Three‐dimensional Speckle Tracking Imaging to Assess Subclinical Myocardial Mechanical Alterations in Patients With Type 2 Diabetes Mellitus Combined With Subclinical Hypothyroidism

ABSTRACT

Objective

This study aimed to evaluate subclinical myocardial mechanical alterations by utilizing three‐dimensional speckle tracking imaging (3D‐STI) in patients with type 2 diabetes mellitus (T2DM) combined with subclinical hypothyroidism (SCH), and evaluate its intergroup discriminatory capacity.

Methods

A total of 147 participants were enrolled and divided into four groups: prediabetes group (n = 37), isolated T2DM group (n = 35), T2DM combined with SCH group (n = 40), and healthy control group (n = 35). Conventional echocardiographic indices, three‐dimensional myocardial strain parameters, and ventriculo‐arterial coupling indicators were compared among groups. Receiver operating characteristic (ROC) curves were plotted to evaluate the ability of myocardial strain parameters to identify individuals presenting early subclinical left ventricular myocardial mechanical abnormalities within the study cohort. Partial correlation analysis was performed to explore the independent correlation between thyroid‐stimulating hormone (TSH) and left ventricular functional parameters after adjusting for confounding variables. Multiple linear regression analysis was applied to assess the independent effect of TSH on myocardial comprehensive index (MCI) after adjusting for confounding factors.

Results

Glycated hemoglobin (HbA1c) was elevated in the prediabetes, T2DM, and T2DM + SCH groups, while TSH was significantly higher in the T2DM + SCH group. Global longitudinal strain (GLS), left ventricular twist angle (LVtw), torsion (Tor), and myocardial comprehensive index (MCI) decreased progressively across the four groups. Effective arterial elastance (Ea) and ventriculo‐arterial coupling index (VAC) were increased in the T2DM and T2DM+SCH groups. ROC analysis demonstrated that MCI yielded the highest AUC (0.967) and sensitivity (97.30%) for prediabetes, and the optimal AUC (0.937), sensitivity (90.00%) and specificity (85.71%) for T2DM+SCH. Partial correlation analysis was performed after adjustment for age, sex, BMI, systolic blood pressure, diastolic blood pressure, triglycerides and HbA1c. The results revealed a significant negative correlation between TSH and MCI (r = −0.616, p < 0.01). Multivariate linear regression analysis was conducted with MCI as the dependent variable. After adjusting for age, sex, BMI, blood pressure, lipid and glucose metabolic parameters, TSH was independently negatively associated with MCI. When TG‐Ab and TPO‐Ab were added for exploratory analysis, elevated TSH remained independently associated with decreased MCI, while thyroid autoantibodies showed no independent effect. Partial correlation analysis indicated a moderate negative correlation between TSH and MCI (r = −0.407, p < 0.01).

Conclusion

3D‐STI enables the identification of early subclinical myocardial mechanical alterations in patients with T2DM combined with SCH.

Type 2 diabetes mellitus combined with subclinical hypothyroidism aggravates ventriculo‐arterial uncoupling. Speckle‐tracking echocardiography identifies elevated VAC and early subclinical myocardial mechanical alterations in patients with preserved LVEF. This dual metabolic comorbidity disrupts cardiac hemodynamic homeostasis.

1. Introduction

With the continuous transformation of residents’ lifestyles and dietary patterns, the global prevalence of diabetes has increased annually, among which type 2 diabetes mellitus (T2DM) accounts for more than 90% of all diabetic cases and has become a prevalent chronic metabolic disorder [1]. Cardiovascular complications represent the leading cause of mortality and disability in patients with T2DM and substantially affect long‐term prognosis [2]. Subclinical hypothyroidism (SCH) is an asymptomatic endocrine disorder identified solely via biochemical testing, affecting roughly 10% of the general population [3]. Emerging evidence indicates that thyroid dysfunction aggravates myocardial metabolic disorders and cardiovascular risk. The coexistence of T2DM and SCH produces synergistic cardiac injury, leading to latent myocardial mechanical alterations that is insensitive to conventional echocardiography. Therefore, early and accurate identification of subclinical myocardial mechanical alterations in T2DM patients with SCH is essential for timely clinical intervention and improved patient prognosis. Three‐dimensional speckle tracking imaging (3D‐STI) enables comprehensive quantitative evaluation of multidimensional myocardial motion and facilitates the early identification of subtle subclinical myocardial mechanical alterations [4]. Previous studies have characterized single‐dimensional myocardial mechanics in patients with T2DM and SCH. Nevertheless, few studies have evaluated multiple myocardial mechanical parameters simultaneously and further adjusted for thyroid autoantibodies to explore the independent effect of TSH. This replication and extension study aimed to explore the value of 3D‐STI in identifying subclinical myocardial mechanical alterations in patients with T2DM complicated by SCH, and to provide imaging evidence for research on cardiovascular subclinical changes in this population.

2. Subjects and Methods

2.1. Study Population

This study enrolled 147 participants who were recruited from outpatient departments of our hospital between January 2024 and March 2026. All subjects were divided into four groups: 37 patients with prediabetes (Pre‐DM group), 35 patients with type 2 diabetes mellitus (DM group), 40 patients with type 2 diabetes mellitus combined with subclinical hypothyroidism (DM+SCH group), and 35 healthy volunteers enrolled during the same period (control group). The diagnostic criteria for prediabetes and type 2 diabetes were based on the 2020 guidelines issued by the American Diabetes Association and the World Health Organization [5]. T2DM was diagnosed based on the following criteria: random plasma glucose ≥ 11.1 mmol/L, fasting plasma glucose (FPG) ≥ 7.0 mmol/L, 2‐hour oral glucose tolerance test (OGTT) plasma glucose ≥ 11.1 mmol/L, or glycated hemoglobin (HbA1c) ≥ 6.5%. Prediabetes was defined by any of the following laboratory thresholds: FPG 6.0–7.0 mmol/L, 2‐h OGTT plasma glucose 7.8–11.0 mmol/L, or HbA1c 5.7%–6.4%. SCH was defined by serum thyroid‐stimulating hormone (TSH) levels > 4.2 mIU/L together with normal free triiodothyronine (FT3) and free thyroxine (FT4). The serum laboratory reference intervals at our institution were 0.27–4.2 mIU/L for TSH, 3.5–6.5 pmol/L for FT3, and 11.5–22.7 pmol/L for FT4.

2.2. Inclusion and Exclusion Criteria

All enrolled participants had a left ventricular ejection fraction (LVEF) ≥ 50%. Exclusion criteria were as follows: a medical history of myocardial infarction, coronary artery disease, essential hypertension, congenital heart disease, severe valvular heart disease, primary cardiomyopathy, or severe arrhythmia; malignant tumors; severe renal failure (estimated glomerular filtration rate < 30 mL/min/1.73 m²); clinically diagnosed Hashimoto thyroiditis accompanied by overt hyperthyroidism or overt hypothyroidism; and recent heavy alcohol consumption, medication use, or thyroid surgery history that could interfere with thyroid function testing results. All procedures were approved by the institutional medical ethics committee, and written informed consent was obtained from all participants prior to enrollment.

2.3. General Data and Laboratory Measurements

General clinical data of all participants were recorded, including gender, age, heart rate, body mass index (BMI), and blood pressure. Fasting peripheral venous blood samples were collected and tested in accordance with standardized laboratory protocols. The measured laboratory indicators included FPG, HbA1c, total cholesterol, triglycerides, and thyroid function parameters (TSH, FT3, and FT4). In addition, thyroid autoantibodies, including thyroglobulin antibody (TG‐Ab) and thyroid peroxidase antibody (TPO‐Ab), were detected. TG‐Ab positivity was defined as a concentration > 110 IU/mL, and TPO‐Ab positivity was defined as a concentration > 40 IU/mL; values below these thresholds were considered negative.

2.4. Echocardiography

All echocardiographic examinations were performed using a GE Vivid E9 color Doppler ultrasound system equipped with M5S and 4V‐D cardiac probes, with a frequency range of 1.7–3.4 MHz and a frame rate > 25 frames per second. The integrated EchoPAC workstation was used for offline image post‐processing and parameter analysis. All participants were placed in the left lateral decubitus position with synchronous electrocardiographic limb lead monitoring and heart rate recording. Conventional echocardiographic parameters were measured in the parasternal long‐axis view, including left atrial end‐systolic diameter (LAD), left ventricular end‐diastolic diameter (LVEDD), interventricular septum end‐diastolic thickness (IVSd), left ventricular posterior wall end‐diastolic thickness (LVPWd), and cardiac output (CO). Left ventricular ejection fraction (LVEF) and stroke volume (SV) were calculated using the biplane Simpson's method. In the apical four‐chamber view, pulsed‐wave Doppler and tissue Doppler imaging were applied to acquire the early diastolic mitral inflow velocity (E) and early diastolic mitral annular velocity (e’), with data averaged over three consecutive cardiac cycles. Subsequently, patients were instructed to hold their breath, and three consecutive cardiac cycles of full‐volume four‐dimensional images were acquired. All images were imported into the EchoPAC workstation for quantitative analysis. The derived strain parameters included global longitudinal strain (GLS), global circumferential strain (GCS), left ventricular twist angle (LVtw), and torsion (Tor). The myocardial comprehensive index (MCI) was calculated according to the formula: MCI = GLS × LVtw. All strain values were analyzed using absolute values. All ultrasonic measurements were performed three times by a single experienced sonographer with more than 10 years of clinical experience, and the average values were adopted for statistical analysis. Systemic vascular resistance (SVR) was calculated as SVR = MAP × 80/CO. Mean arterial pressure (MAP) was defined as DBP + (SBP − DBP) / 3, a formula mathematically equivalent to MAP = (2 × DBP + SBP) / 3. The constant of 80 was applied for unit conversion to obtain SVR in dyn·s·cm^{− 5}. All participants were placed in the supine position at rest. Brachial systolic and diastolic blood pressures were measured three times consecutively, and the average values were recorded. Left ventricular end‐systolic pressure (Pes) was estimated as 0.9 times brachial systolic blood pressure. Based on the calculated Pes, we further computed effective arterial elastance (Ea) and left ventricular end‐systolic elastance (Ees). Ea was defined as Ea = Pes/SV, and Ees was noninvasively calculated via the formula Ees = Pes/ESV, with ESV indicating left ventricular end‐systolic volume. In addition, we assumed for the Ees estimation that the intercept of the pressure‐volume loop is zero. Such approximation is considered reasonable given the actual value is negligible in comparison to ESV and moreover cannot be determined noninvasively. Ventriculo‐arterial coupling index (VAC) was computed as the Ea/Ees ratio [6].

2.5. Statistical Analyses

The IBM‐SPSS Statistics, version 22 software (IBM, Armonk, New York) was used for statistical analysis. Continuous variables conforming to normal distribution were expressed as mean ± standard deviation and compared among the four groups using one‐way analysis of variance (ANOVA) with F‐test. Variables with non‐normal distribution were presented as median (25th, 75th percentile) and analyzed via the Kruskal‐Wallis H test. The least significant difference (LSD) post hoc test was applied for pairwise multiple comparisons. Receiver operating characteristic (ROC) curve analysis was performed to evaluate the intergroup discriminatory capacity of each parameter and identify the optimal cut‐off values, and the DeLong test was used for pairwise comparisons of AUCs. Partial correlation analysis was performed to explore the independent correlation between TSH and left ventricular functional parameters after adjusting for confounding variables. Multiple linear regression analysis was applied to assess the independent effect of TSH on MCI after adjusting for confounding factors. For reproducibility analysis, image data from 10 randomly selected patients were independently measured by two senior sonographers with more than 10 years of working experience to evaluate inter‐observer reliability. Meanwhile, intra‐observer reliability was assessed by repeated measurements of the same 10 cases performed by a single sonographer at a two‐week interval. P < 0.05 was considered significant.

3. Results

3.1. Comparison of General and Baseline Laboratory Data among Four Groups

Baseline clinical and laboratory parameters were compared across the four groups. The results demonstrated that BMI, systolic blood pressure, diastolic blood pressure, and triglyceride levels were significantly higher in the DM group and DM+SCH group than those in the control group (all p < 0.05). FPG levels in the DM and DM+SCH groups were markedly elevated compared with the control and Pre‐DM groups. HbA1c levels increased gradually in the order of control group, Pre‐DM group, and DM group, with significant intergroup differences (all p < 0.05). The DM+SCH group presented significantly higher levels of TSH, TG‐Ab, and TPO‐Ab than the control group, Pre‐DM group, and DM group (all p < 0.05). No significant difference in fasting blood glucose was observed between the Pre‐DM group and the control group. Additionally, fasting blood glucose and HbA1c levels were comparable between the DM group and DM+SCH group. There were no statistically significant differences in TSH, TG‐Ab, and TPO‐Ab levels among the control, Pre‐DM, and DM groups. Age, gender composition, heart rate, total cholesterol, FT3, and FT4 showed no significant differences among all four groups (Table 1, Figure 1).

TABLE 1.

Comparison of general clinical and laboratory parameters among the four groups.

Parameters	Control group (n = 35)	Pre‐DM group (n = 37)	DM group (n = 35)	DM + SCH group (n = 40)	F/χ2 value	p value
Male (n, %)	17(48.57%)	19(51.35%)	20(57.14%)	17(42.50%)	1.594	0.207
Age (years)	49.91 ± 10.91	50.51 ± 7.21	52.51 ± 10.14	51.77 ± 9.13	0.560	0.642
Duration of diabetes (years)	−	−	7.02 ± 4.83	6.89 ± 4.17	3.235	0.076
BMI (kg/m2)	23.89 ± 1.79	24.98 ± 2.24	25.95 ± 3.12 a	26.44 ± 2.80 a , b	7.189	<0.001
Heart rate (beats/min)	71.68 ± 5.37	73.18 ± 6.15	73.82 ± 5.63	73.02 ± 5.13	0.913	0.436
Systolic blood pressure (mmHg)	122.28 ± 5.94	123.08 ± 9.38	126.20 ± 8.31 a , b	127.90 ± 8.88 a , b	5.594	0.001
Diastolic blood pressure (mmHg)	73.37 ± 8.96	75.05 ± 9.09	79.80 ± 9.16 a , b	80.67 ± 8.14 a , b	5.989	0.001
FPG (mmol/L)	5.36 ± 0.65	6.08 ± 0.63	11.21 ± 3.87 a , b	10.38 ± 2.73 a , b	54.857	<0.001
HbA1c (%)	5.06 ± 0.42	5.84 ± 0.25 a	8.84 ± 2.07 a , b	8.62 ± 2.19 a , b	65.161	<0.001
Total cholesterol (mmol/L)	4.65 ± 0.74	4.96 ± 1.15	5.00 ± 0.73	5.04 ± 0.88	1.415	0.241
Triglyceride (mmol/L)	1.69 ± 0.649	2.10 ± 0.77	2.57 ± 1.69 a	2.63 ± 1.45 a	5.034	0.002
TSH (mIU/L)	2.18 ± 1.23	2.20 ± 1.19	2.16 ± 1.22	6.77 ± 2.63 a , b , c	68.870	< 0.001
FT3 (Pmol/L)	4.68 ± 0.81	4.67 ± 0.81	4.57 ± 0.79	4.38 ± 0.71	1.232	0.300
FT4 (Pmol/L)	14.08 ± 2.41	14.10 ± 2.37	14.13 ± 2.56	13.44 ± 1.82	0.834	0.477
TG‐Ab (IU/mL)	92.90 ± 19.89	93.71 ± 20.71	92.43 ± 22.54	284.89 ± 40.53 a , b , c	463.483	< 0.001
TPO‐Ab (IU/mL)	34.11 ± 6.65	34.82 ± 7.18	34.57 ± 7.27	313.13 ± 44.43 a , b , c	1311.480	< 0.001

Note: Values are provided as mean ± SD. Compared with the control group,.

^a P < 0.05; compared with Pre‐DM group,.

^b P < 0.05; compared with DM group,.

^c P < 0.05.

Abbreviations: BMI, Body Mass Index; FPG, Fasting plasma glucose; FT3, Free triiodothyronine; FT4, Free thyroxine; HbA1c, Glycated Hemoglobin; TG‐Ab, Thyroglobulin antibody; TPO‐Ab, Thyroid peroxidase antibody; TSH, Thyroid‐stimulating hormone.

FIGURE 1.

Comparison of TSH, FPG and HbA1c levels among four groups. Legend: Compared with the control group, ^* p < 0.05; compared with Pre‐DM group, ^† p < 0.05; compared with DM group, ^‡ p < 0.05.

3.2. Comparison of Conventional Echocardiographic Parameters

The LAD, IVSd and LVPWd in the DM group and DM+SCH group were significantly increased compared with the Pre‐DM group and control group (all p < 0.05). The E/e’ ratio in the Pre‐DM group, DM group and DM+SCH group was significantly higher than that in the control group (all p < 0.05). The DM+SCH group had significantly lower CO and higher vascular resistance than the control group, Pre‐DM group and DM group (all p < 0.05). In addition, LVEF in the DM+SCH group was significantly lower than that in the control group (p < 0.05). There was no significant difference in LVEDD among the four groups (Table 2).

TABLE 2.

Comparison of conventional echocardiographic parameters among the four groups.

Parameters	Control group (n = 35)	Pre‐DM group (n = 37)	DM group (n = 35)	DM + SCH group (n = 40)	F value	p value
LAD(mm)	30.24 ± 2.78	31.30 ± 2.94	33.42 ± 3.39 a , b	33.70 ± 3.30 a , b	10.513	< 0.001
LVEDD(mm)	45.87 ± 2.80	45.82 ± 3.09	45.25 ± 2.62	45.00 ± 2.49	0.905	0.440
IVSd(mm)	9.03 ± 0.65	8.90 ± 0.67	10.13 ± 1.70 a , b	10.16 ± 1.59 a , b	10.818	< 0.001
LVPWd(mm)	8.74 ± 0.66	8.82 ± 0.77	9.84 ± 1.41 a , b	9.87 ± 1.36 a , b	11.503	< 0.001
E/e'	9.45 ± 1.30	10.29 ± 1.81 a	11.01 ± 1.67 a	11.37 ± 1.36 a , b	10.910	< 0.001
LVEF(%)	63.51 ± 4.51	62.00 ± 4.28	61.57 ± 4.69	60.35 ± 4.01 a	3.324	0.022
CO(L/min)	5.32 ± 0.17	5.29 ± 0.17	5.22 ± 0.25 a	5.11 ± 0.21 a , b , c	8.531	< 0.001
Vascular resistance(dyn.s.cm−5)	1345.35 ± 115.42	1371.82 ± 130.36	1459.33 ± 100.61 a , b	1510.58 ± 98.39 a , b , c	17.631	< 0.001

Note: Values are provided as mean ± SD. Compared with the control group,.

^a P < 0.05; compared with Pre‐DM group,.

^b P < 0.05; compared with DM group,.

^c P < 0.05.

Abbreviations: CO, cardiac output; e', average tissue Doppler diastolic velocity of the mitral annulus (septal and lateral segments); E, peak early diastolic mitral inflow velocity; IVSd, interventricular septum end‐diastolic thickness;LAD, left atrial end‐systolic diameter; LVEDD, left ventricular end‐diastolic diameter; LVEF, left ventricular ejection fraction; LVPWd, left ventricular posterior wall end‐diastolic thickness.

3.3. Comparison of Speckle Tracking Imaging Parameters

GLS, LVtw, Tor and MCI decreased sequentially in the control group, Pre‐DM group, DM group and DM+SCH group, with significant statistical differences among the four groups (all p < 0.05). GCS in the DM group and DM+SCH group was significantly lower than that in the control group and Pre‐DM group (all p < 0.05) (Table 3, Figure 2).

TABLE 3.

Comparison of 3D‐STI parameters among the four groups.

Parameters	Control group (n = 35)	Pre‐DM group (n = 37)	DM group (n = 35)	DM + SCH group (n = 40)	F value	p value
GLS(%)	20.91 ± 1.67	18.97 ± 1.90 a	17.68 ± 1.74 a , b	16.78 ± 1.61 a , b , c	39.262	<0.001
GCS(%)	21.28 ± 1.71	20.43 ± 2.34	18.09 ± 2.33 a , b	17.85 ± 2.17 a , b	22.995	<0.001
LVtw(°)	14.03 ± 1.35	12.48 ± 1.26 a	10.91 ± 2.07 a , b	10.04 ± 1.66 a , b , c	43.801	<0.001
Tor(°/cm)	1.95 ± 0.34	1.46 ± 0.29 a	1.23 ± 0.30 a , b	1.06 ± 0.20 a , b , c	67.588	<0.001
MCI(%×°)	292.82 ± 30.38	236.11 ± 27.06 a	193.45 ± 43.71 a , b	173.48 ± 37.38 a , b , c	82.206	<0.001

Note: Values are provided as mean ± SD. Compared with the control group,.

^a P < 0.05; compared with Pre‐DM group,.

^b P < 0.05; compared with DM group,.

^c P < 0.05.

Abbreviations: GCS, global circumferential strain;GLS, global longitudinal strain; LVtw, left ventricular twist angle; MCI, myocardial comprehensive index; Tor, torsion.

FIGURE 2.

The GLS and 17‐segment bull's‐eye map in the four groups. Legend: A, Control group; B, Pre‐DM group; C, DM group; D, DM+SCH group.

3.4. Comparison of Ventriculo‐Arterial Coupling Parameters

The Ea and VAC values in the DM group and DM+SCH group were significantly higher than those in the control group and Pre‐DM group (all p < 0.05). Additionally, the DM+SCH group exhibited significantly higher VAC levels than the DM group (p < 0.05). There was no significant difference in Ees values among the four groups (Table 4).

TABLE 4.

Comparison of ventriculo‐arterial coupling parameters among the four groups.

Parameters	Control group (n = 35)	Pre‐DM group (n = 37)	DM group (n = 35)	DM + SCH group (n = 40)	F value	p value
Ea(mmHg/ml)	1.86 ± 0.59	1.91 ± 0.51	2.19 ± 0.39 a , b	2.38 ± 0.37 a , b	9.922	< 0.001
Ees(mmHg/ml)	2.30 ± 0.68	2.33 ± 0.65	2.38 ± 0.66	2.26 ± 0.52	0.226	0.878
VAC	0.81 ± 0.11	0.85 ± 0.19	0.96 ± 0.26 a , b	1.09 ± 0.24 a , b , c	13.965	< 0.001

Note: Values are provided as mean ± SD. Compared with the control group,.

^a P < 0.05; compared with Pre‐DM group,.

^b P < 0.05; compared with DM group,.

^c P < 0.05.

Abbreviations: Ea, effective arterial elastance; Ees, left ventricular end‐systolic elastance; VAC, ventriculo‐arterial coupling index.

3.5. ROC Curve Analysis of 3D‐STI Parameters

The AUC values of GLS, LVtw, Tor and MCI all exceeded 0.800; each parameter had sensitivity and specificity greater than 60.00%, along with a Youden's index above 0.5. For individuals with prediabetes, MCI demonstrated the largest AUC (0.967) and highest sensitivity (97.30%), whereas Tor presented the optimal specificity (94.29%). In patients with T2DM complicated by SCH, MCI achieved the maximum AUC (0.937), sensitivity (90.00%) and specificity (85.71%), with Tor exhibiting an identical peak specificity (85.71%). Pairwise AUC comparisons across all parameters in prediabetic subjects and patients with T2DM complicated by SCH were conducted via the DeLong test. All 3D‐STI‐derived myocardial mechanical indices (GLS, LVtw, Tor and MCI) exhibited significantly stronger discriminatory capacity than conventional LVEF (all p < 0.01). There were no statistically significant AUC differences among GLS, LVtw and Tor (all p > 0.05). Subsequent pairwise comparisons indicated that MCI yielded a significantly larger AUC than GLS and LVtw (both p < 0.05), while the AUC difference between MCI and Tor was non‐significant (p > 0.05) (Table 5 and 6, Figure 3).

TABLE 5.

ROC curve analysis of 3D‐STI parameters for subclinical myocardial mechanical abnormalities in Pre‐DM patients.

Parameters	Threshold (%)	AUC	Sensitivity (%)	Specificity (%)	Youden index
LVEF(%)	67	0.591	86.49	31.43	0.18
GLS(%)	19	0.841	83.78	82.86	0.67
LVtw(°)	12.8	0.905	86.49	88.57	0.75
Tor(°/cm)	1.6	0.908	89.19	94.29	0.83
MCI(%×°)	262.2	0.967	97.30	88.57	0.86

Note: GLS, global longitudinal strain; GCS, global circumferential strain; LVtw, left ventricular twist angle; Tor, torsion; MCI, myocardial comprehensive index.

TABLE 6.

ROC curve analysis of 3D‐STI parameters for subclinical myocardial mechanical abnormalities in DM + SCH patients.

Parameters	Threshold (%)	AUC	Sensitivity (%)	Specificity (%)	Youden index
LVEF(%)	63	0.572	82.50	40.00	0.23
GLS(%)	17	0.844	85.00	65.71	0.51
LVtw(°)	11.0	0.861	87.50	74.29	0.62
Tor(°/cm)	1.1	0.875	82.50	85.71	0.68
MCI(%×°)	183.6	0.937	90.00	85.71	0.76

Note: GLS, global longitudinal strain; GCS, global circumferential strain; LVtw, left ventricular twist angle; Tor, torsion; MCI, myocardial comprehensive index.

FIGURE 3.

The ROC curves of echocardiographic variables. Legend: A, Pre‐DM group versus Control group; B, DM+SCH group versus DM group. LVEF, left ventricular ejection fraction; GLS, global longitudinal strain; LVtw, left ventricular twist angle; Tor, torsion; MCI, myocardial comprehensive index.

3.6. Partial Correlations between TSH Level, 3D‐STI Parameters and Ventriculo‐Arterial Coupling Parameters

Partial correlation analyses were performed after adjustment for age, sex, BMI, systolic blood pressure, diastolic blood pressure, triglycerides and HbA1c. The results revealed that TSH was significantly negatively correlated with GLS, LVtw, Tor and MCI, and significantly positively correlated with VAC (r = −0.545, −0.562, −0.591, −0.616, 0.557, all P < 0.01). The 95% confidence bands of the fitted lines in partial residual plots did not cross the zero line (Figure 4).

FIGURE 4.

Composite partial residual scatter plots showing adjusted linear associations between TSH and five myocardial mechanical indicators, including GLS, LVtw, Tor, MCI and VAC.

Partial correlation analysis was performed after adjustment for age, sex, BMI, systolic blood pressure, diastolic blood pressure, triglycerides, HbA1c, TG‐Ab and TPO‐Ab. TSH exhibited a weak negative correlation with GLS (r = −0.236, p > 0.05) and a weak positive correlation with VAC (r = 0.182, p > 0.05), both without statistical significance. Moderate negative correlations were observed between TSH and LVtw, Tor, MCI (r = −0.357, −0.385, −0.407, all p < 0.01). Residual scatter plots revealed consistent linear trends between residual TSH and residual values of all myocardial mechanical indicators. The 95% confidence intervals of regression lines for LVtw, Tor and MCI did not cross the zero line, verifying their significant correlations. In contrast, the 95% confidence bands for GLS and VAC included zero, indicating non‐significant associations, which was consistent with the partial correlation results (Figure 5).

FIGURE 5.

Composite partial residual scatter plots showing adjusted linear associations between TSH and five myocardial mechanical indicators, including GLS, LVtw, Tor, MCI and VAC.

Legend: All analyses were adjusted for age, sex, BMI, systolic blood pressure, diastolic blood pressure, HbA1c and triglycerides. Each dot represents the residual value of an individual participant after eliminating confounding factors. The solid lines indicate fitted linear trends, and the dotted lines represent the 95% confidence bands for the conditional mean. All axes show residual values while retaining the original measurement units.

Legend: All analyses were adjusted for age, sex, BMI, systolic blood pressure, diastolic blood pressure, HbA1c, triglycerides, TG‐Ab and TPO‐Ab. Each dot represents the residual value of an individual participant after adjustment for confounding factors. Solid lines indicate fitted linear trends, and dotted lines represent the 95% confidence bands for the conditional mean. All axes show residual values while retaining the original measurement units.

3.7. Multiple Linear Regression Analysis of Influencing Factors for MCI

Multivariate linear regression analysis was performed to identify independent influencing factors of MCI. MCI was set as the dependent variable, and age, sex, BMI, systolic blood pressure, diastolic blood pressure, triglycerides, HbA1c and TSH were incorporated into the regression model. The results demonstrated that BMI, systolic blood pressure, triglycerides, HbA1c and TSH were independently associated with MCI (β = −0.168, −0.164, −0.159, −0.260, −0.397, all p < 0.05). No independent correlations with MCI were observed for age, sex and diastolic blood pressure (all p > 0.05). After adjustment for age, sex, BMI, blood pressure, lipid and glycemic metabolic parameters, TSH level remained independently negatively correlated with MCI (Table 7).

TABLE 7.

Multiple Linear Regression Analysis of Influencing Factors for MCI.

Variable	B	SE	Standardized β	t value	p value
Age	0.348	0.186	0.073	1.870	0.066
Sex	−1.365	3.435	−0.015	−0.397	0.692
BMI	−2.621	0.722	−0.168	−3.632	0.001
systolic blood pressure	−1.001	0.401	−0.164	−2.498	0.015
diastolic blood pressure	−0.352	0.242	−0.066	−1.453	0.151
triglycerides	−6.054	1.965	−0.260	−3.080	0.003
HbA1c	−4.681	1.950	−0.159	−2.400	0.019
TSH	−5.843	0.652	−0.397	−8.959	< 0.001

Note: Dependent variable is MCI. Age, sex, BMI, systolic blood pressure, diastolic blood pressure, triglycerides, HbA1c and TSH were entered into the multivariate linear regression model. P < 0.05 was considered statistically significant.

Multiple linear regression analysis was performed to identify independent determinants of MCI. MCI was set as the dependent variable, and age, sex, BMI, systolic blood pressure, diastolic blood pressure, triglycerides, HbA1c, TSH, TG‐Ab and TPO‐Ab were entered into the regression model. The results demonstrated that BMI, systolic blood pressure, triglycerides, HbA1c and TSH entered the regression equation (β = −0.161, −0.172, −0.283, −0.164, −0.334, all p < 0.05). No independent correlations between age, sex, diastolic blood pressure, TG‐Ab, TPO‐Ab and MCI were observed (p > 0.05). After adjustment for confounders including age, sex, BMI, blood pressure, lipid parameters, glucose metabolic indicators and thyroid autoantibody levels, TSH was independently negatively associated with MCI (Table 8).

TABLE 8.

Multiple Linear Regression Analysis of Influencing Factors for MCI.

Variable	B	SE	Standardized β	t value	p value
Age	0.359	0.190	0.075	1.884	0.064
Sex	−1.348	3.504	−0.015	−0.385	0.702
BMI	−2.515	0.774	−0.161	−3.248	0.002
systolic blood pressure	−1.052	0.425	−0.172	−2.478	0.016
diastolic blood pressure	−0.343	0.250	−0.064	−1.372	0.175
triglycerides	−6.597	2.384	−0.283	−2.767	0.007
HbA1c	−4.824	2.031	−0.164	−2.375	0.021
TSH	−4.921	2.407	−0.334	−2.044	0.045
TG‐Ab	−0.004	0.051	−0.010	−0.085	0.933
TPO‐Ab	−0.015	0.059	−0.047	−0.251	0.802

Note: Dependent variable is MCI. Age, sex, BMI, systolic blood pressure, diastolic blood pressure, triglycerides, HbA1c, TSH, TG‐Ab and TPO‐Ab were entered into the multivariate linear regression model. P < 0.05 was considered statistically significant.

3.8. Reproducibility Test of GLS, LVtw, MCI and VAC

Intra‐ and inter‐observer reproducibility analysis showed that the intra‐observer ICCs of GLS, LVtw, MCI, and VAC were 0.828, 0.818, 0.816, and 0.823, respectively, and the corresponding inter‐observer ICCs were 0.818, 0.816, 0.810, and 0.812. All ICC values were greater than 0.8, indicating excellent test‐retest reliability and reproducibility of these myocardial mechanical parameters (Figure 6).

FIGURE 6.

GLS, LVtw, MCI and VAC repeatability test (Bland–Altman diagram).

4. Discussion

Diabetes mellitus and thyroid dysfunction are commonly comorbid endocrine disorders that jointly contribute to cardiac structural and functional impairment [7]. Early detection of subclinical cardiac mechanical alterations in these patients is critical for timely clinical intervention and favorable long‐term prognosis. The 3D‐STI technique eliminates the plane limitation of two‐dimensional ultrasound and enables comprehensive analysis of left ventricular wall motion to accurately reflect real ventricular movement, which makes it a sensitive modality for detecting early subtle subclinical myocardial mechanical alterations [8]. Previous studies have characterized single‐dimensional myocardial mechanics in patients with T2DM and SCH. Nevertheless, few studies have evaluated multiple myocardial mechanical parameters simultaneously and further adjusted for thyroid autoantibodies to explore the independent effect of TSH. Based on functional imaging indices, This replication and extension study aimed to evaluate the utility of 3D‐STI for identifying subclinical left ventricular myocardial mechanical alterations in patients with T2DM complicated by SCH, and to provide objective imaging evidence for relevant cardiovascular research.

In the present study, patients in the DM and DM+SCH groups showed significantly greater LAD, IVSd, and LVPWd compared with subjects in the Pre‐DM and control groups (all p < 0.05). Additionally, the E/e’ ratio was significantly higher in the Pre‐DM, DM, and DM+SCH groups than in the control group (all p < 0.05). These findings indicated that elevated blood glucose levels and abnormal thyroid function could induce left ventricular remodeling, manifested as increased ventricular wall thickness and impaired left ventricular diastolic function. Compared with the control group, the DM+SCH group exhibited a slight yet statistically significant reduction in LVEF, while all LVEF values remained within the normal range. Reliance solely on LVEF for left ventricular function monitoring fails to detect early subclinical myocardial damage [9].

In the present study, the Pre‐DM group showed significantly lower GLS, LVtw, Tor and MCI values compared with the control group (all p < 0.05), suggesting that subclinical left ventricular mechanical alterations may emerge as early as the prediabetic stage [10]. Moreover, GLS and LVtw gradually decreased sequentially in the Pre‐DM, DM, and DM+SCH groups (all p < 0.05). These findings suggest that subclinical abnormalities of left ventricular myocardial mechanics can be observed in patients with T2DM complicated by SCH; subclinical alterations in systolic mechanics may emerge prior to or concurrently with diastolic mechanical changes [11]. This phenomenon can be explained by multiple interacting pathological pathways. Persistent hyperglycemia induces cardiac autonomic disturbance and sustained sympathetic activation, resulting in myocardial microvascular rarefaction, chronic myocardial hypoperfusion, oxidative injury of subendocardial cardiomyocytes, and accumulation of advanced glycation end products within the myocardial interstitium. The SCH elevates systemic oxidative stress, reduces nitric oxide synthesis and triggers endothelial dysfunction, further compromising myocardial perfusion and aggravating myocardial inflammation. Together with diabetes‐related insulin resistance, these pathological insults mutually amplify one another through synergistic signaling pathways and may ultimately aggravate subclinical myocardial mechanical alterations, including impaired longitudinal deformation and diminished torsional function. Such superimposed effects are consistent with our findings: patients with combined T2DM and SCH exhibited more prominent reductions in GLS, LVtw, Tor and MCI, characterized by concurrent involvement of longitudinal deformation and ventricular torsional mechanics [12, 13].

To minimize the positional deviation of LVtw caused by the absence of distinct anatomical references at the left ventricular apex, Tor was introduced in this study. Our results demonstrated that Tor values were progressively reduced in the Pre‐DM, DM, and DM+SCH groups compared with the control group, suggesting that Tor serves as a reliable indicator for evaluating myocardial mechanical alterations in patients with prediabetes, type 2 diabetes mellitus, and type 2 diabetes mellitus complicated with SCH. This finding concerning single‐dimensional strain parameters has been validated in previous investigations [14]. Meanwhile, our further observations indicated that left ventricular myocardial motion consists of multiple motion patterns and cannot be fully characterized by a single dimension. GLS and LVtw represent two independent systems of left ventricular systolic mechanics. GLS reflects the mechanical status of the longitudinal myocardium (subendocardial muscle bundles), whereas LVtw characterizes basal‐apical torsion and transmural myocardial synchrony (helical muscle bundles). Accordingly, we constructed a novel composite parameter MCI derived from GLS and LVtw. The multiplicative form endows MCI with synergistic characteristics: MCI decreases when alterations occur in either longitudinal deformation or torsional mechanics. Concurrent changes in both components lead to an amplified reduction of this index. Therefore, MCI can sensitively capture multidimensional concomitant subclinical myocardial mechanical alterations, improve the performance for identifying early occult myocardial abnormalities, and avoid subjectivity introduced by artificial weight assignment. This feature matches the pathological characteristics of multifocal and diffuse myocardial involvement in patients with T2DM combined with SCH. The MCI values decreased gradually across the control, Pre‐DM, DM, and DM+SCH groups with good repeatability. ROC curve analysis revealed that MCI achieved the highest AUC (0.967) and sensitivity (97.30%) in individuals with prediabetes. Among patients with T2DM complicated by SCH, MCI exhibited the maximum AUC (0.937), sensitivity (90.00%) and specificity (85.71%). Pairwise AUC comparisons using the DeLong test demonstrated that MCI possessed superior discriminatory capacity compared with GLS and LVtw, and the differences reached statistical significance. Collectively, intergroup comparisons of 3D‐STI parameters among the four cohorts verified that T2DM combined with SCH exacerbates unfavorable myocardial mechanical alterations. As a novel and sensitive composite imaging index, MCI can detect early subtle subclinical myocardial mechanical alterations characteristic of this metabolically high‐risk population, which provides imaging references for exploring the features of myocardial involvement and subsequent cardiovascular investigations in this cohort [15].

Ventriculo‐arterial coupling reflects cardiac adaptive responsiveness to arterial hemodynamics and functional status under physiological and pathological conditions, and is a robust prognostic biomarker for all‐cause mortality. In this study, Ea and VAC gradually increased across the control, Pre‐DM, DM, and DM+SCH groups. Specifically, Ea and VAC were significantly higher in the DM and DM+SCH groups relative to the control and Pre‐DM groups, and the DM+SCH group showed further elevated VAC compared with the DM group (all p < 0.05). These results indicate that elevated blood glucose promotes arterial elastic impairment, resulting in the premature arrival of reflected waves in the aorta during left ventricular systole and a subsequent increase in ventricular afterload, which triggers subclinical myocardial mechanical alterations. Concurrent SCH exerts synergistic adverse effects and further aggravates arterial elastic impairment and hemodynamic disturbances [16]. Accordingly, the ventriculo‐arterial uncoupling observed in patients with combined T2DM and SCH may stem from the joint effects of reduced arterial elasticity and subclinical left ventricular myocardial mechanical alterations, which is consistent with the significantly elevated VAC levels identified in this group in the present study [17]. The changes in CO and vascular resistance observed in the present study further support this viewpoint.

Baseline data of the present study revealed abnormal glucose and lipid metabolism, blood pressure and body mass in the DM and DM+SCH groups compared with the control group. Partial correlation analysis and multivariate linear regression were jointly adopted to systematically explore the associations between TSH and multiple left ventricular myocardial mechanical indicators after adjustment for confounding factors including age, sex, BMI, systolic blood pressure, diastolic blood pressure, triglycerides and HbA1c. Partial correlation analysis revealed that TSH was significantly negatively correlated with GLS, LVtw, Tor and MCI, and significantly positively correlated with VAC. The 95% confidence bands of the fitted lines in partial residual plots did not cross the zero line, verifying that these linear associations were stable, independent and reliable. Multivariate regression further quantified the independent effects. The results showed that TSH exerted the strongest standardized influence on MCI (β = −0.397, p < 0.001, SE = 0.681), indicating precise effect estimation and better stability of findings relative to other metabolic and blood pressure‐related indicators. Mechanistically, worsening SCH accompanied by sustained TSH elevation exerts superimposed effects when coexisting with hyperglycemia. Together, they increase peripheral vascular resistance and arterial load and mutually reinforce insulin resistance, forming a vicious cycle. Under such conditions, ventricular ejection requires overcoming elevated arterial resistance, thereby triggering ventriculo‐arterial mismatch and subclinical myocardial mechanical alterations. Accordingly, early identification of selective subclinical myocardial mechanical changes synergistically mediated by TSH and hyperglycemia in patients with type 2 diabetes mellitus combined with SCH can provide important imaging evidence for precisely evaluating early cardiovascular risks in this population [18]. In the present study, patients with overt thyroid dysfunction caused by Hashimoto's thyroiditis were strictly excluded. Nevertheless, elevated TSH, TG‐Ab and TPO‐Ab were specifically observed in the DM+SCH group. To further explore the influence of thyroid autoantibodies, TG‐Ab and TPO‐Ab were additionally entered as independent variables into the model on the basis of adjustment for metabolic risk factors. This exploratory analysis revealed that elevated TSH remained independently associated with reduced MCI, whereas thyroid autoantibodies exerted no independent effects. Partial correlation analysis demonstrated moderate negative correlations between TSH and LVtw, Tor and MCI; the 95% confidence bands of fitted lines in partial residual plots did not cross the zero line. Multivariate linear regression indicated that TSH had the strongest standardized effect on MCI (β = −0.334, p < 0.05). Consistent findings derived from these two statistical approaches support an association between elevated TSH and subclinical left ventricular myocardial mechanical alterations after adjustment for multiple confounders. In addition, after full adjustment for all confounding factors, no significant associations were detected between TSH and GLS or VAC. These findings suggest that subclinical myocardial injury exhibits a selective pattern in individuals with type 2 diabetes mellitus and elevated TSH: abnormalities in left ventricular torsional mechanics may emerge earlier than those in GLS. When the synchronization of transmural myocardium (helical myofibers) is disrupted, compensatory mechanisms in longitudinal myocardium can maintain relatively stable GLS values, which may explain why only torsional parameters show moderate correlations with TSH in the present study. Inferred from the present findings, TSH‐mediated myocardial injury is not homogeneous. Left ventricular torsional indices (LVtw, Tor) and composite myocardial index (MCI) are more sensitive to this pathological stimulus, whereas no evident association was observed for GLS. This implies that dysfunction of transmural myocardial synchronization (helical myofibers) occurs at an earlier stage. Nevertheless, given the limited sample size and the inclusion of 10 independent variables in the model, the study lacks sufficient power for stable multivariable estimation and carries a risk of overfitting. Accordingly, inferences derived from this multivariate linear regression model constitute exploratory observations and require further validation in subsequent investigations.

5. Limitations

Several limitations of this study should be noted. First, although major metabolic confounders were adjusted for in the regression models, confounding effects originating from thyroid autoimmune inflammation and residual confounding arising from unmeasured variables could not be fully eliminated. In addition, no a priori power analysis was conducted. Given the inclusion of 10 independent variables in the multivariate model, the relatively limited sample size may result in inadequate statistical power. Further prospective investigations are required to disentangle these associations. Second, peripheral vascular resistance and ventriculo‐arterial coupling indices were indirectly calculated using non‐invasive echocardiographic parameters, without invasive hemodynamic monitoring for validation, which may introduce measurement bias. Third, this is a cross‐sectional observational study and cannot establish causal relationships. Large‐sample prospective cohort studies or interventional basic experiments are needed to further validate the present findings in future work. Fourth, multiple interrelated myocardial mechanical and hemodynamic indicators were analyzed in this study. No correction for multiple comparisons was performed, which may elevate the risk of Type I error.

6. Conclusion

3D‐STI can effectively identify subtle subclinical myocardial mechanical alterations in patients with T2DM complicated by SCH. The findings provide objective imaging evidence for early recognition of subclinical cardiac changes in this high‐risk population, clarify the synergistic effects of hyperglycemia and elevated TSH on myocardial mechanics, and lay a foundation for further prospective exploration of cardiovascular prognostic value in this population.

Conflicts of Interest

All authors have completed the conflict of interest disclosure form and declare that there are no other relationships or activities that may affect the submitted work.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.