Behavior of hyperreflective foci in non-infectious uveitic macular edema, a 12-month follow-up prospective study

Macular edema is the main cause of vision loss in patients with uveitis [1]. Spectral domain optical coherence tomography (SD-OCT) is the gold standard for the diagnosis of this condition. Retinal thickness has come to be recognized as a remarkably valuable measure in the management of patients with uveitic macular edema (UME) and is almost universally used as a main outcome measure in clinical trials evaluating treatments in uveitis. Qualitative data provided by SD-OCT, i.e., the presence of subretinal fluid, distribution of cysts, and ellipsoid zone status, have also been considered in some papers where the analysis of these data has contributed to understanding the pathogenesis and prognosis of UME [2, 3].

In recent years, hyperreflective foci (HRF) have been described in SD-OCT imaging of patients with macular edema secondary to diabetic retinopathy [4], retinal vein occlusions [5], type 2 macular telangiectasia [6], and age-related macular degeneration [7]. Though their origin is not clear, it has been shown that the abundance of such foci may vary after treatment and has been suggested that there may be an association between a decrease in HRF and an improvement in visual function [8, 9]. The aim of this study was to evaluate the presence and behavior of HRF in UME. In addition, we assessed the potential association between these foci on macular thickness and visual acuity (VA).

In this prospective study we describe, to our knowledge for the first time, the behavior of HRF in patients with UME. At baseline, patients had larger numbers of foci and half of them had at least some foci in the outer retina. During follow-up, while macular edema resolved, OCT showed fewer foci and those that remained were more frequently located in the inner retina. Figure 3 illustrates this behavior. Moreover, macular thickness was found to be associated with both the number and the distribution of the foci.

Although HRF have been described in several diseases including diabetic macular edema, age-related macular degeneration, retinal vein occlusions and type 2 macular telangiectasia, the precise nature of these foci and their molecular constituents remain unclear. Three main theories have been put forward in publications concerning these foci. Some authors have hypothesized that HRF represent precursors of hard exudates [4, 11]. Others suggested that they are degenerated photoreceptors or macrophages engulfing such cells, since they have been observed close to disrupted external limiting membrane and ellipsoid zone and have been associated with decreased VA [12]. Finally, other authors have interpreted them as microglial cells activated during an inflammatory reaction [13, 14]. All these theories are plausible and it is possible that all three mechanisms may occur in the same disease but it is likely that each one of them plays a predominant role in a given disease.

Shape of foci is described as round or oval in previous publications regarding HRF in patients with retinal vascular diseases. Our cases showed also round or oval foci. Regarding the size, hyperreflective foci are defined as “small” in publications that evaluate this feature in patients with retinal vascular diseases, though precise size is not reported. In SD-OCT figures provided in these publications displaying foci, variable sizes may be observed. Though it is a subjective judgement, we believe that in our uveitic patients, foci are usually smaller than those observed in patients with diabetic macular edema or retinal vein occlusions. We speculate that this may be explained by a different origin of foci in different diseases. In retinal vascular diseases bigger foci may correspond to lipid exudation; on the other hand smaller foci seen in our patients may correspond to inflammatory cells. A microglial and leukocytic origin of the HRF would seem the most plausible in the context of UME, given the clear inflammatory origin of this condition, and the absence of hard exudates in UME. This may support the view that HRF should not be considered an initial lipidic extravasation in these cases. In this regard, a study assessing SD-OCT imaging of the vitreous and retina of seven patients with posterior segment inflammatory disease described HRF of a size consistent with that expected for inflammatory cells [15]. Interestingly, data from studies performed in murine experimental autoimmune uveoretinitis assessing correlations of OCT imaging of inflammatory lesions and their histopathologic analysis have shown that HRF may represent cellular infiltration [16].

In our patients, HRF decreased over time after starting treatment, whilst macular thickness decreased and edema resolved. Similar findings have been described by other researchers in patients with retinal vasculopathies and age-related macular degeneration. In diabetic macular edema, Vujosevic et al. [9] assessed the presence of HRF and the effect of treatment with anti-vascular endothelial growth factor on their abundance. They observed that the number of foci decreased after treatment, but did not find a correlation between the number of HRF and retinal thickness. In patients with macular edema secondary to diabetic retinopathy or branch retinal vein occlusions treated with intravitreal implant of dexamethasone or ranibizumab, Chatziralli et al. observed a decrease in HRF in parallel with resolution of the macular edema [17]. Framme et al. described a reduction in the number of HRF in 54% of their patients with neovascular choroidal neovascularization after treatment with ranibizumab [7]. Moreover, this reduction correlated with a decrease in CMT. Abri Aghdam et al. assessed the behavior of HRF in patients with neovascular age related macular degeneration after treatment with intravitreal aflibercept [18]. They observed a decrease in the number of foci within radius of 500 and 1500 μm, as well as a correlation between the CMT and number of foci within a 500-μm radius. To explain this behavior, these researchers suggested that HRF were precursors of lipid exudates and hence a sign of hyperpermeability, which might explain the association found between number of foci and macular thickness.

In patients with UME, the inflammatory process induces the invasion of leukocytes into the retina and the activation of microglia. These undergo significant changes in shape and size, from ramified multidirectional extensions to polarized dendrites and then to larger rounded cells which aggregate [19]. Leukocytes and activated microglial cells produce cytokines that increase vascular and epithelial permeability. When the inflammation resolves, the level of retinal cellular infiltrates decreases. These phenomena may support our finding of an association between HRF number and macular thickness.

As mentioned above, almost half of our cases showed HRF in the outer retina at baseline. During follow-up, as the edema resolved, foci were more frequently located in the inner retina. Similar observations have been described in diabetic macular edema. Vujosevic et al. [9]_, in their study assessing the effect of ranibizumab on HRF in diabetic macular edema, reported that the main decrease in foci occurred in the outer nuclear layer when edema resolved. Zheng et al. [19] showed that resting microglia are physiologically located in the inner retinal layers in human eyes. In the same study it was shown that activated microglia migrate towards the outer retinal layers in human eyes with diabetic macular edema and it is suggested that proinflammatory cytokines are responsible for the activation of microglia. Interestingly, in a rat model of experimental autoimmune uveoretinitis, Rao et al. showed that microglia had migrated from the nerve fiber layer and other inner retinal layers to the photoreceptor layer at day 9 after the induction of uveitis [20]. Moreover, Ding X et al. showed that rat microglial cells activated by lipopolysaccharides secreted proinflammatory cytokines (tumour necrosis factor and interleukin beta) which could promote vascular dysfunction and hence permeability [21]. These findings could explain the high frequency of patients with HRF in the outer retina at baseline, when macular edema was present and hence the inflammatory process was active. It has been shown that glucocorticoids inhibit microglial migration [22]. We speculate that the treatment given in our patients (mainly glucocorticoids) may explain the behavior of the HRF after treatment, that is, a more frequent location of foci in the inner retina.

We have not found the number or location of the HRF to have an independent influence on VA. Previous studies have evaluated possible associations between HRF and visual outcomes in diabetic macular edema, retinal vein occlusions and neovascular age-related macular degeneration. In the study by Vujosevic et al., the number of foci was correlated inversely with retinal sensitivity and directly with non-stable fixation, as measured by microperimetry [9]. Uji et al. found an association between the presence of HRF in the outer retinal layers and poor VA in patients with diabetic macular edema [12]. Moreover, both HRF and VA were associated with disruptions in the external limiting membrane and in the junction between the inner and outer segment of the photoreceptors (nowadays known as the ellipsoid zone). In the study by Chatziralli et al. performed in patients with diabetic macular edema and retinal vein occlusions, a higher number of HRF was associated with poorer VA [17]. Kang et al. found that the number of HRF at baseline was inversely associated with final VA in patients with diabetic macular edema treated with intravitreal bevacizumab [23]. The same group reported similar findings in patients with branch retinal vein occlusion [8], neovascular age-related macular degeneration and polypoidal choroidal neovascular vasculopathy [24].

In most of the aforementioned studies, HRF are assumed to be extravasations of lipoproteins and precursors of lipid exudates and the researchers consistently suggest that the underlying pathogenesis of poorer visual outcomes may be related to a toxic effect of lipids on photoreceptors. Regarding UME, it has been shown in a rat model of experimental autoimmune uveoretinitis that activated microglia located at the photoreceptor layer secrete peroxynitrite, which is the most potent biological oxidant known and capable of oxidizing cellular components [20]. Assuming a microglial origin of the foci, one could expect some deleterious effect on photoreceptors and hence on VA, due to the presence of microglia in the outer retina. We failed, however, to demonstrate an association between BCVA and HRF number or distribution. A protective effect of the treatment administered, usually local or systemic steroids, on photoreceptors and/or an insufficient capability of the BCVA test to highlight functional damage may explain this finding.

The main limitation of our study is the subjective assessment and counting of HRF. Nevertheless, agreement between graders of the scans was high. In the future, software able to automatically measure the amount of HRF may help us objectively define the behavior of such foci and clarify their meaning and relevance. Other limitations are the relatively small size of the sample and that the SD-OCT device used lacks software capable of performing consecutive scans in the same retinal section. Strengths of our study are its prospective nature and the long-term follow-up.

In conclusion, SD-OCT scans showed HRF in eyes with UME in our study. After treatment, the number of foci decreased and their distribution changed, remaining foci locating preferentially in the inner retina, and this was associated with a decrease in macular thickness. Further studies with larger numbers of patients are needed to confirm these results and shed light on their implications for clinical practice.