Rhegmatogenous retinal detachment following intravitreal ocriplasmin

We report the presentation and characteristics of eight cases of RRD following OCP, representing the largest series of cases to date. We are unable to give an incidence rate of RRD after OCP, as we do not know the true denominator of cases treated by members during the study period nor the completeness of case reporting to us. As well as the 0.4 % rate in the phase 3 studies of OCP [1], a single case (representing 1.7 %) was reported in the phase I MIVI study day 1 post-injection [3]. There have been reports of three cases of RRD [6‐8] out of 213 [6‐14] treated cases (1.4 %) in recent retrospective institutional studies, two of which occurred at day 6 and one at week 6 post-OCP injection.

The mean age of 72 years in the MIVI trials was similar to our cases. The majority of our cases were phakic (88 %) as opposed to 63 % in the MIVI OCP group. One of our cases had high myopia, which would have been excluded from the phase 3 MIVI trials, which had a cut off point of −8 dioptres of myopia. Two of the patients also had lattice degeneration. There have also been reports of two other patients with lattice degeneration who developed RRD after OCP in recent institutional reviews [7, 8]. This was not listed as an exclusion criterion in the phase II and III MIVI trials [1, 15] report but was in the phase I trials [3]. We would suggest that myopia greater than 8 dioptres and lattice degeneration should be considered as contraindications to OCP treatment. Whether treating lattice preoperatively would reduce the risk of RRD is unknown.

The median time to onset of RRD was 16 days after OCP injection, with 37.5 % presenting within the first 7 days and the rest presenting on days 14, 18, 20, 54, and 131. Photopsia and floaters are common symptoms in the first few days after OCP injection, and consistent with the action of the drug [5]. One of our cases presented whilst still symptomatic with photopsia in the immediate postoperative period, while in four cases there was recurrence of symptoms after the postoperative symptoms had settled. Three of our cases, however, were diagnosed whilst asymptomatic at routine review appointments. Our cases show that RRD post-OCP can be asymptomatic, whilst others with symptoms may be difficult to distinguish from typical post OCP symptoms. We recommend that scheduled appointments at the peak time of RRD in the first few postoperative weeks, as well as careful peripheral retinal examination at all visits, is important even if asymptomatic.

There has been some speculation that breaks occurring soon after ocriplasmin may have been a result of the physical action of the intravitreal OCP injection in yielding vitreoretinal traction [6]. This is supported by the fact that during the repair of the retinal tear and detachment in the MIVI I trials, the vitreous was found still firmly adherent to the retina [3]. In our series, 3/7 presented with RRD within 7 days post-injection; however, five cases presented 14–130 days post-OCP. Given the timing of our five cases and one reported case in the literature at 6 weeks post-OCP injection [7], and the fact that VMT had released in all cases, makes OCP’s pharmacological action the likely culprit, rather than the physical action of the intravitreal injection. It is important to note that although 26.5 % of OCP-injected eyes had vitreo-foveal separation in the MIVI TRUST trials at day 28, complete PVD was noted in only 13.4 % of eyes at the same time point. It is possible, therefore, that more peripheral vitreous separation occurs later in OCP-treated eyes, and the true incidence of tear and RRD formation may ultimately be higher than during the limited follow-up period carried out during the phase 3 studies. We found complete PVD at the time of surgery in 88 % of cases, and it’s clearly important to maintain vigilance for late complete vitreous separation and consequent retinal tears.

Although OCP can induce a clean plane of vitreoretinal separation [2, 16, 17] it does not appear to do this uniformly across the whole VR interface. In a study characterising the effect of OCP on the vitreoretinal interface in an ex-vivo porcine model, the vitreolytic effect was not found to be homogenous throughout the eye [18]. Retinal areas proximal to the site of injection in the mid vitreous frequently appeared to be more devoid of vitreous elements in comparison to more anterior areas. The authors concluded that although this may in part reflect a difference in chemical structure of the vitreous adjacent to the vitreous base, it might reflect a lack of exposure to the enzyme that has to diffuse further to this particular location from its site of injection. This raises the question as to whether other factors, such as variable diffusion through the vitreous and variations in drug preparation and injection technique, might also influence the effect of the drug with its short half-life [19]. The distribution of retinal breaks would also concur with this, in that the breaks were not concentrated in the superotemporal quadrant as found in spontaneous PVD-related tears and RRD [20]. Spontaneous PVD is thought to occur perifoveally and extends first superiorly [21]. Surgically induced vitreous separation results in a different retinal break distribution to spontaneous PVD, with a more evenly distributed tear distribution or possibly a higher incidence of tears inferiorly [22‐24]. It is debated whether intravitreal injection of OCP should be performed with more pressure or deeper into the vitreous cavity than vascular endothelial growth factor inhibitors [12]. In our series, OCP was administered according to the label, without interruption of the freezing cycle and using a standard 30-gauge needle. The depth and site of injection was not recorded, but no complications immediately following injection were noted. It is possible that OCP-mediated vitreous liquefaction can result in an incomplete PVD leaving residual vitreous cortex attached to the retina peripherally, ultimately contributing to late retinal break formation [25].

We asked that surgeons report all cases of retinal tears and RRD to us, but interestingly there were no cases of tears without RRD reported. This is unusual, as it is known that not all tears associated with spontaneous vitreous separation progress to RRD [26]. There have been many reports of subretinal fluid accumulation at the fovea after OCP, and Willekens et al. reported this in 37 % of cases treated including peripapillary SRF in some cases [7]. Increase in MH base diameter has also been widely reported [5, 14, 27]. The aetiology of this is unknown. There is plausible evidence that OCP causes weakening of retinal adhesion by degrading laminin and possibly other proteins in the outer retinal layers including the interphotoreceptor matrix [28], which is known to mediate retinal pigment epithelial-photoreceptor adhesion in primate eyes [29]. This mechanism might also mean that tear formation has a higher incidence of progression to RRD after OCP than after spontaneous PVD. Interestingly, in Pierson syndrome, which is caused by mutations in the laminin b2 gene, the incidence of RRD is higher [30]. This provides further evidence that laminin function is important for maintaining retinal attachment, and perhaps tears induced after PVD from OCP are more prone to progress to RRD.

There are several weaknesses to this study, in particular the absence of a clear denominator number for the number of cases treated in total with OCP that subsequently resulted in the eight retinal detachments. We therefore do not know the true incidence of RRD after OCP in our population. The study was retrospective, and the follow-up after injection was at the discretion of the treating specialist. There is therefore some uncertainty of the timing of retinal detachment in some of the cases.