Association Between Diabetic Macular Edema and Cardiovascular Events in Type 2 Diabetes Patients: A Multicenter Observational Study

Predictive model and risk analysis for diabetic retinopathy using machine learning: a retrospective cohort study in China

OBJECTIVE

Aiming to investigate diabetic retinopathy (DR) risk factors and predictive models by machine learning using a large sample dataset.

DESIGN

Retrospective study based on a large sample and a high dimensional database.

SETTING

A Chinese central tertiary hospital in Beijing.

PARTICIPANTS

Information on 32 452 inpatients with type-2 diabetes mellitus (T2DM) were retrieved from the electronic medical record system from 1 January 2013 to 31 December 2017.

METHODS

Sixty variables (including demography information, physical and laboratory measurements, system diseases and insulin treatments) were retained for baseline analysis. The optimal 17 variables were selected by recursive feature elimination. The prediction model was built based on XGBoost algorithm, and it was compared with three other popular machine learning techniques: logistic regression, random forest and support vector machine. In order to explain the results of XGBoost model more visually, the Shapley Additive exPlanation (SHAP) method was used.

RESULTS

DR occurred in 2038 (6.28%) T2DM patients. The XGBoost model was identified as the best prediction model with the highest AUC (area under the curve value, 0.90) and showed that an HbA1c value greater than 8%, nephropathy, a serum creatinine value greater than 100 µmol/L, insulin treatment and diabetic lower extremity arterial disease were associated with an increased risk of DR. A patient's age over 65 was associated with a decreased risk of DR.

CONCLUSIONS

With better comprehensive performance, XGBoost model had high reliability to assess risk indicators of DR. The most critical risk factors of DR and the cut-off of risk factors can be found by SHAP method to render the output of the XGBoost model clinically interpretable.

Deep Learning Predicts OCT Measures of Diabetic Macular Thickening From Color Fundus Photographs

Purpose

To develop deep learning (DL) models for the automatic detection of optical coherence tomography (OCT) measures of diabetic macular thickening (MT) from color fundus photographs (CFPs).

Methods

Retrospective analysis on 17,997 CFPs and their associated OCT measurements from the phase 3 RIDE/RISE diabetic macular edema (DME) studies. DL with transfer-learning cascade was applied on CFPs to predict time-domain OCT (TD-OCT)-equivalent measures of MT, including central subfield thickness (CST) and central foveal thickness (CFT). MT was defined by using two OCT cutoff points: 250 μm and 400 μm. A DL regression model was developed to directly quantify the actual CFT and CST from CFPs.

Results

The best DL model was able to predict CST ≥ 250 μm and CFT ≥ 250 μm with an area under the curve (AUC) of 0.97 (95% confidence interval [CI], 0.89-1.00) and 0.91 (95% CI, 0.76-0.99), respectively. To predict CST ≥ 400 μm and CFT ≥ 400 μm, the best DL model had an AUC of 0.94 (95% CI, 0.82-1.00) and 0.96 (95% CI, 0.88-1.00), respectively. The best deep convolutional neural network regression model to quantify CST and CFT had an R2 of 0.74 (95% CI, 0.49-0.91) and 0.54 (95% CI, 0.20-0.87), respectively. The performance of the DL models declined when the CFPs were of poor quality or contained laser scars.

Conclusions

DL is capable of predicting key quantitative TD-OCT measurements related to MT from CFPs. The DL models presented here could enhance the efficiency of DME diagnosis in tele-ophthalmology programs, promoting better visual outcomes. Future research is needed to validate DL algorithms for MT in the real-world.

The Association of Severe Diabetic Retinopathy With Cardiovascular Outcomes in Long-standing Type 1 Diabetes: A Longitudinal Follow-up

OBJECTIVE

It is well established that diabetic nephropathy increases the risk of cardiovascular disease (CVD), but how severe diabetic retinopathy (SDR) impacts this risk has yet to be determined.

RESEARCH DESIGN AND METHODS

The cumulative incidence of various CVD events, including coronary heart disease (CHD), peripheral artery disease (PAD), and stroke, retrieved from registries, was evaluated in 1,683 individuals with at least a 30-year duration of type 1 diabetes drawn from the Finnish Diabetic Nephropathy Study (FinnDiane). The individuals were divided into four groups according to the presence of diabetic kidney disease (DKD) and/or SDR (+DKD/+SDR, +DKD/-SDR, -DKD/+SDR, and -DKD/-SDR) at baseline visit. Furthermore, age-specific incidences were compared with 4,016 control subjects without diabetes. SDR was defined as laser photocoagulation and DKD as estimated glomerular filtration rate <60 mL/min/1.73 m2.

RESULTS

During 12,872 person-years of follow-up, 416 incident CVD events occurred. Even in the absence of DKD, SDR increased the risk of any CVD (hazard ratio 1.46 [95% CI 1.11-1.92]; P < 0.01), after adjustment for diabetes duration, age at diabetes onset, sex, smoking, blood pressure, waist-to-hip ratio, history of hypoglycemia, and serum lipids. In particular, SDR alone was associated with the risk of PAD (1.90 [1.13-3.17]; P < 0.05) and CHD (1.50 [1.09-2.07; P < 0.05) but not with any stroke. Moreover, DKD increased the CVD risk further (2.85 [2.13-3.81]; P < 0.001). However, the risk was above that of the control subjects without diabetes also in patients without microvascular complications, until the patients reached their seventies.

CONCLUSIONS

SDR alone, even without DKD, increases cardiovascular risk, particularly for PAD, independently of common cardiovascular risk factors in long-standing type 1 diabetes. More remains to be done to fully understand the link between SDR and CVD. This knowledge could help combat the enhanced cardiovascular risk beyond currently available regimens.

Influence of micro- and macro-vascular disease and Tumor Necrosis Factor Receptor 1 on the level of lower-extremity amputation in patients with type 2 diabetes

BACKGROUND

Patients with type 2 diabetes (T2D) face a high amputation rate. We investigated the relationship between the level of amputation and the presence of micro or macro-vascular disease and related circulating biomarkers, Tumor Necrosis Factor Receptor 1 (TNFR1) and Angiopoietin like-2 protein (ANGPTL2).

METHODS

We have analyzed data from 1468 T2D participants in a single center prospective cohort (the SURDIAGENE cohort). Our outcome was the occurrence of lower limb amputation categorized in minor (below-ankle) or major (above ankle) amputation. Microvascular disease was defined as a history of albuminuria [microalbuminuria: uACR (urinary albumine-to-creatinine ratio) 30-299 mg/g or macroalbuminuria: uACR ≥ 300 mg/g] and/or severe diabetic retinopathy or macular edema. Macrovascular disease at baseline was divided into peripheral arterial disease (PAD): peripheral artery revascularization and/or major amputation and in non-peripheral macrovascular disease: coronary artery revascularization, myocardial infarction, carotid artery revascularization, stroke. We used a proportional hazard model considering survival without minor or major amputation.

RESULTS

During a median follow-up period of 7 (0.5) years, 79 patients (5.5%) underwent amputation including 29 minor and 50 major amputations. History of PAD (HR 4.37 95% CI [2.11-9.07]; p < 0.001), severe diabetic retinopathy (2.69 [1.31-5.57]; p = 0.0073), male gender (10.12 [2.41-42.56]; p = 0.0016) and serum ANGPTL2 concentrations (1.25 [1.08-1.45]; p = 0.0025) were associated with minor amputation outcome. History of PAD (6.91 [3.75-12.72]; p < 0.0001), systolic blood pressure (1.02 [1.00-1.03]; p = 0.004), male gender (3.81 [1.67-8.71]; p = 0.002), and serum TNFR1 concentrations (HR 13.68 [5.57-33.59]; p < 0.0001) were associated with major amputation outcome. Urinary albumin excretion was not significantly associated with the risk of minor and major amputation.

CONCLUSIONS

This study suggests that the risk factors associated with the minor vs. major amputation including biomarkers such as TNFR1 should be considered differently in patients with T2D.

Clinical Decision-Making when Treating Diabetic Macular Edema Patients with Dexamethasone Intravitreal Implants

Diabetes mellitus (DM) is a metabolic disease frequently associated with comorbidities that include diabetic macular edema (DME). The current medical approach to treating DME involves intravitreal injections with either anti-vascular endothelial growth factors or steroids. However, the burden associated with intravitreal injections and DM-derived complications is high, underlining the need to find optimal treatment regimens. In this article we describe the considerations we apply when treating DME patients with dexamethasone intravitreal implants (Ozurdex®), particularly those that influence the clinical decision-making process during the follow-up period. These considerations are based both on the available medical literature and on our clinical experience following the use of these implants in this type of patient, the goal being to optimize the number of injections and the clinical outcome of this therapy. We also provide a general overview of the pathophysiology of DME, highlighting the inflammatory component as a rationale to use steroids in these patients.

LONG-TIME OUTCOME IN PATIENTS TREATED WITH RANIBIZUMAB FOR DIABETIC MACULAR EDEMA: A 4-Year Study

PURPOSE

To investigate the long-time visual outcome in patients with diabetic macular edema treated with ranibizumab in an ordinary clinical setting.

METHODS

One hundred two eyes of 80 patients were followed for 4 years. All patients received a loading dose of 3 monthly ranibizumab 0.5-mg injections. From Month 3 to Month 48, patients received ranibizumab reinjections pro re nata based on disease activity.

RESULTS

Excluding deaths, the 4-year visit was completed by 82% of the study eyes. The best-corrected visual acuity improved by 6.6 Early Treatment Diabetic Retinopathy Study letters at 4 years (P < 0.001). The patients received a mean of 7.7 ± 3.4 ranibizumab injections for 4 years. The number of injections (mean ± SD) given were 4.7 (1.1), 1.4 (1.4), 0.7 (1.1), and 0.9 (1.4) during Years 1 to 4, respectively. No difference in the injections needed was seen between patients who had previously received focal/grid laser and treatment-naive subjects.

CONCLUSION

The gain in the best-corrected visual acuity achieved after the initial loading dose was sustained over time with a pro re nata regimen. The number of injections needed to maintain the best-corrected visual acuity was low during the study period.

Diabetic macular oedema

Diabetic macular oedema, characterised by exudative fluid accumulation in the macula, is the most common form of sight-threatening retinopathy in people with diabetes. It affects one in 15 people with diabetes resulting in more than 20 million cases worldwide. Few epidemiological studies have been done to specifically investigate risk factors for diabetic macular oedema, although poor glycaemic and blood pressure control are associated with the presence and development of the disorder. The pathophysiological processes begin with chronic hyperglycaemia, and interplay between vascular endothelial growth factor (VEGF) and inflammatory mediators. Non-invasive imaging using optical coherence tomography has allowed clinicians to detect mild levels of diabetic macular oedema in order to monitor progress and guide treatment. Although focal or grid laser photocoagulation was the traditional mode of treatment, intraocular pharmacotherapy with anti-VEGF agents is now the standard of care. However, these therapies are expensive and resource intensive. Emerging therapeutic strategies include improving efficacy and duration of VEGF suppression, targeting alternative pathways such as inflammation, the kallikrein-kinin system, the angiopoietin-Tie2 system, and neurodegeneration, and using subthreshold and targeted laser therapy. Ongoing research should lead to improvements in screening, diagnosis, and management of diabetic macular oedema.