Incident Cardiovascular Disease Among Adults With Cancer

Background Patients with cancer and cancer survivors are at increased risk for incident heart failure, but there are conflicting data on the long-term risk for other cardiovascular events and how such risk may vary by cancer site. Objectives The aim of this study was to determine the impact of a new cancer diagnosis on the risk for fatal and nonfatal cardiovascular events. Methods Using administrative health care databases, a population-based retrospective cohort study was conducted among 4,519,243 adults residing in Alberta, Canada, from April 2007 to December 2018. Participants with new cancer diagnoses during the study period were compared with those without cancer with respect to risk for subsequent cardiovascular events (cardiovascular mortality, myocardial infarction, stroke, heart failure, and pulmonary embolism) using time-to-event survival models after adjusting for sociodemographic data and comorbidities. Results A total of 224,016 participants with new cancer diagnoses were identified, as well as 73,360 cardiovascular deaths and 470,481 nonfatal cardiovascular events during a median follow-up period of 11.8 years. After adjustment, participants with cancer had HRs of 1.33 (95% CI: 1.29-1.37) for cardiovascular mortality, 1.01 (95% CI: 0.97-1.05) for myocardial infarction, 1.44 (95% CI: 1.41-1.47) for stroke, 1.62 (95% CI: 1.59-1.65) for heart failure, and 3.43 (95% CI: 3.37-3.50) for pulmonary embolism, compared with participants without cancer. Cardiovascular risk was highest for patients with genitourinary, gastrointestinal, thoracic, nervous system and hematologic malignancies. Conclusions A new cancer diagnosis is independently associated with a significantly increased risk for cardiovascular death and nonfatal morbidity regardless of cancer site. These findings highlight the need for a collaborative approach to health care for patients with cancer and cancer survivors.

A dvances in early diagnosis and treatment have substantially improved clinical outcomes for most patients with cancer in the past 2 decades. As the life expectancy of cancer survivors has increased, so has the likelihood of devel-oping other illnesses after cancer diagnosis. Cancer and cardiovascular (CV) disease share many risk factors, including smoking, lower socioeconomic status, and obesity. 1 This suggests that patients with cancer constitute a high-risk population for CV disease.
Furthermore, many cancer therapies, including chest irradiation as well as systemic therapies such as chemotherapy, are associated with incident CV disease during treatment and in survivorship. It has been hypothesized that the excess CV morbidity of cancer treatments is mediated by direct myocardial and/or vascular injury as well as indirectly through adverse effects on lifestyle behaviors. 2 However, recent population studies have yielded conflicting results on the long-term CV risk of cancer survivors. One study using data from the Surveillance, Epidemiology, and End Results (SEER) program demonstrated a higher risk for CV death among 234,256 cancer survivors compared with the U.S. general population, particularly in the first year after cancer diagnosis. 3 However, another SEER-based study showed that cardiac mortality in a cohort of 347,476 patients with breast cancer was not increased compared with age-matched control subjects during long-term follow-up. 4 Although these studies considered the effects of age, ethnicity, and cancer-related variables, they did not adjust for other risk modifiers, including prior CV disease, hypertension, diabetes mellitus, and dyslipidemia. Indeed, administrative databases confirm a high prevalence of CV disease in patients with cancer relative to cancer-free control subjects. 5 Prior work on CV risk in cancer survivors has also been limited in scope and has typically included study of only 1 cancer type and 1 CV outcome.
Access to multiple health data repositories in Alberta, Canada, permits the construction of wellcharacterized, population-based cohorts with extensive profiling of cancer and CV disease. Therefore, we designed this study to investigate the risk for incident CV events among people with histories of cancer compared with those without cancer after adjustment for baseline CV risk and other potential confounders.
We hypothesized that CV risk is increased in all cancer types and is not limited to incident heart failure.

METHODS
We reported this retrospective population-based cohort study according to the Strengthening the Reporting of Observational Studies in Epidemiology guidelines. 6 The Health Research Ethics Board of Alberta (HREBA.CC-16-0164) provided institutional approval and waived the requirement for participants to provide consent.  April 1994 when records were available). 15 As in our prior work, we used administrative data to identify age, biological sex, and rural residence location. 16 We included the Pampalon index of material deprivation created by Alberta Health Services. 17,18 It categorizes participants at the postal code level into 5 bins of socioeconomic inequalities in health care services and population health, with 5 representing the most deprived neighborhoods.
OUTCOMES. We assessed all-cause mortality, CV mortality, first MI during follow-up, 11,19 first stroke or transient ischemic attack, 11,20 new heart failure, 11,21 and first pulmonary embolism. 22 We defined mortality due to CV causes as in previous work 8

and included International Classification of Diseases-10th
Revision codes for ischemic heart disease, stroke, heart failure, valvular heart disease, and arrhythmia.
Latest potential follow-up for participants was       Table 4). Similarly, participants with more advanced cancer were at higher risk for CV outcomes (Supplemental Table 5).
However, even patients with very early stage disease (stages 0 and I) had higher risk for CV events relative to control subjects without cancer.
Supplemental Table 6 shows the baseline characteristics of participants with and without cancer after matching for age and sex. After matching, the excess risks for nonfatal CV events associated with cancer were similar to those in the primary analysis (