Introduction

The concept of anti-angiogenic therapy was born more than 30 years ago, when Judah Folkman hypothesized that angiogenesis is crucial to tumor growth and to the metastatic process [1]. More recently many studies found that vascular density is a prognostic marker [2]. However, the development and marketing of anti-angiogenic drugs is a recent process. Although the publication of bevacizumab activity in renal cell might have appeared first, the drug was first marketed for treatment of metastatic colorectal cancer [3, 4]. Bevacizumab became widely prescribed for both cancers. Apart from this monoclonal antibody, specifically targeted against isoforms of vascular endothelial growth factor (VEGF), angiogenesis signaling pathways can be blocked by small molecules, namely inhibitors of VEGF receptors with tyrosine-kinase activity (sunitinib or Sutent®, sorafenib or Nexavar®).

These treatments could be expected to induce severe toxicities since they affect vascular development, which is ubiquitous, and are involved in general homoeostasis. However, the first pivotal studies dealing with anti-angiogenic therapies did not find major toxicities, even when these drugs were associated with chemotherapy. Nevertheless, clinical trials published after their marketing, give us more details and information about these cardiovascular side effects, such as hypertension, left ventricular systolic dysfunction (LVSD) and heart failure, conduction abnormalities and QT prolongation (with the risk of torsades de pointe and sudden death) [5]. The major differences between pivotal phase III studies and the post-marketing occurrence of cardiovascular side effects is the nature of the populations of patients included in phase III studies (highly selected patients without co-morbidities), and thereafter of patients included in prospective cohorts (less strictly selected patients), and finally of the general population of patients. Moreover, several meta-analyses were performed in these side effects.

We performed a PubMed search on October 19, 2011, using the following keywords simultaneously: cancer, antiangiogenic therapy, cardiovascular side effects. We focused on major original articles and review articles on this topic.

Hypertension and proteinuria

Hypertension and proteinuria are the most frequent and specific side effects of anti-angiogenic targeted therapies. The prevalence of hypertension appears rather different according to the type of targeted therapy and also to the type of studies. Elevated blood pressure was initially observed with bevacizumab in 20–30% of patients also receiving chemotherapy [6]. It predominated on systolic blood pressure and usually had a rapid onset after bevacizumab initiation. Hypertension crisis was observed in about 1% of patients. Anti-hypertensive treatment was prescribed de novo in about 11–16% of patients. According to a recent meta-analysis assessing 12,949 patients from 19 randomized controlled trials, the overall incidence of elevated blood pressure under bevacizumab reached 8% (6–10%) [7]. Bevacizumab was associated with a significantly increased risk raised blood pressure (RR 5.38; 95% Confidence Interval or CI 3.63–7.97). These authors found a non-significant trend towards a dose-effect relationship, with a risk ratio (RR) of 7.17 (95% CI 3.91–13.13) for a weekly dose of 5 mg/kg compared with a RR of 4.11 (95% CI 2.49–6.78) for a weekly dose of 2.5 mg/kg. The highest RRs were observed under bevacizumab (5 mg/kg/week) for renal cancer (RR 13.77; 95% CI 2.28–83.15) and for breast cancer (RR 18.8; 95% CI 1.23–292.3) [7]. Thus, the increased risk of elevated blood pressure under bevacizumab might be dose-dependent and might vary according to the location of the primitive tumor.

Sunitinib, a tyrosine-kinase VEGF receptor inhibitor, prescribed in 75 patients with imatinib-resistant metastatic GISTs enrolled in a phase I/II trial, induced significant increases in mean systolic and diastolic blood pressure; 47% of patients (35/75) developed hypertension [8]. Sorafenib increased blood pressure by 20 or more mmHg in 16% of patients [9]. For these tyrosine kinase inhibitors, the incidence of hypertension ranged from 19% to 24% in meta-analyses [10, 11].

For anti-angiogenic drugs overall, the incidence of hypertension varies according to the characteristics of patients (age, previous history of hypertension, co-morbidities) and according to the characteristics of the drug prescribed (nature of the drug, dose, therapeutic schedule) [12]. The mechanism of hypertension has not been fully elucidated. Bevacizumab-induced hypertension might be related to a decrease in VEGF-induced NO (nitric oxide) synthesis and/or to the rarefaction of capillary bed, as shown by capillaroscopy [13]. Sorafenib-induced hypertension might not be related to humoral factors [9].

The incidence of mild and asymptomatic proteinuria ranges from 21% up to 63% [14], but major proteinuria has been reported in up to 6.5% of patients with renal cell carcinoma, sometimes in a context of thrombotic microangiopathy [15, 16]. The production of VEGF by podocytes is essential for optimum function of the adjacent glomerular endothelium. By reducing glomerular VEGF, bevacizumab induces thrombotic microangiopathy. The discontinuation of the anti-VEGF drug usually induces a significant reduction in the amount of proteinuria. However, the persistence of proteinuria after drug cessation is common. Although angiotensin-converting enzyme inhibitors and/or angiotensin receptor blockers seem to be preferred, no specific recommendation for an anti-proteinuric agent can be made in this context since no controlled trial addressed this issue. Periodic monitoring of proteinuria should be carried out in patients receiving an anti-VEGF drug. Such patients showing proteinuria should be referred to a nephrologist. We did not find any recommendation for a level of proteinuria over which anti-angiogenic treatment should be discontinued. It seems logical to interrupt therapy when patients present with a nephrotic syndrome.

Arterial and venous thrombosis

When bevacizumab is associated with chemotherapy, a meta-analysis based on 1,745 patients from five randomized controlled trials showed that the risk of arterial thrombo-embolic events was increased (RR 2; 95% CI 1.05–3.75) [17]. Bevacizumab plus chemotherapy induced 5.5 events per 100 patient years, compared with 3.1 events with chemotherapy alone (RR 1.8; 95% CI 0.94–3.33), a non-significant trend towards an increased risk. Predisposing factors are old age (>65 years) and a history of thrombo-embolic event. A more recent meta-analysis assessed arterial thrombo-embolic events (myocardial infarction, stroke) under bevacizumab among 12,617 patients from 20 randomized controlled trials [18]. Incidence of all-grade arterial thrombo-embolic events was 3.3% (2.0–5.6%), whereas incidence of high-grade events was 2% (1.7–2.5%). Patients receiving bevacizumab had a significantly increased risk of arterial thrombo-embolic events (RR 1.44; 95% CI 1.08–1.91). These authors did not find a dose-effect relationship, since risks were similar for 5 and 2.5 mg/kg/week. Renal cell cancers and colorectal cancers had significantly increased risks (3.72, 95% CI 1.15–12.04 and 1.89, 1.28–2.80 respectively). The incidence of high-grade myocardial ischemia was significantly increased under bevacizumab compared with controls (RR 2.14; 1.12–4.08), whereas the risk of stroke was not modified [18].

During sorafenib therapy, the incidence of arterial thrombo-embolic events amounts to 3%, compared with 1% under placebo. During sunitinib therapy, the incidence is similar to that of placebo.

Three meta-analyses examined the issue of venous thrombo-embolism, with conflicting results. In 2007, Scappaticci et al. did not find an increased incidence of venous thrombo-embolism under bevacizumab [17]. This finding was confirmed by another meta-analysis in 2011, based on individual patient data and including a much higher number of patients (6055 from 10 randomized controlled trials) [19]. The unadjusted incidence of venous thrombo-embolic events in the overall population was 10.9% with bevacizumab and 9.8% with controls (odds ratio 1.14; 95% CI 0.96–1.35). The numbers of events per 100 patient-years was 18.5 for bevacizumab and 20.3 for controls [19]. Identified risk factors for venous thrombo-embolism were tumour type, old age, poor performance status, and baseline oral anticoagulant use. Conversely, Nalluri et al. found in 2008, in a third meta-analysis based on 15 studies including 7,956 patients—and not on individual patients—that bevacizumab increased the risk of venous thrombo-embolism [20]. The summary incidences of all-grade and high-grade venous thrombo-embolism were 11.9% and 6.3%. Patients treated with bavacizumab had a significantly increased risk of venous thrombo-embolism (RR 1.33; 95% CI 1.13–1.56) [20]. There was no dose-effect relationship, since risks were similarly increased with 5 and 2.5 mg/kg/week of bevacizumab. This third meta-analysis was not devoid of criticism: it was not based on individual patient data, it did not discriminate between arterial and venous thrombo-embolic events, and finally, the risk ratios were not adjusted on varying durations of observation between studies.

Risk of bleeding

Severe bleeding is rare during treatment with anti-angiogenic drugs. However, lethal cases have been described. In 2010, a meta-analysis (12,617 patients from 20 randomized controlled trials) showed that bevacizumab, when prescribed for cancer, was associated with a statistically significant increase in serious bleeding. The incidence of all-grade hemorrhage was 30.4% (95% CI 21.5–40.9%), with 3.5% being grade 3–5. Bevacizumab significantly increased the risk of all-grade bleeding compared with controls (RR 2.48; 95% CI 1.93–3.18), and also increased the risk of high-grade bleeding (RR 1.91; 95% CI 1.36–2.68). The risk of fatal hemorrhage was low (0.8%) and significantly increased in lung cancer only (RR 5.02; 95% CI 1.52–16.66) [21]. A trend for a dose-effect relationship was found since the risk of bleeding was higher with 5 mg/kg weekly (3.02; 95% CI 2.42–3.78) than with 2.5 mg/kg weekly (2.01; 1.43–2.83). In a recent meta-analysis, published in 2011, including 14,277 patients (7,136 received bevacizumab) from 22 studies, Chinese authors confirmed a significantly increased risk of high-grade bleeding (grade 3 or above) with bevacizumab in cancer patients (2.8%; RR 1.6; 95% CI 1.19–2.15) and of all-grade bleeding (RR 2.65; 95% CI 2.08–3.38) [22]. Again, these authors found a trend for a dose-effect relationship: the risk of bleeding was higher for 5 mg/kg of bevacizumab weekly (RR 3.02; 95% CI 1.85–4.95) than for 2.5 mg/kg where it was not statistically different from controls (RR 1.27; 95% CI 0.95–1.71). The highest risks of bleeding were observed for doses of 5 mg/kg weekly in non-small cell lung cancer (RR 3.41; 95% CI 1.68–6.91), in renal cell cancer (RR 6.37; 95% CI 1.43–28.33) and in colorectal cancer (RR 9.11; 95% CI 1.70–48.79) [22].

Cardiotoxic effects

The first cardiac toxicity of anti-VEGF treatments was the Takotsubo syndrome of cardiomyopathy (ballooning of left ventricle) observed in two men treated with bevacizumab for metastatic cancer [23]. These patients had symptoms of acute coronary syndrome without coronary obstruction. These symptoms could regress rapidly after the acute phase, but life-threatening complications could occur. The precise mechanism of this syndrome is unknown. Recently, Choueiri et al. assessed retrospectively 3,784 patients with breast cancer receiving bevacizumab plus chemotherapy—their data came from the literature of randomized controlled trials [24]. Overall incidence of high-grade chronic heart failure was 1.6% among bevacizumab-treated patients and 0.4% among placebo-treated patients, a statistically significant difference. The risk ratio of chronic heart failure among bevacizumab-treated patients compared with controls was 4.74 (95% CI 1.66–11.18). There was no dose-effect relationship, the incidence of chronic heart failure being similar with 5 and 2.5 mg/kg weekly of bevacizumab.

A decrease in left ventricular ejection fraction (LVEF) was found in 2% of 750 patients with renal-cell carcinoma treated with sunitinib or interferon alfa [25]. Beside hypertension, which was present in 8% of patients receiving sunitinib compared with 1% of those receiving interferon, a grade 3 decline in left ventricular ejection fraction was observed in 2% of patients receiving sunitinib vs. 1% of those under interferon. In another study, fatal or non fatal cardiovascular events were observed in 11% (8/75) of patients treated with sunitinib for imatinib-resistant metastatic GISTs and symptomatic (class III or IV) congestive heart failure was observed in 8% of patients, 28% (10/36) of patients had a decrease in LVEF of at least 10%, and 19% had a decrease of LVEF of 15% or more [8]. Congestive heart failure and left ventricular dysfunction usually resolved after sunitinib cessation. Cardiotoxicity in this study was more marked than in other pivotal studies with sunitinib. Myocardial ischemia or infarction was observed in only 1% of patients treated with second-line sunitinib [26], and in 3% of patients with renal-cell cancer treated with sorafenib [27]. A decline in left ventricular ejection fraction was observed in 8 patients under sunitinib (4.7%); for 5 of them, the decline from baseline reached 20% or more [26]. A prospective series of 74 patients with renal-cell carcinoma treated either by sunitinib or by sorafenib, found that 34% of patients experienced a cardiac event (n = 25) with half of them being symptomatic (7 patients had a serious cardiac illness), and 40.5% having ECG changes (n = 30) [28]. Nine patients (12%) had a decrease in LVEF. All patients recovered after medical management. There was no difference in survival between patients who had a cardiac event and those who did not. Whether or not cardiac toxicity of anti-angiogenic drugs is related to drug-induced hypertension cannot be determined.

It should be mentioned that, among targeted anti-cancer biotherapies, cardiotoxicity is not specific to anti-angiogenic drugs. Trastuzumab (Herceptin®) induces left ventricular systolic dysfunction in breast especially when associated with an anthracycline [29, 30].

Changes in ECG have been described with sunitinib and vandetanib. Sunitinib may induce bradycardia, PR and QT prolongations, torsades de pointe (in less than 0.1% of patients). Vandetanib induces QT prolongation (usually asymptomatic) in 15% of patients with non-small cell lung cancer compared with none in the control group [31].

Recommendations for management of cardiovascular side effects

Information on management of bevacizumab-induced hypertension is scarce. A recent meeting of the National Cancer Institute set a goal of 140/90 mmHg (130/80 mmHg for patients with diabetes or chronic kidney disease) for all patients with hypertension induced by inhibitors of VEGF signaling pathway. Active management of hypertension should be performed before initiation of such therapies. Blood pressure should be measured at least once a week during the first 6 weeks of therapy.

As inhibitors of VEGF signaling pathway frequently induce proteinuria in association with hypertension, it seems logical to systematically perform a screening test for proteinuria before initiation of therapy and to recommend angiotensin-converting enzyme (ACE) inhibitors for the management of hypertension. Verapamil and diltiazem, which are non-dihydropyridine calcium channel blockers inhibiting CYP 3A4, should be avoided in patients taking sunitinib or sorafenib, which are potent inhibitors of CYP3A4. Dihydropyridine calcium channel blockers and ACE inhibitors might be recommended as first-line anti-hypertensive therapy in this population. If, during the first week after initiation of anti-angiogenic therapy, systolic blood pressure rises over 165 mmHg or diastolic blood pressure over 100 mmHg, anti-angiogenic therapy should be held until effective titration of anti-hypertensive therapy [32].

After discontinuation of sunitinib or sorafenib for cardiotoxicity, most but not all patients recovered a baseline left ventricular function [8]. To our knowledge, there is no specific guideline for the management of cardiotoxicity in patients receiving bevacizumab. In patients with bevacizumab-induced cardiotoxicity, therapy should be held and specific cardiologic advice should be sought [24].

We are also unaware of specific recommendations for prevention of venous and arterial thrombo-embolic events during anti-angiogenic therapy. It would seem logical to prescribe preventive low molecular weight heparin (LMWH) in patients with a history or at high risk for venous thrombo-embolism. In their meta-analysis of venous thrombo-embolic events with chemotherapy plus bevacizumab, Hurwitz et al. [19] found that tumor type, old age, poor performance status, history of venous thrombo-embolism, baseline use of anticoagulants and recent surgery were risk factors. However, the benefit of LMWH in the prevention of venous thrombo-embolism in patients taking bevacizumab should be weighed by the increased risk of bleeding in cancer patients. Concerning arterial thrombo-embolism, no study compared aspirin and anticoagulation in this specific population under bevacizumab [18].

Conclusion

Although the cardiovascular toxicities of anti-angiogenic drugs must be taken into account, the risks-to-benefit ratio remains favorable to these agents at least for the indications recognized by FDA and EMA. The incidence and severity of elevated blood pressure among cancer patients vary according to the nature of the drug, its dose, and its prescription modalities, but also according to the age of the patients and the presence of coexisting cardiac diseases.

An Italian meta-analysis specifically assessed the risks-to-benefit ratio of bevacizumab in metastatic colorectal cancer [33]. This meta-analysis, based upon 6 out of 17 eligible studies (these 6 studies included 3,385 patients) found that bevacizumab significantly increased the frequency of 3 side effects only: high blood pressure (RR 2.98, 95% CI 2.32–3.84), gastro-intestinal perforations (RR 5.04, 1.72–14.79), and bleeding (RR 2.07, 1.19–3.62). These side effects should be put in perspective with a statistically significant improvement in Overall Survival (HR 0.80, 0.71–0.91) and Progression Free Survival (HR 0.62, 0.52–0.74). The authors concluded to a favorable risks-to-benefit balance for bevacizumab in this indication.

In 2011, Ranpura et al. published a meta-analysis including 10217 cancer patients from 16 randomized controlled trials [34]. Bevacizumab-related mortality (2.9%) was significantly higher than in the control group receiving chemotherapy alone (2.2%) (RR 1.33; 95% CI 1.02–1.73). In patients receiving taxanes or platinum derivates, bevacizumab increased the risk of fatal adverse events (3.3% vs. 1%; RR 3.49; 95% CI 1.82–6.66). There was no difference in bevacizumab-induced mortality according to the type of cancer or the dose of bevacizumab used. The commonest lethal adverse events were bleeding (23.5%), neutropenia (12.2%) and gastro-intestinal tract perforation (7.1%). Thus, this meta-analysis concluded to an increased mortality related to bevacizumab compared with chemotherapy alone.

In conclusion, although anti-angiogenic drugs have shown several cardiovascular toxicities, the risks-to-benefit balance of these drugs remains favorable. As cancer patients are getting older, the emphasis should be put on the use of anti-angiogenic drugs in the elderly, for whom the risks-to-benefit balance might be less favourable [35]. To improve the diagnosis and management of the cardiovascular side effects of anti-angiogenic drugs, the dialogue between oncologists and cardiologists should be strengthened [36].