The 2013 ACC/AHA risk score and subclinical cardiac remodeling and dysfunction: Complementary in cardiovascular disease prediction

Effect of empagliflozin in peripheral diabetic neuropathy of patients with type 2 diabetes mellitus

BACKGROUND

Diabetic peripheral neuropathy (DPN) is the most dominant cause of neuropathy worldwide, and there has been no specific treatment until now. The aim of the current study was to assess the probable protective effect of empagliflozin in type 2 diabetics who are suffering from DPN.

METHODS

Fifty eligible type 2 diabetes mellitus (T2DM) cases with diabetic peripheral neuropathy were recruited in this study and classified into 2 groups. Group I (n=25) (control group) received placebo tablets once daily. Group II (n=25) (empagliflozin group) received empagliflozin 25mg once daily for three months. Empagliflozin efficacy was evaluated using electrophysiological studies, and HbA1c levels, the brief pain inventory short-form item (BPI-SF) score, the diabetic neuropathy symptom (DNS) score, the atherosclerotic cardiovascular disease (ASCVD) risk score, and the serum levels of neuron-specific enolase (NSE), malondialdehyde (MDA) and calprotectin (Calpro), lipid profile, and random blood glucose level (RBG).

RESULTS

After three months, comparing the results of the empagliflozin arm to the control arm showed a significant improvement in the electrophysiological studies and a significant decrease in the BPI-SF score and the mean serum levels of NSE and MDA. However, no significant difference was determined in HbA1c, Calpro, lipid profile, and RBG levels. In addition, the DNS and ASCVD risk scores were not significantly different. The NSE and MDA levels were significantly negatively correlated with the electrophysiological parameters. However, the BPI-SF score showed a non-significant difference.

CONCLUSIONS

Empagliflozin may be a promising neuroprotective and therapeutic agent for diabetic peripheral neuropathy. Trial registration Identifier: NCT05977465.

Association between oxidative balance score and 10-year atherosclerotic cardiovascular disease risk: results from the NHANES database

Introduction

The oxidative balance score (OBS) is a holistic measure that represents the overall equilibrium between prooxidants and antioxidants in one's diet and lifestyle. Little research has been conducted on the correlation between OBS and 10-year atherosclerotic cardiovascular disease risk (ASCVD). Therefore, the objective of this investigation was to examine the potential correlation between OBS and 10-year risk.

Methods

A total of 11,936 participants from the NHANES conducted between 2001 and 2016 were chosen for the study and their dietary and lifestyle factors were used to assess the OBS score. Logistic regression and restricted cubic splines (RCS) were employed in the cross-sectional study to evaluate the correlation between OBS and the 10-year ASCVD risk. The cohort study utilized Cox proportional hazards models and RCS to assess the correlation between OBS and all-causes and cardiovascular disease (CVD) mortality in individuals with high ASCVD risk.

Results

The cross-sectional study found that the OBS (OR = 0.94, 95% CI = 0.93-0.98), as well as the dietary OBS (OR = 0.96, 95% CI = 0.92-0.96) and lifestyle OBS (OR = 0.74, 95% CI = 0.69-0.79), were inversely associated with the 10-year ASCVD risk. A significant linear relationship was observed between OBS, dietary OBS, lifestyle OBS, and the 10-year ASCVD risk. The cohort study found that the OBS was inversely associated with all-cause (aHRs = 0.97, 95% CI = 0.96-0.99) and CVD (aHRs = 0.95, 95% CI = 0.93-0.98) mortality in individuals with high ASCVD risk. A significant linear correlation was observed between OBS, dietary OBS, lifestyle OBS, and all-cause and CVD mortality in participants with high ASCVD risk.

Conclusion

The findings indicate that OBS, OBS related to diet, and OBS related to lifestyle were significantly inversely correlated with the 10-year ASCVD risk. Adopting a healthy eating plan and making positive lifestyle choices that result in increased OBS levels can help lower the likelihood of all-cause and CVD mortality in individuals with high ASCVD risk.

Serum proteomic profiling of carotid arteriopathy: A population outcome study

BACKGROUND AND AIMS

Circulating proteins reflecting subclinical vascular disease may improve prediction of atherosclerotic cardiovascular disease (ASCVD). We applied feature selection and unsupervised clustering on proteomic data to identify proteins associated with carotid arteriopathy and construct a protein-based classifier for ASCVD event prediction.

METHODS

491 community-dwelling participants (mean age, 58 ± 11 years; 51 % women) underwent carotid ultrasonography and proteomic profiling (CVD II panel, Olink Proteomics). ASCVD outcome was collected (median follow-up time: 10.2 years). We applied partial least squares (PLS) to identify proteins linked to carotid intima-media thickness (cIMT). Next, we assessed the association between future ASCVD events and protein-based phenogroups derived by unsupervised clustering (Gaussian Mixture modelling) based on proteins selected in PLS.

RESULTS

PLS identified 19 proteins as important, which were all associated with cIMT in multivariable-adjusted linear regression. 8 of the 19 proteins were excluded from the clustering analysis because of high collinearity. Based on the 11 remaining proteins, the clustering algorithm subdivided the cohort into two phenogroups. Compared to the first phenogroup (n = 177), participants in the second phenogroup (n = 314) presented: i) a more unfavorable lipid profile with higher total cholesterol and triglycerides and lower HDL cholesterol (p ≤ 0.014); ii) higher cIMT (p = 0.0020); and iii) a significantly higher risk for future ASCVD events (multivariable-adjusted hazard ratio (95 % CI) versus phenogroup 1: 2.05 (1.26-3.52); p = 0.0093). The protein-based phenogrouping supplemented ACC/AHA 10-year ASCVD risk scoring for prediction of a first ASCVD event.

CONCLUSIONS

Focused protein-based phenogrouping identified individuals at high risk for future ASCVD and may complement current risk stratification strategies.

Clinical and Echocardiographic Diversity Associated With Physical Fitness in the Project Baseline Health Study: Implications for Heart Failure Staging

BACKGROUND

Clinical and echocardiographic features may carry diverse information about the development of heart failure (HF). Therefore, we determined heterogeneity in clinical and echocardiographic phenotypes and its association with exercise capacity.

METHODS

In 2036 community-dwelling individuals, we defined echocardiographic profiles of left and right heart remodeling and dysfunction. We subdivided the cohort based on presence (+) or absence (-) of HF risk factors (RFs) and echocardiographic abnormalities (RF-/Echo-, RF-/Echo+, RF+/Echo-, RF+/Echo+). Multivariable-adjusted associations between subgroups and physical performance metrics from 6-minute walk and treadmill exercise testing were assessed.

RESULTS

The prevalence was 35.3% for RF-/Echo-, 4.7% for RF-/Echo+, 39.3% for RF+/Echo-, and 20.6% for RF+/Echo+. We observed large diversity in echocardiographic profiles in the Echo+ group. Participants with RF-/Echo+ (18.6% of Echo+) had predominantly echocardiographic abnormalities other than left ventricular (LV) diastolic dysfunction, hypertrophy and reduced ejection fraction, whereas their physical performance was similar to RF-/Echo-. In contrast, participants with RF+/Echo+ presented primarily with LV hypertrophy or dysfunction, features that related to lower 6-minute walking distance and lower exercise capacity.

CONCLUSIONS

Subclinical echocardiographic abnormalities suggest HF pathogenesis, but the presence of HF risk factors and type of echo abnormality should be considered so as to distinguish adverse from benign adaptation and to stratify HF risk.

The Versatility of Atherosclerotic Cardiovascular Disease Risk Score in Determination of Popliteal Artery Branches Patency in Computed Tomography Angiography

The atherosclerotic cardiovascular disease (ASCVD) risk score is used to estimate coronary artery disease and stroke risk. Atherosclerosis affects arteries throughout the body, including the legs, causing peripheral arterial disease. Atherosclerosis causes luminal stenosis in popliteal artery branches, which affects operative decisions such as intravascular surgery, and lower limb reconstruction. The objective was to investigate the relationship between the ASCVD risk score and degree of stenosis among the popliteal artery and its branches.

Methods

The data regarding all patients who underwent computed tomography angiography (CTA) of the legs during 2016-2021 with complete data for ASCVD risk score assessment were recruited. The association between luminal stenosis from CTA and calculated ASCVD risk score was analyzed.

Results

A total of 383 limbs of 117 men and 81 women, averaged 66.5 years old, were studied. Common comorbidities included hypertension (84.3%), diabetes mellitus (61.1%), and chronic kidney disease (34.3%). Average 10-year ASCVD risks in the greater than or equal to 50% stenosis group of popliteal, anterior tibial, and posterior tibial arteries were significantly higher than the less than 50% stenosis group (P < 0.01). The peroneal artery had no significant difference between stenosis groups. The popliteal artery had significantly higher lifetime ASCVD risks than in the greater than or equal to 50% stenosis group (P < 0.01), but the other arteries showed no statistically significant difference.

Conclusions

The 10-year ASCVD risks showed significant higher values in the greater than or equal to 50% stenosis group of popliteal, anterior tibial, and posterior tibial arteries. These findings can establish the further study on how ASCVD risks can be applied to predict the stenosis of these arteries and guide the rationale of preoperative leg CTA for FFF harvest.

Metabolomics profile and 10-year atherosclerotic cardiovascular disease (ASCVD) risk score

Background

The intermediate metabolites associated with the development of atherosclerotic cardiovascular disease (ASCVD) remain largely unknown. Thus, we conducted a large panel of metabolomics profiling to identify the new candidate metabolites that were associated with 10-year ASCVD risk.

Methods

Thirty acylcarnitines and twenty amino acids were measured in the fasting plasma of 1,102 randomly selected individuals using a targeted FIA-MS/MS approach. The 10-year ASCVD risk score was calculated based on 2013 ACC/AHA guidelines. Accordingly, the subjects were stratified into four groups: low-risk (n = 620), borderline-risk (n = 110), intermediate-risk (n = 225), and high-risk (n = 147). 10 factors comprising collinear metabolites were extracted from principal component analysis.

Results

C₄DC, C_8:1, C₁₆OH, citrulline, histidine, alanine, threonine, glycine, glutamine, tryptophan, phenylalanine, glutamic acid, arginine, and aspartic acid were significantly associated with the 10-year ASCVD risk score (p-values ≤ 0.044). The high-risk group had higher odds of factor 1 (12 long-chain acylcarnitines, OR = 1.103), factor 2 (5 medium-chain acylcarnitines, OR = 1.063), factor 3 (methionine, leucine, valine, tryptophan, tyrosine, phenylalanine, OR = 1.074), factor 5 (6 short-chain acylcarnitines, OR = 1.205), factor 6 (5 short-chain acylcarnitines, OR = 1.229), factor 7 (alanine, proline, OR = 1.343), factor 8 (C_18:2OH, glutamic acid, aspartic acid, OR = 1.188), and factor 10 (ornithine, citrulline, OR = 1.570) compared to the low-risk ones; the odds of factor 9 (glycine, serine, threonine, OR = 0.741), however, were lower in the high-risk group. "D-glutamine and D-glutamate metabolism", "phenylalanine, tyrosine, and tryptophan biosynthesis", and "valine, leucine, and isoleucine biosynthesis" were metabolic pathways having the highest association with borderline/intermediate/high ASCVD events, respectively.

Conclusions

Abundant metabolites were found to be associated with ASCVD events in this study. Utilization of this metabolic panel could be a promising strategy for early detection and prevention of ASCVD events.

Relationship of Noninvasive Assessment of Arterial Stiffness with 10-Year Atherosclerotic Cardiovascular Disease (ASCVD) Risk in a General Middle-Age and Elderly Population

Purpose

As a powerful indicator of arterial stiffening, the brachial-ankle pulse wave velocity (baPWV) has been extensively validated for predicting cardiovascular events. However, whether and how the brachial-ankle pulse wave velocity (baPWV) is correlated with the 10-year atherosclerotic cardiovascular disease (ASCVD) risk is unclear. This study aimed to investigate the association between baPWV and 10-year ASCVD risk in Chinese population.

Methods

A total of 1768 subjects were enrolled from Shanghai, China. They were divided into two groups according to the Pooled Cohorts Equations model made by ACC/AHA as follows: low ASCVD risk (n = 992, 10-year ASCVD risk <7.5%) and high ASCVD risk (n = 776, 10-year ASCVD risk ≥7.5%). The baseline characteristics were obtained via the use of a questionnaire. Measurement of baPWV, laboratory tests, and echocardiography were conducted by trained physicians. The relationship between baPWV and 10-year ASCVD risk was evaluated using multiple logistic regression model and generalized additive model.

Results

The mean age of the subjects was 58.89±8.60 years, 32.69% of which were male. Non-linear relationship analysis revealed threshold effects between baPWV and 10-year ASCVD risk in which a baPWV of approximately 16 m/s might be the threshold effect of 10-year ASCVD risk. After multivariable adjustment, logistic-regression analysis demonstrated that ankle-brachial index (ABI) (OR 5.28, 95% CI 1.20–12.23) and baPWV (OR 9.09, 95% CI 6.84–12.07) were independently correlated with 10-year ASCVD risk. The AUC for baPWV for predicting 10-year ASCVD risk was 0.80 (95% CI 0.78–0.82).

Conclusion

Increased baPWV as an indicator of arterial stiffness correlates strongly with 10-year ASCVD risk in general middle-aged and elderly populations. The association between baPWV and 10-year ASCVD risk is not purely linear but non-linear. Subjects with baPWV above 16 m/s are more likely to encounter a higher 10-year ASCVD risk.

Features and implications of higher systolic central than peripheral blood pressure in patients at very high risk of atherosclerotic cardiovascular disease

Peripheral blood pressure (PBP) is usually higher than central blood pressure (CBP) due to pulse amplification; however, it is not well understood why cuff-measured PBP can be lower than CBP estimated by the late systolic pressure of radial pulse waves. We explored the implications of systolic PBP-CBP (P-CBP) differences for cardiovascular (CV) prognosis. In total, 335 patients at very high risk of atherosclerotic cardiovascular disease (ASCVD) underwent automated applanation tonometry and brachial-ankle pulse wave velocity (baPWV), and they were classified into groups according to positive or negative systolic P-CBP differences. Between-group characteristics and clinical outcomes (the composite of coronary revascularization, stroke, heart failure hospitalization, and CV death) were evaluated. Patients with negative differences had significantly higher frequency of hypertension, coronary artery disease, higher ASCVD risk burden, and elevated N-terminal pro b-type natriuretic peptide. They had higher left atrial volume index (LAVI) and lower systolic mitral septal tissue velocity (TVI-s') than those with a positive difference. These patients showed higher systolic PBP and CBP, and a higher baPWV. Multivariable analysis indicated that TVI-s', LAVI, and ASCVD risk burden were independent determinants of such systolic P-CBP differences. During a median follow-up of 12.6 months, clinical outcomes were significantly related to a negative difference (11.5% vs. 3.4%, p = 0.014), and a systolic P-CBP difference ≤ -8 mmHg was associated with a threefold higher likelihood of poor prognosis. In patients at very high risk of ASCVD, systolic P-CBP difference was associated with cardiac dysfunction and ASCVD risk burden, allowing further risk stratification.

Echocardiographic phenogrouping by machine learning for risk stratification in the general population

Aims

There is a need for better phenotypic characterization of the asymptomatic stages of cardiac maladaptation. We tested the hypothesis that an unsupervised clustering analysis utilizing echocardiographic indexes reflecting left heart structure and function could identify phenotypically distinct groups of asymptomatic individuals in the general population.

Methods and results

We prospectively studied 1407 community-dwelling individuals (mean age, 51.2 years; 51.1% women), in whom we performed clinical and echocardiographic examination at baseline and collected cardiac events on average 8.8 years later. Cardiac phenotypes that were correlated at r > 0.8 were filtered, leaving 21 echocardiographic features, and systolic blood pressure for phenogrouping. We employed hierarchical and Gaussian mixture model-based clustering. Cox regression was used to demonstrate the clinical validity of constructed phenogroups. Unsupervised clustering analyses classified study participants into three distinct phenogroups that differed markedly in echocardiographic indexes. Indeed, cluster 3 had the worst left ventricular (LV) diastolic function (i.e. lowest e' velocity and left atrial (LA) reservoir strain, highest E/e', and LA volume index) and LV remodelling. The phenogroups were also different in cardiovascular risk factor profiles. We observed increase in the risk for incidence of adverse events across phenogroups. In the third phenogroup, the multivariable adjusted risk was significantly higher than the average population risk for major cardiovascular events (51%, P = 0.0028).

Conclusion

Unsupervised learning algorithms integrating routinely measured cardiac imaging and haemodynamic data can provide a clinically meaningful classification of cardiac health in asymptomatic individuals. This approach might facilitate early detection of cardiac maladaptation and improve risk stratification.

Incremental value of diastolic stress test in identifying subclinical heart failure in patients with diabetes mellitus

AIMS

Resting echocardiography is a valuable method for detecting subclinical heart failure (HF) in patients with diabetes mellitus (DM). However, few studies have assessed the incremental value of diastolic stress for detecting subclinical HF in this population.

METHODS AND RESULTS

Asymptomatic patients with Type 2 DM were prospectively enrolled. Subclinical HF was assessed using systolic dysfunction (left ventricular longitudinal strain <16% at rest and <19% after exercise in absolute value), abnormal cardiac morphology, or diastolic dysfunction (E/e' > 10). Metabolic equivalents (METs) were calculated using treadmill speed and grade, and functional capacity was assessed by percent-predicted METs (ppMETs). Among 161 patients studied (mean age of 59 ± 11 years and 57% male sex), subclinical HF was observed in 68% at rest and in 79% with exercise. Among characteristics, diastolic stress had the highest yield in improving detection of HF with 57% of abnormal cases after exercise and 45% at rest. Patients with revealed diastolic dysfunction during stress had significantly lower exercise capacity than patients with normal diastolic stress (7.3 ± 2.1 vs. 8.8 ± 2.5, P < 0.001 for peak METs and 91 ± 30% vs. 105 ± 30%, P = 0.04 for ppMETs). On multivariable modelling found that age (beta = -0.33), male sex (beta = 0.21), body mass index (beta = -0.49), and exercise E/e' >10 (beta = -0.17) were independently associated with peak METs (combined R2 = 0.46). A network correlation map revealed the connectivity of peak METs and diastolic properties as central features in patients with DM.

CONCLUSION

Diastolic stress test improves the detection of subclinical HF in patients with diabetes mellitus.

Proteomic profiling for detection of early-stage heart failure in the community

Aims

Biomarkers may provide insights into molecular mechanisms underlying heart remodelling and dysfunction. Using a targeted proteomic approach, we aimed to identify circulating biomarkers associated with early stages of heart failure.

Methods and results

A total of 575 community‐based participants (mean age, 57 years; 51.7% women) underwent echocardiography and proteomic profiling (CVD II panel, Olink Proteomics). We applied partial least squares‐discriminant analysis (PLS‐DA) and a machine learning algorithm [eXtreme Gradient Boosting (XGBoost)] to identify key proteins associated with echocardiographic abnormalities. We used Gaussian mixture modelling for unbiased clustering to construct phenogroups based on influential proteins in PLS‐DA and XGBoost. Of 87 proteins, 13 were important in PLS‐DA and XGBoost modelling for detection of left ventricular remodelling, left ventricular diastolic dysfunction, and/or left atrial reservoir dysfunction: placental growth factor, kidney injury molecule‐1, prostasin, angiotensin‐converting enzyme‐2, galectin‐9, cathepsin L1, matrix metalloproteinase‐7, tumour necrosis factor receptor superfamily members 10A, 10B, and 11A, interleukins 6 and 16, and α1‐microglobulin/bikunin precursor. Based on these proteins, the clustering algorithm divided the cohort into two distinct phenogroups, with each cluster grouping individuals with a similar protein profile. Participants belonging to the second cluster (n = 118) were characterized by an unfavourable cardiovascular risk profile and adverse cardiac structure and function. The adjusted risk of presenting echocardiographic abnormalities was higher in this phenogroup than in the other (P < 0.0001).

Conclusions

We identified proteins related to renal function, extracellular matrix remodelling, angiogenesis, and inflammation to be associated with echocardiographic signs of early‐stage heart failure. Proteomic phenomapping discriminated individuals at high risk for cardiac remodelling and dysfunction.

Subclinical Heart Dysfunction in Relation to Metabolic and Inflammatory Markers: A Community-Based Study

BACKGROUND

Population studies investigating the contribution of immunometabolic disturbances to heart dysfunction remain scarce. We combined high-throughput biomarker profiling, multidimensional network analyses, and regression statistics to identify immunometabolic markers associated with subclinical heart dysfunction in the community.

METHODS

In 1,236 individuals (mean age, 51.0 years; 51.5% women), we measured 39 immunometabolic markers and assessed echocardiographic indexes of left ventricular diastolic dysfunction (LVDD) and left atrial (LA) reservoir dysfunction. We used partial least squares (PLS) to filter the most relevant biomarkers related to the echocardiographic characteristics. Subsequently, we assessed the associations between the echocardiographic features and biomarkers selected in PLS while accounting for clinical confounders.

RESULTS

Influential biomarkers in PLS of echocardiographic characteristics included blood sugar, γ-glutamyl transferase, d-dimer, ferritin, hemoglobin, interleukin (IL)-4, IL-6, and serum insulin and uric acid. In stepwise regression incorporating clinical confounders, higher d-dimer was independently associated with higher E/e' ratio and LA volume index (P ≤ 0.05 for all). In multivariable-adjusted analyses, the risk for LVDD increased with higher blood sugar and d-dimer (P ≤ 0.048). After full adjustment, higher serum insulin and serum uric acid were independently related to worse LA reservoir strain and higher risk for LA reservoir dysfunction (P ≤ 0.039 for all). The biomarker panels detected LVDD and LA reservoir dysfunction with 87% and 79% accuracy, respectively (P < 0.0001).

CONCLUSIONS

Biomarkers of insulin resistance, hyperuricemia, and chronic low-grade inflammation were associated with cardiac dysfunction. These biomarkers might help to unravel cardiac pathology and improve the detection and management of cardiac dysfunction in clinical practice.

Association of Subclinical Heart Maladaptation With the Pooled Cohort Equations to Prevent Heart Failure Risk Score for Incident Heart Failure

Importance

The Pooled Cohort Equations to Prevent Heart Failure (PCP-HF) estimate the 10-year risk for symptomatic heart failure (HF) from routine clinical data. The PCP-HF score should detect asymptomatic individuals with cardiac maladaptation preceding HF symptoms for it to be a useful HF prediction tool in primary prevention.

Objective

To assess the concordance between PCP-HF risk scoring and the presence of subclinical cardiac maladaptation in the community.

Design, Setting, and Participants

This cross-sectional analysis included participants enrolled in the Flemish Study on Environment, Genes and Health Outcomes who underwent a clinical examination including echocardiography between May 2005 and January 2015. Participants younger than 30 years, older than 79 years, had prevalent cardiovascular disease, and/or had suboptimal echocardiographic imaging quality were excluded. Analysis began February 2020 and ended April 2020.

Exposures

Ten-year HF risk as calculated from routine clinical data using the PCP-HF. Based on tertile limits, participants were categorized as having low (≤0.4%), intermediate (0.4%-2.4%), and high (≥2.4%) 10-year HF risk score.

Main Outcomes and Measures

Echocardiographic profiles of subclinical heart remodeling and dysfunction.

Results

A total of 1020 individuals were analyzed (mean [SD] age, 52.8 [11.4] years; 541 female [53.0%]). The prevalence of left ventricular (LV) remodeling and dysfunction was significantly higher from low to intermediate and high 10-year HF risk score. A doubling in 10-year HF risk score was associated with higher odds for LV concentric remodeling (odds ratio [OR], 1.48; 95% CI, 1.36-1.61; P < .001), LV hypertrophy (OR, 1.66; 95% CI, 1.51-1.83; P < .001), abnormal LV longitudinal strain (OR, 1.12; 95% CI, 1.05-1.19; P < .001), and LV diastolic dysfunction (OR, 2.28; 95% CI, 1.94-2.69; P < .001). Moreover, the PCP-HF score detected echocardiographic abnormalities with an accuracy of 74% (LV concentric remodeling), 78% (LV hypertrophy), 59% (abnormal LV longitudinal strain), and 87% (LV diastolic dysfunction). The likelihood of LV concentric remodeling, hypertrophy, and diastolic dysfunction were 3.1, 3.8, and 9.4 times higher in participants with high 10-year HF risk score than the average population risk, respectively (P < .001). Of all PCP-HF score components, age, body mass index, and systolic blood pressure were key correlates of echocardiographic abnormalities in multivariable-adjusted analysis.

Conclusions and Relevance

PCP-HF risk scoring adequately detected individuals with subclinical heart maladaptation that precedes HF symptoms by years. Thus, it may be a valuable HF prediction tool in primary prevention.

Applying machine learning to detect early stages of cardiac remodelling and dysfunction

AIMS

Both left ventricular (LV) diastolic dysfunction (LVDD) and hypertrophy (LVH) as assessed by echocardiography are independent prognostic markers of future cardiovascular events in the community. However, selective screening strategies to identify individuals at risk who would benefit most from cardiac phenotyping are lacking. We, therefore, assessed the utility of several machine learning (ML) classifiers built on routinely measured clinical, biochemical, and electrocardiographic features for detecting subclinical LV abnormalities.

METHODS AND RESULTS

We included 1407 participants (mean age, 51 years, 51% women) randomly recruited from the general population. We used echocardiographic parameters reflecting LV diastolic function and structure to define LV abnormalities (LVDD, n = 252; LVH, n = 272). Next, four supervised ML algorithms (XGBoost, AdaBoost, Random Forest (RF), Support Vector Machines, and Logistic regression) were used to build classifiers based on clinical data (67 features) to categorize LVDD and LVH. We applied a nested 10-fold cross-validation set-up. XGBoost and RF classifiers exhibited a high area under the receiver operating characteristic curve with values between 86.2% and 88.1% for predicting LVDD and between 77.7% and 78.5% for predicting LVH. Age, body mass index, different components of blood pressure, history of hypertension, antihypertensive treatment, and various electrocardiographic variables were the top selected features for predicting LVDD and LVH.

CONCLUSION

XGBoost and RF classifiers combining routinely measured clinical, laboratory, and electrocardiographic data predicted LVDD and LVH with high accuracy. These ML classifiers might be useful to pre-select individuals in whom further echocardiographic examination, monitoring, and preventive measures are warranted.