Cumulative effect of hyperglycemia and insulin resistance on cardiac dysfunction: The coronary artery risk development in young adults (CARDIA) study

Abstract

We investigated whether cumulative fasting glucose (FG) and insulin resistance (IR) over 20 years are associated with midlife cardiac dysfunction in young adults with or without type 2 diabetes (T2D)/prediabetes. We included young adults with T2D/prediabetes (N = 279) and matched euglycemic individuals (N = 514) who had repeated measures of fasting glucose (FG) and insulin resistance (IR) and echocardiography assessment from the Coronary Artery Risk Development in Young Adults study (CARDIA). We found that cumulative hyperglycemia is associated with midlife diastolic dysfunction in patients with early-onset T2D, and sustained IR negatively affects systolic and diastolic function regardless of T2D status.

Background

Chronic hyperglycemia and insulin resistance (IR) is associated with cardiac function and increase the risk of heart failure (HF). Hyperglycemia has been linked to an excess of advanced glycation end products, alterations in myocardial calcium handling, and modifications of the extracellular matrix. These changes contribute to increased ventricular stiffness, impaired relaxation, and ultimately diastolic dysfunction.^1–5 Meanwhile, IR affects substrate metabolism, promotes cellular injury, and contributes to microvascular dysfunction, as well as sympathetic and neurohormonal hyperactivity, leading to functional and structural changes in the heart.^1,2However, despite the biological link between chronic hyperglycemia and IR and the development of HF, epidemiologic evidence has been mostly extrapolated from older adult cohorts.^6,7 Early-onset type 2 diabetes (T2D, diagnosed <age 40) has been shown to be disproportionately associated with HF. Yet⁸,; little is known about the cumulative effect of hyperglycemia and IR on cardiac dysfunction among young adults with early-onset T2Dor those with prediabetes (pre-DM). We evaluated the effects of glucose and IR accumulated over 20 years and cardiac dysfunction among individuals with early-onset T2D or pre-DM and compared the effects among those with euglycemic status.

Methods

Data source: The Coronary Artery Risk Development in Young Adults (CARDIA) study is a prospective study of 5,115 healthy young adults enrolled in 1985 across 4 U.S. field centers.⁹ We used data from CVD-free non-pregnant participants with early-onset T2D/pre-DM or euglycemia (age-, sex-, and race-matched) (Supplemental Figure 1). CARDIA data were obtained from the National Heart, Lung, and Blood Institute, Biologic Specimen, and Data Repository Information Coordinating Center (BioLINCC).

Measures

Early-onset T2D was defined by self-reported diabetes or antidiabetic medication (every examination), fasting glucose (FG) ≥126 mg/dL (CARDIA years 0, 7, 10), or 2-hour oral glucose tolerance test (OGTT) ≥200 mg/dL (year 10) at or before year 10 (maximum age 40). Pre-DM was defined by FG 100 to 125 mg/dL or OGTT 140 to 199 mg/dL at or before year 10. Insulin users were excluded from the study.

Euglycemic controls (matched 1:2 to cases based on sex, age, and race at the ‘onset of hyperglycemia’) were selected based on the absence of elevated glucose levels throughout the study period and the availability of 4 or more risk factor measurements, which was necessary for applying a robust quadratic random-effects model (see Analysis section for details). Risk factors: Glucose and insulin assay methods were described previously.^3,10 The index of HOMA-IR was expressed by the formula: $\mathrm{f}\mathrm{a}\mathrm{s}\mathrm{t}\mathrm{i}\mathrm{n}\mathrm{g}\mathrm{i}\mathrm{n}\mathrm{s}\mathrm{u}\mathrm{l}\mathrm{i}\mathrm{n}(\mathrm{\mu }\mathrm{U}/\mathrm{d}\mathrm{L})\times \mathrm{f}\mathrm{a}\mathrm{s}\mathrm{t}\mathrm{i}\mathrm{n}\mathrm{g}\mathrm{g}\mathrm{l}\mathrm{u}\mathrm{c}\mathrm{o}\mathrm{s}\mathrm{e}(\mathrm{m}\mathrm{g}/\mathrm{d}\mathrm{L})/405$.⁷

Outcomes: At Year 25, doppler echocardiography and 2D-guided M-mode echocardiography were performed (Toshiba Medical Systems, Otawara, Japan) by trained sonographers. Left ventricular (LV) systolic function was estimated with ejection fraction and global longitudinal strain. The strain was calculated as the change in segment length relative to its end-diastolic length, and the peak systolic value was recorded. More negative values of strain indicate greater shortening or better function. Peak early diastolic mitral annular velocity (e’) was calculated from the average of the septal and lateral mitral annular velocities. E/e′ ratio was calculated as an index of LV filling pressures. ¹¹

Analysis

We used the area under the growth curve (AUC) between year 10 and year 25 derived from quadratic random-effects models of ≥4 repeated FG and HOMA-IR measures to estimate their cumulative burden. The AUC, computed from longitudinal growth curve models, uses multiple measurements of risk factors throughout time to reduce within-person variability and has been widely adopted for estimating risk burden in life-course cohorts.^2,12 Because the growth curves of the risk factors may be cubic models, a minimum of 4 measures were needed.¹² This model computes all participants’ maximum likelihood estimates of curve parameters. We selected the most parsimonious model using p-values (<0.05) of the independent variable (age). We calculated the AUCs using an integral calculus formula based on the fixed and random-effect parameters of the model during follow-up for each subject, dividing them by follow-up years to reflect the varying follow-up intervals between participants.^2,12 We computed AUC to characterize the overall risk factor from the first recorded risk factor to the last recorded risk factor.²

For individual i:

$$\begin{array}{l}{\mathrm{A}\mathrm{U}\mathrm{C}}_{\mathrm{i}}=\left[\left(\beta 0+\mathrm{b}{0}_{\mathrm{i}}+{\epsilon }_{\mathrm{i}},\right)\text{Vage}+\left(\beta 1+\mathrm{b}{1}_{\mathrm{i}}\right){\text{Vage}}^2\right.\\ \left.+(\beta 2+\mathrm{b}2\mathrm{i}){\text{Vage}}^3+\left(\beta 3+\mathrm{b}{3}_{\mathrm{i}}\right){\text{Vage}}^4\right]/\\ \left({\text{lastage}}_{\mathrm{i}}-{\text{firstage}}_{\mathrm{i}}\right)\end{array}$$

where $\beta =(\beta 0,\beta 1,\beta 2,\beta 3)$ is a vector of fixed-effect parameters, ${\mathrm{b}}_{\mathrm{i}}=\left(\mathrm{b}{0}_{\mathrm{i}},\mathrm{b}{1}_{\mathrm{i}},\mathrm{b}{2}_{\mathrm{i}},\mathrm{b}{3}_{\mathrm{i}}\right)$ is a vector of random-effect parameters, and ɛ_i is an unknown error for individual $\mathrm{i}$. We performed linear regressions to examine the associations of the total AUC of FG or HOMA-IR with cardiac functional outcomes, adjusting for sex, race, field center, midlife (year 25) age, education, smoking, oral antidiabetic medications, lipid-lowering medications, and resting heart rate. We further adjusted for the cumulative burden of hypertension and obesity (AUC of SBP or BMI) because of their confounding role in the association of interest.

Results

Among individuals with early-onset T2D/pre-DM, the mean onset age was 32 years. Sociodemographic and lifestyle factors were comparable in the 2 glycemic groups at baseline (year of diagnosis) (Table 1). At year 25, the mean age of 2 groups was 51 years. The mean of e’ was significantly lower, but e/e’ and longitudinal peak systolic strain were significantly higher in the early-onset DM/pre-DM group than in the euglycemic group (9.8 vs. 10.9, P = .001 and 8.4 vs.7.6 %, and −14.6 vs. −15.3, P ≤ .0002, respectively) at year 25. Total AUCs of FG and HOMA-IR were higher in the early-onset T2D/pre-DM than in the euglycemic group (Table 1).

Table 1.

Characteristics of participants with early-onset DM/PreDM versus euglycemia.

	Early-Onset DM/PreDM N = 279	Matched-Euglycemia N = 514	P-value
Year of diagnosis/matched euglycemia
Age, year, mean SD	32.3 (4.8)	31.9 (4.8)	0.32
Non-Hispanic Blacks, n %	129 (46)	235 (46)	0.88
Women, n %	181 (65)	339 (66)	0.76
≥ College education, n %	127 (46)	307 (60)	0.002
Current smoker, n %	67 (29)	110 (26)	0.08
Alcohol drinking (mL/day), mean SD	10.8 (23.1)	8.7 (15.5)	0.17
Antihypertensive medication, n %	11 (3.9)	6 (1.2)	0.0107
Lipid-lowering medication, n %	0 (0)	3 (0.6)	0.5557
Antidiabetic oral medication, n %	0 (0)	0 (0)
Metabolic risk factors,
Fasting glucose (mg/dL), mean SD	110 (41)	83 (7.4)	<0.0001
Log (fasting glucose), mean SD	4.7 (0.3)	4.4 (0.1)	<0.0001
HOMA-IR, mean SD	1.4 (1.5)	0.7 (0.3)	<0.0001
Log (HOMA-IR), mean SD	0.1 (0.6)	−0.4 (0.3)	<0.0001
Fasting insulin, mIU/L, mean SD	25.5 (33.2)	11.4 (6.6)	<0.0001
Log (fasting insulin), mean SD	2.9 (0.7)	2.3 (0.4)	<0.0001
BMI (kg/m2), mean SD	31.0 (8.0)	25.3 (5.6)	<0.0001
HDL-c (mg/dL), mean SD	46.9 (13.5)	55.1 (15.3)	<0.0001
LDL-c (mg/dL), mean SD	110.5 (31.7)	104.8 (28.5)	0.01
Triglycerides (mg/dL), mean, SD	96 (36)	77 (43)	<0.0001
Systolic blood pressure (mmHg), mean SD	111.3 (13.6)	106.6 (10.9)	<0.0001
Diastolic blood pressure (mmHg), mean SD	71.8 (10.8)	68.2 (9.4)	<0.0001
Year 25
Age, year, mean SD	50.6 (3.6)	50.4 (3.6)	0.45
Current smoker, n %	46 (20.4)	69 (15.7)	0.18
Antihypertensive medication, n %	132 (47.7)	100 (19.5)	<0.0001
Lipid-lowering medication, n %	99 (36.4)	38 (7.4)	<0.0001
Antidiabetic oral medications, n %	109 (39.5)	0 (0)	<0.0001
Metabolic risk factors,
Fasting glucose (mg/dL), mean SD	131.4 (57.2)	88.9 (6.3)	<0.0001
Log (fasting glucose), mean SD	4.8 (0.4)	4.5 (0.1)	<0.0001
HOMA-IR, mean SD	1.0 (0.6)	0.6 (0.3)	<0.0001
Log (HOMA-IR), mean SD	−0.1 (0.5)	−0.6 (0.4)	<0.0001
Fasting insulin, mIU/L, mean SD	15.9 (13.2)	8.6 (5.8)	<0.0001
Log (fasting insulin), mean SD	2.5 (0.8)	2.0 (0.6)	<0.0001
BMI (kg/m2), mean SD	33.9 (8.4)	28.8 (6.4)	<0.0001
HDL-c (mg/dL), mean SD	54 (17.5)	62.2 (17.7)	<0.0001
LDL-c (mg/dL), mean SD	104.7 (34.3)	113.1 (33.0)	0.001
Triglycerides (mg/dL), mean, SD	121 (58)	82 (32)	<0.0001
Systolic blood pressure (mmHg), mean SD	121.9 (16.0)	117.8 (15.2)	0.0004
Diastolic blood pressure (mmHg), mean SD	76.3 (10.4)	73.8 (11.4)	0.0024
LV outcomes
LV ejection fraction (%) mean, SD	69.6 (8.8)	69.7 (7.6)	0.92
LV ejection fraction < 50% n, %	5 (2.0)	3 (0.6)	0.13
e’ (cm/s) mean, SD	9.8 (2.3)	10.9 (2.2)	<0.0001
e/e’ mean, SD	8.4 (2.4)	7.6 (2.1))	<0.0001
Longitudinal peak systolic strain (s−1), mean, SD	−14.6 (2.5)	−15.3 (2.3)	0.0002
Resting heart rate (beats per minute), mean SD	69.1 (11.6)	64.4 (10.5)	<0.0001
Total AUC of main exposures
Log (fasting glucose), mean SD	4.7 (0.2)	4.4 (0.1)	<0.0001
Log (HOMA-IR), mean SD	0.03 (0.35)	−0.38 (0.23)	<0.0001

The mean onset age was estimated based on age at the first appearance of an elevated glucose marker (diabetes and prediabetes) or self-reported diagnosis or medication use (for T2D only).

Table 2 shows results from the adjusted linear regressions for the association between AUC of FG, HOMA-IR, and cardiac function outcomes. One standard deviation (SD) higher AUC of log (FG) (per 0.17) was inversely associated with higher e’ (β coefficient 0.18, 95% confidence interval −0.23, −0.05) in participants with early-onset T2D/pre-DM. One SD increment of AUC of log (HOMA-IR) (per 0.34) was inversely associated with e’ in both early-onset T2D/pre-DM (−0.2, −0.34, −0.06) and euglycemia groups (−0.18, −0.34, −0.04). Higher AUC of log (FG) was associated with higher e/e’ in the early T2D/pre-DM group but not in the euglycemia group (0.27, 0.12, 0.39 vs. −0.14, −0.45, 0.17, P-interaction 0.02). Higher AUC of log (HOMA-IR) was associated with Ell in both early T2D/pre-DM (0.16, 0.001, 0.32) and euglycemia (0.2, 0.04, 0.36) groups.

Table 2.

Adjusted β coefficients (95% CI) of 1 SD increase in AUCs of risk factors and LV function outcomes in participants with or without early-onset diabetes or pre-diabetes.

	Early T2D/PreDM	FDR-P	Euglycemia	FDR-P	P-interaction#
e’
Log (FG), per 0.17	−0.18 (−0.23, −0.05) **	0.007	−0.07 (−0.38, 0.24)	0.66	0.81
Log (HOMA-IR), per 0.34	−0.20 (−0.34, −0.06) **	0.007	−0.18 (−0.34, −0.04) *	0.04	0.91
e/e’
Log (FG), per 0.17	0.27 (0.12, 0.39) **	0.0006	−0.14 (−0.45, 0.17)	0.37	0.02 *
Log (HOMA-IR), per 0.34	0.12 (−0.05, 0.285)	0.23	0.12 (−0.03, 0.26)	0.35	0.37
Ejection Fraction
Log (FG), per 0.17	0.02 (−0.14, 0.17)	0.84	−0.05 (−0.38, 0.29)	0.79	0.69
Log (HOMA-IR), per 0.34	0.09 (−−0.09, 0.27)	0.46	−0.06 (−0.22, 0.1)	0.79	0.10
Longitudinal peak systolic strain
Log (FG), per 0.17	0.13 (−0.01, 0.27)	0.07	−0.13 (−0.47, 0.21)	0.44	0.41
Log (HOMA-IR), per 0.34	0.16 (0.01, 0.32) *	0.07	0.20 (0.04, 0.36) *	0.022	0.66

Adjusted for sex, race, education, field center, midlife age, obesity, smoking, alcohol drinking, lipid-lowering medications, antidiabetes medications, resting heart rate, and cumulative burden of hypertension (AUC of SBP) and obesity (AUC of BMI).

^*P<.05

^**P<.005

^#P for the interaction of AUC of fasting glucose or IR and glycemic status.

Discussion

This study characterized the longitudinal impact of glucose and IR on cardiac dysfunction among young adults with or without T2D/pre-DM. Despite the extensive literature on T2D and cardiac dysfunction, most evidence was derived from older adult cohorts.^6,7 Our study extended the literature on the early development of HF in T2D to young adult patients by linking the longitudinal glucose and IR burden based on robust repeated measures.

We found a significant positive association between cumulative glucose and diastolic dysfunction in participants with early T2D/pre-DM but not in healthy controls. Patients with T2D are predominantly affected by HF with preserved ejection fraction (HFpEF) resulting from diastolic dysfunction.⁴ Hyperglycemia has been linked to alterations in myocardial calcium handling and extracellular matrix modifications, leading to increased ventricular stiffness, impaired relaxation, and, consequently, diastolic dysfunction.⁴ We also found that cumulative HOMA-IR was significantly associated with diastolic dysfunction (e’) and systolic dysfunction (Ell) in participants regardless of early-onset T2D status. This appears to be in contrast with previous evidence indicating the excess risk of cardiac dysfunction associated with IR in those with diabetes. One possible explanation for this finding may be due to the imprecise role of HOMA-IR, rather than the gold standard, the Hyperinsulinemic-Euglycemic Clamp, in measuring IR. HOMA-IR only reflects IR under fasting conditions and doesn’t capture postprandial or dynamic insulin action. It is known that IR is associated with substrate metabolism, cellular injury, microvascular dysfunction, and sympathetic and neurohormonal hyperactivity, resulting in impaired structural, functional, and metabolic changes.⁷ Our findings support the concept of early and potentially aggressive intervention to maintain optimal glycemic control and overall metabolic health via healthy lifestyle and proper pharmacotherapies among those with early T2D/pre-DM to prevent premature HF.

Limitations

Because a minimum of 4 risk factor measures were needed for the quadratic random-effects models to examine the association of AUCs of glucose or IR with echocardiographic outcomes at year 25, we excluded 38 early-onset T2D cases after year 10. However, this potential selection bias may be minimal given the small number of exclusion cases and our use of the demographic-matched case-control design. Furthermore, due to data limitations, postprandial glycemia and HbA1c were not included in the diabetes definition, which may have limited the precision in identifying T2D cases. Additionally, IR estimated by the HOMA-IR model is not as accurate as “gold standard” tests such as the hyperinsulinemia-euglycemic clamp. However, it is a reliable and practical tool to estimate IR in large cohort studies. Another consideration regarding the AUCs is that burden typically refers to the range exceeding the ’gold standard’ for either glucose or HOMA-IR. However, by not restricting the AUC to only the range above the “gold standard”, we can analyze the cumulative effects of glucose and IR levels below standard thresholds. This approach provides valuable insights within the euglycemic group and enables a meaningful comparison to the diabetic group.

Conclusion

Our study shows that cumulative glucose and IR adversely affect midlife cardiac function in those with early-onset T2D/pre-DM. Potentially early and more aggressive risk factor management may mitigate future HF risk in young adults with T2D/pre-DM.

Supplementary Material

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.ahj.2025.09.008.