Four-Year Outcomes in Protocol W

May 2023


Kyle Davis Kovacs, MD

Weill Cornell Medicine, New York, NY

Maturi RK, Glassman AR, Josic K, Baker CW, Gerstenblith AT, Jampol LM, Meleth A, Martin DF, Melia M, Punjabi OS, Rofagha S, Salehi-Had H, Stockdale CR, Sun JK; DRCR Retina Network. Four-Year Visual Outcomes in the Protocol W Randomized Trial of Intravitreous Aflibercept for Prevention of Vision-Threatening Complications of Diabetic Retinopathy. JAMA. 2023 Feb 7;329(5):376-385. doi: 10.1001/jama.2022.25029. Erratum in: JAMA. 2023 Mar 28;329(12):1034. PMID: 36749332.


Anti-vascular endothelial growth factor intravitreal (VEGF) injections have well demonstrated clinical efficacy in the setting of center-involving diabetic macular edema (CI-DME) and proliferative diabetic disease (PDR). However, more recent investigations have sought to delineate the implications of beginning injection treatments in the setting of non-proliferative disease (NPDR) without CI-DME. The question of whether to treat diabetic retinal disease which is not yet impacting vision has remained: does early intervention ultimately lead to better visual outcomes? Fortunately, the Diabetic Retinopathy Clinical Research (DRCR) Network designed a prospective trial, Protocol W, to compare prophylactic anti-VEGF treatment with a “just-in-time” approach where treatment was initiated only after vision-threatening complications occurred.


Protocol W was designed as a 4-year prospective, randomized, masked, multicenter study by the DRCR Retina Network assessing the role of intravitreal aflibercept in eyes with moderate-severe NPDR without CI-DME. The study had anatomic endpoints at year 2 including rates of CI-DME and the progression to PDR, as well as rates of vision-threatening disease progression, and a visual outcome endpoint at 4 years. Patients with either type 1 or type 2 diabetes were included and were randomized to receive either aflibercept or sham injections at baseline and months 1, 2, 4, 8, 16, and 20 during the first 2 years of the study. After 2 years the patients were followed and given either aflibercept or sham injection every 16 weeks, with deferral of the injection if the diabetic retinopathy severity score was mild NPDR or better. Either arm received aflibercept with increases in CI-DME (either 10% increase anatomically or a functional decrease in vision) or the development of high-risk PDR.


Prior Findings: 2-year Data

399 eyes from 328 patients were enrolled in the study, with 1:1 distribution in randomization. The 2-year data previously reported a significantly lower probability of developing CI-DME or PDR (16% aflibercept versus 43% sham). This was broken down to reveal lower rates of CI-DME (4.1% aflibercept versus 14.8% sham) and lower rates of conversion to PDR (13.5% aflibercept versus 33.2% sham). On average, patients in the treatment arm received 8 injections over 2 years, with 19.2% of patients in the sham arm receiving injections (and of those who did received 5.7 injections over 2 years). There was no significant difference in change in visual acuity between groups, nor a difference in the percentage of eyes losing 10 or more letters at 2 years.


4-year Visual Outcomes

Excluding deaths, 271/366 (74.0%) of eyes completed the study and 300/374 (80.2%) either completed the study or met the primary outcome. Treatment effect on visual acuity was not significantly different between any groups or preplanned subgroups. Visual acuity decreased 2.7 letters from baseline in the aflibercept groups compared with 2.4 letters in the sham group. 13.1% of the aflibercept group lost 10 letters or more compared with 12.7% in the sham group.

Anatomically at 4-years, the cumulative incidence of PDR or CI-DME with vision loss was 33.9% in the aflibercept group versus 56.9% in the sham group, which was significantly different (though notably less than the difference observed at the 2-year timepoint). PDR developed in 27.9% in the aflibercept group versus 49.0% in the sham group, and CI-DME occurred in 11.3% of the aflibercept group and 19.1% in the sham group, both of which were significant.

Those assigned to the aflibercept group received a mean of 13.0 injections at 4-years, while the sham group saw 40.3% of eyes have aflibercept initiated with a mean of 8.7 injections of that group. There was no significant difference in cardiovascular or cerebrovascular adverse event rates between the cohorts.


Limitations and Conclusions

The authors note 4-year retention of 75%, despite closely monitoring study participants and a proactive approach to facilitating visit ease. Their practices do not necessarily reflect real-world practices, where follow-up would undoubtedly be far lower. As the authors also note, the regression of some of the anatomic benefit seen in the 2-year results by the end of year 4, which could be related to the change in dosing scheme in these years. As the authors note, it is possible that patients were being under dosed at a 16-week interval. It is also possible that the lack of significant effect may also be due to the high efficacy of aflibercept in rapidly controlling disease, which makes rescue or ‘just-in-time’ therapy a reasonable approach assuming adequate patient follow-up.

In conclusion, while early/prophylactic treatment with aflibercept improved anatomic outcomes in the 2-year data, this did not correlate with visual functional benefit at the final 4-year endpoint for Protocol W. As such, the authors conclude that aflibercept use in such a preventative strategy may not be warranted.


Obtain CME credit: