Open access Original research BMJ Open Diab Res Care: first published as 10.1136/bmjdrc-2019-000845 on 29 December 2019. Downloaded from http://drc.bmj.com/ on February 19, 2021 at Univesity Autonomous of Diabetic retinopathy as an independent predictor of subclinical cardiovascular disease: baseline results of the PRECISED study Rafael Simó ‍ ‍,1,2 Jordi Bañeras,3,4 Cristina Hernández,1,2 José Rodríguez-­Palomares,3,4 Filipa Valente,3,4 Laura Gutierrez,3,4 Teresa González-­Alujas,3,4 Ignacio Ferreira,3,4 Santiago Aguadé-Bruix,5 Joan Montaner,6 Daniel Seron,7,8 Joan Genescà,9,10 Anna Boixadera,11 José García-­Arumí,11 Alejandra Planas,1 Olga Simó-Servat,1,2 David García-­Dorado3,4 To cite: Simó R, Bañeras J, Abstract Hernández C, et al. Objective Detection of subclinical cardiovascular disease Significance of this study Diabetic retinopathy as an (CVD) has significant impact on the management of type independent predictor of 2 diabetes. We examined whether the assessment of What is already known about this subject? subclinical cardiovascular ►► Diabetic retinopathy is associated to cardiovascular Barcelona. Protected by copyright. diabetic retinopathy (DR) is useful for identifying patients at disease: baseline results events in diabetic population. of the PRECISED study. a higher risk of having silent CVD. BMJ Open Diab Res Care Research design and methods Prospective case–control What are the new findings? 2019;7:e000845. doi:10.1136/ study comprising 200 type 2 diabetic subjects without history ►► Diabetic retinopathy is a powerful and independent bmjdrc-2019-000845 of clinical CVD and 60 age-­matched non-­diabetic subjects. risk factor for identifying diabetic subjects with sub- The presence of subclinical CVD was examined using two clinical cardiovascular disease. parameters: (1) calcium coronary score (CACs); (2) composite ►► The presence of diabetic retinopathy confers a higher ►► Additional material is of CACs >400 UA, carotid plaque ≥3 mm, carotid intima– published online only. To view risk of subclinical cardiovascular disease than factors media thickness ratio >1, or the presence of ECG changes contained in contemporary risk equations such as please visit the journal online suggestive of previous asymptomatic myocardial infarction. blood pressure, low-­density lipoprotein (LDL) choles- (http://​dx.​doi.​org/​10.​1136/​ In addition, coronary angio-­CT was performed. DR was bmjdrc-​2019-​000845). terol and HbA1c. assessed by slit-­lamp biomicroscopy and retinography. ►► There is a clear relationship between the presence and Results Type 2 diabetic subjects presented higher CACs degree of diabetic retinopathy and coronary stenosis. In the memory of the beloved than non-­diabetic control subjects (p<0.01). Age, male Dr. Garcia-­Dorado. Passed gender, and the presence of DR were independently How might these results change the focus of away on August 16th 2019. related to CACs >400 (area under the receiver operating research or clinical practice? characteristic curve (AUROC) 0.76). In addition, an inverse ►► The presence and degree of diabetic retinopathy RS and IF contributed equally. relationship was observed between the degree of DR and should be considered a better tool for identifying type Received 22 August 2019 CACs <10 AU. The variables independently associated 2 diabetic subjects at risk of cardiovascular disease Revised 1 November 2019 with the composite measurement of subclinical CVD than the traditional risk factors. Accepted 11 November 2019 were age, diabetes duration, the glomerular filtration rate, ►► Type 2 diabetic patients with diabetic retinopathy and microalbuminuria, and the presence of DR (AUROC 0.71). in particular those patients with moderate–severe In addition, a relationship (p<0.01) was observed between degree represent a good target population for a cost-­ the presence and degree of DR and coronary stenosis. effective cardiovascular screening. © Author(s) (or their Conclusions The presence and degree of DR is ►► The detection and grading of diabetic retinopathy per- employer(s)) 2019. Re-­use independently associated with subclinical CVD in type mit us to identify a high-­risk subset of diabetic pop- permitted under CC BY-­NC. No 2 diabetic patients. Our results lead us to propose a ulation who might benefit most from tight control of commercial re-­use. See rights rationalized screening for coronary artery disease in type 2 and permissions. Published cardiovascular risk factors. In addition, given that these diabetes based on prioritizing patients with DR, particularly by BMJ. patients are more prone to develop cardiovascular those with moderate–severe degree. For numbered affiliations see disease, they could be used to enrich the cohorts for end of article. future intervention trials, thus reducing sample size, duration, and costs of studies. Correspondence to Introduction Dr Rafael Simó; Diabetes mellitus (DM) is one of the leading ​rafael.​simo@v​ hir.​org causes of mortality and major morbidities, blindness. In fact, diabetic complications Dr Ignacio Ferreira; including cardiovascular disease (CVD), account for much of the social and financial ​nachoferreira@​secardiologia.​es kidney failure, lower limb amputation, and burden of diabetes.1 2 BMJ Open Diab Res Care 2019;7:e000845. doi:10.1136/bmjdrc-2019-000845 1 Cardiovascular and Metabolic Risk BMJ Open Diab Res Care: first published as 10.1136/bmjdrc-2019-000845 on 29 December 2019. Downloaded from http://drc.bmj.com/ on February 19, 2021 at Univesity Autonomous of Patients with diabetes generally have a greater extent widespread screening for silent CAD in diabetes cannot of atherosclerosis with a higher prevalence of multivessel be recommended at this time19 and, consequently, the coronary artery disease (CAD) and infarction compared identification of a more targeted population in which the with the non-­ DM population.3 4 Once CAD manifests CACs would be more cost-­efficient seems warranted. clinically, diabetic patients continue to have a worse prog- Emerging evidence indicates that microangiopathy nosis compared with non-­diabetic patients, both during is independently associated with an increased risk of acute ischemic events and during long-­term follow-­up.5 CV events in diabetic patients.20 A recent population-­ However, the increase in CVD risk is not homogeneous based cohort study showed that the cumulative burden in patients with DM. In fact, a significant number of these of microvascular disease increases the risk of future patients will never experience CV complications, and CVD among individuals with type 2 diabetes.21 However, current guidelines consistently advise against the routine whereas microalbuminuria and macroalbuminuria are use of prophylactic aspirin in patients with DM.6 7 There- frequently included among the risk factors in the studies fore, the early identification of diabetic patients at risk of addressed to evaluate CV events, the presence and degree developing CVD remains a challenge.8 9 of diabetic retinopathy (DR) is often missing. The high prevalence of type 2 diabetes precludes imple- On this basis, the main aim of the present study was to menting a generalized screening for CVD in asymptom- determine whether the assessment of DR could help in atic patients.6 7 On the other hand, the traditional risk identifying those in the diabetic population who should factors associated with diabetes (ie, hypertension, dyslip- be candidates for CACs assessment. Since the screening idemia, obesity) are estimated to contribute in less than of DR is an essential part of diabetes care, this approach 40% of CV events. In fact, these factors are not useful for would not increase the economic burden associated with identifying the coronary asymptomatic patients in which diabetes management. In addition, the usefulness of DR the screening could be cost-­effective.8 9 Therefore, cheap in identifying subclinical CVD measured by the composite and routinely available measures that identify those of a CACs >400, carotid plaque ≥3 mm, CIMT >1, or the patients at high risk of developing CVD are needed. presence of ECG changes suggestive of previous asymp- Barcelona. Protected by copyright. Carotid ultrasound to measure carotid intima–media tomatic myocardial infarction (MI) was also examined. In thickness (CIMT) and a coronary artery calcium scan the present work, we show the first results of the baseline to quantify coronary calcification have been used as data of the PRECISED study which is mainly addressed to non-­invasive techniques to identify subclinical athero- further clarify the impact of the presence and degree of sclerosis.10 CIMT has been widely used as a marker to microangiopathy on CV events in type 2 diabetic patients. identify subclinical atherosclerosis, as increased CIMT has been associated with a high prevalence of CAD and future CV events.3 A recent study has found that carotid Research design and methods plaque is predictive of underlying silent coronary athero- Study design, data source, and patient enrollment sclerosis prevalence, severity, and extent in asymptomatic This was a prospective case–control study comprising a type 2 diabetic patients.11 Extensive data indicate a close total of 200 type 2 diabetic subjects with no history of relationship between the coronary artery calcium score clinical CVD and 60 non-­diabetic subjects matched by (CACs) and clinical coronary events among individuals age (control group) (PRECISED study: C ​ linalTrial.​gov with and without diabetes.12–15 In addition, it has been NCT02248311). shown that the prognostic significance of elevated CACs The inclusion criteria were (1) age from 50 to 79 in predicting coronary events appears to be greater in years and (2) a history of type 2 diabetes diagnosed at patients with diabetes than in those without diabetes, least 1 year prior to the day of screening. The exclusion and it can further enhance the predictions provided by criteria were (1) a medical history of a CV event, (2) type established risk models.15 16 Moreover, the CACs has been 1 diabetes, (3) any contraindication for the performance related to moderate to large perfusion defects assessed by of PET/CT or MRI, and (4) any concomitant disease scintigraphy.4 For all these reasons, the Imaging Council associated with a short life expectancy. of the American College of Cardiology concluded that The subjects were recruited from the Outpatient CAC screening is the most sensitive risk stratification tool Diabetic Clinic of Vall d’Hebron Hospital and the Primary among asymptomatic persons with diabetes.17 Patients Healthcare centers within its catchment area (North are typically stratified by Agatston units (AU), yielding Barcelona). A total of 2631 clinical records of subjects a CACs <100 (low risk), 100 to 400 (moderate risk), with type 2 diabetes were reviewed: 1912 did not meet and >400 (high risk). Notably, the finding of a CACs of the inclusion criteria, leaving 719 eligible for screening; <10 AU may facilitate risk stratification by enabling the out of these, 200 subjects with type 2 diabetes agreed to identification of people at very low risk within the overall participate in the study. A total of 60 age-­matched non-­ high-­risk population of diabetic patients.18 In this regard, diabetic subjects without a history of CVD served as a it should be noted that the absence of coronary calcium control group. These individuals were recruited from the portends a remarkably favorable prognosis despite the same Primary Healthcare centers and most were relatives presence of DM, with no patients experiencing adverse of diabetic subjects. A flowchart showing the recruitment cardiac events during 5 years of follow-­up.15 Nevertheless, process is displayed in online supplementary figure S1. 2 BMJ Open Diab Res Care 2019;7:e000845. doi:10.1136/bmjdrc-2019-000845 Cardiovascular and Metabolic Risk BMJ Open Diab Res Care: first published as 10.1136/bmjdrc-2019-000845 on 29 December 2019. Downloaded from http://drc.bmj.com/ on February 19, 2021 at Univesity Autonomous of Classic risk factors (common carotid artery, internal and external carotid Anthropometric measurements and classical risk factors artery) by means of a high-­frequency linear probe (vivid Clinical data were obtained on the first visit by an endo- 7-­GE and vivid 9, Medical Systems, GE Healthcare, with crinologist. Anthropometric data were obtained by stan- linear probe 7 MHz). dardized protocols at the same visit. Weight and height Conventional ultrasound (B-­mode and color Doppler) were measured with a balance with a fixed stadiometer was used to measure common IMT and to identify the to determine the body mass index (BMI). Hypertension presence of carotid plaques according to the Mannheim was defined as systolic blood pressure ≥140 mm Hg or consensus.24 Extracranial carotid arteries were explored diastolic blood pressure ≥90 mm Hg, or when patients axially and longitudinally. We measured the IMT of the were under treatment with antihypertensive agents. far wall of the common carotid artery at the level of 1 cm A history of smoking habits (non-­ smoker/current proximal to the bifurcation bilaterally. We used the mean smoker/ex-­smoker) was recorded. Smokers who stopped value obtained from three measurements for our anal- smoking ≥1 year prior to recruitment were considered ysis. Frequency of carotid plaques was defined as the ex-­smokers. presence of plaques in any of the explored territories. Dyslipidemia was defined by the use of lipid-­lowering The assessment of all the measurements and the perfor- drugs, decreased values of high-­density lipoprotein (HDL) mance of the ultrasound studies were performed by two cholesterol (men <0.9 mmol/L, women <1.0 mmol/L), researchers (TG-­A and LG), who were blinded to the or by at least one increased value of total cholesterol conditions of the participants. (>5.2 mmol/L), low-­ density lipoprotein (LDL) choles- terol (>4.3 mmol/L), or triglycerides (>1.7 mmol/L). CT-CAC scanning Framingham and United Kingdom Prospective Diabetes An ECG synchronized prospective non-­enhanced cardiac Study-­coronary heart disease (UKPDS-­CHD) were calcu- CT was performed with Siemens Biograph mCT 64 s lated as described elsewhere.22 equipment. The calcium score was analyzed using semi-­ automatic methodology with “Syngo.Via” cardiac CT Barcelona. Protected by copyright. Fundus examination software, the global and individual Agatston score for DR was evaluated by experienced ophthalmologists in coronary vessels being calculated. Patients were classi- mydriasis using slit-­lamp biomicroscopy and retinog- fied by AU as low risk (<100 AU), moderate risk (100–400 raphy with the same camera (Topcon-­ DRI-­OCT-­ AU), or high risk (>400 AU).18 TRITON). The examiners classified DR according to the International Clinical Diabetic Retinopathy Disease Severity Scale23: (1) no apparent retinopathy, (2) mild Coronary angio-CT non-­ proliferative retinopathy (NPDR), (3) moderate After patient preparation with beta blockers for NPDR, (4) severe NPDR, and (5) proliferative diabetic decreased heart rate, and nitroglycerin for vasodilatation retinopathy (PDR). if required, an ECG synchronized prospective contrast-­ enhanced coronary CT was performed with Siemens Laboratory tests Biograph mCT 64 s equipment. Automatic coronary A venous blood sample was collected from the antecu- vessel extraction of all coronary vessels with visual anal- bital vein, separated by centrifugation (2000×g at 4°C for ysis of coronary stenosis was performed by researchers 20 min) and frozen at −80°C for batched storage and anal- blind to the patient’s condition with “Syngo.Via” cardiac ysis. Glycosylated hemoglobin was determined by the Cobas CT software as described elsewhere.25 B 101 (Roche) system. The remaining biochemical param- eters were measured using an Olympus AU5400 automatic Statistical analysis biochemistry analyzer (Olympus, Tokyo, Japan). The categorical variables are presented as percentages. Assessment of subclinical CVD For the quantitative variables, the mean and SD are Subclinical CVD was defined as the composite of a CACs displayed, except for triglycerides, lipoprotein (a), and >400, carotid plaque ≥3 mm, or CIMT >1. The presence homocysteine in which median and range were used. of ECG changes suggestive of previous asymptomatic Differences among groups were assessed using the χ2 MI (Minnesota codes 1.1. and 1.2) was also considered test for qualitative variables, while t-­test quantitative vari- subclinical CVD. ables with a normal distribution, and non-­ parametric tests were used for those quantitative variables without a Electrocardiogram normal distribution. A 12-­lead ECG was taken after the patient had been lying Diabetic subjects were divided into two groups: those down for at least 5 min. The ECG was evaluated by a with calcium scores higher and lower than 400 AU. cardiologist. Logistic regression analysis to predict a CACs higher than 400 AU was performed using the variables that were Carotid ultrasonography significant at the univariant analysis. ROC curves were All study participants underwent a standard echo-­color calculated and the χ2 test for ROC area comparison was Doppler examination of the extracranial carotid arteries performed, the complete logistic model being compared BMJ Open Diab Res Care 2019;7:e000845. doi:10.1136/bmjdrc-2019-000845 3 Cardiovascular and Metabolic Risk BMJ Open Diab Res Care: first published as 10.1136/bmjdrc-2019-000845 on 29 December 2019. Downloaded from http://drc.bmj.com/ on February 19, 2021 at Univesity Autonomous of Table 1 Characteristics of subjects with type 2 diabetes and non-­diabetic control subjects Type 2 diabetes (n=186) Control group (n=57) P value Sex (woman) (n, %) 107 (57.53%) 37 (64.91%) 0.32 Ethnicity (Caucasian n, %) 178 (95.7%) 56 (98.25%) 0.64 Age (years) 65.7±6.47 66.02±6.63 0.75 2 BMI (kg/m ) 30.25±4.9 26.83±4.77 <0.00001 Waist circumference (cm) 103.9±13.53 91.2±13.92 <0.00001 Smoking 0.59  No (n, %) 99 (53.23%) 34 (59.65%)  Current smoker (n, %) 25 (13.44%) 7 (12.3%)  Ex-­smoker (n, %)  62 (33.33%)  15 (26.32%) CV family history (n, %)  22 (11.83%)  8 (14.04%) 0.65 Hypertension (n, %)  134 (72.04%)  28 (49.12%) 0.001 Use of ARB/ACEi (n, %)  118 (63.44%)  18 (31.58%) <0.0001 Dyslipidemia (n, %)  148 (78.57%)  25 (43.86%) <0.00001 Use of statins (n, %)  133 (71.51%)  19 (31.67%) <0.00001 Use of ezetimibe (n, %)  10 (5.38%)  0 (0%) 0.074 HbA1c (%)  7.43±1.18  5.55±0.31 <0.00001 HbA1c (mmol/mol)  57.85±12.96  37.12±3.31 Creatinine (mg/dL)  0.82±0.24  0.76±0.19 0.76 Barcelona. Protected by copyright. GFR mL/min  81.76±16  85.58±10.88 0.09 AST (IU/L)  25.51±15.71  23.48±5.73 0.34 ALT (IU/L)  25.94±16.88  21.12±10.55 0.043 GGT (IU/L)  44.46±71.82  31.04±29.77 0.17 Data are expressed as % or mean±SD. ACEi, ACE inhibitors; ALT, alanine transaminase; ARB, angiotensin receptor blockers; AST, aspartate aminotransferase; BMI, body mass index; CV, cardiovascular; GFR, glomerular filtration rate; GGT, gamma-­glutamyl transferase. with a logarithmic transformation of the microalbumin- CVD, or in laboratory parameters of kidney function uria:creatinine ratio. (creatinine and glomerular filtration rate (GFR), aspar- The differences between diabetic subjects with the tate aminotransferase (AST) and gamma-­glutamyl trans- combined endpoint defined above and those without it ferase (GGT). Subjects with type 2 diabetes presented were assessed. Selected variables that were significant at a significantly higher BMI and waist circumference, as the univariate analyses were used to predict the combined well as a higher prevalence of hypertension and dyslipid- endpoint by forward stepwise selection method in a emia. Therefore, as expected, significantly more subjects binary logistic regression analysis. ROC curves were were treated with angiotensin receptor blockers, ACE calculated. The area under the receiver operating char- inhibitors, and statins in the diabetic group in compar- acteristic curves (AUROCs) for the UKPDS-­ CHD and ison with non-­diabetic controls. As expected, HbA1c Framingham risk scores for predicting the presence of was significantly higher in the diabetic population than the combined endpoint and the effect of the addition of in control subjects. We also found significantly higher DR and/or microalbuminuria were calculated. levels of alanine transaminase (ALT) in diabetic subjects Significance was accepted at the level of p value <0.05 in comparison with controls. for all the analyses. Statistical analyses were performed The specific characteristics of subjects with diabetes are with the Stata statistical package. displayed in online supplementary table S1. In summary, the type 2 diabetic patients included in the study exhib- Results ited a long-­ term duration of diabetes with relatively Clinical characteristics good metabolic control (HbA1c: 57.85±12.56 mmol/mol The general characteristics as well as the main labora- (7.43%±1.18%)). Around 60% were under combined tory findings in subjects with type 2 diabetes and non-­ treatment with metformin plus insulin or insulin alone, diabetic controls are shown in table 1. We did not find and up to 30% presented at least one microangiopathic any significant difference between groups regarding complication. One out of 200 included patients (0.5%) age, gender, ethnicity, smoking habit, family history of presented ECG signs of a previous MI. 4 BMJ Open Diab Res Care 2019;7:e000845. doi:10.1136/bmjdrc-2019-000845 Cardiovascular and Metabolic Risk BMJ Open Diab Res Care: first published as 10.1136/bmjdrc-2019-000845 on 29 December 2019. Downloaded from http://drc.bmj.com/ on February 19, 2021 at Univesity Autonomous of Table 2 Characteristics of subjects with type 2 diabetes according to the calcium score Calcium score <400 n=137 Calcium score ≥400 n=41 P value Sex (woman) (n, %) 85 (62.04%) 18 (43.90%) 0.039 Age (years) 64.83±6.69 68.24±4.83 0.002 2 BMI (kg/m ) 30.01±5.02 31.42±4.62 0.11 Diabetes duration (years) 13.60±9.17 16.54±9.36 0.075 Waist circumference 102.87±13.81 108.07±12.86 0.034 Smoking 0.69  No 76 (55.47%) 19 (46.34%)  Current smoker 20 (14.60%) 5 (12.20%)  Ex-­smoker (n, %) 41 (29.93%) 17 (41.46%) Hypertension (n, %) 94 (68.61%) 33 (80.49%) 0.14 Dyslipidemia (n, %) 110 (80.29%) 32 (78.05%) 0.75 HbA1c (%) 7.41±1.21 7.42±1.08 0.97 HbA1c (mmol/mol) 57.58±13.26 57.64±11.8 Total cholesterol (mmol/L) 4.79±0.96 4.74±0.84 0.77 HDL cholesterol (mmol/L) 1.31±0.32 1.2±0.29 0.051 LDL cholesterol (mmol/L) 2.72±0.81 2.72±0.70 0.97 Triglycerides (mmol/L) 1.47 (0.64–5.68) 1.78 (0.64–4.55) 0.14 Homocysteine (µmol/L) 10.9 (5.8–37.6) 14 (6.3–127) 0.005 Barcelona. Protected by copyright. Lipoprotein (a) (mg/dL) 8.93 (1–162.9) 6.89 (1–129) 0.64 GFR (mL/min) 82.32±16.66 79.37±16.35 0.32 Creatinine (mg/dL) 0.82±0.36 0.87±0.24 0.45 Albumin/:creatinine ratio  <30 mg/g (n, %) 88 (64.71%) 27 (65.85%) 0.37  30–300 mg/g (n, %) 41 (30.37%) 10 (24.39%)  >300 mg/g (n, %) 6 (4.44%) 4 (9.76%) Log albumin:creatinine ratio 1.28±0.64 1.37±0.72 0.45 Retinopathy (n, %) 27 (22.13%) 17 (39.02%) 0.014  NPDR 26 16    Mild 14 8    Moderate 7 8    Severe 5 0  PDR 4 1 Neuropathy (n, %) 21 (15.33%) 11 (26.83%) 0.09 Data are expressed as %, mean±SD, or median and (range). BMI, body mass index; GFR, glomerular filtration rate; HDL, high-­density lipoprotein; LDL, low-­density lipoprotein; NPDR, non-­ proliferative diabetic retinopathy; PDR, proliferative diabetic retinopathy. DR is related to the CACs of DR were independently related to a CACs >400 AU The distribution of control and diabetic subjects, taking (table 3A). The model for identifying type 2 diabetic into account the CACs, is shown in online supplementary subjects with a CACs >400 showed an AUROC of 0.77 table S2. As expected, type 2 diabetic patients presented (95% CI 0.72 to 0.83; p<0.00001). higher CACs than non-­diabetic control subjects. Among We also observed an inverse relationship between the type 2 diabetic patients, gender, age, waist circumference, degree of DR and a CACs <10 AU. In this regard, 93.9% serum levels of homocysteine, and the presence of DR were of patients with no or only mild DR presented a CACs significantly different in those patients with a CACs ≥400 AU <10 whereas this percentage was as low as 6.1% in those in comparison with patients with <400 AU (table 2). patients with moderate–severe DR. The logistic regression model, which included the Relationship between the CACs and coronary stenosis variables with statistical significance in the univariate Type 2 diabetic patients presented a significantly higher study, showed that only age, gender, and the presence proportion of coronary stenosis (stenosis >50% in at BMJ Open Diab Res Care 2019;7:e000845. doi:10.1136/bmjdrc-2019-000845 5 Cardiovascular and Metabolic Risk BMJ Open Diab Res Care: first published as 10.1136/bmjdrc-2019-000845 on 29 December 2019. Downloaded from http://drc.bmj.com/ on February 19, 2021 at Univesity Autonomous of Table 3 (A) Results of the logistic regression analysis for predicting the presence of a calcium coronary score (CACs) ≥400 AU in subjects with type 2 diabetes. (B, C) Results of the logistic regression analysis for predicting the presence of subclinical cardiovascular disease (CVD) in subjects with type 2 diabetes assessed by the combined endpoint (CACs ≥400 AU, carotid plaque ≥3 mm, carotid intima–media thickness >1, or ECG changes suggestive of previous asymptomatic myocardial infarction). Variables that were significant at a p value <0.05 in the univariate analysis of online supplementary table S3 (comparison between the presence or not of the combined endpoint) were entered in the model as independent variables (age, diabetes duration, hypertension, homocysteine, glomerular filtration rate (GFR), the presence of microalbuminuria and/ or diabetic retinopathy (DR)), adjusting for gender and dyslipidemia. A forward stepwise selection method was used with a significance level of 0.05. In table 3C, the same analyses have been performed but DR was classified into no-­DR or mild and moderate to severe DR/proliferative DR. A Coefficient SE OR (95% CI) P value Age (years) 0.112 0.037 1.12 (1.04 to 1.21) 0.002 Gender (M/F) −1.14 0.407 0.32 (0.14 to 0.71) 0.005 Homocysteine (µmol/L) 0.063 0.036 1.07 (0.99 to 1.14) 0.084 DR (no/yes) 0.898 0.412 2.46 (1.11 to 5.51) 0.029 Constant −9.189 2.462 Dependent variable: CACs ≥400 AU. Independent variables included in the model: age, gender, waist circumference, homocysteine, and DR (no/yes). B Age (years) 0.077 0.027 1.06 (1.02 to 1.11) 0.005 DR and/or microalbuminuria 0.011 Barcelona. Protected by copyright. Microalbuminuria (no/yes) 0.944 0.458 2.70 (1.05 to 6.30) DR (no/yes) 0.932 0.523 2.54 (0.91 to 7.08) Microalbuminuria and DR 1.199 0.621 3.32 (0.98 to 11.20) Constant −4.854 1.81 Dependent variable: combined endpoint. Independent variables included in the model: age, gender, diabetes duration, GFR, homocysteine, hypertension (no/yes), dyslipidemia (no/yes), and the presence of microalbuminuria and/or DR (no/yes). C Age (years) 0.077 0.028 1.06 (1.02 to 1.11) 0.006 DR and/or microalbuminuria 0.001 Microalbuminuria 0.816 0.407 2.26 (1.02 to 5.02) DR (no–mild/moderate–severe) 2.144 1.096 8.53 (1.05 to 69.12) Microalbuminuria and DR 2.202 1.096 9.04 (1.05 to 77.57) Constant −4.873 1.842 Dependent variable: combined endpoint. Independent variables included in the model: age, gender, diabetes duration, GFR, homocysteine, hypertension (no/yes), dyslipidemia (no/yes), and the presence of microalbuminuria and/or DR (no or mild /moderate to severe). least in one coronary artery) than non-­diabetic control Subclinical CVD subjects (24.3% vs 9.1%; p=0.03). Online supplementary The presence of subclinical CVD assessed by the figure S2 displays the localization of these stenoses in the combined endpoint was higher in type 2 diabetic patients coronary arteries. than in the control group (65.7% vs 44%; p=0.05). The As expected, we found a clear relationship between main clinical characteristics of type 2 diabetic patients the CACs and the presence of coronary stenosis in both according to the presence or absence of the combined diabetic and non-­diabetic control subjects (figure 1A). endpoint is displayed in online supplementary table 3. Diabetic subjects with coronary stenosis presented a Type 2 diabetic patients with the combined endpoint significantly higher CACs (AU) than diabetic subjects were older, presented longer diabetes duration, more without coronary stenosis (337 (95% CI 10 to 2236) vs 5 frequent hypertension, higher homocysteine levels, (95% CI 0 to 293); p<0.0001). Also, a clear relationship higher levels of creatinine, a lower GFR, higher microal- between the presence and degree of DR and the presence buminuria levels, and a higher frequency of DR. In addi- of one or more coronary stenoses was found (figure 1B). tion, those patients with moderate–severe DR presented 6 BMJ Open Diab Res Care 2019;7:e000845. doi:10.1136/bmjdrc-2019-000845 Cardiovascular and Metabolic Risk BMJ Open Diab Res Care: first published as 10.1136/bmjdrc-2019-000845 on 29 December 2019. Downloaded from http://drc.bmj.com/ on February 19, 2021 at Univesity Autonomous of The probability of having a CACs ≥400 AU, carotid plaque ≥3 mm, CIMT >1, and the composite endpoint according to the presence and degree of DR is repre- sented in online supplementary table 4. The presence of DR significantly increases the risk of CACs ≥400 AU and the composite endpoint. Finally, the discrimination abilities of the United Kingdom Prospective Diabetes Study (UKPDS) and Fram- ingham cardiovascular risk scores to predict the combined endpoint of subclinical cardiovascular disease, adding DR (categorized as either “yes or no” or “moderate–severe vs no or mild DR”) alone or in combination with microal- buminuria, were calculated and compared employing the AUROC. We found that it was only when DR was catego- rized as moderate to severe versus no or mild DR that the AUROC increased significantly (table 4). Discussion The present study provides evidence that DR is a powerful and independent risk factor for identifying diabetic subjects with subclinical CVD. In fact, we have found that the assessment of DR is a good and independent predictor of both a CACs >400 AU and the composite of Barcelona. Protected by copyright. a CACs >400 AU, carotid plaque ≥3 mm, and CIMT >1. In addition, a clear relationship between the presence and degree of DR and coronary stenosis was found. These findings suggest that type 2 diabetic subjects with DR represent a subset of patients at very high risk of CVD who need a specific program aimed at reducing CV risk factors, optimizing metabolic control, and providing a periodic assessment of CVD. It should be noted that the Figure 1 (A) Calcium coronary score (CACs, AU) in subjects presence of DR confers a higher risk of subclinical CVD with coronary stenosis (black columns) and without coronary than factors contained in contemporary risk equations, stenosis (white columns). Data are median (95% CI 5% to such as blood pressure, LDL cholesterol, and HbA1c. 95%). (B) Association between coronary stenosis and the In this regard, we have found that moderate–severe DR presence and degree of diabetic retinopathy (DR). NPDR, significantly increases the AUROC of the UKDPS and non-­proliferative diabetic retinopathy; PDR, proliferative Framingham risk scores CHD in predicting the presence diabetic retinopathy. of the combined endpoint of subclinical CVD. Consistent with our findings, previous reports have documented an increase in CV risk in patients with a higher prevalence of subclinical CVD assessed by the DR, particularly in those with advanced DR.26–29 For combined endpoint in comparison with patients with instance, de Kreutzenberg et al,30 in a large cohort of mild DR or the absence of DR (91.3% vs 60.8%; p=0.004). type 2 diabetic subjects found that DR alone or in combi- The logistic regression model in which DR was intro- nation with nephropathy was independently associated duced as a categorical variable (yes/no) showed that with the presence of carotid plaques, and that severity DR was not an independent variable accounting for the of microangiopathy correlates with severity of carotid combined endpoint (table 3B). However, when DR was atherosclerosis. In addition, coronary atherosclerosis introduced as absence–mild DR versus moderate–severe, and plaque vulnerability are more severe in patients it was strongly associated with the combined endpoint with DR.31 Overall, this information suggests that CAD (OR 8.53 (95% CI 1.05 to 69.12)) (table 3C). The is mediated by microvessel damage within the arterial AUROC of this model for identifying subclinical CVD wall (the vasa vasorum) induced by diabetes.32 Recent using the combined endpoint was 0.69 (95% CI 0.61 to evidence indicates that in subjects with type 2 diabetes, 0.76; p<0.05). When, in the same model, microalbumin- the vasa vasorum presents evolutionary changes similar uria was added to the DR categorized as moderate–severe to those observed in the retina: an initial stage in which versus no or mild DR, the strength of the association endothelium dysfunction and loss of capillaries predom- increased (9.04 (95% CI 1.05 to 77)) as well as the inates,33 and more advanced stages in which ischemia AUROC (0.72 (95% CI 0.64 to 0.78; p<0.05)). plays a key role, leading to angiogenesis and plaque BMJ Open Diab Res Care 2019;7:e000845. doi:10.1136/bmjdrc-2019-000845 7 Cardiovascular and Metabolic Risk BMJ Open Diab Res Care: first published as 10.1136/bmjdrc-2019-000845 on 29 December 2019. Downloaded from http://drc.bmj.com/ on February 19, 2021 at Univesity Autonomous of Table 4 Comparisons between UKPDS-­CHD (CHD=coronary heart disease) (A) and Framingham risk scores (B) with and without the addition of microalbuminuria (no/yes), diabetic retinopathy (DR) (no/yes), DR+microalbuminuria, moderate–severe DR, and moderate–severe DR+microalbuminuria in assessing the presence of the combined endpoint A ROC area SE 95% CI P value UKPDS-­CHD 0.658 0.044 0.58 to 0.75 UKPDS-­CHD+microalbuminuria 0.683 0.044 0.60 to 0.77 0.191 UKPDS-­CHD+DR 0.687 0.045 0.60 to 0.77 0.198 UKPDS-­CHD+microalbuminuria and DR 0.701 0.044 0.62 to 0.79 0.074 UKPDS-­CHD+moderate–severe DR 0.729 0.041 0.65 to 0.818 0.001 UKPDS-­CHD+microalbuminuria and moderate–severe DR 0.733 0.042 0.65 to 0.82 0.008 B Framingham 0.649 0.045 0.56 to 0.78 Framingham+microalbuminuria 0.66 0.045 0.57 to 0.75 0.52 Framingham+DR 0.678 0.046 0.59 to 0.77 0.269 Framingham+microalbuminuria and DR 0.683 0.045 0.60 to 0.77 0.223 Framingham+moderate–severe DR 0.717 0.042 0.67 to 0.83 0.007 Framingham+microalbuminuria and moderate–severe DR 0.714 0.043 0.63 to 0.80 0.034 UKPDS, United Kingdom Prospective Diabetes Study. Barcelona. Protected by copyright. neovascularization.33–35 This change in plaque phenotype develop a cardiovascular outcome. However, given that the results in a more inflamed and unstable plaque, favoring prevalence of DR in type 2 diabetic subjects is around 30%, plaque rupture and a poor outcome of CV events. Thus, it still remains a huge population as regards recommending microcirculation represents a “common soil” between DR screening for silent CVD. The high risk in those patients with and CAD, and it may no longer make sense to consider moderate–severe DR of exhibiting subclinical CVD points microangiopathy and macroangiopathy as entirely sepa- to this specific subset of patients as the most cost-­effective rated entities in the setting of DM. candidates for screening. Therefore, healthcare professionals The screening of CVD in the diabetic population involved in the treatment of patients with diabetes should continues to be a challenge. An analysis of pooled contemplate those individuals with moderate–severe DR as population-­based studies showed that in adults older a high-­risk population independently of the association of than 60 years without known atherosclerotic CVD at other traditional CV risk factors. baseline, the CACs has a greater association with inci- CT angiography images of coronary artery stenosis dent CHD (follow-­up 11 years) and a modestly improved provide a better prediction risk in asymptomatic type 2 prediction of incident stroke.36 In addition, in the diabetics than traditional risk factors and the CACs.38 39 diabetic population, the severity of the CACs appeared Notably, we have found for the first time a relationship to be a more important clinical prognostic indicator between DR and the presence of coronary stenosis assessed than measures of disease severity, such as insulin use, by coronary CT angiography. Patients with DR presented a glycemic control, and diabetes duration.37 Our study higher proportion of having one or more coronary stenoses further supports this concept and, in addition, shows ≥50% than those patients without DR (38.5% vs 20.2%). In a clear relationship between a CACs >400 AU and coro- addition, this proportion was even higher (50%) in those nary artery stenosis in type 2 diabetic subjects. However, patients with moderate–severe NPDR and PDR. given the high prevalence of diabetes, it is desirable to Apart from DR, we found that microalbuminuria was identify a more targeted population in which screening associated with subclinical CVD assessed by the combined by using the CACs could be recommended. This would endpoint, but it was unrelated to the CACs or the presence increase the cost-­effectiveness of screening and avoid of coronary stenosis. This finding suggests that microalbu- unnecessary irradiation. We found that by using simple minuria reflects a more widespread atherosclerotic process, clinical parameters such as age, gender, and the assess- whereas the presence and degree of DR might be more ment of DR, the AUROC for predicting a CACs >400 directly related to CAD. It should be noted that there is was 0.77. In addition, CACs <10 was also associated with extensive evidence indicating that microalbuminuria is an the absence of, or just mild DR. independent CV risk factor and it is generally considered Overall, our results point to type 2 diabetic patients with in the design and the analysis of the results of clinical trials DR as a subset of the diabetic population at a high risk of on CVD performed in the diabetic population. However, having silent CVD and, in consequence, more prone to this is not the case for DR. This is a significant drawback 8 BMJ Open Diab Res Care 2019;7:e000845. doi:10.1136/bmjdrc-2019-000845 Cardiovascular and Metabolic Risk BMJ Open Diab Res Care: first published as 10.1136/bmjdrc-2019-000845 on 29 December 2019. Downloaded from http://drc.bmj.com/ on February 19, 2021 at Univesity Autonomous of that could lead to a serious bias on the results obtained in algorithms (eg, QRISK, Framingham, Reynolds, UKDPS) both epidemiologic and interventional studies. are for calculating risk estimates and not for the detection Our results underline the importance of screening for of subclinical atherosclerosis. Therefore, further studies DR, which should be envisaged as a method for studying the with a larger sample size are required to evaluate whether microvasculature of the retina and its potential implications the addition of DR grading to conventional risk algorithms in vision loss, and also as a herald of other systemic compli- will confer any additional benefits. cations of diabetes, including CVD. Retinal imaging has the Our findings could have both therapeutic and investiga- significant advantage of not using ionizing irradiation and tional implications. In fact, the detection and grading of DR it is performed as routine screening in the diabetic popula- permits us to identify patients at high CV risk who might, tion. Given the alarming increase in the number of people under current guidelines, benefit most from further lipid with diabetes and the shortage of trained retinal specialists and blood glucose lowering, or it could be used to enrich and graders of retinal photographs, an automated approach the cohorts for future intervention trials with patients more involving a computer-­ based analysis of the fundus image prone to develop clinical outcomes, thus reducing sample would reduce the burden of health systems in screening for size, duration, and the cost of studies. DR.40 41 There is hence an increasing interest in the devel- In summary, the presence of DR is an independent opment of automated analysis software using artificial intelli- predictor of CACs >400 AU, and moderate–severe DR is gence/deep neural learning for the analysis of retinal images independently associated with subclinical CVD. There- in people with diabetes,41 and it is foreseeable that specific fore, the presence and degree of DR should be consid- software will be developed to better define the cardiovascular ered a better tool for identifying type 2 diabetic patients risk of diabetic individuals based on the retinal structural and at risk of CVD than the conventional CV risk factors used functional findings of microvessels. in current risk equations. Our results have significant The low percentage of silent MI detected by ECG in clinical implications in terms of proposing a rationalized our cohort, 1 out 200 (0.5%), is quite surprising. In the screening of CAD based on prioritizing patients with DR, UKPDS, 1 in 6 (16%) patients with newly diagnosed type Barcelona. Protected by copyright. in particular those with a moderate–severe degree. 2 diabetes had evidence of silent MI on the baseline surface ECG.42 In older studies, the prevalence of ECG Author affiliations abnormalities in patients with type 2 diabetes and no 1 Diabetes and Metabolism Research Unit, Endocrinology Department, Vall d’Hebron known CAD was even higher, approaching 20%.43 44 The Research Unit, Vall d’Hebron Hospital, Autonomous University of Barcelona, low rate of silent MI in our diabetic population could not Barcelona, Spain 2 be attributed to low diabetes duration or a selected group CIBERDEM, ISCIII, Madrid, Spain 3 Cardiology Research Group, Cardiology Department, Vall d’Hebron Research with low diabetes-­ related systemic complications. In Institute, Vall d’Hebron Hospital, Autonomous University of Barcelona, Barcelona, fact, the mean of known diabetic duration was 12 years, Spain around 30% of the subjects already presented microan- 4 CIBER CV, ISCIII, Madrid, Spain giopathic complications, and more than half required 5 Medical and Metabolic Research Area, Nuclear Medicine, Vall d’Hebron Research treatment with insulin. It should be noted that this low Institute, Vall d’Hebron Hospital, Autonomous University of Barcelona, Barcelona, rate of ECG signs of asymptomatic ischemic heart disease Spain 6 Neurovascular Research Laboratory, Vall d’Hebron Research Institute, Vall did not mean the presence of low myocardial microan- d’Hebron Hospital, Autonomous University of Barcelona, Barcelona, Spain giopathy. In this regard, we have previously reported 7 Nephrology Department, Vall d’Hebron Research Institute, Vall d’Hebron Hospital, that DR was independently associated with myocardial Autonomous University of Barcelona, Barcelona, Spain perfusion defects.45 Altogether, these findings suggest 8 REDinREN, ISCIII, Madrid, Spain 9 that our population could have a high degree of protec- Liver Unit, Department of Medicine, Vall d’Hebron Research Institute, Vall d’Hebron tion against CV outcomes despite having a high rate of Hospital, Autonomous University of Barcelona, Barcelona, Spain 10 CIBEREHD, ISCIII, Madrid, Spain cardiovascular risk factors including microangiopathy. A 11 Ophthalmology Department, Vall d’Hebron Research Institute, Vall d’Hebron Mediterranean diet and the high proportion of patients Hospital, Autonomous University of Barcelona, Barcelona, Spain under treatment with statins might help explain why we detected a very low rate of silent MI; however, specific Contributors RS, IF, and DG-­D conceived the study concept and design, studies aimed at examining this issue are needed. interpreted data, and contributed to critically revising the manuscript. JB, CH, and This study has several limiting factors. First, the analyses OS-­S analyzed data and drafted the manuscript. JR-­P, FV, LG, TG-­A, SA-­B, AB, and OS-­S contributed to the collection and interpretation of data. SA-­B, JM, DS, were restricted to individuals for whom complete informa- JG, and JG-­A contributed to the study design, data interpretation, and revised the tion was available, and this may have resulted in selection manuscript for important intellectual content. All authors approved the final article. bias. Second, the results could have been affected by vari- RS, SA-­B, JM, DS, JG, and DG-­D obtained funding. RS and IF are the guarantors ables such as diet or ambient factors not considered in this of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. analysis. Third, the study population was relatively small Funding This work was supported by an Integrative Excellence Project by the and the data presented are observational in nature. Finally, Spanish Institute of Health, Instituto de Salud Carlos III, grant PIE 2013/27, CIBER although our results suggest that the assessment of DR may CV, CIBERDEM, and the European Regional Development Fund (ERDF-­FEDER). offer a simple tool to identify very high-­risk individuals with The Neurovascular Research Laboratory is part of the Spanish Stroke Research type 2 diabetes who are currently perceived to be at low Network INVICTUS+ (RD16/0019/0021). absolute risk using contemporary risk models, all the risk Competing interests None declared. BMJ Open Diab Res Care 2019;7:e000845. doi:10.1136/bmjdrc-2019-000845 9 Cardiovascular and Metabolic Risk BMJ Open Diab Res Care: first published as 10.1136/bmjdrc-2019-000845 on 29 December 2019. Downloaded from http://drc.bmj.com/ on February 19, 2021 at Univesity Autonomous of Patient consent for publication Obtained. 16 Elkeles RS, Godsland IF, Feher MD, et al. Coronary calcium measurement improves prediction of cardiovascular events in Ethics approval The study was conducted according to the Declaration of Helsinki asymptomatic patients with type 2 diabetes: the PREDICT study. Eur and was approved by the local ethics committee. Heart J 2008;29:2244–51. Provenance and peer review Not commissioned; externally peer reviewed. 17 Budoff MJ, Raggi P, Beller GA, et al. Noninvasive cardiovascular risk assessment of the asymptomatic diabetic patient: The Imaging Data availability statement All data relevant to the study are included in the Council of the American College of Cardiology. JACC Cardiovasc article or uploaded as online supplementary information. Imaging 2016;9:176–92. 18 Kramer CK, Zinman B, Gross JL, et al. Coronary artery calcium Open access This is an open access article distributed in accordance with the score prediction of all cause mortality and cardiovascular events in Creative Commons Attribution Non Commercial (CC BY-­NC 4.0) license, which people with type 2 diabetes: systematic review and meta-­analysis. permits others to distribute, remix, adapt, build upon this work non-­commercially, BMJ 2013;346:f1654. and license their derivative works on different terms, provided the original work is 19 Selvin E, Steffes MW, Zhu H, et al. Glycated hemoglobin, diabetes, properly cited, appropriate credit is given, any changes made indicated, and the and cardiovascular risk in nondiabetic adults. N Engl J Med use is non-­commercial. See: http://c​ reativecommons.​org/​licenses/​by-​nc/​4.​0/. 2010;362:800–11. 20 Pearce I, Simó R, Lövestam-­Adrian M, et al. Association between ORCID iD diabetic eye disease and other complications of diabetes: implications for care. A systematic review. Diabetes Obes Metab Rafael Simó http://​orcid.​org/​0000-​0003-​0475-​3096 2019;21:467–78. 21 Brownrigg JRW, Hughes CO, Burleigh D, et al. Microvascular disease and risk of cardiovascular events among individuals with type 2 diabetes: a population-­level cohort study. Lancet Diabetes References Endocrinol 2016;4:588–97. 1 International Diabetes Federation. IDF diabetes atlas. 3rd edition, 22 Ahn H-­R, Shin M-­H, Yun W-­J, et al. Comparison of the Framingham 2017. http://www.​diabetesatlas.​org/​resources/​previous-​editions.​html Risk Score, UKPDS Risk Engine, and SCORE for Predicting Carotid 2 Zhang P, Gregg E. Global economic burden of diabetes and its Atherosclerosis and Peripheral Arterial Disease in Korean Type 2 implications. Lancet Diabetes Endocrinol 2017;5:404–5. Diabetic Patients. Korean J Fam Med 2011;32:189–96. 3 Haffner SM, Lehto S, Rönnemaa T, et al. Mortality from coronary 23 Wilkinson CP, Ferris FL, Klein RE, et al. Proposed International heart disease in subjects with type 2 diabetes and in nondiabetic clinical diabetic retinopathy and diabetic macular edema disease subjects with and without prior myocardial infarction. N Engl J Med severity scales. Ophthalmology 2003;110:1677–82. 1998;339:229–34. 24 Touboul P-­J, Hennerici MG, Meairs S, et al. Mannheim carotid 4 Anand DV, Lim E, Hopkins D, et al. Risk stratification in intima–media thickness consensus (2004–2006). An update on uncomplicated type 2 diabetes: prospective evaluation of the behalf of the Advisory Board of the 3rd and 4th Watching the combined use of coronary artery calcium imaging and selective Barcelona. Protected by copyright. myocardial perfusion scintigraphy. Eur Heart J 2006;27:713–21. Risk Symposium, 13th and 15th European Stroke Conferences, 5 Becker A, Bos G, Vegt F. Cardiovascular events in type 2 diabetes: Mannheim, Germany, 2004, and Brussels, Belgium, 2006. comparison with nondiabetic individuals without and with prior Cerebrovasc Dis 2007;23:75–80. cardiovascular disease: 10-­year follow-­up of the Hoorn Study. Eur 25 Mark DB, Berman DS, Budoff MJ, et al. ACCF/ACR/AHA/NASCI/ Heart J 2003;24:1406–13. SAIP/SCAI/SCCT 2010 expert consensus document on coronary 6 American Diabetes Association. 9. Cardiovascular disease and computed tomographic angiography: a report of the American risk management: standards of medical care in diabetes—2018. College of Cardiology Foundation Task force on expert consensus Diabetes Care 2017;41:S86–104. documents. J Am Coll Cardiol 2010;55:2663–99. 7 Piepoli MF, Hoes AW, Agewall S, et al. 2016 European Guidelines 26 Cheung N, Wang JJ, Klein R, et al. Diabetic retinopathy and the risk on cardiovascular disease prevention in clinical practice: The Sixth of coronary heart disease: the Atherosclerosis Risk in Communities Joint Task Force of the European Society of Cardiology and Other Study. Diabetes Care 2007;30:1742–6. Societies on Cardiovascular Disease Prevention in Clinical Practice 27 Targher G, Bertolini L, Zenari L, et al. Diabetic retinopathy is (constituted by representatives of 10 societies and by invited associated with an increased incidence of cardiovascular events in experts) Developed with the special contribution of the European type 2 diabetic patients. Diabetic Med 2008;25:45–50. Association for Cardiovascular Prevention & Rehabilitation (EACPR). 28 Gerstein HC, Ambrosius WT, Danis R, et al. Diabetic retinopathy, Eur Heart J 2016;37:2315–81. its progression, and incident cardiovascular events in the ACCORD 8 Fox CS, Golden SH, Anderson C, et al. Update on prevention of trial. Diabetes Care 2013;36:1266–71. cardiovascular disease in adults with type 2 diabetes mellitus in light 29 Alonso N, Traveset A, Rubinat E, et al. Type 2 diabetes-­associated of recent evidence: a scientific statement from the American Heart carotid plaque burden is increased in patients with retinopathy Association and the American Diabetes Association.. Circulation compared to those without retinopathy. Cardiovasc Diabetol 2015;132:691–718. 2015;14:33. 9 Low Wang CC, Hess CN, Hiatt WR, et al. Clinical Update: 30 de Kreutzenberg SV, Coracina A, Volpi A, et al. Microangiopathy is Cardiovascular Disease in Diabetes Mellitus: Atherosclerotic independently associated with presence, severity and composition Cardiovascular Disease and Heart Failure in Type 2 Diabetes of carotid atherosclerosis in type 2 diabetes. Nutr Metab Cardiovasc Mellitus—Mechanisms, Management, and Clinical Considerations. Dis 2011;21:286–93. Circulation 2016;133:2459–502. 31 Kurihara O, Takano M, Mizuno K, et al. Impact of diabetic retinopathy 10 Nambi V, Chambless L, Folsom AR, et al. Carotid intima–media on vulnerability of atherosclerotic coronary plaque and incidence of thickness and presence or absence of plaque improves prediction acute coronary syndrome. Am J Cardiol 2016;118:944–9. of coronary heart disease risk: the ARIC (Atherosclerosis Risk In 32 Gerstein HC, Werstuck GH. Dysglycaemia, vasculopenia, and the Communities) study. J Am Coll Cardiol 2010;55:1600–7. chronic consequences of diabetes. Lancet Diabetes Endocrinol 11 Jeevarethinam A, Venuraju S, Dumo A, et al. Relationship between 2013;1:71–8. carotid atherosclerosis and coronary artery calcification in 33 Gerstein HC, Nair V, Chaube R, et al. Dysglycemia and the density of asymptomatic diabetic patients: a prospective multicenter study. the coronary vasa vasorum. Diabetes Care 2019;42:980–2. Clin Cardiol 2017;40:752–8. 34 Arcidiacono MV, Traveset A, Rubinat E, et al. Microangiopathy of 12 Malik S, Budoff MJ, Katz R, et al. Impact of subclinical large artery wall: a neglected complication of diabetes mellitus. atherosclerosis on cardiovascular disease events in individuals Atherosclerosis 2013;228:142–7. with metabolic syndrome and diabetes: the multi-­ethnic study of 35 Sedding DG, Boyle EC, Demandt JAF, et al. Vasa vasorum atherosclerosis. Diabetes Care 2011;34:2285–90. angiogenesis: key player in the initiation and progression 13 Agarwal S, Morgan T, Herrington DM, et al. Coronary calcium score of atherosclerosis and potential target for the treatment of and prediction of all-­cause mortality in diabetes: the diabetes heart cardiovascular disease. Front Immunol 2018;9:70. study. Diabetes Care 2011;34:1219–24. 36 Yano Y, O'Donnell CJ, Kuller L, et al. Association of coronary artery 14 Wong ND, Nelson JC, Granston T, et al. Metabolic syndrome, calcium score vs age with cardiovascular risk in older adults: diabetes, and incidence and progression of coronary calcium: the an analysis of pooled population-­based studies. JAMA Cardiol multiethnic study of atherosclerosis study. JACC Cardiovasc Imaging 2017;2:986–94. 2012;5:358–66. 37 Malik S, Zhao Y, Budoff M, et al. Coronary artery calcium score for 15 Raggi P, Shaw LJ, Berman DS, et al. Prognostic value of coronary long-­term risk classification in individuals with type 2 diabetes and artery calcium screening in subjects with and without diabetes. J Am metabolic syndrome from the multi-­ethnic study of atherosclerosis. Coll Cardiol 2004;43:1663–9. JAMA Cardiol 2017;2:1332–40. 10 BMJ Open Diab Res Care 2019;7:e000845. doi:10.1136/bmjdrc-2019-000845 Cardiovascular and Metabolic Risk BMJ Open Diab Res Care: first published as 10.1136/bmjdrc-2019-000845 on 29 December 2019. Downloaded from http://drc.bmj.com/ on February 19, 2021 at Univesity Autonomous of 38 Halon DA, Azencot M, Rubinshtein R, et al. Coronary computed 42 Davis TME, Coleman RL, Holman RR, et al. Prognostic significance tomography (CT) angiography as a predictor of cardiac and of silent myocardial infarction in newly diagnosed type 2 diabetes noncardiac vascular events in asymptomatic type 2 diabetics: a 7‐ mellitus: United Kingdom prospective diabetes study (UKPDS) 79. year population‐based cohort study. J Am Heart Assoc 2016;5:pii: Circulation 2013;127:980–7. e003226. 43 Scheidt-­Nave C, Barrett-­Connor E, Wingard DL. Resting 39 Lee KY, Hwang B-­H, Kim T-­H, et al. Computed tomography electrocardiographic abnormalities suggestive of asymptomatic angiography images of coronary artery stenosis provide a better ischemic heart disease associated with non-­insulin-­dependent prediction of risk than traditional risk factors in asymptomatic diabetes mellitus in a defined population. Circulation 1990;81:899–906. individuals with type 2 diabetes: a long-­term study of clinical 44 Boland LL, Folsom AR, Sorlie PD, et al. Occurrence of unrecognized outcomes. Diabetes Care 2017;40:1241–8. myocardial infarction in subjects aged 45 to 65 years (the ARIC 40 Abramoff MD, Suttorp-­Schulten MSA, Suttorp-­Schulten MSA, study). Am J Cardiol 2002;90:927–31. et al. Evaluation of a system for automatic detection of diabetic 45 Hernández C, Candell-­Riera J, Ciudin A, et al. Prevalence and risk retinopathy from color fundus photographs in a large population of factors accounting for true silent myocardial ischemia: a pilot case– patients with diabetes. Diabetes Care 2008;31:193–8. control study comparing type 2 diabetic with non-­diabetic control 41 Gulshan V, Peng L, Coram M, et al. Development and validation of subjects. Cardiovasc Diabetol 2011;10:9. a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs. JAMA 2016;316:2402–10. Barcelona. Protected by copyright. BMJ Open Diab Res Care 2019;7:e000845. doi:10.1136/bmjdrc-2019-000845 11