Atypical retinal pigment epithelial defects with retained photoreceptor layers: a so far disregarded finding in age related macular degeneration

We here describe a so far disregarded presentation of AMD with atypical RPE-defects not attributable to GA or RPE-tears with preserved overlying photoreceptor-layers. The majority had SRF which persisted or resolved independent to intravitreal treatment. This fact strengthens the assumption that not a CNV was the underlying cause of SRF but rather the local loss of the RPE and therefore the inability to pump ions and fluid out of the subretinal space. RPE-defect enlargement was associated with persistent SRF and lack of migration of subretinal RPE/drusenoid material, whereas migration of subretinal RPE/drusenoid material correlated with resolution/absence of SRF.

RPE-tears in the classical sense were first reported by Hoskin et al. and are characteristically found in patients with fibrovascular PED [14, 15]. They can occur spontaneously or may be related to various treatments [13, 15‐18]. Contractile and hydrostatic forces inherent within the vascularized PED are described to be the most likely factors leading to the evolution of a RPE-tear, which usually involves the photoreceptor layers [15]. In general, RPE-tears in exudative AMD have a poor visual prognosis. Despite this fact, some cases with subfoveal RPE-tears and preserved visual function and retained photoreceptor layers were reported [11, 12, 19]. But our cases lacked characteristics of typical RPE-tears, which are typically found near the base of the fibrovascular PED with retracted RPE. There was also no evidence of exudative AMD, lacking the causative tractional forces of the CNV membrane. Further reasonable visual function and retained photoreceptors overlying the RPE-defect were evident and could be preserved over a long period of time.

GA may be the other well-known condition in AMD presenting with large RPE-defects. In contrast to our findings, RPE atrophy is associated with photoreceptor loss in GA. In fact, the loss of photoreceptors, in particular the loss of the rods and cone outer segments precedes RPE atrophy in GA and the area of photoreceptor loss may be much larger than the area of RPE atrophy [20‐23]. Typical morphological patterns such as subsidence and thinning of the outer plexiform layer and loss of the ELM, the EZ- and the interdigitation zone is associated with RPE atrophy in GA [24].

The first report describing two patients with similar atypical RPE-defects was published 1987 [7]. These so called “blow-outs” of the RPE demonstrated intense leakage through the defects on FA. The resulting vision was 20/30 and 20/70, respectively. Similar to our cases and in contrast to characteristic RPE-tears, the RPE-defects were rather found at the dome of the detachment and not along the margins of the PED [7]. A case report published 1990 found two patients with spontaneous RPE-tears in long-standing serous RPE-detachments without evidence of a CNV who could keep a visual acuity of at least 20/40 over an observation period of 3 years after the rip had occurred [8]. During follow up, pigment migration along the tear edges was noted [8]. A RPE-tear in a patient with a mixed serous/drusenoid PED with intact photoreceptor layers and a visual acuity of 20/25 during the course of observation has recently been described [9]. Also in this patient, no definite evidence of a CNV was found [9]. Another report published 2014 showed spontaneous bilateral RPE-defects. After 2.5 years of follow-up the overlying photoreceptor layers were still intact and visual acuity were 6/9 and 6/6, respectively [25]. These examples all describe atypical, subfoveal RPE-defects with long-term survival of the overlying photoreceptors in patients with AMD and may depict the same disease process described here. A recent report described so called RPE-apertures [10]. Similar to our cases, well circumscribed discontinuity of the RPE without retracted RPE was found and avascular PEDs preceded respected lesions [10]. They enlarged homogeneously over time and focal hyperautofluoscent lesions were found prior to the onset [10]. Given the fact that 4 of our cases showed a drusenoid PED preceding the RPE-defect and hyperautofluosescent lesions, it is very likely that the RPE-apertures described, resemble the same disease process. However, none of the previously reported RPE apertures revealed migration of subretinal material.

Although 4 of our cases (case 2, 4 and 5, 6) had hyperautofluorescent lesions, an AMD related genesis, in particular a drusenoid PED rather than an underlying CSC or pattern dystrophy seems most likely as the age ranged between 71 and 87 years and they had large, soft drusen in the study as well as in the fellow eyes. However, a pattern dystrophy, CSC, Morbus Stargardt or a former Morbus Best may also be associated with such RPE alterations and cannot be excluded as possible underlying diseases.

Apparently photoreceptors seem capable to survive without normal RPE. A previous report already speculated that RPE may not obligatory be essential for the survival of photoreceptor-cells [8]. In AMD, rods seem more vulnerable, while cones seem to survive for longer periods also in the absence of outer or inner segments [23, 26, 27]. The reason for the earlier involvement of rods relative to cones in AMD is unknown, however it was speculated that the localized deficiency of retinoid, which is essential for photoreceptor survival, may be an underlying cause leading to impaired retinoid transfer from blood to the RPE due to debris accumulation [22]. Another explanation may be found in a study performed 2002, which demonstrated that within the neurosensory retina of non-human vertebrates there is an RPE-independent visual cycle for cone photopigment regeneration and day-light vision [28]. In accordance with those findings our long-term observation of preserved photoreceptor layers and reasonable visual function might uphold the hypothesis that (at least) the cone photopigment regeneration can occur independently from the RPE. The same previous study however also speculated that photoreceptors may be still apposed to some remaining RPE cells [8]. Unfortunately even in the OCT era, this question cannot be conclusively determined; additional image modalities such as polarization sensitive OCT or adaptive optics OCT may be able to further elaborate this question.

Beside the proposed mechanisms, other potential mode of actions may contribute to the long-term survival: Four of our cases showed migration of subretinal RPE/drusenoid material. The migration and recovery of pigmented tissue was initially noted and described 1988 in 4 patients after classic RPE-tears and it was speculated that it might resemble a repair mechanism [5] Bressler et al. also noticed, migration of pigmented material in both reported patients [8]. Caramoy et al. observed 2012 that patients with RPE-tears revealed migration and repopulation of RPE as shown in FAF and SD-OCT. However, this migrated RPE did not form a functional RPE-layer. They concluded that a small RPE-defect may be repaired by cell proliferation, but that respective RPE-cell proliferation is not sufficient in covering larger defects [29]. Mukai et al. recently described potential repair mechanisms in patients after RPE-tears and postulated 2 main modes of actions: The first mechanism included persistent SRF after the RPE-tear leading to thickened proliferative tissue at the affected area. The second mechanism was described to present with early and complete resolution of the SRF, the outer retina directly being attached to the Bruch’s membrane and attenuation of the normal hyperreflective band attributable to “normal” RPE [12]. The RPE/drusenoid material found in this report was also postulated to represent RPE-regeneration [12]. This assumption is strengthened by the fact that animal models have proven RPE-regeneration [8, 12, 30]. Accordingly, our cases showed SRF resolution and 2 cases had concomitant visual function improvement accompanied with the ingrowth of RPE/drusenoid material above the Bruch’s membrane, whereas in cases of persistent SRF no ingrowth was noted. Further, in cases of persistent SRF and no RPE ingrowth, enlargement of the RPE-defects has been observed. As the resolution of SRF was associated with the presence and ingrowth of the subretinal material it is likely that the tissue resembles RPE-regeneration, as previously suggested by abovementioned reports with the capability of normal RPE to actively pump ions and fluid out of the subretinal space. RPE regeneration may also explain the survival of the photoreceptors in patients with respective ingrowth.

Limitation of this study includes its retrospective design and the small number of patients which is mainly due to the fact that this finding is rather rare in contrast to GA.