Cardiovascular toxicity of angiogenesis inhibitors in treatment of malignancy: a systematic review and meta-analysis

Husam Abdel-Qadir, Josee-Lyne Ethier, Douglas S. Lee, Paaladinesh Thavendiranathan, Eitan Amir

Abstract

Background: The cardiovascular risk of angiogenesis inhibitors is not well-quantified. We hypothesized that, compared to direct vascular endothelial growth factor (VEGF) inhibitors (antiVEGF antibodies or decoy receptors), small molecule agents have higher risk due to their less specific mechanism.
Methods: We searched the MEDLINE, EMBASE, and Cochrane Central Register of Controlled Trials for phase III randomised controlled trials comparing angiogenesis inhibitor-based therapy to other systemic therapy. Outcomes evaluated were hypertension, severe hypertension, cardiac dysfunction, congestive heart failure, cardiac ischemia, arterial thromboembolism, venous thromboembolism, and fatal cardiovascular events. Data were pooled using Mantel-Haenszel random effects method to generate odds ratios (OR).
Results: We identified 77 studies meeting inclusion criteria. Compared to routine care, angiogenesis inhibitors were associated with a higher risk of hypertension (OR 5.28 [4.53 - 6.15], number needed to harm [NNH] 6), severe hypertension (OR 5.59 [4.67 - 6.69], NNH 22), cardiac ischemia (OR 2.83 [1.72 - 4.65], NNH 84) and cardiac dysfunction (OR 1.35 [1.06 - 1.70], NNH 139). VEGF inhibitors were associated with an increased risk of arterial thromboembolism (OR 1.52 [1.17, 1.98], NNH 140). No significant interaction was observed between the two drug subgroups for any outcomes. We identified no significant increase in the risk of the other outcomes evaluated.
Conclusion: Angiogenesis inhibitors increase the risk of hypertension, arterial thromboembolism, cardiac ischemia and cardiac dysfunction. There was no significant difference in cardiovascular risk between direct VEGF inhibitors and small molecule agents.

Introduction

Angiogenesis is a key factor for tumor growth and survival[1, 2]. Accordingly, angiogenesis inhibition is an attractive target for suppressing tumour growth, and angiogenesis inhibitors have been shown to improve outcomes in a variety of malignancies. Most angiogenesis inhibitors exert their effect by inhibiting the vascular endothelial growth factor (VEGF) signalling pathway, via one of two mechanisms[3, 4]. The first is direct inhibition of the VEGF ligan d’s ability to bind to its target receptor. The monoclonal antibodies, bevacizumab[5] and ramucirumab[6], as well as the VEGF decoy receptor aflibercept[7] inhibit angiogenesis via this mechanism. The second class of angiogenesis inhibitors are small molecules that inhibit the tyrosine kinases which would be activated by the VEGF ligand-receptor interaction. Agents in this class include sunitinib[8], sorafenib[9], pazopanib[10], vandetanib[11], vatalanib[12], cabozantinib[13], axitinib[14], and regorafenib[15] amongst others. In contrast to monoclonal antibodies, small molecules typically target multiple tyrosine kinases other than VEGF.
The inhibition of VEGF has deleterious effects on the cardiovascular system[3, 4]. VEGF inhibitors have been associated with an increased risk of adverse cardiovascular events such as hypertension, thromboembolic events, myocardial ischemia, left ventricular dysfunction, heart failure, and QT interval prolongation (with possible associated arrhythmias). However, these are rare events among patients enrolled in clinical trials, and individual studies designed to demonstrate efficacy of each of these agents are underpowered to detect statistically significant differences in the incidence of adverse cardiovascular events other than common toxicities such as hypertension. Moreover, it is unclear if the mechanistic differences between the two classes translate into clinically relevant differences in the rate of these cardiovascular events. Finally, it has been difficult to quantify how many of these adverse events lead to death, which is an important consideration in patients who are being treated with these drugs for advanced malignancies that pose an important competing risk of death. To determine the risk of adverse cardiovascular events associated with these agents, we conducted a systematic review of phase III and IV randomized controlled trials of adult patients with malignancy treated with routine care with or without an angiogenesis inhibitor.

Methods

Search Strategy

We conducted a systematic review and meta-analysis of Phase III randomised controlled trials (RCTs) and phase IV post-marketing studies according to PRISMA guidelines[16]. The main comparison was standard therapy, as determined by individual studies for the given malignancy being treated, compared with standard therapy plus an angiogenesis inhibitor. The angiogenesis inhibitors considered were: bevacizumab, aflibercept, ramucirumab, sunitinib, sorafenib, pazopanib, vandetanib, cabozantinib, axitinib, ponatinib, and regorafenib. Inclusion was not restricted based on language of publication. The studies were limited to those assessing patients aged 18 years or above, who are being treated for malignancy (solid or haematologic) at the time of study enrolment.
The MEDLINE, EMBASE, and Cochrane Central Register of Controlled Trials (CENTRAL) databases was searched with the aid of a librarian in June 2014 to identify all eligible studies published after 1990. The search was done in accordance with PRISMA recommendations. The search strategy combined three concepts: (1) malignancy; (2) angiogenesis inhibitors (individual drug names, as well as Medical Subject Heading related to Angiogenesis inhibitors, tyrosine kinase inhibitors, and VEGF), and (3) phase III/IV RCTs. To increase sensitivity, the search strategy did not include terms for cardiovascular or adverse events. The detailed search strategy is provided in Appendix 1.

Data extraction and management

The citations obtained from the electronic search were initially de-duplicated by the librarian conducting the search, and were subsequently manually de-duplicated by one author (HAQ) who then reviewed the titles and abstracts to select papers for more detailed review. Studies were reviewed independently by HAQ and JLE and data extracted independently using a common data abstraction form in Microsoft Excel[17]. Disagreements between the two reviewers were resolved by consensus, or by deference to a third author (EA) if needed. Reasons for exclusion were documented. If there were multiple reports of the same trial, we included data from the most up-to-date reference possible.
The outcomes of interest were: (1) hypertension; (2) severe hypertension (Grade 3 or higher); (3) arterial thromboembolic events; (4) cardiac dysfunction; (5) congestive heart failure; (6) cardiac ischemia and (7) fatal cardiovascular events. For all outcomes, we used the definition of the adverse event used by the individual clinical trials. We also initially aimed to include venous thromboembolism as an eighth outcome. Due to variable outcome definitions within trials, we assessed the following related outcomes: (1) deep vein thrombosis (DVT); (2) pulmonary embolism (PE); (3) VTE with site unspecified; (4) unspecified thromboembolism. At study onset, we had planned to study QT prolongation and arrhythmias as an additional outcome but this was rarely reported and this objective was thus abandoned.
The number of patients in each treatment arm who experienced the outcome of interest and the number of patients in the safety analysis were extracted from each study. Data were extracted from intention-to-treat analyses wherever possible. In cases where there were multiple intervention groups, we assessed the ones that were deemed closest to our primary intention of comparing background therapy without an angiogenesis inhibitor compared to the same therapy with the addition of the agent of interest. If there were arms that exposed patients to different doses of an angiogenesis inhibitor, those arms were pooled together. We also aimed to collect data on the weighted average of the median follow-up duration for adverse events in the study arms (or the mean if the median was unavailable). This was often not explicitly stated.
Accordingly, we collected data on treatment duration, or when that was unavailable, progressionfree survival as a surrogate. If none of these data were reported, we collected median follow-up instead. The risk of bias for each study was assessed using a modified version of the Cochrane Risk of Bias tool[18] that was modified to take into account funding by the pharmaceutical industry.

Data synthesis and statistical analysis

The extracted data were transferred to the Review Manager (RevMan) software package, Version 5.3[19]. The event numbers and numbers at risk were used to generate an odds ratio for the adverse cardiovascular outcome of interest in angiogenesis –inhibitor-treated patients compared to those who were not. It was anticipated that event rates would be low with frequent zero events within study arms. Thus, the Mantel-Haenszel random effects model with zero-cell corrections was used to pool the results for the purposes of this report. A random effects model was chosen to account for heterogeneity in included patients, underlying malignancies, and treatment regimens. Funnel plots for adverse outcomes were inspected visually for subjective evidence of substantial publication bias.
We performed subgroup analyses comparing VEGF ligand/ receptor inhibitors (typically monoclonal antibodies or decoy receptors) with small-molecule tyrosine kinase inhibitors. We also performed subgroup analyses based on whether treatment was delivered in the adjuvant or palliative setting, and study blinding status. We studied if the risk of heart failure, cardiac dysfunction, and fatal cardiovascular events was affected by concurrent anthracycline exposure as part of the standard treatment. In addition, we performed meta-regression analyses using linear regression weighted by study sample size to explore the influence of mean/ median age and sex (the proportion of women in individual studies) on the OR for each outcome of interest. Differences between the three cohorts (subgroups) were assessed using methods described by Deeks et al[20]. Absolute risks of each adverse event were calculated as the number of events per person over the follow-up period of the trial. The difference in absolute risk between the angiogenesis inhibitor group and the control group was also presented as the number needed to harm (NNH). All statistical tests were two-sided, and statistical significance was defined as P <0.05.

Results

As illustrated in figure 1, the literature search yielded 9830 potentially relevant citations, from which 77 citations were deemed to fit our inclusion criteria. The characteristics of these studies are summarized in Table 1, with detailed descriptions provided in Appendix 2. Forty studies evaluated bevacuzimab[9, 21-59], 10 sorafenib[60-68], 9 sunitinib[8, 69-76], 5 vandetanib[77-81], 4 aflibercept[7, 82-84], while 2 studies each evaluated ramucirumab[6, 85], pazopanib[86, 87], and vatalanib[12, 88], and one study evaluated each of cabozantinib[89], regorafenib[15], and axtinib[90].
The quality of studies, as assessed by the modified Cochrane risk of bias tool, is summarized in Appendix 3. All studies were well-randomized with excellent follow-up. Most studies had prospective adverse event monitoring using well-described, objective criteria, although the types of adverse events studied and their definition varied between trials . The risk of bias within the studies was most threatened by two factors. First, many studies were unblinded. Second, the majority of studies were funded by pharmaceutical companies, who were often involved in data collection and analysis as well as manuscript preparation. However, examination of funnel plots for adverse outcomes suggested against the presence of substantial publication bias.
Hypertension was a well-recognized complication that was consistently reported, either Grade 3 or higher, or less commonly as any newly incident hypertension. Other adverse cardiac events were reported less frequently or consistently. Thromboembolic events in particular were difficult to extract in a consistent manner due to multiple classifications within trials as any thromboembolism, arterial thrombosis, venous thrombosis, or events restricted to specific vascular beds (such as cerebrovascular events or deep vein thrombosis). Thus, only a limited number of studies were deemed to have data that could be categorized into our pre-specified outcome definitions. Fatal cardiovascular events were also inconsistently reported, and their rates were mostly determined from a review of all reported Grade 5 events, and subsequent determination by the two reviewers whether the event qualified as a primary cardiovascular event. We also noted that small molecule inhibitors had substantially higher rates of noncardiovascular toxicity, particularly in hematologic malignancies. Thus, trials reporting small molecule inhibitors were more likely to limit the reporting of adverse events to those that occurred above a certain frequency (e.g. 3% in many studies).
The results of the meta-analysis are summarized in Figure 2 and Table 2, while the detailed forest plots are provided in Appendix 4. Angiogenesis inhibitors were associated with a significantly higher odds ratio of hypertension (5.28 [95% CI 4.53 - 6.15]), severe hypertension (5.59 [95% CI 4.67 - 6.69]), arterial thromboembolism (1.52 [95% CI 1.17 - 1.98]), cardiac dysfunction (1.35 [95% CI 1.06 - 1.70]), and cardiac ischemia (2.83 [95% CI 1.72 - 4.65]). We did not observe an increased odds ratio for of fatal cardiovascular events, other forms of thromboembolism, or congestive heart failure. When comparing NNH, the most substantial increase in adverse events was noted for hypertension (NNH=6), followed by severe hypertension (NNH=16), and cardiac ischemia (NNH=84). Heterogeneity, as determined by the I2 statistics was moderate for arterial thromboembolism (17%) and severe hypertension (39%), while it was substantial for any hypertension (66%) and other thromboembolism outcomes. It was assessed as low for other evaluated outcomes.
The results of subgroup and meta-regression analyses are summarized in Appendices 5 and 6. The subgroup analysis examining VEGF ligand inhibitors relative to small molecule agents showed no significant differences for most outcomes, with the exception of unspecified venous thromboembolism, where a protective effect was suggested for small molecule agents (OR 0.51; 95% CI 0.26 – 0.98) compared with a neutral risk for VEGF ligand inhibitors (OR 1.10; 95% CI 0.99 – 1.22). When comparing the estimated cardiovascular risk reported in studies performed in the adjuvant versus advanced disease settings, a significantly higher risk of hypertension, severe hypertension, and unspecified venous thromboembolism was noted in the adjuvant setting. There was no difference in the risk of cardiac dysfunction, congestive heart failure, or fatal cardiovascular events in studies with concurrent anthracycline exposure relative to its absence. Blinded studies were noted to have a lower reported odds ratio for severe hypertension, unspecified venous thromboembolism, and unspecified thromboembolism compared to open-label studies. Finally, the majority of meta-regression analyses showed no significant associations with outcomes risk. However, we observed statistically significant associations linking increased age with a lower odds ratio for hypertension (beta = -0.41 / one year; p= 0.03), congestive heart failure (beta = -0.29 / one year; p= 0.004), and unspecified thromboembolism (beta = -0.10 / one year; p= 0.03). We also observed significant associations between the percentage of female patients with higher odds ratios for hypertension (beta = 0.079 / percentage of females; p= 0.02) and severe hypertension (beta = 0.13 / percentage of females; p= 0.03).

Discussion

This meta-analysis of 77 Phase III trials confirms that patients treated for malignancy with an angiogenesis inhibitor containing regimen are at higher risk of hypertension, cardiac ischemia, arterial thromboembolism, and cardiac dysfunction relative to control. The numbers needed to harm were lowest for hypertension, severe hypertension and cardiac ischemia suggesting a clinically relevant increase in the risk of these events. Other cardiovascular events occurred less frequently with less substantial risk increases. Fatal cardiovascular events were rare and were not significantly increased among patients treated with these agents. Moreover, we did not observe meaningful differences in the risk of cardiovascular events within the two broad mechanistic classes of angiogenesis inhibitor agents.
Newly incident hypertension related to use of angiogenesis inhibitors has been described to mostly respond well to routine antihypertensive therapy[91]. Some studies have suggested that the development of hypertension may serve as a predictive marker of disease responsiveness to angiogenesis inhibitor therapy[92]. However, a more comprehensive review of data from seven trials suggested that the development of early hypertension was not significantly associated with treatment benefit, whether measured by progression free survival or overall survival[93]. The additional medications may compromise the quality of life of patients with limited survival time, as would the development of cardiac ischemia or heart failure. These agents may be used in patients at higher baseline cardiovascular risk than those included within clinical trials. The odd ratios and numbers needed to harm derived from this meta-analysis should thus be interpreted in light of a patient’s baseline risk to estimate a specific patient’s anticipated event rate. Accordingly, we anticipate that our findings may help physicians and their patients gauge the risk-benefit of adding an angiogenesis inhibitor to a patient with advanced malignancy.
Interestingly, we observed an increased risk of thromboembolism in the arterial beds (including cardiac ischemia) but not the venous system. This may indicate that these events are mediated through hypertension-induced endothelial damage. VEGF is also important in promoting endothelial cell survival under stress[91], which may further increase the susceptibility to endothelial damage secondary to hypertension. Alternatively, this may be mediated by the inhibition of nitric oxide metabolism, which is important in maintenance of endothelial function[94, 95]. However, this observation should be interpreted cautiously given the inconsistent classification of thromboembolic events in reports of adverse events from the reviewed trials. The pathways inhibited by angiogenesis inhibitors are also important in modulating the cardiac response to stress, increasing the susceptibility to heart failure secondary to hypertension[96]. This risk of heart failure did not appear to be importantly affected by concurrent anthracycline use, although a higher risk has been reported with anthracycline use prior to angiogenesis therapy initiation[44].
Our subgroup and meta-regression analyses showed that the estimated odds ratios for developing certain adverse events were lower with increasing age and greater proportion of male patients. This may be a reflection of the lower baseline incidence of cardiovascular events at baseline in women and younger patients. For some events, the odds were of greater magnitude in the adjuvant setting and in unblinded trials. These observations may inform adverse event monitoring protocols of future trials. This will be particularly relevant if patients treated in the adjuvant setting, whose risk of cancer-related death is markedly lower than in those with metastatic disease. The risk of adverse cardiovascular events may be underestimated with data from patients with advanced disease, for whom cancer-related death provide a high level of competing risks that preclude the development of a cardiac event[97]. As the expected risk of these competing risks are lower in the adjuvant setting, we may expect a higher absolute incidence of cardiovascular events, including fatal ones, if these drugs become used in the setting of earlier stage cancer.
This study has several limitations. First, we relied on adverse event data provided in published clinical trials. Reported adverse events were selected by the authors of individual studies based on varying criteria, such as perceived clinical relevance or frequency occurring above a certain threshold. Therefore, many of the adverse events of interest to our study may have not been deemed clinically relevant, and thus not reported[98]. Moreover, some events may have occurred at rates below the threshold for reporting while still occurring at a higher rate compared to standard therapy, which may be clinically relevant when these drugs are considered for less selected patients outside of clinical trials. This may have been particularly relevant for the small molecule agents where a larger variety of adverse events were reported, and the less frequently occurring cardiovascular events may have been unreported. Second, there was significant heterogeneity between studies. This was likely driven by different patient populations with diverse malignancies who underwent different baseline treatment regimens. However, the underlying impact of angiogenesis inhibitors on the vascular system is driven by physiology at the level of the endothelium that should be consistent across these settings, such that the odds ratio can be validly pooled across these different studies. The incorporation of a random effects model acknowledges the substantial heterogeneity in the background environment within which these cardiovascular events are occurring. Finally, the determination of cause of death in cancer trials may have been inaccurate[99] such that cardiovascular death was attributed to other causes including to the underlying cancer.

Conclusion

Our systematic review and meta-analysis confirms and quantifies the increased cardiovascular risk associated with angiogenesis inhibitors as a class effect. While hypertension is the most common and clinically recognized adverse event, there is a substantial increase in the risk cardiac ischemia and a smaller increase in the risk of cardiac dysfunction and arterial thromboembolism. Our reported odds ratios should be considered along with a patient’s baseline cardiovascular risk and the anticipated benefit from an angiogenesis inhibitor and other competing risks prior to making treatment decisions with these agents. Moreover, the anticipated event rate should be used to guide follow-up plans and possible risk reduction measures in patients treated with an angiogenesis inhibitor. This increase in cardiovascular risk should also be taken into account in future trial design, particularly if these drugs are to be used in the adjuvant setting.

Reference List

1. Folkman J. Tumor angiogenesis: therapeutic implications. N Engl J Med 1971;285(21):1182-6.
2. Dvorak HF, Brown LF, Detmar M, et al. Vascular permeability factor/vascular endothelial growth factor, microvascular hyperpermeability, and angiogenesis. (REGARD): an international, randomised, multicentre, placebo-controlled, phase 3 trial. Lancet 2014;383(9911):31-9.
3. Van Cutsem E, Bajetta E, Valle J, et al. Randomized, placebo-controlled, phase III study of oxaliplatin, fluorouracil, and leucovorin with or without PTK787/ZK 222584 in patients with previously treated metastatic colorectal adenocarcinoma. J Clin Oncol 2011;29(15):2004-10.
4. Smith DC, Smith MR, Sweeney C, et al. Cabozantinib in patients with advanced prostate cancer: results of a phase II randomized discontinuation trial. J Clin Oncol 2013;31(4):412-9.
4. Spano JP, Chodkiewicz C, Maurel J, et al. Efficacy of gemcitabine plus axitinib compared with gemcitabine alone in patients with advanced pancreatic cancer: an open-label randomised phase II study. Lancet 2008;371(9630):2101-8.
5. Grothey A, Van Cutsem E, Sobrero A, et al. Regorafenib monotherapy for previously treated metastatic colorectal cancer (CORRECT): an international, multicentre, randomised, placebo-controlled, phase 3 trial. Lancet 2013;381(9863):303-12.
6. Moher D, Liberati A, Tetzlaff J, et al. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med 2009;6(7):e1000097.
7. Microsoft Excel. In. 2010 ed. Redmond, Washington: Microsoft; 2010.
8. Cochrane Handbook for Systematic Reviews of Interventions. In: S HJG, (ed). Version 5.1.0 [updated March 2011]: The Cochrane Collaboration; 2011.
9. Review Manager (RevMan) [Computer program]. In. 5.3 ed. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration; 2014.
10. Deeks JJ. Systematic reviews in health care: Systematic reviews of evaluations of diagnostic and screening tests. Bmj 2001;323(7305):157-62.
11. Aghajanian C, Blank SV, Goff BA, et al. OCEANS: a randomized, double-blind, placebo-controlled phase III trial of chemotherapy with or without bevacizumab in patients with platinum-sensitive recurrent epithelial ovarian, primary peritoneal, or fallopian tube cancer. J Clin Oncol 2012;30(17):2039-45. 12. Allegra CJ, Yothers G, O'Connell MJ, et al. Initial safety report of NSABP C-08: A randomized phase III study of modified FOLFOX6 with or without bevacizumab for the adjuvant treatment of patients with stage II or III colon cancer. J Clin Oncol 2009;27(20):3385-90.
13. Bear HD, Tang G, Rastogi P, et al. Bevacizumab added to neoadjuvant chemotherapy for breast cancer. N Engl J Med 2012;366(4):310-20.
14. Boutsikou E, Kontakiotis T, Zarogoulidis P, et al. Docetaxel-carboplatin in combination with erlotinib and/or bevacizumab in patients with non-small cell lung cancer. Onco Targets Ther 2013;6:125-34.
15. Brufsky AM, Hurvitz S, Perez E, et al. RIBBON-2: a randomized, double-blind, placebo-controlled, phase III trial evaluating the efficacy and safety of bevacizumab in combination with chemotherapy for second-line treatment of human epidermal growth factor receptor 2-negative metastatic breast cancer. J Clin Oncol 2011;29(32):4286-93.
16. Burger RA, Brady MF, Bookman MA, et al. Incorporation of Vandetanib bevacizumab in the primary treatment of ovarian cancer. N Engl J Med 2011;365(26):2473-83.
17. Cameron D, Brown J, Dent R, et al. Adjuvant bevacizumab-containing therapy in triple-negative breast cancer (BEATRICE): primary results of a randomised, phase 3 trial. Lancet Oncol 2013;14(10):933-42.
18. Chinot OL, Wick W, Mason W, et al. Bevacizumab plus radiotherapy-temozolomide for newly diagnosed glioblastoma. N Engl J Med 2014;370(8):709-22.
19. Corrie PG, Marshall A, Dunn JA, et al. Adjuvant bevacizumab in patients with melanoma at high risk of recurrence (AVAST-M): preplanned interim results from a multicentre, open-label, randomised controlled phase 3 study. Lancet Oncol 2014;15(6):620-30.
20. Cunningham D, Lang I, Marcuello E, et al. Bevacizumab plus capecitabine versus capecitabine alone in elderly patients with previously untreated metastatic colorectal cancer (AVEX): an open-label, randomised phase 3 trial. Lancet Oncol 2013;14(11):1077-85.