A total of 473 hypertensive patients were enrolled between January 2020 and December 2024, all of whom underwent CMR examinations. Hypertension was defined as a systolic blood pressure (SBP) > 140 mmHg and/or diastolic blood pressure (DBP) > 90 mmHg based on at least two office readings (14). The exclusion criteria were as follows (Figure 1): heart failure with left ventricular ejection fraction (LVEF) < 50% (n = 194), hypertrophic cardiomyopathy (n = 29), restrictive cardiomyopathy (n = 13), dilated cardiomyopathy (n = 13), severe valvular heart disease (n = 8), myocardial infarction (n = 19), myocarditis (n = 27), malignant tumors (n = 11), pericarditis (n = 4), atrial fibrillation (n = 10) and inadequate image quality (n = 15). Finally, 130 HTN patients were included in the study and divided into two groups on the basis of the absence or presence of T2DM: HTN-only (n = 74) and HTN-T2DM (n = 56). The diagnosis of T2DM was based on the current American Diabetes Association guideline recommendations (15). The control group included 42 age- and sex-matched individuals who underwent CMR for the evaluation of non-specific symptoms (e.g., chest pain or tightness) and were found to have entirely normal cardiac structure and function.

Patient clinical data were collected from medical records. The following covariates were included in the analysis: age, sex, body mass index (BMI), heart rate (derived from the ECG recorded during the CMR scan), systolic and diastolic blood pressure (SBP/DBP, retrieved from routine clinical records closest to the date of the CMR examination), hypertension classification, duration of hypertension or diabetes, glycated haemoglobin A1c (HbA1c), fasting blood glucose (FBG), total cholesterol (TC), triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C). The triglyceride‒glucose (TyG) index serves as a biomarker of insulin resistance and predicts heart failure risk. It is calculated as (16):TyG=ln[(TG(mgdL)*FBG(mgdL))2]

All CMR examinations were performed via a 3.0 Tesla scanner (MAGNETOM Prisma, Siemens Healthineers, Erlangen, Germany) equipped with an 18-channel body phased-array coil. Participants were positioned supine, and ECG-gated acquisitions were performed during breath‒hold intervals to minimize motion artifacts. Cine imaging was acquired via a balanced steady-state free precession sequence with the following parameters: TR/TE = 40.35/1.22 ms; slice thickness = 8.0 mm (gap: 2 mm); FOV = 250 × 300 mm²; matrix = 208 × 139; flip angle = 40°. The protocol comprised cine imaging in short-axis, 2-, 3-, and 4-chamber views.

The CMR images of 30 HTN patients and 15 controls were randomly selected to assess interobserver variability (ZL vs. YC). The intraobserver measurements were performed by Dr. ZL in a random order, with an interval of at least one month between assessments, to minimize recall bias. Two radiologists were blinded to the clinical information of the patients with HTN.

The normality of continuous variables was assessed via the Shapiro‒Wilk test. Accordingly, normally distributed continuous variables are presented as the mean ± standard deviation and non-normally distributed variables as the median (interquartile range, IQR). Categorical variables are expressed as frequencies (percentages). One-way analysis of variance (ANOVA) or the Kruskal‒Wallis rank sum test was used to compare the baseline characteristics among groups, and Tukey's HSD test or Dunn's test with Holm correction was applied for post-hoc multiple comparisons when the overall test was significant. The chi-square test was used to analyse differences in categorical variables. Groupwise comparisons of CMR-derived parameters were performed by analysis of covariance (ANCOVA), adjusted for age, sex, BMI, heart rate, and SBP, with multiple comparisons corrected using the Holm-Bonferroni method. Spearman's rank correlation (ρ) was performed across the entire cohort to examine associations between key metabolic indices (diabetes duration, HbA1c) and cardiac function, as well as between LACI and left atrial/ventricular strain parameters. Multivariable regression analyses were performed in the overall population to identify independent determinants of atrial and ventricular function and atrioventricular coupling status after covariate adjustment, and to assess the independent effect of study group. Analyses were also conducted in hypertensive patients alone to determine independent predictors of these cardiac parameters, focusing on comorbid diabetes and metabolic profiles. Collinearity diagnostics confirmed no severe multicollinearity among variables, with all variance inflation factors (VIF) < 3. Intraclass correlation coefficients [ICCs; two-way mixed-effects model, ICC (3,1)] were calculated to evaluate interobserver and intraobserver agreement. A two-sided P value of <0.05 was considered statistically significant. All the statistical analyses were performed via R software, version 4.5.1.

Table 1 summarizes the baseline clinical and metabolic characteristics of the study participants. No significant differences were observed in age, sex distribution, BMI, BSA, or heart rate among the three groups. Both the HTN-only and HTN-T2DM groups had significantly higher SBP and DBP compared with the control group (all P < 0.001). Among patients with hypertension, the majority were classified as stage II or III, with a median hypertension duration of 5–6 years. With regard to metabolic profiles, the HTN-T2DM group showed significantly worse glycemic control, with higher HbA1c and FBG levels compared to both the control and HTN-only groups (all P < 0.001). They also exhibited higher TyG index [median 8.87 (IQR 8.44, 9.31)] than both the control [8.39 (7.97, 8.80); adjusted P = 0.0007] and HTN-only groups [8.60 (8.26, 9.02); adjusted P = 0.047]. In terms of lipid profiles, with the HTN-T2DM group exhibiting significantly higher TG and lower HDL-C levels compared to the control group (adjusted P = 0.036 and P = 0.0006, respectively). TC and LDL-C levels did not differ significantly among the groups. In the HTN-only and HTN-T2DM groups, the usage rates of antihypertensive drugs were as follows: ACEI/ARB (14.9% vs. 19.6%), ARNI (9.5% vs. 10.7%), β-blockers (36.5% vs. 33.9%), CCB (36.5% vs. 42.9%), and diuretics (9.5% vs. 7.1%). Statins were used for lipid-lowering in 20.3% of HTN-only patients and 30.4% of HTN-T2DM patients. Hypoglycemic medications were only administered in the HTN-T2DM group, with metformin (30.4%), SGLT2i (23.2%), insulin (16.1%), α-glucosidase inhibitors (10.7%) and sulfonylureas (10.7%) being the main agents.

After adjustment for age, sex, BMI, heart rate, and SBP, conventional LV volumetric and functional parameters including EDVi, ESVi, SVi, CI, and CO showed no significant differences among the three groups (all overall ANCOVA P > 0.05). However, both hypertensive subgroups exhibited significantly higher LVMi and LVRi compared with the control group (both P < 0.001). post-hoc comparisons revealed no significant difference in LVMi between the two hypertensive groups (P = 0.3632), while LVRi showed a trend toward further elevation in the HTN-T2DM group (P = 0.0938).

Regarding LA phasic function, both hypertensive groups had significantly lower LA-TEF and LA-AEF than the controls (all P < 0.001, Figure 3). LA-PEF did not reach statistical significance (overall P = 0.0589). LA volume indices demonstrated selective alterations: LAVmini and LAVprei were significantly elevated in hypertensive groups compared to controls (overall P < 0.01), while LAVmaxi showed borderline significance (overall P = 0.0669).

Left atrioventricular coupling assessed by LACI was significantly impaired in both hypertensive subgroups vs. controls (overall P < 0.001). Notably, LACI was further elevated in the HTN-T2DM group compared with the HTN-only group (P = 0.0445), suggesting additional deterioration with superimposed diabetes (Figure 4). The details are shown in Table 2.

Table 3 summarizes alterations in cardiac deformation parameters. After adjustment for covariates, LV strain analysis revealed a graded deterioration across groups. The HTN-T2DM group demonstrated significantly reduced GLS (P < 0.0001), GCS (P = 0.0035), and PDSR.l (P < 0.0001) compared to controls, with GRS showing borderline reduction (P = 0.0201). PDSR.r and PDSR.c were also lower in the HTN-T2DM group vs. controls (P = 0.0295 and P = 0.0159, respectively). In the HTN-only group, only GLS and PDSR.l were significantly impaired relative to controls (both P < 0.0001), with PDSR.c showing a trend toward reduction (P = 0.0665). Notably, GLS was further reduced in the HTN-T2DM group compared with the HTN-only group (P = 0.0017).

LA deformation analysis revealed significant impairment in reservoir and conduit function across the control, HTN-only and HTN-T2DM groups (Figure 3). Compared to controls, both hypertensive groups had significantly lower εs, εe, SRs, and SRe (all P < 0.01). Importantly, the HTN-T2DM group exhibited further deterioration in reservoir (εs: P = 0.0175; SRs: P = 0.0337) and conduit (εe: P = 0.0011; SRe: P = 0.0364) function relative to the HTN-only group. In contrast, the pump function parameter (εa and SRa) was preserved among all groups (overall P > 0.05).

Spearman correlation analysis revealed significant associations among cardiac MRI-derived parameters, as summarized in the heatmap (Figure 5). Key findings include the following: LACI demonstrated moderate to strong negative correlations with all LA phasic strain parameters, particularly with reservoir strain (εs: ρ = −0.54, P < 0.001), conduit strain (εe: ρ = −0.49, P < 0.001), and peak global longitudinal strain of the left ventricle (GLS: ρ = −0.59, P < 0.001). In contrast, GLS showed positive correlations with LA deformation parameters, including reservoir strain (εs, ρ = 0.67, P < 0.001), conduit strain (εe, ρ = 0.61, P < 0.001), and active strain (εa: ρ = 0.51, P < 0.001).

Details are shown in Supplementary Table S1. To explore the relationship between metabolic dysregulation and cardiac alterations, Spearman correlation analyses were performed between key metabolic indicators and CMR-derived parameters. DM.duration showed prominent associations with cardiac structure and function: positively correlated with LV remodelling (Mi; ρ = 0.283, Ri; ρ = 0.354) and LACI (ρ = 0.372, all P < 0.001); negatively correlated with LV deformation (GLS; ρ = −0.399, εs; ρ = −0.396, εe; ρ = −0.448, all P < 0.001), indicating longer DM duration links to atrioventricular uncoupling and impaired myocardial deformation. HbA1c exhibited similar adverse patterns: positively associated with LVMi (ρ = 0.181, P = 0.018), Ri (ρ = 0.372) and LACI (ρ = 0.250, both P < 0.001); negatively correlated with GLS (ρ = −0.332), εs (ρ = −0.249) and εe (ρ = −0.313, all P < 0.001). FBG was positively correlated with LVRi (ρ = 0.197, P = 0.010) and negatively with GLS (ρ = −0.176, P = 0.021). Notably, HDL had protective effects: positively correlated with GLS (ρ = 0.249, P < 0.001), εs (ρ = 0.171, P = 0.025) and εe (ρ = 0.163, P = 0.033); negatively associated with LVRi (ρ = −0.265, P < 0.001) and LACI (ρ = −0.185, P = 0.015).

Multivariable linear regression models were employed to examine the independent associations between clinical variables and cardiac deformation parameters (εs, εe, GLS), and the left atrioventricular coupling index in both the overall study population and a subgroup of hypertensive patients.

In the model incorporating the entire cohort (Table 4), both HTN and HTN-T2DM were independently associated with significantly reduced left atrial reservoir function (εs), conduit function (εe), and reduced left ventricular GLS (all P < 0.05). The reduction in strain parameters was more pronounced in the HTN-T2DM group compared to the HTN-only group (e.g., εs: β = −11.96 vs. −5.67). Conversely, both HTN-only and HTN-T2DM were associated with a significant increase in LACI (β = 5.52 and 8.10, respectively; both P < 0.001), indicating impaired left atrioventricular coupling. Additionally, age and male sex were consistently associated with myocardial deformation and LACI.

Multivariable linear regression analyses within the hypertensive subgroup (Table 5) revealed that comorbid T2DM was independently associated with impaired left atrioventricular deformation parameters after adjusting for age, sex, BMI, heart rate, SBP, hypertension duration, HDL-C, LDL-C, and TyG. Specifically, T2DM was a significant independent predictor of reduced LA reservoir function (εs: β = −6.09, P = 0.018), impaired LA conduit function (εe: β = −5.58, P = 0.002), and worsened LV systolic function (GLS: β = −1.37, P = 0.010). The models explained 17.4%, 28.6%, and 14.4% of the variance in εs, εe, and GLS, respectively.

Reproducibility analysis confirmed the robustness of CMR-FT measurements across all deformation parameters (Supplementary Table S2). Intraobserver ICCs ranged from 0.92 (GCS) to 0.98 (εs), while interobserver ICCs varied between 0.87 (GCS) and 0.97 (εs).