Secondary Logo

Journal Logo

Original Study

RETINAL NONPERFUSION IN THE EARLY TREATMENT DIABETIC RETINOPATHY STUDY SEVEN FIELDS COMPARED WITH WIDEFIELD FLUORESCEIN ANGIOGRAPHY

Correlation and Use of Extrapolation Factor

Martin-Gutierrez, Maria P. MD*; Vermeirsch, Sandra MD*; Chandra, Shruti MD*; Agarwal, Aditi A. K. MS(Ophthal), MRCS (Edin)*; Selvam, Senthil PhD, FRCOphth*; Thottarath, Sridevi MBBS*; Vazquez-Alfageme, Clara MD; Sen, Piyali MBBS*; Sivaprasad, Sobha DM, FRCOphth*,‡; Hykin, Philip G. FRCOphth*,‡; Nicholson, Luke FRCOphth*

Author Information
doi: 10.1097/IAE.0000000000003498
  • Open

Central retinal vein occlusions (CRVO) represent a frequent pathology seen in everyday Medical Retina practice.1,2 It typically presents with superficial and deep retinal hemorrhages in all four quadrants of the retina, and characteristically shows venous system tortuosity because of blood flow obstruction, with or without optic disk edema.3,4

Blood flow impairment after CRVO may result in capillary drop out and retinal nonperfusion, especially in the peripheral retina, where retinal vascularization is naturally less profuse. Deprivation of oxygen to the retina stimulates the production of vascular endothelial growth factor by mainly the retinal pigment epithelium, but also other types of cells, such as glial cells, endothelial cells, ganglion cells, or Müller cells, in an attempt to increase the retinal vascularization, and thus, retinal blood supply. However, increased intraocular vascular endothelial growth factor concentrations may result in the appearance of cystoid macular edema and retinal neovascularization. The latter, in turn, may lead to vitreous hemorrhage, anterior segment neovascularization, neovascular glaucoma, and visual impairment. Therefore, optimal management of patients with CRVO is crucial to prevent vision-threatening complications.

Central retinal vein occlusions may be classified into ischemic and nonischemic, depending on the extension of capillary nonperfusion on Early Treatment Diabetic Retinopathy Study (ETDRS) 7F fundus fluorescein angiography (FFA).3 This distinction is important because the risk of neovascularization development is different for these two groups. Ischemic CRVOs exhibit 10 or more disk areas (DAs) of nonperfusion on ETDRS 7F FFA, and may be associated with poorer visual acuity, relative afferent pupillary defect, and electrophysiological abnormalities, but most importantly, with a higher risk of neovascularization (35% of iris and angle neovascularization over 3 years, in contrast to 10% of nonischemic CRVO).5 According to the Central Vein Occlusion Study, the risk of anterior neovascularization defined as two clock hours of iris neovascularization or any angle neovascularization, in ischemic CRVO may be lowered to 20% when panretinal photocoagulation is administered prophylactically,6 but the difference was not statistically significant. Subsequently, there is some ongoing controversy regarding prophylactic panretinal photocoagulation for ischemic CRVOs.

The ETDRS 7F is a photographic method used for the Central Vein Occlusion Study study and many other clinical trials before the development of ultra-widefield imaging. It combines 7 standard photographic views for Zeiss (Carl Zeiss, Oberkochern, Germany) fundus camera, obtained during FFA.3 The seven stereographic photographs were assembled and assessed as a single image, which facilitated the task of determining nonperfusion within the photographed areas. This technique allowed the visualization of the central posterior 90° of the retina, which represents approximately 30% of the retinal surface.7 This was certainly relevant for prognosis estimations, as nonperfusion in the posterior pole constitutes the main risk factor for neovascularization,8 but was unable to provide information about the perfusion status of the far peripheral retina, where most nonperfusion occurs. The recent development of ultra-widefield imaging techniques9 allows a visualization of the peripheral retina up to 200°, constituting approximately 80% of the entire retinal surface,10 and possibly rendering the ETDRS 7F obsolete. Widefield angiography (WFA) is now commonplace in clinical practice and clinical research; however, our understanding of the management and definition of an ischemic CRVO derived from angiography is based on the 7F. Several studies have attempted to redefine ischemic CRVO using WFA, identifying 75 DA or more as an alternative criterion.8,11 Despite these efforts, such studies lack the robust nature of the landmark Central Vein Occlusion Study study. There is a limitation in our knowledge base on the correlation of 7F angiography and WFA in assessing retinal nonperfusion which is essential to bridge the traditional evidence base with current imaging modalities.

Nevertheless, widefield imaging may not be available in all clinical practices, and so, assessing nonperfusion in the peripheral retina remains a challenging task for many clinicians. In this study, we aim to investigate the relationship between area of nonperfusion identified with ETDRS 7F and WFA and if possible define a simple linear equation to extrapolate the areas of nonperfusion observed in the ETDRS 7 standard fields to widefield. This will provide guidance for the estimation of overall retinal nonperfusion, and help optimize clinical management of CRVO in centers with no access to WFA.

Methodology

This study is a post hoc analysis of the LEAVO study, a randomized clinical trial that compared the clinical effectiveness of ranibizumab, aflibercept, and bevacizumab intravitreal injections for the treatment of cystoid macular edema secondary to CRVO. The LEAVO study was approved by the United Kingdom National Ethics Committee Service (14/LO/1043) and the study was conducted in accordance with the tenets of the Declaration of Helsinki. Written informed consent was obtained from participants before entry into the study.

Inclusion Criteria and Data Collection

The eligibility criteria for the LEAVO study have been described.5 For this post-hoc study, only participants who had WFA performed using the Optos ultra-widefield system (Optos Plc, Dumfermline, Scotland) with available and gradable angiograms at baseline and 100 weeks were included. To eliminate any possible grading bias because of the presence of retinal hemorrhages at onset, we selected images from the WFA at exit only.

Image Acquisition

Widefield angiography were performed with Optos ultra-widefield system using a standard protocol after intravenous bolus infusion of 5 mL of 20% fluorescein sodium. The protocol consisted of acquiring images in the transit phase, arteriovenous phases, and late frames. A single investigator (L.N.) identified the best macula centered FFA image in the arteriovenous phase from the FFA series of each patient.

Image Processing

From 235 participants who had WFA in the LEAVO study, 184 had baseline and exit WFA. Of these, 31 were excluded because of ungradable images mainly because of media opacities. All 153 images were graded to evaluate retinal nonperfusion using the concentric rings method12 (Figure 1A) and nonperfusion within the posterior pole, temporal peripheral, nasal peripheral, and the total area of retinal nonperfusion were determined using the average measurements from the results of two consultant medical retinal specialists (L.N. and C.V.A.). The segments corresponding to the posterior pole, temporal peripheral, and nasal peripheral retina have been described previously.13

F1
Fig. 1.:
A. ETDRS 7 standard fields template superimposed on the ultra-widefield fundus fluorescence angiogram in an eye with CRVO; (B) superimposed template on an ultra-widefield fundus fluorescein angiogram to measure the extent of total retinal nonperfusion using the concentric rings method.

A template simulating the ETDRS 7F was then elaborated on Photoshop CS2 (Adobe, San Jose, CA) and was superimposed on the original image (Figure 1B). The distance between the fovea and the optic disk was measured in each patient and then used as radius for each of the 7F. Each image was then assessed by two graders (S.V. and M.P.M-G), who had over 6 years of experience in the field of Ophthalmology. Both graders were blinded to the patient information and to the assessment of the fellow grader. The average measurement between the two assessors were used for analysis. An open-source software by the National Institutes of Health, ImageJ, was used to measure the extension of nonperfusion within the ETDRS 7 standard fields. This software allows the grader to adjust the magnification, brightness, and contrast of the images to facilitate the task of identifying areas of nonperfusion. This was defined as absence of retinal arterioles and/or capillaries, and is detected by characteristics such as pruned appearance of adjacent arterioles and a darker appearance of the choroid, as per the SCORE study.14 The disk area for each image was measured, and later used to calculate the extension of nonperfusion for each image (expressed as disk areas); this was performed to reduce any possible bias secondary to the variability of the optic disk size in our cohort. All 153 images were graded on the same computer, HP Desktop Workstation Z230, with an Eizo FlexScan S2433W class color LCD monitor.

Statistical Analysis

Descriptive statistics were used to analyze the data. The intergrader agreement was expressed as intraclass correlation coefficient. The Shapiro–Wilk normality test was used to validate the normal distribution of the data, and Pearson correlation test was calculated to find linear correlation between variables. A regression line equation was calculated to approximately extrapolate the DA of nonperfusion in the peripheral retinal from the measurements in the ETDRS 7F and posterior pole, respectively. All statistical operations were conducted on R, a free software environment for data manipulation, statistical computing and graphical display. Statistical significance was set at 0.05 for all statistical calculations.

The 153 images were classified into five subgroups according to the extension of nonperfusion within the ETDRS 7F: 1) one group comprising all images with zero DA of nonperfusion, 2) a second group comprising all images with between 0 and 10 DA of nonperfusion, 3) a third group comprising all images with 10 to 20 DA of nonperfusion, 4) a fourth group comprising all images with 20 to 30 DA of nonperfusion, and 5) a fifth group comprising all images with over 30 DA of nonperfusion.

Results

From a total of 235 participants with widefield imaging, 184 eyes of 184 participants had baseline and week 100 widefield angiograms. Thirty-one eyes were excluded because of poor-quality and ungradable images resulting in a total of 153 eyes. The intergrader agreement between the two graders was calculated using the intraclass correlation coefficient, with a significance level of 0.05. The intraclass correlation coefficient calculated for our study was 0.98.

Eyes with no areas of nonperfusion on 7F angiography manifested a mean area of nonperfusion of 23.1 DA 95% CI (19.20, 27.06) in WFA, whereas eyes with any area of nonperfusion on 7F had 59.5 DA 95% CI (48.40, 70.53) of nonperfusion on WFA. Eyes with areas of nonperfusion less than 10 DA on 7F angiography manifested 45.5 DA 95% CI (35.75, 55.18) of nonperfusion on WFA. These eyes would normally be classified as nonischemic using the traditional criteria set in Central Vein Occlusion Study. Eyes with >10 DA of nonperfusion on 7F angiography had 115.5 DA 95% CI (88.89, 142.05) of nonperfusion on WFA. This is detailed in Table 1.

Table 1. - The Data From Our Cohort has Been Subdivided Into Five Groups to Facilitate Calculations
Groups N ETDRS 7 Standard Fields (DA) Posterior Pole (DA) Total Retina (DA)
0 73 0 0.07; 95% CI (0.07–0.21) 23.13; 95% CI (19.20–27.06)
0–10 64 2.19 2.12; 95% CI (1.02–3.23) 45.46; 95% CI (35.75–55.18)
10–20 11 13.66 11.77; 95% CI (8.46–15.08) 94.44; 95% CI (69.30–119.50)
20–30 2 25.49 34.46; 95% CI (12.03–56.88) 125.28; 95% CI (−42.18 to 292.74)
Over 30 3 48.79 35.74; 95% CI (15.65–55.80) 186.02; 95% CI (88.46–283.58)
The groups were classified as follows: Group 0 includes those patients who did not show ischemia within the ETDRS 7 standard field; Group 0 to 10 corresponds to those patients whose nonperfusion within the ETDRS 7 standard fields range from over 0 and 10 disk areas of ischemia, and so forth. The second column of the table, “ETDRS standard fields,” corresponds to the average of nonperfusion for each of the groups. The disk areas of nonperfusion within the posterior pole and total retina had been previously measured in a previous study.11 The columns “Posterior pole” and “Total Retina” express the average of nonperfusion found in the posterior pole and total retina, respectively, for each of the groups.

Correlation for our data was calculated using the Pearson correlation coefficient. We found that there was a linear correlation between DA of nonperfusion for ETDRS 7F and posterior pole (R = 0.921; 95% CI [0.208, 0.994], P-value = 0.026), DA of nonperfusion in the ETDRS 7F and total retina (R = 0.985, 95% CI [0.793, 0.999], P-value = 0.002), and DA of nonperfusion between posterior pole and total retina (R = 0.936; 95% CI [0.311, 0.995], P-value = 0.019) (Table 2).

Table 2. - Pearson's Correlation Test was Found to be Statistically Significant for Comparisons of Extension of Nonperfusion Between ETDRS Seven Standard Fields and Posterior Pole, ETDRS Seven Standard Fields and Total Retina, and Posterior Pole and Total Retina
Pearson's Correlation Test Statistic P
Nonperfusion in ETDRS 7 standard fields versus posterior pole R = 0.921; 95% CI (0.208–0.994) 0.03
Nonperfusion in ETDRS 7 standard fields versus total retina R = 0.985; 95% CI (0.793–0.999) 0.002
Nonperfusion in posterior pole versus total retina R = 0.936; 95% CI (0.311–0.995) 0.02

For each correlation, we also calculated a regression line equation. This would facilitate the task of approximately extrapolating the DA of nonperfusion measured within the ETDRS 7F to the posterior pole (y = 2.5 + 0.8x) or total retina (y = 37 + 3.2x), and posterior pole to total retina (y = 36 + 3.5x) (Figure 2).

F2
Fig. 2.:
Pearson correlation between variables. For each comparison, there is a near perfect, positive linear correlation between variables. A regression line equation has been calculated for each of the correlations, which will be helpful to estimate the areas of nonperfusion in the posterior pole from the area of nonperfusion in the ETDRS 7 standard fields (A), the total area of nonperfusion from the measurements in the ETDRS fields (B) and the total area of nonperfusion from the measurements in the posterior pole (C).

Eyes with no areas of nonperfusion on 7F, 79.5% manifested more than 10 DA of temporal peripheral nonperfusion, whereas only 1.4% had more than 10 DA of nasal peripheral nonperfusion. This was similar for eyes with areas of nonperfusion between 0 and 10 DA, with 82.8% of eyes having temporal peripheral nonperfusion exceeding 10 DA and 20.3% nasally. Interestingly, eyes with more than 10 DA of nonperfusion on 7F angiography, 100% had more than 10 DA of temporal peripheral nonperfusion, and 62.5% of eyes had more than 10 DA of nasal peripheral nonperfusion. For eyes manifesting more than 10 DA of temporal peripheral nonperfusion on WFA, 12.6% had more than 10 DA of nonperfusion on 7F angiography, whereas for eyes with more than 10 DA of nasal peripheral nonperfusion on WFA, 41.7% had >10 DA of nonperfusion on 7F angiography.

Discussion

This study identified a positive correlation between retinal nonperfusion measured on 7F angiography and WFA in eyes with CRVO. Eyes with <10 DA of nonperfusion on 7F angiography still had at least 35.8 DA of retinal nonperfusion on WFA, whereas eyes with >10 DA of nonperfusion on 7F angiography correlated with at least 88 DA of total nonperfusion on WFA. There is a positive and linear relationship between area of nonperfusion in 7F and WFA in CRVO with more than 3-times the amount of nonperfusion identified on WFA (Figure 3).

F3
Fig. 3.:
Fluorescein angiography exhibiting the area of retinal capillary nonperfusion in four different patients affected of CRVO; 70 DA of total retinal nonperfusion and 0.81 DA in 7F (A), 123 DA of total retinal nonperfusion and 15 DA in 7F (B), 130 DA of total retinal nonperfusion and 18 DA in 7F (C) and 175 DA of total retinal nonperfusion and 46 DA in 7F (D). These four examples may serve to judge clinically the extent of ischemia in patients with CRVO. Note that the areas of capillary nonperfusion within the seven ETDRS increase proportionally as the area of total area of nonperfusion also increases, according to the positive linear correlation between capillary nonperfusion in 7F and total retina demonstrated by our study.

A 153 WFA images were collected and analyzed. The agreement between our two graders in measuring the extension of nonperfusion within ETDRS 7F for these data was measured by intraclass correlation coefficient, which showed an agreement of 0.98 showing an almost perfect agreement between our graders, which adds consistency to our study.14 We subdivided the images into five different groups according to the extension of nonperfusion in the ETDRS 7F, and later correlated with the extension of nonperfusion within the posterior pole and total retina for each image. Despite the small number of variables for each category and in some of the subgroups, we preferred to continue analyzing our data when classified into the five subgroups, because it is our aim to offer manageable, comprehensible results that may be easily applied in everyday practice. We consider that not simplifying the data in this manner would have defeated this purpose (Figure 4).

F4
Fig. 4.:
Distribution of areas of nonperfusion for groups 0, 0 to 10 and over 10 DA of nonperfusion in the 7F (A), and areas of nonperfusion in total retina for all five subgroups (B). Median is shown for each group in each box. Group 0 to 10 is the most heterogeneous group in both cases, showing more variability than the rest of groups. The n in the remaining groups gets depleted as the areas of ischemia increase in each group.

The Pearson test was calculated to correlate the areas of nonperfusion (classified into the five subgroups) between ETDRS 7F and posterior pole, ETDRS 7F and total retina, and posterior pole and total retina. In all cases, we found a statistically significant, positive correlation. Given that the coefficient value was near one for all three cases, this constitutes a near-perfect correlation. However, because the n was small in those subgroups with a greater extent of ischemia (over 20 DA of nonperfusion), we recommend caution when working with these data, because we cannot offer a strong clinical conclusion in these cases, as opposed to the subgroups showing less than 20 DA of ischemia.

We hereby include a simple regression line that will allow clinicians to quickly and easily extrapolate the extent of total retinal nonperfusion from the data obtained from the ETDRS 7F or posterior pole. This, we think, should be extremely helpful, especially in those clinical practices lacking WFA, to optimize clinical management for patients with CRVO.

The prevailing criteria to consider a CRVO as ischemic is the presence of more than 10 DA of nonperfusion within the ETDRS 7F. In more recent WFA studies of nonperfusion in CRVO, >75 DA on WFA seems to be a consistent threshold for an “ischemic” phenotype with more than 75% of eyes with >75 DA found to have neovascular complications.8,11 This is reflected in our findings of eyes with <10 DA nonperfusion on 7F had at most, upper 95% CI, of 55.2 DA of nonperfusion on WFA, whereas eyes with >10 DA of nonperfusion on 7F had at least, lower 95% CI, 88 DA of total nonperfusion, further corroborating the WFA threshold of an ischemic CRVO of 75 DA. In addition, the use of intravitreal anti–vascular endothelial growth factor injections as a treatment for CMO may alter the natural course of those CRVOs that present with this complication. Therefore, prognostic markers for neovascularization are required, and the revision for the need of prophylactic panretinal photocoagulation.

It was previously reported in this cohort that temporal peripheral nonperfusion is common in CRVO on WFA13 and not surprisingly, temporal nonperfusion was common in all eyes irrespective of eyes with less or more than 10 DA of nonperfusion on 7F angiography. However, nasal peripheral nonperfusion was associated with more than 10 DA of nonperfusion on 7F. This suggests, despite the temporal preponderance to nonperfusion, the nasal peripheral retina is more protected, and its involvement suggests a more widespread and posterior spread of nonperfusion in eyes with CRVO.

In conclusion, there is a positive and linear correlation for area of nonperfusion in eyes with CRVO measured using ETDRS 7F and WFA with more than 3 times the amount of nonperfusion identified with the latter modality. A conversion factor of ×3 may give clinicians using seven field FFA an approximate estimate of the ischemia that widefield would have identified with a result of greater than 75 DA indicative of iCRVO and need for intervention.

References

1. Klein R, Klein BEK, Moss SE, Linton KLP. The Beaver Dam Eye Study. Ophthalmology 1992;99:58–62.
2. Mitchell P, Smith W, Chang A. Prevalence and associations of retinal vein occlusion in Australia. The Blue Mountains Eye Study. Arch Ophthalmol 1996;114:1243–1247.
3. Baseline and early natural history report: the central vein occlusion study. Arch Ophthalmol 1993,111:1087–1095.
4. Schmidt-Erfurth U, Garcia-Arumi J, Gerendas BS, et al. Guidelines for the management of retinal vein occlusion by the European Society of Retina Specialists (EURETINA). Ophthalmologica 2019;242:123–162.
5. Hykin P, Prevost AT, Vasconcelos JC, et al. Clinical effectiveness of intravitreal therapy with ranibizumab vs aflibercept vs bevacizumab for macular edema secondary to central retinal vein occlusion: a randomized clinical trial. JAMA Ophthalmol 2019;137:1256–1264.
6. Oishi A, Holz FG. Validation of concentric rings method as a topographic measure of retinal nonperfusion in ultra-widefield fluorescein angiography. Am J Ophthalmol 2016;161:220.
7. Silva PS, Cavallerano JD, Sun JK, et al. Peripheral lesions identified by mydriatic ultrawide field imaging: distribution and potential impact on diabetic retinopathy severity. Ophthalmology 2013;120:2587–2595.
8. Nicholson L, Vazquez-Alfageme C, Patrao NV, et al. Retinal nonperfusion in the posterior pole is associated with increased risk of neovascularization in central retinal vein occlusion. Am J Ophthalmol 2017;182:118–125.
9. Tan CS, Sadda SR, Hariprasad SM. Ultra-widefield retinal imaging in the management of diabetic eye diseases. Ophthalmic Surg Lasers Imaging Retina 2014;45:363–366.
10. Friberg TR, Pandya A, Eller AW. Non-mydriatic panoramic fundus imaging using a non-contact scanning laser-based system. Ophthalmic Surg Lasers Imaging 2003;34:488–497.
11. Tsui I, Kaines A, Havunjian MA, et al. Ischemic index and neovascularization in central retinal vein occlusion. Retina 2011;31:105–110.
12. Nicholson L, Vazquez-Alfageme C, Ramu J, et al. Validation of concentric rings method as a topographic measure of retinal nonperfusion in ultra-widefield fluorescein angiography. Am J Ophthalmol 2015;160:1217–1225.e2.
13. Nicholson L, Vazquez-Alfageme C, Sen P, et al. The clinical relevance of ultra-widefield angiography findings in patients with central retinal vein occlusion and macular oedema receiving anti-VEGF therapy. Eye 2022;36:1086–1093.
14. Blodi BA, Domalpally A, Scott IU, et al. Standard care vs corticosteroid for retinal vein occlusion (SCORE) Study system for evaluation of stereoscopic color fundus photographs and fluorescein angiograms: SCORE study report 9. Arch Ophthalmol 2010;128:1140–1145.
Keywords:

central retinal vein occlusion; retinal capillary nonperfusion; 7 fields ETDRS; fluorescein angiography; widefield angiography