Impact of glycemic optimization on myocardial steatosis and cardiac remodeling in patients with newly diagnosed type 2 diabetes: a longitudinal study

Abstract

Background

Cardiac remodeling is common in individuals with type 2 diabetes (T2D) and is influenced by glycemic and metabolic factors. However, myocardial steatosis, a proposed contributor to diabetic cardiomyopathy, has been inconsistently related to glycemic control. This study aimed to characterize longitudinal changes in myocardial triglyceride content (MTGC) and cardiac remodeling following glycemic optimization in newly diagnosed T2D.

Methods

In this uncontrolled, exploratory longitudinal study, twenty adults with newly diagnosed T2D underwent a 12-month standardized glycemic optimization protocol including insulin, metformin, and empagliflozin, in addition to nutritional and lifestyle counseling. Cardiac magnetic resonance imaging (CMR) and proton magnetic resonance spectroscopy (¹H-MRS) were performed at baseline and after 12 months to assess cardiac structure, function, and MTGC. Longitudinal changes and associations between clinical, biochemical, and imaging parameters were assessed.

Results

Participants (mean age 54.8 ± 9 years, 72.3% male) achieved significant reductions in HbA1c levels, body mass index (BMI) and waist circumference (WC). No significant changes in MTGC were found at follow-up (p = 0.23). CMR evaluation revealed increases in left ventricle (LV) ejection fraction (59.0% [54.8–61.5] vs. 63.1% [56.9–66.3], p = 0.01) and decreases in ventricular volumes: (LV) end-systolic volume (29.9 mL/m² [26.4–35.1] vs. 27.3 mL/m² [22.5–31.7]; p = 0.007), right ventricular (RV) end-systolic volume (30.6 mL/m² [25.9–35.7] vs. 28.7 mL/m² [25.5–32.6], p = 0.02) and RV end-diastolic volume (76.5 mL/m² [64.6–82.4] vs. 72.4 mL/m² [66.1–77.7], p = 0.03). The indexed LV mass increased (46.1 g/m² [35.1–54.2] vs. 49.5 [39.5–54.3], p = 0.006). No associations were found between HbA1c improvement and the MTGC or CMR parameters. Reductions in BMI and WC were associated with greater left atrial strain (ρ = − 0.78 and − 0.77; p < 0.001), whereas reductions in WC were also associated with greater LV end-diastolic volume (ρ = -0.59; p = 0.024).

Conclusions

In patients with newly diagnosed T2D, 12 months of glycemic optimization was associated with changes in cardiac remodeling parameters despite no detectable changes in myocardial steatosis. The observed cardiac changes were more closely associated with concurrent reductions in adiposity markers than with changes in HbA1c, emphasizing weight management as a key target for early prevention of diabetic cardiomyopathy.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12933-026-03105-3.

Research insights

What is currently known about this topic?

• Diabetes increases cardiovascular risk. The effect of glycemic optimization on myocardial fat remains unclear.

What is the key research question?

• How does glycemic optimization affect myocardial steatosis and cardiac remodeling in newly diagnosed type 2 diabetes?

What is new?

• Following glycemic optimization, cardiac remodeling parameters change, whereas myocardial steatosis does not.

How might this study influence clinical practice?

• Reinforces that early metabolic optimization is associated with favorable cardiac remodeling in T2D.

Introduction

Type 2 diabetes (T2D) represents a significant challenge to cardiovascular health, as it is associated with an increased risk of heart failure (HF) and cardiovascular mortality [1–3]. Among the T2D population, the most prevalent HF phenotypes are asymptomatic left ventricular diastolic dysfunction and HF, with preserved ejection fraction affecting 43% and 17% of individuals, respectively [4, 5]. Diabetes duration, obesity, and smoking habits are known to be independent risk factors for heart failure, contributing alongside arterial hypertension and coronary artery disease to its development.

In addition to these classical factors, myocardial steatosis, defined as excessive triglyceride (TG) deposition within cardiomyocytes, has emerged as a potential pathological factor involved in cardiac dysfunction in individuals with T2D [6–10]. Lipid accumulation in the myocardium can induce lipotoxic damage via the accumulation of ceramides or diacylglycerol, triggering oxidative stress and mitochondrial dysfunction, eventually leading to myocyte apoptosis, cardiac fibrosis, and ultimately, cardiac dysfunction [11–13]. The myocardial triglyceride content (MTGC) varies widely among individuals and is increased in people with T2D, metabolic syndrome, and obesity. However, the mechanisms underlying this increase are not yet fully understood [14, 15].

The gold standard for evaluating MTGC is a cardiac biopsy, an invasive technique associated with complications, which limits its use in clinical research for ethical reasons. Therefore, proton magnetic resonance spectroscopy (¹H-MRS) has become the noninvasive method of choice for quantifying myocardial lipid content in experimental animal models and in humans, and it has demonstrated good sensitivity and reproducibility compared with cardiac biopsy [16, 17]. However, few studies have evaluated myocardial steatosis in individuals with T2D using ¹H-MRS, yielding mixed results [18–23].

Although improved glycemic control has been shown to have a favorable effect on cardiac structure and function as assessed by echocardiography in subjects with T2D [24, 25], its specific effect on myocardial steatosis remains uncertain. Most existing data are derived from cross-sectional or short-term studies, and the longitudinal relationships between changes in metabolic control, myocardial lipid content, and cardiac remodeling in newly diagnosed patients with T2D have not been fully elucidated. Therefore, we aimed to characterize longitudinal changes in myocardial triglyceride content and cardiac remodeling following 12 months of glycemic optimization combined with nutritional and lifestyle counseling in individuals with newly diagnosed T2D.

Materials and methods

Study subjects

This substudy was conducted in a subset of participants from a previously published cohort of individuals with newly diagnosed T2D who were recruited from the diabetes outpatient clinic at the Hospital de la Santa Creu i Sant Pau (Barcelona, Spain) between 2018 and 2020 [26, 27]. The inclusion criteria were an age ≥ 18 years and a diagnosis of T2D according to the American Diabetes Association (ADA) criteria at the time of the study [28]. We excluded individuals who were positive for GAD65 antibodies; who received steroids or immunosuppressive therapy; who had reduced renal function (estimated glomerular filtration rate < 60 mL/min/1.73 m²); who had elevated liver enzymes ≥ 3 times above the upper limit of normal; and who had liver cirrhosis, pancreatic disease, known coronary artery disease, heart failure, or active malignancy.

Study design

This prospective before-after longitudinal study included four visits: an initial visit at the time of T2D diagnosis (baseline) and at 3, 6, and 12 months of follow-up. All participants received guideline-based glycemic management combined with nutritional and lifestyle counseling and standard recommendations for physical activity.

The initial pharmacological therapy consisted of insulin glargine U-100 at a dosage of 0.2 IU/kg/day, sitagliptin at a dosage of 100 mg/day, and metformin at a dosage of 1700 mg/day initiated at the baseline visit. Afterward, basal insulin was suspended after 2 weeks, and sitagliptin was replaced with empagliflozin (10 mg/day).

Anthropometric and biochemical evaluations were performed at each visit, and CMR with ¹H-MRS was conducted at baseline and after 12 months.

Given the exploratory nature of the study and the absence of robust prior data on expected longitudinal changes in MTGC assessed by ¹H-MRS in patients with newly diagnosed type 2 diabetes, no formal a priori sample size calculation was performed.

Clinical and biochemical evaluation

For all the subjects, the following clinical data were collected: age, sex, blood pressure, smoking status, dyslipidemia status, body mass index (BMI) and waist circumference (WC). Fasting blood samples were analyzed for lipid profile (total cholesterol [TC], triglyceride [TG], high-density lipoprotein cholesterol [HDL-c], low-density lipoprotein cholesterol [LDL-c]), glucose profile (fasting plasma glucose [FPG], HbA1c, and fasting plasma C-peptide), lipoprotein (a) [Lp(a)] and apolipoprotein B (ApoB), and C-reactive protein (CRP).

Hypertension was defined according to the criteria established by the European Society of Hypertension/European Society of Cardiology [29].

Dyslipidemia was defined as the presence of any of the following: a TG concentration of 150 mg/dL or greater, an HDL-c concentration of less than 40 mg/dL in men or less than 45 mg/dL in women, or an LDL-c concentration greater than 160 mg/dL [30].

Insulin resistance (IR) was estimated using the triglyceride and glucose index (TyG), which was calculated as ln (fasting TG [mg/dL]) × FPG [mg/dL])/2. IR was defined as a TyG ≥ 4.68 [31].

Remnant cholesterol (Rem-c) was calculated as follows: Total cholesterol – [HDL-c -LDL-c] [32].

The duration of hyperglycemia was defined as the time when the criteria for diabetes were identified in primary care, prompting referral to specialized care.

Cardiovascular magnetic resonance data acquisition and analysis

CMR imaging was performed on a 3T scanner (Verio, Siemens, Erlangen, Germany) following a standardized protocol, with the patient in the supine position and a 16-element phased-array coil placed over the chest. Images were acquired during breath-holds with electrocardiographic gating. Cine long-axis (2-, 3-, and 4-chamber views) and short-axis reconstructions (contiguous slices of 8-mm thickness covering from base to apex) were acquired using segmented k-space steady-state free precession (SSFP) sequences. Delayed enhancement images were acquired with a segmented gradient-echo inversion-recovery sequence at matching cine-image slice locations 10 to 20 min after intravenous gadolinium-diethylenetriamine penta-acetic acid (DTPA) administration (Gadovist, 0.15 mmol/kg). The inversion time was optimized to achieve myocardial nulling.

All the images were analyzed offline with commercially available postprocessing software (QMass-MR, v.8.1; Medis Medical Imaging Systems, Leiden, the Netherlands). A CMR expert performed the CMR data analysis. Left (LV) and right ventricular (RV) volumes were obtained from the short-axis cine images tracing the endocardial borders in the end-diastolic and end-systolic frames, excluding the papillary muscles from the tracing. The LV mass was calculated by subtracting the endocardial volume from the epicardial volume at the end of diastole and multiplying it by the tissue density (1.05 g/ml). In accordance with current recommendations, LV mass and volume were indexed to body surface area [33]. Myocardial fibrosis was visually assessed by signal intensity on late gadolinium enhancement (LGE) sequences, and the distribution, location, and number of segments affected by LGE were reported. No LGE positivity was detected in any participant at baseline or at the 12-month follow-up; therefore, LGE status was not included in the statistical analyses.

A feature-tracking CMR-derived analysis of the LV, RV, and left atrium (LA) was performed to measure global longitudinal strains, utilizing commercially available software (QStrain Version 8.1; Medis, Leiden, The Netherlands). For LV global longitudinal strain (GLS), endocardial borders were automatically traced in cine longitudinal 4-, 2- and 3-chamber views. For RV-GLS, endocardial borders were automatically traced in a cine longitudinal 4-chamber view. Finally, for LA-GLS, LA endocardial borders were automatically traced in a cine longitudinal 2-chamber view, excluding the LA appendage and pulmonary veins. The tracking quality was visually assessed, and manual adjustments were made for all strain analyses, if necessary.

Cardiac ¹H-MRS protocol

Cardiac ¹H-MRS was performed concurrently with CMR. A single-voxel, ECG-triggered, and self-navigated ¹H-MRS technique was used to place the region of interest within the septum. To avoid voxel contamination from the blood pool signal, a short-axis view and a 4ch view were used for planning [34].

The resulting ¹H-MRS spectrum was analyzed using dedicated spectroscopy software (SyngoVia, Siemens, Erlangen, Germany). Water peaks were assessed at 4.7 ppm (W), while fat peaks were assessed at 0.9 ppm (Fat1) and 1.3 ppm (Fat2). Total cardiac fat content was calculated as MTGC (%) = ([Fat1 + Fat2]/W) ∗ 100, as previously described [34]. A formal interobserver variability assessment was not performed. Quantification was fully automated and based on previously validated protocols [35]. To minimize measurement noise, fitted baseline-corrected spectral lines were used for MTGC estimation.

Statistical analysis

All anthropometric measures and biochemical and imaging variables were described at baseline (T₀) and after 12 months of treatment (T₁₂). Given the small sample size, continuous variables are expressed as medians (interquartile ranges), and categorical variables are expressed as counts and percentages. Nonparametric tests were used for longitudinal comparisons regardless of distribution.

Given the single-arm longitudinal design, each participant served as their own control. The primary analysis evaluated within-subject changes in the MTGC assessed by ¹H-MRS. The analyses were restricted to participants whose baseline and follow-up imaging data were complete. Within-subject changes between baseline and follow-up were assessed using the Wilcoxon signed-rank test for anthropometric, biochemical, and imaging variables. For each variable, change scores were calculated as T₁₂–T₀. Spearman’s rank correlation coefficients were used to explore associations between changes in anthropometric or laboratory parameters and imaging outcomes. Correlation heatmaps were generated to visually summarize the relationships between variables. Cardiac remodeling parameters derived from cardiac magnetic resonance imaging were analyzed in an exploratory manner. Given the number of secondary outcomes and the hypothesis-generating nature of these analyses, no formal adjustment for multiple comparisons was applied. To complement the p values and facilitate interpretation of the magnitude of the observed changes, effect sizes for nonparametric paired comparisons were calculated using the Wilcoxon r statistic (r = Z/√N), along with 95% confidence intervals obtained through bootstrapping (1000 repetitions). Statistical significance was defined as a two-tailed p value < 0.05.

Post hoc sensitivity analyses were conducted to explore the potential influence of insulin exposure on longitudinal changes in cardiac remodeling parameters and MTGC. First, Spearman rank correlation analyses were used to examine the association between the duration of insulin therapy (expressed in weeks) and changes in imaging-derived outcomes. Second, a group-based sensitivity analysis was performed by comparing participants who discontinued insulin early (≤ 2 weeks) with those who remained on insulin therapy for a longer duration. Between-group comparisons were conducted using nonparametric tests, and effect sizes were estimated using the Wilcoxon r statistic. Third, exploratory adjusted linear regression models were fitted to assess whether insulin exposure independently predicted changes in cardiac outcomes after accounting for relevant baseline covariates. These models included insulin exposure duration as the predictor of interest and were adjusted for baseline body mass index, systolic blood pressure, and HbA1c.

Given the exploratory nature of these analyses, the limited sample size, and the absence of a prespecified hypothesis regarding insulin exposure, all sensitivity analyses were conducted post hoc and interpreted descriptively.

Statistical analyses were performed using the R-based software jamovi v.2.5, and figures were generated using the ggplot2, corrplot, and Hmisc packages in RStudio version 4.4.2 [36].

Missing data rates across imaging variables were minimal. Complete CMR datasets were available for all participants; MTGC assessed by ¹H-MRS was unavailable for one participant because of insufficient spectral quality. No imputation was performed.

Results

Sample description and characteristics

Among the 25 eligible individuals, 20 completed both baseline and follow-up CMR studies and were included in the analysis (Fig. 1). The study subject characteristics are detailed in Table 1. In brief, the participants were middle-aged adults (median age 55 years) and predominantly male (80%), with a high incidence of hypertension (45%) and dyslipidemia (40%). 40% of the participants were active smokers, and 35% reported regular alcohol consumption (median 0 [0–5] standard drinks per week).

Fig. 1.

Flowchart of study participants

Table 1.

Patient characteristics

	New onset T2D (n = 20)
Sex (male), n (%)	16 (80)
Age (years)	55.5 (48.5–60.5)
Smoking status
Nonsmoker, n (%)	6 (30)
Former smoker, n (%)	6 (30)
Active smoker, n (%)	8 (40)
Alcohol consumption, n (%)	7 (35)
Hypertension, n (%)	9 (45)
SBP, mmhg	117.5 (110.0-134.0)
DBP, mmhg	81.0 (74.0–89.0)
Dyslipidemia, n (%)	6 (40)
Duration of hyperglycemia, days	3.0 (1.5–7.5)
Treatment
ACEi, n (%)	6 (30)
ARB, n (%)	2 (10)
Calcium channel blocker, n (%)	4 (20)
Beta blocker, n (%)	1 (5)
Statins, n (%)	1 (5)

ACEi = angiotensin II converting enzyme inhibitor. ARB = angiotensin II receptor blocker

Changes in anthropometric and biochemical parameters during glycemic optimization

Anthropometric and biochemical parameters at baseline and after 12 months of follow-up are shown in Table 2. Briefly, HbA1c levels declined substantially over the follow-up period, with 90% of the participants achieving their glycemic target (HbA1c < 7%). Reductions in anthropometric measures were observed, along with increases in HDL-c and ApoB levels. Liver enzymes and CRP levels also decreased, whereas other biochemical parameters remained unchanged.

Table 2.

Baseline and 12-month follow-up clinical variables

	T0 (N = 20)	T12 (N = 20)	p value
Weight (kg)	96.50 (80.7–114.1)	93.40 (83.3–102.9)	0.016
WC, cm	112.5 (102.5–125.5)	107.25 (101.5–117.5)	0.040
BMI, kg/m2	32.83 (29.7–39.3)	32.56 (29.6–34.3)	0.011
HbA1c, %	11.10 (10.2–13.1)	6.05 (5.7–6.5)	< 0.001
FPG, mg/dL	122.94 (110.3–208.7)	109.80 (104.1–131.1)	0.044
FCP, pmol/L	944.00 (761.15)	1070.00 (287.98)	0.594
Creatinine, mg/dL	0.82 (0.71–0.98)	0.81 (0.73–0.94)	0.271
Total cholesterol, mg/dL	189.25 (158.9–215.8)	175.89 (148.4–206.5)	0.546
HDL-c, mg/dL	41.02 (32.5–44.9)	41.22 (37.7–50.3)	0.036
LDL-c, mg/dL	125.15 (92.5–144.3)	101.00 (85.0–123.3)	0.104
TG, mg/dL	111.52 (99.1–193.8)	132.75 (90.3–188.9)	0.481
Rem-c, mg/dL	29.81 (24.9–39.2)	30.38 (20.1–41.1)	0.622
ApoB, mg/dL	1.02 (0.87–1.25)	0.91 (0.76–1.05)	0.048
Lp(a), mg/dL	11.8 (5.9–27.9)	11.7 (6.8–43.9)	0.922
Total bilirubin, µmol/L	12.00 (8–14)	10.50 (6.5–13)	0.266
AST, IU/L	29.50 (23.5–53)	19.00 (17.5–24.5)	0.001
ALT, IU/L	38.00 (34–66.5)	21.50 (16–31.5)	0.002
GGT, IU/L	35.00 (28–69.5)	19.50 (7–38)	0.002
ALP, IU/L	94.00 (83–107)	78.50 (73.5–102)	0.016
TyG index	4.83 (4.7–5.2)	4.83 (4.5–5.0)	0.388
CRP, mg/L	7.85 (4.55–11.9)	2.35 (1.15–4.75)	< 0.001
US TrT, ng/L	7.4 (5.87–12.03)	8.3 (6.2–11.2)	0.28
NT-proBNP, ng/L	32.6 (12.6–49.2)	24.4 (10.0–55.2)	0.95

Values are expressed as medians (interquartile ranges). Two-tailed p values are reported in the table, with significant values (p < 0.05) shown in bold.

WC = waist circumference, BMI = body mass index, HbA1c = glycated hemoglobin, FGC = fasting plasma glucose, FCP = fasting C-peptide, HDL-c = high-density lipoprotein cholesterol, LDL-c = low-density lipoprotein cholesterol. VLDLc = very low-density lipoprotein cholesterol, TG = triglyceride, RC = remnant cholesterol, Lp(a) = lipoprotein (a), ApoB = apolipoprotein B, GGT = gamma glutamyltransferase, AST = aspartate transaminase, ALT = alanine transaminase, ALP = alkaline phosphatase, TyG index= triglyceride-glucose index and CRP = C-reactive protein

Changes in spectroscopy and CMR parameters

The median values of the ¹H-MRS and CMR imaging parameters are presented in Table 3. No significant change was observed in MTGC after glycemic optimization (T₀: 0.52% [0.25–1.44] vs. T₁₂: 1.05% [0.43–3.06], p = 0.23). In contrast, several CMR parameters significantly changed over the follow-up period (Table 3). In particular, after 12 months of glycemic optimization, the LV ejection fraction (LVEF) increased, whereas the LV end-diastolic volume index (LVEDVi) and RV volume decreased. The indexed LV mass (LVMi) also increased. These changes were associated with moderate-to-large effect sizes, with Wilcoxon r values ranging from 0.46 to 0.59 Supplementary Material. Figure 1).

Table 3.

Spectroscopy and CMR parameters during the glycemic optimization protocol

	T0 (N = 20)	T12 (N = 20)	p value
MTGC - %	0.52 (0.25–1.44)	1.05 (0.43–3.06)	0.225
LVEDVi - ml/m2	74.1 (69.1–80.9)	72.6 (68.3–76.5)	0.097
LVESVi - ml/m2	29.9 (26.4–35.1)	27.3 (22.5–31.7)	0.007
LVEF - %	59.0 (54.8–61.5)	63.1 (56.9–66.3)	0.012
LVMi - g/m2	46.1 (35.1–54.2)	49.5 (39.5–54.3)	0.006
RVEDVi - ml/m2	76.5 (64.6–82.4)	72.4 (66.1–77.7)	0.033
RVESVi - ml/m2	30.6 (25.9–35.7)	28.7 (25.5–32.6)	0.021
RVEF - %	59.3 (55.2–61.6)	60.5 (55.9–61.1)	0.277
LA GLS - %	32.8 (27.3–36.8)	33.1 (28.4–41.2)	0.648
LV GLS - %	23.5 (21.9–26.2)	23.3 (0.8–26.3)	0.765
RV GLS - %	28.3 (26.0-31.9)	29.9 (28.4–34.5)	0.090

Values are expressed as medians (interquartile ranges). Two-tailed p values are reported in the table, with significant values (p < 0.05) shown in bold

MTGC = myocardial triglyceride content. LVMi = left ventricular mass index. EF: ejection fraction. RVESVi = right ventricular end-systolic volume index, RVEDVi = right ventricle end-diastolic volume index, LVESVi = left ventricular end-systolic volume index, LVEDVi = left ventricular end-diastolic volume index. LA GLS = left atrium global longitudinal strain. LV GLS = left ventricular global longitudinal strain. RV GLS = right ventricle global longitudinal strain

Correlations between anthropometric, biochemical and imaging variables

An exploratory, unadjusted Spearman correlation heatmap between MTGC, CMR parameters, and anthropometric and laboratory parameters is presented in Fig. 2. All the rho coefficients and p values are detailed in the Supplementary Material. No associations were found between baseline and 12-month changes in HbA1c and MTGC, between HbA1c and CMR parameters, or between MTGC and CMR parameters (Fig. 2). Similarly, no correlations were observed between changes in anthropometric measures and changes in MTGC. In contrast, significant correlations were observed between the anthropometric measures and some CMR parameters. Changes in BMI and WC were associated with changes in LA GLS (ρ = − 0.78, p < 0.001 and ρ = − 0.77, p < 0.001, respectively). Changes in WC were also associated with changes in LVEDVi (ρ = − 0.5; p = 0.024).

Fig. 2.

Correlation heatmaps between myocardial fat content, cardiac structure and function parameters and anthropometric and laboratory parameters. MTGC = myocardial triglyceride content. LVMi = left ventricular mass index. EF: ejection fraction. RVESVi = right ventricular end-systolic volume index, RVEDVi = right ventricle end-diastolic volume index, LVESVi = left ventricular end-systolic volume index, and LVEDVi = left ventricular end-diastolic volume index. LA GLS = left atrium global longitudinal strain. LV GLS = left ventricular global longitudinal strain. RV GLS = right ventricle global longitudinal strain. WC = waist circumference, BMI = body mass index, HbA1c = glycated hemoglobin, FGC = fasting plasma glucose, FCP = fasting C-peptide, HDLc = high-density lipoprotein cholesterol, LDL-c = low-density lipoprotein cholesterol. VLDLc = very low-density lipoprotein cholesterol, TG = triglyceride, REM-C = remnant cholesterol, Lp(a) = lipoprotein (a), ApoB = apolipoprotein B, GGT = gamma glutamyltransferase, AST = aspartate transaminase, ALT = alanine transaminase, ALP = alkaline phosphatase, TyG index= triglyceride-glucose index and CRP = C-reactive protein

Sensitivity analysis

Post hoc sensitivity analyses revealed that insulin exposure duration was not consistently associated with longitudinal changes in cardiac structure, function, myocardial triglyceride content, or strain parameters. Across complementary analytical approaches—including correlational analyses, group-based comparisons according to early versus longer insulin exposure, and exploratory adjusted regression models—no major or consistent associations were observed. Although isolated moderate associations were detected for selected strain parameters and myocardial triglyceride content, these findings were not replicated across related outcomes and should be interpreted cautiously in the context of multiple exploratory comparisons.

Discussion

The present study revealed that, in individuals with newly diagnosed T2D, twelve months of glycemic optimization combined with nutritional and lifestyle counseling improved cardiac remodeling parameters without altering MTGC. The cardiac benefits observed were closely related to reductions in adiposity (BMI and WC) rather than glycemic normalization, emphasizing the close association between weight reduction and early favorable cardiac adaptations.

Effect of optimized glycemic control on myocardial steatosis in patients with newly diagnosed T2D

In this cohort of individuals with newly diagnosed T2D, improvements in HbA1c were not accompanied by changes in MTGC. To our knowledge, this is among the few longitudinal studies using ¹H-MRS to jointly assess MTGC and cardiac remodeling parameters in individuals with T2D undergoing contemporary glycemic optimization. Previous interventional studies have yielded heterogeneous results, likely reflecting differences in study design, disease stage, and therapeutic context [19–23]. Notably, Zib et al. reported a reduction in MTGC following HbA1c improvement with the addition of pioglitazone, a drug with well-established antisteatotic effects, despite concomitant weight gain [18]. In contrast, van der Meer et al. did not observe a reduction in MTGC after treatment with pioglitazone, despite the effects on myocardial metabolism and cardiac structure [19]. Importantly, both studies were conducted in individuals with long-term T2D and focused on relatively short interventions, with follow-up limited to 6 months. Collectively, these observations suggest that changes in MTGC are not uniformly linked to improvements in glycemic control per se and that treatment class, disease stage, and metabolic effects beyond a reduction in the level of HbA1c may play a relevant role in modulating the myocardial lipid content.

In support of this view, a recent 12-month randomized trial evaluated the effect of a fasting-mimicking diet in patients with type 2 diabetes using ¹H-MRS to measure myocardial triglyceride content. The results revealed a significant reduction in the myocardial triglyceride content following the intervention, suggesting that lifestyle modification may influence myocardial steatosis in this patient population [37]. In another randomized study investigating the effect of intensive glycemic control in patients with HF and T2D, although myocardial steatosis was not directly measured, improvements in muscle strength and body composition were observed in the intensive glycemic control group, suggesting possible indirect effects on myocardial health [38]. Although previous cross-sectional studies [11] have reported associations between MTGC and BMI, we did not observe these associations in our cohort. A possible explanation is our small sample size and the modest 6% reduction in adiposity measures, which may have been insufficient to modify MTGC. Indeed, the extent of fold changes in MTGC was considerably lower than that in other ectopic fat depots, such as the liver, where the lipid content is four- to tenfold greater [7]. No association was detected between IR and MTGC. These findings are consistent with those of Krššák et al., who reported that IR was not associated with MTGC in either healthy or T2D women [39]. Similarly, Van der Meer did not observe a reduction in MTGC following improvements in glycemic control and IR, suggesting that myocardial fat may respond differently than other fat depots with an established relationship with IR, such as liver or skeletal muscle [19].

Effects of optimized glycemic control on cardiac remodeling and function in patients with newly diagnosed T2D

Suboptimal glycemic control has been associated with an increased risk of cardiovascular events [40, 41]. Our study revealed significant improvements in cardiac structure and function after glycemic control; however, these changes were not directly related to reductions in HbA1c but rather to the TyG index. The TyG index is regarded as a reliable surrogate marker for insulin resistance and is strongly correlated with more direct measures, such as HOMA‑IR and the hyperinsulinemic–euglycemic clamp [37–39]. These findings suggest that the relationship between glycemic control and cardiac function is more complex and cannot be fully explained solely by favorable changes in HbA1c. In this context, intensive glycemic therapy in T2D patients has demonstrated mixed effects on cardiac remodeling. Although accumulating evidence from echocardiographic studies suggests that improving glycemic control may have a beneficial effect on cardiac structure and function in individuals with T2D [24, 25, 42], other studies have reported no diastolic improvement, which could be partly attributed to shorter follow-up periods (i.e., 4 and 6 months, respectively); neither showed an association with HbA1c reduction [43, 44].

An increase in LVMi after optimization of glycemic control was observed. Although increased LV mass is commonly interpreted as a marker of left ventricular hypertrophy [48], this change was not accompanied by deterioration in systolic or diastolic function or by adverse changes in ventricular volume or ejection fraction and therefore does not appear to reflect maladaptive remodeling or clinically relevant myocardial dysfunction. Moreover, the observed increase in LVMi may be explained by changes in body size rather than a true increase in absolute myocardial mass. Participants experienced modest but consistent weight loss (approximately 6% of initial body weight), which likely resulted in a reduction in body surface area. Given that LVM was related to body surface area, this reduction may have led to an apparent increase in LVMi despite the stable absolute LV mass. This interpretation is supported by the concomitant improvement in other functional and volumetric remodeling parameters.

The concurrent use of empagliflozin and insulin may influence cardiac structure and functional outcomes. Several studies have shown that empagliflozin, an SGLT2 inhibitor with confirmed cardiovascular benefit across the spectrum of heart failure, can reverse cardiac remodeling [45–47]. In parallel, insulin treatment has been associated with heterogeneous effects. Experimental evidence from animal models suggests that insulin signaling through the PI3K/Akt-1 pathway exerts dual effects, including cardiomyocyte hypertrophy and apoptosis prevention [49, 50], while simultaneously improving sarcoplasmic reticulum function and calcium handling [51], thereby supporting a potential cardioprotective role under certain conditions. Consistent with these findings, large cardiovascular outcome trials such as ORIGIN [52] and DEVOTE [53] have reported no excess cardiovascular risk with insulin therapy when glycemic control is appropriately achieved in patients at high cardiovascular risk. Post hoc exploratory sensitivity analysis did not suggest a significant association between the duration of insulin exposure and the observed changes in cardiac structure, function, or MTGC.

Finally, in an exploratory, unadjusted analysis, we observed that reductions in anthropometric parameters were associated with changes in left atrial global longitudinal strain and LV end-diastolic volume. Excess adiposity has been increasingly implicated in the pathophysiology of heart failure and other cardiac conditions [49]. Adipose tissue dysfunction has been linked to adverse cardiometabolic pathways—including altered adipokine secretion, low-grade inflammation, neurohormonal activation, and insulin resistance—which have been associated with structural and functional cardiac alteration [50] In line with existing evidence, these observations support the concept that obesity is a clinically relevant correlate of adverse cardiac remodeling and heart failure risk [51–54].

This study has several limitations. First, the single-center, before–after design without a control group limits causal inference and precludes attribution of the observed changes to specific components of the intervention. As with any single-arm longitudinal study, regression to the mean cannot be completely excluded as a potential contributor to changes in selected outcomes. Second, the exploratory nature of the study, small sample size, and male predominance limit both the statistical robustness of the analyses and the extrapolation of the findings to broader populations. Third, the naturalistic design did not allow full control of potential confounders, including concomitant use of lipid-lowering or antihypertensive agents, as well as unmeasured lifestyle factors such as diet and physical activity. Although post hoc sensitivity analyses were conducted to explore the potential influence of insulin exposure, these analyses were exploratory in nature and should be interpreted descriptively. Finally, the use of GLP-1 receptor agonists—a drug class with established cardiovascular benefits and weight-reducing effects—was not available because of national health insurance restrictions at the time of the study. In addition, no formal interobserver variability, reproducibility analyses, or phantom calibration data were available for myocardial triglyceride content quantification. Although standardized acquisition and automated postprocessing were used to minimize operator-dependent variability, residual measurement variability—particularly at low MTGC signal levels—cannot be fully excluded.

Despite these limitations, our study has notable strengths. It is among the few longitudinal studies to use ¹H-MRS to evaluate myocardial steatosis alongside detailed CMR-derived measures of cardiac structure and function in individuals with newly diagnosed T2D, with a 12-month follow-up. The standardized treatment protocol and the integrated imaging approach allowed a detailed longitudinal characterization of myocardial lipid content and cardiac remodeling over one year of early glycemic optimization. Together, these features provide valuable descriptive insight into the temporal associations between glycemic control, anthropometric changes, and cardiac remodeling in the early stages of T2D.

Conclusions

In individuals with newly diagnosed T2D, reductions in adiposity parameters, measured as BMI and WC, achieved during glycemic optimization were consistently associated with favorable changes in cardiac remodeling indices, whereas myocardial steatosis remained unchanged. These findings suggest that weight-related mechanisms may be linked to early cardiac adaptations in T2D.

Accordingly, therapeutic strategies that combine glycemic improvement with effective adiposity reduction—including lifestyle interventions and pharmacologic agents with cardiometabolic profiles such as SGLT2 inhibitors or GLP-1 receptor agonists—may be relevant for promoting early cardiometabolic health from the time of diabetes diagnosis.

Further studies with larger, controlled cohorts and longer follow-up periods are needed to determine the time course of myocardial lipid adaptation and to confirm the independent contribution of adiposity reduction to early cardiac remodeling in patients with diabetes.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ApoB

ApoLipoprotein B

BMI

Body mass index

CMR

Cardiovascular magnetic resonance

EF

Ejection fraction

FCP

Fasting C-peptide

FPG

Fasting plasma glucose

HbA1c

Glycosylated hemoglobin

HF

Heart failure

HDL-c

High-density lipoprotein cholesterol

CRP

C-reactive protein

LA GLS

Left atrial global longitudinal strain

LDL-c

Low-density lipoprotein cholesterol

LGE

Late gadolinium enhancement

LVEDVi

Left ventricular end diastolic volume index

LVESVi

Left ventricular end systolic volume index

LVMi

Left ventricular mass index

LP(a)

Lipoprotein(a)

LV GLS

Left ventricular global longitudinal strain

MTGC

Myocardial triglyceride content

Rem-c

Remnant cholesterol

RVEDVi

Right ventricular end diastolic volume index

RVESVi

Right ventricular end systolic volume index

RV GLS

Right ventricular global longitudinal strain

TC

Total cholesterol

TG

Triglycerides

T2D

Type 2 diabetes mellitus

WC

Waist Circumference

TyG

Triglycerides- glucose index

¹H-MRS

Proton magnetic resonance spectroscopy

Author contributions

JJ and NA participated in and designed the study. NA and J.J. conceptualized the study. All the authors contributed to the research data. AR-R, AT, [PG-M](https:/pubmed.ncbi.nlm.nih.gov/?sort=date&term=Gil-Millan+P&cauthor_id=40649060) , AP, JLS-Q, and JJ contributed to the clinical assessment and management and development of the databases. AR-R, AT, JR, BP-P, DM, JJ, and NA analyzed and interpreted the data. BP-P conducted the statistical analysis. All the authors contributed to the discussion and reviewed the manuscript; AR-R, AT, BP-P, JR, JJ and NA wrote the original draft of the manuscript; AR-R, AT, JR, DM, JJ, and NA reviewed/edited the manuscript; and DM supervised the study. All authors read and approved the manuscript. D.M., J.J. and NA provided the funding support.

Funding

This research was funded by grants FIS PI15/00625, PI18/00328, and PI21/01163 (to D.M.), PI16/00471 (to J.L.S.-Q.), PI17/00232, PI21/00770, and PI24/00156 (to J.J.) and PI17/01362 (to N.A.) from the Instituto de Salud Carlos III (ISCIII) (cofinanced by the European Regional Development Fund) and by grants from Fundació La MARATÓ de TV3 (201602.30.31 to N.A. and J.J.). We also acknowledge the support received from CIBER de Diabetes y Enfermedades Metabólicas Asociadas (CIBERDEM) (leading group CB15/00071) and CIBER de Enfermedades Cardiovasculares (CIBERCV), which are projects of the ISCIIII. J.J. was a recipient of a Miguel Servet Type 1 contract (CP13/00070; ISCIII) and a Miguel Servet Type 2 contract (CPII18/00004; ISCIII) and has received financial support from Agencia Estatal de Investigación (MCIN/AEI/10.13039/501100011033 and the European Union “NextGeneration EU”/PRTR) within the action “Consolidación Investigadora 2022” (CNS2022-135559). D.M., J.J., J.R. and N.A. are members of the coordinated consolidated quality research group of the Agència de Gestió d’Ajuts Universitaris i de Recerca (AGAUR) (2021 SGR 00857, and 2021 SGR 01211) from Generalitat de Catalunya. N.A. and D.M. were members of the Quality Research Group 2017-SGR-1149 from the Generalitat de Catalunya. Additionally, J.R., D.M. and J.J. belong to the XARTEC Salut network. The Institut de Recerca de l’Hospital de la Santa Creu i Sant Pau and the Germans Trias i Pujol Research Institute are accredited by the Generalitat de Catalunya as Centers de Recerca de Catalunya (CERCA).

Data availability

Data of the study may be available for research collaboration purposes upon reasonable request to the corresponding authors and will require the completion of a data processing agreement.

Declarations

Ethics approval and consent to participate

The study protocol was approved by our local Ethics Committee of the Hospital de SANT PAU (IIBSP-REL-2017-27; date: Jul 26, 2017); all the procedures fully complied with the Declaration of Helsinki. All the subjects provided written informed consent.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.