Ethics statement
This retrospective study was deemed exempt from institutional review board (IRB) review by the SingHealth Centralised Institutional Review Board (CIRB). This study adhered to the tenets of the Declaration of Helsinki. Written informed consent was obtained from the participants of the original studies [10, 11].
Study population
We used clinical data and retinal photographs from the UK Biobank, a prospective population-based cohort in the UK [12]. The UK Biobank protocol is available online [13].
We excluded (1) duplicated retinal photographs (n = 18,423), (2) those who had type 1 diabetes (n = 288), (3) those with pre-existing CVD at baseline (n = 7624), (4) poor-quality photographs (n = 11,115), and (5) those who were < 40 years old (n = 1) (Additional file 1: eFigure 1). Pre-existing CVD was defined as previous history of coronary heart disease, other heart diseases, stroke, transient ischemic attack, peripheral arterial disease, or cardiovascular surgery, and patients who have undergone cardiovascular procedures based on the Classification of Interventions and Procedures version 4 (OPCS-4) [10].
A total of 48,260 participants, representing the general population without a history of CVD, were included for analysis. In addition, we also defined three at-risk subgroups: (1) the non-statin cohort, individuals not taking a statin; (2) the stage 1 hypertension cohort, individuals with stage 1 hypertension and not taking any antihypertensive medications; and (3) the middle-aged cohort, individuals aged 40–64 years at baseline (Additional file 1: eFigure 1). Since the purpose of Reti-CVD is aimed at primary prevention of CVD, we focus on individuals who display a lack of awareness towards the risk factors associated with cardiovascular disease. In a previous publication, the authors reported based on age-stratified analyses that unawareness rates were highest in individuals aged between 40 and 49 years old and lowest in individuals aged above 70 years old [14]. Therefore, our middle-aged cohort is classified based on vulnerability.
Retinal photographs included in the study were taken using the Topcon 3D OCT-1000 Mark II (Topcon Corporation) between 7 December 2009 and 21 July 2010. Retinal cameras used in the training set include AFP-210 non-mydriatic auto retinal camera (NIDEK Corporation, Aichi, Japan), TRC-NW8 non-mydriatic retinal camera (Topcon Corporation, Tokyo, Japan), and Nonmyd A-D (Kowa Co. Ltd., Shizuoka, Japan). We did not include Topcon 3D OCT-1000 Mark II (Topcon Corporation) in our training set.
The other variables used in this study were defined as follows. Pre-diabetes and diabetes were defined based on (1) medical history and (2) glucose ≥ 5.5 mmol/L. Medical history of high cholesterol, type 1 diabetes, antihypertensive were self-declared and collected from a baseline assessment questionnaire on medical conditions. Smoking status was self-declared as well and categorized into “life-time smoker” and “never.”
Definition of cardiovascular disease events
In the UK Biobank, we used hospitalization and mortality data provided by the National Health Service (NHS) registers. The main outcome of interest in the current study reflected the outcome used in the QRISK3 risk score: fatal CVD events (ICD-10 I00-99, F01, Q20-Q28, C38.0, P29, G45) [15] or nonfatal coronary heart disease, ischemic stroke, or transient ischemic attack (ICD-10 G45, I20–24, and I63–64) [16].
Calculation of the QRISK3 score and borderline-QRISK3 group
For each individual, the QRISK3 score was calculated using R package, version 3.6 [17]. The distribution of the QRISK3 score is provided in Additional file 2: eFigure 2. For comparison with three-strata Reti-CVD groups, we divided the subjects into 5 groups based on the QRISK3 score (%) (≥ 0 to < 5; ≥ 5 to > 10; ≥ 10 to > 15; ≥ 15 to < 20; and ≥ 20). In addition, as the recommended threshold for initiating statins and antihypertensive medication from a 10-year risk of CVD is 10%, we defined “borderline-risk” group as those who had QRISK3 score between 7.5 and 10%. Specifically, we divided the subjects into 5 groups as compared to the previous three-risk-strata group of QRISK3 as outlined in the NICE guidelines. Herein lies the difference in intervals used to evaluate CVD events. We used 5% intervals compared to the previous 10% intervals so that a more detailed comparison could be done with Reti-CVD.
New retinal-based cardiovascular disease risk stratification system ≈
Details of RetiCAC model development and previous validations have been described elsewhere [9]. Briefly, the RetiCAC score was defined based on a probability score derived from our deep-learning algorithm of binary classification (absence vs presence of coronary artery calcium [CAC]). The probability scores ranged from zero to one, with a high value indicating a high probability of the presence of CAC. First, we enhanced the RetiCAC algorithm using more datasets which includes both retinal photographs and CT scans from Korea (Additional file 3: eDocument 1). Second, we proposed new cardiovascular disease risk stratification groups (i.e., Reti-CVD) with optimized cut-off values based on 40th percentile and 95th percentile of Reti-CVD score among 48,260 participants after exclusion. The cut-off values were determined to have similar incidence rate: the low-risk group of Reti-CVD was designed to have similar incidence rate to those of the 0 to 5% QRISK3 risk group (i.e., 2.5 per 1000 person-years), and the moderate-risk of Reti-CVD group was designed to have similar incidence rate to those of the 5 to 10% QRISK3 risk group (i.e., 7.0 per 1000 person-years). We used these proposed cut-off values to further stratify the CVD risk in the UK Biobank participants.
Statistical analysis
Analyses were done using p < 0.05 as the significance level, Stata/MP version 14.0 for survival analysis, and R version 3.4.4 for estimation of net reclassification index (NRI) using the R package survIDINRI [18]. Descriptive statistics are provided for all participants and also according to the three risk groups by Reti-CVD.
In the UK Biobank, hospitalization and mortality data were available up to May 05, 2021, at the time of analysis and each participant was followed up to 11.4 years from the date of baseline visit. In survival analysis, each patient was followed up to 11.4 years (median follow-up, 11.0 years) from the date of baseline visit to the last follow-up date or the date of the CVD events.
In all populations, the cumulative incidence of cardiovascular events rate was evaluated across the three groups (low, moderate, and high risk) defined by the Reti-CVD using Kaplan-Meier method. Cox proportional hazards model was used to estimate the hazard ratios (HRs) and trends in HRs and respective p-values were examined by fitting a linear model for the three categories. Unadjusted HR was provided according to three risk groups of Reti-CVD, and QRISK3-ajudted HR trend was provided.
To help future patients make an informed decision on statin and antihypertensive medication initiation, we only included the borderline-QRISK3 group who had QRISK3 score between 7.5 and 10% of 10-year CVD risk. In the borderline QRISK3 group, cumulative incidence of cardiovascular events rate was evaluated across the three groups (low, moderate, and high risk) according to the Reti-CVD and compared with participants with QRISK3 5–7.5% and 10–12.5%. The same analysis was repeated for middle-aged group (40 to 64 years).
The incremental prognostic value of the Reti-CVD over the QRISK3 in the prediction of CVD events was assessed using C-statistics and continuous net reclassification index (NRI) [18]. In addition, decision curve analysis was used to compare the net-benefit of models at different thresholds over QRISK3 model and Reti-CVD plus age and gender model. Age and gender were included for fair comparison because QRISK3 is based on survival model including various risk factors including age, gender, smoking, and comorbidities. Continuous models were presented as decision-analysis curves (demonstrating potential outcomes of using any threshold in that model) and models with higher net-benefit were considered higher performing.