What We Don't Know About Avastin Might Hurt Us

The emerging vascular endothelial growth factor (VEGF) antagonists appear so far to represent a genuine breakthrough in the treatment of subretinal neovascularisation. Pegaptanib sodium (Macugen; OSI Eyetech, Melville, NY) is the first to have its efficacy and safety demonstrated by phase III clinical trials.¹ Inhibiting receptor binding by attaching around amino acid 137, it is selective for the VEGF-A₁₆₅ isoform. The as yet unpublished 2-year results of the phase III studies of ranibizumab (Lucentis; Genentech, South San Francisco, Calif) may be even more promising. Ranibizumab is an antibody fragment that binds to all isoforms of VEGF-A at amino acids 86-89. Bevacizumab (Avastin) is an IgG1-antibody that binds VEGF-A the same way as ranibizumab. It is available now because it was approved for the treatment of metastatic colorectal cancer in 2004, but how do we know whether it is safe and efficacious for the treatment of subretinal neovascularisation?

While ranibizumab is derived from bevacizumab, the 2 molecules have quite different pharmacokinetics and VEGF-binding affinity.^{2 - 6} Ranibizumab (48 kDa)² is an antibody fragment approximately one third of the size of bevacizumab (148 kDa)³ that penetrated the retina much better than a full-sized antibody after intravitreal injection in monkeys.⁴ Ranibizumab has also been modified to increase 100 times its affinity for the VEGF-A receptor binding domain (Kd = 0.14 nM vs. 20 nM for bevacizumab).² The vitreous and serum half-lives of bevacizumab are much longer than those of ranibizumab (5.6 vs 3.2 days and 21 days vs 15 hours, respectively^{4 - 6} ), raising the possibility of both local and systemic overdosage if bevacizumab is used in the same way as ranibizumab.

Because subretinal neovascularisation causes breakdown of the blood-retinal barrier, it is inevitable that drugs injected into the eye will appear in the systemic circulation, where they appear to inhibit important physiological functions of VEGF such as wound healing and the formation of collateral circulations in myocardial and peripheral vascular ischemia.⁷ Patients receiving systemic bevacizumab treatment in combination with 5-fluorouracil have twice the risk of serious thromboembolic events compared with those treated with 5-fluorouracil alone.⁸ A phase III trial including avastin in postsurgical adjuvant colon cancer recently suspended enrollment because of a cluster of sudden deaths in 1 of the treatment arms (http://www.roche.com/med-cor-2006-02-13). The safety data from the pegaptanib studies have been good so far, but postmarketing surveillance has been requested in view of a perceived risk of serious adverse events arising from long-term therapy.⁹ The safety of ranibizumab and bevacizumab cannot be inferred from that of pegaptanib because they bind VEGF-A differently. If ranibizumab and bevacizumab have a wider action then they will also have a higher risk profile. Nor will the 2-year safety data for ranibizumab, if they are acceptable when they are eventually published as anticipated, automatically apply to bevacizumab because of the different pharmacokinetics and affinities of the 2 drugs as outlined above.

One of the reasons for a caution which may seem excessive to some is the possibility that VEGF inhibitors may eventually cause retinal atrophy by blocking the cytokine's recently discovered neuroprotective actions, which have received scant attention in the ophthalmic literature. In the normal nervous system, VEGF is now recognized as an important signaling molecule.^{10 - 11} It stimulates the proliferation of neuronal precursor cells in vitro,¹² promotes the growth of neuronal processes,¹³ and protects neurons from a range of acute¹² and chronic¹⁴ injuries. Transgenic mice in which hypoxic regulation of VEGF is disturbed develop a motor neuron degeneration similar to amyotrophic lateral sclerosis.¹⁴ Thus VEGF itself is proposed as a treatment of a range of neurological diseases.^{10 - 11 ,15} Direct evidence for a potentially relevant trophic effect of VEGF in the retina comes from a study in which it stimulated the proliferation of photoreceptors in early postnatal rodent retinal cultures.¹⁶ These observations in laboratory models will not necessarily prove clinically relevant, but they are cause for caution particularly when the optimal dose and therapeutic range of bevacizumab are yet to be established

Several reports have appeared on the use of bevacizumab for the treatment of macular disease.^{17 - 18} Experience has shown that preliminary, unmasked, uncontrolled studies of new treatments for subretinal neovascularsation (interferon alfa, radiation, triamcinolone, submacular surgery) almost invariably produce highly favorable results that are not confirmed by phase III trials. The consensus of anecdotal evidence is that bevacizumab is efficacious for subretinal neovascularization in the short term; however, open-label “pilot” studies track long-term local and systemic safety data rarely, if ever.

Treatment with intravitreal bevacizumab is certainly affordable, but that is because it has not been rigorously tested for use in the eye. Pegaptanib and ranibizumab were designed specifically for intravitreal use and have been or are being tested for safety and efficacy in long-term, closely monitored phase III randomized clinical trials that were preceded by appropriate preclinical testing. The widespread use of intravitreal triamcinolone, which we can now see seems to have an acceptable adverse event profile, might be invoked as an example of how such a treatment can develop without the need for extensive clinical research. However, it is one thing to inject into the vitreous a drug that had been injected around the eye for decades, and quite another to inject a drug that has never been near the eye and that does not have any clinical safety data much longer than a few months.

On the other hand, there is nothing legally or ethically wrong in medical practitioners recommending whatever treatment that they feel is appropriate for their patients' conditions. Patients considering treatment with bevacizumab need to know not just that there are risks, but also that we cannot be certain yet what they actually are. Given the extreme circumstances in which patients with subretinal neovascularisation often find themselves, many will accept this. Practitioners who do choose to use bevacizumab must make every effort to track their patients assiduously to collect all available data on local and systemic safety as well as efficacy for as long as possible.

Correspondence: Dr Gillies, Save Sight and Eye Health Institute, University of Sydney, GPO Box 4337, Sydney New South Wales 2001, Australia.

Financial Disclosure: Dr Gillies has participated as a member of advisory boards for both Pfizer and Novartis.