Opposing pathophysiological concepts in multiple evanescent white dot syndrome: The great MEWDS schism

Abstract

Multiple evanescent white dot syndrome (MEWDS) is a benign posterior uveitis involving the outer retina-choriocapillaris complex. The clinical picture, since its first description in 1984, is well-established. Classified at the benign end of the choriocapillaritis spectrum, including, among others, idiopathic multifocal choroiditis, serpiginous choroiditis, and acute posterior multifocal pigment epitheliopathy, a challenge to this classical choriocapillaritis classification of the disease was attempted, attributing the primary lesion process to a photoreceptoritis. The aim of this perspective and review article was to gather a sufficient body of evidence for a credible pathophysiological explanation of MEWDS. We reviewed the literature and integrated the results of some of our own studies on MEWDS. The crucial role of indocyanine green angiography (ICGA) and its correct interpretation is highlighted, giving clear indications that uptake failure of ICG by the retinal pigment epithelial (RPE) cannot explain ICGA hypofluorescence. In contrast to what is often claimed, ICGA hypofluorescence is already present in the early angiographic phases in MEWDS, speaking for a perfusion problem rather than the RPE nonuptake thesis. Hypoxic damage principally affects the photoreceptor cell but only minimally the RPE cell in MEWDS, explaining why primary photoreceptoritis has been privileged by some. The principles of optical coherence tomography angiography, being based on the presence of flow, it cannot detect the presence or absence of flow in low-flow end-capillaries and is, therefore, an inappropriate imaging modality for MEWDS. There is a continuum between MEWDS and other choriocapillaritis entities, which are all based on inflammatory choroidal circulatory dysfunctions of diverse severity, including MEWDS. We found a substantial body of evidence indicating that end-choriocapillaris nonperfusion is at the origin of MEWDS with limited secondary hypoxic damage mainly to the vulnerable photoreceptors and functional preservation of the metabolically more resistant RPE cell. As only small end-choriocapillary vessels are involved in MEWDS, spontaneous resolution of the disease occurs mostly without sequels.

INTRODUCTION

Multiple evanescent white dot syndrome (MEWDS) was first described by Lee Jampol et al. in 1984.[1] While the clinical characteristics were clearly determined since the first report, the pathophysiological explanation has been swaying from primary choriocapillaritis to primary retinal pigment epithelial (RPE)/primary photoreceptor disease. From time to other, episodically, articles appear defending one or the other hypothesis. On the occasion of the publication of a recent article by Thibaut Mathis et al. on MEWDS,[2] our aim here was to review the evidence-based arguments available to understand the mechanisms at the origin of MEWDS, by presenting and analysing the solidity of the opposing arguments.

THE CRUCIAL ROLE OF INDOCYANINE GREEN ANGIOGRAPHY AND ITS CORRECT INTERPRETATION

Since the publication in 2016 of the first article reversing the established cause of choriocapillaris-induced photoreceptor damage and attributing the primary lesion process in MEWDS to causes other than choriocapillaris nonperfusion, it looks as if we have entered the era of an “intellectual schism” concerning MEWDS. On one side, some are speaking of photoreceptoritis (a misnomer, as there is no objective proof of an inflammatory process), basing their belief on the argument that indocyanine green angiography (ICGA) hypofluorescence is due to the supposed failure of ICG uptake by dysfunctioning RPE cells, allegedly measuring RPE damage, rather than caused by choriocapillaris nonperfusion.[3] In the latter study, only less than half of the patients had an ICGA performed making it difficult to draw meaningful conclusions.[3] Interpreting ICGA is difficult and needs more than superficial understanding of its imaging mechanism. Since ICGA became available, misinterpretations of this imaging modality were manifold. The first particularity that many retina specialists but also uveitis specialists did not understand was the fact that the large ICG-protein molecular complex with a size of approximately 70,000 Daltons remained indeed within the large choroidal vessels as indicated by the early pioneering research performed by Bob Flower et al.[4] However, these researchers had to limit their analysis of the choroidal circulation to the early phase of angiography, as their instruments had a low sensitivity to capture ICG fluorescence. Thereafter, the notion that ICG remained intravascularly in the choroidal circulation (and the retinal circulation) prevailed and it was difficult to have the explanation accepted that the large ICG-protein molecular complex egressed freely from the fenestrated choriocapillaris and impregnated the choroid during the intermediate and late angiographic phases unless there were areas of vascular closure determining dark areas devoid of ICG. This became an accepted explanation for choriocapillaritis entities, such as acute posterior multifocal placoid pigment epitheliopathy (APMPPE), idiopathic multifocal choroiditis, serpiginous choroiditis or MEWDS, the difference of patterns being explained by the size and flow of the choriocapillaris vessels involved, namely large dark areas of nonperfusion seen both on ICGA and optical coherence tomography angiography (OCT-A) such as in APMPPE and smaller paler areas in case of end-choriocapillary low-flow vessel involvement such as in MEWDS seen on ICGA but not on OCT-A. Indeed, as flow is needed to be “seen” by OCT-A to determine whether there is perfusion or not this imaging method is not able to give this information in low or very low-flow vessels such as in MEWDS. In other words, OCT-A is unfit to analyze endchoriocapillaris low flow circulation that is impaired in MEWDS. Therefore, absence of altered choriocapillaris on OCT-A is not an argument to determine whether the whole choriocapillaris is intact or not in MEWDS.

UPTAKE FAILURE OF INDOCYANINE GREEN BY THE RETINAL PIGMENT EPITHELIAL CANNOT EXPLAIN INDOCYANINE GREEN ANGIOGRAPHY HYPOFLUORESCENCE

In the recent article cited hereabove,[2] the authors performed painstaking work and statistical calculations comparing ICGA hypofluorescence and blue light autofluorescence (BAF) hyperautofluorescence and showed that hypofluorescent ICGA areas were more extended than BAF hyperautofluorescent areas, which may well be the case. Like in many articles putting forward the primary involvement of the RPE and/or the photoreceptors, the well-analyzed data in the article by Mathis and Coll.[2] fail to prove that the RPE is primarily involved but is just a morphological observation not allowing to draw conclusions on the mechanism. For all the articles that believe in the controversial thesis of “alleged nonuptake of ICG” by the RPE explaining the dark areas of ICGA in MEWDS, it is this “dogma” that represents the flaw in the theories primarily attributing hypofluorescence to other causes than choriocapillaris nonperfusion. The primary photoreceptoritis theory is based on one single article that indicates that there is altered uptake of ICG by RPE cells “in vitro,” either decreased or even increased.[5] How strong is the RPE damage in MEWDS supposed to be and what is at the origin of the damage? To what extent can this “in vitro” study be applied to the “in vivo” condition of MEWDS taking into account the fact that the ICG molecule used in the “in vitro” study was the native ICG molecule with a molecular weight of 625 Daltons, whereas “in vivo” ICGA is bound in up to 99% to large proteins forming a large molecular complex. When the primary photoreceptoritis theory was elaborated, one among the alternative conjectures to choriocapillaritis put forward, the authors had a hard time to explain ICGA hypofluorescence in MEWDS in the limited number of cases that had ICGA performed. Hence, they chose to take the part of the article by Chang et al. that said that RPE uptake was diminished, in their “in vitro” study.[5] Such a misinterpretation is explained by the widespread difficulty to correctly interpret ICGA by many. Nevertheless, since the article princeps at the origin of the primary photoreceptoritis theory was published, the circumstances were set for the start of a vivid intellectual confrontation regarding MEWDS pathophysiology.

HYPOFLUORESCENCE IS ALREADY PRESENT IN THE EARLY ANGIOGRAPHIC PHASE IN MULTIPLE EVANESCENT WHITE DOT SYNDROME SPEAKING FOR A PERFUSION PROBLEM INVALIDATING THE RETINAL PIGMENT EPITHELIAL NONUPTAKE THESIS

The authors of the recent article published on comparison of ICGA and BAF,[2] and others, try to explain the difference of mechanisms between MEWDS and APMPPE by claiming that MEWDS is only hypofluorescent in the late angiographic phase, which is absolutely incorrect and invalidates ipso facto the hypothesis of RPE nonuptake of ICG and strongly advocates choriocapillaris hypo or nonperfusion. The origin of the belief that there is no early ICGA hypofluorescence, again, is based on a single article reporting a single case.[6] Moreover, when analysing the images of this article, there is indeed early although faint ICGA hypofluorescence.[6] In all of our 15 MEWDS cases with good quality imaging, ICGA showed not only hypofluorescence in the late angiographic phase (24–28’) but the circulation was already disturbed with hypofluorescent areas in the very early phase as well as in the intermediate phase (8–10’) in all of them [Figure 1]. It is true that hypofluorescence is more pronounced in the late angiographic phase when the contrast between perfused and nonperfused areas increases. We are presently performing a study analyzing quantitatively early, intermediate and late ICGA hypofluorescence, clearly indicating that hypofluorescence is not limited to the late angiographic phase (MS in preparation). Once the obvious evidence that there is indeed also early/intermediate ICGA-hypofluorescence is admitted as shown in Figures 1 and 2, to say that hypofluorescence is due to non-uptake of the ICG molecule by the RPE is an undefendable theory for two reasons. (1) Some time is needed in order that the alleged nonuptake of possibly diseased RPE occurs; therefore, early hypofluorescence can only come from nonperfusion, as impregnation or not of the RPE cells takes some time. Persisting fluorescence 24–48 h after injection can probably be explained by RPE ICG impregnation (2) As already mentioned, unlike in the “in vitro” experiments by Chang et al. that used the native ICG molecule with a molecular weight of 625 Daltons, the molecule in ICGA is not comparable as, after intravenous injection, it is strongly linked, up to 99%, to large proteins constituting a large molecular complex all the more unable to enter quickly and impregnate RPE cells or other cells whether healthy or damaged, leaving only the circulatory hypothesis to explain ICGA hypofluorescent areas during the angiographic phases in MEWDS as in other choriocapilaritis entities. It is true that the early hypofluorescence is fainter in MEWDS than in APMPPE because it involves small end-capillary vessels whereas in APMPPE and other choriocapillaritis conditions larger vessels are involved. If the primary RPE/photoreceptor damage hypothesis should be retained, how can this primary damage be explained? Where does this damage come from; is it toxic or other? This makes little sense and is difficult to explain, when hypoxia due to nonperfusion, like in all other choriocapillaritis entities, makes much more sense, is logical and self-explanatory.

Figure 1.

Blue light autofluorescence, optical coherence tomography (OCT), OCT-angiography (OCT-A), and indocyanine green angiography (ICGA) sequence in a case of multiple evanescent white dot syndrome (MEWDS). BAF (top left and top middle pictures) shows area of hyperautofluorescence corresponding to the loss of photoreceptor outer segments shown on spectral-domain-OCT (bottom picture), this loss of screen giving access to the normal/possibly slightly increased fluorophores in the RPE cells. OCT-A (top right) fails to show choriocapillaris drop out as endcapillaries lack flow that is necessary to be imaged by OCT-A. The ICGA sequence shows areas of perfusion disturbance in early frames (30”) (left frame, middle row, and crimson arrows), distinct presence of hypofluorescence in the intermediate phase (middle frame in middle row) which is more pronounced in the late phase (right frame, middle row), these images clearly indicating that hypofluorescence is not limited to the late phase in MEWDS. BAF: Blue light autofluorescence, OCT-A: Optical coherence tomography angiography, ICGA: Indocyanine green angiography

Figure 2.

Indocyanine green angiography (ICGA) hypofluorescence is present since the early angiographic phases and points toward a circulatory explanation of ICGA hypofluorescence in MEWDS. Perfusion disturbance is already present in early time points (21”) (left). Nonperfused areas are clearly present in the intermediate phase (10’47”) (second from left) and best identified on late phase frames as the contrast between perfused and nonperfused is maximal (angiographic time = 27’43”) (second from right). BAF (right) more precisely shows the diseased areas. ICGA: Indocyanine green angiography, FAF: Fundus autofluorescence

HYPOXIC DAMAGE PRINCIPALLY AFFECTS THE PHOTORECEPTOR CELL BUT ONLY MINIMALLY THE RETINAL PIGMENT EPITHELIAL CELL IN MULTIPLE EVANESCENT WHITE DOT SYNDROME

The photoreceptor cell is the weak element in the RPE-photoreceptor complex because it is metabolically very demanding with high energy needs, making it vulnerable to metabolic stress.[7] It was also shown to be much more sensitive to hypoxic effects.[8] In contrast, the RPE cell is metabolically much more resistant.[9,10] Limited hypoxia-induced effects on RPE cells do not harm it significantly but are sufficient to induce photoreceptor damage.[11] As only small endchoriocapillary vessels are affected in MEWDS, the hypoxic effect is very faint, allowing sufficient RPE metabolic function and no RPE cell loss. On the other hand, more severe choriocapillaritis entities, such as APMPPE, involve larger vessels, causing a more severe hypoxic effect with dysfunction and potential loss of RPE cells. In MEWDS, impairment of RPE function is minimal at most and fails to produce RPE cell death, whereas the vulnerable photoreceptors are maximally impacted by secondary nonperfusion hypoxia, causing loss of outer photoreceptor segments which regenerate in most cases.[12,13]

HOW BLUE LIGHT AUTOFLUORESCENCE IS GENERATED AND HOW IT IS RELATED TO INDOCYANINE GREEN ANGIOGRAPHY HYPOFLUORESCENCE

BAF autofluorescence is generated from the fluorophores, such as lipofuscin, contained in the RPE. (1) Either fluorophores are increased, causing bright hyperautofluorescence in case of massive photoreceptor outer segment digestion by the RPE or (2) the RPE is metabolically impaired, causing accumulation of fluorophores resulting in increased autofluorescence or (3) there is fatal damage and loss of RPE cells, causing decreased or absent autofluorescence. The RPE damage theory, as the initiating step in MEWDS, is very unlikely, as the RPE cell is metabolically much more resistant than the photoreceptor cell.[9,10] Another mechanism of hyperautofluorescence, which is most probably occurring in MEWDS, is caused by the ischemic-induced loss of photoreceptor outer segments that unmasks the physiological RPE autofluorescence.[14] The latter is seen in pure photoreceptor diseases, such as AZOOR, autoimmune retinopathy, or cancer-related autoimmune retinopathy, as well as in choriocapillaritis.[15,16] Multimodal imaging helps us to differentiate the 2 causes as we can detect the presence or not of damage in photoreceptor outer segments through OCT and compare it with BAF and ICGA. In case of MEWDS, the multimodal imaging allows to identify the tissues affected but more importantly to suggest where the lesion process starts, choriocapillaris nonperfusion causing photoreceptor damage, a metabolically vulnerable cell.[11] The well-accepted diagnostic triad in MEWDS includes loss of photoreceptor outer segments on OCT, end-choriocapillary non-perfusion causing ICGA hypofluoresent areas, and BAF hyperautofluoresence.[17] In case of proven primary photoreceptor disease, there is photoreceptor damage seen on OCT and BAF hyperautofluorescence, but no ICGA hypofluorescence as the choriocapillaris is not involved. Figure 3 shows the sequence between fundus autofluorescence (FAF) and ICGA hypofluorescence, the latter preceding the secondary damage to photoreceptor outer segments shown on FAF. In one area, FAF presents only faint hyperautofluorescent dots, meaning that the damage to photoreceptor outer segments is not well advanced yet [Figure 3, red arrows], whereas on ICGA, already deep dark hypofluorescent lesions are visible in the exact same area [Figure 3, crimson arrows], meaning well-established choriocapillaris nonperfusion before full-blown loss of outer segments and hyperautofluorescence. In another area ICGA dark dots [Figure 3, green arrows] are fainter, whereas hyperautofluorescent areas [Figure 3, yellow arrows] are more clearly visible and extended indicating temporally more advanced photoreceptor lesions, suggesting that ICGA hypofluorescence precedes BAF hyperautofluorescence.

Figure 3.

Matching and time course between indocyanine green angiography (ICGA) hypofluorescent dots and BAF in a case of multiple evanescent white dot syndrome (ICGA angiographic time = 28’). When performing imaging examination, lesions in the fundus are not at the same stage and time point. In one area (red arrows and circle), hyperautofluorescence is still only faint, whereas ICGA hypofluorescence is marked (crimson arrows) indicating well-established hypo or nonperfusion. In the area nasal to the optic disc, hyperautofluorescent areas are brighter and more spreadout (yellow arrows), wheras corresponding ICGA areas (green arrows) are already fading when compared to the more dark temporal areas (crimson arrows)

The correlation between these 2 imaging methods showed that the encircled lesion (red circle on the right image) appeared as a well-established perfusion void on ICGA (darkly hypofluorescent), whereas on BAF, the lesion (red circle on the left) is barely visible, meaning that the photoreceptor outer segment damage is a secondary consequence of the preceding choriocapillaris nonperfusion. This case highlights the fact that in the same patient, lesions are at different stages of evolution. This is even more the case in the study by Mathis et al.,[2] as patients were included at different times with symptom durations ranging from 0 to 4 weeks. This makes it more difficult and sometimes very approximative to compare the two imaging modalities and draw conclusions.

Things were clearly attributed to choriocapillaritis[18,19,20,21,22] until the first photoreceptoritis article was published.[3] From then on, the great schism occurred, and part of the clinicians convinced in the past by the fact that MEWDS was a choroidal circulatory problem as for other choriocapillaritis entities, converted to the photoreceptoritis belief.[23]

OPTICAL COHERENCE TOMOGRAPHY ANGIOGRAPHY CANNOT DETECT LOW-FLOW END-CAPILLARIES

One of the reasons that oriented toward the primary RPE/photoreceptor disease theory in MEWDS was the development of OCT-A that showed an apparently normal choriocapillaris. The imaging capacity of OCT-A relies on the presence of sufficient flow. Structures such as low-flow or quasi-no-flow endchoriocapillaries do not appear on OCT-A imaging. Hence, it is not possible to determine whether there is perfusion or not in these structures. The flow in larger choriocapillaris vessels (1 mm/s) is already much slower than in retinal vessels (4.8 mm/s)[24] and this is even more so for endchoriocapillaries. Therefore, on OCT-A, the choriocapillaris appears normal in MEWDS as only normal flow/normal size choriocapillaris vessels are shown, where circulation remains normal in most MEWDS cases. However, the severity of MEWDS is not uniform, and more pronounced forms exist in which choriocapillaris dropouts have been demonstrated, indirectly showing that it is indeed the absence of perfusion that is causing ICGA hypofluorescent areas in MEWDS.[25] These MEWDS cases with choriocapillary dropouts clearly imply that choriocapillaris perfusion impairment is the primary mechanism at the origin of MEWDS.

THERE IS A CONTINUUM BETWEEN MULTIPLE EVANESCENT WHITE DOT SYNDROME AND OTHER CHORIOCAPILLARITIS ENTITIES ALL BASED ON INFLAMMATORY CHOROIDAL CIRCULATORY DYSFUNCTIONS

There are numerous articles reporting on cases that initially present as MEWDS and on follow-up develop recurrences and scars needing to be re-diagnosed as cases of multifocal choroiditis, which clearly shows that there is a continuum between these entities and, hence, a common pathophysiological process.[26,27,28,29] After a first report on the involvement of the choroidal circulation in MEWDS,[30] numerous articles showed the implication of the choroidal circulation at the origin of MEWDS and other choriocapillaritis entities.[18,20,31,32,33,34,35,36] A multicenter study showed that choroidal thickness and vascularity was significantly increased during the acute stage of MEWDS indicating involvement signs of the whole choroid.[37] This was also shown in two other similar studies.[38,39] It is certainly very unlikely that primary photoreceptor disease could have such an impact on the whole choroid but speaks much more for a choroidal origin of the disease. Finally, the great number of articles reporting on cases where MEWDS co-existed with other choriocapillaris entities represents a strong argument linking MEWDS to the other choriocapillaritis entities, thus speaking for a common pathophysiological process.[22,26,27,28,39,40,41,42] Figure 4 illustrates the case of a patient who had a typical presentation of MEWDS which later evolved to multifocal choroiditis with repeated episodes. Moreover, MEWDS can take an atypical presentation and “secondary forms” exist, associated with infectious chorioretinitis cases.[43] Such MEWDS-like findings were interpreted as satellite parainfectious choriocapillaritis in some cases such as toxoplasmosis.[44]

Figure 4.

(a) First episode of choriocapillaritis after a febrile episode with photopsies, visual field disturbance and typical multiple evanescent white dot syndrome (MEWDS) findings. Damage and loss of ellipsoid layer on spectral-domain optical coherence tomography (SD-OCT) with inlay showing that section goes through indocyanine green angiography (ICGA) hypofluorescent and blue light autofluorescence (BAF) hyperautofluorescent zone (Top right) (Middle row) ICGA shows extended area of hypofluorescence at presentation in the late angiographic phase (angiographic time = 23’13”–25’33”) (left) with almost complete resolution after 9 weeks (right) (Bottom row). BAF hyperautofluorescent areas corresponding to ICGA hypofluorescence at presentation (left) with partial resolution after 9 weeks (second frame from left) (bottom right). Visual field scotoma corresponding to the ICGA hypofluorescent and BAF hyperauoifluorescent area that recovered after 9 weeks (right). (b) Reccurrence of choriocapillaritis 10 months after the first episode is shown on 4a. Fundus shows development of scars (Top left, A). Damage and loss of the ellipsoid layer on SD-OCT (Top right, A). Extended area of ICGA hypofluorescence around optic disc and scattered dots beyond (angiographic times = 23’–24’) (Middle left quartet of frames, A). BAF hyperautofluorescence corresponding to ICGA hypofluorescence (A-BAF) and corresponding visual field loss (middle right, A) The patient made several more episodes with new fundus scars (Bottom left, B), ICGA hypofluorescence and damage to ellipsoid layer (not shown) as well as recurrent hyperaufluorescences (Bottom right, B). The diagnosis had to be revised to multifocal choroiditis. This example shows the continuum between MEWDS and other choriocapillaritis entities that have a common pathophysiology. SD-OCT: Spectral-domain optical coherence tomography, ICGA: Indocyanine green angiography, BAF: Blue light autofluorescence

PRIMARY PHOTORECEPTOR DISEASE DOES NOT PRODUCE INDOCYANINE GREEN ANGIOGRAPHY HYPOFLUORSCENCE

As indicated earlier in this review, multimodal imaging allows to differentiate choriocapillaritis-induced photoreceptor disease from primary photoreceptor disease.[15,16] As illustrated in Figure 5, primary photoreceptor disease is morphologically analyzed by spectral-domain-OCT showing damage/loss of the ellipsoid layer [Figure 5, bottom left] meaning loss of the screen in front of the RPE that causes unmasking of the RPE lipofuscin fluorescence appearing as hyperautofluorescent areas on BAF [Figure 5, top, second from right]. What differentiates primary photoreceptor damage from a choriocapillaritis induced secondary photoreceptor damage such as in MEWDS is ICGA which appears normally perfused in primary photoreceptor disease [Figure 5, top right]? Consequently, MEWDS cannot be considered a primary photoreceptoritis, as it should show no ICGA hypofluorescence, hence, no involvement of the choriocapillaris which obviously is not the case. Unfortunately, in many recent publications, ICGA is no more performed and no more part of multimodal imaging leading to incomplete appraisal of MEWDS. If the fanciful theory of failure of ICG uptake is considered, why does it not occur in proven primary photoreceptor disease, while it occurs in alleged “MEWDS photoreceptoritis”?

Figure 5.

Suspected cancer-associated photoreceptoritis. Patient presented with visual acuity reduced to hand movements. Fundus showed a perifoveal clear halo (top, middle left) corresponding to loss of photoreceptor outer segments on spectral-domain optical coherence tomography (bottom left); no hypofluorescence on indocyanine green angiography, indicating normal choroidal perfusion (top right), normal fluorescein angiography (botton, second from right) and severe scotoma on visual field (bottom right). SD-OCT: Spectral-domain optical coherence tomography, ICGA: Indocyanine green angiography

CONCLUSION

The most probable pathophysiology in MEWDS is endchoriocapillary nonperfusion, causing hypoxemic damage to the photoreceptors. The latter are damaged because they are metabolically vulnerable, whereas RPE cells are intact because metabolically more resistant. The only manner to show hypo or nonperfusion is to perform ICGA, as very slow or noflow endchoriocapillaries cannot be examined by OCT-A. The limited involvement of choriocapillaris nonperfusion explains the benign evolution of MEWDS with spontaneous recovery.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

Nil.