Comparison of Early Treatment Diabetic Retinopathy Study Standard 7-Field Imaging With Ultrawide-Field Imaging for Determining Severity of Diabetic Retinopathy

Key Points

Question

Is ultrawide-field imaging comparable with Early Treatment Diabetic Retinopathy Study (ETDRS) imaging when assessing the severity of diabetic retinopathy within the 7 standard fields?

Findings

In this cross-sectional study, there were 737 gradable eyes on both ETDRS 7-field images and ultrawide-field images masked to contain the same 7 fields after adjudication; 435 eyes (59.0%) had exact agreement, and 714 eyes (96.9%) were within 1 step of agreement.

Meaning

These findings could justify ultrawide-field imaging for assessing the severity of diabetic retinopathy end points in future studies.

This cross-sectional study analyzes severity of diabetic retinopathy in patients whose eyes were examined via the Early Treatment Diabetic Retinopathy Study (ETDRS) 7-field imaging and ultrawide-field (UWF) imaging techniques.

Abstract

Importance

Moderate to substantial agreement between Early Treatment Diabetic Retinopathy Study (ETDRS) 7-field imaging and ultrawide-field (UWF) imaging has been suggested in single-center studies. Comparing images obtained by multiple centers could increase confidence that UWF images can be used reliably in place of ETDRS imaging in future clinical trials.

Objective

To compare diabetic retinopathy (DR) severity from modified ETDRS 7-field imaging and UWF imaging.

Design, Setting, and Participants

This preplanned, cross-sectional analysis included modified ETDRS 7-field images obtained using the Diabetic Retinopathy Clinical Research Network acquisition protocol and UWF images obtained captured with the Optos 200Tx system (Optos, PLC) from adult participants (≥18 years old) with type 1 or type 2 diabetes. Both image types were evaluated by trained graders masked to clinical data. Data collection occurred from February 2015 to December 2015, and data analysis from June 2016 to December 2017.

Main Outcomes and Measures

Agreement between UWF images, UWF images masked to include only the ETDRS 7-field area, and ETDRS 7-field images were calculated using κ statistics.

Results

A total of 764 eyes from 385 participants were included; participants had a median (IQR) age of 62.2 (53.6-69.2) years, 194 (50.4%) were women, and 256 (66.5%) were white. Of 742 eyes with both ETDRS 7-field images and UWF masked images graded, 359 (48.4% [95% CI, 44.4%-52.4%]) eyes had exact agreement, and 653 eyes (88.0% [95% CI, 85.2%-90.3%]) agreed within 1 step (weighted κ, 0.51 [95% CI, 0.44-0.58]). After open adjudication by an independent senior grader of all images with more than a 2-step discrepancy, perfect agreement was found in 435 eyes (59.0% [95% CI, 55.1%-62.8%]) and agreement within 1 step in 714 eyes (96.9% [95% CI, 95.1%-98.0%]; κ, 0.77 [95% CI, 0.73-0.82]). Ability of the imaging modalities to detect retinopathy severity in an individual eye was considered similar in 59 eyes (50.9% [95% CI, 41.3%-60.4%]), better for ETDRS 7-field imaging in 22 eyes (19.0% [95% CI, 12.5%-27.7%]), and better for UWF-masked images in 31 eyes (26.7% [95% CI 18.8%-36.5%]). Comparing UWF masked and unmasked images, 94 of 751 eyes (12.5%) had DR graded as at least 1 step more severe on UWF unmasked images vs UWF masked images. Predominantly peripheral DR lesions were present in 308 of 751 eyes (41.0%); this suggested increased DR severity by 2 or more steps in 34 eyes (11.0%).

Conclusions and Relevance

Imaging by the ETDRS 7-field and UWF imaging systems have moderate to substantial agreement when determining the severity of DR within the 7 standard fields. Disparities in an individual eye are equivalently distributed between imaging modalities and can be better or worse on 1 or the other. Longitudinal follow-up will evaluate the primary outcome of this study to determine if peripheral retinal findings are associated with future retinopathy outcomes.

Introduction

For 2 decades, the standard for determining severity of diabetic retinopathy (DR) has been the extended modified Airlie House classification used in the Early Treatment Diabetic Retinopathy Study (ETDRS).¹ In this defined classification system, the location and extent of specific retinal lesions are evaluated in the posterior pole, using 7 stereoscopic pairs of photographs per eye (ETDRS 7 standard fields). The DR severity determined in this manner is highly correlated with the risk of DR progression.² Current DR therapeutic and management guidelines are based on clinical trials using the ETDRS classification system. A condensed version of this classification has been proposed internationally for general clinical DR assessment.³

Substantial diabetic retinal pathology can exist in the retinal periphery located outside the ETDRS 7 standard fields, which in total compose about 34% of the retinal surface.^4,5,6 Advances in retinal imaging technology now allow ultrawide-field (UWF) cameras to evaluate up to 82% of the retinal surface in a single image.⁷ Several single-center studies have suggested moderate to substantial agreement between UWF and ETDRS imaging.^8,9,10 Furthermore, UWF studies have shown that DR occurs in areas peripheral to the ETDRS fields in up to 40% of eyes and is extensive enough to imply a more severe level of DR in 9% to 15% of eyes.^5,11,12,13

This study is a preplanned cross-sectional analysis of baseline data from an ongoing 4-year evaluation of 764 eyes of 385 participants to assess how the retinal far periphery observed using mydriatic UWF images (with a 200° view) affects the ability to assess DR severity and is associated with rates of DR worsening over time compared with ETDRS 7-field imaging. Severity grading between the ETDRS 7-field and UWF images within the ETDRS 7-field area was compared to assess whether UWF images can be used reliably in place of ETDRS imaging in future clinical trials. We also evaluated the peripheral area in the UWF images to determine presence, type, severity, and location of DR lesions outside the ETDRS 7-field area and how these peripheral lesions might change grading of DR severity compared with ETDRS images.

Methods

This study is a prospective observational study conducted by the Diabetic Retinopathy Clinical Research Network (DRCR.net) at 38 clinical sites in the United States and Canada. (The protocol, labeled AA, can be found at http://drcrnet.jaeb.org.) The study adhered to the tenets of the Declaration of Helsinki and was approved by multiple institutional review boards. Study participants provided written informed consent.

Enrolled participants were adults (≥18 years) with type 1 or type 2 diabetes. Study eyes had nonproliferative DR (NPDR; defined as ETDRS retinopathy severity at levels 35 through 53), no history of panretinal photocoagulation, and no central-involved diabetic macular edema on optical coherence tomography, based on DRCR.net sex-based and machine-based thresholds or clinical examination.¹⁴ This article includes data from the eyes included in the study (n = 579) and additional nonstudy eyes (n = 185) that had available ETDRS and UWF images at baseline.

Image Acquisition

At baseline, after pupil dilation, ETDRS 7-field stereoscopic images, UWF images, and a fundus examination were performed on both eyes of each participant. Fundus photographs were obtained based on the DRCR.net image acquisition procedure. The UWF images were obtained using the Optos 200Tx (Optos, PLC). The UWF imaging procedure included two 200° central images and four 200° steered images. In the present report, only the central images were evaluated. The professionals capturing ETDRS 7-field images were certified by the Fundus Photograph Reading Center at the University of Wisconsin, and those capturing UWF images were certified by Optos, PLC.

Image Grading

Ultrawide-field and ETDRS 7-field images were transferred digitally to the Joslin Diabetes Reading Center (Boston, Massachusetts) for standardized grading by 3 senior readers masked to all clinical characteristics and the grading of the other imaging modality. All readers had extensive experience grading retinopathy images (range, 14-37 years). To ensure reproducibility, all grading was performed using high-resolution displays that are color calibrated and gamma corrected biannually. More than 43 000 UWF images have been evaluated by the reading center. Grader agreement for DR severity (weighted κ) was 0.88 to 0.93 for UWF images and 0.83 to 0.84 for ETDRS 7 standard field images.¹⁵ Using images that were not part of this study, agreement between the 2 primary readers for UWF vs ETDRS images was substantial (κ, 0.80 ± 0.13).⁸

An individual reader graded either UWF or ETDRS images but never both modalities for the same eye. When grading the UWF images, images were initially presented with a template mask, which had been digitally overlaid automatically to obscure the retinal periphery such that only the ETDRS 7-field area was visible (UWF masked; Figure 1A). After complete grading of the UWF-masked image, the mask was removed, and the full image including the UWF periphery was evaluated (UWF unmasked; Figure 1B).

Figure 1. Ultrawide-Field Imaging and Masking.

Ultrawide-field (200°) images from the Optos 200Tx (Optos, PLC): A, Image periphery masked such that only the idealized Early Treatment of Diabetic Retinopathy Study (ETDRS) 7-field area is visible; B, Ultrawide-field unmasked (200°) view.

The UWF images stereographically projected by Optos software were used to determine DR severity. For all 3 image approaches, the presence and severity of the following lesions were assessed in temporal, superior temporal, inferior temporal, superior nasal, and inferior nasal fields (fields 3-7): hemorrhages and/or microaneurysms, intraretinal microvascular abnormalities, venous beading, new vessels on the disc, new vessels elsewhere, hard exudates, retinal thickening, preretinal hemorrhage, vitreous hemorrhage, traction retinal detachment, and central fields.

For the unmasked UWF images, each lesion type was characterized as (1) not observed, (2) predominantly observed within the ETDRS fields, (3) predominantly observed outside the ETDRS fields, or (4) observed uniformly distributed within and outside the ETDRS fields, or (5) ETDRS field obscured, (6) periphery obscured, or (7) ungradable. For each field, a lesion was considered predominantly peripheral if more than 50% of the lesion was observed in the retinal periphery compared with the area within the modified ETDRS fields. A lesion was considered uniformly distributed if it was equivalently located both within and outside the ETDRS field. A lesion was considered predominantly posterior if 50% of the lesion was within the modified ETDRS fields.

Based on the extent and severity of the lesions, a composite ETDRS DR grade was derived: no DR (level 10-12), minimal NPDR (level 14-20), mild NPDR (level 35), moderate NPDR (level 43), severe NPDR (level 47-53), very severe NPDR (level 53E), inactive proliferative DR (PDR) (level 60), non–high-risk PDR (level 61-65), and high-risk PDR (level 71-75). Proliferative DR was defined as new vessels in at least 1 field. In the absence of new vessels, if the eye had NPDR characteristics in at least 1 field, then the eye was considered to have NPDR regardless of any areas obscured. An eye was classified as having no DR if all fields were gradable and did not have DR characteristics. Sensitivity analyses using other approaches for missing fields (eg, assessing only eyes with data available for at least 3 fields) did not appreciably change the results.

The ETDRS 7-field and UWF masked images do not necessarily visualize the exact same retinal area because of asynchronous image capture and the unique imaging methodology of each instrument. Thus, preplanned side-by-side adjudication of all eyes with DR severity differing by 2 or more steps between the ETDRS and UWF masked imaging was performed by a senior grader (L.P.A.), who was not involved in prior grading, to assess the reasons underlying the discrepancy, including grader error, suboptimal image quality, or differences in visualized areas. The senior grader also determined from all available image information which modality provided the overall more accurate DR severity grade (recorded as “ETDRS more accurate,” “UWF more accurate,” or “modalities similar”).

Statistical Analyses

Percentages and 95% CIs were calculated using generalized estimating equations to account for the correlation between eyes of the same participant. Levels of agreement were calculated using κ statistics. For multilevel outcomes, quadratic weights were used to estimate κ statistics and 95% CIs. Strength of agreement beyond chance alone was determined using the Landis and Koch interpretation of κ statistics (0.20 indicates slight agreement; 0.21-0.40, fair agreement; 0.41-0.60, moderate agreement; 0.61-0.80, substantial agreement; and 0.81-1.00, almost perfect agreement). A sensitivity analysis was conducted to adjust for the correlation between eyes of the same participant, but the results did not change appreciably.¹⁶ All statistical analyses were conducted using SAS version 9.4 (SAS Institute Inc).

Results

Baseline characteristics are shown in eTable 1 in the Supplement. Among the 385 participants, the median (IQR) age was 62 (53-69) years; 194 (50.4%) were women, 256 (66.5%) were white, and 51 (13.2%) had type 1 diabetes. A total of 764 eyes were included, with a median (IQR) visual acuity letter score of 85 (80-89) letters (Snellen equivalent, 20/20). The distribution of DR severity graded for each image type is shown in eTable 2 in the Supplement. Based on ETDRS imaging, NPDR was present in 695 of 752 eyes (92.4%), with 559 eyes (74.3%) having minimal or moderate NPDR.

Comparison of Imaging Modalities Within the Same Fields (ETDRS and UWF Masked Images)

Comparison Prior to Adjudication

A total of 19 images were ungradable on either the ETDRS 7-field system (n=9) or UWF masked images (n=10), while 3 images were ungradable on both modalities, leaving 742 eyes for analyses. Severity of DR classified into 8 categories (with severe and very severe NPDR combined into 1 category) matched exactly between ETDRS 7-field images and UWF masked images in 359 eyes (48.4% [95% CI, 44.4%-52.4%]) and was within 1 severity level in 653 eyes (88.0% [95% CI, 85.2%-90.3%]), with moderate agreement (weighted κ, 0.51 [95% CI, 0.44-0.58]; Table 1, Figure 2A). Discrepancies of 2 or more steps were equally distributed between ETDRS and UWF masked images.

Table 1. Diabetic Retinopathy Severity Grading Comparing 7-Field Images With Masked Ultrawide Field Images Before Adjudication.

Ultrawide-Field Masked Photographs, No. (%)	ETDRS Photos, No. (%)
	No DR	Minimal NPDR	Mild NPDR	Moderate NPDR	Severe or Very Severe NPDR	Inactive PDR	Less Than High-risk PDR	High-risk PDR	Overall
Before Adjudicationa
No DR	2 (28.6)	0	11 (4.4)	1 (0.3)	2 (1.5)	0	0	0	16
Minimal NPDR	0	0	1 (0.4)	0	0	0	0	0	1
Mild NPDR	5 (71.4)	0	133 (52.8)	64 (21.3)	13 (9.6)	1 (4.3)	3 (20.0)	0	219
Moderate NPDR	0	0	87 (34.5)	142 (47.3)	56 (41.2)	1 (4.3)	3 (20.0)	2 (22.2)	291
Severe or very severe NPDR	0	0	16 (6.3)	83 (27.7)	55 (40.4)	0	3 (20.0)	3 (33.3)	160
Inactive PDR	0	0	3 (1.2)	1 (0.3)	0	20 (87.0)	2 (13.3)	0	26
Less than high-risk PDR	0	0	1 (0.4)	7 (2.3)	8 (5.9)	0	4 (26.7)	1 (11.1)	21
High-risk PDR	0	0	0	2 (0.7)	2 (1.5)	1 (4.3)	0	3 (33.3)	8
Overall	7	0	252	300	136	23	15	9	742
After Adjudicationb,c
No DR	8 (88.9)	0	4 (1.7)	1 (0.3)	0	0	0	0	13
Minimal NPDR	0	0	1 (0.4)	0	0	0	0	0	1
Mild NPDR	1 (11.1)	0	143 (60.1)	68 (22.3)	4 (3.1)	0	0	0	216
Moderate NPDR	0	0	86 (36.1)	151 (49.5)	42 (32.1)	0	0	0	279
Severe or very severe NPDR	0	0	4 (1.7)	81 (26.6)	82 (62.6)	0	1 (4.8)	1 (14.3)	169
Inactive PDR	0	0	0	0	0	26 (100)	0	0	26
Less than high-risk PDR	0	0	0	4 (1.3)	2 (1.5)	0	20 (95.2)	1 (14.3)	27
High risk PDR	0	0	0	0	1 (0.8)	0	0	5 (71.4)	6
Overall	9	0	238	305	131	26	21	7	737

Abbreviations: DR, diabetic retinopathy; ETDRS, Early Treatment of Diabetic Retinopathy Study; NPDR, nonproliferative diabetic retinopathy; PDR, proliferative diabetic retinopathy; UWF, ultrawide field.

^aExact agreement, 48.4%; agreement within 1 step, 88%.

^bExact agreement, 59.0%; agreement within 1 step, 96.9%.

^cFive eyes had missing DR severity grades after adjudication (excluded from the analyses).

Figure 2. Differences in Diabetic Retinopathy Severity Grade Between Early Treatment of Diabetic Retinopathy Study (ETDRS) 7-Field and Ultrawide-Field Masked (UWF) Images.

Differences in diabetic retinopathy severity grade on the derived 8-point scale, excluding all eyes that could not be graded by either the ETDRS 7-field photographs, the UWF masked photographs, or both. A, No difference accounted for 48.4% of the total; ETDRS 1 step worse, 16.7%; ETDRS 2 or more steps worse, 5.8%; UWF masked images 1 step worse, 22.9%; UWF masked images 2 or more steps worse, 6.2%. B, No difference accounted for 59.0% of the total; ETDRS 1 step worse, 15.2%; ETDRS 2 or more steps worse, 1.5%; UWF masked images 1 step worse, 22.7%; UWF masked images 2 or more steps worse, 1.6%.

Comparison After Adjudication

There were 130 eyes (17.0% [95% CI, 14.3%-20.1%]) that were 2 or more steps discrepant. Images could not be compared in 14 eyes because the ETDRS or UWF masked image was ungradable, leaving 116 eyes for adjudication.

After adjudication of these 116 eyes, DR severity between ETDRS and UWF masked images (n = 737) matched exactly in 435 eyes (59.0% [95% CI, 55.1%-62.8%]) and within 1 severity grade in 714 eyes (96.9% [95% CI, 95.1%-98.0%]). There was substantial agreement between the 2 imaging modalities (weighted κ,0.77 [95% CI, 0.73-0.82]; Table 2, Figure 2B).

Table 2. κ Statistics of Agreement Between Early Treatment of Diabetic Retinopathy Study 7-Field Images and Ultrawide-Field Images in Severity Grading, Before and After Adjudication.

Statistic	κ (95% CI)
Before adjudication
Unweighted	0.26 (0.21-0.31)
Weighted
With linear weights	0.38 (0.33-0.44)
With quadratic weights	0.51 (0.44-0.58)
After adjudication
Unweighted	0.42 (0.37-0.47)
Weighted
With linear weights	0.61 (0.56-0.65)
With quadratic weights	0.77 (0.73-0.82)

Among adjudicated images, the ability of ETDRS and UWF masked images to accurately detect retinopathy was considered equivalent in 59 eyes (50.9% [95% CI, 41.3%-60.4%]), ETDRS 7-field images were considered more accurate in 22 eyes (19.0% [95% CI, 12.5%-27.7%]), and UWF masked images were considered more accurate in 31 eyes (26.7% [95% CI, 18.8%-36.5%]). For the 59 eyes with similarly accurate images, grading error was the major reason for the difference, which was equivalently distributed between ETDRS images (n = 25) or UWF masked images (n = 29). When either ETDRS images (n = 22) or UWF masked images (n = 31) were identified as the more accurate modality, the major reason for the difference was suboptimal image quality of the other modality (UWF, n = 17; ETDRS, n = 27). The distribution of the lesions underlying these differences is in eTable 3 in the Supplement.

Additional Lesions Observed in the Retinal Periphery by UWF Imaging

There were 751 gradable eyes on both UWF masked and UWF unmasked images. The severity of DR between UWF masked images and UWF unmasked images (both 200°) matched exactly in 648 eyes (86.3% [95% CI, 83.4%-88.7%]), was more severe in the masked grade in 9 eyes (1.2% [95% CI, 0.6%-2.3%]), more severe by 1 step in the unmasked grade in 59 eyes (7.9% [95% CI, 6.0%-10.2%]), and more severe by 2 or more steps in the unmasked grade in 35 eyes (4.7% [95% CI, 3.3%-6.6%]) (eTable 4 and eFigure 1 in the Supplement). Comparing ETDRS images with the unmasked UWF images (n = 743), DR severity was more severe by 2 or more steps in the unmasked grade in 76 eyes (10.2% [95% CI, 8.1%-12.9%]) (eFigure 2 in the Supplement).

Predominantly peripheral DR lesions were present in 1 or more fields for 308 of 751 eyes (41.0%). In 222 of these eyes (72.1% [95% CI, 66.5%-77.1%]), masked and unmasked UWF DR grades matched exactly. In 83 eyes (26.9% [95% CI, 22.0%-32.5%]), greater DR severity was observed on the UWF unmasked image (eTable 5 in the Supplement), and 34 eyes (11.0%) had greater DR severity by 2 steps or more.

Distribution of Predominantly Peripheral Lesions

Lesions were observed in 4504 fields: 2312 (51.3%) were predominantly within the ETDRS fields, 660 (14.7%) were predominantly peripheral to the ETDRS fields, and 1532 (34.0%) were uniformly distributed. Of these fields, 3056 had hemorrhages and/or microaneurysms, with 458 fields (15.0%) predominantly peripheral; 1299 intraretinal microvascular abnormalities, with 168 fields (12.9%) predominantly peripheral; 110 venous beading, with 13 fields (11.8%) predominantly peripheral; and 39 new vessels elsewhere, with 21 fields (53.8%) predominantly peripheral (Figure 3). Of the 660 predominantly peripheral fields, 458 (69.4%) had hemorrhages and/or microaneurysms, 168 (25.5%) had intraretinal microvascular abnormalities, 21 (3.2%) had new vessels elsewhere, and 13 (2.0%) had venous beading. The distribution of specific predominantly peripheral DR lesions within individual eyes is shown in eFigure 3 in the Supplement.

Figure 3. Location of Observed Lesions on Ultrawide-Field Unmasked Images.

Percentage of fields with lesions observed within the Early Treatment of Diabetic Retinopathy Study (ETDRS) area, outside the ETDRS area, and uniformly distributed (approximately equivalent) within and outside the ETDRS area. The total number of fields with any lesion observed is also presented. With respect to any lesion (n = 4504), those predominantlywithin the ETDRS 7 fields was 51.3%, those uniformly distributed were 34.0%, and those predominantly outside the ETDRS fields were 14.7%. For hemorrhage or microaneurysms (n = 3056), those values were 46.0%, 39.0%, and 15.0%, respectively; for intraretinal microvascular abnormalities (n = 1299), 63.3%, 23.8%, and 12.9%, respectively; for venous beading (n = 110), 63.6%, 24.6%, and 11.8%, respectively; for new vessels elsewhere (n = 39), 33.3%, 12.8%, and 53.9%, respectively.

Discussion

Nearly all modern formal imaging assessments of DR severity have been based on grading the location and extent of key DR lesions within the ETDRS 7-standard field area.¹ However, given the technological advances now enabling UWF imaging and the potential benefits of this approach,¹⁷ there may be substantial impetus for moving to UWF imaging if it is comparable in determining DR severity and if pathology in the retinal periphery provides additional clinically useful information on prospective worsening of retinopathy.^8,11,12

In this study, we found that ETDRS images had moderate agreement with UWF images that were masked to reveal only the ETDRS 7-field area, and substantial agreement after open adjudication of known discrepancies by 2 masked graders. Unpublished data from a Wisconsin reading center report exact agreement and agreement within 1 step between 2 graders evaluating the exact same ETDRS image to be 78% and 99%, respectively (Barbara Blodi, MD, written communication, December 26, 2017). Predominantly peripheral lesions were common on UWF imaging, and results from this study showed that the UWF peripheral area may suggest increased DR severity by 2 or more steps in approximately 11.0% of eyes.

Although these data suggest that ETDRS and UWF imaging can both be used in clinical studies for assessing DR severity, it is important that the 2 imaging approaches do not visualize the exact same retinal area because of asynchronous image capture, manual positioning of each of the 7 ETDRS fields, and idealized ETDRS masking area used on the UWF masked images. Side-by-side comparison of the images derived from the 2 approaches was a crucial, preplanned component of this study to determine whether discrepancies were owing to inherent image quality, grading differences, or areas of retina visualized in 1 modality but not the other. Adjudication assessed whether each modality was more likely to accurately demonstrate presence of disease in an individual eye, or if the 2 imaging modalities were similar.

Both approaches to imaging were judged to similarly demonstrate retinopathy findings in more than half of the discrepant eyes. The major reason for the difference in eyes with 2 or more step discrepancy was grading errors. These cases were equivalently distributed between ETDRS and UWF masked images. Correcting these grading errors resulted in a 2 or more step difference in DR severity in 3.1% of eyes. When either modality was identified as the more accurate images, the major reason for the difference was suboptimal image quality of the other modality. The suboptimal image quality was equally distributed between modalities. These findings further suggest that (1) the 2 imaging modalities are comparable for determining DR severity when following the protocol used in this study and (2) maximizing image quality is of paramount importance. Differences in field definition accounted for only a small percentage of discrepant cases.

As reported previously, unmasked UWF images frequently identified additional DR lesions in the retinal periphery.^12,15,18 Given that imaging of the ETDRS fields identifies a more severe DR level based on grading variability in approximately 1.0% of cases, UWF unmasked images likely suggest a net increase of DR severity in 10.0% of eyes. This is similar to prior reports of 9.0% to 15.0% of eyes.^5,11,12,13

Limitations

A potentially important and yet unresolved issue regarding comparisons of 7-field and UWF imaging in management of DR concerns how an ophthalmologist might evaluate DR severity in the clinic without an expert grader or adjudication and how such evaluation may change over time. However, important for clinical trials, the data from this analysis suggest that, following the protocol for this study, moderate to substantial agreement of DR severity is possible when comparing 7-field and UWF modalities across a diversity of DR stages.

Another unresolved issue is the extent to which DR findings peripheral to the area of retina visualized with ETDRS 7-field imaging can affect the risk of DR progression. Future findings from the longitudinal data in this study might provide a definitive answer as to whether UWF images can improve methods to assess and triage eyes at risk for DR worsening.

Conclusions

The identification of a subset of patients at greatly increased risk of experiencing DR progression and onset of proliferative DR that cannot be assessed by ETDRS 7-field imaging would have important implications for the evaluation and care of diabetic eye disease. Not only would UWF devices be the preferred imaging modality, but their use would be important in clinical trial settings requiring precise assessment of prospective DR progression rates, in clinical care for accurate patient counseling, and in teleophthalmology programs to improve risk assessment and triage in eyes that otherwise would not have the peripheral retina evaluated. The current study suggests that enough patients have predominantly peripheral DR lesions at baseline to answer this question at the conclusion of this trial.

These baseline data show that ETDRS 7-field and UWF imaging can have moderate to substantial agreement when determining DR severity within the ETDRS area. These findings could justify UWF imaging for assessing DR severity end points in future studies. The use of UWF imaging in clinical settings not only increases the frequency of DR identification nearly 2-fold, but also reduces acquisition time by more than half, ungradable image rate by 71% (to <3%), and image evaluation time by 28% compared with nonmydriatic fundus photography.¹¹ However, this study was performed with the assistance of a dedicated image grading center. Additional investigations may determine if these data hold true within the clinical setting or how UWF images perform in the development of artificial intelligence approaches. Furthermore, UWF imaging identifies peripheral lesions not captured by ETDRS imaging in a large percentage of eyes. Whether identification of these peripheral lesions will substantially affect the ability to assess the risk of future DR progression awaits final data from this ongoing study.

Supplement.

eFigure 1. Summary of Differences of Diabetic Retinopathy Severity Graded with the Ultra-Wide Field Masked Unmasked Images

eFigure 2. Summary of Differences of Diabetic Retinopathy Severity Graded with the Early Treatment Diabetic Retinopathy Study and Ultra-Wide Field Unmasked Images

eFigure 3. Distribution of Predominantly Peripheral Lesions by Type

eTable 1. Participant Characteristics at Baseline-Overall

eTable 2. Distribution of Diabetic Retinopathy Severity Grading by Type of Image

eTable 3. Distribution of the Type of Lesions and Reason for Differences between Diabetic Retinopathy Grades with Ultra-Wide Field Masked Images and Early Treatment Diabetic Retinopathy Study Photographs

eTable 4. Comparison of Diabetic Retinopathy Severity Grading by Ultra-Wide Field Masked and Unmasked Images

eTable 5. Comparison of Diabetic Retinopathy Severity Level with and Without Peripheral Lesions Outside the Early Treatment Diabetic Retinopathy Study Modified Fields