This cross-sectional study included asymptomatic participants in an ongoing cardiovascular health screening program from June 2009 through December 2012 at a tertiary-care teaching institute in Northern Taiwan. The primary aim of this program was to examine the hypothesis that certain demographic characteristics, behavioral factors, or biochemical data are associated with subclinical cardiac dysfunction in otherwise healthy individuals. All participants underwent a thorough evaluation, including general physical examination, baseline anthropometric measurements, blood sampling, and comprehensive echocardiography on the day of appointment. As described in our previous work (18), clinical symptoms, baseline comorbidities, smoking status, and exercise habits were obtained from a detailed structured questionnaire. This study was approved by the institutional review board of MacKay Memorial Hospital (14MMHIS202), and conducted in accordance with the Declaration of Helsinki. Informed consent was obtained from all participants.

Baseline comorbidities collected included diabetes, hypertension, dyslipidemia, and CVD. CVD constituted a group of diseases including coronary artery disease (CAD), stroke, and peripheral arterial disease. Laboratory parameters measured included hemoglobin, fasting blood sugar, lipid profile, renal function, N-terminal pro-brain natriuretic peptide (NT-proBNP), and urinalysis. All biochemical tests were conducted using a Hitachi 7170 Automatic Analyzer (Hitachi Corp., Hitachinaka, Ibaraki, Japan), and NT-proBNP was measured using an electrochemiluminescence immunoassay “ECLIA” assay (Roche Diagnostics GmbH, D-68298, Mannheim, Germany) in a standardized central laboratory. Renal function in terms of estimated glomerular filtration rate (eGFR) was calculated using CKD-EPI and four-variable MDRD equations, and was categorized as 30 to < 60, 60 to < 90, and ≧ 90 ml/min/1.73 m². For simplicity, eGFR is referred to as CKD-EPI eGFR if not otherwise specified. We defined proteinuria, measured with a dipstick, as negative, mild (trace to 1+), or severe (2+ to 3+). Test strips were measured using an automatic dipstick analyzer (CLINITEK Novus^®, Siemens). Validation of results with quantitative urine albumin amount was good ( Supplemental Figure 1 ). As per the Kidney Disease: Improving Global Outcomes (KDIGO) guidelines, participants were further classified based on eGFR and proteinuria categories (19). Subjects with missing data for serum creatinine or dipstick proteinuria were excluded from analysis.

Conventional echocardiography and TDI were performed on all participants, based on the American Society of Echocardiography and European Association of Cardiovascular Imaging guidelines (20, 21) using a GE system (Vivid i, GE Vingmed Ultrasound, Norway) equipped with a 2- to 4-MHz transducer (3S-RS). LV and left atrial (LA) structural parameters measured included LV end-diastolic and end-systolic diameters, wall thickness, LA/LV volume by modified biplane Simpson’s method, and LV mass by the Devereux formula (22). Maximum LA volume (LAVmax) was measured at ventricular end-systole just before opening of the mitral valve, while minimum LA volume (LAVmin) was measured at end-diastole, just before closure of the mitral valve. LV ejection fraction (LVEF) was calculated as 100 × (maximal LV volume − minimal LV volume)/maximal LV volume. LVEF was considered abnormal if < 50%. LV mass was further indexed to body surface area (BSA) as LV mass index (LVMI), and LAV was similarly indexed to BSA. LVH was defined as an LVMI greater than 115 g/m² in men and 95 g/m² in women (23).

The most important modalities to evaluate diastolic function are transmitral pulsed-wave Doppler flow and tissue Doppler mitral annular velocity profile (16, 17). The former helps to assess the presence and severity of DD, which alters the relationship between peak velocity flow in early diastole (E-wave) and that in late filling (A-wave), the time taken from the maximum E to baseline (deceleration time [DT]), and the interval between closure of the aortic valve and opening of the mitral valve (isovolumetric relaxation time [IVRT]). TDI measures the velocity of mitral annular motion, characterized by peak systolic velocity (s’), early diastolic velocity (e’), and late diastolic velocity (a’) in apical four-chamber view. Average e’ was taken as the average of septal e’ and lateral e’. LV filling pressure was estimated using the E/e’ ratio (average e’). DD was defined as E/e’> 15 or average e’ <9 cm/s when E/e’ is between 8 and 15 (24). Composite diastolic score was calculated based on TDI e’ velocity, E/e’ ratio, LAV index, and pulmonary artery pressure (16). Scores ranged from 0 to 2, where 0 was normal, 2 abnormal, and 1 in-between.

All echocardiographic images were performed blind to clinical information by an experienced technician, and stored digitally and reviewed offline using proprietary software (EchoPAC version 10.8, GE Vingmed Ultrasound, Norway). The reproducibility analysis has been reported in our previous article (18). We randomly selected 50 subjects for coefficient of variation analysis of a number of measured parameters ( Supplemental Table 1 ). For instance, the intra-class correlation coefficients for LAVmax were 92% between analyzers (interobserver) and 98.5% for the same analyzer (intraobserver).

This study analyzed the relationship between degree of renal dysfunction and cardiac deformational functional changes. The cohort was divided into eGFR and proteinuria categories. Trend tests were performed for continuous variables across categories of eGFR and proteinuria using one-way analysis of variance (ANOVA) and for categorical ones using the Cochran–Armitage test. Continuous variables are presented as mean ± standard deviation (SD); discrete variables are described as counts and percentages.

Multivariate linear regression was performed for markers of DD and renal function. Model 1 was adjusted for baseline clinical features (age and gender). Model 2 was additionally adjusted for baseline comorbidities (hypertension, diabetes, and CVD), body mass index (BMI), systolic blood pressure (SBP), current smoking, and laboratory data (fasting glucose, high-density lipoprotein [HDL], and total cholesterol). Model 3 added proteinuria to model 2. As for sensitivity tests, key echocardiographic variables (LVMI, LVEF, and stroke volume [SV]) were separately added to models 2 and 3. The final results of multivariate analyses were summarized by β-coefficient and 95% confidence intervals (CI).

Because NT-proBNP is a powerful indicator of HF (25), we also tested whether associations between renal function and diastolic parameters vary with NT-proBNP as an a priori hypothesis; therefore, possible interactions was evaluated with or without interaction terms between renal function (i.e., eGFR and proteinuria categories) and diastolic parameters (i.e., average e’, composite diastolic score, and LAV index) with NT-proBNP in factorial (two-way ANOVA in SPSS) and linear (ggplot2 package in R) designs.

All statistical analyses were carried out using Microsoft Excel 2013, IBM SPSS Statistics for Windows, Version 22.0 (IBM Corp. Released 2013. Armonk, NY), and R (R Core Team (2022). A two-sided p-value <0.05 was considered significant.

No funding was used for this study.

Our study included 5526 asymptomatic participants, and 584 were excluded for lack of serum creatinine or urine dipstick test ( Supplemental Tables 2 ; 3 ). Among 4942 enrollees, 65.8% were men, mean age was 49.4 ± 11.2 years, and mean CKD-EPI eGFR was 90.6 ± 15.7 ml/min/1.73 m² at enrollment ( Table 1 ). Hypertension was the most prevalent systemic disease in this cohort, reported in 18.7% of the enrollees. All participants were categorized into three groups based on eGFR and into three groups based on proteinuria on a dipstick ( Table 1 ). Great heterogeneity was observed between groups in terms of patient characteristics, baseline comorbidities, and laboratory data. As eGFR declined or proteinuria increased, there were trends of greater age, larger BMI, higher blood pressure, higher fasting glucose, higher uric acid, higher triglyceride, and higher NT-proBNP levels (all p for trends < 0.05).

On echocardiographic assessment, the systolic function of our participants was preserved (overall LVEF was 62.7 ± 5.4%) ( Table 2 ). Overall E/A ratio was 1.2 ± 0.4, E/average e’ 7.9 ± 2.6, septal e’ 8.0 ± 2.2 cm/s, lateral e’ 10.4 ± 2.9 cm/s, average e’ 9.2 ± 2.4 cm/s, LVMI 76.9 ± 14.8 g/m², and NT-proBNP 46.9 ± 109.9 pg/ml. LVMI in our cohort did not meet the criteria for LVH. LV geometry differed significantly by renal function status with higher LVMI, LAV indices, and LV end-diastolic and end-systolic diameters among individuals with lower CKD-EPI eGFR (or higher proteinuria) when compared with their counterparts. In parallel with the severity of renal dysfunction, E/A ratio and e’ gradually decreased, while peak A-wave velocity, DT, IVRT, E/e’, and composite diastolic score all gradually increased (all p for trends < 0.05). Similar trends of altered cardiac structures and functions were observed across MDRD eGFR categories ( Supplemental Table 4 ). Of note, participants in the worst categories (i.e., having eGFR between 30 and < 60 ml/min/1.73 m² or severe dipstick proteinuria) showed the lowest TDI-determined e’ values (septal e’ < 7 cm/s, lateral e’ < 10 cm/s, and average e’ < 9 cm/s), suggestive of highly abnormal diastolic relaxation ( Table 2 ; Supplemental Table 4 ) (16).

Figure 1 illustrates the levels of average e’, E/e’, LVMI, and NT-proBNP across categories of eGFR and proteinuria. We demonstrated a graded pattern of average e’, E/e’, and LVMI with the severity of renal function. In Table 3 , average e’ is summarized by CKD-EPI/MDRD eGFR and proteinuria category. The levels of average e’ did not meet the risk classification for prognosis of CKD and cardiovascular mortality as per the KDIGO guidelines (19).

In several multivariate regression models adjusted for clinical risk factors, CKD-EPI eGFR was positively correlated with average e’, and negatively correlated with NT-proBNP ( Table 4 ) and maximum LAVi ( Supplemental Table 5 ). CKD-EPI eGFR had no significant effect on LV filling E/e’ and LVMI. The adjusted models remained significant with respect to markers of DD when CKD-EPI eGFR was replaced by MDRD eGFR ( Table 4 ; Supplemental Table 5 ).

As shown in Figure 2 , a significant interaction exists between renal function and diastolic markers with reference to NT-proBNP, (interaction p < 0.05). Individuals with lower average e’ or higher composite diastolic score, rather than E/e’, present with higher NT-proBNP levels across worsening eGFR category (or having severe proteinuria) ( Figure 2 ; Supplemental Figure 2 ).