Before the recent explosion of intravitreal pharmacotherapy, management of central retinal vein occlusion (CRVO) had long been guided by the Central Vein Occlusion Study. The results of this landmark study, published in 1995, demonstrated no visual benefit with grid-pattern laser, and observation therefore was recommended with panretinal photocoagulation on detection of neovascularization. Without any proven treatment options before the incorporation of intravitreal therapy, management of CRVO had been straightforward.
In this issue of the American Journal of Ophthalmology , Brown and associates present 12-month data from the Vascular Endothelial Growth Factor (VEGF) Trap-Eye: Investigation of Efficacy and Safety in Central Retinal Vein Occlusion (COPERNICUS) trial, whose visual benefits prompted approval of aflibercept (VEGF Trap-Eye or Eylea; Regeneron Pharmaceuticals Inc, Tarrytown, New York, USA) by the Food and Drug Administration (FDA) for treatment of CRVO in September 2012. Aflibercept is the most recent member of the anti-VEGF armamentarium against CRVO, which also includes ranibizumab (Lucentis; Genentech Inc, San Francisco, California, USA), approved by the FDA in June 2009 based on results from the Ranibizumab for the Treatment of Macular Edema after Central Retinal Vein Occlusion (CRUISE) study, and bevacizumab (Avastin; Genentech Inc), whose extensive, international use remains off label. Other proven treatment options include a sustained-release intravitreal dexamethasone delivery system (Ozurdex; Allergan Inc, Irvine, California, USA), which was approved by the FDA in June 2009 based on results from the Global Evaluation of Implantable Dexamethasone in Retinal Vein Occlusion with Macular Edema (GENEVA) trial, and triamcinolone acetonide, whose continued off-label use was supported by results from the Standard Care vs Corticosteroid for Retinal Vein Occlusion (SCORE) trial.
We, as ophthalmologists, are privileged to witness a profound advance in our ability to alter definitively the natural history of a visually disabling condition for which previously there had been no real treatment. In the COPERNICUS trial, 56.1% of eyes treated every 4 weeks with aflibercept gained 15 letters or more after 24 weeks, with a mean improvement in central retinal thickness of 457 μm, compared with a 15-letter or more gain in only 12.3% of sham-injected eyes and a mean improvement of 145 μm. As reported by Brown and associates, these visual and anatomic improvements were maintained with a favorable side-effect profile through 52 weeks, with continued aflibercept treatment on an as-needed basis. Effective treatment profiles also have been reported in the pivotal clinical trials with ranibizumab, dexamethasone, and triamcinolone acetonide, and, anecdotally and in smaller trials, with bevacizumab.
But suddenly, we have become inundated with choices. Bevacizumab, ranibizumab, dexamethasone, triamcinolone, and now aflibercept—how does one choose? The evidence-based practice of medicine would direct us toward the primary trial data, but these studies compared treatment only with sham or observation and not with each other. Furthermore, comparison of these studies is limited by a lack of uniformity among primary end points. Although the primary end point in the CRUISE trial was mean change in best-corrected visual acuity at 6 months, the primary end point in the GENEVA trial was time to a 15-letter or more improvement at 6 months, and in the SCORE trial, the primary end point was the proportion of subjects with a 15-letter or more improvement at 1 year; in the recent COPERNICUS trial, the primary end point was proportion of subjects with a 15-letter or more improvement at 6 months.
In lieu of these nonequivalent primary end points, it may be tempting to compare common clinical end points shared across trials. For example, although 56% of aflibercept-treated eyes gained 15 letters or more at 6 months in the COPERNICUS trial (versus 12% of sham-treated eyes), 48% of ranibizumab-treated patients at 6 months in the CRUISE trial (versus 17% sham), 29% of patients after a single injection of Ozurdex at 30 days in the GENEVA trial (versus 11% sham), and 27% of triamcinolone acetonide-treated patients at 1 year in the SCORE trial (versus 7% sham) gained 15 letters or more. Before comparing these end points, however, it is critical to recognize that these trials also lack uniformity of enrollment criteria. The GENEVA trial was unique in its analysis of both central and branch retinal vein occlusion patients as a single assessed group. Additionally, differences in the numerous inclusion and exclusion criteria among these studies result in distinct trial populations with respect to baseline visual acuities, chronicity of disease, and proportion of ischemic cases, among other factors. These differences are reflected by the variable responses among sham-treated eyes and prevent direct comparison among trials of even these identical end points.
Pediatric ophthalmologic clinical trials generally are overseen by the rubric of the Pediatric Eye Disease Investigator Group. Other areas of ophthalmology and medicine undoubtedly would benefit from such a collaborative network, which could ensure uniformity in otherwise seemingly random enrollment criteria and end point analyses. Certainly, the bar should be raised to expect a standardization of clinical trials at least for same-class therapies targeting the same disease. Without such standardization, the evidence-based practice of medicine is relegated to guesswork.
Even with such standardization, however, regional, temporal, or other differences may preclude accurate comparisons of outcomes. Perhaps what is needed are large, randomized, double-blind trials to compare treatment options. The landmark Comparison of Age-related Macular Degeneration Treatments Trial provided such a comparison of bevacizumab with ranibizumab for treatment of neovascular age-related macular degeneration (AMD). In a survey of retina specialists, 79% in the United States (admittedly including the authors) and 89% internationally reported that the 2-year Comparison of Age-related Macular Degeneration Treatments Trial data did not change their treatment paradigms for neovascular AMD (Jumper JM, Mittra RA, eds. American Society Retina Specialists 2012 Preferences and Trends Membership Survey). Perhaps most of us are treating neovascular AMD the so-called right way. More likely, as much as we extol the practice of evidence-based medicine, we each are susceptible to developing intrinsic biases, particularly in the absence of definitive data, that inextricably shape the way we interpret evidence and practice medicine.
In the authors’ practices, macular edema associated with CRVO typically is treated initially with anti-VEGF agents, and consideration for steroids generally is reserved for refractory cases, particularly in pseudophakic patients without glaucoma. There are compelling scientific and nonscientific arguments for the use of bevacizumab versus ranibizumab, and it will be interesting to follow how the addition of aflibercept to the anti-VEGF armamentarium affects use and pharmacotherapy selection. In the treatment of neovascular AMD, the option to dose bimonthly is uniquely ascribed among these medications to aflibercept. However, as with ranibizumab, FDA approval of aflibercept was granted for monthly treatment of CRVO. Ultimately, unless a clear difference in efficacy emerges, it is likely that the choice of anti-VEGF agent will continue to be driven similarly not by any substantive evidence, but by individual experiences in training or practice, socio-politico-economic biases, or simply availability.
Genentech, Allergan, and now Regeneron certainly should be commended for revolutionizing our treatment of CRVO and other retinal diseases. And although we can never have too many treatment choices, it will be important to develop evidence-based determinations of the individual merits (and risks) of each choice before intrinsic biases become ingrained. Particularly because multiple treatments are now being developed rapidly for the same conditions, the need for standardization of trials becomes increasingly essential. More important, however, we must remember that most of these treatment options were adapted secondarily for use in CRVO based on primary development as therapy for neovascular AMD or other diseases. All of these treatments, although effective, are temporizing measures that target downstream mediators of CRVO sequelae and not the underlying causes. Ultimately, our understanding of CRVO pathogenesis needs to be advanced to develop true treatment options that not only address the underlying pathophysiology, but also can be compared judiciously with other available therapies. Until then, perhaps we are just spinning in place.