Associations between glycated haemoglobin and multi-modal imaging markers of early cardiac aging

Abstract

Background: Glycated haemoglobin (HbA1c) is a well-established biomarker for diabetes diagnosis and management and is linked to risk of cardiovascular death. However, among adults without cardiovascular disease (CVD) and diabetes, the value of HbA1c in predicting distinct signatures of myocardial ageing has not been explored. Methods: Subjects, from among older adults without CVD, underwent comprehensive cardiovascular and metabolic assessment. Transthoracic echocardiography measured left ventricular structure and function. Longitudinal left atrial (LA) strain comprising reservoir strain (Ɛs), conduit strain (Ɛe) and booster strain (Ɛa) and their corresponding peak strain rates (SRs, SRe, SRa) were measured using cardiac magnetic resonance (CMR). Blood sampling for biomarkers and cardiovascular examinations were performed. Results: 247 subjects (mean age 71 years, 44.1% female, mean HbA1c 6.0%) were included. HbA1c was significantly associated with E/A ratio (p < 0.0001), conduit strain (Ɛe) (p < 0.0001), conduit strain rate SRe (p < 0.0001), and conduit strain rate to booster strain rate SRe:SRa ratio (p < 0.0001). Multivariate models adjusting for clinical variables such as body mass index, blood pressure, heart rate, diabetes mellitus, smoking, and associated cardiac parameters, demonstrated a persistent independent association. Each unit increase in HbA1c was associated with lower E/A ratio, lower Ɛe, higher SRe and lower SRe:SRa ratio. These associations remained significant after diabetic subjects were excluded. Conclusion: Distinct associations were found between HbA1c and myocardial functions of interest in the ageing heart. HbA1c may be useful biomarker for stratifying risks associated with myocardial ageing, independent of diabetes status.

Trial registration: ClinicalTrials.gov Identifier: NCT02791139.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11357-024-01387-7.

Introduction

Haemoglobin A1c (HbA1c), measurement of glycated haemoglobin, indicates plasma glucose levels over three months. A simple and low-cost laboratory test widely available and used in medical practice for diagnosis and monitoring of diabetes mellitus. It is not subject to daily fluctuations in plasma glucose levels, making it a useful longitudinal assessment. HbA1c levels have been linked to risk of cardiovascular death in men with type 2 diabetes in a large prospective cohort [1]. However, recent studies have unveiled a compelling association between HbA1c levels and ageing beyond its conventional role in diabetes management. This emerging link has sparked interest in exploring the potential utility of HbA1c as a novel biomarker for ageing and ageing associated cardiovascular disease.

Cardiovascular diseases (CVDs) remain a leading cause of morbidity and mortality worldwide. Their prevalence continues to rise, particularly in the ageing population. With ageing, the heart undergoes complex structural and functional changes collectively known as cardiac ageing. The ageing heart is associated with increased risks of cardiovascular disease [2]. Some physiological changes associated with early cardiac ageing were previously reported [3, 4]. Of note, diastolic dysfunction as characterized by abnormalities in relaxation and filling of the heart during diastolic phase and upstream changes in the left atrium as represented by left atrial strain identifies early functional changes in the heart of asymptomatic older adults.

As the global demographic shifts towards an increasingly older population, there is a growing imperative to identify reliable biomarkers that can aid in early detection, monitoring, and management of age-related cardiovascular changes. In diabetes, HbA1c has been shown to predict risk of future cardiovascular events [1]. Importantly, HbA1C tends to increase with ageing and this relationship is independent an individual’s glycemic status [5]. We believe that this accessible and reliable measurement could be developed as a biomarker for cardiovascular ageing.

In this study we examined the relationship between HbA1c levels and cardiovascular function in a cohort of older adults with cardiovascular ageing. We hypothesize that HbA1c levels correlate with presence of cardiac dysfunction and can serve as a biomarker for cardiovascular ageing.

Methods

Study population

Participants were derived from Cardiac Ageing Study (CAS), a prospective cohort study designed to examine characteristics and determinants of cardiac ageing in older adults [6]. Subjects were local community adults who volunteered for the study. Current analyses consisted of men and women who were enrolled in the baseline 2014 examination without history of physician-diagnosed cardiovascular disease (such as coronary heart disease and stroke) or cancer. Recruitment of subjects was conducted by trained study coordinators and informed written consent was obtained. Study protocol was approved by the local Institutional Review Board (CIRC/2014/628/C).

Data acquisition

Enrolled subjects underwent extensive clinical and cardiovascular examinations at our study site which is a tertiary cardiovascular hospital. Anthropometric measurements comprising of height (cm), weight (kg) and waist circumference (cm) were obtained to compute body surface area (BSA) (mg/m2) and body mass index (BMI) (kg/m2). Each subject filled in a standardized questionnaire for declaration of pre-existing cardiovascular risk factors including medical diagnoses of hypertension, dyslipidemia and diabetes mellitus, and their smoking status (active, former or never smokers). Electrocardiogram was performed to confirm sinus rhythm status. Clinical data and blood sampling were obtained same day as cardiovascular examinations. Participants did not have atrial fibrillation or significant valvular heart disease on echocardiography.

Transthoracic echocardiography

Standard echocardiography, including two-dimensional, M-mode, pulse Doppler, tissue Doppler imaging, was performed in standard parasternal and apical views (apical four-chamber, apical two-chamber, and apical long). Three cardiac cycles were recorded for each subject. Left ventricular ejection fraction, left atrial volume, and left atrial volume index were measured. Doppler echocardiography recorded the trans-mitral flow E and A waves from the apical four-chamber view. E/A ratio was calculated as the ratio of peak velocity flow in early diastole (E) to peak velocity flow in late diastole by atrial contraction (A). E/A ratio has previously been shown to be prognostic of clinical outcomes in community older adults [7]. Pulsed wave tissue Doppler imaging was performed with sample volume at septal and lateral annulus from the apical four-chamber view. Frame rate ranged between 80 and 100 frames per second. Tissue velocity patterns (E' and A') were recorded. All measurements were taken by the same operator, averaged over three cardiac cycles, adjusted by the RR interval.

Cardiac magnetic resonance imaging

We developed an in-house semi-automatic algorithm to track distance (L) between left atrioventricular junction and a user-defined point located at the mid posterior left atrial (LA) wall on standard cardiac magnetic resonance (CMR) two- and four-chamber views [8]. Both views were utilized to calculate average longitudinal strain (Ɛ) and strain rate (SR) results. Longitudinal strain (ε) at any given time point (t) in the cardiac cycle, starting from end-diastole (time 0), was computed as Ɛ(t) = (L[t]—L0) / L0. LA reservoir strain (Ɛs), conduit strain (Ɛe) and booster strain (Ɛa) were determined at specific time points during the cardiac cycle: left ventricular end-systole, diastasis, and pre-LA systole, respectively. Reservoir strain (Ɛs) represents the LA's ability to passively expand and accommodate blood during ventricular systole, while conduit strain (Ɛe) reflects the LA's role in facilitating blood flow from pulmonary veins to left ventricle during early diastole. On the other hand, booster strain (Ɛa) indicates the LA's contractile function during late diastole, aiding in additional blood transfer into the left ventricle just before atrial systole. Peak values of the first-time derivative of strain–time curve at systole, diastasis, and LA contraction corresponded to respective peak strain rates (SRs). Strain and SR parameters from both two- and four-chamber views were averaged to derive mean results for further analysis.

Statistical methodology

Subjects were categorized into young and old using age cut-off of 71 years (sample mean). Continuous variables were presented as mean with standard deviation, while categorical variables were expressed as n(%). Bivariable associations between clinical characteristics, HbA1c, and cardiac indices were examined by linear regression model. For each cardiac indices, E/A ratio, Ɛe, SRe and SRe:SRa, univariate linear regression was applied to clinical variables, HbA1c and cardiac characteristics (Model 1). Further multiple linear regression was applied to HbA1c and those clinical variables with p < 0.05 in univariate linear regression to adjust for possible confounding effect (Model 2, 4, 5, 6). Furthermore, multiple linear regression was applied to include those cardiac characteristics with p < 0.05 (Model 3). We additionally conducted quadratic regression to assess whether the association between HbA1c and cardiac indices were predominantly linear or non-linear. Likelihood ratio test was used to compare the two regression models (linear and non-linear). All statistical analyses were conducted using STATA 18 software (College Station, TX, USA), and a two-tailed p-value of less than 0.05 was considered statistically significant.

Results

Baseline, cardiac characteristics and associations with HbA1c

A total of 247 participants [mean age 71 ± 9.2 years, n = 138 males (65.9%)] were studied in this sample. Participants’ mean body mass index (BMI) was 24 ± 3.2 kg/m². Prevalence of hypertension, dyslipidaemia, diabetes mellitus and smoking were 51.4%, 48.6%, 19% and 18.2% respectively. Mean HbA1c was 6.0 ± 0.8%. (Table 1).

Table 1.

Clinical and cardiac characteristics

Clinical	Mean ± SD or n (%)
Age, year	71 ± 9.2
Female	109 (44.1%)
Weight, kg	60 ± 9.9
Height, cm	160 ± 7.8
BSA	1.6 ± 0.2
Body mass index	24 ± 3.2
SBP	148 ± 25.5
DBP	75 ± 10.6
Pulse	74 ± 12.7
Waist circumference	82 ± 9.1
Hypertension	127 (51.4%)
Dyslipidemia	120 (48.6%)
Diabetes_mellitus	47 (19.0%)
Smoking	45 (18.2%)
HbA1c	6.0 ± 0.8
Echocardiographic variables	Mean ± SD
Interventricular septum thickness at end diastole (IVSD) (cm)	0.82 ± 0.2
Interventricular septum thickness at end systole (IVSS) (cm)	1.23 ± 0.2
Left ventricular internal diameter end diastole (LVIDD) (cm)	4.42 ± 0.6
Left ventricular internal diameter end systole (LVIDS) (cm)	2.50 ± 0.5
Left ventricular posterior wall end diastole (LVPWD) (cm)	0.77 ± 0.1
Left ventricular posterior wall end systole (LVPWS) (cm)	1.42 ± 0.2
Left ventricular outflow tract (LVOT) (cm)	2.05 ± 0.2
Left ventricular ejection fraction (LVEF) (%)	65 ± 7.7
Left ventricular fractional shortening (LVFS) (%)	44 ± 6.4
Left ventricular mass (grams)	127 ± 55
Left ventricular mass index (grams/m2)	78 ± 31
Left atrial volume (ml)	37 ± 13
Left atrial volume index (ml/m2)	23 ± 7.6
Peak velocity flow in early diastole E (MV E peak) (m/s)	0.67 ± 0.2
Peak velocity flow in late diastole by atrial contraction A (MV A peak) (m/s)	0.78 ± 0.2
Ratio of MV E peak velocity to MV A peak velocity	0.91 ± 0.34
Mitral valve flow deceleration time (MV DT) (ms)	207 ± 37
Pulmonary artery systolic pressure (PASP) (mmHg)	27 ± 6.4
Peak systolic septal mitral annular velocity (Septal S′) (m/s)	0.076 ± 0.01
Peak early diastolic septal mitral annular velocity (Septal E’) (m/s)	0.073 ± 0.02
Septal mitral annular velocity during atrial contraction (Septal A’) (m/s)	0.10 ± 0.02
Peak systolic lateral mitral annular velocity (m/s) (Lateral S)	0.092 ± 0.03
Peak early diastolic lateral mitral annular velocity (m/s) (Lateral E)	0.092 ± 0.03
Lateral mitral annular velocity during atrial contraction (m/s) (Lateral A)	0.11 ± 0.02
Ratio of Peak velocity flow in early diastole E (MV E peak) to Peak early diastolic septal mitral annular velocity (Septal E’)	9.67 ± 3.1
Ratio of Peak velocity flow in early diastole E (MV E peak) to Peak early diastolic lateral mitral annular velocity (Lateral E’)	7.77 ± 2.7
Left atrial function	Mean ± SD
Reservoir strain, ɛs (%)	31 ± 7.2
Conduit strain, ɛe (%)	14 ± 4.8
Booster strain, ɛa (%)	17 ± 4.6
Reservoir strain rate, SRs (1/ ɛs)	1.6 ± 0.5
Conduit strain rate, SRe (1/ ɛe)	-1.4 ± 0.6
Booster strain rate, SRa (1/ ɛa)	-2.2 ± 0.7
Ratio of SRe/SRa	0.7 ± 0.3

Echocardiography revealed normal left ventricular dimensions and systolic function. However, we noted reduced peak velocity flow in early diastole (0.67 ± 0.2 m/s) and E/A ratio (0.91 ± 0.34). In conjunction with peak early diastolic septal (0.073 ± 0.02 m/s) mitral annular velocity, low ratio of peak velocity flow in early diastole to peak early diastolic septal (9.67 ± 3.1), and non-enlarged left atrial volumes, these cardiac characteristics represent low grade diastolic dysfunction in absence of elevated left ventricular filling pressures, a finding typically observed in myocardial ageing.

Left atrial function parameters obtained via cardiac magnetic resonance imaging are summarized in Table 1. Mean values of LA reservoir strain (Ɛs), conduit strain (Ɛe) and booster strain (Ɛa) were 31 ± 7.2%, 14 ± 4.8% and 17 ± 4.6% respectively.

We performed linear regression analysis to determine associations between various cardiovascular parameters and HbA1c. HbA1c was significantly associated with E/A ratio (β = -0.10, 95% CI -0.15, -0.05, p < 0.0001), conduit strain (Ɛe) (β = -1.35, 95% CI -2.08, -0.61, p < 0.0001), conduit strain rate SRe (β = 0.16, 95% CI 0.073, 0.253, p < 0.0001), and conduit strain rate to booster strain rate SRe:Sra ratio (β = -0.10, 95% CI -0.14, -0.05, p < 0.0001).

Scatter plots demonstrated positive association between HbA1c and cardiac measurements, appearing mostly linear in nature (Fig. 1a, 1b and 1c). Explained variances between linear and quadratic regression analyses were minimally different, E/A (linear R² = 0.055; quadratic R² = 0.070; p = 0.047), Ɛe (linear R² = 0.054; quadratic R² = 0.061; p = 0.18), SRe/Sra (linear R² = 0.063; Quadratic R² = 0.067; p = 0.28) which suggested the model fit of quadratic regression model was not superior to the linear regression model.

Fig. 1.

a Scatter plot depicting the linear and quadratic relationship between HbA1c and E/A ratio. Linear regression model: E/A ratio = 1.51—0.099xHbA1c; Quadratic regression model: E/A ratio = 3.37—0.68xHbA1c + 0.045xHbA1c²; Linear R² = 0.055 and Quadratic R² = 0.070, p = 0.047 b Scatter plot depicting the linear and quadratic relationship between HbA1c and left atrial conduit strain Se. Linear regression model: Se = 21.6—1.35xHbA1c; Quadratic regression model: Se ratio = 39.5—6.98xHbA1c + 0.43xHbA1c²; Linear R² = 0.054 and Quadratic R² = 0.061, p = 0.18 c Scatter plot depicting the linear and quadratic relationship between HbA1c and left atrial strain ratio SRe:SRa. Linear regression model: SRe/SRa ratio = 1.25—0.095xHbA1c; Quadratic regression model: SRe/SRa ratio = 2.2—0.39xHbA1c + 0.023xHbA1c²; Linear R² = 0.063 and Quadratic R.² = 0.067, p = 0.28

Relationship between clinical variables, HbA1c and various cardiac indices

We performed univariate and multivariate analyses to examine contribution of various factors altering cardiovascular function (Table 2). Based on univariate analysis, age (p < 0.0001), BMI (p < 0.0001), systolic blood pressure (SBP) (p = 0.001), heart rate (HR) (p < 0.0001), diabetes mellitus (p = 0.044), smoking status (p = 0.021), and HbA1c levels (p < 0.0001) were associated with E/A (Model 1). Subsequently, multivariate models were constructed to adjust for potential confounders. In Model 2, adjustments were made for age, BMI, SBP, HR, diabetes mellitus, smoking, and HbA1c. In Model 3, additional adjustments were made for left ventricular ejection fraction (LVEF), interventricular septum thickness at end systole (IVSS), and HbA1c. LVEF and IVSS were associated with E/A (Online Resource 1, Supplementary Table S1). After correction for clinical variables (Model 2) and left ventricular measurements (Model 3), HbA1c was still independently associated with E/A ratio in all studied models (β = -0.062, 95%CI -0.115, -0.0097, p = 0.02). Each unit increase in HbA1c was associated with lower E/A ratio (-0.062, -0.115 to -0.0097).

Table 2.

Univariate and Multivariable Associations between Clinical Variables, HbA1c and Cardiac Indices

Cardiac Indices	Clinical and HbA1c covariates
E/A ratio	Age Beta (95%CI)	Age P value	Body mass index Beta (95%CI)	Body mass index P value	Systolic blood pressure Beta (95%CI)	Systolic blood pressure P value	Heart rate Beta (95%CI)	Heart rate P value	Diabetes mellitus Beta (95%CI)	Diabetes mellitus P value	Smoking Beta (95%CI)	Smoking P value	HbA1c Beta (95%CI)	HbA1c P value
E/A ratio: Each unit increase: Lower index indicating poorer cardiac function
Model 1	-0.16 (-0.25, -0.071)	< 0.0001	-0.024 (-0.037, -0.011)	< 0.0001	-0.003 (-0.005, -0.001)	0.001	-0.007 (-0.01, -0.003)	< 0.0001	-0.11 (-0.22, -0.003)	0.044	-0.13 (-0.24, -0.019)	0.021	-0.10 (-0.15, -0.047)	< 0.0001
Model 2	-0.13 (-0.21, -0.045)	0.003	-0.022 (-0.035, -0.01)	< 0.0001	-0.002 (-0.004, -0.0004)	0.012	-0.007 (-0.01, -0.004)	< 0.0001	-0.012 (-0.1, 0.12)	0.84	-0.057 (-0.16, 0.046)	0.27	-0.072 (-0.13, -0.017)	0.011
Model 3	-0.13 (-0.21, -0.04)	0.004	-0.025 (-0.038, -0.013)	< 0.0001	-0.0017 (-0.003, -0.0002)	0.026	-0.0066 (-0.0097, -0.0035)	< 0.0001	-0.020 (-0.13, 0.09)	0.721	-0.064 (-0.169, 0.041)	0.231	-0.062 (-0.115, -0.0097)	0.020
Se	Age Beta (95%CI)	Age P value	Height (cm) Beta (95%CI)	Height (cm) P value	Waist circumference (cm) Beta (95%CI)	Waist circumference (cm) P value	Hypertension Beta (95%CI)	Hypertension P value	Body mass index Beta (95%CI)	Body mass index P value	Smoking Beta (95%CI)	Smoking P value	HbA1c Beta (95%CI)	HbA1c P value
Se: Each unit increase: Lower index indicating poorer cardiac function
Model 1	-3.67 (-4.92, -2.42)	< 0.0001	0.10 (0.025, 0.185)	0.011	-0.087 (-0.155, -0.018)	0.013	-1.70 (-2.94, -0.47)	0.007	-	-	-2.94 (-4.53, -1.34)	< 0.0001	-1.35 (-2.08, -0.61)	< 0.0001
Model 4	-2.79 (-4.04, -1.54)	< 0.0001	0.13 (0.052, 0.207)	0.001	-0.066 (-0.13, -0.0002)	0.049	-0.49 (-1.65, 0.671)	0.405	-	-	-2.40 (-3.94, -0.866)	0.002	-0.790 (-1.49, -0.094)	0.026
SRe	Age Beta (95%CI)	Age P value	Height (cm) Beta (95%CI)	Height (cm) P value	Waist circumference (cm) Beta (95%CI)	Waist circumference (cm) P value	Hypertension Beta (95%CI)	Hypertension P value	Body mass index Beta (95%CI)	Body mass index P value	Smoking Beta (95%CI)	Smoking P value	HbA1c Beta (95%CI)	HbA1c P value
SRe: Each unit increase: Higher index indicating poorer cardiac function
Model 1	0.53 (0.381, 0.676)	< 0.0001	-	-	0.013 (0.0045, 0.021)	0.003	0.296 (0.148,0.444)	< 0.0001	0.031 (0.008, 0.055)	0.009	0.32 (0.124, 0.514)	0.001	0.163 (0.073, 0.253)	< 0.0001
Model 5	0.46 (0.309, 0.607)	< 0.0001	-	-	-0.004 (-0.016, 0.008)	0.527	0.14 (-0.0002, 0.277)	0.050	0.035 (0.001, 0.069)	0.043	0.22 (0.032, 0.402)	0.022	0.105 (0.023, 0.188)	0.013
SRe: SRa	Age Beta (95%CI)	Age P value	Heart rate Beta (95%CI)	Heart rate P value	Waist circumference (cm) Beta (95%CI)	Waist circumference (cm) P value	Systolic blood pressure Beta (95%CI)	Systolic blood pressure P value	Body mass index Beta (95%CI)	Body mass index P value	Smoking Beta (95%CI)	Smoking P value	HbA1c Beta (95%CI)	HbA1c P value
SRe/SRa: Each unit increase: Lower index indicating poorer cardiac function
Model 1	-0.244 (-0.33, -0.16)	< 0.0001	-0.006 (-0.009, -0.003)	< 0.0001	-0.008 (-0.013, -0.004)	< 0.0001	-0.002 (-0.004, -0.0009)	0.002	-0.015 (-0.028, 0.003)	0.018	-0.187 (-0.292, -0.082)	0.001	-0.095 (-0.143, 0.0472)	< 0.0001
Model 6	-0.214 (-0.293, -0.136)	< 0.0001	-0.006 (-0.009, -0.003)	< 0.0001	-0.003 (-0.0097, 0.0034)	0.346	-0.0014 (-0.0028, -0.00001)	0.048	-0.0081 (-0.026, 0.01)	0.379	-0.084 (-0.183, 0.015)	0.095	-0.063 (-0.107, -0.020)	0.005

E/A ratio Ratio of Peak velocity flow in early diastole to Peak velocity flow in late diastole by atrial contraction

Model 1 unadjusted (only significant covariates at univariate level are shown in the table)

Model 2 adjusted for age, body mass index, systolic blood pressure, heart rate, diabetes mellitus, smoking, HbA1c

Model 3 adjusted for age, body mass index, systolic blood pressure, heart rate, diabetes mellitus, smoking, LVEF: left ventricular ejection fraction, Interventricular septum thickness at end systole (IVSS), HbA1c

Model 4 adjusted for age, waist circumference, height, hypertension, smoking, HbA1c

Model 5 adjusted for age, waist circumference, hypertension, body mass index, smoking, HbA1c

Model 6 adjusted for age, heart rate, waist circumference, body mass index, systolic blood pressure, smoking, HbA1c

We performed univariate and multivariate analyses to examine association between cardiac MRI findings (conduit strain (Ɛe), conduit strain rate (SRe) and conduit strain rate to booster strain rate ratio (SRe:SRa) and clinical variables. At the univariate level, age (p < 0.0001), height (p = 0.011), waist circumference (p = 0.013), hypertension (p = 0.007), smoking (p < 0.0001) and HbA1c (p < 0.0001) were associated with conduit strain (Ɛe) (Model 1). Adjusting for these clinical variables, HbA1c remained associated with conduit strain (εe) (β = -0.79, 95%CI -1.49, -0.094, p = 0.026) (Model 4). Adjusted models for SRe (Model 5) and SRe:Sra ratio (Model 6) demonstrated similar independent associations with HbA1c. Each unit increase in HbA1c was associated with lower Ɛe (-0.79, -1.49 to -0.094), higher SRe (0.105, 0.023 to 0.188) and lower SRe:SRa ratio (-0.063, -0.107 to -0.02).

Sensitivity analysis performed without diabetic participants yielded similar results (Online Resource 1, Supplementary Table S2). HbA1c remained significantly associated with E/A ratio (β = -0.098, 95%CI -0.17, -0.025, p = 0.009) after adjustments were made for clinical variables (Model 2), and left ventricular measurements (β = -0.085, 95%CI -0.155, -0.015, p = 0.0017) (Model 3). HbA1c was similarly independently associated with conduit strain (Ɛe) (β = -1.345, 95%CI -2.323, -0.367, p = 0.007) (Model 4), conduit strain rate (SRe) (β = 0.207, 95%CI 0.087, 0.327, = 0.001) (Model 5) and conduit strain rate to booster strain rate ratio (SRe:SRa) (β = -0.07, 95%CI, -0.154, -0.025, p = 0.007) (Model 6) in all adjusted models.

Discussion

Our study demonstrated positive associations between HbA1c levels and specific myocardial functions of the ageing heart, independent of age and diabetes status.

Biological ageing refers to physiological changes the body undergoes as part of the ageing process. A complex process influenced by a host of genetic and environmental factors and thus varies between individuals. As opposed to chronological ageing which follows strictly the passage of time. Two individuals with the same chronological age might have different biological ages and therefore be predisposed to different ageing related diseases. Thus, biological ageing is a concept becoming increasingly relevant given the ageing population globally, and with it increasing prevalence of ageing related diseases, It has sparked interest in the search for biomarkers that can identify risk for ageing relating diseases. Ageing reviews have defined a good biomarker of biological ageing as one with measurement reliability and feasibility, representative of biologic ageing processes, robust and consistent association with risk of death and clinical or functional trial endpoints, and responsive to intervention, for which HbA1c was identified among top candidates [9, 10]. Given widespread availability and use of HbA1c in clinical practice including primary care settings, HbA1c may be an accessible laboratory test for ageing.

Several studies have already demonstrated an association between HbA1c, ageing and age-related health conditions. Frailty is a clinical syndrome characterized by increased vulnerability and decreased physiological reserves in older adults. One study showed that HbA1c of 6.5% and greater was significantly associated with higher likelihood of prefrail and frail status, independent of BMI, inflammation and comorbidities including osteoarthritis, coronary artery disease and chronic obstructive pulmonary disease [11]. A second study showed the association of presence of diabetes with high risk of frailty in comparison to non-diabetics, however was partially explained by unhealthy behaviours, obesity, poor glucose control and dyslipidaemia [12].

Given effects of ageing will vary from individual to individual, HbA1c represents a quantifiable marker that could be used to risk stratify older adults at risk of diastolic dysfunction.

Interestingly there are studies that evaluated relationship of HbA1c and ageing in non-diabetic subjects. Using data from two independent cohort studies, Masuch et al. confirmed increasing chronological age was associated with high HbA1c levels and derived age adjusted reference ranges for HbA1c, thereby showing the potential of measurement for HbA1c outside of its intended use in the diabetic population [5]. A cross-sectional analysis by Dubowitz et al. conducted in adults without known diabetes demonstrated both glucose intolerance and HbA1c increased with age, and this was independent of other clinical factors including race, BMI, waist circumference and lipid profile [13]. Finally results from EPIC-Norfolk showed HbA1c was continuously related to subsequent all cause, cardiovascular and ischemic heart disease mortality [1].

Our study population included subjects with a mean age of 71 years, an appropriate representation of the older adult population in which cardiac ageing might be prominent. These subjects were without any reported symptoms or prior medical diagnosis of cardiovascular disease. Subsequent examination with transthoracic echocardiography also revealed they had normal left ventricular function with a mean left ventricular ejection fraction (LVEF) of 65% (± 7.7 SD). In addition, mean HbA1c was 6.0% in a population with predominantly non-diabetics (91.0%). Despite this our results demonstrated a significant association between age, HbA1c and early stages of cardiac ageing, with measurable cardiac indices representative of impaired myocardial relaxation [14] typically observed with ageing and left atrial strain abnormalities. This suggests a role for use of HbA1c as a biomarker of cardiac ageing, independent of the presence of poorly controlled diabetes and cardiovascular disease.

In the initial univariate analysis (Model 1), significant associations were observed between HbA1c levels and cardiac indices, including the E/A ratio (from echocardiography), conduit strain (Ɛe), conduit peak strain rate (SRe), and SRe:SRa ratio (from CMR). Indicating that HbA1c is linked to various cardiac measures, supporting its potential role as a biomarker for cardiac ageing. In subsequent multivariate analyses (Model 2, Model 3, Model 4, Model 5 and Model 6), notably, even after accounting for effects of other clinical variables and cardiac measures, HbA1c remained significantly associated with the E/A ratio, Ɛe, SRe, and SRe:SRa ratio. This suggests the relationship between HbA1c with these various cardiac indices is independent of other clinical factors associated with ageing, further supporting its relevance and strengthens its potential as a valuable biomarker for assessing age-related changes in cardiac function.

We postulate that associations between HbA1c and cardiac ageing, independent of effects of diabetes, may be biologically linked to systemic accumulation of advanced glycated end products (AGE). AGE, like HbA1c, is a result of the glycation process, a non-enzymatic reaction between molecules such as sugars, proteins, lipids, or nucleic acids [15]. Similar to the ageing process, AGE occurs over an extended period of time, accumulating in walls of vessels and other tissues, leading to organ complications in affected tissues [16]. Accumulation of AGE in the vasculature could lead to increasing arterial stiffness and reduction in cardiac perfusion, and subsequently result in structural modifications leading to cardiac dysfunction [17]. Interestingly, AGE accumulation has been observed among patients with or without prevalent heart failure, and patients without heart failure but with cardiac systolic dysfunction [18].

Our findings expand on this field, demonstrating association between HbA1c and cardiac diastolic function in the older ageing adult population, upstream prior to prevalent heart disease. Given the ageing cardiovascular system is associated with heightened risks of future cardiovascular disease, HbA1c may serve as a biomarker for future cardiovascular risk in the general ageing population, like other serum biomarkers which have provided insights into cardiovascular risks in ageing [19]. Further research to corroborate these findings as well as further quantify relationship between level of HbA1c and corresponding age-related cardiac changes will be helpful. This will potentially enable integration of HbA1c measurements into routine cardiovascular assessments and improve early detection and management of cardiac ageing, even in individuals without diabetes and established cardiovascular disease.

Limitations

We acknowledge some limitations in the current study. Firstly, sample size of 247 subjects may be considered relatively small for drawing definitive conclusions. Further studies with larger and more diverse populations are warranted to validate and strengthen these findings. Additionally, study subjects were predominantly of Chinese ethnicity from the local community, potentially limiting application of the results to other ethnic groups. Influence of genetic and race-related factors on the association between HbA1c and cardiac ageing should be considered in future investigations. We did not correct for medication data such as drugs. However, the low levels of glycated haemoglobin coupled with low proportion of patients with diabetes suggest that observations are unlikely driven by diabetes, which was additionally adjusted for in the multivariable associations. The cross-sectional study design precludes causal inferences and implies an epidemiological association between HbA1c and cardiac ageing. While the value of using HbA1c in older adults with type 2 diabetes mellitus on stratifying risks of cardiac ageing is unclear, our findings provide useful estimates that could be used to track progression of cardiac ageing using HbA1c prior to onset of diabetes in older adults.

Conclusion

Our study provides evidence of a significant association between HbA1c levels and early stages of cardiac ageing, independent of presence of diabetes and cardiovascular disease. This suggests HbA1c may serve as a valuable biomarker for age related changes in cardiac function, inclusive of early changes such as impaired diastolic function and evidence of left atrial stain.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

All authors read and approved the final manuscript.

Funding

The Cardiac Ageing Study has received funding support from the National Medical Research Council (MOH-000153; HLCA21Jan-0052; MOH-001193; MOH-001200) and Hong Leong Foundation. The funder had no role in the design and conduct of the study; collection; management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript.

Data Availability

Availability of data and materials: The datasets generated and/or analysed during the current study are not publicly available due to institutional restrictions but are available from the corresponding author on reasonable request.

Declarations

Competing interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.