The spectrum of polypoidal choroidal vasculopathy in Caucasians: clinical characteristics and proposal of a classification

Idiopathic polypoidal choroidal vasculopathy (PCV) was first described in 1990 by Yannuzzi et al. as peculiar polypoidal subretinal lesions that were associated with haemorrhagic detachments of the retinal pigment epithelium (RPE) [1]. Later, Yannuzzi suggested that PCV may be a self-contained clinical entity, which involves elevated reddish to orange lesions on fundus examination, dilated inner choroidal vessels, and polypoidal vascular structures beneath a RPE detachment [2].

Recently, controversy has arisen regarding the definition of PCV and its association with neovascular age-related macular degeneration (nAMD) [2‐5]. PCV and nAMD have been found to have similar environmental risk factors and to share molecular and genetic determinants involving the complement pathway [6]. Furthermore, several studies have shown that polypoidal lesions in PCV are often associated with a branching vascular network (BVN) of subretinal neovascularization between the RPE and Bruch’s membrane, indicating that PCV could be a variant of type 1 neovascularization of any origin [4, 7]. Patients with type 1 (sub-RPE) subretinal neovascularization can have an aneurysmal dilation (the ‘polyp’), often at the edge of the neovascularization. This aneurysmal dilation is also known with the term PCV, which strictly speaking is a misnomer because we have learned that neovascular lesions that characterize PCV do not arise directly from the choroid [3]. The aneurysmal dilation or PCV typically shows as a well-circumscribed hyperfluorescent lesion already in the early phase on indocyanine green angiography (ICGA). The non-aneurysmal part of the type 1 neovascular lesion is often referred to as a BVN, which can therefore be present in every macular neovascular condition [3, 4, 6]. In comparison with nAMD without PCV, patients with PCV may show relative resistance to anti-vascular endothelial growth factor (VEGF) treatment [8, 9]. Based on imaging studies, some authors have suggested that the term PCV should be reserved for patients in whom findings characteristic for AMD, such as drusen, pigmentary changes, and geographic atrophy, are absent [7, 10, 11].

PCV has been studied mostly in Asian populations. These studies have shown a relatively high prevalence of PCV of up to 61.6% in Asian patients who presumably had nAMD [10, 12, 13]. However, this prevalence has been shown to be only 8.7% in Caucasian patients with a suspicion of nAMD [14, 15], although a recent study described a prevalence of up to 31.9% [16]. A study in which the occurrence of PCV was compared directly between an Asian and non-Asian nAMD cohort showed a large difference: 48% of Japanese patients and 9% of French nAMD patients had PCV [17]. Despite the fact that drusen have traditionally been considered hallmark and even mandatory lesions for the diagnosis of AMD [18], we have also described a spectrum of nAMD in elderly Caucasian patients without drusen in the fellow eye, in whom 5 patients (10%) were shown to have evidence of PCV in the affected eye [19]. This may indicate that the aetiology of non-drusen-associated nAMD and PCV may be different from drusenoidal neovascular subtypes [19]. Little is known about the risk factors, clinical characteristics, and treatment outcome of PCV in Caucasians, as previously conducted studies had small sample sizes [20‐23].

The purpose of this study is to describe the anatomical characteristics and the clinical course of PCV in the largest Caucasian population published so far. We show that there are different clinical manifestations of PCV in Caucasians, and propose a PCV classification into 3 subtypes, comparing our findings in the present study population to the available data on Caucasian nAMD patients, and to previously published characteristics of Asian PCV patients.

To the best of our knowledge, we describe the largest group of Caucasian PCV patients to date. We found a spectrum of PCV, and propose a classification, based on a previous description of Coscas et al. [7]. In the current Caucasian patient cohort, we were able to distinguish 3 types of PCV: the first and largest group of patients had PCV within the context of typical drusen-associated nAMD, with the presence of ≥ 5 drusen with a diameter of ≥ 63 μm in the affected and fellow eye (type A PCV); the second group comprised patients with PCV in combination with a non-polypoidal type 1 CNV or BVN component, but without drusen as typical signs of AMD (type B PCV); the third group included patients with idiopathic PCV with neither drusen nor signs of non-polypoidal type 1 CNV or BVN (also not in the fellow eye; type C PCV). While drusen-associated PCV type A generally has a normal to thin choroid without pachyvessels, types B and C may also have a pachychoroid in a considerable proportion (but not all) of the cases [7, 9]. However, a remarkably high number of patients who received SD-OCT in our study had a thin choroid, also in cases of type B and C PCV.

Our study reveals additional differences between Caucasian and Asian PCV patients. PCV in our study tends to manifest at a slightly older age (73 years versus 65–73 years in Asians) [12, 13, 32‐35]. A male preponderance (42–83%) is reported in Asians, while in the current cohort the opposite was found, with a female preponderance of 66% [12, 13, 32‐35], which is also in contrast to some previous studies on PCV in Caucasians, in whom men and women were in general evenly represented [8, 36]. PCV accounts for about 22–55% of cases manifesting as nAMD in Asians, compared with 8.7% of Caucasian population [15], but some authors have described a prevalence of PCV in Caucasians up to 31.9%. Polypoidal lesions in Asians are often located in the macular region (78–94%) rather than in the peripapillary region (6–14%), while this difference is less pronounced in Caucasians (75% in the macula and 19% in the peripapillary region). [12, 13, 32, 33, 35] Most remarkably, half of the patients in the current Caucasian PCV study showed clinical similarities with nAMD in Caucasians, including the following: age [37, 38], laterality [33, 38], presence of drusen, and presence of a type 1 neovascularization [39, 40]. This is in line with the increasingly established idea of PCV being a variant manifestation of a type 1 neovascularization within the spectrum of AMD [4, 11, 15], which is also in line with findings from Balaratnasingam et al., who consider PCV/pachychoroid neovasculopathy to be part of the nAMD spectrum [3]. Remarkably, drusen have also been found to be present in 14.7–27% of PCV patients in previous studies [7, 15]. There is ample evidence that many, if not all, cases of PCV actually constitute an aneurysmal type 1 macular neovascularization, that is located between Bruch’s membrane and the RPE, not in the choroid. [3, 4] However, this is still subject of controversy, and a recent expert panel effort could not reach consensus on the terminology of PCV versus aneurysmal type 1 neovascularization, because of differences in opinion on whether polypoidal structures represent ‘simple’ aneurysmal dilations or more complex structures [5]. Based on our findings on multimodal imaging, we also postulate that most cases of PCV are actually aneurysmal type 1 neovascularization. Still, in several patients in our study we had evidence of a potential connection of these lesions to the underlying choroid through an interrupted Bruch’s membrane.

Traditionally, AMD by definition is associated with some kind of drusen, whether these are ‘traditional’ sub-RPE drusen and/or subretinal drusenoid deposits (also known as reticular pseudodrusen) [18, 41, 42]. However, we have previously shown that there is a spectrum of subretinal neovascularization in the macula in elderly without drusen in the affected eye as well as fellow eye [19]. In the current study on PCV, we had similar observations: 50% of PCV patients did not show drusen in the affected and fellow eye. Remarkably, 88.2% of patients with sufficiently gradable multimodal imaging had a choroidal thickness below 350 μm, without any drusen, and often without pachyvessels. This indicates that it is not uncommon for Caucasians to have PCV with a thin-to-normal choroidal thickness, and no drusen as signs of ‘typical’ AMD. Other studies on PCV in Caucasians have also described an absence of drusen in a considerable proportion of patients [7, 36]. The pathophysiology of these nAMD and PCV subtypes without drusen may therefore be different from drusen-associated AMD. For instance, pachychoroid neovasculopathy and pachychoroid-associated PCV may be pathophysiologically and clinically more close to central serous chorioretinopathy [9, 43]. In our study, only 6 out of 22 patients with OCT imaging of sufficient quality showed pachyvessels. The mean subfoveal choroidal thickness in these patients was thin (245 μm), and only 2 out of the 22 patients with SD-OCT measurements had a pachychoroid of > 350 μm (396 and 420 μm, respectively) [31]. Therefore, we postulate that there is a third form of PCV, which is associated with a relatively thin choroid, and is neither associated with drusenoid AMD, nor with pachychoroid and/or pachyvessels.

All PCV patients, regardless of whether there is a background of drusen or pachychoroid, had in common that they present with a neovascular and/or polypoidal degeneration at an age that is on average well over 70 (73 years in the current study). Hence, it seems rather arbitrary to exclude these forms of age-related neovascular degeneration of the macula without drusen (and without pachychoroid) from the definition of AMD. It is not always possible to assess whether these age-related macular neovascularizations have a background of drusenoid AMD, pachychoroid, or neither of the aforementioned, while virtually all forms require some form of anti-VEGF therapy. Thus, it may be reasonable to consider these 3 subtypes as different clinical manifestations of age-related macular neovascularization, or nAMD, when other causes such as myopic neovascularization and multifocal choroiditis have been excluded. However, we and others have hypothesised that the pathogenesis and genetic background of macular neovascularization associated with drusenoid AMD is different from that of for instance pachychoroid-associated macular neovascularization (pachychoroid neovasculopathy, with or without a polypoidal component) [3, 4, 7, 19]. The clinical subtype of nAMD and PCV could have therapeutic consequences: for instance, patients with PCV have been described to have a higher likelihood of resistance to anti-VEGF treatment in some studies [8, 9].

We have previously found differences in complement levels and genetic risk factors between central serous chorioretinopathy and AMD [44, 45]. In an Asian population, single nucleotide polymorphisms in chromosomal regions of the complement factor H, age-related maculopathy susceptibility 2 (ARMS2), and high-temperature requirement factor A1 genes have been associated both with PCV and nAMD [46]. However, an association between the rs10490924 variant of the ARMS2 gene and PCV in combination with a BVN was recently found, suggesting that PCV in combination with BVN shares genetic features with nAMD, which is in contrast with PCV without BVN [47, 48]. Little is known on genetic associations in Caucasian PCV patients [15]. While there is clinical and genetic evidence of common disease determinants and pathways between PCV and AMD, we hypothesise that these disease risk factors and pathways may diverge considerably between drusen-associated and non-drusen-associated PCV. Drusen have been described to be a hallmark of complement activation, whereas other factors such as a pachychoroid background may be more important in non-drusen-associated PCV [49]. Likewise, there may be significant differences in (patho)genetic background between PCV associated with drusen, PCV with non-polypoidal type 1 CNV or BVN but without drusen, and idiopathic solitary PCV without drusen.

In the present analysis of Caucasian PCV cases, we observed a good anatomical response to treatment. Here, the most frequently performed initial therapy was conventional thermal laser (36%), closely followed by combination therapy of PDT and anti-VEGF injections (33%), and PDT monotherapy (17%). In 51% of patients, the initially performed therapy showed good anatomical recovery (Table 3, Supplementary Table 3). However, our study has several limitations, inherent to the fact that this was a retrospective study that also included follow-up of PCV patients who were initially phenotyped before the advent of novel imaging modalities such as SD-OCT. It should be noted that no firm conclusions on clinical outcome can be drawn from this analysis, given the inclusion bias (conventional laser is only used in extramacular polypoidal lesions, and nowadays performed only rarely), and the retrospective nature of this study. Due to the relatively small number of included patients, possible differences between both treatment regimens and between disease subtypes could not be studied. Moreover, choroidal thickness should be taken into account in more detail in the future, since OCT scans of sufficient quality were only available for a minority of our patients, also because of the fact that the majority of our patients have been included in the pre-(SD-)OCT era. OCT angiography could also provide additional information on the neovascular component of lesions to be studied. However, currently available devices would not be optimal to assess the PCV component, since direction of flow and low flow phenomena cannot be determined reliably [50]. Assessing pachyvessels in our study also proved to be challenging, without both the availability of a clear cut-off value for the definition of pachyvessels, and the difficulty of distinguishing several choroidal layers in patients with a relatively thin choroid [3].

Collectively, these findings support the hypothesis that PCV in Caucasians is phenotypically and presumably pathophysiologically more associated with AMD than in Asians with PCV, in whom a pachychoroid background is more common [7‐9]. In Caucasian PCV, a spectrum of 3 subtypes of PCV can be distinguished (A, B, and C), out of which only type A PCV has a background of typical, drusen-associated AMD. Future prospective studies should aim at unravelling the underlying (patho)genetic background and optimal treatment in these 3 subgroups of PCV in the Caucasian population.