ABSTRACT
Aim
Limited data on the link between various measures of glycemia and metabolism and autonomic dysfunction exist. We assessed the association of fasting plasma glucose [FPG], glycated haemoglobin [HbA1C], fasting insulin, and HOMA‐IR and cardiac autonomic function measures separately for men and women in the Diabetes Prevention Program Outcomes Study (DPPOS).
Methods
We included DPPOS participants with ECG‐based heart rate variability (HRV) measured at the 10‐year examination. Resting heart rate (HR), standard deviation of all normal‐to‐normal intervals (SDNN), and root mean square of successive differences between normal‐to‐normal intervals (rMSSD) were measured. Multivariable linear regression models assessed the cross‐sectional associations of FPG, HbA1C, and fasting insulin and HOMA‐IR with cardiac autonomic function measures.
Results
Among 2266 participants (mean age 65 years, 68.2% women, 53.4% White, 43.7% with prediabetes and 56.3% with diabetes at the time of evaluation), HbA1C, FPG, fasting insulin and HOMA IR were positively associated with high resting HR in men and women. HbA1C and FPG were inversely associated with rMSSD (β [95% CI] per SD: −0.08 [−0.14, −0.03] and −0.08 [−0.13, −0.03]) and SDNN (−0.08 [−0.12, −0.03] and −0.07 [−0.11, −0.02]) among women, but not in men. Fasting insulin was inversely associated with rMSSD in the men (−0.09 [−0.18, 0.00]). HOMA‐IR was inversely associated with rMSSD in men (−0.11 [−0.20, −0.02] and women −0.07 [−0.12, −0.02]). HOMA‐IR was associated with SDNN only in women (−0.05 [−0.10, 0.00]).
Conclusions
Markers of dysglycaemia are associated with worse ECG‐based autonomic dysfunction in adults with prediabetes and diabetes in the DPPOS, more so for some glycaemic markers and in women.
Keywords: autonomic dysfunction, HbA1C , prediabetes, type 2 diabetes
1. Introduction
Type 2 diabetes is common in the US and is associated with high morbidity and mortality [1]. Autonomic dysfunction has been described as a serious and common complication of diabetes [2], which in turn confers a high risk of adverse outcomes. Among individuals with diabetes, cardiac autonomic neuropathy (CAN) has been associated with increased risks of arrhythmias, sudden cardiac death, and cardiovascular‐ and all‐cause mortality, as compared with individuals without CAN [3, 4, 5, 6].
Autonomic neuropathy may also occur during the early stages of diabetes [7] with evidence of decreased heart rate variability (HRV) [8, 9, 10, 11, 12, 13] or postural changes in heart rate [8] among people with prediabetes. Furthermore, hyperinsulinemia and insulin resistance, which are common in hyperglycaemic states (prediabetes and diabetes), may affect HRV [14, 15]. Previous studies of hyperglycaemia‐related autonomic dysfunction have seldom assessed the more than single measures of glycemia [16]. No prior studies have explored the relationship between different measures of glycemia (e.g., fasting plasma glucose [FPG], haemoglobin A1C [HbA1C]) and of insulin levels and insulin resistance measures and autonomic neuropathy. Indeed, prior studies have heavily relied on single parameters of glycemia, such as FPG or HbA1C, which may mask critical pathophysiological shifts in dysglycaemia. Specifically, these fail to capture intrinsic changes in insulin secretion and sensitivity, which can drive microvascular risks. Assessing a comprehensive suite of concurrent glucose regulation markers, can help bridge that gap—offering a more robust evaluation of CAN across the glycaemic spectrum.
Using data from the Diabetes Prevention Program (DPP), and its long‐term longitudinal follow‐up, the DPP Outcomes Study (DPPOS), we examined the cross‐sectional association of various measures of glycaemic markers (FPG and HbA1C), fasting insulin levels, and insulin resistance (homeostasis model assessment of insulin resistance [HOMA‐IR]) with measures of cardiac autonomic function.
2. Methods
2.1. Study Population
The Diabetes Prevention Program (DPP) study (1996–2001) is a randomised controlled trial that enrolled 3234 adults aged ≥ 25 years who were at high risk for developing diabetes (i.e., body mass index [BMI] ≥ 24 kg/m2, fasting glucose 5.3–6.9 mmol/L, and 2‐h glucose 7.8–11.0 mmol/L, and no previous diagnosis of diabetes) recruited from 27 U.S. clinical centres. The participants were randomly assigned to three groups: standard lifestyle recommendations plus placebo twice daily; standard lifestyle recommendations plus 850 mg of metformin twice daily; and an intensive program of lifestyle modification [17]. At study entry, all the study participants underwent a standardised 2‐h 75‐g oral glucose tolerance test (OGTT) after an overnight fast, which allowed the classification of the glycaemic status. The glycaemic categories were classified according to the American Diabetes Association (ADA) classification [18]. The OGTT was repeated annually and a FPG measured at 6‐month intervals. Diabetes was defined as a FPG ≥ 126 mg/dL or a 2hPG ≥ 200 mg/dL [18]. Prediabetes was defined as a FPG of 100 to 125 mg/dL or 2hPG of 140 to 200 mg/dL without criteria for diabetes being met [18]. Diabetes was diagnosed as an outcome when the above criteria were met and confirmed by a repeat test within 6 weeks. Once a diagnosis of diabetes was confirmed, OGTTs were discontinued but the fasting blood glucose assessment continued.
At the end of the DPP trial phase, all surviving participants regardless of their diabetes status were invited to enrol in the DPPOS (2002–present), which is an observational study, including 2779 participants. The DPPOS focused on the development of microvascular complications, cancer, and cardiovascular disease.
The present study represents a cross‐sectional analysis of individuals who attended the 10‐year examination of DPPOS (N = 2266), who completed a multi‐metric assessment of glycemia regulation and autonomic function testing through measures of HRV assessed at this exam.
The study was conducted according to the guidelines set out in the Declaration of Helsinki and all procedures involving research study participants. Prior to initiating the study protocol, each participant provided written informed consent and each study center obtained approval from its respective institutional review board. The trials are registered at ClinicalTrials.gov (Diabetes Prevention Program: NCT00004992; Diabetes Prevention Program Outcomes Study: NCT00038727; Diabetes Prevention Program Outcomes Study AD/ADRD: NCT05704309).
2.2. Glycaemic Status
The measures of glycemia assessed included FPG and HbA1C. FPG was measured using the glucose oxidase method. HbA1C was measured using high‐performance liquid chromatography (Tosoh G7 analyzer, Tosoh Corporation, Tokyo, Japan) standardised to the Diabetes Control and Complications Trial assay [19]. Insulin was measured by a radioimmunoassay method using an anti–guinea pig antibody that measures total immunoreactive insulin. The assay is a 48‐h polyethylene glycol–accelerated method with coefficients of variation (CVs) of 4.5% for high‐concentration quality control samples and 6.9% for low‐concentration quality control samples. The CV for masked split duplicates in this assay was < 8.5% [14]. Insulin‐related measures included fasting insulin levels and insulin resistance assessed by the homeostasis model assessment of insulin resistance (HOMA‐IR), which was calculated as (fasting insulin (μIU/ml) × fasting glucose (mmol/L))/22.5 [20].
2.3. Cardiac Autonomic Function Measures
Cardiac autonomic function measures were derived from 10‐s digital resting 12‐lead ECG recordings. These included mean heart rate across 12 leads (beats/min), standard deviation of all normal‐to‐normal R‐R intervals (SDNN) (milliseconds), the root mean square of successive differences between all normal‐to‐normal R‐R intervals (rMSSD) (milliseconds), and QT Index. The QT index was computed as QTI = QT*(HR + 100)/656 as defined by Rautaharju et al. [21].
Heart rate (assessed at rest) represents both overall autonomic function and cardiorespiratory fitness [22]. SDNN represents joint sympathetic and parasympathetic modulation of heart rate; rMSSD represents parasympathetic modulation of heart rate. Higher heart rate indicates overall poor autonomic function and low fitness, whereas lower SDNN and rMSSD represent poor parasympathetic and sympathetic functions, and poor sympathetic function only, respectively [23].
The overall presence of autonomic dysfunction was defined based on the following combination of criteria: SDNN < 17.13 ms and/or rMSSD < 24.94 ms [24] and/or QTI > 112 [25]. The SDNN and rMSSD cut‐points used for the detection of CAN were based on a recent study that validated these against CAN detected by cardiovascular reflex tests, the gold standard for CAN assessment [24]. The QTI cut point for CAN has been shown to be robustly associated with mortality and adverse cardiovascular outcomes [25].
2.4. Covariates
The covariates were assessed by standardised questionnaires, physical examination and laboratory tests. The main covariates used as adjustment variables in our models included the main factors known to affect HRV (identified based on prior studies), which are age, sex, race and treatment arm (placebo, metformin or lifestyle), as well as other factors including body mass index (BMI), systolic blood pressure, use of antihypertensive medications (including beta‐blockers, non‐dihydropyridine calcium channel blockers, and Angiotensin‐Converting Enzyme [ACE] inhibitors/Angiotensin II Receptor Blockers [ARBs]), ratio of total cholesterol to HDL‐cholesterol, use of glucose lowering medications, and estimated glomerular filtration rate (eGFR).
2.5. Statistical Analyses
Given the known influence of sex on cardiac autonomic function, we conducted analyses in the overall sample and stratified by sex. We presented the characteristics (at year 10 of DPPOS) of participants by sex and glycaemic status (prediabetes and diabetes). We assessed the differences in characteristics using the t‐test or Mann–Whitney U test for continuous variables and the χ 2 test for categorical variables.
We assessed the associations of glycaemic markers (or insulin‐related measures) and cardiac autonomic function measures using multivariable linear regression analyses to model each measure of cardiac autonomic function (heart rate, SDNN, rMSSD, and QTI) as a function of each glycemia‐ or insulin‐related measure (glycemia categories [prediabetes vs. diabetes (prior diagnosis)], FPG, HbA1C, fasting insulin, and HOMA‐IR), accounting for demographics (age and race), original treatment arm (lifestyle, metformin, or placebo), BMI, systolic blood pressure, use of antihypertensive medications, ratio of total cholesterol to HDL‐cholesterol, use of glucose lowering medications, and eGFR.
Each of the glycemia‐and insulin‐related markers was examined in models to evaluate their individual association with the autonomic function measures. To ensure a comparable scale for effect‐size interpretations, all exposure variables were standardised (z‐scored) across the full study sample prior to main and sex‐stratified analyses. Due to the marked right‐skewness inherent to metabolic and autonomic data, fasting insulin, HOMA‐IR, and the continuous metrics of SDNN and rMSSD were natural log‐transformed (ln) to satisfy the assumptions of normality. We tested for the effect modification of original treatment group (lifestyle or metformin) on the association of glycaemic measures and cardiac autonomic function measures by testing the interaction of glycaemic measures and treatment group.
We also assessed the prevalence of autonomic dysfunction in the two glycaemic groups (prediabetes and diabetes) and used multivariable logistic regression analyses to model autonomic dysfunction (binary variable) as a function of each glycemia‐ or insulin‐related measure (glycemia categories [prediabetes vs. diabetes], FPG, HbA1C, fasting insulin, and HOMA‐IR) using the same adjustment variables as done in the aforementioned multivariable linear models.
We also fitted separate models, adjusting for age and race, to explore the potential effect modification of sex on the association between glycaemic measures and cardiac autonomic function. Interaction terms between glycemia‐ or insulin‐related measures and sex were tested to assess whether the relationship between glycemia‐ or insulin‐related measures and cardiac autonomic function differs by sex. We also investigated the associations between the measures of glycaemic regulations and cardiac autonomic function measures by menopausal status (pre‐menopause vs. menopause [defined as age ≥ 55 years]).
To account for multiple testing and control the false discovery rate (FDR), the Benjamini‐Hochberg (BH) adjustment was applied. We defined the ‘family’ of tests for each adjustment procedure as follows: (1) primary cross‐sectional models evaluating all metabolic exposures against primary autonomic outcomes; (2) sex‐stratified models; and (3) menopausal‐status stratified models. Interaction terms were adjusted within their own distinct exploratory family. Statistical significance after correction was defined as a BH‐adjusted p‐value (q‐value) < 0.05. All analyses were performed using R version 4.0.0.
3. Results
3.1. Characteristics of Participants
Among the 2266 eligible participants (mean age: 65 [SD: 9.9] years, 68.2% women, 53.4% White), 991 (43.7%) had prediabetes and 1275 (56.3%) had a diagnosis of diabetes by year 10 of DPPOS. A total of 2132 participants had complete data for the measures of cardiac autonomic function and glycaemic measures.
Compared to men, women were younger, had higher BMI, eGFR, cholesterol, fasting insulin, heart rate and QTI (Table S1). Among men, compared to participants with prediabetes, those with diabetes were older, had a lower eGFR and higher BMI, SBP, DBP, glycaemic markers, cholesterol level and heart rate (Table 1). Among women, compared to participants with prediabetes, those with diabetes were older, less likely to be White, and more likely to have a lower eGFR and higher BMI, glycaemic markers, cholesterol level and heart rate (Table 1). The comparison of the characteristics of individuals with diabetes to those with prediabetes is shown in Table S2.
TABLE 1.
Characteristics of the Diabetes Prevention Program Outcomes Study participants by sex and diabetes status at year 10 of DPPOS.
| Male N = 720 | p value | Female N = 1546 | p | |||||
|---|---|---|---|---|---|---|---|---|
| Prediabetes N = 310 | Diabetes N = 410 | Total N = 720 | Prediabetes N = 681 | Diabetes N = 865 | Total N = 1546 | |||
| Age at visit, years | 69.0 (10.2) | 67.5 (10.1) | 68.2 (10.2) | 0.052 | 64.4 (9.6) | 62.8 (9.1) | 63.5 (9.4) | 0.001 |
| Race or ethnic group | ||||||||
| White | 192 (61.9) | 232 (56.6) | 424 (58.9) | 0.064 | 386 (56.7) | 401 (46.4) | 787 (50.9) | < 0.001 |
| Non‐Hispanic Black | 37 (11.9) | 74 (18.0) | 111 (15.4) | 129 (18.9) | 233 (26.9) | 362 (23.4) | ||
| Hispanic | 47 (15.2) | 64 (15.6) | 111 (15.4) | 94 (13.8) | 134 (15.5) | 228 (14.7) | ||
| American Indian | 9 (2.9) | 4 (1.0) | 13.1 (8) | 57 (8.4) | 69 (8.0) | 126 (8.2) | ||
| Asian or Pacific Islander | 25 (8.1) | 36 (8.8) | 61 (8.5) | 15 (2.2) | 28 (3.2) | 43 (2.8) | ||
| Body mass index, kg/m2 | 29.2 (4.8) | 31.9 (6.6) | 30.7 (6.0) | < 0.001 | 32.7 (6.6) | 34.3 (7.2) | 33.6 (7.0) | < 0.001 |
| SBP, mm Hg | 119.9 (14.0) | 123.1 (14.6) | 121.7 (14.4) | 0.003 | 120.9 (14.9) | 121.5 (15.0) | 121.2 (15.0) | 0.393 |
| Fasting plasma glucose, mg/dL | 104.5 (9.4) | 133.2 (34.3) | 120.9 (30.2) | < 0.001 | 102.5 (9.8) | 132.3 (43.2) | 119.2 (36.2) | < 0.001 |
| HbA1C, % | 5.5 (0.4) | 6.5 (1.2) | 6.1 (1.1) | < 0.001 | 5.6 (0.4) | 6.6 (1.4) | 6.1 (1.2) | < 0.001 |
| Fasting insulin | 24.8 (17.0 to 33.7) | 31.7 (21.7 to 48.2) | 28.3 (19.7 to 42.4) | < 0.001 | 26.2 (17.9 to 37.2) | 34.0 (23.4 to 50.0) | 30.4 (20.6 to 44.3) | < 0.001 |
| HOMA‐IR | 6.5 (4.4 to 8.8) | 10.3 (6.6 to 16.5) | 8.0 (5.3 to 12.9) | < 0.001 | 6.5 (4.4 to 9.7) | 10.4 (6.7 to 16.6) | 8.6 (5.4 to 13.4) | < 0.001 |
| BP‐lowering med(s) | 175 (56.5) | 303 (73.9) | 478 (66.4) | < 0.001 | 352 (51.7) | 606 (70.1) | 958 (62.0) | < 0.001 |
| Glucose‐lowering med(s) | 11 (3.5) | 210 (51.2) | 221 (30.7) | < 0.001 | 36 (5.3) | 428 (49.5) | 464 (30.0) | < 0.001 |
| Heart rate‐(bpm) | 60.1 (10.0) | 64.2 (11.6) | 62.4 (11.1) | < 0.001 | 64.3 (9.4) | 66.2 (11.0) | 65.3 (10.4) | < 0.001 |
| SDNN | 26.1 (19.6) | 24.7 (15.0) | 25.3 (17.2) | 0.323 | 25.5 (16.0) | 25.0 (16.5) | 25.2 (16.3) | 0.545 |
| rMSD | 25.8 (25.8) | 22.5 (16.4) | 23.9 (21.1) | 0.058 | 24.2 (19.1) | 23.4 (19.4) | 23.7 (19.3) | 0.462 |
| QTI | 101.4 (5.6) | 101.7 (5.5) | 101.6 (5.5) | 0.575 | 102.3 (4.6) | 102.3 (4.6) | 102.3 (4.6) | 0.907 |
| Autonomic dysfunction | 209 (67.6) | 258 (64.0) | 467 (65.6) | 0.354 | 443 (65.4) | 595 (69.0) | 1038 (67.4) | 0.151 |
Note: Autonomic Dysfunction: SDNN < 17.13 ms and/or rMSSD < 24.94 ms and/or QTI > 112.
3.2. Glycemia‐ and Insulin‐Related Markers and Indices of Cardiac Autonomic Function
The multivariable adjusted linear regression analyses of the associations of glycaemic markers and indices of cardiac autonomic function are shown in Figure 1 and Table S3. Increasing levels of HbA1C and FPG were significantly and positively associated with heart rate among both men (beta coefficient [95% CI] per SD: 2.09 [0.74, 3.45] and 2.38 [0.98, 3.77]) and women (β [95% CI] per SD: 1.75 [0.91, 2.58] and 1.99 [1.16, 2.83]). Higher levels of HbA1C and FPG were significantly inversely associated with rMSSD (β [95% CI] per SD: −0.08 [−0.14, −0.03] and −0.08 [−0.13, −0.03]) and SDNN (β [95% CI] per SD: −0.08 [−0.12, −0.03] and −0.07 [−0.11, −0.02]) among women, but not in men.
FIGURE 1.
Associations of glycemia‐and insulin‐related measures and measures of cardiac autonomic function at year 10 of Diabetes Prevention Program Outcomes Study (linear regression models). Panel A: Overall sample. Panel B: Women. Panel C: Men. Each model follows the structure: One measure of cardiac autonomic function ~ one measure of glycemia + adjustment variables. FPG: Fasting plasma glucose, HOMA‐IR: Homeostasis model assessment of insulin resistance, rMSSD: Root mean square of successive differences between all normal‐to‐normal R‐R intervals, SDNN: Standard deviation of all normal‐to‐normal R‐R intervals.
Increasing levels of fasting insulin were significantly and positively associated with heart rate among men (β [95% CI] per SD: 2.65 [1.27, 4.03]) and women (β [95% CI] per SD: 2.73 [1.86, 3.60], p < 0.05). Fasting insulin was significantly inversely associated with rMSSD among men (β [95% CI] per SD: −0.09 [−0.18, 0.00]), but not women. No significant association was observed between fasting insulin and SDNN.
HOMA‐IR was significantly and positively associated with heart rate among both men (β [95% CI] per SD: 3.04 [1.67, 4.42]) and women (β [95% CI] per SD: 3.05 [2.18, 3.91]). Higher levels of HOMA‐IR were significantly inversely associated with rMSSD among men (β [95% CI] per SD: −0.11 [−0.20, −0.02]) and women (β [95% CI] per SD: −0.07 [−0.12, −0.02]). HOMA‐IR was associated with SDNN in women only (β [95% CI] per SD: −0.05 [−0.10, 0.00]).
Except for heart rate in men, diabetes status was not associated with measures of cardiac autonomic function. Overall, no effect modification was found from testing the interaction of glycaemic markers and treatment groups for any of the glycaemic markers and individual autonomic function parameters associations in linear multivariable models.
In analyses testing the interaction between glycaemic markers and sex was significant for the associations between heart rate and HbA1C (p = 0.011), and between heart rate and diabetes status (p = 0.029). Similarly, significant interactions were observed for the associations between rMSSD and fasting insulin (p = 0.005), rMSSD and HOMA‐IR (p = 0.006), and between SDNN and fasting insulin (p = 0.028).
When we examined premenopausal women and post‐menopausal women, the associations of HbA1C, FPG, fasting insulin, HOMA‐IR, and glycaemic status with heart rate tended to be stronger in magnitude for pre‐menopausal women (Table S4). Similar patterns were observed for the HbA1C, FPG, fasting insulin, and HOMA‐IR associations with rMSSD and with SDNN (Table S4).
3.3. Glycemia‐ and Insulin‐Related Markers and Cardiac Autonomic Dysfunction
In multivariable adjusted logistic regression analyses (Figure 2 and Table 2), HbA1C and FPG were significantly associated with autonomic dysfunction among all study participants (OR [95% CI] per SD: 1.27 (1.09, 1.51) and 1.20 (1.02, 1.42), p < 0.05). The corresponding associations were non‐significant in men. There was no significant association between fasting insulin or HOMA‐IR and overall autonomic dysfunction in both men and women.
FIGURE 2.
Associations of glycemia‐and insulin‐related measures and the presence of autonomic dysfunction (cardiac autonomic neuropathy) at year 10 of Diabetes Prevention Program Outcomes Study (logistic regression models). Each model follows the structure: Autonomic Dysfunction ~ one measure of glycemia + adjustment variables. FPG: Fasting plasma glucose, HOMA‐IR: Homeostasis model assessment of insulin resistance, rMSSD: Root mean square of successive differences between all normal‐to‐normal R‐R intervals, SDNN: Standard deviation of all normal‐to‐normal R‐R intervals.
TABLE 2.
Adjusted estimates of the association of glycemia‐and insulin‐related measures and autonomic dysfunction at year 10 of DPPOS.
| Markers | Group | Autonomic dysfunction | |
|---|---|---|---|
| Odds ratio (95% CI) | p value | ||
| HbA1C (%) | Overall | 1.27 (1.08, 1.51) | 0.005 |
| Female | 1.26 (1.03, 1.54) | 0.024 | |
| Male | 1.32 (0.97, 1.80) | 0.075 | |
| FPG (mg/dL) | Overall | 1.20 (1.02, 1.42) | 0.028 |
| Female | 1.21 (0.99, 1.48) | 0.062 | |
| Male | 1.21 (0.90, 1.63) | 0.209 | |
| Fasting insulin (μU/mL) | Overall | 1.09 (0.93, 1.27) | 0.279 |
| Female | 1.06 (0.88, 1.27) | 0.537 | |
| Male | 1.15 (0.87, 1.51) | 0.338 | |
| HOMA IR | Overall | 1.15 (0.98, 1.34) | 0.085 |
| Female | 1.12 (0.93, 1.35) | 0.235 | |
| Male | 1.20 (0.91, 1.59) | 0.199 | |
| Diabetes status | Overall | 1.12 (0.82, 1.52) | 0.481 |
| Female | 1.09 (0.75, 1.58) | 0.646 | |
| Male | 1.20 (0.68, 2.12) | 0.530 | |
Note: Autonomic Dysfunction: SDNN < 17.13 ms and/or rMSSD < 24.94 ms and/or QTI > 112.
Adjustment variables include age, race, original treatment arm (lifestyle, metformin, or placebo), body mass index (BMI), systolic blood pressure, use of antihypertensive medications, ratio of total cholesterol to HDL‐cholesterol, use of glucose lowering medications, and estimated glomerular filtration rate (eGFR).
Each model follows the structure: Autonomic Dysfunction~one measure of glycemia + adjustment variables. Glycaemic status: Two categories (diabetes and prediabetes) with prediabetes as reference. Fasting insulin and HOMA IR are in log scale. HbA1C, FPG, Fasting insulin and HOMA IR are standardised.
4. Discussion
In DPPOS, we found that markers of glycaemic control (higher levels of FPG and HbA1C) were positively associated with a high resting heart rate among women and men, and inversely associated with rMSSD, SDNN, and the overall presence of autonomic dysfunction, mainly among women. HOMA‐IR was inversely associated with rMSSD and SDNN in women.
Our findings provide additional insight on the links between measures of glycemia and cardiac autonomic function parameters. Our results complement the body of knowledge on the adverse effects of dysglycaemia on cardiac autonomic function. Our results showing an association between fasting glucose and HRV are consistent with those of prior community‐based studies including individuals with or without diabetes [9, 10, 11, 16, 26]. Our findings on the relation of glycaemic markers and high resting heart rate are also congruent with prior studies [27]. Our results are also concordant with prior studies of hyperinsulinemia [28] or insulin resistance and HRV [29, 30, 31] or heart rate [29]. While the vast majority of these prior studies used similar approaches to assess HRV (based on ECG), these did not always focus on those with prediabetes nor did they include the full array of glycaemic markers [16]. Indeed, our study attempted to capture the whole spectrum of dysglycaemia (including prediabetes and diabetes), and all aspects of the metabolic milieu associated with hyperglycaemia, as captured by markers of insulin resistance. Importantly, our study highlighted two important aspects. First, the sex differences in terms of cardiac autonomic neuropathy risk associated with dysglycaemia. Second, autonomic neuropathy was present in both prediabetes and diabetes, with a lack of major differences between prediabetes and diabetes. The absence of a strict linear gradient across worsening clinical categories suggests that pathophysiological pathways disrupting cardiovagal tone and ventricular repolarisation are engaged early in glucose intolerance. Furthermore, subclinical autonomic alterations may plateau by the time individuals meet diagnostic criteria for diabetes, highlighting the value of early metabolic tracking before overt disease manifests. The prevalence of cardiac autonomic dysfunction observed in our study is higher than that previously described in other studies among individuals with prediabetes or diabetes [13]. The reasons for these differences are probably manifold. First, there are no agreed upon cut‐points for the short resting ECG‐based HRV indices used for determining CAN. Hence a variety of cut‐points for the various HRV indices have been used [13]. We used contemporary rMSSD and SDNN cut‐points that have recently been validated against CAN detected by the gold‐standard of cardiovascular reflex tests in a population with type 1 diabetes [24]. The population with type 1 diabetes is known to have higher levels of CAN. Our ECG‐based CAN definition additionally included the QT index criterion, which has not always been used in prior studies. Second, we evaluated resting heart rate alongside HRV metrics; it is strongly influenced by non‐autonomic determinants such as cardiorespiratory fitness levels, physical activity habits, underlying subclinical inflammation, daily volume status, and chronotropic medications. Hence, resting heart rate is a broad indicator of autonomic balance rather than a surrogate for distinct autonomic pathways. Third, the high prevalence of cardiac autonomic dysfunction among individuals with prediabetes may reflect the impact of historical clinical management within this cohort, with those who developed diabetes subject to closer monitoring and glucose‐lowering interventions, which may have stabilised or slowed further autonomic degradation. Conversely, those remaining in the prediabetes category did not receive targeted pharmacological glucose‐lowering therapy, potentially narrowing the gap in autonomic outcomes between the two groups. Finally, the prevalence depends on the HRV cut‐off points defining abnormality; hence, our results should be interpreted in the context of the cut‐off points chosen for the analysis.
The exact mechanisms through which hyperglycaemia leads to autonomic dysfunction are unclear. The possible mechanisms linking hyperglycaemia to neuronal damage and subsequent autonomic dysfunction include, among others, oxidative stress, endothelial dysfunction, the formation of advanced glycation end‐products (AGEs), as well as the polyol pathway, poly(ADP ribose) polymerase (PARP), hexosamine pathway, protein kinase C pathway, hedgehog (Hh) signalling, and pro‐inflammatory pathways [32]. Several conceivable mechanisms may explain the observed sex differences. Prior to menopause, women typically exhibit a relative parasympathetic (vagal) dominance, largely mediated by the cardioprotective effects of oestrogen, which enhances nitric oxide bioavailability and suppresses sympathetic overactivity.
This study has potential implications for patients with dysglycaemia including prediabetes and diabetes. Our findings suggest that decreases in autonomic function are present at early stages of glycaemic dysregulation, which suggests that primordial prevention to avoid the development of any form of chronic hyperglycaemia may be the best approach to preventing autonomic nerve damage. Addressing autonomic dysfunction occurrence and its progression among individuals with dysglycaemia may help reduce the burden of long‐term complications including cardiovascular outcomes and mortality, which have been shown to be strongly associated with autonomic dysfunction [4, 33]. Additionally, a tool as simple as the standard 12‐lead resting ECG may help identify individuals with autonomic dysfunction.
There are limitations to our study. First, autonomic dysfunction measures were derived from short ECG recordings with HRV assessed using time‐domain indices only, and we did not use ECG data recorded over a longer period of time or perform dynamic cardiovascular autonomic reflex tests [33], the gold standard for autonomic dysfunction detection. Short‐term recordings do not account for circadian variations or longer‐term sympathetic‐parasympathetic shifts provided by 24‐h ambulatory monitoring. Furthermore, we lack data on frequency‐domain and non‐linear indices, and thus on spectral power. Nevertheless, time‐domain parameters, particularly rMSSD, are highly reliable, stable, and less susceptible to artefacts in short tracings compared to frequency‐domain metrics. It is possible that we missed some cases of autonomic dysfunction and possibly underestimated the true extent of the contribution of dysglycaemia to autonomic dysfunction. However, HRV indices have been validated against the gold standard for assessing autonomic neuropathy [13]. Because resting short‐term ECGs cannot capture the dynamic responsivity of the autonomic nervous system, our measures should be strictly interpreted as indicators of ECG‐derived subclinical autonomic dysregulation rather than definite CAN. Second, our study did not include any baseline assessment of autonomic dysfunction. It was a cross‐sectional study, so we cannot make direct causal inferences. Third, we may also have not captured all individuals with prediabetes, as the individuals included only those with impaired glucose tolerance and an elevated fasting glucose > 95 mg/dL (who were individuals at particularly high risk for diabetes development). Indeed, HbA1C was not used as a criterion for identifying prediabetes or diabetes at the start of the study. Fourth, the current study may have been subject to selection bias (healthy survivor effect), as DPPOS participants with severe cardiovascular comorbidities or profound autonomic degradation may have been disproportionately lost to follow‐up as they may have suffered premature cardiovascular events. Fifth, classifying the menopausal status strictly by an age proxy (age > 55) is a crude metric that may misclassify women undergoing early or late menopause, and those with a history of surgical menopause or hormone replacement therapy. Finally, there is a possibility of residual or unmeasured confounding, especially as our adjustment focused on the most relevant covariates.
Our study has several important strengths. First, we explored a large, diverse sample of adults with prediabetes or type 2 diabetes and conducted the assessment of various aspects of autonomic dysfunction using several measures of HRV and multiple glycaemic markers. We also tested the robustness of our findings by accounting for therapy received within the framework of the trial.
5. Conclusion
In conclusion, in a large cohort of individuals with prediabetes or type 2 diabetes, there was a high prevalence of autonomic dysfunction in dysglycaemia (including prediabetes and diabetes), and glycaemic markers were associated with ECG‐based autonomic dysfunction, especially among women. Our study highlights a potentially high burden of cardiac autonomic neuropathy in the early stages of dysglycaemia, especially at the prediabetes stage, pointing to the need for future research to investigate whether short‐term recordings from a standard 12‐lead ECG could offer an accessible avenue for screening for subclinical autonomic dysfunction.
Author Contributions
Justin B. Echouffo‐Tcheugui: conceptualisation, formal analysis, investigation, methodology, writing – original draft, review and editing. Xun Liu: formal analysis, writing – review and editing. Elsayed Z. Soliman, Xavier Pi‐Sunyer, Sherita H. Golden, Sam Dagogo‐Jack, Neil H. White, Matthew Budoff, Uzoma Ibebuogu, Christine Darwin, William C. Knowler, Marinella Temprosa: conceptualisation, investigation, methodology, writing – review and editing. William C. Knowler and Marinella Temprosa: access to all data. Justin B. Echouffo‐Tcheugui and Marinella Temprosa are the guarantors of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Funding
The research reported in this publication was supported by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) of the National Institutes of Health (NIH) under award numbers U01 DK048489, U01 DK048339, U01 DK048377, U01 DK048349, U01 DK048381, U01 DK048468, U01 DK048434, U01 DK048485, U01 DK048375, U01 DK048514, U01 DK048437, U01 DK048413, U01 DK048411, U01 DK048406, U01 DK048380, U01 DK048397, U01 DK048412, U01 DK048404, U01 DK048387, U01 DK048407, U01 DK048443, and U01 DK048400, and in part by the National Institute on Aging of the NIH under award number 5 U19 AG07855, by providing funding during the Diabetes Prevention Program (DPP) and DPPOS (Diabetes Prevention Program Outcomes Study) to the clinical centres and the Coordinating Centre for the design and conduct of the study, and collection, management, analysis, and interpretation of the data. Funding was also provided by the National Institute of Child Health and Human Development, the National Institute on Aging, the National Eye Institute, the National Heart Lung and Blood Institute, Division of Cancer Epidemiology, the Office of Research on Women's Health, the National Institute on Minority Health and Health Disparities, the Centers for Disease Control and Prevention, and the American Diabetes Association. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The Southwestern American Indian Centers were supported directly by the NIDDK, including its Intramural Research Program, and the Indian Health Service. The General Clinical Research Center Program, National Center for Research Resources, and the Department of Veterans Affairs supported data collection at many of the clinical centers. Merck KGaA provided medication for DPPOS. DPP/DPPOS have also received donated materials, equipment, or medicines for concomitant conditions from Bristol‐Myers Squibb, Parke‐Davis, and LifeScan Inc., Health O Meter, Hoechst Marion Roussel Inc., Merck‐Medco Managed Care Inc., Merck and Co., Nike Sports Marketing, Slim Fast Foods Co., and Quaker Oats Co. McKesson BioServices Corp., Matthews Media Group Inc., and the Henry M. Jackson Foundation provided support services under subcontract with the Coordinating Center. The sponsor of this study was represented on the Steering Committee and played a part in study design, how the study was done, and publication. All authors in the writing group had access to all data. The opinions expressed are those of the study group and do not necessarily reflect the views of the funding agencies. Dr. Echouffo Tcheugui was supported by NIH/NHLBI grant K23 HL153774.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Table S1: Characteristics of the Diabetes Prevention Program Outcomes Study participants at Visit 10 by sex.
Table S2: Characteristics of the Diabetes Prevention Program Outcomes Study participants at Visit 10 by glycaemic status.
Table S3: Adjusted estimates of the association of glycemia‐and insulin‐related measures and cardiac autonomic function measures at year 10 of DPPOS.
Table S4: Associations of glycemia‐and insulin‐related measures and measures of cardiac autonomic function at year 10 of DPPOS by menopausal status.
Appendix S1: DPP, DPPOS I & DPPOS II Research Group (1996‐2013).
Acknowledgements
The authors thank the staff and participants of the Diabetes Prevention Program (DPP) and DPPOS (Diabetes Prevention Program Outcomes Study). The sponsor of this study was represented on the Steering Committee and played a part in study design, how the study was done, and publication. All authors in the writing group had access to all data. The opinions expressed are those of the study group and do not necessarily reflect the views of the funding agencies.
Echouffo‐Tcheugui J. B., Liu X., Soliman E. Z., et al., “Dysglycaemia and Autonomic Dysfunction: The Diabetes Prevention Program Outcomes Study,” Diabetes, Obesity and Metabolism 28, no. 11 (2026): 10743–10752, 10.1111/dom.71111.
Handling Editor: Huilin Tang
Data Availability Statement
In accordance with the NIH Public Access Policy, we continue to provide all manuscripts to PubMed Central including this manuscript DPP/DPPOS has provided the protocols and lifestyle and medication intervention manuals to the public through its public website (https://www.dppos.org). The DPPOS abides by the NIDDK data sharing policy and implementation guidance as required by the NIH/NIDDK (https://repository.niddk.nih.gov/studies/dppos/). New materials generated in the conduct of this research are available from authors upon reasonable request.
References
- 1. Menke A., Casagrande S., Geiss L., and Cowie C. C., “Prevalence of and Trends in Diabetes Among Adults in the United States, 1988‐2012,” JAMA 314, no. 10 (2015): 1021–1029. [DOI] [PubMed] [Google Scholar]
- 2. Pop‐Busui R., Boulton A. J. M., Feldman E. L., et al., “Diabetic Neuropathy: A Position Statement by the American Diabetes Association,” Diabetes Care 40, no. 1 (2017): 136–154. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Chowdhury M., Nevitt S., Eleftheriadou A., et al., “Cardiac Autonomic Neuropathy and Risk of Cardiovascular Disease and Mortality in Type 1 and Type 2 Diabetes: A Meta‐Analysis,” BMJ Open Diabetes Research & Care 9, no. 2 (2021): e002480. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Maser R. E., Mitchell B. D., Vinik A. I., Freeman R., and Freeman R., “The Association Between Cardiovascular Autonomic Neuropathy and Mortality in Individuals With Diabetes: A Meta‐Analysis,” Diabetes Care 26, no. 6 (2003): 1895–1901. [DOI] [PubMed] [Google Scholar]
- 5. Rizzo M. R., Sasso F. C., Marfella R., et al., “Autonomic Dysfunction Is Associated With Brief Episodes of Atrial Fibrillation in Type 2 Diabetes,” Journal of Diabetes and Its Complications 29, no. 1 (2015): 88–92. [DOI] [PubMed] [Google Scholar]
- 6. Sardu C., Paolisso P., Santamaria M., et al., “Cardiac Syncope Recurrence in Type 2 Diabetes Mellitus Patients vs. Normoglycemics Patients: The CARVAS Study,” Diabetes Research and Clinical Practice 151 (2019): 152–162. [DOI] [PubMed] [Google Scholar]
- 7. Pfeifer M. A., Weinberg C. R., Cook D. L., et al., “Autonomic Neural Dysfunction in Recently Diagnosed Diabetic Subjects,” Diabetes Care 7, no. 5 (1984): 447–453. [DOI] [PubMed] [Google Scholar]
- 8. Wu J.‐S., Yang Y.‐C., Lin T.‐S., et al., “Epidemiological Evidence of Altered Cardiac Autonomic Function in Subjects With Impaired Glucose Tolerance but Not Isolated Impaired Fasting Glucose,” Journal of Clinical Endocrinology & Metabolism 92, no. 10 (2008): 3885–3889. [DOI] [PubMed] [Google Scholar]
- 9. Gerritsen J., Dekker J. M., TenVoorde B. J., et al., “Glucose Tolerance and Other Determinants of Cardiovascular Autonomic Function: The Hoorn Study,” Diabetologia 43, no. 5 (2000): 561–570. [DOI] [PubMed] [Google Scholar]
- 10. Singh J. P., Larson M. G., O'Donnell C. J., et al., “Association of Hyperglycemia With Reduced Heart Rate Variability (The Framingham Heart Study),” American Journal of Cardiology 86, no. 3 (2000): 309–312. [DOI] [PubMed] [Google Scholar]
- 11. Schroeder E. B., Chambless L. E., Liao D., et al., “Diabetes, Glucose, Insulin, and Heart Rate Variability: The Atherosclerosis Risk in Communities (ARIC) Study,” Diabetes Care 28, no. 3 (2005): 668–674. [DOI] [PubMed] [Google Scholar]
- 12. Ziegler D., Voss A., Rathmann W., et al., “Increased Prevalence of Cardiac Autonomic Dysfunction at Different Degrees of Glucose Intolerance in the General Population: The KORA S4 Survey,” Diabetologia 58, no. 5 (2015): 1118–1128. [DOI] [PubMed] [Google Scholar]
- 13. Eleftheriadou A., Williams S., Nevitt S., et al., “The Prevalence of Cardiac Autonomic Neuropathy in Prediabetes: A Systematic Review,” Diabetologia 64, no. 2 (2021): 288–303. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Hoffman R. P., Hausberg M., Sinkey C. A., and Anderson E. A., “Hyperglycemia Without Hyperinsulinemia Produces Both Sympathetic Neural Activation and Vasodilation in Normal Humans,” Journal of Diabetes and Its Complications 13, no. 1 (1999): 17–22. [DOI] [PubMed] [Google Scholar]
- 15. Frontoni S., Bracaglia D., Baroni A., et al., “Early Autonomic Dysfunction in Glucose‐Tolerant but Insulin‐Resistant Offspring of Type 2 Diabetic Patients,” Hypertension 41, no. 6 (2003): 1223–1227. [DOI] [PubMed] [Google Scholar]
- 16. Coopmans C., Zhou T. L., Henry R. M. A., et al., “Both Prediabetes and Type 2 Diabetes Are Associated With Lower Heart Rate Variability: The Maastricht Study,” Diabetes Care 43, no. 5 (2020): 1126–1133. [DOI] [PubMed] [Google Scholar]
- 17. The Diabetes Preventio Program Research Group , “The Diabetes Prevention Program. Design and Methods for a Clinical Trial in the Prevention of Type 2 Diabetes,” Diabetes Care 22, no. 4 (1999): 623–634. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. American Diabetes Association Professional Practice Committee , “Diagnosis and Classification of Diabetes: Standards of Care in Diabetes‐2024,” Diabetes Care 47, no. Suppl 1 (2024): S20–S42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Diabetes Control and Complications Trial Research Group , Nathan D. M., Genuth S., et al., “The Effect of Intensive Treatment of Diabetes on the Development and Progression of Long‐Term Complications in Insulin‐Dependent Diabetes Mellitus,” New England Journal of Medicine 329, no. 14 (1993): 977–986. [DOI] [PubMed] [Google Scholar]
- 20. Matthews D. R., Hosker J. P., Rudenski A. S., Naylor B. A., Treacher D. F., and Turner R. C., “Homeostasis Model Assessment: Insulin Resistance and Beta‐Cell Function From Fasting Plasma Glucose and Insulin Concentrations in Man,” Diabetologia 28, no. 7 (1985): 412–419. [DOI] [PubMed] [Google Scholar]
- 21. Rautaharju P. M., Warren J. W., and Calhoun H. P., “Estimation of QT prolongation. A Persistent, Avoidable Error in Computer Electrocardiography,” Journal of Electrocardiology 23, no. Suppl (1990): 111–117. [DOI] [PubMed] [Google Scholar]
- 22. Blomqvist C. G. and Saltin B., “Cardiovascular Adatations to Physical Training,” Annual Review of Physiology 4 (1983): 169–189. [DOI] [PubMed] [Google Scholar]
- 23. Malik M., Bigger J. T., Camm A. J., et al., “Heart Rate Variability: Standards of Measurement, Physiological Interpretation, and Clinical Use,” European Heart Journal 17, no. 3 (1996): 354–81.l. [PubMed] [Google Scholar]
- 24. Pop‐Busui R., Backlund J. Y. C., Bebu I., et al., “Utility of Using Electrocardiogram Measures of Heart Rate Variability as a Measure of Cardiovascular Autonomic Neuropathy in Type 1 Diabetes Patients,” Journal of Diabetes Investigation 13, no. 1 (2022): 125–133. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Pop‐Busui R., Evans G. W., Gerstein H. C., et al., “Effects of Cardiac Autonomic Dysfunction on Mortality Risk in the Action to Control Cardiovascular Risk in Diabetes (ACCORD) Trial,” Diabetes Care 33, no. 7 (2010): 1578–1584. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Meyer M. L., Gotman N. M., Soliman E. Z., et al., “Association of Glucose Homeostasis Measures With Heart Rate Variability Among Hispanic/Latino Adults Without Diabetes: The Hispanic Community Health Study/Study of Latinos (HCHS/SOL),” Cardiovascular Diabetology 15, no. 1 (2016): 45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Casagrande S. S., Cowie C. C., Sosenko J. M., Mizokami‐Stout K., Boulton A. J. M., and Pop‐Busui R., “The Association Between Heart Rate and Glycemic Status in the National Health and Nutrition Examination Surveys,” Journal of Clinical Endocrinology and Metabolism 105, no. 3 (2020): e858–e870. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Emdin M., Gastaldelli A., Muscelli E., et al., “Hyperinsulinemia and Autonomic Nervous System Dysfunction in Obesity: Effects of Weight Loss,” Circulation 103, no. 4 (2001): 513–519. [DOI] [PubMed] [Google Scholar]
- 29. Festa A., D'Agostino R., Hales C. N., Mykkänen L., and Haffner S. M., “Heart Rate in Relation to Insulin Sensitivity and Insulin Secretion in Nondiabetic Subjects,” Diabetes Care 23, no. 5 (2000): 624–628. [DOI] [PubMed] [Google Scholar]
- 30. Poon A. K., Whitsel E. A., Heiss G., et al., “Insulin Resistance and Reduced Cardiac Autonomic Function in Older Adults: The Atherosclerosis Risk in Communities Study,” BMC Cardiovascular Disorders 20, no. 1 (2020): 217. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Watanabe K., Sekiya M., Tsuruoka T., et al., “Relationship Between Insulin Resistance and Cardiac Sympathetic Nervous Function in Essential Hypertension,” Journal of Hypertension 17, no. 8 (1999): 1161–1168. [DOI] [PubMed] [Google Scholar]
- 32. Dewanjee S., Das S., Das A. K., et al., “Molecular Mechanism of Diabetic Neuropathy and Its Pharmacotherapeutic Targets,” European Journal of Pharmacology 833 (2018): 472–523. [DOI] [PubMed] [Google Scholar]
- 33. Vinik A. I. and Ziegler D., “Diabetic Cardiovascular Autonomic Neuropathy,” Circulation 115, no. 3 (2007): 387–397. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1: Characteristics of the Diabetes Prevention Program Outcomes Study participants at Visit 10 by sex.
Table S2: Characteristics of the Diabetes Prevention Program Outcomes Study participants at Visit 10 by glycaemic status.
Table S3: Adjusted estimates of the association of glycemia‐and insulin‐related measures and cardiac autonomic function measures at year 10 of DPPOS.
Table S4: Associations of glycemia‐and insulin‐related measures and measures of cardiac autonomic function at year 10 of DPPOS by menopausal status.
Appendix S1: DPP, DPPOS I & DPPOS II Research Group (1996‐2013).
Data Availability Statement
In accordance with the NIH Public Access Policy, we continue to provide all manuscripts to PubMed Central including this manuscript DPP/DPPOS has provided the protocols and lifestyle and medication intervention manuals to the public through its public website (https://www.dppos.org). The DPPOS abides by the NIDDK data sharing policy and implementation guidance as required by the NIH/NIDDK (https://repository.niddk.nih.gov/studies/dppos/). New materials generated in the conduct of this research are available from authors upon reasonable request.