This was a prospective, multicenter cohort study conducted from January 2022 to June 2024 across three specialized cardiometabolic outpatient centers in Bulgaria. The objective was to evaluate the clinical, biomarker, and functional effects of treatment with glucagon-like peptide-1 receptor agonists (GLP-1RA), sodium–glucose cotransporter-2 inhibitors (SGLT2i), or their combination in patients with a cardiometabolic profile characterized by heart failure with preserved ejection fraction (HFpEF), type 2 diabetes mellitus (T2DM), obesity, and obstructive sleep apnea (OSA).

A total of 435 patients were screened for eligibility. Of these, 162 met inclusion criteria. Sixteen declined participation, and 12 were lost to follow-up before baseline, resulting in 134 patients for final analysis. Patients were stratified into three therapeutic arms: GLP-1RA + SGLT2i (n = 54), GLP-1RA monotherapy (n = 31), and SGLT2i monotherapy (n = 49) (Fig. 1).

Inclusion criteria were:

Exclusion criteria included:

Obstructive sleep apnea was assessed using ApneaLink™ (ResMed, San Diego, CA, USA), a validated portable type III monitoring device that measures nasal airflow, respiratory effort, pulse oximetry, and heart rate. The following parameters were analyzed: apnea–hypopnea index (AHI), oxygen desaturation index (ODI), minimum and mean SpO₂, and total recording time. The diagnostic accuracy of ApneaLink has been validated against polysomnography, demonstrating high sensitivity and specificity in detecting moderate-to-severe OSA (Ng et al. 2009).

Subjective daytime sleepiness was assessed using the Epworth Sleepiness Scale (ESS), a validated 8-item self-report questionnaire widely used in sleep research globally. A score ≥10 was considered indicative of excessive daytime sleepiness (Johns 1991).

Cardiac function was evaluated by transthoracic echocardiography at baseline and after 6 months of follow-up. Measured parameters included left ventricular ejection fraction (LVEF), E/eʹ ratio, indexed left atrial volume (LAVi), mitral inflow velocities, tricuspid annular plane systolic excursion (TAPSE), systolic pulmonary artery pressure (sPAP), and the right ventricular–pulmonary arterial coupling ratio (TAPSE/sPAP).

Laboratory assessments included NT-proBNP, hsCRP, HbA1c, lipid profile, and renal function markers. Clinical data, including weight, BMI, systolic and diastolic blood pressure, and NYHA functional class, were recorded at each visit.

Statistical analyses were performed using IBM SPSS Statistics v26.0. Data normality was tested using the Shapiro–Wilk test. Continuous variables were expressed as mean ± standard deviation (SD) or median and interquartile range (IQR), as appropriate. Categorical variables were presented as frequencies and percentages. Comparisons between the three groups were performed using one-way ANOVA or the Kruskal–Wallis test for continuous variables and the chi-square test or Fisher’s exact test for categorical variables. Paired t-tests and Wilcoxon signed-rank tests were used for within-group comparisons. Correlations between clinical, echocardiographic, and laboratory variables were explored using Pearson or Spearman correlation coefficients. Time-to-event analyses for predefined outcomes (cardiovascular death) were performed using Kaplan–Meier survival analysis, with differences between survival curves evaluated by the log-rank test. Patients lost to follow-up were censored at the last known contact. A two-tailed p-value < 0.05 was considered statistically significant. Missing data were managed via multiple imputation. All analyses were conducted on an intention-to-treat basis.

The study protocol was approved by the Ethics Committee of the Medical University of Sofia (Protocol No. 1316, approval date: December 13, 2021). All procedures were conducted in accordance with the principles of the Declaration of Helsinki, and all participants provided written informed consent prior to enrollment.

A total of 134 patients were included in the final analysis following stratification into three treatment groups: dual therapy with GLP-1RA and SGLT2i, GLP-1RA monotherapy, and SGLT2i monotherapy. There was a predominance of male participants across all groups, with no significant difference in sex distribution. Most anthropometric measures, such as BMI and waist circumference, did not differ significantly at baseline. Measures related to breathing during sleep – such as the frequency of apneas and hypopneas, oxygen desaturation episodes, and daytime sleepiness – also did not differ significantly between the groups. Similarly, patient-reported quality of life scores and most echocardiographic parameters showed no substantial variation at baseline.

Nevertheless, some important differences were observed. Levels of NT-proBNP, a marker of cardiac strain, were significantly different among the groups at baseline. Differences were also noted in diastolic function (E/eʹ ratio) and left ventricular volume. Certain blood lipid values, including HDL, triglycerides, and total cholesterol, as well as resting heart rate, also showed variation. Overall, despite these few variations, most baseline characteristics were well balanced across the groups, ensuring a comparable starting point for further evaluation (Table 1).

In addition to baseline clinical and physiological profiles, the distribution of comorbidities and background pharmacotherapies was systematically recorded for all three treatment groups. The collected data included key cardiovascular conditions and commonly prescribed drug classes relevant to the management of heart failure and diabetes. An overview of these characteristics is provided in Table 2.

At the 6-month follow-up, noticeable improvements in anthropometric measures were observed. Participants receiving combination therapy showed the most pronounced reductions in body mass index, followed by those on GLP-1RA alone, while individuals treated with SGLT2i exhibited minimal change. A comparable pattern was seen for waist circumference, with the greatest reductions noted in the dual therapy group.

Respiratory outcomes also reflected significant progress over time. Reductions in both the apnea–hypopnea index and the oxygen desaturation index were most evident among patients treated with GLP-1RA-containing regimens. Furthermore, a modest but significant improvement in average nocturnal oxygen saturation was detected. In contrast, changes in the lowest oxygen saturation and time spent below 90% saturation remained similar across groups. A comprehensive overview of these findings is presented in Table 3.

At six months, significant improvements were observed in several cardiac markers. Ejection fraction increased notably in the dual therapy (+3.25 ± 0.50%) and GLP-1RA groups (+3.38 ± 0.43%), while the SGLT2i group showed only a minor change (+0.35 ± 0.45%; p < 0.001). NT-proBNP levels decreased in all groups, with the largest reduction seen in the GLP-1RA arm (−333 ± 736 pg/mL), followed by SGLT2i (−250 ± 791 pg/mL), although these differences did not reach statistical significance (p = 0.094). A modest drop in the E/eʹ ratio was observed, most evident in the GLP-1RA group (−1.95 ± 1.98), compared to dual therapy (−1.25 ± 1.64) and SGLT2i (−1.13 ± 0.22; p = 0.035). Changes in LVEDV and LAVi were minimal and did not differ significantly between groups. The TAPSE/sPAP ratio improved slightly with dual therapy (+0.08 ± 0.09), remained stable in GLP-1RA (+0.04 ± 0.05), and showed no meaningful change in SGLT2i (−0.003 ± 0.07; p < 0.001), as illustrated in Fig. 2.

Patient-reported outcomes, as assessed by the KCCQ, demonstrated a significant overall improvement at six months, with the most notable change observed in the dual therapy group (Δ = 10.85 ± 7.78). This was significantly greater than in the GLP-1RA monotherapy group (Δ = 5.36 ± 8.58) and the SGLT2i monotherapy group (Δ = 2.92 ± 5.03; p < 0.001).

At six months, participants across all groups exhibited measurable improvements in several key metabolic indicators. A significant reduction in HbA1c was observed, with the greatest decrease in the dual therapy group (−1.23 ± 0.56), followed by GLP-1RA monotherapy (−1.26 ± 0.47) and SGLT2i monotherapy (−0.93 ± 0.48; p = 0.004). Renal function, as assessed by eGFR, improved most notably in the SGLT2i group (+8.96 ± 3.95), while changes were more modest in the dual (+2.67 ± 3.23) and GLP-1RA (+1.76 ± 3.76) groups (p < 0.001). Inflammatory status, reflected by hsCRP levels, showed a marked reduction in both GLP-1RA-based groups (−1.19 ± 0.54 in dual therapy; −1.10 ± 0.47 in monotherapy), compared to a smaller decline in the SGLT2i group (−0.32 ± 0.48; p < 0.001).

Lipid profile improvements were also more pronounced in the dual and GLP-1RA groups, including greater reductions in total cholesterol, LDL, and triglycerides, as well as increases in HDL. Systolic and diastolic blood pressure decreased significantly in the GLP-1RA and dual therapy groups, while changes in pulse were minimal and not statistically significant (Table 4).

Survival analysis over the 6-month follow-up period revealed differences among the treatment groups. Patients receiving combination therapy exhibited the highest survival rate (92.6%), followed by those on GLP-1RA monotherapy (83.9%). The SGLT2i group had the lowest survival rate (75.5%). The mean survival times were 5.92 months for the combination group, 5.63 months for GLP-1RA, and 5.35 months for SGLT2i. The log-rank test showed a statistically significant difference between the survival distributions (χ² = 6.23; p = 0.044), favoring the combination approach (Fig. 3).

This study has several limitations that warrant consideration. First, the absence of patient randomization may introduce selection bias. Additional bias is possible due to the lack of a control group. However, from an ethical standpoint, random assignment to treatment or no-treatment arms in patients already indicated for GLP-1RA or SGLT2i therapy would not have been feasible. Instead, the study aimed to reflect real-world clinical practice and therapeutic decision-making, thereby enhancing external validity. Second, sleep-disordered breathing was assessed using a home-based ApneaLink device rather than a full polysomnography study. While this may reduce diagnostic precision, the home-monitoring approach offers significant advantages in terms of accessibility, patient comfort, and cost-effectiveness. It also enabled the inclusion of a larger cohort, which would have been more challenging with resource-intensive in-lab diagnostics. Third, the statistical power of the study is influenced by group size. Although meaningful differences were observed across key parameters, certain comparisons may be underpowered, particularly for biomarker-based outcomes. Future studies with larger sample sizes are needed to confirm and expand upon these findings. Lastly, the duration of follow-up was relatively short. While six months is sufficient to detect functional and metabolic changes, it may not be long enough to capture structural cardiac remodeling or long-term clinical outcomes. Extended observation periods will be essential to establish the durability and prognostic implications of the observed improvements.