The role of anti-VEGF in the management of proliferative diabetic retinopathy and diabetic vitreous haemorrhage

This post will explore the role of anti-VEGF injections in the management of proliferative diabetic retinopathy. It will cover the effect of anti-VEGF on diabetic retinal neovascularization and vitreous haemorrhage. It will also address the common concerns regarding the use of anti-VEGF in proliferative disease, including the risk of tractional retinal detachment (crunch) and the danger in loss to follow up.

Efficacy of anti-VEGF injections in proliferative diabetic retinopathy

The CLARITY trial and the DRCR.net protocol S study are the two seminal studies on the effect of anti-VEGF injections in the management of proliferative diabetic retinopathy.

CLARITY trial

This is a phase 2b, non-inferiority trial comparing intravitreal aflibercept (eylea) to panretinal photocoagulation (PRP) in the management of proliferative diabetic retinopathy. Patients with type 1 or 2 diabetes, aged 18 years or older, with clinical evidence of previously untreated proliferative diabetic retinopathy or persistent retinal neovascularisation after initial PRP requiring additional PRP (ie, previously treated) were included in this study. They were randomized into two groups

-          2mg / 0.05ml aflibercept every 4 weeks for 3 injections then as required (112 patients)

-          Standard PRP treatment (109 patients)

The primary outcome was visual acuity at 52 weeks.

This study demonstrated that at 52 weeks, there was a significant difference in visual acuity with the aflibercept group being superior to the PRP group by 4 letters (p < 0.0001). 5% of the aflibercept group had 10 or more letters of worsening comparing to 15% of the PRP group (p < 0.009). The median number of injections required in the aflibercept group was 4 over 52 weeks. The aflibercept group had less vitreous haemorrhage (9% vs 18% in PRP group, p < 0.034). Macular thickness and volume significantly increased in the PRP group compared with the aflibercept group. 89% of the aflibercept group compared to 71% of the PRP group had no macula oedema by week 52. Furthermore, the Esterman visual field score was worse in the PRP group comparing to the aflibercept group.

A number of further analyses of the CLARITY study have been published. One analysis showed that eyes treated with aflibercept showed higher rates of regression of NVE compared with those treated with PRP (96.2% vs 78%, p = .01), but no difference was found for NVD (64.7% vs 46.2%, p = .23). This implies that there can be persistent NVE and NVD despite adequate laser, and injection therapy is likely to be of benefit to induce further regression in NV, with its effect more prominent for NVE.

Another analysis showed that the median baseline area of neovascularization decreased from 0.98 disc area (DA) to 0.68 DA in the PRP group and from 0.70 DA to 0 DA in the aflibercept group at week 12 (p = 0.019). At week 52, this measured 0.24 DA in the PRP group and 0 DA in the aflibercept group (p = 0.45). However, there was no improvements in retinal arterio-venous oximetry differences or area of retinal non-perfusion in either group at week 52. This shows that although treatment with aflibercept is able to achieve an earlier and complete regression of neovascularization, the underlying ischaemia is not modified by aflibercept therapy and ongoing treatment is likely required.

DRCR.net protocol S study

This study compares intravitreal ranibizumab (lucentis) therapy to PRP for the treatment of proliferative diabetic retinopathy (2 year result here and 5 year result here). Study participants were at least 18 years old and had type 1 or type 2 diabetes, at least 1 eye with PDR, no previous PRP, and a best corrected visual acuity letter score of 6/96 or better. Patients were randomized into two groups

-          0.5mg intravitreal ranibizumab every 4 weeks for 3 injections, then as required (191 eyes)

-          Standard PRP treatment (203 eyes)

The primary outcome was mean visual acuity change in 2 years.

The results of this study are very similar to those in the CLARITY study. By 2 years, mean visual acuity change in the ranibizumab group was +2.8 letters comparing to +0.2 letters in the PRP group (p < 0.001 for non-inferiority). Peripheral visual field was significant worse in the PRP group (-422 dB vs -23dB in the ranibizumab group, p < 0.001). Eyes in the PRP group were more likely to require vitrectomies (15% vs 4%, p < 0.001). There was a trend of higher retinal detachment rate in the PRP group as well, though this did not quite reach statistical significance (5% vs 8% for tractional retinal detachment, and 6% vs 10% for any retinal detachment, p = 0.09). As expected, there was less diabetic macula oedema in the ranibizumab group (cumulative risk 9% vs 28%, p < 0.001).

In terms of injection load, eyes in the ranibizumab group without diabetic macula oedema received a median of 7 injections in the first year, and a median of 10 injections over the entire 2 year period. For the eyes with diabetic macula oedema at baseline, the median number of injections was 9 in the first year and 14 over the entire 2 year period.

However, by 5 year, despite this being a well-conducted clinical trial, only 66% of patients completed the 5 year review. The visual acuity change was similar between the two groups (3.4 letters in the ranibizumab group vs 3.0 letters in the PRP group). There was significant further deterioration in the visual field result in the ranibizumab group. The mean change in cumulative visual field total point score was −330 vs −527 dB in the ranibizumab and PRP groups respectively.

DRCR.net protocol AB study for diabetic vitreous haemorrhage

The DRCR.net protocol AB study addressed the use of aflibercept, as compared to early vitrectomy, in the treatment of vitreous haemorrhage due to PDR. Patients with type 1 or 2 diabetes with PDR and vitreous haemorrhage causing vision loss (visual acuity 6/9.6 or worse) were recruited. They were randomized into two groups.

-          Vitrectomy with PRP within 2 weeks (104 eyes)

-          Intravitreal aflibercept injections every 4 weeks for 4 injections, then as required (100 eyes)

The primary outcome was mean visual acuity score over 24 weeks.

The results showed a trend of better mean visual acuity score in the vitrectomy group over the aflibercept group (63 lettes vs 59 letters, p = 0.06) but this fell just outside of statistical significance. However, for patients with a baseline visual acuity of 6/240 or worse, there was a clear benefit with early vitrectomy over aflibercept at 24 weeks (16.7 letter difference, 95% confidence interval outside null). There was no difference in visual acuity at 2 years between the two groups. As expected, recovery of vision was quicker in the vitrectomy group (62.3 letters in the vitrectomy group and 52.6 letters in the aflibercept group at 4 weeks, p = 0.003). Furthermore, there were significant amount of cross-over between the two groups (32-33%).

It is of concern that if there is a fundus obscuring vitreous haemorrhage and aflibercept was given, tractional retinal detachment could develop undetected. In this study, there appears to be a higher risk of tractional retinal detachment in the aflibercept group (23% vs 14% in the PRP group). 5.7% of the PRP group eyes and 15% of the aflibercept group eyes had macula threatening or involving tractional retinal detachment. However no statistical analysis was given for this figure, though a chi-square test performed by myself showed the chi-square statistic as 6.0979 with the p-value as 0.013534, suggesting this difference is statistically significant.

Lack of reperfusion and loss to follow up

Most practicing retina specialists would have noticed that the risk of loss to follow up is high in patients with diabetes. The loss to follow up rate of 33% over 5 years in DRCR.net protocol S study confirms this. This is concerning if the proliferative disease is treated only with anti-VEGF, as there does not appear to be restoration of reperfusion and these patients remain at risk of proliferative disease once anti-VEGF treatment is interrupted (see discussion on CLARITY study above, further evidence RECOVERY study and here).

A study looked at the outcome of patients with proliferative diabetic retinopathy treated with anti-VEGF or PRP, and whom had been lost to follow up for 6 months or more. In those treated with PRP, although there was some worsening in visual acuity the visit immediately after lost to follow up as compared to baseline, there was no difference in visual acuity between the final visit of the study and the baseline visit. However, patients treated with anti-VEGF only had significantly worsening of visual acuity both at the visit immediately after lost to follow up and at the final visit (mean acuity at final visit 0.92 logMAR, i.e. slightly worse than 6/48). The anti-VEGF group also had a much higher risk of tractional retinal detachment (10 out of 30 eyes in the anti-VEGF group vs 1 out of 46 eyes in the PRP group, p < 0.001) and neovascular glaucoma (4 eyes vs 0 eyes, p = 0.02).

Tractional retinal detachment risks with anti-VEGF

Many ophthalmologists are concerned about the risk of tractional retinal detachment with anti-VEGF therapy for PDR. However, one must realise that tractional retinal detachment can develop following both PRP treatment and anti-VEGF treatment for PDR, and it is reassuring that in DRCR.net protocol S study, the risk of tractional retinal detachment is similar between the two groups. On a pooled analysis of 5 DRCR.net trials (Protocols I, J, N, S, and T), the 1-year cumulative probability of tractional retinal detachment was 6.8% in control-group eyes and 4.8% in anti-VEGF group eyes (p = 0.86). This study concluded that ‘These findings do not support the hypothesis that anti-VEGF therapy for diabetic macular edema or PDR increases the risk of TRD among eyes with PDR similar to those enrolled in five DRCR Retina Network protocols for which prompt vitrectomy was not planned’.

One reason for the concern is that the development of tractional retinal detachment is rapid following anti-VEGF therapy (the mean time to development or progression is 13 days, with a range of 3 to 31 days), whilst tractional retinal detachment tends to be less rapidly progressing following PRP. For patients with PDR treated with PRP, the treating ophthalmologist will likely detect early tractional changes by the second session of PRP and management modified accordingly. However, if a patient was given a single anti-VEGF injections for PDR and left for 4 or more weeks without review, this patient can develop significant tractional retinal detachment undetected with poor outcome.

Take home message

Anti-VEGF therapy for PDR appears to be superior to PRP within the first 2 years of treatment, with better visual field outcome, less diabetic macular oedema and less requirement for vitrectomy. Therefore, it is reasonable to offer a course of anti-VEGF therapy for patients presenting with PDR with a good view to the fundus as their initial therapy.

There is a risk of early tractional retinal detachment following anti-VEGF therapy for PDR, so all patients with PDR will need to be reviewed within 1-2 weeks following their first anti-VEGF injection. One should not perform anti-VEGF therapy in patients without any fundal view due to media opacity (e.g. vitreous haemorrhage), unless surgical intervention is planned within 1-2 weeks to visualise the retina.

Given the need for ongoing injections for patients with PDR treated only with anti-VEGF due to retinal non-perfusion, and the risk of severe permanent vision loss if these patients are lost to follow up, one should offer PRP therapy in all patients with PDR initially treated with anti-VEGF. This can be performed once the retina is stabilised (usually after the initial course of 3-4 anti-VEGF injections). Patients must be warned of the risk of severe and permanent vision loss in the case of loss to follow up if they decline PRP.