CC BY
9
International Heart and Vascular Disease Journal Volume 4, Number 10, June 2016
Journal of the Cardioprogress Foundation
ORIGINAL ARTICLES
The probabilistic calculator for prediction
of coronary atherosclerosis risk in patients
with obesity
Veselovskaya N.G., Chumakova G.A.*, Shenkova N.N., Osipova E.S.
Scientific Institution "Institute for Complex Problems of Cardiovascular Disease," Kemerovo, Russia
Altai State Medical University, Barnaul, Russia
Galina A. Chumakovа, MD, professor of Postgraduate Medical Faculty, Altai Medical University, leading researcher at the Department of multifocal atherosclerosis, Institute for Complex Problems of cardiovascular disease; Barnaul, Kemerovo, Russia
Nadezhda G. Veselovskaya, PhD, senior researcher at the Research Institute of the Department of multifocal atherosclerosis, cardiologist at Altai Regional Cardiology Clinic, Barnaul, Kemerovo, Russia
Natalia N. Shenkova, Ph.D student of the Department of hospital and outpatient therapy, Altai State Medical University, Barnaul. Russia
Elena S. Osipova, Ph. D. student of the Department of hospital and outpatient therapy, Altai State Medical University, Barnaul, Russia
Summary Objective
To create a method of coronary atherosclerosis prediction in patients with obesity.
Material and methods
This study involved 85 men, 39-65years (average age of47,68±6,65years) with absence of clinical manifestations of coronary heart disease and atherosclerosis of other localizations. Patients had the obesity of the I-III degree, BMI 36,23±4,31 kg/m2 and visceral obesity in case of epicardial fat tissue thickness^7 mm. 2 groups of comparison were identified according with the performed coronary angiography or multislice spiral computer tomography of coronary arteries. Group I (n=35) included patients with existence of coronary atherosclerosis, Group II (n=50) included patients with absence of coronary atherosclerosis.
As the result of comparison of two groups of an arterial hypertension the existence of carbohydrate violations, triglycerides, leptin, adiponectin and C -reactive protein have been identified as possible predictors of coronary
Results
Corresponding author: Tel. 8-903-910-80-40. E-mail: g.a.chumakova@mail.ru
atherosclerosis risk. Each predictor received its coefficient of importance after the regression analysis with optimal scaling importance. The size of right classifications as a result of logistic regression was 79,1% that indicates a good predictive ability of this regression model.
Conclusion
The created scale allows to estimate risk of coronary atherosclerosis in the absence of disease clinical manifestations, that is important in terms of well-timed preventive actions and the prevention of the disease progression.
Key words
visceral obesity, coronary risk, scale
It has been proved that neurohumoral activity of visceral fat including epicardial fat tissue has an important role in cardiovascular complications (CVC) development in patients with obesity [1, 2]. Due to this the identification of high coronary risk group considering the presence of normal obesity and in particular visceral one would allow to plan and perform prophylaxis interventions in order to prevent CVC with well timing.
At the same time existing scales of coronary risk stratification (Framingham, PROCAM, SCORE) do not take into account main pathogenetic mechanisms that connect obesity and CVC [3-7].
The aim of this study is to create probabilistic calculator for coronary atherosclerosis prognosis in patients with obesity.
Materials and methods
85 males of 38-65 years (average age 47,68±6,65 years) without clinical manifestations of stenocardia and atherosclerosis were included in this study. All patients had I-III grade obesity, body mass index (BMI) around 36,23±4,31 kg/m2 and visceral obesity according with the epicardial fat tissue thickness (EFTS) >27 mm. Patients with severe concomitant diseases, diabetes mellitus 2 type (DM-2) and bad quality of echocardiography (EchoCG) visualization were excluded from this study.
During enrollment of the study we measured height and weight of patient, quantified BMI according with the formula (weight (kg)*height (m2). When BMI was s 30 kg/m2 patient was diagnosed with normal obesity. Total cholesterol (TC), triglycerides (TG), high density lipids cholesterol (HDLC), low density lipids cholesterols (LDLC) and glucose levels were measured in all patients. Lipoprotein A (LPA), apoli-poprotein B (Apo B) and apolipoprotein A1 (apo A1) levels were measured with immune precipitation method. Leptin, adiponectin, resistin levels and in-terleukin-6 (IL-6) and tumor necrosis factor-a (TNFa) concentration in blood serum were identified using
enzymoimmunoessay (EIE) kit (BioSource, Belgium). C-reactive protein (CRP) concentration was measured using highly sensitive latex-enhanced immune precipitation kit (Thermo Fisher Scientific, Finland). Epicardial fat tissue thickness was estimated with transthoracic EchoCG and Vivid 5 ultrasound machine (General Electrics, USA) with mechanical sector sensor 3,5 MHz. Three cardiac cycles in long parasternal position were registered. EFTS was measured behind free wall of the left ventricle in the end of systole. on the line that was maximally perpendicular to the fibrous ring of aortic valve used as an anatomic reference point [8, 9]. To estimate subclinical coronary atherosclerosis we performed selective coronaroan-giography (CAG) with angiographic machine INNOVA 3100 (USA) or multislice computer tomography (MCT) of coronary arteries (CA) on Aquilion-64 ("Toshiba", Japan) tomography with the proceeding of data at the working station VITREA.
According with CAG and MCT results, we divided the patients with epicardial obesity into two groups: group 1 with the signs of coronary atherosclerosis (n=35) and group 2 (n=50) without signs of coronary atherosclerosis.
STATISTICA 10 and SPSS-21 software were used for statistical analysis. 0,05 was taken as the critical level of statistical significance during checking of the null hypothesis. Normality estimation for quantitative characteristics in compared groups was performed with Kolmogorov-Smirnov and Shapiro-Wilk tests. Descriptive statistical values are present in the text of the article as M ± SD in case of characteristic's normal distribution, where M is an average value, SD -standard deviation, or as Med in case of not normal distribution. To compare central group parameters we used parametric and non-parametric methods: Student's T-test or Mann-Whitney U-test. To create the scale of coronary atherosclerosis prognosis and as a regression model we chose Regression with Optimal Scaling (CATREG) model that was performed with SPSS software.
Results
In order to analyze the relation between coronary atherosclerosis with possible predictors we made preliminary comparative analysis of two comparison groups: group I (n=35) and group II (n=50) in major and additional metabolic, neurohumoral risk factors (RF) and vascular inflammation markers that after were used in current study.
We defined the list of characteristics that had statistical relation with dependent variable - the presence of coronary atherosclerosis, and after formed the list of variables for regression analysis.
Thus, the list of possible predictors included: presence of arterial hypertension (AH), presence of carbohydrate metabolism abnormalities (impaired fasting hyperglycemia or impaired glucose tolerance (IGT)), TG, leptin, adiponectin and CRP. We performed receiving operating characteristic (ROC) analysis to define threshold levels of quantitative predictors and interval variables' reduction.
Optimal level of TG that was taken as a cut-off point was 1,8 mmol/mL(sensitivity 72%, specificity 66,7%), for leptin - 12,8 ng/mL (sensitivity 80%, specificity 64%).
Cut-off point for adiponectin was 10 pmol/mL (sensitivity 84%, specificity 45%), for CPR cut-off point was 5 mg/mL (sensitivity 64%, specificity 76%).
Table 1. Regression analysis results of coronary
After obtaining cut-off points we made regression analysis with optimal scaling to estimate significance of predictors. These importance coefficients were chosen as weight values for creating the scale. For each one of 6 predictors included in regression model we quantified the points by multiplying absolute value of appropriate importance coefficient and 100 and its rounding to integer (Table 1).
Thus we created preliminary version of this risk scale (Table 1). We tested if the created regression model was adequate by binary logistical regression. The value of true classifications was 79,1%, that is considered high enough and indicates of good prognostic ability of this regression model.
Using this equation, we quantified theoretical values of subclinical coronary atherosclerosis presence probability for each patient. Dispersion diagram that reflects this relation is present at Image 1.
Optimal cut-off value for sum of points according with the results of ROC-analysis that allowed to divide the patients into two groups was 58 points. Thus, if the probability of coronary atherosclerosis is s40%, its risk can be considered as high (Image 1).
To make this risk calculator easier to use we created its version for MS Excel, MS Office 2010 software (Table 2).
lerosclerosis predictor significance estimation
Predictor Cut-off values Standardized coefficients p-value Partial correlation coefficient Importance coefficient Points
Beta Standard error
TG s1.8 0.262 0.105 0.015 0.314 0.234 +23
Leptin S12.8 0.240 0.147 0.107 0.213 0.246 +25
Adiponectin «10.0 0.060 0.087 0.493 0.063 0.043 +4
CRP s5.0 0.233 0.128 0.074 0.251 0.222 +22
AH Present 0.189 0.089 0.039 0.233 0.126 + 13
Carbohydrate metabolism abnormalities Present 0.236 0.102 0.024 0.278 0.129 + 13
Comment: Beta-coefficient reflects summarized impact of predictor on response value, partial correlation coefficient reflects independent impact of predictor on response value.
Logistical regression (coronary atherosclerosis) P-l/(l+exp(-4,607+0.075*points)
Image 1. Equation and logistic regression function plot that reflects the relation between the sum of points and probability of coronary atherosclerosis presence.
Table 2. MS Excel interface of the risk calculator for sublinical coronary atherosclerosis prognosis
А B C D F G H I J K М
Predictors TG Leptin Adiponectin CRP AH IGT A0-4.607 B 0.075 Sum of ponts P (%)
Cut-off values s1.8 S12.8 «10.0 s5 1 (0) 1 (0)
Weight points 23 25 4 22 13 13
Patient's characteristic
Comments: 1. RF information of each patient should be put to the cells of the table. 2. Probability of coronary atherosclerosis presence (%) would appear in «M» cell.
This calculator contains the data of logistical regression. Numerical characteristics of analyzed prognostic criteria for each patient should be put into this calculator and the value of prognostic risk would automatically appear in percentage format in the "M" cell of this table.
Discussion
In this study coronary atherosclerosis in clinically unsuspected patients with visceral obesity was verified in 35 (41%) of patients. In other studies it was identified that 61% of patients of morbid obesity group and BMI s40 kg/m2 without coronary heart disease (CHD) clinical manifestations (average age 50,4±10,0 years, BMI 43,8±4,8 kg/m2) had stenosis at least of one CA [10]. Another study that was made in Latin American population, where 88,7% of participants had obesity and 53,2% of participants had metabolic syndrome, identified carotid arteries atherosclerosis signs in 34,8% of cases according with the results of Doppler ultrasound scanning [11]. Patients with metabolic syndrome that was diagnosed according with the ATP III classification and without CHD clinical manifestations had CA calcinosis in 24,7% of cases [12].
In one of Russian studies atherosclerosis plaques were found in 35% of patients of 30-55 years with abdominal obesity [13].
In our study coronary atherosclerosis predictors that are appropriate for risk prognosis were: AH presence, carbohydrate metabolism abnormalities - impaired fasting hyperglycemia or IGT, TG, leptin, adiponectin and CRP.
The correlation between CRP levels and intima-media complex thickness in carotid arteries was found in one study [14]. Previously the connection between CRP and atherosclerotic lesions of CA and other peripheral arteries was demonstrated in another study [15]. It was also proved that CRP and oxidized LDLC have direct relation to inflammatory lesions of arteries in CHD [16]. Another study identified the association of proinflammatory marker IL-6
and CA calcinosis [17]. Leptin levels were associated with CA calcinosis independently from weight and other RF that proves proatherogenic role of leptin [18]. It is known that CRP is one of the main chronic inflammation markers and it participates directly in CA atherosclerosis progression [19]. One study demonstrated that visceral fat tissue stimulates CRP synthesis [20].
Patients with CHD and low adiponectin levels have more prominent atherosclerotic CA lesions according with CAG results comparing with the patients with higher concentrations of this protein [21]. Low adiponectin concentrations in combination with high IL-6 levels in patients with obesity and metabolic syndrome were associated with the risk of cardiovascular disease development, and the highest risk of DM-2 and CHD development was found in patients with the combination of low adiponectin and HDLC levels [22]. Low adiponectin concentration had positive correlation with the degree of CA calcinosis and asymptomatic stenosis that was identified with angiography technique in patients with DM-2 or without it [23].
Conclusions
Investigation of neurohumoral and proinflammatory activity of visceral fat tissue proved the connection of these factors and coronary atherosclerosis. At the same time visceral obesity degree, adipokines and proinflammatory markers still haven't been used in any scale of coronary risk estimation that reduces significantly the precision of cardiovascular risk evaluation in patients with obesity.
Probabilistic calculator of coronary atherosclerosis risk prognosis in patients with visceral obesity that have been created by the authors of this study allows to take into account main pathogenetic mechanisms that link obesity and coronary atherosclerosis. Evaluation of TG, leptin, adiponectin, CRP, presence of AH and carbohydrate metabolism disorders in every single patient with visceral obesity allows to predict the presence or absence of subclinical coronary artery atherosclerosis with the probability of 79,1%,
that is particularly important for early prophylaxis and prevention of disease progression.
Conflict of interest: None declared
References
1. Juge-Aubry CE, Henrichot E, Meier СА. Adipose tissue: a regulator of inflammation. Best Pract Res Clin Endocrinol Metab. 2005;19(4): 547-66.
2. Chumakova GA, Veselovskaya NG, Kozarenko AA, et al. Features of morphology, structure and function of heart at obesity. Russian Cardiology J. 2012;4:93-9. Russian
3. Assmann G, Cullen P, Schult H. The Munster Heart Study (PROCAM) Eur Heart J. 1998;19 (Suppl A):2-11.
4. Brindle P, Emberson J, Lampe F, et al Predictive accuracy of the Framingham coronary risk score in British men: prospective cohort study. BMJ. 2003;327:1267-70.
5. Conroy RM, Pyorala K, Fitzgerald AP, et al Estimation of ten-year risk cardiovascular disease in Europe: the SCORE project. Eur Heart J. 2003; 24:987- 1003.
6. Pocock SJ, McCormack V, Gueyffier F, et al. A score for predicting risk of death from cardiovascular disease in adults with raised blood pressure based on individual patient data from randomized controlled trials. BMJ. 2001;323: 75-81.
7. Thomsen TF, McGee D, Davidsen M, et al. A cross -validation of risk -scores for coronary heart disease mortality based on data from the Glostrup Population Studies and Framingham Heart Study. Int J Epidemiol. 2002;31: 817-22.
8. Thomsen TF, McGee D, Davidsen M, et al. A cross -validation of risk -scores for coronary heart disease mortality based on data from the Glostrup Population Studies and Framingham Heart Study. Int J Epidemiol. 2002;31:817-22.
9. Iacobellis G, Willens HJ. Echocardiographic epicardial fat: a review of research and clinical applications. J Am Soc Echocardiogr. 2009; 22:1311-9.
10. Lubanski MS, Vanhecke TE, Chinnaiyan KM. Subclinical coronary atherosclerosis identified by coronary computed tomo-graphic angiography in asymptomatic morbidly obese patients. Heart Int. 2010; 5(2):15-9.
11. Laing ST, Smulevitz B, Vatcheva KP, et al. High Prevalence of Subclinical Atherosclerosis by Carotid Ultrasound among Mexican Americans: Discordance with 10-Year Risk Assessment using the Framingham Risk Score. Echocardiography. 2012;11: 56-8.
12. Hunt ME, O'Malley PG, Feuerstein I, et al. The relationship between the 'metabolic score' and sub-clinical atherosclerosis detected with electron beam computed tomography. Coron Artery Dis. 2003;14(4): 317-22.
13. Belyaeva OD, Mandal V, Ananeva NI. Carotid artery intima-media wall thickness as early marker atherosclerosis at patient with abdominal obesity. Arterial hypertension. 2008;14(1):71-6. Russian
14. Hayaishi-Okano R, Yamasaki Y, Katakami N, et al. Elevated C-reactive protein associates with early-stage carotid atherosclerosis in young subjects with type 1. Diabetes Care. 2002; 25(8);1432-8.
15. Panugova EV, Alexandrova EN, Nasonov EL. Atherosclerotic defeat of vessels at patients with the stable coronary heart disease: communication with C-reactiv protein. Kardiologiia. 2009;49(4): 40-5. Russian
16. Alekperov EZ, Nadzhafov RN. Modern concepts about an inflammation role at atherosclerosis. Kardiologiia. 2010;50(6):88-91. Russian
17. Saremi A, Anderson RJ, Luo P, et al. Association between IL-6 and the extent of coronary atherosclerosis in the veterans affairs diabetes trial (VADT). Atherosclerosis. 2009 ;203(2) :610-14.
18. Reilly MP, Iqbal N, Schutta M et al. Plasma leptin levels are associated with coronary atherosclerosis in type 2 diabetes. J Clinical Endocrinol & Metabol. 2004;89(8):3872-8.
19. Momiyama Y, Ohmori R, Fayad ZA, et al Associations between plasma C-reactive protein levels and the severities of coronary and aortic atherosclerosis. J Atheroscler Thromb. 2010;17(5):460-7.
20. Ouchi N, Kihara S, Funahashi T, et al. Reciprocal association of C-reactive protein with adiponectin in blood stream and adipose tissue. Circulation.2003; 107(5):671-4.
21. Otsuka F, Sugiyama S, Kojima S, et al. Plasma adiponectin levels are associated with coronary lesion complexity in men with coronary artery disease. JACC. 2006;48:1155-62.
22. Gnacinska M, Matgorzewicz S, Lysiak-Szydtowska W, et al. The serum profile of adipokines in overweight patients with metabolic syndrome. Endokrynol Pol. 2010;61: 36-41.
23. Maahs DM, Ogden LG, Kinney GL, et al. Low plasma adiponectin levels predict progression of coronary artery calcification. Circulation. 2005;11: 747-53.
