How to choose the right anti-VEGF

With three to choose from, it’s important to know the seminal studies and what works for DME patients.

By Sanket U. Shah, MD, and Raj K. Maturi, MD

With three anti-VEGF drugs now available to treat diabetic macular edema, the question for the clinician becomes which one to use. In our practice, we have paid close attention to recent prospective studies to answer this question. One notable finding is that for subjects with visual acuity better than 69 letters (20/40), no difference was noted among the three anti-VEGF drugs tested, Avastin (bevacizumab), Lucentis (ranibizumab) and Eylea (aflibercept).¹

Studies, however, don’t always reflect real life. In clinical practice, for example, we do not perform strict high-contrast ETDRS (early treatment diabetic retinopathy study) acuity testing. Thus, we have found that a study refraction-based visual acuity score of 20/40 generally translates into a chairside, unrefracted acuity (wearing current prescription) that is likely in the 20/60 range. Thus, in our clinic, a large group of patients with chairside acuity of less than 20/60 and with severe edema, are begun with aflibercept. Those with acuity better than this are generally begun on bevacizumab. Additionally, we switch patients who are only partially responsive to anti-VEGF (about 30% of subjects) to intraviteral Ozurdex (dexamethasone implant).

That said, we have derived the following protocols from the findings of recent studies. Here’s a look at how they have informed our choices.

Pathophysiology of DME

Multiple biochemical factors and pathways interplay to initiate and propagate DME.

These include upregulation of vascular endothelial growth factor (VEGF), intercellular adhesion molecule-1, monocyte chemotactic protein-1, interleukin-6, pigment epithelium-derived factor, prostaglandins, other cytokines, leukocyte migration, and loss of inner retinal barrier function of capillary vascular tight junctions.^2-7

But it is VEGF upregulation that has emerged as the key factor in the pathogenesis of DME.⁵ Inflammation mediated by several pro-inflammatory cytokines plays an important role as well.^6,7 The inhibition of VEGF by these anti-VEGF therapies has demonstrated remarkable success in control of this disease, which affects up to 7% of diabetic patients resulting in varying levels of visual loss.^1,8-11 In the United States, one out of 25 diabetic patients aged 40 years or older has DME in at least one eye.¹² As the number of diabetic patients increases, so too will the disease burden. New-onset DME has been shown to be associated with about 29% increase in medical costs in the first three years of diagnosis.¹³

Rezaei KA, Stone TW, eds. 2014 Global Trends in Retina Survey: Chicago, IL. American Society of Retina Specialists; 2014.

These anti-VEGF agents have replaced laser photocoagulation as the gold standard first-line treatment for DME.^14,15

On their first visit to our clinic, patients undergo a review of their diet, renal status and HbA1c level. We have found that measures such as controlling salt intake and minimizing diet soda can be additionally helpful to the overall health of the patient. We also ask these patients to check their blood pressure daily if they are taking any antihypertensive treatment. Patients wearing a boot (for the treatment of a diabetic foot ulcer) are asked about the status of the ulcer. Should the ulcer be large, new or nonhealing, we avoid anti-VEGF agents and begin with a dexamethasone implant (Ozurdex, Allergan). We have also found that a combined steroid and anti-VEGF treatment are helpful in limited cases, although we generally avoid concurrent treatment with both agents.¹⁶

For a majority of patients, we generally begin them on the most affordable of the three, bevacizumab. For those whose eyes are responsive we continue this treatment modality. For the others, especially for those with severe edema, we switch to aflibercept. We begin with aflibercept if they have especially severe edema. For patients who are minimally responsive to anti-VEGF agents (less than a 20% decrease of excess edema despite four injections on a monthly basis), we switch to dexamethasone implant. We also consider switching to a dexamethasone implant for those patients who have residual edema with visual loss or those who balk at frequent treatments. We give serious consideration to thermal laser treatments for those who have diffuse edema, or especially focal edema, if “leaky” microaneurysms are present on angiography.

Details of the DRCR.net Protocol T

The landmark DRCR.net Protocol T compared the three anti-VEGF agents in DME.¹ This was a one-year, multicenter randomized clinical trial that enrolled adult diabetic patients with center-involving DME in at least one eye, with visual acuity letter score ranging from 24 to 78 letters, and who had no anti-VEGF treatment in the previous 12 months. The eyes were randomized to 2.0 mg aflibercept, 1.25 mg bevacizumab, or 0.3 mg ranibizumab in the ratio of 1:1:1, with q4 weeks injection for the first 24 weeks and using prespecified criteria thereafter. Macular laser photocoagulation was allowed at or after the 24 weeks for persistent diabetic macular edema.

An interesting outcome was the dependency of visual gain over a period of one year on baseline visual acuity. When the baseline visual-acuity letter score was 78 to 69 (corresponding to the Snellen acuity of 20/32 to 20/40, with an average acuity of 20/25), the mean improvement was 8.0 letters with aflibercept, 7.5 letters with bevacizumab, and 8.3 letters with ranibizumab (p>0.5 for each pairwise comparison). However, when the initial letter score was less than 69 (corresponding to Snellen acuity of 20/50 or worse, with an average acuity of 20/80), the mean improvement was 18.9 letters with aflibercept, 11.8 letters with bevacizumab, and 14.2 letters with ranibizumab (p<0.001 for aflibercept vs. bevacizumab, p = 0.003 for aflibercept vs. ranibizumab, and p = 0.21 for ranibizumab vs. bevacizumab). With regard to the mean change in central subfield thickness, irrespective of the baseline visual acuity, both aflibercept and ranibizumab significantly performed better than bevacizumab. Researchers, however, found no difference between aflibercept and ranibizumab. The rates of serious adverse events were similar in all three groups.

On cost-effectiveness

The DRCR.net Protocol T shows that aflibercept has greater efficacy in a certain subgroup of DME eyes than either ranibizumab or bevacizumab. But bear in mind that decisions in health care are based often on not just a drug’s effectiveness, but also its cost-effectiveness. In addition, it is difficult to quantify the extent of the impact a small difference (although statistically significant) the absolute visual acuity change has on the actual quality of life, unless we take into account indices such as quality-adjusted life-years (QALY).

In their study using the Markov model for the cost-effectiveness analysis of Protocol T, Holekamp et al. focused only on two agents — aflibercept and ranibizumab — and found that aflibercept jettisoned costs with minimal gain in QALYs compared with ranibizumab.³ When the baseline visual acuity was 20/40 or better, aflibercept rose $8,000 more than ranibizumab for one year of treatment, with no difference in QALY.¹⁷ When the baseline visual acuity was 20/50 or worse, the treatment costs were the same as for the baseline visual acuity, and there was only a minimal difference in QALY — 0.0103 — in favor of aflibercept.¹⁷ In view of the large cost difference between the two agents, and only a marginal QALY difference, the authors said ranibizumab appeared to be more cost-effective than aflibercept, irrespective of the baseline visual acuity.¹⁷

Researchers in the United Kingdom reported a similar cost-effectiveness outcome, favoring ranibizumab over aflibercept.¹⁸ Bevacizumab was excluded from these analyses for obvious reasons — it would be the low-cost beneficiary, with the largest QALY by far in any analysis. These findings are difficult to place into current practice; for example, a patient might note minimal gain if only the poorly seeing eye were to be treated, in the context of the healthy fellow eye.

Another thought on T

Protocol T also found that aflibercept and ranibizumab both reduced macular thickness significantly better than bevacizumab, and found this positive point translated into a better visual acuity gain than bevacizumab, but only in eyes with baseline visual acuity of 20/50 or worse.¹

As discussed by Martin and Maguire, about half of those with DME have a presenting visual acuity of 20/40 or better.

Considering the phenomenal difference in cost per dose among the three — with bevacizumab at (<$50), ranibizumab ($1,200), and aflibercept ($1,950), they recommended use of bevacizumab as first-line therapy in DME eyes with visual acuity of 20/40 or better, and aflibercept as first-line therapy in DME eyes with visual acuity of 20/50 or worse.¹⁷ Considering the QALY outcomes as outlined above, ranibizumab — and bevacizumab, given its statistical equivalence to ranibizumab in terms of visual acuity gain — would remain viable alternatives for this second group of eyes with worse acuity.¹⁹

Analysis of QALY outcomes of bevacizumab and comparison to the other two agents would provide more answers.

Other concerns

Publication of the DRCR.net Protocol T results prompted lots of discussion. Since the study involved a 0.3-mg dose of ranibizumab (which is FDA approved) instead of the initially approved 0.5-mg dose, eye-care professionals wondered if aflibercept performs better than the 0.5 mg dose of ranibizumab.²⁰ Also, the applicability of this single study was sorely questioned: It lacked confirmation from a parallel independent trial, and perhaps was tainted from investigator-bias due to “partial blinding”.²¹ A Cochrane review of randomized controlled trials and meta-analysis of outcomes reported that no significant subgroup difference existed between bevacizumab, ranibizumab and aflibercept for the two primary outcomes of visual acuity gain of three lines or more, and visual acuity loss of three lines or more.²²

Of course, even with protocol-based regular treatment of subjects, 30% – 50% of subjects in Protocol T had continued edema at the end of the first year of study. These subjects may have non-VEGF mediated sources for at least part of their visual dysfunction. There is a role for other modalities in such cases including intravitreal steroids or thermal laser treatment.

It will be interesting to follow the two-year results of protocol T, which are expected in November. While there are no clear answers, this study has provided us with an impetus to identify factors for differential treatment of DME eyes. There may be factors, such as and possibly in addition to baseline visual acuity, that could help us stratify DME eyes into subgroups in order to predict response to treatment, and ultimately achieve the best visual outcome for our DME patients. OM

REFERENCES

About the Authors
	Raj K. Maturi, MD, is an associate professor of ophthalmology at Indiana University School of medicine and practices at the Midwest Eye Institute in Indianapolis. He has conducted numerous clinical trials in an effort to advance treatment options available to patients in diabetes and retinal disease.
	Sanket U. Shah, MD, is a PGY4 Resident in Ophthalmology at Indiana University School of Medicine, Indianapolis, IN. He is planning to pursue a Vitreoretinal Surgery fellowship upon graduation.