Background To investigate possible associations between diabetic retinopathy (DR) and systemic vascular endothelial function and arterial stiffness measured using reactive hyperaemia peripheral arterial tonometry.
Methods This was a cross-sectional observational clinical study. Subjects with diabetes were recruited and DR was graded from retinal photographs. Systemic endothelial function was measured using reactive hyperaemia peripheral arterial tonometry (EndoPAT) and expressed as the reactive hyperaemia index (RHI). Peripheral arterial stiffness was measured using the same device and expressed as the augmentation index (AI).
Results In total, 164 eyes of 95 Chinese patients were evaluated. The mean age of the subject eyes was 60.1±8.2 years and 76.8% were men. The mean duration of diabetes was 15.5±9.8 years, and the mean HbA1c was 8.1±1.4%. In age–gender-adjusted models, increasing severity of DR was associated with increasing mean RHI (p=0.001) and increasing mean AI (p<0.001). In multivariate models, adjusting additionally for smoking, mean duration of diabetes, HbA1c and hypertension, the associations with RHI and AI persisted (p=0.011 and 0.001, respectively). In analyses of the dichotomous outcomes clinically significant macular oedema (CSME), moderate DR and vision-threatening DR, AI was a significant predictor of CSME and vision-threatening DR. In multivariate-adjusted models, for every SD increase in AI, the odds of having CSME was 1.78 times higher (95% CI 1.05 to 2.99; p=0.029). For every SD increase in AI, the odds of having vision-threatening DR was 1.73 times higher (95% CI 1.17 to 2.56; p=0.003).
Conclusions Subjects with more severe DR have larger peripheral reactive hyperaemic responses and greater peripheral vascular stiffness. These findings support the link between the microvascular changes of diabetes and macrovascular disease.
- Diagnostic tests/Investigation
Statistics from Altmetric.com
Diabetic retinopathy (DR) is a microvascular complication of diabetes mellitus (DM) that is a leading cause of visual loss and blindness. Subjects with DR are also at increased risk of developing the macrovascular complications of diabetes. For example, in the Framingham Eye Study, subjects with DR were 14 times more likely to have cardiovascular disease. The frequent coexistence of the microvascular and macrovascular complications of diabetes suggests the possibility of a common pathogenic pathway.
There is increasing recognition that endothelial dysfunction may be an early step in the pathogenesis of diabetes and its complications, and may thus provide a link between its microvascular and macrovascular complications.1–4 Endothelial dysfunction is the earliest clinically identifiable event in the pathogenesis of cardiovascular disease. Endothelial dysfunction has been shown to precede the development of diabetes5 and linked to the development of the macrovascular complications of diabetes including cardiovascular risk. However, the associations of endothelial dysfunction with the microvascular complications of diabetes, and DR in particular, have been less consistent.
The most widely used method for assessing endothelial dysfunction is by measurement of flow-mediated dilation (FMD) of the brachial artery after a period of occlusion of blood flow. FMD measurement is however operator dependent and may be affected by measurement-induced changes in systemic haemodynamics. Studies using FMD have thus far identified inconsistent relationships between DR and endothelial dysfunction, with some reporting reduced FMD in DR,1 ,6–8 and others reporting no relationship.3 ,9
The EndoPAT (Itamar Medical, Israel) is a novel device for the measurement of endothelial dysfunction in vivo. The device is a form of reactive hyperaemia peripheral arterial plethysmography that measures the vascular endothelial response to temporary vascular deprivation in the arm by quantifying blood volume in the finger. The advantages of the device over conventional FMD include standardised measurements that are non-operator dependent, and simultaneous measurement of responses in the fellow arm to serve as a control for measurement-induced haemodynamic changes. The device is also able to derive the augmentation index (AI), a measure of peripheral arterial stiffness and macrovascular disease. Data from large population-based studies including the Framingham Heart Study have shown associations between EndoPAT indices of endothelial function and vascular stiffness and DM, but associations with DR have not been explored.10
The aim of this study was to investigate possible associations between the severity of diabetic microvascular disease as indicated by the presence and severity of DR, and systemic vascular endothelial cell dysfunction and peripheral arterial stiffness as measured by the EndoPAT.
This was a cross-sectional observational study. Adult subjects with diabetes were recruited prospectively from the DR screening clinics at a tertiary eye care centre. All subjects were of Chinese ethnicity. Subjects were given standardised instructions on the preparations required prior to the study visit. These included no alcohol consumption for 48 h prior to the study visit; no cardioactive medications for four half-lives or a maximum of 48 h prior to the study visit; no tea/coffee for 24 h prior to the study visit; no exercise for 12 h prior to the study visit and no food for 12 h (overnight fast) prior to the study visit. On the morning of the study visit, no smoking was allowed and all study investigations were performed between 08:00 and 11:00.
Assessment of diabetic retinopathy
Retinal photography was performed using a digital retinal camera (Canon CR-DGi with a 10-D SLR back; Canon, Tokyo, Japan), following a standardised protocol.11 Photographs were then graded by trained, masked graders at the Singapore Eye Research Institute.
DR was considered present if any characteristic lesion as defined by the Early Treatment Diabetic Retinopathy Study (ETDRS) severity scale was present.12 For each eye, a retinopathy severity score was assigned according to a scale modified from the Airlie House classification system.12 ,13 Macular oedema (ME) was defined by hard exudates in the presence of microaneurysm and blot haemorrhage within 1 disc diameter from the foveal centre or the presence of focal photocoagulation scars in the macular area. Clinically significant macular oedema (CSME) was considered present when ME was within 500 µm of the foveal centre or if focal laser photocoagulation scars were present in the macular area. Retinopathy severity was categorised as minimal non-proliferative diabetic retinopathy (NPDR; ETDRS levels 14–20), mild NPDR (level 35), moderate NPDR (levels 43–47), severe NPDR (level 53) and proliferative retinopathy (level 61 or higher).
We defined two additional outcomes for analyses. Moderate DR was defined as moderate NPDR, and vision-threatening retinopathy as the presence of severe NPDR, proliferative retinopathy or CSME using the Eye Diseases Prevalence Research Group definition.14 If an eye was ungradable, it was excluded from that particular analysis.
Assessment of endothelial function and augmentation index using EndoPAT
The EndoPAT is a type of computerised plethysmograph.15 A sensor was applied to a finger on each hand. An inflatable blood pressure cuff was applied to one upper arm. A baseline measurement of finger volumes (pulse amplitudes) in the fingers to which sensors had been applied was recorded for 5 min. The arm cuff was then inflated to 60 mm Hg above systemic systolic blood pressure or 200 mm Hg, whichever is higher, for 5 min. The cuff was then deflated and changes in finger blood volume recorded from each finger for four more minutes. The period of circulatory deprivation in one arm is followed in normal subjects by a marked nitric oxide(NO)-induced vasodilatation in the ipsilateral finger 90–120 s after cuff deflation.
This vasodilatation following a period of arm occlusion is due to a local release of NO and prostaglandins as a response to shear stress and/or tissue hypoxia. As some of the released mediators enter the systemic circulation, a smaller increase in finger blood volume in the non-ischaemic arm occurs. The computer program identifies the preocclusion and postocclusion differences in finger blood volume in each hand and calculates a reactive hyperaemia index (RHI) that is a measure of the endothelial response to occlusion and reactive hyperaemia corrected for systemic effects on the basis of volume changes in the contralateral finger. AI, a measure of vascular stiffness, is also automatically generated, as well as AI@75, which is AI standardised to a heart rate of 75 bpm.
Other study evaluations
Fasting venous blood samples were collected for biochemistry tests, including serum lipids (total cholesterol, high-density lipoprotein, low-density lipoprotein cholesterol), glycosylated haemoglobin A1c (HbA1C), creatinine and glucose.
Models were based on analyses by eyes using generalised estimating equation models to account for correlations between the two eyes. Multivariable logistic regression models were constructed with the DR outcomes as the dependent variables to assess the relationship with RHI, AI and AI@75. The natural logarithm of RHI, which is closer to a normal distribution, was also used as it has been shown to provide a better double-sided distribution than RHI itself. Initial adjustments were made for age and gender. Smoking, mean duration of diabetes, glycosylated haemoglobin and hypertension were added in a second multivariable model.
In total, 164 eyes of 95 patients with DR were evaluated. The mean age of the subject eyes was 60.1±8.2 years, the majority were men (76.8%) and they were all of Chinese ethnicity. The mean duration of diabetes was 15.5±9.8 years, and the mean HbA1c level was 8.1±1.4% (table 1).
In age–gender-adjusted models, increasing severity of DR was associated with increasing mean RHI and mean lnRHI (p=0.001 and 0.004, respectively). Increasing severity of DR was also associated with increasing mean AI and mean AI@75 values (p<0.001 for both). In multivariate-adjusted models with the addition of smoking, mean duration of diabetes, HbA1c and hypertension as covariates, the significant associations with RHI and lnRHI persisted (p=0.011 and 0.027, respectively), as did the associations with AI and AI@75 (p=0.001 and 0.004, respectively) (table 2).
Neither RHI nor lnRHI was significantly associated with the presence of CSME. AI was a significant predictor of CSME. In multivariate-adjusted models, for every SD increase in AI, the odds of having CSME was 1.78 times higher (95% CI 1.05 to 2.99; p=0.029). In analyses of the dichotomous outcomes moderate DR and vision-threatening DR, RHI was a significant predictor of vision-threatening DR in age–gender-adjusted models. For every SD increase in RHI, the odds of having vision-threatening DR was 1.59 times higher (95% CI 1.05 to 2.42; p=0.03). However, this association was nullified after multivariate adjustment (p=0.11). AI and AI@75 were significant predictors of vision-threatening DR. In age–gender-adjusted models, for every SD increase in AI, the odds of having vision-threatening DR was 2.00 times higher (95% CI 1.29 to 3.11; p=0.002). For every SD increase in AI@75, the odds of having vision-threatening DR was 2.47 times higher (95% CI 1.58 to 3.85; p<0.001). These associations persisted with multivariate adjustment. For every SD increase in AI, the odds of having vision-threatening DR was 1.73 times higher (95% CI 1.17 to 2.56; p=0.003). For every SD increase in AI@75, the odds of having vision-threatening DR was 2.02 times higher (95% CI 1.22 to 3.34; p=0.02) (table 3).
Our study has shown associations between peripheral microvascular and macrovascular changes and DR in an adult population with diabetes. Subjects with more severe DR had greater peripheral reactive hyperaemic responses, as well as greater peripheral vascular stiffness.
Studies evaluating the association between endothelial dysfunction measured using FMD and DR have yielded conflicting results. Gibney et al6 evaluated the forearm vasodilatory response to reactive hyperaemia (RH) in patients with type 1 diabetes. RH was evaluated using strain-gauge plethysmography, and subjects with retinopathy showed reduced forearm RH responses compared with those with no retinopathy. Yun et al7 assessed the relationship between brachial artery FMD and DR in 167 subjects with diabetes. FMD responses were lower in subjects with DR compared with those without, and participants with reduced FMD were 11.8 times more likely to have DR. Brachial artery FMD has also been evaluated by Sogawa et al.8 In a cohort of 74 subjects with varying DR severity, FMD was significantly decreased in patients with DM compared with healthy control subjects. While no significant differences were found among the groups with no DR, mild NPDR and moderate NPDR, FMD was significantly decreased in the severe NPDR and PDR groups compared with the group with no DR. Malecki et al1 also evaluated endothelial function by FMD and found diminished responses in subjects with DR. Some studies have however yielded conflicting results. In the Hoorn study,3 endothelial function was measured using brachial artery FMD. The study found that the presence of retinopathy was not significantly associated with FMD responses. Similarly, Ugurlu et al9 analysed FMD responses in patients with type 1 diabetes. Subjects with diabetes and no DR, diabetes and early DR and normal controls were compared, with no differences found in any of the markers of endothelial dysfunction.
Operator-dependent variability and the lack of standardised measurement protocols are likely to have contributed to the inconsistencies in study outcomes. By using EndoPAT, we hoped to minimise these factors by employing a non operator-dependent, internally controlled measurement device. Our study showed increasing reactive hyperaemic responses with increasing DR severity. These findings are somewhat surprising as they suggest that endothelial function is enhanced in subjects with more severe DR. A possible biological explanation could be the autonomic dysregulation that occurs in subjects with more severe diabetes, analogous to the increased blood flow from arteriovenous shunting associated with the diabetic Charcot foot despite macrovascular occlusive disease. However, it is also uncertain whether EndoPAT measures endothelial function in a physiologically similar manner as FMD.16 While some studies have validated RHI as a marker of endothelial function, it reflects alterations in flow and digital microvascular dilatation that may not be purely NO dependent.17 As with FMD, the total shear-stress stimulus may also be important in determining dilatory response, but this cannot be assessed with the EndoPAT technique.18 As such, RHI may not be a simple index of endothelial function. Dhindsa has shown that EndoPAT measurements are only moderately related to FMD measurements.19 In an assessment of the ability of EndoPAT to detect changes in endothelial function induced by interventions such as smoking or glucose loading, Moerland was also unable to demonstrate significant changes in RHI parameters.20 Subjects with diabetes and renal impairment were also found to have similar or greater RHI readings compared with normal controls, similar to the increased RHI readings we found in subjects with more severe DR. The exaggerated reactive hyperaemic responses seen in subjects with more severe microvascular disease may represent endothelial dysfunction rather than enhanced endothelial function. Subjects with more severe diabetes have been shown to have a degree of relative tissue hypoxia as oxygen is more strongly bound to glycosylated haemoglobin with a resulting shift in the oxygen dissociation curve to the right and decreased availability of oxygen to tissues.21 Hypoxia can alter the molecular biology of vascular endothelium, most notably by upregulation of all three NO synthases (iNOS, nNOS and eNOS).22 Relative tissue hypoxia from diabetes could increase the sensitivity of affected endothelium to a further hypoxic insult such as occurs with the use of EndoPAT with an increased production of NO and a correspondingly increased vasodilation response.
DM is associated with increased peripheral arterial stiffness, a factor that may contribute to increased cardiovascular morbidity and mortality. AI is derived from the arterial pulse waveform, determined by the ratio between augmentation pressure and pulse pressure. This ratio represents the enhancement (augmentation) of the central aortic pressure by the peripherally reflected pulse wave, and is thus intimately related to the biomechanical properties of the arterial wall.23 Measurements of AI using various devices have shown differences between diabetic subjects and non-diabetic subjects. Wilhelm et al24 showed increased AI and arteriosclerosis in subjects with DM and in non-diabetic subjects with cardiovascular disease. In a comparison of young patients with type 1 diabetes and normal controls, Palombo et al25 showed that diabetic patients had higher AI values than controls. There is limited data on the relationships between microvascular disease and peripheral vascular stiffness. Araszkiewicz et al26 evaluated the relationship between DR and peripheral AI in 87 patients with type 1 diabetes. Patients with retinopathy had increased AI even after adjustment for numerous systemic confounders. Our study has shown similar associations between more severe DR and greater arterial stiffness. These findings support the notion that the microvascular changes of DR may be a risk marker for the macrovascular complications of diabetes and additionally that increasing AI values in diabetics may be a risk factor for sight-threatening DR.
Our study benefited from standardised assessments of DR severity from retinal photography as well as systemic covariates. General limitations of our study include the relatively small sample size, the possibility of selection bias and the cross-sectional design that restricts inferences of causality. Sample size calculations were based on FMD differences observed in studies with significant findings. The reproducibility of EndoPAT has also been shown to be only moderate.27 However, this is a non-differential bias that tends to bias the results towards null instead of a positive association.
In conclusion, our study has shown associations between the microvascular changes of DR and greater reactive hyperaemic responses as well as greater peripheral vascular stiffness in the peripheral circulation of subjects with more severe DR. Further study is needed to define the role of endothelial function measured using EndoPAT in the management of DR.
Contributors Design and conduct of the study: LSL, LHL, WF and TYW. Collection, management, analysis and interpretation of the data: LSL, LHL, CMGC, PGO, LG, EST, RM, DW, WF and TYW. Preparation, review or approval of the manuscript: LSL, LHL, CMGC, EST, RM, DW, WF and TYW.
Funding This study was supported by National Medical Research Council grant R710/60/2009.
Competing interests None.
Patient consent All study procedures were performed in accordance with the tenets of the Declaration of Helsinki as revised in 1989..
Ethics approval Singapore Eye Research Institute IRB.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.