Real-World Anti-VEGF DME Data Differ from Randomized Results

Improvements in visual acuity (VA) were less pronounced with anti-VEGF therapy in the "real-world" setting compared with randomized controlled trials (RCTs) for patients with diabetic macular edema (DMA), according to a presentation at the 2017 American Society of Retina Specialists Annual Meeting.

Differences in outcomes were primarily driven by decreased treatment intensity and worse patient characteristics in the real world versus RCTs. Moreover, the review found that those with better VAs did not benefit as substantially from anti-VEGF therapy and that a clinically relevant difference between antibodies was not observed.

"Patients in the real world getting anti-VEGF therapy for DME aren't doing as well as they are in trials. Part of this is under treatment and the other part is comorbidities," said Thomas A. Ciulla, MD, MBA, volunteer clinical professor of ophthalmology, Indiana University School of Medicine. "These real-world studies become more and more important as our healthcare system evolves."

As a comparison, Ciulla looked to the 660-patient DRCR protocol T trial. The mean VA at baseline was 64.8 letters, which was higher than seen in the real-world data. Patients received a median of 9 injections in the aflibercept (Eylea) arm and 10 injections for bevacizumab (Avastin) and ranibizumab (Lucentis).

Overall, the mean improvement in VA at 12 months was 13.3 letters with aflibercept, 9.7 with bevacizumab, and 11.2 with ranibizumab. After 24 months, the mean VA improvement was 12.8 letters in the aflibercept group, 10 with bevacizumab, and 12.3 for ranibizumab.

Data for the real-world analysis, which included 5872 patients, came from electronic medical records from retina specialists across the United States via the Vestrum Health Retina Database. Patients were divided into 3 mutually exclusive cohorts based on the duration of follow-up (6-, 12-, and 24 months). All patients had received at least 3 monthly injections of an anti-VEGF therapy and macular and pan-retinal laser treatment rates were similar in each group.

The study did not have strict exclusive criteria, as would be seen in a RCT. "These patients have comorbidities that aren't allowed in a clinical trial, such as prior laser or diabetes that isn't quite as well controlled," Ciulla noted.

There were 2503 patients at an initial age of 62 years in the 6-month group. The mean number of injections was 5.2 and the mean baseline VA was 57.5, with a 6.1 mean letter improvement across treatments. Although a statistically significant difference was not seen between each group, the mean improvement in VA was 6.9 letters with aflibercept, 4.9 with ranibizumab, and 6.1 with bevacizumab.

In the 12-month cohort (n = 2315), the initial age of patients was 62 years and the mean number of injections was 8.5. The mean baseline VA was 57.4 letters and there was a 6.3 mean letter improvement across treatments. VA improvement was 7.4, 6.1, and 6.0 letters in the aflibercept, ranibizumab, and bevacizumab groups, respectively.

"In trials, patients get about 10 injections. In the real-world, physicians are administering 8 on average, so there is slight under treatment," said Ciulla. "Under treatment is very common in this space. There's a real treatment burden. Patients coming back every month or 4 to 6 weeks is a real burden in this population."

In the 24-month cohort (n = 1041), the initial age was 61 years and the mean number of injections was 14.8. There were fewer than 50 patients receiving aflibercept in this cohort compared with 487 for ranibizumab and 554 for bevacizumab. The mean VA improvement was 6.2 letters with ranibizumab and 6.5 letters for bevacizumab.

"Whether it was Avastin, Eylea, or Lucentis, results were very similar," said Ciulla. "The anti-VEGF agent may not matter as much. We think the under treatment may supersede the VEGF agent used. The VEGF agent may have a role but we didn't see in this study where patients may be under treated."

When looking deeper at findings from the 12-month cohort, those with a VA of 20/201 or worse or a range of 20/71 to 20/200 had the most significant improvements with anti-VEGF therapy. Those with VAs of 20/41 to 20/70 had letter improvements of 3.0, 4.5, and 4.3 with bevacizumab, ranibizumab, and aflibercept, respectively.

Interestingly, Ciulla noted, those with VA of 20/40 or better did not show improvements in vision following treatment with anti-VEGF therapy. At 12 months, the VAs declined by 2.3, 1.9, and 2.0 letters in the bevacizumab, ranibizumab, and aflibercept, respectively.

"Patients who start with good vision, 20/40 or better, actually lose 2 letters of vision at 1 year. They do poorly," Ciulla explained. "We tend not to follow patients who are doing well as closely. If we have patients who start with good vision we may need to follow them as closely as other patients."

Ciulla TA, Williams DF. “Real-World” Outcomes for Anti-VEGF Therapy of Diabetic Macular Edema in the United States. Presented at: 35th Annual Meeting of the American Society of Retina Specialists; August 11-15, 2017; Boston, Massachusetts.