Evidence-based prediction and prevention of cardiovascular morbidity in adults treated for cancer

Background Cancer treatment-related morbidity relevantly compromises health status in cancer survivors, and efforts to optimise health-related outcomes in this population are vital to maximising healthy survivorship. A pre-treatment assessment – and possibly preventive management strategies – of cancer patients at increased risk for cardiovascular disease (CVD) seems a rational approach in this regard. Definitive evidence for such strategies is largely lacking, thereby impeding the formulation of firm recommendations. Results The current scoping review aims to summarise and grade the evidence regarding strategies for prediction and prevention of CVD in adults in relation to oncological treatments. We conducted a scoping literature search for different strategies for primary prevention, such as medical and lifestyle interventions, as well as the use of predictive risk scores. We identified studies with moderate to good strength and up to now limited evidence to recommend primary preventive strategies in unselected patients treated with potentially cardiotoxic oncologic therapies. Conclusion Efforts to minimize the CVD burden in cancer survivors are needed to accomplish healthy survivorship. This can be done by means of robust models predictive for CVD events or application of interventions during or after oncological treatments. Up to now there is insufficient evidence to implement preventive strategies in an unselected group of patients treated with potential cardiotoxic oncological treatments. We conclude that randomised controlled trials are needed that evaluate medical and lifestyle interventions in groups at increased risk for complications, in order to be able to influence chronic illness risks, such as cardiovascular complications, for cancer survivors.


Highlights
-Minimising treatment-related cardiovascular disease (CVD) in cancer survivors is crucial to maximising healthy survivorship. -Predictive models that aid in identifying persons at highest risk for such toxicities as well as primary prevention may be approached to decrease CVD after cancer treatments. -We performed a scoping literature review on prediction and primary prevention of CVD in cancer survivors and graded the studies with a Level of Evidence.
-We identified studies with moderate to good strength and up to now limited evidence to recommend primary preventive strategies in unselected patients treated with potentially cardiotoxic oncologic therapies. Some predictive models exist, but those lack methodological robustness to be implemented in clinical use.

Background
Health prospects after a cancer diagnosis have improved impressively over the past decades. There is a steadily improving life-expectancy for patients with cancer, with an age-standardised 10-year survival of 70% for patients diagnosed with breast cancer, 80% for Hodgkin's disease, and 90% for testicular cancer in Western countries [1,2]. Worldwide, the number of new cancer cases is rising and was estimated to be 18.1 million in 2018, leading to a large and growing group of cancer survivors [3,4]. Several factors contribute to these improved cure rates, including earlier detection of some cancers due to screening strategies, the organisation of care in centres of expertise and more effective treatment regimens [3]. Relatively recently a trend towards de-escalation of (curative) oncologic treatments has begun, e.g. in breast cancer [5], meaning that studies are designed to reduce the amount and intensity of treatment patients receive while maintaining equally good cancer outcomes. However, still, many patients with early-stage malignancies will receive treatment combinations of surgery, radio-, chemo-and targeted therapies. Different treatment modalities have different modes of action, levels of effectiveness, as well as the scope of acute and late side-effects. Consequently, a large and growing group of cancer survivors is at risk to develop treatment-related toxicity. Long-term complications include the development of cardiovascular disease (CVD), sub−/infertility, neurotoxicity and second malignancies [6]. Efforts to reduce disease burden, both in terms of risk for disease relapse and long-term treatment-related toxicity, in this population are vital to maximise healthy survivorship and improve quality of life. CVD can arise during oncological treatment, as well as up to decades after treatment completion. Significantly, not only direct treatment-induced toxicity can contribute to this development, but also (shared) risk-factors for both CVD and malignancy can increase the risk for this morbidity [7][8][9][10][11][12][13].
To achieve an optimal health status for cancer survivors, detection andwhen possibleprevention of these issues is of high priority. The development of CVD is a gradual process that progresses from subclinical changes to clinical morbidity, a process that may be accelerated by damage arising from oncological treatments (Fig. 1). Therefore, a baseline risk assessment followed by early interventions in case of therapies associated with a high risk for CVD morbidity as well as for persons at increased risk might be effective to impede or stop the progression towards overt CVD in cancer patients.
The current review aims to concisely summarise methods for baseline risk assessment, primary prevention with medication as well as lifestyle interventions in adult cancer patients who are receiving oncological treatments that may cause cardiovascular morbidity.

Methods
The literature was reviewed on the different topics covered in this article, such as baseline risk prediction tools, primary prevention with medication and lifestyle interventions. The PubMed and Cochrane databases were used for this purpose as well as ClinicalTrials.gov. Peerreviewed papers published in medical journals on clinical trials in human adults and literature reviews, written in the English language between January 1990 -September 2020 were included. Figure 2 provides a PRISMA flow chart of the selection process and the included manuscripts for the topics prediction tools and primary prevention with medication. The available data for lifestyle interventions is too scarce to assign a level of evidence to, and therefore, we cover this part as a narrative section. Each item is assigned a level of evidence (LOE) backing up its applicability, according to the system used by the Oxford Centre of Evidence-Based Medicine (Table 1). Level 1a evidence is viewed as definite, and level 5 as weak. To maintain a concise summary of the literature, we decided to only refer to papers within the highest level of evidence for each reviewed topic, summarised in Table 2.

Prediction of future CVD in cancer patients
An accurate and practically applicable model that predicts the baseline risk for development of future CVD is needed in patients with a newly diagnosed malignant disease scheduled for anti-cancer treatment (Fig. 1). Timely and accurate risk prediction is crucial to adapt treatment plans and initiate (preventive) CVD treatment to balance optimal cancer therapy whilst minimising the risks of development of CVD.
Up to date, it is not known whether predictive models for CVD that are used in the general population, e.g. the Framingham Risk Score, are representative of the future CVD risk in cancer survivors treated with therapies that are potentially harmful to the cardiovascular system. Theoretically, one could argue that such models are not entirely representative, as there are additional pathogenic factors that contribute to the development of CVD in the specific population (e.g. chemotherapy toxicity causing direct cardiomyocyte damage and accelerated atherosclerosis) [21]. A cross-sectional study among testicular cancer survivors revealed no differences in the risk score on the Framingham Risk Score when compared to the score of age-matched persons in the general population, 1 year after chemotherapy treatment [22]. However, long-term follow-up studies have confirmed the increased risk of late CVD in this population [23][24][25].
The recently published guideline from the European Society of Medical Oncology (ESMO) recommends a thorough baseline screening for CVD risk factors before an anti-cancer treatment is initiated, but does not include a discussion about the use of predictive models [26]. The guideline of the American Society of Clinical Oncology (ASCO) on prevention and monitoring of cardiotoxicity in adult cancer survivors provides an approach on how patients can be divided into different risk groups for future CVD, based on oncological treatment, age and presence of risk factors for CVD. This risk staging is though mainly evidence-based with intermediatemoderate quality evidence [27].
The only area in which such predictive models, based on clinical features and treatment factors, for future CVD have been investigated are breast cancer patients. To the best of our knowledge, seven studies have been published on this topic (Fig. 2, Table 2) [8,[15][16][17][18][19][20]. Table 2 summarises the respective study characteristics and the populations included in the analyses. The study by Ezaz et al. showed a clinically useful model where patients with HER2-positive breast cancer were divided into low, medium and high-risk groups for heart failure/ cardiomyopathy, with relatively good discriminative ability [15]. This study used part of their cohort as an internal validation of the model, and later, Rushton et al. used this model in their data set of a very selected group of breast cancer patients referred to a tertiary care centre after they had developed cardiac evens in relation to cancer treatment [8]. The most important conclusion from the latter analysis was that the application of this cardiac risk score helped identify those at low risk of permanent cardiac dysfunction. However, it did not perform as well in identifying high-risk patients.
A similar approach was used in the analysis by Fogarassy et al., where a cohort of over 8000 breast cancer patients treated with epirubicin was analysed [8]. Here, risk factors for late heart failure were identified in a training set of 70% of this cohort and internally validated in the other 30%. The paper by Hermann et al. gives an expert-    [19]. This model had the good discriminative ability and was also consistent in the validation set, but no cancer treatment-related factors were included in this model. In addition, one study investigated a predictive risk score for development of cardiac toxicity from anthracyclines in patients with metastatic breast cancer [20]. In summary, three baseline predictive models for heart failure in breast cancer patients have been composed, and the one developed by Ezaz et al. in HER2-positive breast cancer patients [15] was validated in highly selected external cohort with moderate performance [16]. Our literature search did not identify comparable studies that have been performed in other cancer patient populations.

Prevention of future CVD during oncological treatments
Primary prevention refers to delaying or preventing the onset of CVD, whereas secondary prevention aims to reduce the number of new or severe cases of CVD. Table 3 summarises the currently published studies in this area, where we have only included papers with the highest level of evidence (according to Table 1) within a specific class of medication. The role of intervention with angiotensin-converting enzyme (ACEiI) inhibitors, angiotensin-receptor blockers (ARB), beta-blockers, statins, aspirin/anticoagulants and other drugs will be discussed, respectively. Most intervention studies have been performed in patients treated with anthracyclines, mainly breast cancer patients. The majority of the studies investigated preventive measures during and up to shortly after completion of a potentially toxic therapy in unselected patients.

Preventive medication: (ACEI/ARB/beta-blockers)
Recently, a meta-analysis reported on the available evidence about primary prevention with reninangiotensin-aldosterone inhibitors such as ACEI and ARB, and beta-blockers, in breast cancer patients treated with anthracyclines with and without trastuzumab [28]. A total of N = 2301 patients was included, and the cardioprotective effect of the different drugs was estimated. Median changes in LVEF at three time-points (immediate after completion of chemotherapy, at six-and 12months post-treatment) and the incidence of heart failure (CHF), was statistically significantly lower in the intervention groups compared to the control. However, the magnitude of the effect was modest and the confidence intervals (CI) for the last two time-points rather broad (standardized mean difference in LVEF -chemotherapy completion − 2.36 [95% CI: − 3.23 to − 1.49]; 6 months − 6.54 [95% CI: − 10.74 to − 2.34]; 12 months − 5.37 [95% CI: − 9.31 to − 1.43]). In the three trials included in the meta-analysis that included patients with HER2-positive breast cancer treated with trastuzumab, no effect of the intervention was noted in terms of LVEF or rate of CHF. There was neither a persistent effect seen of beta-blockers in terms of one-year LVEFoutcomes in the entire group. The authors conclude that prophylactic treatment with ARB/ACEI and betablockers of unselected breast cancer patients scheduled for anthracycline +/− trastuzumab should not be routinely recommended. This finding was also noted in a large prospective study in unselected patients with HER2-positive breast cancer that were randomized to candesartan or placebo during and after adjuvant systemic therapy [29]. Contrarily, selecting high-risk patients that are candidates for interventions aimed to decrease rates of cardiotoxicity might be a rationale strategy, as was demonstrated in large Italian study [42]. Here, patients with rises in troponin shortly after highdose chemotherapy for breast cancer hade lower rates of late cardiotoxicity (decreases in LVEF of > 10%) when they received treatment with enalapril. In line with these findings, we believe future studies should aim to identify high-risk patients and specifically target studies in such a population.
Of interest, three prospective randomized trials in early breast cancer patients have directly compared the role of ACEi/ARB versus beta-blockers to placebo [36- (Table 3).
A meta-analysis that focused on beta-blockade included six trials (in total N = 633 patients) that investigated the role of carvedilol in primary prevention of anthracycline-induced cardiac toxicity. This study revealed a protective effect as apparent from reduced rates of pathological LVEF (defined as a LVEF < 50%) favouring the carvedilol group (3.2% vs 5.8%; odds ratio 0.42 [95% CI: 0.18 to 0.99]). Furthermore, there were significantly smaller reductions in LVEF in carvedilol-treated patients than in placebo-treated patients (mean differences: 2.41% [95% CI: 0.01 to 4.81; p = 0.05) [43]. The NNT of 38 implies that 38 patients have to be treated with carvedilol to prevent a LVEF-decrease to < 50% in one patient.
Small studies have been performed in patients with other non-breast cancers ( Table 2); in a cohort of N = 40 non-Hodgkin lymphoma patients, treatment with the ARB valsartan resulted in ameliorated changes in subclinical parameters for cardiac toxicity (such as echocardiographic parameters and biomarkers), compared to controls who did not receive valsartan [31].
In the OVERCOME trial (NCT01110824), the combination of carvedilol plus enalapril resulted in preserved LVEF 6 months after chemotherapy treatment for hematologic malignancies, compared to a control group without this medication combination (N = 90) [33].

Preventive medication: statins
Statins can lower the level of low-density lipoprotein (LDL) cholesterol in the blood. High levels of LDL cholesterol can lead to atherosclerosis and CVD. A small randomised controlled trial on atorvastatin 40 mg/daily for 6 months during anthracycline therapy for various malignancies demonstrated that the intervention group had a preserved LVEF, which was one of the secondary endpoints of the study, compared to a decline of almost 8% in LVEF in the control group [39]. There was no statistically significant difference in the primary endpoint of the trial, namely a LVEF < 50% after anthracycline treatment.
Highest LOE for the use of statins: 1b (no difference in LVEF < 50%).

Preventive medication: aspirin/anticoagulants
Acetylsalicylic acid (ASA, aspirin) inhibit platelet aggregation. Low dose ASA prevent myocardial infarction and stroke in high risk individuals. To our knowledge, no clinical studies have been conducted or are ongoing regarding the role of aspirin or anticoagulants in the primary prevention of cancer therapy-related cardiotoxicity. However, the preventive effect on the development of venous thrombosis during oncological treatments is established for high-risk groups [44,45].
Some preclinical studies have reported on the role of prostacyclins in the development of anthracyclineinduced cardiomyocyte damage, and therefore theoretically, cyclooxygenase 1 and 2 inhibitors could be effective in prevention cardiotoxicity [46,47]. However, there are no available clinical studies with this class of drugs.
Highest LOE for aspirin/anticoagulants: not available.

Preventive medication: dexrazoxane and sildenafil
The use of the iron-chelator dexrazoxane is mainly established in pediatric cancer patients as a drug for primary prevention of cardiotoxicity related to anthracycline therapy. In 2019, a systematic literature review on the use of dexrazoxane in adults was published [40]. This review identified seven trials in breast cancer patients, and the meta-analysis of these trials with in total 2177 breast cancer patients revealed a statistically significant and clinically relevant protective effect on the rate of CHF and cardiac events, without signs of impaired oncological treatment efficacy. In a small prospective study in women treated with doxorubicin (NCT01375699), treatment with sildenafil (100 mg daily, 300 mg at treatment days) did not prevent declines in LVEF or rises in high sensitivity cardiac troponin I [41]. This drug is used very rarely in clinical practice, and the available evidence does not support use as of now.

Lifestyle interventions before and during treatment initiation
Recently, a survey among various medical professionals involved in the care of cancer survivors reported a high rate of health promotion mainly amongst primary care physicians, but a much lower rate among oncologists and other specialists [48]. Many reported lack of time as well as concerns with compliance among cancer survivors as reasons for abstaining health-promoting recommendations.
Several retrospective cohort studies have shown associations between lifestyle habits (such as dietary intake, physical activity, alcohol consumption and smoking status) and the risk for long-term cardiovascular morbidity and mortality after cancer therapy, e.g. in a cohort of Hodgkin lymphoma, hematopoietic cell transplantation and colorectal cancer survivors [49][50][51]. A retrospective analysis from the Women's Health Initiative cohort showed that higher levels of self-reported physical activity were associated with a significant graded reduction in CV events in breast cancer survivors [52]. Various sources can introduce bias and Influence these findings, and a causal relationship has, therefore, not been established yet. To our knowledge, no prospective trials have been performed that investigate lifestyle interventions, such as physical exercise, smoking cessation and diet changes, concerning objective and clinically relevant cardiovascular endpoints.
The ASCO guidelines on breast cancer survivorship [53], as well as the ASCO and ESMO guidelines on prevention of cardiac dysfunction in cancer survivors [26,27], underscore the importance of screening for and educating (breast) cancer survivors about abovementioned lifestyle modifications, that may reduce the risk for or severity of cardiotoxicity or cardiovascular diseases in this population. Associations, and to some extent also causality, between such modifications and health outcomes in the general population, is evident from several extensive studies.

Smoking cessation
A meta-analysis showed that smoking cessation rates after a cancer diagnosis are rather low, even after intervention programmes [54]. No prospective data are supporting the hypothesis that smoking cessation might improve health prospects after a cancer diagnosis, including a lower rate of CVD as well as improved oncologic outcomes. However, indirect evidence originates, for example, from a large meta-analysis that found evidence for an increased risk for radiotherapy-induced cardiac toxicity in smokers [13]. The current lack of evidence should, however, not discourage professionals from stimulating smoking cessation.
Highest LOE for smoking cessation and CVD endpoints: n.a.

Diet changes and weight loss
Weight gain after a breast cancer diagnosis is common. Cespedes et al. reported a weight gain of > 4.5 kg/10 lbs. among 25% of a total of 2000 breast cancer patients [55]. However, this did not translate into increased long-term risk for CVD. A large prospective intervention study in early breast cancer patients on dietary changes, with increases in fruit and dietary fibre consumption and reductions in dietary fat intake, revealed no difference in breast cancer-related outcomes nor overall mortality [56]. A cohort study in breast cancer survivors showed that women who had a prudent (balanced) diet, consisting of high intakes of fruits, vegetables, whole grains, and poultry, had improved survival as well as a lower non-breast cancer-related survival, which may thus relate to a lower rate of death due to CVD [57]. On the contrary, a systematic review on the impact of dietary patterns in breast cancer and colorectal cancer survivors was not able to find any positive associations between dietary patterns and risk for disease recurrence and overall mortality [58]. However, the review did not specify CVD-related endpoints.
Highest LOE for diet interventions/weight loss and CVD endpoints: n.a. (circumstantial: reduction non-BC mortality in BC survivors).

Physical exercise
Most intervention studies on physical exercise interventions have been performed in early breast cancer patients, as well as in smaller populations of patients with colorectal and prostate cancer. Physical exercise interventions improve physical performance, cardiorespiratory fitness and fatigue during and shortly after breast cancer chemotherapy treatment completion [59,60]. A systematic review and random-effects meta-analysis in almost 50,000 breast and colorectal cancer patients showed that patients who increased their physical activity by any level from pre-to post-diagnosis showed a total reduction in mortality risk (RR = 0.61; 95% CI = 0.46-0.80) compared with those who did not change their physical activity level or were physically inactive/insufficiently active before diagnosis [35]. As of now, effects of physical exercise interventions on early or late cardiovascular endpoints in cancer survivors have not been published, although preclinical evidence demonstrates that exercise and caloric restriction can independently reduce anthracycline-related injury to the heart [61].
In an Australian intervention study, twenty-eight patients with early-stage breast cancer undergoing anthracycline chemotherapy were able to choose between exercise training or standard care [62]. In the intervention group, no differences were seen in LVEF, but decreases in V·O2 peak (representing peak aerobic fitness) during chemotherapy were attenuated, and functional status was preserved, compared to control. Some indirect evidence that exercise might preserve cardiovascular status in cancer patients comes from a prospective study in colon cancer survivors reporting that physical exercise caused lower insulin levels and insulin resistance, compared to standard of care [63]. A prospective study in breast cancer patients revealed that an exercise-based intervention programme protected from treatment-induced tachycardia and hypotension, but was only a temporary effect during the intervention [64].
Even though exercise or the initiation of such, after a cancer diagnosis may decrease risks for late cardiovascular disease, there is still lack of knowledge in among health professionals, that may well be due to some considerable heterogeneity and risk for bias in published studies, but also the various approaches that can be used and confirm the safety of exercise interventions [65]. In a recently updated evidence-based guideline, it is recommended that women with breast cancer follow the recommendations of 150 min of aerobic activity per week and strength exercise 2-3 times a week [66].
Highest LOE for physical exercise and CVD endpoints: 5 (rationale for ongoing RCT's).

Strategies for prevention of CVD after cancer treatment completion
To our knowledge, no intervention studies have been performed that studied the role of strategies to prevent development of CVD after completion of oncological treatments. The guidelines from ASCO and ESMO on prevention of cardiac dysfunction related to oncological treatments recommend that clinicians complete a careful history and physical examination in survivors of cancer previously treated with potentially cardiotoxic therapies, and additional cardiac investigations in case of clinical signs and symptoms of heart failure [26,27]. These guidelines also advocate regular evaluation and management of traditional cardiovascular risk factors and obesity in patients previously treated with cardiotoxic cancer therapies. A so-called 'heart-healthy' lifestyle, including advice on the importance of diet and exercise, should be discussed as part of long-term follow-up care. This is a The need for taking cancer treatments into account in patients with CVD, and the increased risk for CVD in cancer survivors, was also recently addressed in a study and accompanying editorial in Lancet [67,68]. It is, however unclear whether target values, e.g. lipids and blood pressure should be equal or perhaps more stringent in cancer survivors than the general population, especially regarding primary prevention.
Highest LOE for strategies for CVD prevention in cancer survivors: 5.
Ongoing intervention trials aiming to decrease CVD in relation to cancer treatments At clinicaltrials.gov, we identified 17 actively recruiting prospective trials in adult cancer patients; almost all concern patients treated with anthracyclines and (in part) anti-HER2 agents as well (Table 4). Five studies are offering a physical exercise intervention, with training supervised by experts, whereas nine concern an intervention with medication from different classes. Interestingly, two studies are investigating the role of ivabradine. This is a class of drugs that, to our knowledge, has not yet been investigated in a cancer patient population with the aim to decrease treatment-related cardiac toxicity. It belongs to a class of medication called hyperpolarization-activated cyclic nucleotide-gated channel blockers, and the mechanism of action is that it slows the heart rate and thereby increases diastolic perfusion time and myocardial perfusion. Almost all studies (15 out of 17) have cardiac function assessed by ultrasound or cardiac MRI, either LVEF or global strain, as primary endpoint.

Conclusion
Our literature review identified studies with moderate to good strength and up to now limited evidence to recommend primary preventive strategies in unselected patients treated with potentially cardiotoxic oncologic therapies (Table 5). Some predictive models exist, but all lack methodological robustness to be implemented in clinical use.
In current clinical practice, based on the data reviewed here, we recommend refraining from offering patients with a cancer diagnosis primary prevention with medication concomitantly with a potential cardiotoxic treatment. Meta-analyses support the role of beta-blockers or ACEI/ARB, although the magnitude of effect is very modest in unselected patients, and in our opinion too little to justify the potential side-effects, interactions with oncological therapies and costs.
At the same time, based mainly on data from retrospective analyses, we do emphasize that all clinicians should advocate the beneficial role of life-style optimization by means of a heart-healthy diet and meeting recommendations for level of physical exercise in all patients scheduled for such treatments. Although there is clearly less evidence to support such interventions in terms of cardioprotective capacities, is there ample evidence on other clinically relevant outcomes to pursue healthy life-style behaviour in our patients. Currenlty ongoing trials will hopefully reveal whether such interventions will be effective in decreasing cardiovascular complications of oncological treatments.
In the future, a patient at increased risk for treatmentrelated CVD (both acute and late onset) should ideally be identified through an objective, practically usable and externally validated baseline risk estimation score (Fig. 1). Such a risk score should be easily accessible and based on patient-and treatment-related factors, and in addition, biomarkers for subclinical cardiovascular damage (at baseline or during treatment) could be included. Patients that are, based on such a risk assessment, deemed at highest risk would be ideal candidates to include in an intervention trial that aims to decrease cancer treatment-related cardiovascular toxicity in order to improve cancer survivorship by optimizing outcomes whilst minimizing the risk for unwanted side-effects.