Aims To describe optical coherence tomography angiography (OCT-A) abnormalities of patients with pseudophakic cystoid macular oedema (PCMO) before and after pharmacological resolution, compared with diabetic macular oedema (DMO) and normal eyes.
Methods In this retrospective, observational study, 44 eyes (30 patients) were included: 15 eyes (15 patients) affected by PCMO; 14 healthy fellow eyes used as negative control group; 15 eyes (15 age-matched and sex-matched patients) with DMO used as positive control group. All patients underwent a complete ophthalmological examination at baseline, including OCT-A scans of the macula through AngioPlex CIRRUS-5000 (Carl Zeiss Meditec, Dublin, USA). Patients with PCMO and DMO were re-evaluated after the pharmacological resolution of cystoid macular oedema (CMO).
Results Disruption of parafoveal capillary arcade and cystoid spaces in deep capillary plexus (DCP) were frequent in patients with PCMO and DMO (73% and 100%, 87% and 100%). Capillary abnormalities and non-perfusion greyish areas in DCP were more frequent in DMO (P<0.001 and P=0.014). Patients with PCMO showed a larger foveal avascular zone area in DCP at baseline (P<0.001), which significantly reduced after treatment (P=0.001). Vessel density of full-thickness retina and DCP was reduced in patients with PCMO (P=0.022 and P=0.001), and no changes were observed after treatment. Interestingly, DCP appeared less represented in patients with DMO than PCMO subjects (P=0.001).
Conclusions Patients with PCMO have an impairment of mainly DCP, partially reversible after treatment. Furthermore, we disclosed that different alterations of the retinal vasculature characterise CMO derived from two different diseases, namely PCMO and DMO, and this could be due to their distinct pathophysiology.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Pseudophakic cystoid macular oedema (PCMO), also called Irvine-Gass syndrome, is the most common cause of suboptimal visual acuity after cataract surgery.1 Classic perifoveal petaloid pattern with or without late leakage of the optic disk have been observed on fluorescein angiography (FA) in 20%–30% of patients undergoing modern cataract surgery and the percentage increase up to 40% using optical coherence tomography (OCT).2 3 Characteristics of PCMO on OCT include macular thickening and cystoid spaces located in the inner nuclear layer (INL) and Henle’s fibre layer.4 Also in uncomplicated cataract surgery, there is a progressive increase in retinal thickness up to 6 months with a peak at 4–8 weeks after the surgery, first in the parafoveal area (at 1 month) and then in the foveal area (at 2 months).3 5
Optical coherence tomography-angiography (OCT-A) is a new technique that provides unprecedented clues to the state of retinal vascular flow in a depth resolved manner6, thanks to the possibility of visualising the vascular networks in separate layers. It has been previously used to describe the microvasculature in normal eyes7 and in various retinal diseases.8–14 Recently, Spaide15 has demonstrated, using OCT-A, areas of decreased or absent flow in the deep capillary plexus (DCP) in cystoid macular oedema (CMO) secondary to different retinal vascular diseases, gathering insight in its complex pathogenesis.
The aim of this study is to describe, using OCT-A, retinal vascular network abnormalities in patients affected by PCMO, compared with CMO related to diabetic retinopathy (DR) and to normal eyes. In addition, we investigated the changes in superficial capillary plexus (SCP) and DCP after pharmacological resolution of PCMO.
We reviewed the charts of consecutive patients with the diagnosis of PCMO that presented at the Medical Retina & Imaging Unit of the Department of Ophthalmology, University Vita-Salute, Ospedale San Raffaele in Milan between January and June 2016. This study adhered to the tenets of the Declaration of Helsinki. All patients signed a written consent.
The criteria for inclusion were: (1) age greater than 18 years, (2) diagnosis of PCMO (defined as foveal CMO on structural OCT and perifoveal petaloid pattern of the optic disk on FA after cataract surgery).2 3 The exclusion criteria were: (1) presence of any other retinal disorder, (2) myopia greater than 6 dioptres, (3) any previous treatment administered for PCMO, (4) presence of media opacities to ensure proper images quality.
Fellow eyes of patients affected by PCMO were used as negative control group if they were not affected by any ocular disease.
An additional age-matched and sex-matched cohort of patients with treatment-naïve diabetic macular oedema (DMO), diagnosed as CMO secondary to DR visualised by structural OCT, was used as positive control group. Ocular exclusion criteria included: (1) diagnosis of DMO from more than 6 months, (2) any other retinal disease than DR, (3) previous laser photocoagulation or vitrectomy, (4) positive history for previous intravitreal treatments (eg, antivascular endothelial growth factor and corticosteroids). If both eyes of a patient met all inclusion criteria, only one eye was randomly chosen for study.
Clinical charts and multimodal imaging data, at baseline and after treatment, were reviewed. All patients (patients with PCMO and control subjects) underwent a complete ophthalmological examination, including best-corrected visual acuity (BCVA) using ETDRS charts, infrared reflectance (IR), spectral domain-OCT (SD-OCT), FA and OCT-A. IR, SD-OCT and FA were performed using Spectralis (Heidelberg Engineering, Heidelberg, Germany). All examinations were performed between the 9.00 am and 12.00 pm. Central macular thickness (CMT) in the central 1 mm diameter circle of ETDRS thickness map was recorded with the Spectralis Software (Heidelberg Eye Explorer V.220.127.116.11, Heidelberg, Germany).
OCT-A images acquisition and analysis
OCT-A examinations were performed using AngioPlex (CIRRUS HD-OCT model 5000, Carl Zeiss Meditec, Dublin, USA). We used a 3×3 mm scanning area centred on the macula. All acquisitions were performed using FastTrac retinal-tracking technology to reduce motion artefacts. The minimum strength of OCT-A images was 7 out of 10. Image segmentation of SCP and DCP was assessed and manually adjusted by one of the senior authors (GQ) to reduce the segmentation errors. According to the recent literature, the SCP included a slab extending from the internal limiting membrane to the inner plexiform layer (IPL) and the DCP included a slab extending from the IPL to the outer plexiform layer (OPL).16
Two trained examiners (EC and AC), masked to each other, analysed the abnormalities of the SCP and DCP surrounding the foveal avascular zone (FAZ). Several parameters were used: capillary abnormalities (defined as the presence of capillary network disruption and/or capillary dilation and/or macular shunting vessels), presence of intraretinal cystoid spaces (evaluated as well-defined black roundish areas without any signal on OCT-A) and presence of non-perfusion greyish areas. In SCP, we also evaluated the disruption/integrity of parafoveal capillary arcade (disruption was defined when extending over 1-quadrant of the entire length). In case of discordance, the results were adjudicated by a senior retinal specialist (GQ).
All images were exported into ImageJ 1.50 (National Institutes of Health, Bethesda, Maryland, USA) software. The FAZ area was manually measured both in SCP and DCP using a previously published method.17 Vessel density (VD) was calculated through image thresholding and binarisation.12 18 Otsu’s thresholding was used to binarise each image; FAZ area was contoured and coloured to pure blue. White pixels were considered as vessel and black pixels as background. VD was calculated as the ratio between the white pixel and the total pixels after FAZ exclusion, in all selected retinal plexuses (full-thickness, SCP and DCP).
Statistical analyses were performed using SPSS Statistics V.20 (IBM, Armonk, New York, USA). Quantitative variables were reported as means±SD and categorical variables as counts (percentage). Association between categorical variables was performed using Fisher’s exact test. The Gaussian distribution of continuous variables was verified with the Kolmogorov-Smirnov test. Comparisons of mean BCVA, CMT, FAZ area, VD, signal strength of patients with PCMO and DMO between baseline and the end of the treatment were performed using Student paired t-test. Comparisons of mean BCVA, CMT, FAZ area, VD, signal strength between patients with PCMO, patients with DMO and fellow eyes were performed using Student independent samples t-test. Pearson’s correlation analyses were performed to analyse the correlation between FAZ area, VD, BCVA and CMT. In all analyses, P values<0.05 were considered as statistically significant.
Fifteen eyes of 15 Caucasian patients (9 females, 6 males) were included. The mean age was 73±8 years (range 58–84 years) and patients developed a visual decline related to PCMO after a mean of 7.2±3.0 weeks (range 2–11 weeks). All cases were uncomplicated cataract operations (there was no capsule rupture, iris manipulation or other complication). One fellow eye of a patient was excluded from the control group due to the presence of an epiretinal membrane.
Twelve out of 15 eyes were treated (9 eyes with systemic and topical nonsteroidal anti-inflammatory drugs (NSAIDs) (25 mg of indomethacin orally four times a day for a period of 4 weeks associated with a topical diclofenac 0.1% four times a day)19 20 and 3 eyes with 700 µg intravitreal dexamethasone implant) and completed the follow-up. After 16±5 weeks (range, 12–24 weeks), all patients showed a resolution of CMO on SD-OCT. A strong response to the treatment was observed: CMT decreased significantly to 316±49 µm (P<0.001) and BCVA improved to almost 20/25 Snellen equivalent (0.15±0.12 LogMAR (P=0.023)) (table 1).
Fifteen eyes (15 patients) affected by DMO were included in the positive control group. The subjects were homogenous for age and sex: the mean age was 69±9 years (range, 58–83 years; P=0.18), with eight females and seven males (P=0.71). BCVA was ~20/40 Snellen equivalent (0.30±0.21 LogMAR; P=0.92) and CMT was 429±92 µm (range 308–578 µm; P=0.11). After the treatment (mean follow-up of 17±8 weeks; range, 12–24 weeks), CMT of DMO eyes decreased significantly to 314±70 µm (P=0.007) and BCVA improved to more than 20/40 Snellen equivalent (0.24±0.14 LogMAR (P=0.045)) (table 1). Comparing CMT after the treatment between patients affected by PCMO and DMO, no significant differences were disclosed (CMT 316±49 µm and 314±70 µm, respectively (P=0.964)). However, although not statistically significant, the mean BCVA at the end of the follow-up was higher in patients with PCMO compared with patients with DMO (0.15±0.12 LogMAR and 0.24±0.14 LogMAR, respectively (P=0.119)). No significant differences in signal strength were disclosed between all subgroups (table 1).
Abnormalities of SCP and DCP at OCT-A
Considering OCT-A images, table 2A shows SCP and DCP findings in PCMO group and DMO group. The parafoveal capillary arcade disruption was a frequent finding both in patients with PCMO and DMO (73% and 87%, respectively). Cystoid spaces were located in DCP in all patients when visible on OCT-A (figures 1C and 2C). Capillary abnormalities of both SCP and DCP were more frequent in patients with DMO compared with PCMO (P=0.001 and P<0.001, respectively) (figures 1A–C and 2A–C). Also non-perfusion greyish areas in DCP were more represented in patients with DMO compared with PCMO (P=0.014). In patients with PCMO, we observed a significant reduction of central cysts in DCP after treatment (from 93% to 8%, P<0.001), and a trend reduction of non-perfusion greyish areas in SCP and DCP (although the difference was not statistically significant; P=0.069 and P=0.19, respectively) (table 2A, figures 1A–C and 3A–C). In patients with DMO, we observed a significant reduction of central cysts in DCP after the treatment (from 93% to 30%, P=0.020) without any other significant change. No abnormities were observed in normal eyes (online supplementary figure 1A–C).
FAZ and VD analysis
Regarding FAZ area analysis, no significant difference was found in SCP comparing PCMO group at baseline to after treatment, DMO and normal groups (P=0.095, P=0.46 and P=0.37, respectively) (table 2B). Instead, PCMO eyes at baseline showed a larger FAZ area in DCP compared with fellow eyes (P<0.001), which was significantly reduced after treatment (P=0.001). Also in patients with DMO, FAZ area in DCP slightly reduced after the treatment (from 1.452±0.251 to 1.217±0.271 mm2 (P=0.001)); however, FAZ area in DCP after the treatment was significantly larger in patients with DMO compared with PCMO (P=0.035). No significant correlation was found between BCVA and FAZ area of both SCP and DCP in patients with PCMO (P=0.091 and P=0.37, respectively) and between CMT and FAZ area (P=0.32 and P=0.13, respectively).
With regard to full-thickness retina, patients with PCMO presented lower VD compared with fellow eyes (P=0.022). This difference was more significant by analysing only DCP (P=0.001), whereas no significant difference was disclosed in SCP (figure 1A–C, online supplementary figure 1A–C). VD in full-thickness retina, SCP and DCP did not exhibit any significant modification after treatment (P=0.46, P=0.68 and P=0.52, respectively). No significant difference was found in FAZ area and VD of different retinal plexuses at baseline and after treatment between patients with PCMO treated with NSAIDs and with intravitreal dexamethasone implant (P>0.3 in all analyses).
Also in DMO eyes, no significant difference was disclosed in VD of full-thickness retina, SCP and DCP between the baseline and after the treatment (P=0.88, P=46 and P=0.18, respectively). However, DCP in DMO eyes appeared less represented and with larger non-perfusion greyish areas compared with patients with PCMO; this difference was confirmed by VD analysis: patients affected by DMO showed a lower VD compared with PCMO subjects at the baseline (P=0.001), and this difference was maintained even after the treatment (P<0.001) (figures 1C and 2C). No other difference was disclosed between patients with PCMO and patients with DMO before and after the treatment. All other VD analyses are listed in table 2B.
A positive correlation was found between VD and FAZ area in DCP of PCMO eyes at baseline (P=0.014). Furthermore, CMT positively correlated with VD in full-thickness retina of PCMO eyes at baseline (P=0.024). These correlations were not disclosed in patients affected by DMO (P=0.46 and P=0.21, respectively). No significant correlation was found between BCVA and VD in all groups.
Histological and clinical studies supported the hypothesis that PCMO results from proinflammatory soluble factors inducing mid-capillary plexus hyperpermeability.21 22 FA demonstrated retinal vessel hyperpermeability showing the classic perifoveal petaloid-leakage. Furthermore, Sigler et al 22 supposed that the earliest retinal damage involved the INL with cystic changes and then the OPL and subretinal space. This pattern was very similar to that of active uveitic CMO,23 supporting again the inflammatory pathogenesis of PCMO. Nevertheless, FA has the limitation that cannot well visualise the DCP,10 which is the main retinal vascular layer altered in PCMO.
Recently, the introduction in clinics of OCT-A has provided the possibility of visualising the vascular networks in separate layers (SCP and DCP), overcoming some limitation of FA. To the best of our knowledge, no studies have been published about the investigation of retinal vascularisation using OCT-A in PCMO.
Several vascular network abnormalities were described in other forms of CMO using OCT-A.13–15 24 25 In our study, we confirmed that OCT-A is able to show the retinal vascular abnormalities and cystoid spaces, generally located in the DCP and non-perfusion greyish areas in PCMO.
Moreover, OCT-A offers the possibility to well measure the FAZ.26 An enlargement of the FAZ area has been shown in vascular diseases and some authors reported that it could contribute to BCVA impairment.25 27 28 In the present study, FAZ area in patients with PCMO showed no enlargement in SCP compared with the fellow eye. Nevertheless, the foveal capillary network in SCP showed several signs of impairment, as disruption of the parafoveal capillary arcade (73% of eyes) that may represent a beginning sign of SCP damage in PCMO and it may partially contribute to BCVA impairment.
In healthy subjects, histological studies with confocal imaging, speckle variance OCT and OCT-A detected vascular network in the retinal nerve fibre layer, ganglion cell layer (GCL), IPL, INL and OPL, generating a SCP (composed chiefly by one or more layers of capillaries inside the GCL) and a DCP (composed mainly by two subcomponents, one on either side of the INL).16 29–31 In eyes affected by PCMO, our results show a decreased VD in full-thickness retina and DCP compared with the fellow eye. Since the SCP is directly connected to the retinal arterioles, it has a great perfusion pressure compared with DCP and thus it is less vulnerable.15 Furthermore, at OCT-A, cystoid spaces were located only in DCP and not in SCP, and also FAZ area of DCP was larger compared with fellow eyes (P<0.001). These functional and anatomical alterations may explain why DCP is more compromised in patients with PCMO. Interestingly, our data showed that while enlarged FAZ of DCP was significantly reduced after the treatment (P=0.001), the impaired VD did not improve after CMO resolution. Based on these results, we suppose that retinal vasculature damage was partially not reversible with a permanent decreased flow on OCT-A (figures 1C and 3C). Although the evaluation of PCMO recurrences was not the purpose of our study, we speculated that this non-resolution of impaired VD could be implicated in the recurrence of some cases of PCMO.
In patients affected by CMO (both PCMO and DMO), OCT-A showed impaired blood flow in the regions affected by oedema on structural SD-OCT. One can argue that this could be an artefact of OCT-A signal due to the increased retinal thickness, but there was no evidence that attenuation of OCT-A signal was due to artefactual loss of signal for the CMO. Indeed, by analysing the corresponding structural B-scan, there was no evidence of decreased reflectivity in the mid-retinal layer (figure 1A-C). Therefore, we believe that in these cases, there is a true impairment of DCP blood flow. We report that non-perfusion greyish areas in DCP were more frequent in patients affected by DMO than PCMO (100% vs 67%, P=0.014), and also DMO subjects had a lower VD compared with patients with PCMO in DCP both at the baseline and after the treatment (P=0.001 and P<0.001, respectively) (figures 1C and 2C). Based on these observations, we suppose that this disparity was related to the different pathogenesis of the two diseases. Recently, Spaide15 has speculated that in DMO, the DCP vessels are not simply displaced by the cystoid spaces, but the flow is nearly absent or absent in the sites of CMO due to the occlusion of small vessels resulting from ischaemia, inflammation and leukostasis. If there is absent flow in DCP, the Müller cells could not traffic excess fluid out of the retina and thus cause CMO.15 However, in other forms of CMO, such as in PCMO and other inflammatory CMO, the DCP does not necessarily have to be occluded and possible leakage from both vascular layers can lead to cystoid fluid accumulations. Our data support this revolutionising theory on the different pathogenesis of CMO between vascular and inflammatory disease, by reporting several differences in non-perfusion greyish areas and VD of DCP between PCMO and DMO eyes. Moreover, this theory could also explain the different correlation found between CMT, FAZ area and VD in PCMO and DMO eyes. In PCMO eyes a strong correlation was found between VD, CMT and FAZ area. We speculate that a mechanic effect exerted by the intraretinal fluid could cause a displacement of retinal vessels, which could generate enlargement of FAZ and VD increase. Instead, in DMO eyes, no correlation was disclosed probably due to the occlusive pathogenesis of CMO and not to a mechanical vessel displacement. This theory on the different pathogenesis of CMO could also explain the different restoration of the FAZ area in DCP between patients with DMO and PCMO after the treatment (P=0.035) and the different BCVA at the end of the follow-up, although not statistically significant (P=0.119).
Though we acknowledge that this study has several limitations, mainly due to its retrospective nature and the relative short follow-up to evaluate the recurrences, this is the first study to compare different OCT-A features between these two types of CMO. Another limitation of our study is that OCT-A segmentation could be influenced by CMO; nevertheless, image segmentation was manually adjusted by a senior author and we did not report the absolute value of our findings, but we compared two groups affected by the same amount of oedema (no significant differences in CMT were disclosed). For this reason, we believe that the possible presence of artefacts influenced the analysis of both groups in the same way and then the two groups were comparable to each other.
In conclusion, using OCT-A we reported the impairment of mainly the DCP and the alteration of both retinal vascular plexuses in patients affected by PCMO. We showed that some OCT-A alterations were only partially reversible after therapy. Furthermore, we disclosed that different alterations of the retinal vasculature characterise CMO derived from two different diseases, namely PCMO and DMO, and this could be due to their distinct pathophysiology.
Contributors RS, EC, GQ: research design, data acquisition and analysis, interpretation of data, drafting the manuscript and critical revision of the manuscript. SM, AC, AR, LQ: data acquisition and analysis, critical revision of the manuscript. FB, GM: interpretation of data and critical revision of the manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests RS, EC, AC, SM, LQ, AR: none declared. GM is the consultant for: Alcon (Fort Worth, Texas, USA), Allergan Inc (Irvine, California, USA), Bausch and Lomb (Rochester, New York, USA), Santen (Osaka, Japan). FB is the consultant for: Alcon (Fort Worth, Texas, USA), Alimera Sciences (Alpharetta, Georgia, USA), Allergan Inc (Irvine, California, USA), Farmila-Thea (Clermont-Ferrand, France), Bayer Shering-Pharma (Berlin, Germany), Bausch And Lomb (Rochester, New York, USA), Genentech (San Francisco, California, USA), Hoffmann-La-Roche (Basel, Switzerland), NovagaliPharma (Évry, France), Novartis (Basel, Switzerland), Sanofi-Aventis (Paris, France), Thrombogenics (Heverlee, Belgium), Zeiss (Dublin, USA) Giuseppe Querques consultant for: Alimera Sciences (Alpharetta, Georgia, USA), Allergan Inc (Irvine, California, USA), Bayer Shering-Pharma (Berlin, Germany), Heidelberg (Germany), Novartis (Basel, Switzerland), Sandoz (Berlin, Germany), Zeiss (Dublin, USA).
Patient consent Obtained.
Ethics approval Ethics committee of San Raffaele Hospital.
Provenance and peer review Not commissioned; externally peer reviewed.