Journal Club

Recent Noteworthy Studies to Stimulate Discussion and Debate

■ Avastin for RVO. With ranibizumab already approved for the treatment of macular edema following retinal vein occlusion and with the results of the CATT trial showing comparable results of ranibizumab and bevacizumab in wet AMD, whether bevacizumab is effective in macular edema following RVO is a hot topic.

To investigate this issue, retinal physicians at the Cole Eye Institute in Cleveland and at Duke University collaborated on a trial of intravitreal bevacizumab for macular edema secondary to RVO. They report their results in the September 2011 issue of Retina.

In a retrospective case series, 53 patients with branch RVO were examined for variables including best-corrected visual acuity and central foveal thickness. Final mean BCVA improved from 20/137 at baseline to 20/96, for a mean final change of +1.6 lines at an average follow-up of nine months. Twenty-eight percent of the patients studied had their BCVA improve by more than three lines; however, losses of three lines or more were seen in 6% of eyes. Mean central foveal thickness decreased from 425 µm to 289 µm, and the mean number of injections was 2.5. Effective treatment appeared to be limited by the length of symptoms, as patients treated after more than six months showed worse functional outcomes than those treated at six months or less.

The study is the largest to date on the use of bevacizumab in BRVO. The most important finding, according to the authors, is the contribution of more evidence to the idea of treating RVO earlier, rather than undertaking three months or more of watchful waiting.

■ OCT in AMD. If OCT measurements of retinal thickness do not always correlate with visual acuity, then what do these measurements mean for visual prognosis? An article in the October 2011 issue of Graefe's Archive of Clinical and Experimental Ophthalmology, written by a research team in Japan, sought to answer this question.

Twenty eyes with wet AMD received an intravitreal injection of ranibizumab and were divided into two groups: one of treatment-naïve eyes and the other receiving repeated ranibizumab injections. At baseline and at one month, the authors gathered data on BCVA and several OCT readings, including foveal thickness retinal volume at 3- and 5-mm central circles, RPE elevation, type of fluid collections and type of AMD lesion. Correlations were sought between BCVA and all OCT parameters.

In the treatment-naïve eyes, there was a strong correlation between BCVA and nearly all of the OCT variables measured, as well as the size of fibrovascular lesions and patient age. In the other patient group, only foveal volume was correlated with BCVA. Further statistical analysis demonstrated that fibrovascular lesion size and foveal volume maintained their significant correlation with BCVA in the respective groups.

The authors write that prospective studies enrolling larger groups of patients need to be undertaken before any firm conclusions can be made. Nevertheless, they note the comparative nature of their own study in emphasizing the importance of their data. They also urge that similar studies be undertaken with a wide variety of OCT instruments.

■ Distinguishing wet from dry AMD. If earlier treatment of AMD means better outcomes, then a better method of detection is necessary. Doctors in California and Arizona collaborated on a test of a three-dimensional contrast threshold Amsler grid to distinguish dry AMD from the exudative form of the disease, reporting their findings in the October 2011 issue of the British Journal of Ophthalmology. Ninety eyes (63 with AMD — 34 dry and 29 wet — and 27 controls) were tested, first with the conventional Amsler grid and then with the 3D version. The number and volume of 3D defects were analyzed statistically.

Seventy-four percent of the eyes with dry AMD and 21% of those with wet AMD had no distortions using the conventional Amsler grid. Of these eyes initially without defects detected, 20% of the dry AMD eyes and 100% of the wet AMD eyes exhibited visual field defects using the three-dimensional Amsler grid. Eyes with wet AMD were significantly more likely to display stepped defects and all volumetric indices were greater in wet AMD eyes. The new technology also had 83.9% positive and 90.6% negative predictive value for wet AMD.

Thus, in both forms of AMD, the new 3D Amsler grid had a greater likelihood of detecting visual field defects, and this likelihood was even greater in wet AMD eyes than in eyes with dry AMD. Using qualitative and quantitative criteria, the new grid may be useful in distinguishing dry from wet AMD and, therefore, detecting when conversion from dry to wet has taken place. The grid may also prove useful in quantifying disease severity and measuring treatment outcomes.

■ Vital dye in PDR surgery. The Surgeon's Corner column in the October 2011 Archives of Ophthalmology features a description of the surgical treatment of proliferative diabetic retinopathy with viscosurgery using vital dye, by Jeroni Nadal, MD, and María José Capella, MD, of Barcelona. They argue that vital dye offers advantages compared with traditional viscosurgery because the intensity of the dye color provides an indication of the degree of fibrovascular membrane adherence to the retinal surface.

■ OCT-guided Avastin dosing. As uncertainty persists regarding the optimal dosing for anti-VEGF agents in wet AMD, physicians in Lebanon and Qatar have collaborated on a trial comparing the efficacy of as-needed dosing of bevacizumab to continuous fixedinterval dosing. The American Journal of Ophthalmology published their findings online in mid-October.

One hundred twenty patients with wet AMD were randomized to either fixed-interval dosing (given over four or six weeks) or variable dosing; the presence of recurrence of subretinal fluid on OCT was the principal indicator for retreatment in the as-needed arm. The main outcome measurements of the trial were improvements in best-corrected visual acuity and central retinal thickness at 12 months.

The variable-dosing arm saw a mean improvement of 11 ETDRS letters at 12 months, while the fixedinterval arm resulted in a mean improvement of 9.2 letters. This difference was not statistically significant. Both treatment arms in the trial also resulted in decreases in central retinal thickness: -80.7 µm for the variabledosing arm vs 100.5 µm for the fixeddosing arm, again with the results not statistically significant. The fixed-dosing arm had a mean of 9.5 injections given vs 3.8 in the as-needed arm, and this difference was highly statistically significant.

Clearly, while fixed and variable dosing of bevacizumab were equivalent over one year in terms of improvements in BCVA and reductions of central retinal thickness, the treatment burden for variable dosing was much smaller. Acknowledging that their results are appearing in the shadow of the CATT trial, the authors of this study recommend larger, randomized studies on variable dosing regimens for bevacizumab, as well as continued testing of these factors in ranibizumab.

■ FAF in geographic atrophy. Graefe's Archive for Clinical and Experimental Ophthalmology has published online a study undertaken in Barcelona to assess intra- and interobserver agreement in the evaluation of fundus autofluorescence (FAF) patterns in geographic atrophy secondary to AMD. The authors studied a consecutive series of 69 eyes of 49 patients with geographic atrophy, submitting all patients to FAF and then having four observers evaluate the FAF images.

The results regarding intraobserver agreement were impressive, with the results indicating substantial to almost perfect agreement (a range of 0.51 to 0.83 using kappa coefficients). However, interobserver agreement was not as good, with the results rated as poor to substantial (a range of 0.30 to 0.62).

The use of more simple classifications resulted in better reproducibility. Importantly, comparisons among the four raters seemed to diverge in agreement in part because of the lack of a set of shared criteria for assessment. Different weighting for types of FAF features may have also played a role in the variation.

The authors of the study urge that a uniform set of criteria be adopted for the assessment of FAF images in patients with geographic atrophy secondary to AMD.

■ Open-globe injury study. A recent article published online by Retina investigated the risk factors for no light perception after open-globe injury, hoping to determine whether such factors would be predictors for unfavorable visual outcomes. In a casecontrolled study, 72 eyes with NLP were matched to two control eyes per case, with the injured eyes chosen from the database of the Eye Injury Vitrectomy Study.

The authors found that ciliary body damage, closed funnel retinal detachment and choroidal damage were all independent risk factors for NLP after open globe injury. Sixtyseven of the 72 traumatized eyes had at least one of these three factors. Forty-three of the 72 eyes eventually recovered light perception, and five of them were submitted to vitreoretinal surgery with favorable outcomes.

The authors recommend that trauma surgeons pay greater attention to injuries to the ciliary body and choroid and be aware of close funnel retinal detachment when devising interventions in such patients. Rather than concluding that NLP eyes cannot be saved, vitreoretinal surgery may be a possible treatment that can result in restoration of vision.

■ Faster OCT in development. The departments of bioengineering and ophthalmology at the University of Washington in Seattle have collaborated on a project to develop and test ultrahigh-speed spectral-domain OCT for imaging at 850-nm central wavelength using two high-speed line scan cameras running at 250 kHz. They report their findings online at Biomedical Optics Express.

The new system realizes an imaging speed of 500,000 A-lines per second, an order of magnitude faster than many commercially available SDOCT units and much faster than the earlier time-domain models. With this technology, the authors suggest two scanning protocols for retinal imaging in humans. The first would achieve isotropic dense sampling, enabling three-dimensional imaging within 0.72 seconds for a region 16 mm² in area. The second would scan the retina in a large field of view, capturing 1,200 A-lines over 10 mm².

At this point, motion artifacts are a problem, so the authors suggest using the relatively high reflecting RPE layer as a reference to flatten 3D data. RP