Risk factors for progression of subclinical diabetic macular oedema

A Diabetic Retinopathy Classification Framework Based on Deep-Learning Analysis of OCT Angiography

Purpose

Reliable classification of referable and vision threatening diabetic retinopathy (DR) is essential for patients with diabetes to prevent blindness. Optical coherence tomography (OCT) and its angiography (OCTA) have several advantages over fundus photographs. We evaluated a deep-learning-aided DR classification framework using volumetric OCT and OCTA.

Methods

Four hundred fifty-six OCT and OCTA volumes were scanned from eyes of 50 healthy participants and 305 patients with diabetes. Retina specialists labeled the eyes as non-referable (nrDR), referable (rDR), or vision threatening DR (vtDR). Each eye underwent a 3 × 3-mm scan using a commercial 70 kHz spectral-domain OCT system. We developed a DR classification framework and trained it using volumetric OCT and OCTA to classify eyes into rDR and vtDR. For the scans identified as rDR or vtDR, 3D class activation maps were generated to highlight the subregions which were considered important by the framework for DR classification.

Results

For rDR classification, the framework achieved a 0.96 ± 0.01 area under the receiver operating characteristic curve (AUC) and 0.83 ± 0.04 quadratic-weighted kappa. For vtDR classification, the framework achieved a 0.92 ± 0.02 AUC and 0.73 ± 0.04 quadratic-weighted kappa. In addition, the multiple DR classification (non-rDR, rDR but non-vtDR, or vtDR) achieved a 0.83 ± 0.03 quadratic-weighted kappa.

Conclusions

A deep learning framework only based on OCT and OCTA can provide specialist-level DR classification using only a single imaging modality.

Translational Relevance

The proposed framework can be used to develop clinically valuable automated DR diagnosis system because of the specialist-level performance showed in this study.

Comparison of Central Macular Fluid Volume With Central Subfield Thickness in Patients With Diabetic Macular Edema Using Optical Coherence Tomography Angiography

IMPORTANCE

Diabetic macular edema (DME) is the predominant cause of visual impairment in patients with type 1 or 2 diabetes. Automated fluid volume measurements using optical coherence tomography (OCT) may improve the diagnostic accuracy of DME screening.

OBJECTIVE

To assess the diagnostic accuracy of an automated central macular fluid volume (CMFV) quantification using OCT for DME.

DESIGN, SETTING, AND PARTICIPANTS

A cross-sectional observational study was conducted at a tertiary academic center among 215 patients with diabetes (1 eye each) enrolled from January 26, 2015, to December 23, 2019. All participants underwent comprehensive examinations, 6 × 6-mm macular structural OCT horizontal raster scans, and 6 × 6-mm macular OCT angiography volumetric scans. From January 1 to March 30, 2020, 2 retinal specialists reviewed the structural OCT scans independently and diagnosed DME if intraretinal or subretinal fluid was present. Diabetic macular edema was considered center involved if fluid was present within the central fovea (central 1-mm circle). A third retinal specialist arbitrated any discrepancy. The mean central subfield thickness (CST) within the central fovea was measured on structural OCT horizontal raster scans. A deep learning algorithm automatically quantified fluid volumes on 6 × 6-mm OCT angiography volumetric scans and within the central foveas (CMFV).

MAIN OUTCOMES AND MEASURES

The area under the receiver operating characteristic curve (AUROC) and the sensitivity and specificity of CST and CMFV for DME diagnosis.

RESULTS

We enrolled 1 eye each of 215 patients with diabetes (117 women [54.4%]; mean [SD] age, 59.6 [12.4] years). Diabetic macular edema was present in 136 eyes; 93 cases of DME were center involved. The AUROC of CMFV for diagnosis of center-involved DME (0.907 [95% CI, 0.861-0.954]) was greater than the AUROC of CST (0.832 [95% CI, 0.775-0.889]; P = .02). With the specificity set at 95%, the sensitivity of CMFV for detection of center-involved DME (78.5% [95% CI, 68.8%-86.3%]) was higher than that of CST (53.8% [95% CI, 43.1%-64.2%]; P = .002). Center-involved DME cases not detected by CST but detected by CMFV were associated with a thinner CST (290.8 μm [95% CI, 282.3-299.3 μm] vs 369.4 μm [95% CI, 347.1-391.7 μm]; P < .001), higher proportion of previous macular laser treatment (11 of 28 [39.3%; 95% CI, 21.5%-59.4%] vs 12 of 65 [18.5%; 95% CI, 9.9%-30.0%]; P = .03), and female sex (20 of 28 [71.4%; 95% CI, 51.3%-86.8%] vs 31 of 65 [47.7%; 95% CI, 35.1%-60.5%]; P = .04).

CONCLUSIONS AND RELEVANCE

These findings suggest that an automated CMFV is a more accurate diagnostic biomarker than CST for DME and may improve screening for DME.

Risk factors for fellow eye treatment in protocol T

PURPOSE

To identify risk factors for fellow eye treatment of diabetic retinopathy with Vascular Endothelial Growth Factor (VEGF) injections during the Diabetic Retinopathy Clinical Research Network (DRCR.Net) Protocol T trial METHODS: In this post-hoc analysis of randomized clinical trial data, Cox regression analysis was performed at 52 and 104 weeks to determine risk factors for treatment in 360 fellow eyes. Survival analysis was performed to determine mean time to treatment based upon medication used.

RESULTS

Of 360 fellow eyes, 142 (39.4%) required treatment between weeks 4 and 104. Risk factors predicting a lower likelihood of year 1 treatment included older subject age (Hazard Ratio [HR]=0.98, 95% CI 0.96-0.99; p = 0.02) and higher baseline study eye ETDRS score (HR=0.98, 95% CI 0.97-0.99, p = 0.04). Center-involving DME at baseline in the fellow eye was predictive of a higher treatment need at both 52 (HR=1.89, 95% CI 1.42-2.51, p < 0.0001) and 104 weeks (HR=2.68, 95% CI 1.75-4.11, p < 0.0001). Subjects treated in the study eye with aflibercept (HR=0.574, 95% CI 0.371-0.887, p = 0.013) and ranibizumab (HR=0.58, 95%CI 0.36-0.94, p = 0.03) were less likely to require first year fellow eye injection than subjects treated with bevacizumab although this difference was no longer significant at week 104 (aflibercept HR=0.77, 95% CI 0.52-1.16, p = 0.21; ranibizumab HR=0.66, 95% CI 0.43-1.00, p = 0.05). Mean time to treatment was significantly shorter in the bevacizumab group (bevacizumab 25.83 weeks, aflibercept 38.75 weeks, ranibizumab 34.70 weeks (p=0.012)).

CONCLUSION

Bilateral treatment with intravitreal anti-VEGF injections was common during the DRCR.net Protocol T. Medication choice may impact the risk of fellow eye treatment.

Correlations of blood lipids with early changes in macular thickness in patients with diabetes

PURPOSE

This study aimed to investigate the correlations of blood lipids with changes in macular volume and thickness in patients with diabetes.

METHODS

Central subfield macular thickness (CSMT) and central subfield macular volume (CSMV) were measured in 85 patients with type 2 diabetes (DM2) using spectral-domain optical coherence tomography (SD-OCT). In addition, serum creatinine (Cr), total cholesterol (TC), triglyceride (TG), high-density lipoprotein cholesterol (HDLC), low-density lipoprotein cholesterol (LDLC), glycated hemoglobin (HbA1c), and proteinuria were measured in all patients.

RESULTS

The mean CSMT of the 85 patients was 272.8±16.9μm, and CSMV was 215.1±14.6×10^-3mm³. TC, LDLC, HbA1c, and proteinuria were closely correlated to CSMT, and HDLC, HbA1c, and proteinuria were correlated to CSMV. After adjustment for sex, age, DM2 course, proteinuria, and HbA1c, LDLC was significantly correlated to CSMT (95% confidence interval (CI), 1.93-11.05, P=0.008) and CSMV (95% CI, 0.92-8.41, P=0.021); however, HDLC, TC, and TG showed no significant correlation with CSMT or CSMV.

CONCLUSIONS

Increased LDLC was closely correlated to changes in CSMT and CSMV in early diabetic patients, and increased CSMT was also accompanied by increased LDLC; therefore, this type of patient would be more likely to develop macular edema.

Plasma homocysteine levels are associated with macular thickness in type 2 diabetes without diabetic macular edema

PURPOSE

To explore the relationships between macular thickness and the plasma concentrations of homocysteine, vitamin B12, folate, and other known risk factors for patients with diabetes without diabetic macular edema (DME).

METHODS

Fasting venous blood samples were collected from 252 subjects (126 relatively healthy subjects with type 2 diabetes without diabetic macular edema and 126 age- and gender-matched controls). Measurement of macular thickness and volume was performed for those subjects using SD-OCT. The plasma concentrations of homocysteine, vitamin B12, folate, and other known risk factors were analyzed in all the patients and controls using multiple linear regression models.

RESULTS

An increase in the serum levels of homocysteine was present within patients with type 2 diabetes compared to healthy individuals. The mean total plasma homocysteine levels were associated with a greater central subfield macular thickness (CSMT), average macular thickness (AMT), and average macular volume (AMV) in patients with type 2 diabetes without DME, after adjusting for age, sex, duration of diabetes, and HbA1c. Each 1 mmol/L increase in tHcy level was associated with a 6.57 µm greater CSMT (95% confidence interval [CI] 1.78, 11.36), a 4.51 µm greater AMT (95% CI 1.05, 7.98), and a 4.72 mm³ greater AMV (95% CI 1.23, 8.21).

CONCLUSIONS

Higher homocysteine levels were associated with an increased CSMT, AMT, and AMV in diabetic patients without DME. This link may indicate that patients with type 2 diabetes with increased levels of plasma tHcy are more prone to develop a clinical manifestation of DME.

Optical coherence tomography (OCT) for detection of macular oedema in patients with diabetic retinopathy

BACKGROUND

Diabetic macular oedema (DMO) is a thickening of the central retina, or the macula, and is associated with long-term visual loss in people with diabetic retinopathy (DR). Clinically significant macular oedema (CSMO) is the most severe form of DMO. Almost 30 years ago, the Early Treatment Diabetic Retinopathy Study (ETDRS) found that CSMO, diagnosed by means of stereoscopic fundus photography, leads to moderate visual loss in one of four people within three years. It also showed that grid or focal laser photocoagulation to the macula halves this risk. Recently, intravitreal injection of antiangiogenic drugs has also been used to try to improve vision in people with macular oedema due to DR.Optical coherence tomography (OCT) is based on optical reflectivity and is able to image retinal thickness and structure producing cross-sectional and three-dimensional images of the central retina. It is widely used because it provides objective and quantitative assessment of macular oedema, unlike the subjectivity of fundus biomicroscopic assessment which is routinely used by ophthalmologists instead of photography. Optical coherence tomography is also used for quantitative follow-up of the effects of treatment of CSMO.

OBJECTIVES

To determine the diagnostic accuracy of OCT for detecting DMO and CSMO, defined according to ETDRS in 1985, in patients referred to ophthalmologists after DR is detected. In the update of this review we also aimed to assess whether OCT might be considered the new reference standard for detecting DMO.

SEARCH METHODS

We searched the Cochrane Database of Systematic Reviews (CDSR), the Database of Abstracts of Reviews of Effects (DARE), the Health Technology Assessment Database (HTA) and the NHS Economic Evaluation Database (NHSEED) (The Cochrane Library 2013, Issue 5), Ovid MEDLINE, Ovid MEDLINE In-Process and Other Non-Indexed Citations, Ovid MEDLINE Daily, Ovid OLDMEDLINE (January 1946 to June 2013), EMBASE (January 1950 to June 2013), Web of Science Conference Proceedings Citation Index - Science (CPCI-S) (January 1990 to June 2013), BIOSIS Previews (January 1969 to June 2013), MEDION and the Aggressive Research Intelligence Facility database (ARIF). We did not use any date or language restrictions in the electronic searches for trials. We last searched the electronic databases on 25 June 2013. We checked bibliographies of relevant studies for additional references.

SELECTION CRITERIA

We selected studies that assessed the diagnostic accuracy of any OCT model for detecting DMO or CSMO in patients with DR who were referred to eye clinics. Diabetic macular oedema and CSMO were diagnosed by means of fundus biomicroscopy by ophthalmologists or stereophotography by ophthalmologists or other trained personnel.

DATA COLLECTION AND ANALYSIS

Three authors independently extracted data on study characteristics and measures of accuracy. We assessed data using random-effects hierarchical sROC meta-analysis models.

MAIN RESULTS

We included 10 studies (830 participants, 1387 eyes), published between 1998 and 2012. Prevalence of CSMO was 19% to 65% (median 50%) in nine studies with CSMO as the target condition. Study quality was often unclear or at high risk of bias for QUADAS 2 items, specifically regarding study population selection and the exclusion of participants with poor quality images. Applicablity was unclear in all studies since professionals referring patients and results of prior testing were not reported. There was a specific 'unit of analysis' issue because both eyes of the majority of participants were included in the analyses as if they were independent.In nine studies providing data on CSMO (759 participants, 1303 eyes), pooled sensitivity was 0.78 (95% confidence interval (CI) 0.72 to 0.83) and specificity was 0.86 (95% CI 0.76 to 0.93). The median central retinal thickness cut-off we selected for data extraction was 250 µm (range 230 µm to 300 µm). Central CSMO was the target condition in all but two studies and thus our results cannot be applied to non-central CSMO.Data from three studies reporting accuracy for detection of DMO (180 participants, 343 eyes) were not pooled. Sensitivities and specificities were about 0.80 in two studies and were both 1.00 in the third study.Since this review was conceived, the role of OCT has changed and has become a key ingredient of decision-making at all levels of ophthalmic care in this field. Moreover, disagreements between OCT and fundus examination are informative, especially false positives which are referred to as subclinical DMO and are at higher risk of developing clinical CSMO.

AUTHORS' CONCLUSIONS

Using retinal thickness thresholds lower than 300 µm and ophthalmologist's fundus assessment as reference standard, central retinal thickness measured with OCT was not sufficiently accurate to diagnose the central type of CSMO in patients with DR referred to retina clinics. However, at least OCT false positives are generally cases of subclinical DMO that cannot be detected clinically but still suffer from increased risk of disease progression. Therefore, the increasing availability of OCT devices, together with their precision and the ability to inform on retinal layer structure, now make OCT widely recognised as the new reference standard for assessment of DMO, even in some screening settings. Thus, this review will not be updated further.

Influence of diabetes on macular thickness measured using optical coherence tomography: the Singapore Indian Eye Study

PURPOSE

To determine the influence of diabetes, diabetic retinopathy (DR), and other factors on macular thickness, measured using optical coherence tomography (OCT), in a population-based sample.

METHODS

Data from the population-based Singapore Indian Eye Study were analyzed. We measured macular thickness using Stratus OCT Fast Macular Thickness scan protocol in 228 participants with diabetes mellitus (including 167 without DR, 44 with mild DR, 17 with moderate or severe DR) and 72 non-diabetic controls without macular oedema or other macular lesions. Analysis was done on right eyes.

RESULTS

The mean age of participants was 60.1 ± 10.1 years, with 53.8% men. Macular thickness measurements did not differ significantly between diabetic participants with no or mild DR and non-diabetic participants. Diabetic participants with moderate or severe DR had greater foveal and temporal outer macula thickness compared with those with no or mild DR (P=0.003). In a multivariate linear regression model, older age (P=0.009), male gender (P=0.005), and lower spherical equivalent (P=0.001) were other factors associated with greater foveal thickness in all participants after controlling for body mass index, glycosylated haemoglobin, total cholesterol, and mean systolic blood pressure.

CONCLUSION

This population-based study showed that diabetic participants with moderate or severe DR had thicker foveal measurements, even in the absence of diabetic macula oedema, than non-diabetic controls. Other factors that influenced macular thickness measurements were age, gender, and spherical equivalent. These data may aid the interpretation of OCT findings in persons with diabetes and DR.