve diseases or aortic valve insufficiency or aortic valve stenosis or mitral valve insufficiency or mitral valve stenosis or pulmonary valve insufficiency or pulmonary valve stenosis or tricuspid valve insufficiency or tricuspid valve stenosis or angina pectoris or angina, unstable or angina, stable or myocardial infarction or stroke or venous thromboembolism or pulmonary embolism or pericarditis or peripheral vascular disease or arrhythmias, cardiac Keywords cardiovascular* or CVD or cardiac or cardiotoxi* or heart disease* or coronary artery dis* or revascular* or coronary bypass or artery bypass or aorta bypass or cardiomyopathy* or cardiopulmonary arrest* or cardiac arrest* or heart arrest* or heart failure or valvular*disease or valve disease or valve stenosis or valve insufficiency or angina* or heart infarc* or myocardial infarc* or heart attack or coronary infarc* or stroke or tia or transient ischemic attack or cerebrovascular accident or venous thromboembolism or deep*thrombo* or thromboem* or pulmonary embolism or pericarditis or peripheral vascular or peripheral art* or arrhythmia* or fibrillation or heart*flutter Limits • English language • Humans • 1960 -Current year Embase Breast Cancer Indexed terms breast cancer or breast tumour or basal like breast cancer or breast adenocarcinoma or breast carcinogenesis or breast carcinoma or breast sarcoma or estrogen receptor positive breast cancer or inflammatory breast cancer or triple negative breast cancer Keywords breast cancer or breast neoplasm* or breast neoplasm or breast tumour or breast adenocarcinoma or breast carcinogenesis or breast carcinoma or breast sarcoma Endocrine Therapy Indexed terms aromatase inhibitor or anastrozole or exemestane or letrozole or tamoxifen Keywords chemotherapy or anthracycline or daunorubicin or doxorubicin or epirubicin or cyclophosphamide or fluorouracil or methotrexate or taxoid* or taxane* or paclitaxel or docetaxel or tamoxifen or aromatase inhibitor* or anastrazole or exemestane or letrozole or endocrine therapy or trastuzumab or Herceptin or breast cancer treatment Cardiovascular Disease Indexed terms heart disease/ or cardiovascular disease/ or cardiotoxicity/ or heart arrhythmia/ or heart atrium arrhythmia/ or heart ventricle arrhythmia/ or atrial fibrillation/ or heart atrium flutter/ or heart ventricle arrhythmia/ or heart ventricle flutter/ or heart ventricle fibrillation/ or heart fibrillation/ or heart failure/ or acute heart failure/ or congestive heart failure/ or diastolic heart failure/ or systolic heart failure/ or heart ventricle failure/ or heart left ventricle failure/ or heart right ventricle failure/ or Ischemic cardiomyopathy/ or cardiomyopathy/ or congestive cardiomyopathy/ angina pectoris/ or stable angina pectoris/ or unstable angina pectoris/ or heart infarction/ or acute heart infarction/ or heart atrium infarction/ or pericarditis/ or valvular heart disease/ or aorta valve disease/ or mitral valve disease/ or pulmonary valve disease/ or tricuspid valve disease/ or aorta valve stenosis/ or mitral valve stenosis/ or heart valve stenosis/ or pulmonary valve stenosis/ or tricuspid valve stenosis/ or revascularization/ or heart arrest/ or cardiopulmonary arrest/ or cerebrovascular accident/ or venous thromboembolism/ or deep vein thrombosis/ or thromboembolism/ or embolism/ or vein thrombosis/ or peripheral vascular disease/ Keywords cardiovascular* or CVD or cardiac or cardiotoxi* or heart disease* or coronary artery dis* or arrhythmia* or fibrillation or heart*flutter or heart failure or cardiomyopathy or angina or heart*infarc* or myocardial infarc* or heart attack or coronary infarc* or pericarditis or valvular*disease or valve disease or valve stenosis or valve insufficiency or revascular* or coronary bypass or artery bypass or aorta bypass or cardiopulmonary arrest* or cardiac arrest* or heart arrest* or cerebrovascular accident or stroke or tia or transient ischaemic attack or venous thromboembolism or deep*thrombo* or thromboem* or pulmonary embolism or peripheral vascular or peripheral art* Limits Appendix 3 -Bias assessment criteria -cohort studies Exposure Low • Minimum exposure period or need for several prescriptions before classified • Exposure ascertained through prescription or pharmacy records High • Exposure ascertainment not clearly defined, or defined by patient or physician recall • Future information used to inform exposure status at baseline • Potential for exposure misclassification due to no information of exposure prior to index • No minimum exposure period or need for several prescriptions • Non exposed or referent group from a different population to exposed Outcome Assessment

Low
• Well defined diagnosis using hospital records, GP diagnosis, or similar methods • Method of outcome ascertainment has been clearly validated High • Unclear method of diagnosis, or diagnosis defined by patient or physician recall • Potential for differential misclassification due to different methods of outcome ascertainment being used for different exposure groups All women with a first invasive breast cancer diagnosed at KPSC between January 1, 1980, and July 1, 2000.
Women with a first-time diagnosis of breast carcinoma who were treated with tamoxifen or with bladder carcinoma, colorectal carcinoma, or non-melanoma skin cancer between January 1, 1991 andDecember 31, 1999. Women with other cancers (bladder, colorectal, and non-melanoma skin cancer), were selected to provide an unexposed population, because most women with breast carcinoma in the GPRD were treated with tamoxifen, and to increase the comparability of the exposure reference group to the tamoxifen-exposed group with respect to ongoing medical surveillance.

Exclusions
Any other malignancies besides breast cancer, a history of VTE or thrombophlebitis, stroke, angina pectoris, myocardial infarction, diabetes mellitus, chronic renal disease, hypertension, hyperlipidemia, intermittent claudication, systemic lupus erythematosus, epilepsy, connective tissue disorders or cystic fibrosis. Furthermore all potential cases were excluded if they underwent mastectomy, chemotherapy, radiotherapy, trauma (i.e. accident, bone fracture) or major surgery (i.e. abdominal surgery, hip replacement) within 6 months prior to the index date, who had recurrent or metastatic breast cancer, or who were in their terminal phase and died within 6 months after the index date (subjects who died from pulmonary embolism were included).
Patients with a subsequent primary cancer diagnosis (other than a second primary breast cancer, cervical cancer in situ, or basal or squamous cell skin cancer) before their stroke diagnoses were excluded from the study because the other cancer could alter their breast cancer treatment or their stroke risk. Patients with thromboembolic disease diagnoses other than stroke (i.e., myocardial infarction, venous thromboembolism, or pulmonary embolism) were excluded.
Women were excluded if they had a history of cancer, MI, angina pectoris, congestive heart failure, or HIV/acquired immunodeficiency syndrome before the study entry date. Women with known HIV infection were excluded because HIV infection may complicate cancer therapy, including adjuvant therapy. Women were required to have at least 1 year of recorded follow-up after their study entry date to assure adequate follow-up.

Intervention arm
Tamoxifen (any, currently exposed in VTE case-control analysis at index date) (n=133) Any tamoxifen (in stroke case-control analysis) (n=286) Current tamoxifen. Women who received 2 or more tamoxifen prescriptions within 1 year of their index date were considered current users (n=49)

Reference arm
Never or past tamoxifen (in VTE case-control analysis at index date) (n=64) Unexposed to tamoxifen (in stroke case-control analysis). Unlikely to have been prescribed AIs due to study period being before approval of AIs (n=246) Unexposed to tamoxifen. Unlikely to have been prescribed AIs due to study period being before approval of AI (n=158) A matched analysis was conducted by using conditional logistic regression models, and relative risk estimates (odds ratios) of developing VTE with regard to current and past use were obtained, using never users as reference group.
Case patients were compared with their individually matched control subjects using univariate and multivariable conditional logistic regression methods. Crude and adjusted odds ratios were estimated, and 95% confidence intervals were calculated. These analyses were limited to case patients who had their first stroke after their breast cancer diagnosis and their matched control subjects.
The risk of IHD was assessed for current tamoxifen users and according to the dose-response measures among all cases combined and among cases stratified by diagnosis of angina or MI. Odds ratios and 95% CIs were estimated using conditional logistic regression modeling Adjustments Cases and controls matched on age (within two years), duration of breast cancer (same year of breast cancer) and calendar year of VTE (same index date). Then analyses adjusted for BMI ( < 30, 30+ kg m−2, unknown), smoking status (never, ex, current, unknown), and hysterectomy status (yes, no) Menopausal status (pre-or perimenopausal, naturally postmenopausal, or menopausal because of surgery); history of hypertension (no, yes but not requiring medication, and yes requiring medication); history of diabetes (no, yes but not requiring medication, and yes requiring medication); chemotherapy (yes, no) Cases and controls were matched on the date of the case's IHD diagnosis, age (1 year), and study entry date (6 months). Analyses were further adjusted for BMI (kg/m2), treated hypertension, use of hormone replacement therapy, and smoking status. Information concerning these risk factors was ascertained from the data base on or before the index date Any tamoxifen (in MI case-control analysis) (n=216) Any tamoxifen during follow up (n=8232) Currently exposed to Tamoxifen (n=4710) or AI (n=9067). Patients who were simultaneously prescribed both drugs contributed analysis time to both the tamoxifen and AI group.

Reference arm
Unexposed to tamoxifen (in case-control analysis). Unlikely to have been prescribed Ais due to study period being before approval of AIs (n=165) Unexposed to tamoxifen. Unlikely to have been prescribed AIs due to study period being before approval of AIs (n=8057) Not currently exposed to either tamoxifen or AI therapy (n=29497) Case patients were compared with their individually matched control subjects using univariate and multivariable conditional logistic regression methods. Crude and adjusted odds ratios were estimated, and 95% confidence intervals were calculated. These analyses were limited to case patients who had their first stroke after their breast cancer diagnosis and their matched control subjects.
Follow-up was initiated 3 months after the surgery date. Follow-up ended on December 31, 2005. Risks of events were analyzed individually by year for the first 5 years of follow-up, and then cumulatively for Years 1 to 5. RRs and 95% confidence intervals were calculated as estimates of the association between tamoxifen therapy and incident thromboembolic events. Cox proportional hazards models were used to estimate crude HRs and adjusted HRs controlling for confounding, for years 1 to 5 individually, and for Years 5 to 10 taken together. the proportional hazards assumption was Propensity score matching was used. Cox proportional hazards models with time varying treatment variables were used to assess whether treatment with AIs or tamoxifen was associated with MI and stroke among women with breast cancer and to assess the association of breast cancer with the outcomes of interest. The time-varying treatment variables allowed women to contribute information to the treatment group when on treatment and to the control group when not on treatment; women who received both AIs and tamoxifen contributed to both groups. For each outcome event, women were followed from the time of their first diagnosis code only tested by adding a covariate to the model to represent the interaction between exposure and the log of survival time until the occurrence of the event or the censoring of their observation.

Adjustments
Menopausal status (pre-or perimenopausal, naturally postmenopausal, or menopausal because of surgery); history of hypertension (no, yes but not requiring medication, and yes requiring medication); history of diabetes (no, yes but not requiring medication, and yes requiring medication); chemotherapy (yes, no) Age, surgical procedures (other than breast cancer surgery), metastatic tumors other than breast cancer, radiotherapy, chemotherapy, diabetes, stroke, chronic obstructive pulmonary disease, and heart failure were assessed at baseline Patients with pre-existing cardiovascular disease, such as coronary artery disease, ischemic stroke, hemorrhagic stroke or peripheral artery disease were excluded Women were excluded if they were treated with tamoxifen or AIs in the year preceding breast cancer diagnosis. Women were also excluded if they had substantial exposure to both tamoxifen and an aromatase inhibitor. This was defined as >10% of the days during which either tamoxifen or an aromatase inhibitor was prescribed. Accordingly, women were only included if they were exposed to one of the drug categories for 90% of the days during which a study drug was dispensed.

Intervention arm
At least one tamoxifen prescription after the index date (n=17874) at least one tamoxifen prescription associated with the breast cancer diagnosis (n=2056) Aromatase inhibitors at index date (n=7049). Patients were analysed based on the drug category they were predominantly exposed to. Outcomes were compared using the Cox proportional hazard model for estimating hazard ratios and 95 % confidence intervals.

Reference arm
Survival analysis was assessed using Kaplan-Meier analysis, with the significance based on the log-rank test. The survival time was calculated from the date of enrollment to the development of AMI, ischemic stroke or hemorrhagic stroke. Multiple regression analysis was carried out using Cox proportional hazard regression analysis to evaluate the effect of tamoxifen use on determining the occurrence of AMI, stroke, or total cardiovascular events.
Time-to-event analyses were performed for MI, using tamoxifen as the reference treatment. Cumulative incidence function curves were used to estimate the cumulative incidence of MI over time after accounting for the competing risk of death. This allowed us to estimate the incidence of MI, given that some subjects will die before the occurrence of a cardiac event. IPTW using the propensity score was used to reduce the effects of measured confounding variables when estimating the effect of AIs versus tamoxifen. The PS model was estimated using a logistic regression model with receipt of AIs as the dependent variable and all covariates as the independent or explanatory variables. The variables that were chosen included markers of cardiovascular disease, cardiovascular risk factors, cancer severity, major non-cardiovascular comorbidities, health care utilisation, factors that increase risk of adverse cardiac events with breast cancer (left-sided disease, chemotherapy, trastuzumab, radiation), as well as medications that could impact risk of cardiovascular disease. Truncated weights were used to minimise undue influence from atypical individuals with very high weights. The distribution of measured baseline covariates was compared between treatment groups in the sample weighted by the inverse probability of treatment using standardised differences. Variables were determined to be well balanced if the standardised difference was Adjustments Congestive heart failure, rheumatic disease, renal disease, diabetes mellitus, hypertension, Charlson comorbidity index. Patients had to have been diagnosed with the comorbidity within 1 year prior to the index date. Information was obtained from the NHI database, and all of the diagnoses were identified from either a single report in the inpatient medicinal claims file or from no less than two reports in the outpatient medicinal claims files Diabetes mellitus, cardiac arrhythmia, hyperlipidemia, congestive heart failure, and chronic obstructive pulmonary disease, 365 days before the date of diagnosis of breast cancer. The diagnosis code of any comorbidity must have appeared at least twice and lasted longer than 30 days before officially being regarded as a comorbidity. Medications before enrollment were also reviewed within the database, which included angiotensin-converting enzyme inhibitors, βadrenergic antagonists, calcium-channel blockers, diuretics, statins, antiplatelet agents (aspirin or clopidogrel) and thiazides. primary operable, histologically node-negative, ERnegative breast cancer and a life expectancy of at least 10 years Eligible patients were postmenopausal women aged 80 years or younger (ABCSG trial 8) or 75 years or younger (ARNO 95) with histologically verified, locally radically treated invasive or minimally invasive breast cancer without previous chemotherapy, hormone therapy, or radiotherapy, and absence of organ metastases. Women must have had 2yrs of tamoxifen (20mg ) daily.

Exclusions
Women who had previously been diagnosed with cancer, except for those who had had in-situ carcinoma of the cervix or squamous-cell or basal-cell carcinoma of the skin, were not eligible.
No information given indeterminate menopausal status (or menopausal status maintained by medication), presence of secondary malignant disease, tumour infiltration of skin or breast muscle (T4 tumours), and presence of other concomitant serious medical conditions-eg, those involving bone marrow function, the central nervous system, uncompensated cardiac insufficiency, or uncontrolled local or systemic infection.

Intervention arm
Radiation therapy followed by tamoxifen (10mg) twice daily for 5yrs (n=891) CMF and tamoxifen (10 mg) twice a day (n=498) Anastrozole (1mg) daily for remainder of 5 year endocrine treatment following 2 years of tamoxifen (n=1602). Follow up began after initial tamoxifen.

Reference arm
Radiation therapy followed by placebo (n=890). Unlikely to have been prescribed AIs due to study period being before approval of AIs.
CMF and placebo (n=499). Unlikely to have been prescribed AIs due to study period being before approval of AIs.
Tamoxifen (20mg) daily for remainder of 5yr endocrine treatment following 2 initial years of tamoxifen (n=1597 Previous adjuvant tamoxifen therapy lasting 4.5 -6 years; histologically confirmed primary breast cancer; a tumor that was positive for estrogen receptor, progesterone receptor, or both (defined by a level of 10 fmol/mg protein or a positive result on immunohistochemical analysis of ER or PR); discontinuation of tamoxifen therapy less than 3 months before enrollment; an Eastern Cooperative Oncology Group performance status of 0, 1, or 2 (scored on a scale of 0 to 4, with lower scores indicating better function); a life expectancy of more than 5 years; and postmenopausal status. Women were defined as being postmenopausal if they were at least 50 years of age at the start of adjuvant tamoxifen therapy, were younger than 50 years at the start of tamoxifen therapy but postmenopausal at the initiation of tamoxifen therapy, were younger than 50 years at the start of tamoxifen therapy but had undergone bilateral oophorectomy, were premenopausal and younger than 50 years of age at the start of tamoxifen therapy but became amenorrheic during chemotherapy or treatment with tamoxifen, or were any age but had postmenopausal levels of luteinizing hormone or folliclestimulating hormone prior to study enrollment. Women with unknown hormone receptor status were eligible, provided an effort was made to determine the receptor status of the primary tumor.
Histologically confirmed primary breast cancer, tumor estrogen receptor positivity , positive axillary nodes, and no evidence of recurrent or metastatic disease, who were receiving adjuvant treatment with tamoxifen for the last 2-3 years Patients were eligible if they had histologically confirmed, completely resected unilateral invasive breast carcinoma that was positive for estrogen receptors or that was of unknown receptor status. Patients were postmenopausal and had received adjuvant tamoxifen therapy for at least two years but not more than three years and one month. Patients were required to have adequate hematologic, renal, and liver function at the time of randomization. b

Exclusions
No information given Patients with a history or presence of any other cancer (except adequately treated skin cancer or carcinomain-situ of the cervix) and patients with any condition that may jeopardize their compliance to treatment or follow-up The presence of a tumor with known negative estrogen-receptor status; evidence of local relapse or a distant metastasis since the time of diagnosis; a clinically significant skeletal, cardiac, or endocrine disorder; and the use of hormone-replacement therapy within four weeks before randomization. Patients were also excluded if they had clinical evidence of severe osteoporosis or a history of a previous neoplasm other than carcinoma in situ of the cervix or basal-cell skin carcinoma or if they were taking concomitant anticoagulant agents, a selective estrogen-receptor modulator other than tamoxifen, or any other form of hormonal therapy.

Intervention arm
Letrozole (2.5mg) daily for 5 years, following previous adjuvant tamoxifen for 4.5-6 years. Follow up began after initial tamoxifen.
1mg of anastrozole daily for remainder of 5yr endocrine treatment following 2-3yrs of tamoxifen (n=223). Follow up began after initial tamoxifen.

Reference arm
Placebo daily for 5 years, following previous adjuvant tamoxifen for 4.5-6 years. Follow up began after initial tamoxifen.
20mg of tamoxifen daily for the remainder of their 5yr endocrine treatment following 2-3 initial yrs of tamoxifen (n=225). Follow up began after initial tamoxifen.
Tamoxifen (20 or 30mg ) daily for the remainder of their of 5yr endocrine treatment following 2-3 initial years of tamoxifen (n=2338 Histologically proven operable breast cancer confined to the breast and ipsilateral axilla, with the exception of internal-mammary-node involvement detected by means of sentinel-node biopsy, and tumor that expressed estrogen or progesterone receptors in at least 10% of the cells, as assessed with the use of immunohistochemical testing. Patients with synchronous bilateral hormonereceptor-positive breast cancer were eligible. Patients had undergone either a total mastectomy with subsequent optional radiotherapy or breast-conserving surgery with subsequent radiotherapy. Either axillary dissection or a negative sentinel-node biopsy was required. Macrometastasis in a sentinel node required axillary dissection or irradiation. Women eligible for the study were diagnosed with International Union Against Cancer stage I or stage II estrogen receptor-positive breast cancer between 1990 and 2004 at ages 45 to 69 years, as reported to the Danish Breast Cancer Cooperative Group (DBCG) clinical database

Exclusions
Women with no existing cardiovascular disease (defined using ICD-8 and ICD-10 codes) as of the date of breast cancer surgery

Intervention arm
In the TEXT study, 5 yrs of exemestane (25mg daily) plus triptorelin. In the SOFT study 5 yrs of exemestane plus ovarian supression. Any tamoxifen during follow up (n=8232)

Reference arm
In the TEXT study 5 yrs of tamoxifen (20mg daily) plus triptorelin. In the SOFT study 5 yrs of tamoxifen plus ovarian supression.
Unexposed to tamoxifen. Unlikely to have been prescribed Ais due to study period being before approval of AIs (n=8057 Number of outcomes reported for adverse events.
Follow-up was initiated 3 months after the surgery date. Follow-up ended on December 31, 2005. Risks of events were analyzed individually by year for the first 5 years of follow-up, and then cumulatively for Years 1 to 5. RRs and 95% confidence intervals were calculated as estimates of the association between tamoxifen therapy and incident CVD events. Cox proportional hazards models were used to estimate crude HRs and adjusted HRs controlling for confounding, for years 1 to 5 individually, and for Years 5 to 10 taken together. The proportional hazards assumption was tested by adding a covariate to the model to represent the interaction between exposure and the log of survival time

Adjustments
Age group, diabetes, renal disease, hypertension, chronic obstructive pulmonary disease, radiation therapy, and chemotherapy Relative risk taken from paper, or calculated from raw numbers Calculated Paper CVD outcome(s) Stroke, MI, Thromboembolic events Angina, stroke, MI, heart failure