Participants were examined in the Outpatient Department of the 4th Clinic of Internal Medicine at LP University Hospital in Kosice. The examinations were carried out in the morning, under standard conditions. We performed blood sampling, urine tests, electrocardiograms, subclinical ATS markers, and conducted medical interviews to detect major RFs for ATS and pharmacotherapy. On physical examination, anthropometric parameters and office blood pressure were measured. A 10-year total CV mortality (SCORE) and Framingham risk score were calculated for each subject. Blood and urine samples were analysed by standard laboratory methods. Metabolic parameters included: fasting glucose, glycated haemoglobin (HbA1c), uric acid, serum total cholesterol (T-C), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), triglycerides (TAG) and serum creatinine. The estimated glomerular filtration rate was calculated according to the “Modification of Diet in Renal Disease” formula [19]. The following values were considered abnormal for the primary prevention cohort, based on accredited laboratory reference values: T-C $>$5.0 mmol/L, LDL-C $>$3.0 mmol/L, HDL-C 1.0/1.2 mmol/L (males/females), TAG $>$1.7 mmol/L [20], creatinine $>$90 µmol/L, eGFR 90 mL/min/m${}^2$, uric acid $>$357/428 µmol/L (males/females). Non-modifiable RFs for ATS as well as arterial hypertension (AH), dyslipoproteinemia (DLP), obesity/central obesity, diabetes mellitus (DM), impaired fasting glucose, metabolic syndrome (MetS) were defined according to current recommendations [1, 18]. Smoking status was characterized as current smoking $\ge$1 cigarette/day. To compute an individual’s 10-year fatal risk, the SCORE charts for high-risk countries, applicable to Slovakia (low risk 1%/moderate risk $\ge$1% and 5%/high risk $\ge$5% and 10%/very high risk $\ge$10%) were used. The total CV event risk was computed as follows: fatal risk $\times$ 3(4) for males (females). The SCORE system estimates the 10-year risk of a first fatal atherosclerotic CVD in populations of countries at low/high/very high CV risk, using sex, smoking status, age, systolic blood pressure and T-C as variables [1]. For comparison, the 10-year risk for hard coronary events (derived from the Framingham Heart Study—FHS) was also calculated, using age, sex, T-C, HDL-C, smoking status, DM, systolic blood pressure, and hypertensive treatment. FHS scores were classified as low (10%), moderate (10–20%), or high (20%) risk [21]. Based on the latest CVD prevention guidelines [13], we refined the calculated SCORE 10-year fatal CV risk by using carotid plaque burden to specify the personalized CV risk at baseline and after follow-up. The targeted dietary and pharmacological management of AH and DLP was satisfactory at the enrolment visit, no polypharmacotherapy was observed, and subjects were mainly treated with one prescribed drug. Based on personalized CV risk assessment, preventive measures (predominantly lifestyle modifications, or individually, if it was indicated according to the results on regular clinical and laboratory check-ups, pharmacological treatment) were recommended for each subject, to which they agreed. Adherence to treatment instructions was regularly checked by family doctors and by the study investigators at the end of follow-up.

Carotid arteries were examined using ultrasonography (USG) by the same experienced physician for all patients and all years and were blinded to the subjects’ health status and RFs. USG methodology as well as CIMT and carotid plaque definitions followed the Mannheim consensus and European recommended protocols [22, 23]. Bilateral carotid arteries were scanned using high-resolution B-mode USG (Philips HD 15) with the 7.5-MHz probe in real-time, at 5$\times$ magnification. IMT was measured automatically, on distinct plaque-free common carotid artery (CCA) posterior wall, 10 mm proximal to the flow divider, during end-diastole, at its presumed maximum thickness. We used the mean of 4 measured values for each side. The definition of a plaque was a focal wall protrusion into the arterial lumen of at least 1.5 mm or $>$50% of the surrounding IMT value. Plaques were recorded in transverse and longitudinal planes in 4 segments (CCA, bulb, internal carotid artery (ICA), external carotid artery (ECA)). No patient had significant carotid plaque with pathological peak systolic velocity (PSV) acceleration. Generally, carotid plaques were stable, with smooth surface and were isoechogenic at baseline and during follow-up. CCA parameters evaluated in our study included: mean right, left IMT (CIMTdx, sin), maximum IMT (CIMTmax), right or left IMT $>$0.9 mm (CIMTbilat $>$0.9) [24], abnormal age and sex adjusted mean right or left IMT (asCIMTbilat), i.e., in males/females on the left side, aged in years (y): 31–40 y: 0.57/0.51 mm, 41–50 y: 0.61/0.57 mm, $>$50 y: 0.70/0.64 mm; on the right side: 31–40 y: 0.5/0.49 mm, 41–50 y: 0.57/0.53 mm, $>$50 y: 0.62/0.59 mm [25, 26], CCA-IMT progression (mm/year), $\mathrm{\Delta }$ CIMT (difference in mean CIMT during follow-up) and presence of carotid plaque. Due to the identical progression rate and values (mean $\pm$ SD) of CIMT on both sides at 2 visits, for statistical analysis we used CIMTsin. The rate of IMT change per year was calculated by using the change of mean CCA IMT (between end and baseline of the study) divided by the time interval between the two ultrasound scans (approximately 5 years) [16]. Our intra-observer variability was acceptable (mean absolute difference = 0.085 $\pm$ 0.069; correlation coefficient = 0.88; coefficient of variation = 7.2%).

Patient’s data are summarized at baseline and at the end of follow-up. Continuous variables were presented as mean $\pm$ standard deviation (SD), categorical variables as the number of cases with percent frequency. Continuous clinical variables, including markers of subclinical carotid ATS between patients at baseline and at follow-up visit were compared using a paired samples t-test. A McNemar’s test was used to compare the frequencies of categorical variables in time between paired samples. The crude effect of traditional RF on dichotomized/continuous markers of carotid subclinical ATS was tested by binary logistic/linear regression analysis. Normal distribution of selected parameters was confirmed in CIMTdx and CIMTsin, however, our distribution tends to be bimodal or multimodal and still followed a normal curve on the graph. Crude effect of categorical (positive family history for CVD, CV risk age: in men/women $\ge$45/$\ge$55 years [27], sex, AH, DLP, smoking, central obesity, MetS) and continuous (duration of AH, T-C, HDL-C, LDL-C, TAG, number of RFs, SCORE) were assessed using univariate regression analysis. Correlations were tested with follow-up parameters, with suspected higher prevalence of subclinical ATS markers. In multivariate analysis we tested a mutually adjusted influence of all predictors from the univariate model on the analysed morphological parameters. Significant predictors of subclinical carotid ATS were established from the multivariate model using the backward (Wald) method. Parameters used in the univariate analysis were entered into the multivariate model. Pharmacotherapy as a covariate was not tested in the study due to the wide spectrum of analysed predictors and relatively small number of subjects on medication. During the follow-up a progression rate of CIMT was also calculated. The changes in computed (using SCORE charts) and personalized CV risk stratification were also compared at baseline and after follow-up. A value of p 0.05 was considered significant. Data analyses were performed in IBM SPSS 23.0 Statistics (IBM Corp., Armonk, NY, USA).

Out of the 187 initial, healthy, non-diabetic, 35–55-year-old (mean age 45.6 $\pm$ 5 years at baseline) subjects, 141 persons were examined after a follow-up of 4.67 $\pm$ 0.95 years. The characteristics of the study population at baseline and after follow-up are shown in Table 1. Worsening of the persons’ risk profile within 5 years is documented in Table 2. After follow-up we documented a significantly higher prevalence of DLP, central obesity, AH, as well as the number of RFs (3.78 $\pm$ 6.06; p 0.05). The risk SCORE and FHS score were increased from baseline to the end of follow-up (0.57 $\pm$ 0.93–1.16 $\pm$ 1.56 and 3.66 $\pm$ 4.73–6.18 $\pm$ 6,84, respectively), but still remained at a low-moderate risk level. Changes in analysed markers of subclinical carotid ATS during follow-up are listed in Table 3. The mean values of CIMT left and right (0.62 $\pm$ 0.10 mm; p 0.001, both) remained under the cut off level of 0.9 mm at follow-up but were significantly increased (by 0.08 $\pm$ 0.11–0.12 mm; p 0.001). A similar significant increase in maximum CIMT values (0.08 $\pm$ 0.12 mm; p 0.001) was also detected. The mean right and left CCA-IMT change/year was the same: 0.017 mm. The occurrence of CIMT $>$0.9 mm was rare (2.1%) and not significantly altered. However, the prevalence of asCIMTbilat was significantly increased from 52.9% to 78.8% (p 0.001). A similar increase in the rate of carotid plaque burden in the CCA, bulb, ICA and ECA was also observed (from 4.8% to 17.9%; p 0.001).

Based mainly on the surprisingly high prevalence of asCIMTbilat and carotid plaque burden in middle-aged, healthy population, we analysed the possible associations between classical RFs for ATS and carotid USG parameters, to determine their suitability as screening tests for subclinical ATS and eventually for personalized CV risk prediction. Statistically inconsistent associations were found (no relationship; borderline/weak significance) between classical RFs and CIMT parameters either continuous or dichotomous (including the mean CIMT difference between baseline and follow-up) in the univariate analysis (data are not shown). In contrast, a strong relationship was confirmed between all classical RFs (except for sex and positive family history) and the occurrence of carotid plaque (Table 4). Insignificant predictors are not shown. In the multivariate analysis we assessed the predictive power of the influence of classical RFs on markers showing progression of subclinical ATS. Total cholesterol was the only factor with a significant effect on the mean CIMT (p = 0.013; B = 0.024). A significant, but inconclusive influence of 2 lipid parameters was also detected on the pathological value of age- and sex-adjusted CIMT bilaterally (T-C: p = 0.019; Exp(B) = 58; LDL-C: p = 0.044; Exp(B) = 0.017). The multivariate regression model from the CIMT max showed statistical significance in non-lipid variables: CV risk score (p 0.001; B = 0.028) and male sex (p = 0.002; B = –0.087). Nonetheless, the higher number of RFs (as the only independent variable) increased the probability of plaque occurrence on carotid arteries by 1.7 (Table 5). The number of RFs was a weak determinant of individual 5-year CIMT progression (Table 5). In comparison with risk charts, by using the personalized approach (carotid plaque presence in addition to computed risk), the high CV risk was 4–5$\times$ more prevalent at baseline and at follow-up. At baseline, every subject was at low-to moderate (5%) calculated 10-year fatal CV (SCORE) and hard coronary event (FHS) CV risk. By detecting carotid plaque, in 4.8% of subjects the CV risk was reclassified into high. After 5-year follow-up, high CV risk according to the SCORE chart was present in 4.3% of subjects, but using personalized stratification, the prevalence of high CV risk was 17.9%. Due to the low number of individuals on hypolipidemic treatment at the time of enrollment, we did not monitor the effect of hypolipidemic treatment on the progression of carotid ATS.

The risk profile of our study group is comparable with the literature [17, 28], but obesity and DLP are increased in our study due to the fact, that we followed central obesity and had tighter cut-offs for DLP. In the large on-going PESA study with enrollment of participants without CVD, with no exclusion of diabetics, the study group had a better risk profile in term of DLP (40.9%) and obesity (13.3%), but the proportion of lipid-lowering therapy was similar (6.6%) [29].

Based on a systematic review, in low-to-intermediate risk individuals (mean age of 60 $\pm$ 7.6 years) the mean CIMT varied between 0.62–1.07 mm, and CIMTmax between 0.78–1.8 mm [7]. In the large ongoing Progression of Early Subclinical Atherosclerosis (PESA) study, similar to our results with a similar mean age, the mean CIMT value was 0.59 mm [4, 29]. The progression rate of the mean (SD) CCA-IMT in our study was in line with published data, with some limitations due to varying progression rates for CIMT reported in different population-based studies, ranged between 0.0038–0.060 mm/year [30, 31]. Comparable progression rates were reported in other studies [16, 32]. A mildly higher rate of CCA-IMT (0.025 mm/year) was observed in the large ARIC study [33]. The CAPS study reported a 0.001 mm/year progression rate of the CCA-IMT [17]. Increased CIMT represents subclinical vascular disease, may be related to intimal or medial hypertrophy or both, and may be an adaptive response to changes and is not clearly synonymous with subclinical ATS but is related to it due to similar alterations in the progression of both processes [34]. CIMT in subclinical vascular disease is a marker of CVD risk [35].

For CVD risk assessment, instead of normative values (i.e., pathological IMT $>$0.9 mm, reflecting primarily ATS at the carotid bifurcation and hypertension mediated hypertrophy at the level of CCA), carotid USG imaging and measurements should follow protocols with CIMT values in percentiles by age, sex, race/ethnicity and mostly also by side [25, 26, 36]. In comparison with previous data [28, 37], the occurrence of CIMT $>$0.9 mm was rare in our study and not significantly changed after 5-year follow up. CIMT $>$0.9 mm was detected in 1% of participants in the PESA study [4, 29]. There was a 36.7% incidence of CIMT $>$0.9 mm reported by Mitu et al. [28] among apparently healthy individuals, classified mainly in high risk SCORE, and an incidence of 34% was reported by Novo et al. [37] in an older study group, with a relatively high prevalence of diabetic and hypertensive patients.

Similarly to our results, the 75th percentile of the CCA-IMT distribution was established at 0.58 and 0.59 mm in healthy females and males without CV RFs, over 40 years of age [38, 39]. In a recent study of an apparently healthy population aged 57.7 $\pm$ 10.4 years, without exclusion of DM, the distribution of pathological CIMT $>$0.74 mm (75% percentile) was 25.96% (lower than in our study), but it followed a higher cut-off level in comparison with our study [40]. The prevalence of CIMT $>$75th percentile for the patient’s age, sex and race/ethnicity was approximately 12% across the Framingham study, but at intermediate FRS, 22–58% of patients had increased CIMT [41]. However, no data are available on the progression rate of pathological age- and sex-adjusted CIMT in the literature.

USG measures of carotid IMT and plaque are non-invasive methods for measuring ATS burden and strongly associated with vascular RFs and the incidence of CV events [35]. ATS progression predicts CV events [14]. The occurrence of carotid plaques seems to be variable in the general population and might be explained by geographical influence, age and the presence of CV RFs [28]. According to a systematic review [7], the occurrence of plaque in asymptomatic, low-to-intermediate risk cohorts, with different age and risk profile was an average of 35% (from 1.4% to 65.3%). Some studies [11, 28, 37, 40] in comparison to our results, reported a higher prevalence of carotid plaque (78%, 40%, 25%, 34%, resp.) probably due to the enrollment of older subjects. Data from studies with asymptomatic, middle-aged individuals documented higher occurrence of carotid plaques (29.3% in subjects with risk SCORE 5% [28], 31% in the PESA Study [29]). In the Refine study among 50–69-year-old participants, after a 4.2-year follow-up, in those patients without plaque at the first visit, the rate of plaque burden was 29.7%, which is a higher progression than in our study, but in a population with worse risk profile, with no exclusion of CVD [32]. Similar to our data, 20.5% of subjects developed new carotid artery plaques during a 5-year follow-up in a community in Taiwan (older subjects, no exclusion of DM) [16].

CIMT is associated with CVD RFs, the prevalence and incidence of CVD, and the degree of ATS in several different arterial beds [42]. In line with various studies in healthy populations, we documented associations between almost all classical CV RFs and CIMT parameters, however, in univariate analysis the associations were dominantly weak (datasets from univariate analysis are available from the corresponding author on request). Some studies showed robust correlation between age and the CIMT [43, 44]. In the Happy study, it was relatively better for the female cohort, which is partially in line with our findings [43]. Although the CIMT is thicker in men [26, 36, 38], sex does not independently predict the CIMT [45]. In other small cross-sectional studies of healthy subjects, age, BMI, waist circumference, systolic blood pressure (SBP) and diastolic blood pressure as well as TAG, HDL-C, glycaemia, and histories of CVD and type 2 DM [40, 44] were significantly associated with CIMT. Central obesity was significantly associated with CIMTmax and mean CIMT, while AH was only associated with CIMTmax in our study. There are non-consistent results in the literature regarding CIMT and lipoproteins. Most single-centre studies indicate the relationship between higher CIMT and higher levels of T-C, LDL-C, and non-HDL-C, as well as inverse associations with HDL-C [44, 45] (comparable to us); however, meta-analyses fail to show any associations [46, 47]. Similar controversies in association between HbA1c and the CIMT were revealed in non-diabetic individuals [48]. Alizargar et al. [44] showed a significant and strong correlation between HbA1c and CIMT, which we confirmed for MetS. With an increasing number of RFs, the increase of mean IMT in all carotid arterial segments was found in the Bogalusa Study [49], we confirmed weak associations between risk SCORE and mean as well as maximum CCA IMT values.

In the multivariate analysis, age appeared to be the most common independent predictor of CIMT [44, 45]. Apart from age, Alizargar et al. [40] also found, that waist circumference, SBP and C-reactive protein (CRP), and Paul et al. [49], also found that male sex, T-C/HDL-C ratio and smoking were common independent predictors of CIMT. Mitu et al. [28, 50] reported, that risk SCORE positively, significantly and also independently correlated with CIMT and the presence of carotid plaques in a small, clinically healthy, middle-aged cohort. In contrast, we found only a weak prediction of mean CIMT with T-C, and CIMT max with risk SCORE, but we found no positive influence of other RFs on CIMT. This is probably due to the limited range of age and our small study sample.

In a univariate analysis by Novo et al. [37], CIMT $\ge$0.9 mm or carotid plaque presence were related to the major CV RFs (age, AH, DM, HDL-C) and were independently associated with a major incidence of cerebro- and CV events. Similar data were published from a meta-analysis of 75 studies on the increased CIMT $>$1 mm and carotid plaque incidence with CV RFs [46]. In other studies age and waist circumference were predictors of high CIMT after adjustment for confounders [40, 45], with waist circumference being the strongest independent predictor [44]. Individuals with CIMT values $>$0.9 mm have a 4.1 times higher risk for being at a high SCORE CVD risk in a multivariate analysis [28]. We did not confirm any impact on CIMT $>$0.9 mm probably due to its low prevalence. In our subjects with central obesity there was 2.4$\times$ higher probability of CIMT detection over the age-and sex-specific cut off level, but after adjustment for confounders, the association disappeared. There are no comparable data in the literature.

Carotid ATS (IMT, plaques) is independently associated with all traditional RFs and CVD [35, 51]. Thickening of the CIMT reflects early stages of ATS, but plaque formation indicates later stages [52]. A meta-analysis of 76 cross-sectional studies with evaluation of 11 RFs, showed an association between the incidence of carotid plaque and AH, DM, DLP, current smoking, hypertriglyceridemia, LDL-C, hyperuricemia, hyperhomocysteinemia, and MetS [47]. We did not analyse the impact of DM and hyperhomocysteinemia, but found similar significant associations of RFs (also central obesity and SCORE risk) with the occurrence of carotid plaque. In line with our findings, Mitu et al. [50] documented an association of increased CV risk scores with the presence of carotid plaque and suggested the screening of subclinical ATS in subjects with a risk SCORE $\ge$3. Sex, T-C, LDL-C, TAG, non-HDL-C, waist, smoking and CVD risk score were associated with the risk of plaque formation in plaque-free subjects at baseline, however, in multivariable analyses, LDL-C was the only RF associated with plaque formation [32]. In addition, in the study from Taiwan, subjects with new carotid artery plaques during follow-up were older, hypertensive, and diabetic, but there was no association after controlling for CV RFs [16]. In asymptomatic individuals without DM, a positive association between HbA1c levels (also bellow the pre-diabetes cut off range) and subclinical ATS was identified even after adjustment for potential confounders (except for T-C) [53]. Fasting glucose levels showed a positive association with the prevalence and extent of subclinical ATS in univariate, but not in multivariate analysis [53]. In the PESA study, all traditional RFs as well as ATS CVD 10-year risk contributed to the progression of subclinical ATS across coronary and multiple noncoronary territories in the univariate analysis. In contrast, only age, male sex, and DLP were significantly associated with new ATS onset in disease-free participants [29]. Age, male sex, and all other CV RFs except obesity, DM, and estimated risk showed an independent association with ATS progression in multivariate analysis, with DLP as the strongest modifiable RF [29]. In our study, the only significant predictor of the presence of carotid plaque was the number of RFs. We only assessed the carotid region and analysed the correlation between the RFs and the occurrence of carotid plaque at the end of follow-up.

The presence of subclinical ATS is the major causal RF for CVD in asymptomatic individuals rather than a prominent additional predictive factor [54]. Carotid plaque burden may detect early stages of disease even before coronary calcification [11]. The Multi-Ethnic Study of Atherosclerosis (MESA) study showed that adding different plaque metrics with CIMT measurements to RFs significantly increased the association with the incidence of CVD events [55]. The BioImage study showed a significantly higher risk prediction performance of manual three-dimensional (3D) quantification of plaque thickness compared with two-dimensional (2D) measurements of plaque and CIMT [3]. A first clinical event in the 10 years follow-up was reported in 32% of subjects with carotid wall thickening and 62% with asymptomatic carotid plaque, moreover carotid subclinical ATS was related to the major CV RFs enhancing the predictive value of risk scores especially in the low- risk population [37]. Similarly, in the Heinz Nixdorf Recall (HNR) study CAC, CIMT, and ankle-brachial index (ABI) were associated with stroke in addition to established RFs [56]. CV disease primary prevention guidelines prioritize risk stratification by using clinical risk scores; beyond traditional RFs, CAC scoring, or the presence of carotid plaque as a high risk finding [1]. USG are sufficient for determining an accurate prediction of the CV risk in asymptomatic patients. Carotid USG results should be combined with other ATS factors, and a comprehensive risk assessment may help to guide CV prevention decisions. The observation in the Refine study that risk scores are predictive of new plaque formation in patients with no plaque at the first visit [32], the proposal of Mitu et al. [50] to screen populations with a risk SCORE $\ge$3, the characteristics of a screened population with $\ge$3 RFs for measurement of subclinical vascular disease with the aim of improving CV risk prediction [57], in line with our conclusions, may be of value to determine who should be given more attention to modify or treat individual RFs.