VCAM-1 Is Associated With High Cardiovascular Risk Predicted by the Framingham Score: A Cross-Sectional Study With Diabetic Women

The best strategy to establish cardiovascular risk (CVR) in women has yet to be dened, although risk scores are widely used. The inclusion of endothelial biomarkers, such as vascular cell adhesion molecule-1 (VCAM-1), to the risk scores could increase their discriminatory power and improve risk assessment. Objective To evaluate the association between endothelial biomarkers and CVR in women with type 2 DM (T2DM) and without previous cardiovascular disease (CVD). - BMI; blood pressure - SBP and blood pressure - glycemic and lipid proles and serum biomarkers VCAM-1, broblast growth factor-23, Syndecan-1 and Angiopoietin-2). The CVR was stratied using the Framingham Risk Score - FRS (version with laboratory tests - laboratory FRS and the one with the BMI - non-laboratory FRS) and the United Kingdom Prospective Diabetes Study (UKPDS) risk engine.


Introduction
The world's population of individuals with diabetes mellitus (DM) is approximately 463 million people. [1] Among women, the overall prevalence of DM is lower than in men (9.0% and 9.6%, respectively); however, in Central and South America, the number of diabetic women is higher than that of men. [1,2] About 1.6 million deaths are attributed to DM worldwide, with cardiovascular diseases (CVD) being the main cause of death in diabetic individuals, which is equivalent to about 2/3 of the total deaths. Considering gender differences, globally, in 2019, there were more deaths associated with DM in women (2.3 million) than in men (1.9 million). [1,2] Moreover, it is suggested that diabetic women have a higher risk of coronary artery disease (CAD), cerebrovascular disease, cardiac death and all-cause mortality compared to men. [3] The cardiovascular risk assessment in women population is essential to prevention of cardiovascular events. Risk scores can be carried out, such as the Framingham Risk Score (FRS) and the score derived from the United Kingdom Prospective Diabetes Study (UKPDS). [4,5] However, the literature shows con icting results about risk estimate when using these scores in women, with a previous risk overestimation by the FRS and the UKPDS having been demonstrated. The latter, although being a speci c score for the diabetic population, has already demonstrated a risk overestimation of 51.3% in women. 6,7] There is yet no de nition regarding the best CVR strati cation strategy in diabetic individuals [4,7] A metaanalysis evaluated different tools for CVR strati cation in diabetics and found that their performance varied signi cantly, even among those developed speci cally for this population. [8] Moreover, there are important differences between genders regarding the epidemiological and statistical determinants of the performance of risk prediction models. [9] Therefore, improvements in the predictive capacity of these models, such as the possible addition of speci c biomarkers, such as vascular cell adhesion molecule-1 (VCAM-1), broblast growth factor-23 (FGF-23), Syndecan-1 (Sdc-1) and Angiopoietin-2 (Ang-2), would be required to understand the short and long-term risks before implementing them into clinical practice. [8] However, there is no de nition of which biomarkers could be used for the best estimate of the occurrence of cardiovascular events and the studies still bring con icting results regarding the role of biomarkers in predicting CVR in the overall population and the diabetic populations. [4,10,11] Research conducted by Ren et al. [12] and Llauradó et al. [13] found no association between VCAM-1 and the occurrence of CVD, respectively, in the overall population and in diabetics. As for the study by Gardner et al. [14] , it showed higher levels of VCAM-1 in Caucasian women, when compared with Caucasian men with peripheral obstructive arterial disease (POAD). Women in this study still had more severe POAD than men, with worse exercise performance and daily outpatient activity.
Considering the existing gaps in CVR strati cation in diabetic women, this study aimed to stratify cardiovascular risk by different scores and evaluate the association between the predicted risk and serum vascular and cardiorenal biomarkers in women with T2DM.

Study type and location
This is an observational, cross-sectional study that evaluated women with T2DM, aiming at assessing their CVR and its association with new serum endothelial biomarkers. The study was carried out from January to October 2017, in a Primary Health Care (PHC) unit in the city of Fortaleza, state of Ceará, a large urban center located in northeastern Brazil.

Study population
The study population comprised women with T2DM, followed at the aforementioned PHC unit. The criteria de ned by the American Diabetes Association [15] were adopted for the diagnosis of T2DM.
Women with T2DM, aged 30 to 74 years, who lived in the area assigned to PHC unit and who signed the free and informed consent form, were included in the study. This age limit was de ned, since the FRS to be used in primary care was validated with individuals who were included in this age group. [16] The exclusion criteria comprised women with a previously diagnosed CVD (CAD, stroke, POAD or heart failure -HF) or who had symptoms possibly related to CVD; those with malignant neoplasms of any kind and / or active in ammatory disease; those who had an infection with systemic effects at the time of the assessment; those whose electrocardiogram (EKG) examination performed during the study was suggestive of the presence of CVD, according to previously established criteria [17] and pregnant or postpartum women.

Patient recruitment
The participants' recruitment was carried out through a public announcement with posters and pamphlets and through a direct invitation to possible participants. Interested individuals were shown the research objectives, their risks and bene ts, and explanations of possible doubts were provided.
Women were evaluated according to the following sequence for the sample selection: electrocardiogram (EKG) to assess possible alterations compatible with CVD, performed on a portable 3-channel Philips EP-The participants were instructed to come on a scheduled day and time, so that data collection could be carried out, according to the following steps: 1) Anamnesis -the following parameters were evaluated: sociodemographic data (age, ethnicity, level of schooling) and clinical history (time since the T2DM diagnosis; presence of systemic arterial hypertension -SAH; use of antihypertensive medication; personal pathological history and smoking status); 2) Physical examination -the following data were assessed: height and weight measurement, to calculate the body mass index (BMI), as recommended [18] and blood pressure (BP) measurement, with the recording of systolic BP (SBP) and diastolic BP (DBP), as previously recommended. [19] Women who stated being hypertensive and who regularly used antihypertensive medications for BP control purposes in the past month were considered as having previous SAH. Women who smoked at least one cigarette in the last 30 days were considered to be current smokers.
Blood sample collection and processing Blood was collected from the participants using the venipuncture technique, after a 12-hour fast. The samples were placed in vacuum collection tubes, with separating gel, for further processing of these samples in the biochemical tests and serum biomarkers, with the exception of glycemia, for which the sample was placed in a tube with sodium uoride and ethylenediaminetetraacetic acid (EDTA) and glycated hemoglobin A1c (hbA1c), for which the samples were stored in tubes with EDTA. The collected samples were taken for processing within an average time of 2 hours.
The other blood samples were centrifuged at 3500-4000 rpm for 15 minutes to obtain the serum, which was aliquoted for storage in 3 1mL-Eppendorf tubes. Subsequently, these samples were stored in a freezer at -80 o C.

Laboratory analysis
HbA1c was measured in the total blood sample of each participant, using the high performance liquid chromatography technique. For the overall biochemical evaluation, the following laboratory tests were performed in the stored serum samples, using the colorimetric enzymatic method: fasting glycemia (FG), total cholesterol (TC), high-density lipoprotein cholesterol (HDL-c) and triglycerides (TG). Levels of lowdensity lipoprotein cholesterol (LDL-c) were obtained using Friedewald's formula. [20] Analysis of biomarkers All of serum endothelial biomarkers were analyzed using the "sandwich" type Enzyme-Linked Immunosorbent Assay (ELISA) technique, according to the manufacturer's instructions for the assays. Speci c kits were acquired for each biomarker: for FGF-23 (Duoset DY2604; R&D Systems®), VCAM-1 (ab47355; Abcam®), Sdc-1 (ab47352; Abcam®) and Ang-2 (Duoset DY623; R&D Systems®). All analysis were performed in isolated serum samples.

Cardiovascular risk strati cation through the scores
For the analysis of CVR in this study, the following tools were used: the FRS for primary care using lipid fractions (laboratory FRS), the FRS using the BMI (non-laboratory FRS) and the UKPDS risk engine.
The CVR by laboratory FRS was assessed using the online calculator. [21] For its calculation, the following data were used: age, presence of diabetes, smoking status, SBP value in mmHg, presence of treatment for SAH, and HDL-c and TC values. [16,22] The non-laboratory FRS was also calculated using a tool available online. [21] For this purpose, the same data used for the laboratory FRS were utilized, with the exception of the TC and HDL-c, which were replaced by the BMI. The risk of CAD, stroke, POAD or HF) in 10 years was assessed using these scores. Regarding the CVR analyzed by these scores, the participants were classi ed as having low (< 5%), intermediate (calculated risk ≥ 5% and ≤ 10%) or high risk (> 10%).
Additionally, the vascular age of each participant was estimated according to both scores. [16,22] Data on age, T2DM duration, gender, presence of atrial brillation, ethnicity, smoking status, hbA1c, SBP, TC and HDL-c values were used to calculate the UKPDS score. With this tool, the risk of fatal or non-fatal CAD and stroke in 10 years was analyzed for each participant. [23,24] According to UKPDS risk engine, the participants were classi ed as having high (> 20%), intermediate (10 to 20%) or low (< 10%) risk for any of the outcomes. [23,24] Statistical analysis Categorical data were expressed as absolute count and relative frequency in percentages and were compared using the chi-square test. All quantitative variables were assessed regarding a normal distribution, using the Kolmogorov-Smirnov normality test and evaluation of variance and histograms to verify data asymmetry. Variables with normal distribution were presented as mean ± standard deviation and non-normal data were shown as median and interquartile range. The Analysis of variance (ANOVA) test with Tukey's post-test were applied to compare means of continuous variables. Additionally, Spearman's correlation analyses were performed to assess the association between quantitative variables. Signi cance was set at p < 0.05 (two-way). All analyses were performed using the Statistical Package for Social Sciences (SPSS) version 23.0 for Macintosh (IBM, Armonk, NY, USA).

Ethical aspects
This research was carried out in accordance with Resolution n. 466/2012 of the National Health Council, after approval by the Research Ethics Committee of University of Fortaleza (Opinion n. 1.843.144 / 2016).

Results
A total of 88 women with T2DM were included in this study. Table 1 shows the main characteristics of the assessed population. The mean age of the women was 56 ± 10 years, with the majority being classi ed as brown (81.8%). Only 4.6% of the sample had a higher education level; 17.2% were illiterate and 70.1% had not entered high school. Quantitative data expressed as mean ± standard deviation or as median and interquartile range in parentheses. Categorical data expressed as absolute count and percentages in parentheses.
When evaluating the clinical characteristics, 47.7% were smokers at some point in their lives (current or past smoker) and 71.6% had SAH as a comorbidity. The median time of T2DM diagnosis was 5 years (IQR 3-9 years). The analysis of body composition showed that most of the study sample was overweight (32%) or obese (42%).
The laboratory data showed high blood glucose levels, with a mean fasting plasma glucose of 8.53 ± 3.65 mmol/L and hbA1c of 65 ± 1 mmol/mol. The analysis of lipid levels showed high mean levels of LDL-c (27.06 ± 8.99 mmol/L).  Quantitative data expressed as median and interquartile range in parentheses. Categorical data expressed as absolute count and percentages in parentheses. Table 3 shows the correlations between CVR scores and vascular biomarkers. Among the biomarkers, VCAM-1 was the one that showed the best correlation with the assessed risk scores, although the correlation was weak.  When performing a more detailed analysis (Table 4) and when correlating the values of VCAM-1 with the risk strati cation categories, we observed that, for the laboratory FRS, there was a statistically signi cant association between the CVR categories and VCAM-1 (p = 0.024), so that individuals with a higher CVR also had higher values of serum VCAM-1. For the other assessed scores, there was no statistical signi cance in the observed associations. * ANOVA-test with Tukey's posttest was used for all comparisons. The signi cance occurred among "low risk" versus "high risk" group. Figure 1 shows that, when the low-risk group is compared with the high-risk group according to the Laboratory FRS, those with the highest CVR have, in fact, higher values of VCAM-1 (p < 0.05).

Discussion
The present study is the rst to attempt to establish a relationship between biomarkers and the CVR predicted by different risk tools in diabetic women using the scores and biomarkers evaluated by us. We demonstrated that the majority of these women had high CVR according to the scores derived from the Framingham cohort and low CVR according to the UKPDS score. Moreover, serum levels of VCAM-1 were directly associated with a high CVR estimated by the laboratory FRS, even in patients without previous cardiovascular disease. three models was consistently better in women than in men, which makes us think that the reproducibility degree in the implementation of these scores in the population of our study may have been reliable, as our sample was consisted by women only. Moreover, great differences were observed regarding performance between the validations of the same model in different populations, [25] which shows the need to create or adapt a score validated for speci c populations, such as the Brazilian one. However, all of these models overestimated the risk of CVD development, especially in high-risk populations, [25] which corroborates the idea that it is necessary to correlate other factors, by adding other criteria to the scores, as in the case with biomarkers, to increase their degree of reliability.
In a study designed to assess the risk of CVD in adults with T2DM and metabolic syndrome and to compare the Framingham risk scores and the UKPDS, Korean adults were evaluated and these scores were compared in this population. No signi cant differences were observed between the two scores and it was shown that approximately a quarter of the adults had a high risk of CVD (> 20%). [26] In another study that compared several scores (UKPDS, FRS, Atherosclerotic Cardiovascular Disease -ASCVD, and Joint British Societies for the prevention of Cardiovascular Disease -JBS3) in South Asian and Caucasian populations with T2DM, a high prevalence of subclinical CAD was veri ed in high-risk patients through all scores, with JBS3 showing the highest correlation, despite the higher rate of low-risk classi cation in the studied population according to all scores. [27] Another study that evaluated the performance of the Framingham and UKPDS scores showed that the UKPDS was the score with the best 10-year CVD risk prediction in patients with T2DM, when compared to the Framingham equation. [28] As for the comparison of scores in the female population, the study by Cook et al. [29] showed that Reynolds score was more effective than the Framingham score for risk strati cation, but this result was not observed in diabetic women. In the present study, when the CVR was classi ed by the laboratory FRS and the non-laboratory FRS, it was observed that the majority of the population was classi ed as high risk in both of them (72.7% and 81.8%, respectively). When using the UKPDS, both for CAD risk and for the risk of stroke (fatal or non-fatal), the highest percentage of the women was classi ed as low risk. With the con icting results in the analysis of these scores, including the discrepancies found in the results of the present study, it will be necessary to improve the scores in an attempt to enhance their predictive power, so they can be used with greater reliability in the female population with T2DM.
Regarding the Framingham risk scores and the laboratory pro le, our sample showed that most of the participants were classi ed as high risk, with the laboratory data showing high glycemic levels, with a mean FG of 8.53 ± 3.65 mmol/L and hbA1c of 65 ± 1 mmol/mol. It has been de ned in the literature that goals not reached by diabetic patients imply a higher risk of unfavorable outcomes. This was also demonstrated in the study by Kim et al, [26] in which the inadequate glycemic control was associated with a high risk of CVD. The mean level of hbA1c in the high-risk group (hbA1c = 69 mmol/mol) was higher than in the low-risk group (hbA1c = 56 mmol/mol), corroborating this association.
Because it is a disease that results in endothelial dysfunction, T2DM can predispose to cardiovascular complications and thus, some biomarker measurements can be considered for CVD assessment and their alterations may constitute an increased risk of future complications. [30] A study evaluated the relationship of T2DM and glycemic control with circulating cell adhesion molecules (CAMs) and showed that VCAM-1 was signi cantly higher in patients with T2DM than in healthy individuals. [31] Another study evaluated 23 biomarkers of different pathophysiological pathways to improve the risk prediction of cardiovascular events in patients with T2DM, in addition to the traditional risk factors. It was observed that markers such as the N-terminal fragment of B-type natriuretic peptide (NT-proBNP), osteopontin, metalloproteinase-3 of the extracellular matrix and their combination improved the prediction of the cardiovascular event risk in this population. [32] These studies point to the fact that the use of serum biomarkers in predicting CVD can be an objective tool in the evaluation of these patients, considering that the traditional risk scores do not show uniform results.
In a prospective study with diabetic patients without manifest macrovascular disease followed for 5 years, some biomarkers (intercellular adhesion molecule-1 -ICAM-1, VCAM-1, P-selectin and E-selectin) were measured in beginning and during follow-up. Baseline ICAM-1 was found to be signi cantly higher in patients who developed macrovascular disease than in the ones who did not. [33] Kocijancic et al. [34] demonstrated, in patients with chronic dialysis kidney disease, that the concentrations of ICAM-1 and VCAM-1 had a strong independent correlation with carotid intima-media thickness (IMT) and that ICAM-1 and omentin-1 were strong predictors of cardiovascular death and progression of IMT. The present study does not correlate biomarkers with the incidence of CVD but assesses the relationship between such biomarkers and the event risk prediction assessed by scores. This association can generate hypotheses that can be con rmed with subsequent studies of more robust incidence analysis.
Another study assessed the baseline activity of a disintegrin and metalloproteinase domain 17 (ADAM17), by measuring the levels of the four main circulating substrates (VCAM-1, ICAM-1, Interleukin-6 and the soluble tumor necrosis factor-receptor 1) and correlated them with a second major cardiovascular event (cardiovascular death, peripheral arterial surgery and non-fatal acute myocardial infarction and stroke). A score was created based on the substrates of ADAM17, correlating it to the Framingham Recurring-Coronary-Heart-Disease-Score and a signi cant increase was observed in the score prediction accuracy, with an important improvement of the correct classi cation in 10% of events and 41% of non-events. [35] Few studies in the literature have attempted to correlate the use of biomarkers, especially those analyzed in the present study, and the CVR predicted by scores in the diabetic population. [36][37][38][39][40][41] One of these studies evaluated the association between a set of biomarkers that analyze different metabolic pathways, such as Asymmetric dimethylarginine (ADMA), Soluble endothelin-1 (ET-1), Placental growth factor-1 (PIGF-1) and NT-pro-BNP and the CVR predicted by the UKPDS and Action in Diabetes and Vascular disease (ADVANCE) scores in diabetic patients. This study showed that ADMA and PIGF-1 were not associated with CVR strati cation with any of the scores, while ET-1 was associated with the risk of stroke by the UKPDS and NT-proBNP was associated with CVR predicted by both tools. [39] Different from our research, these studies did not evaluate the female diabetic population, or the biomarkers analyzed by us.
In our study, it was observed that, of the assessed biomarkers, VCAM-1 is the one that showed the best correlation with the risk scores, although this correlation was weak, which may have been due to the small study sample. We also observed that higher levels of VCAM-1 were found in patients classi ed as having high CVR by the laboratory FRS. VCAM-1 is a protein that mediates the adhesion of lymphocytes, monocytes, eosinophils and basophils to the vascular endothelium and can play a role in atherosclerosis development. It is thought that one of the reasons why this can occur in patients with T2DM is the fact that their HDL-c has less anti-in ammatory capacity. This fact was demonstrated in the study carried out by Ebtehaj et al., [42] which showed that, in individuals with T2DM, the anti-in ammatory capacity of HDLc was strongly impaired, with a greater increase in the expression of VCAM-1 messenger ribonucleic acid, which indicates less anti-in ammatory capacity.
Studies are contradictory when they try to correlate the CAMs, such as VCAM-1 and ICAM-1, with cardiovascular events in a population of individuals without previous CVD and it seems that the same occurs with studies in the subpopulation of diabetic patients, as previously reported. [30] According to Derosa et al., [43] pro-atherogenic CAMs (ICAM-1, VCAM-1 and E-selectin) are elevated in T2DM and their increased expression and release would contribute to accelerated atherogenesis in diabetic patients. Also correlating VCAM-1 and diabetes, a study group evaluated patients with a recent diagnosis of DM and measured several markers in serum, including VCAM-1, and correlated them with cardiovascular function (represented by the measurement of ow-mediated vasodilation, intima-media thickness and arterial stiffness). In this study, a positive relationship was found between IMT and between arterial stiffness with VCAM-1. [44] This demonstrates the presence of an increase in the expression of VCAM-1 in diabetic patients with atherosclerotic disease evidenced by these methods.
The studies show that CAMs such as VCAM-1 are established markers of endothelial dysfunction, corroborating the data from our study, which showed higher levels of this biomarker in patients at higher cardiovascular risk. It will be necessary to monitor our studied sample to document future cardiovascular events and verify the discriminatory power of both risk scores and the measurement of biomarkers, especially VCAM-1.

Study limitations
The main limitation of this study is related to its cross-sectional design, not being possible to establish an association between the assessed biomarkers and cardiovascular outcomes, which would be able to assess the discriminating power of both the biomarkers and the applied risk scores in this population. Moreover, other important points are the fact that no risk scores have been validated for the Brazilian population and the fact that the present study was carried out in a single center. These factors may also have caused bias in the present study. However, the data described by us in this study are important because they correlate biomarkers, especially VCAM-1, with CVR predicted by scores widely used worldwide in clinical practice for the risk strati cation in the female population and constitute an important topic to be further studied in future research.

Conclusion
We evaluated women with T2DM without previous CVD and found that most of them had high CVR according to the Framingham scores and low risk according to the score that is more speci c for diabetics (UKPDS). Moreover, we found that the VCAM-1 biomarker was directly associated with the CVR estimated by the laboratory FRS, which may indicate the presence of endothelial injury and subclinical atherosclerosis. The ndings described here may point to the need to nd biomarkers, such as VCAM-1, that can further re ne the risk analysis by cardiovascular risk scores, tools that are widely used in CVR strati cation in the female population with and without diabetes in the most diverse levels of health care. The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Ethics approval and consent to participate
The present study is in full agreement with Resolution n. 466/2012 of the National Health Council and obtained approval from the Research Ethics Committee of Universidade de Fortaleza (Opinion n. 1.843.144 / 2016). Each study participant provided written informed consent.

Consent for publication
Not applicable.

Availability of data and materials
Not applicable.

Competing interests
Not applicable.

Funding
Not applicable.
Authors' contributions AKMN and GBSJ contributed with study concept and design, data collection, analysis and interpretation, as well as the writing of the manuscript. GCM contributed with data collection, statistical analysis and review of the manuscript. DOCL, AMCM, APPL, RVBMC, JHSJ and JMOL contributed with data collection and writing of the manuscript. RPS contributed with data interpretation, review of the manuscript and nal approval of the version to be published. All authors read and approved the nal manuscript.