Pharmaceutically treated anxiety but not depression prior to cancer diagnosis predicts the onset of cardiovascular disease among breast cancer survivors

This population-based observational study showed that pharmaceutical therapy for anxiety but not depression in the year prior to breast cancer diagnosis was associated with an increased hazard of incident CVD in 1-year breast cancer survivors. This relation remained significant after adjustment for depression, traditional cardiovascular risk factors, and clinical characteristics, and was restricted to younger breast cancer patients. Associations did not differ by type of cancer treatment or the presence of traditional cardiovascular risk factors in the year prior to breast cancer diagnosis.

The association between pharmaceutical treatment for anxiety present in the year prior to cancer diagnosis and the increased risk for CVD found in the current study is consistent with studies in healthy individuals, showing that anxiety predicts the onset of CVD [11, 12]. We focused on pharmaceutically treated anxiety and depression prior to breast cancer to examine whether anxiety and depression prior to the major life event of a breast cancer diagnosis are related to an increased risk of CVD. However, we know that being diagnosed with cancer and undergoing treatment is known to have a major impact on one’s life, as many cancer survivors experience feelings of anxiety or depression [23, 24]; hence, an additional proportion of breast cancer survivors may start to use pharmaceutical treatment for anxiety. We found a higher increased risk for anxiety in our study (48%) than a meta-analysis among healthy individuals did (26%) [12]. The increased risk of anxiety in our study remained statistically significant despite adjustment for hypertension, hypercholesterolemia, diabetes mellitus, tumor stage, and received chemo-, radio-, or hormone therapy. Hence, this association seems not to be driven by the presence of traditional cardiovascular risk factors, cancer stage, or cancer treatment. It is interesting that anxiety is only associated with an increased CVD risk in younger survivors (≤65 years). This is in line with previous research where psychological factors especially seem to play a role among younger individuals, whereas among older individuals aging—and likely physiological factors related to the aging and disease process—is suggested to drive the relationship with poor health outcomes [25]. Several behavioral and pathophysiological mechanisms have been suggested to underlie the associations between anxiety and increased CVD risk in non-cancer populations. Anxiety has, for example, been related to unhealthy lifestyle behaviors such as smoking and limited exercise [26]. Furthermore, the autonomous nervous system and hypothalamic–pituitary–adrenal axis known for their role in the pathogenesis of CVD [27, 28] have been suggested to be involved, as anxious individuals have a lower heart rate variability [29] and higher cortisol levels [30]. As medication use was used as a proxy for both anxiety and CVD, we cannot rule out that the association may partly be explained by a pharmacokinetic interaction between both drug types, although little is known about this association and the usage of anxiolytics is common among CVD populations [31, 32].

The lack of an association between depression and CVD risk was unexpected, as previous studies have demonstrated that depression is a risk factor for both incident CVD and CVD progression [11, 13‐15]. This dissimilarity could be because often other studies have examined either depression or anxiety, and hence they are unable to disentangle the role of each of depression and anxiety separately and together. Alternatively, the result of the low prevalence of pharmaceutically treated depressed individuals in our study (8%) could play a role in us not confirming previous results. Using drug dispense information as an indicator of depression, we could be underestimating the true prevalence of depression, as a depression and its milder form depressive symptoms are often not treated with medication. Alternatively, depression may comprise different subtypes. It has been suggested that some manifestations of depression may partly reflect cardiac disease severity [33], and hence it may have distorted the strength of the previously found associations between depression and prognosis of CVD. Nevertheless, the fact that anxiety, but not depression, was significantly associated with increased CVD risk is in line with previous findings studying poor outcomes among CVD populations [34‐36].

Surprisingly, there was no effect of chemo- or radiotherapy on CVD risk, despite the known cardiotoxicity of these treatments [3, 4]. This finding may be attributed to the lack of detailed information about type and dose of systemic therapy given. As not all chemotherapeutic agents are equally cardiotoxic, cardiotoxicity of chemo- and radiotherapy is dose dependent [3]. It is thus possible that grouping all chemotherapeutic agents irrespective of type and dosage will result in an underestimation of their true effect. Moreover, we found a protective effect of hormonal cancer treatment on CVD risk. A previous study also found that hormone treatment in breast cancer survivors can lower the risk of CVD, yet only several years after diagnosis, as it increases the risk of CVD in the first years after cancer diagnosis [37]. Previous studies have suggested that estrogen hormone treatment has favorable effects on lipoproteins, coronary arteriosclerosis, endothelial function, and arterial thrombosis [37]. Exploratory post hoc analyses, indeed, showed a protective effect of hormone treatment on incident CVD 6–10 years after cancer diagnosis in our sample, whereas there was no significant relation in the first 5 years after diagnosis or after ≥11 years.

A general limitation inherent to the observational study design is the lack of information on residual confounders. Furthermore, we used drug dispense information as a proxy for anxiety, depression, and CVD. We did not have additional information on medical diagnosis or patient-reported information on anxiety or depression. Nevertheless, algorithms based on pharmacy drugs are known to be more specific, yet less sensitive than medical diagnoses [38]. Also, we used a rather tight algorithm to define CVD, as this was based on a minimum of two C01 drug dispenses within 6 months, possibly leading to an underestimation of the incidence of CVD. It is also possible that we missed CVD patients who use other drugs, such as ACE inhibitors or beta-blockers but no C01 drug, although we expect this number to be small. In addition, as we were interested in incident CVD, breast cancer survivors with CVD in the year prior to their cancer diagnosis were excluded. Hence, we are looking at a subpopulation of 1-year breast cancer survivors.

Strengths of our study include the large population-based sample of breast cancer survivors and the usage of high-quality databases of the Netherlands Cancer Registry and PHARMO enabling a 13-year follow-up period. Additionally, this is to our knowledge the first study that examined the association between pharmaceutically treated anxiety and depression with incident CVD among breast cancer survivors. Furthermore, we estimated CVD risk 1 year after breast cancer diagnosis, striking a compromise between not starting too late and missing incident CVD due to ongoing cancer treatment or increased clinical checkups.

In conclusion, 1-year breast cancer survivors with pharmaceutically treated anxiety in the year prior to their cancer diagnosis had a 48% increased risk of incident CVD, after adjustment for depression, traditional cardiovascular risk factors, tumor stage, and cancer treatment. This increased risk seems to be limited to those breast cancer survivors who were 65 years or younger at cancer diagnosis. Depression was not related to an increased risk of incident CVD. Future studies unraveling these associations are warranted in order to provide the best optimal care for women treated for breast cancer.