Purpose
To compare non perfusion area (NPA) and the ischemic index (ISI) between 2 fields of view using ultra-widefield optical coherence tomography angiography (UWF-OCTA) and evaluate their correlation with diabetic retinopathy (DR) severity and predominantly peripheral lesions (PPL).
Design
Cross-sectional, retrospective study.
Participants
Sixty-two patients (101 eyes) with DR without prior pan-photocoagulation (PRP) treatment, imaged with 12 × 12-mm and 21 × 26-mm scans on UWF-OCTA. Forty-nine patients (79 eyes) with same-day ultra-widefield color fundus photography (UWF-CFP) were included in the PPL analysis.
Methods
NPA was calculated from UWF-OCTA scans using a semi-automatic algorithm on FIJI. ISI was determined as NPA divided by total scan area. A subset of eyes with same-day UWF CFP were assessed by 2 graders for the presence and extent of fields with PPLs- defined as >50% of diabetic lesions in extended versus Early Treatment of Diabetic Retinopathy Study (ETDRS) fields. Eyes were classified as having PPLs if ≥ 1 field met this criterion.
Main Outcome Measures
NPA and ISI from 12 × 12-mm and 21 × 26-mm scans and their association with DR severity.
Results
NPA and ISI were significantly greater in PDR versus NPDR for both scans and the NPA/ISI was significantly higher in the 21 × 26-mm scan ( P <.001 for all). Both the 12 × 12-mm (OR: 1.26 [95% CI: 1.07-1.48, P =.006]) and the 21 × 26-mm ISI (OR: 1.20 [95% CI: 1.10-1.29, P <.001]) were significantly associated with increasing DR severity. ROC analysis showed similar performance between both scan ISI at predicting DR severity (AUC: 0.868 vs 0.878, P =.558). In the sub-analysis, presence of PPL was only significantly associated with ISI on the 21 × 26-mm scan (B = 0.43901, P =.009). However, the interaction of both presence (B = 0.16, P =.006) and extent (B = 0.2,5 P =.044) of PPLs with increasing DR severity was positively associated with increasing ISI on the 21 × 26-mm scan.
Conclusions
NPA/ISI on both the 12 × 12-mm and 21 × 26-mm scans is predictive of DR severity, highlighting the importance of OCTA detection of ischemia. The wider scan area of the 21 × 26-mm scan can better assess far peripheral ischemia that correlates with PPLs, supporting its potential role in tracking DR progression.
INTRODUCTION
Diabetic retinopathy (DR) is a leading microvascular complication of diabetes and a significant cause of vision impairment among working-age adults in the United States. , As diabetes prevalence continues to rise, with projections estimating that 60.6 million U.S. adults will be affected by 2060, the burden of DR is expected to increase correspondingly, necessitating advancements in detection and treatment strategies. ,,, The most widely used staging systems for DR include the Early Treatment Diabetic Retinopathy Study (ETDRS) severity scale and the International Classification of Diabetic Retinopathy (ICDR) relay on seven-standard field (7SF) color fundus photography, which captures only ∼30% of the retina. , Ultra-widefield (UWF) imaging address this limitation by capturing up to 82% of the retina, allowing for the detection of peripheral retinal lesions (predominantly peripheral lesions, PPLs). ,, UWF color fundus photography (UWF-CFP) imaging has facilitated the extension of the ETDRS DR severity scale by incorporating PPLs presence, which has been linked to an increased risk of disease progression independent of baseline severity. However, the biological mechanisms driving this association remain poorly understood.
The advent of swept-source optical coherence tomography angiography (SS-OCTA) has transformed DR assessment, offering a non-invasive alternative to fluorescein angiography (FA) and CFP, with an expanding scanning range. One of the key advantages of SS-OCTA is its ability to detect non-perfusion areas (NPAs), which serve as important biological markers especially in terms of DR progression and have been extensively studied using both FA and OCTA. ,,,, However, many of these previous studies were limited by smaller expanded field scanning ranges (less than 12 × 12), as NPAs are primarily located in the mid-periphery. Both PPLs and increased NPAs have been shown to contribute to an elevated risk of DR progression, yet the relationship between them remains under investigation. Previous research found no significant differences in expanded-OCTA (12 × 12 mm) NPA between groups with and without PPLs. Yet, other studies point to eyes with PPLs having significantly higher non-perfusion indexes (NPI) than eyes without, although variation in the OCTA scan size might have contributed.
In this study, we aim to cf NPAs and ischemic index (ISI) across different scan sizes using UWF-OCTA to further explore their correlation with DR severity. Additionally, we investigate the relationship between NPAs and PPLs to better understand the mechanisms by which PPLs may drives DR progression. By refining DR staging methodologies and leveraging advanced imaging technologies, this research seeks to improve the precision of DR assessment, allowing for more effective risk stratification and individualized treatment planning for patients with DR.
METHODS
Subjects and Study Design
This observational cross-sectional study was conducted at Massachusetts Eye and Ear (MEE) from May 2024 to January 2025. The study was approved by the Institutional Review Board of Massachusetts General Brigham (2019P001863), and informed consent was obtained from all subjects. All procedures adhered to the tenets of the Declaration of Helsinki and Health Insurance Portability and Accountability Act regulations.
Inclusion criteria for our study were (1) adult patients aged greater than 18 years, (2) a diagnosis of NPDR or PDR in at least one eye, (3) visual acuity (VA) of at least 20/200 Snellen.
Exclusion criteria included (1) inadequate image quality (defined as significant image artifact obscuring the view, and the peripheral field of view not fully exposed due to obstructions such as eyelashes or eyelids), (2) severe media opacity, such as corneal scarring and dense cataracts and vitreous hemorrhage, (3) the presence of other ocular comorbidities such as retinal vein or artery occlusion, retinal vasculitis, glaucoma, optic neuropathy, and pathological myopia, or (4) prior panretinal photocoagulation (PRP). A total of 116 eyes were screened, of which 15 were excluded due to inadequate image quality.
Procedures and Data Collection
All included participants (n = 62, 39 with bilateral data) underwent a comprehensive ophthalmic examination. VA, intraocular pressure (IOP), slit-lamp examinations, and relevant medical history were obtained. Participants were then imaged using a 400-kHz UWF SS-OCTA instrument (DREAM OCT; Intalight Inc., CA, USA). The DREAM OCT system provides an axial resolution of 5.5 µm (optical) and 2.0 µm (digital), and a lateral resolution of 15 µm (optical) and 5.8 µm (digital). Angiography 12 × 12-mm and 21 × 26-mm scans centered on the macula were performed. For the 12 × 12 mm protocol, there are 1024 × 1024 scans per volume. The 26 × 21 mm ultra-widefield protocol utilizes 1536 × 1240 scans per volume and provides an estimated 130° field of view. The 21 × 26-mm scans required approximately 15 seconds for image acquisition and were well tolerated. Same day UWF CFP (California, Optos plc., Dunfermline, UK), which captures approximately a 200° retinal field, were obtained for a subset of participants (n = 49). DR grading was carried out by experienced senior retina specialists (J.B.M., D.G.vs., D.H., L.A.K., N.A.P. D.M.W and J.W.M) based on both clinical findings and multimodal imaging using the International Clinical Diabetic Retinopathy Disease Severity Scale, which served as the gold standard for disease classification.
A comprehensive set of demographic and clinical data were collected from electronic medical records at baseline including age, sex (biological sex), smoking status (never, former, or current), type and duration of diabetes, most recent glycated hemoglobin (HbA1c) at time of image collection, presence of hypertension, prior treatment with anti-vascular endothelial growth factor (anti-VEGF) and pars plana vitrectomy (PPV).
Image Processing and Analysis
NPAs were calculated on retina slab of 12 × 12-mm and 21 × 26-mm angiograms, using a semi-automatic validated algorithm on FIJI, as described in a previous study ( Figure 1 ). Briefly, the FIJI algorithm was employed to enhance vessel visibility in the angiography scans through background subtraction and directional filtering. Following the application of global thresholding, NPA exceeding a predefined minimum size (250 pixel) were automatically identified via particle analysis. These threshold-derived NPA served as the regions of interest for subsequent quantification. The ISI was determined by dividing the NPA by the total scan area.
Non-perfusion areas (NPA) and ischemic index (ISI) measured by FIJI in 12 × 12-mm and 21 × 26-mm OCT angiography.
Eyes were evaluated for the presence and extent of PPLs (number of fields with PPLs) by independent reviewers (Y.Z, S.G) using established criteria. Any discrepancies between evaluators were resolved by a third reviewer (X.D). Only 79 of the 101 total study eyes had same-day UWF-CFP available for PPL analysis. A mask representing the ETDRS seven-field region was added using the built-in program of Optos Advance (Optos, Dunfermline, UK) ( Figure 2 ). Peripheral fields 3–7 were designated to the peripheral area adjacent to corresponding ETDRS field 3–7. The following lesions were considered for analysis: microaneurysms, hemorrhages, cotton wool spots, intra-retinal microvascular abnormalities, neovascularization, and fibrovascular proliferations. If more than 50% of diabetic lesions were subjectively considered to reside in the extended field compared to its respective ETDRS field, that field pair (ETDRS + extended field) was designated to have PPL. Only one field with PPLs was required for the eye to be graded as having PPLs. The extent of PPLs was determined by counting the number of field pairs exhibiting PPLs.
Peripheral predominant lesions on fundus photo.
Statistical Analysis
Statistical analysis was preformed using Stata version 18.0 (StataCorp; College Station, TX, USA) and R version 4.4.3 (R Foundation for Statistical Computing; Vienna, Austria). Descriptive statistics were reported as mean ± SD, median (IQR) or frequency (percentage, %), as appropriate. The differences in NPAs and ISI between the NPDR and PDR groups, across both 12 × 12-mm and 21 × 26-mm angiography, were analyzed using the Mann-Whitney U Test. The correlations between 12 × 12-mm and 21 × 26-mm ISI and the differences across various DR severities were analyzed using the Pearson correlation test and the Wilcoxon signed-rank test, respectively. Ordered logistic regression was performed to analyze the correlation between ISI of both scans sizes (12 × 12-mm and 21 × 26-mm) with DR severity, adjusting for age and diabetes duration, with clustering at the patient level. Receiver Operating Characteristic (ROC) analysis compared the predictive value of the 12 × 12-mm and 21 × 26-mm ISI in distinguishing mild-to-moderate NPDR from severe NPDR-to-PDR, considering sample size limitations and clinical relevance. The agreement between the 2 graders for the presence of PPL was assessed using Cohen’s Kappa (κ) coefficient. The association of the ISI of both scan sizes (12 × 12-mm and 21 × 26-mm) with the presence and extent of PPL was determined using multi-level mixed-effects linear regression. ISI values were rescaled by a factor of 100. All analyses accounted for nesting of eyes within patients . P -values were reported as 2-tailed, with values less than 0.05 considered statistically significant.
RESULTS
Demographic and ocular characteristics at baseline
The population included 101 eyes from 62 patients. Average age of patients was 62 years, with 37% of patients being female, 82% with T2DM, a baseline HbA1c of 7.70 (IQR: 6.80, 8.80) and a diabetes pathology duration median of about 18 years (IQR:13, 22). About 79% of patients had hypertension and about 39% of patients had been or were current smokers ( Table 1 ).
Table 1
DEMOGRAPHIC Characteristics of Patients with Diabetic Retinopathy
| Demographic Characteristics | All Patients | Patients with PPL | Patients without PPL | P -values |
|---|---|---|---|---|
| No. of patients (No. of eyes) | 62 (101) | 21 (29) | 32 (50) | |
| Age(years), Median (IQR) | 62 (51, 71) | 65 (53,73) | 62 (51, 68) | .397 |
| Sex | .855 | |||
| Male, n (%) | 39 (62.9) | 13 (61.9) | 19 (59.4) | |
| Female, n (%) | 23 (37.1) | 8 (38.1) | 13 (40.6) | |
| Race/Ethnicity | .946 | |||
| White, not Hispanic, n (%) | 38 (61.3) | 12 (57.1) | 17 (53.1) | |
| Black/African American, n (%) | 12 (19.4) | 4 (19.0) | 8 (25.0) | |
| Hispanic/Latin American, n (%) | 9 (14.5) | 4 (19.0) | 5 (15.6) | |
| Asian, n (%) | 3 (4.8) | 1 (4.8) | 2 (6.2) | |
| Hypertension, n (%) | 48 (77.4) | 18 (85.7) | 23 (71.9) | .239 |
| Smoking | .598 | |||
| Never, n (%) | 38 (61.3) | 11 (52.4) | 19 (59.4) | |
| Former, n (%) | 21 (33.9) | 8 (38.1) | 1 (3.1) | |
| Current, n (%) | 3 (4.8) | 2 (9.5) | 12 (37.5) | |
| Diabetes Types | .456 | |||
| Type 1, n (%) | 11 (17.7) | 5 (23.8) | 5 (15.6) | |
| Type 2, n (%) | 51 (82.3) | 16 (76.2) | 27 (84.4) | |
| Diabetes Duration (years), Median (IQR) | 18 (13, 22) | 19 (15, 30) | 15 (12, 19) | .007* |
| Most Recent Hgb A1c, Median (IQR) | 7.70 (6.80, 8.80) | 7.80 (7.00, 8.80) | 7.10 (6.68, 8.88) | .175 |
In terms of DR severity, 37.6% (38 eyes) had mild NPDR, 25.7% with moderate NPDR (26 eyes), 14.9% with severe NPDR (15 eyes) and 21.8% with PDR without PRP (22 eyes). Prior DR treatments included IV injections (32.7%), and PPV (4.0%). Among the 4 eyes that underwent PPV, 2 procedures were performed for vitreous hemorrhage secondary to PDR and 2 for vitreous hemorrhage associated with retinal tears.
Correlation of Retinal Non-Perfusion between 2 Scan Ranges
In the NPDR group, median NPA was 1.05 mm² ([IQR] 0.00, 5.52) for 12 × 12-mm scans and 11.32 mm² ([IQR] 0.00, 38.99) for 21 × 26-mm scans. The PDR group showed higher values: 17.00 mm² ([IQR] 8.59, 22.31) and 92.09 mm² ([IQR] 65.56, 133.54), respectively. The differences in all ischemia-related parameters between NPDR and PDR were statistically significant (all P <.001).
Similarly, ISI was significantly elevated in the PDR group. For 12 × 12-mm scans, median ISI was 0.01 ([IQR] 0.00, 0.04) in NPDR and 0.12 ([IQR] 0.06, 0.15) in PDR ( P =.002). For 21 × 26-mm scans, values were 0.02 ([IQR] 0.00, 0.08) in NPDR and 0.20 ([IQR] 0.14, 0.29) in PDR (all P <.001).
NPA and ISI in 21 × 26-mm OCTA field of view was significantly higher than the 12 × 12-mm field of view for both NPDR and PDR groups ( P <.001 for all) ( Figure 3 ).
Boxplots comparing non-perfusion area (NPA) and ischemia index (ISI) between 12 × 12-mm and 21 × 26-mm OCTA scan areas in patients with NPDR and PDR. Both NPA and ISI values were significantly higher in the 21 × 26-mm scan area compared to the 12 × 12-mm scan in both diagnostic groups. Additionally, PDR eyes exhibited consistently elevated NPA and ISI levels relative to NPDR eyes across both scan sizes.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
