GLP-1/GIP Receptor Agonists in Obesity-Related HFpEF: The GLIDE-HF Registry

Background and rationale Obesity is a principal driver of heart failure with preserved ejection fraction (HFpEF) and defines a distinct clinical phenotype characterized by pronounced congestion, limited exercise reserve, and particular susceptibility to metabolic intervention. GLP-1 and dual GLP-1/GIP receptor agonists produce substantial weight loss and have shown favorable effects on symptoms and quality of life in this population. However, how these drugs affect cardiac mechanics during exercise, multiparametrically assessed congestion, and right ventricular-pulmonary hemodynamics remains poorly characterized, as do these effects in real-world practice and across different molecules. Existing large-scale registries rely on hard endpoints (hospitalizations, deaths) and administrative data but lack deep mechanistic phenotyping. The niche and scientific value of GLIDE-HF is precisely this phenotyping: serial exercise echocardiography, lung ultrasound, myocardial strain analysis, a full iron panel, and right ventricular-pulmonary assessment within a single, consistently conducted cohort. The registry provides real-world data, spans multiple molecules, and enables comparison of disease course between treated and untreated patients. Study design and regulatory classification GLIDE-HF is a prospective, single-center, non-interventional observational cohort registry conducted within routine outpatient heart failure care. The overriding principle is that all patients with obesity-related HFpEF meeting eligibility criteria are enrolled regardless of whether they receive a GLP-1/GIP receptor agonist. The decision to initiate or withhold such therapy, the choice of molecule, dosing, and any modification are made solely by the treating physician based on clinical, regulatory, and reimbursement indications, within standard care and independently of registry participation. The protocol does not propose, allocate, or modify any pharmacological intervention. GLP-1/GIP receptor agonist therapy is therefore an observed, not an assigned, variable. The registry does not randomize, does not blind treatment allocation, and does not create a control group through protocol allocation. Patients not receiving such therapy constitute a naturally occurring observational comparator, not a control arm in the interventional sense. Imaging and laboratory assessments performed within the registry correspond to a deepened but clinically appropriate evaluation of an obesity-related HFpEF patient in a heart failure clinic. The scope of assessment is identical across all enrolled patients regardless of treatment, ensuring data comparability between treated and untreated patients. Regulatory classification: non-interventional/observational study with diagnostic components. It is not a clinical trial of a medicinal product; drugs are used in accordance with their Summary of Product Characteristics within standard practice, and treatment allocation is not protocol-defined. The registry is subject to bioethics committee opinion and to personal data protection regulations. Objectives The registry has an explicit hierarchy of objectives. The primary axis concerns treated patients; the observational comparator (untreated patients) is used exploratorily, to the extent permitted by available sample size. Primary objective (treated patients): to characterize, under routine care, the trajectory of exercise diastolic reserve and the accompanying panel of mechanistic and clinical parameters in obesity-related HFpEF patients during GLP-1/GIP receptor agonist therapy. Parameters of particular interest include change over time in exercise E/e' (a measure of left ventricular filling pressures), peak tricuspid regurgitation velocity (a measure of pulmonary pressures), systolic reserve assessed by left ventricular outflow tract velocity-time integral (LVOT VTI), post-exercise lung ultrasound B-line count (a measure of congestion), quality of life (KCCQ), NT-proBNP, and 6-minute walk distance. Secondary objective (observational comparator): exploratory comparison of disease course and clinical status between treated and untreated patients, using the naturally occurring observational comparator. The reason for non-treatment is categorized, with the preferred primary comparator being patients not receiving the drug for economic or access reasons only, as the group most clinically similar to treated patients. All such comparisons are hypothesis-generating. Specific objectives include description of change in multiparametrically assessed congestion (lung ultrasound B-lines, NT-proBNP, CA-125, urinary sodium, clinical congestion score) during treatment; change in quality of life and functional capacity; exploratory comparison of effects and tolerability across individual molecules (semaglutide, tirzepatide, liraglutide, dulaglutide, and others), accounting for confounding by indication; characterization of iron status and its change over time and its association with exercise capacity and diastolic reserve; characterization of the right ventricular-pulmonary system and its change over time; and characterization of treatment safety and tolerability in this population under real-world conditions. Methodological objective (feasibility): to assess the feasibility of reliable serial exercise echocardiography and lung ultrasound in an unselected, real-world obesity-related HFpEF clinic population, specifically the proportion of interpretable measurements of key parameters and its change over the observation period. Because obesity substantially complicates echocardiographic imaging, feasibility is itself a valuable scientific outcome. Analytic character and causal caveat: given the observational design, all analyses of treatment effect are descriptive and hypothesis-generating. Changes observed in patients during treatment (before-after comparison) cannot be interpreted as evidence of a causal drug effect, owing to the absence of randomization, possible regression to the mean, and confounding by indication. Comparisons between treated and untreated patients, despite corrective methods, remain subject to residual selection bias. The registry provides characterization of disease course, scientific signals, data across molecules, and comparison with the natural course, not causal proof. Population and HFpEF diagnosis Working definition of obesity-related HFpEF: a patient with chronic heart failure symptoms or exertional dyspnea; obesity defined as BMI at least 30 kg/m2; preserved left ventricular ejection fraction (LVEF at least 50%); objective features of HFpEF confirmed by a contemporary diagnostic algorithm (H2FPEF and HFA-PEFF scores); and no dominant alternative cause of dyspnea. Diagnosis rests on two complementary validated scores. The H2FPEF score assesses HFpEF probability from clinical and echocardiographic features (BMI greater than 30 kg/m2, 2 points; use of at least 2 antihypertensive drugs, 1 point; atrial fibrillation, 3 points; estimated pulmonary artery systolic pressure greater than 35 mmHg, 1 point; age greater than 60 years, 1 point; E/e' greater than 9, 1 point), interpreted as low (0-1), intermediate (2-5), or high (6-9) probability. The HFA-PEFF score, developed by the Heart Failure Association of the European Society of Cardiology, is based on functional, morphological, and biomarker domains with a total of 0-6 points, interpreted as low (0-1), intermediate (2-4), or high (5-6) probability. Patients qualify if neither score indicates low probability of HFpEF. Exposure and cohort classification The characterized exposure is treatment with a GLP-1 receptor agonist or dual GLP-1/GIP receptor agonist, initiated and conducted by the treating physician within routine care per the Summary of Product Characteristics and indications. The registry documents but does not determine: molecule name, indication, dose and escalation schedule, start date, changes, reductions, and temporary or permanent discontinuations, as well as tolerability and adverse events. For registration purposes the cohort is structured into two groups by exposure status: an exposed cohort (patients receiving GLP-1/GIP receptor agonist therapy) and an unexposed cohort (patients not receiving such therapy). Within these groups, patients are further sub-classified by exposure timing and, for untreated patients, by reason for non-treatment: treated, newly initiating (therapy started at or after enrollment; baseline defined as treatment start; the core of trajectory analyses); treated, therapy started before enrollment (baseline at enrollment; analyzed separately to avoid bias from the pre-observation period); untreated for economic or access reasons (no treatment despite indication and no contraindication, for cost, lack of reimbursement, or unavailability; the preferred primary comparator, most clinically similar to treated patients); untreated because of clinical contraindication or intolerance (classified separately, possible confounding by indication); and untreated by patient preference (declined despite indication and no contraindication; possible self-selection). Exposure status may change over time; the registry records the date and direction of each change, enabling analyses that account for time-varying exposure. Standard HFpEF care is provided to all patients according to current European Society of Cardiology guidelines and center practice, including diuretic treatment of congestion to maintain euvolemia, use of SGLT2 inhibitors and mineralocorticoid receptor antagonists per indication, blood pressure control, management of atrial fibrillation, coronary artery disease, and diabetes, and lifestyle recommendations. The registry does not impose a treatment optimization algorithm beyond center standard; the course of care and any pharmacotherapy changes are documented. Assessment schedule and scope Observation is embedded in the routine outpatient visit rhythm and applies equally to treated and untreated patients. Assessment points are the baseline visit and 12 weeks (window plus/minus 2 weeks), 24 weeks (plus/minus 2 weeks; the key trajectory time point), and 52 weeks (plus/minus 4 weeks; durability of change), with optional continuation at subsequent routine visits. Because of the naturalistic character, time windows are flexible and fitted to the real visit rhythm; the actual date of each visit is recorded and analyses account for deviations from the nominal point. At every point a full, uniform laboratory panel is obtained, identically in treated and untreated patients: cardiovascular biomarkers (NT-proBNP, CA-125); complete blood count with red cell indices, RDW, leukocytes, platelets; renal parameters and electrolytes (creatinine with estimated GFR, sodium, potassium); glucose and HbA1c; high-sensitivity C-reactive protein; liver panel (ALT, AST, GGT, total bilirubin, albumin); urinary albumin-to-creatinine ratio and spot urinary sodium and creatinine with sodium-to-creatinine ratio (assessing natriuretic effect, congestion phenotype, and adherence to sodium restriction); and a full iron panel enabling diagnosis of absolute and functional iron deficiency per ESC definition (ferritin, total iron-binding capacity, serum iron, transferrin, transferrin saturation, soluble transferrin receptor, and hepcidin). Iron deficiency is defined as ferritin below 100 micrograms/L, or ferritin 100-299 micrograms/L with transferrin saturation below 20%. Resting echocardiography is performed to a uniform protocol on a high-end system with DICOM archiving, covering left ventricular dimensions and volumes and biplane Simpson ejection fraction; wall thickness, indexed left ventricular mass, and relative wall thickness; indexed left atrial volume; filling parameters (mitral E and A velocities, deceleration time, septal and lateral e' by tissue Doppler, and calculated E/e'); peak tricuspid regurgitation velocity and estimated pulmonary artery systolic pressure where measurable; right ventricular systolic function (TAPSE) and the TAPSE/PASP coupling index; inferior vena cava diameter and respiratory collapsibility with right atrial pressure estimation, assessed at rest only; and LVOT VTI as a measure of stroke volume. Offline longitudinal strain analysis of left atrial reservoir strain and right ventricular free-wall strain is performed on archived DICOM images by an experienced echocardiographer on anonymized images, with left atrial strain assessed from the apical four-chamber view given frequently limited image quality of other views in obese patients. Exercise echocardiography is the key distinguishing tool. It is performed on a semi-supine bicycle ergometer with continuous ECG monitoring, serial blood pressure measurement, and a physician present with the ability to stop the test immediately. The patient maintains 60 rpm; exercise begins at 20 W and increases by 20 W every 3 minutes until symptom-limited exercise or another stopping indication. Measurements are taken at predefined stages: rest, 20 W, 40 W, peak (highest achieved stage), and, if needed, early recovery. At each stage, where image quality allows, the following are measured: mitral E velocity, septal and lateral e' and calculated E/e', LVOT VTI, peak tricuspid regurgitation velocity where measurable, right ventricular systolic function, heart rate, and blood pressure. A "highest interpretable value" rule addresses limited image quality at peak in obese patients (from hyperventilation and E/A fusion): for each parameter independently, the value is taken from the highest exercise stage at which the measurement was reliably interpretable, following the hierarchy peak greater than 40 W greater than 20 W. The 20 W stage serves as a rescue point; validation studies show that E/e' measured at 20 W is highly concordant with the peak value (feasibility rising from about 75% at peak to over 95% at 20 W) while preserving diagnostic value. Calculated reserve parameters include the rest-to-exercise increase in E/e' (diastolic reserve), LVOT VTI reserve (percent increase over rest), tricuspid regurgitation velocity reserve, and estimated exercise pulmonary artery systolic pressure (calculated from tricuspid regurgitation velocity using a fixed right atrial pressure from the resting measurement, since reliable assessment of inferior vena cava collapsibility during exercise is not feasible and the measured regurgitation velocity is the dominant component of the estimate). Lung ultrasound provides semiquantitative assessment of pulmonary congestion by counting B-lines. A hybrid protocol optimizes feasibility in the immediate post-exercise window: at rest, a full 8-zone protocol (four anterior and four lateral-posterior zones bilaterally) per the European Association of Cardiovascular Imaging consensus, with per-zone B-line count capped at 10; immediately post-exercise, an abbreviated 4-zone anterior protocol (upper and lower anterior zones bilaterally), as anterior zones are most sensitive to dynamic exercise congestion and feasible in the short post-exercise window. Lung ultrasound is performed first after exercise, as B-lines resorb rapidly, with a target acquisition within 2 minutes. The principal dynamic parameter is the post-exercise increase in total anterior B-lines relative to rest (using the same four anterior zones at both points); the resting 8-zone total (baseline congestion) and the number of positive zones are also reported. Quality of life is assessed with the validated Kansas City Cardiomyopathy Questionnaire (Clinical Summary Score) completed by the patient at each point, and functional capacity with the 6-minute walk test per standard procedure. Echocardiographic quality assurance To ensure measurement reliability and objectivity, echocardiographic assessment uses blinded offline analysis: archived images are anonymized, coded, and read by an echocardiographer unaware of clinical data or treatment, providing research-grade measurement quality despite the observational design. Reproducibility is assessed at two levels: inter-observer (about 20% of studies re-read by a second independent blinded echocardiographer on the same images) and intra-observer (about 10% read twice by the main reader at least 4 weeks apart, blinded to prior results). For both levels, coefficient of variation and intraclass correlation coefficient are reported for key parameters (E/e', LVOT VTI, tricuspid regurgitation velocity, TAPSE, left atrial volume). Sample size, recruitment, and duration As an observational registry, GLIDE-HF is not based on a formal power calculation for a predefined confirmatory hypothesis. Target cohort size is pragmatic and dependent on clinic patient flow; planned target recruitment is at least 150 patients, conducted continuously as eligible obesity-related HFpEF patients are enrolled. Recruitment begins in January 2027 and is conducted continuously. Given the registry character, no fixed end date is set; a first data review and analysis are anticipated 24 months after enrollment of the first patient, then at a rhythm dependent on cohort size and scientific aims. The untreated subcohort is gathered opportunistically; its size and composition are reported transparently, and comparative analyses are conducted only to the extent sample size permits. Statistical analysis Analysis populations: the full registry cohort (all enrolled patients, for baseline characterization, feasibility, and safety); the newly-initiating treated cohort (core of trajectory analyses); the observational comparator cohort (untreated patients, primary comparator being the economically excluded group; exploratory analyses); and per-molecule subcohorts (exploratory between-drug comparisons). Primary (descriptive) analysis: trajectories of change over time in treated patients, for mechanistic and clinical parameters (exercise E/e', peak tricuspid regurgitation velocity, LVOT VTI reserve, post-exercise B-lines, KCCQ, NT-proBNP, 6-minute walk test), modeled with linear mixed models with a random patient effect, accounting for actual time from the baseline visit. Changes are reported as mean estimates with 95% confidence intervals, without confirmatory testing. The registry describes trajectories; it does not perform confirmatory testing of a treatment effect. Comparative (exploratory) analysis: comparison of treated and untreated patients, with the primary comparator being the economically excluded group, adjusted for confounding by indication using propensity score methods (matching or weighting), to the extent sample size permits. Sensitivity analyses include alternative comparator definitions. Results are explicitly labeled hypothesis-generating and do not support inference of treatment superiority. Detailed analyses of iron status (trajectories of ferritin, transferrin saturation, soluble transferrin receptor, hepcidin) and of the right ventricular-pulmonary system (tricuspid regurgitation velocity, TAPSE, TAPSE/PASP, right ventricular strain) are conducted as descriptive and exploratory analyses, accounting for treatment status. Missing data: the strategy distinguishes two sources, loss to follow-up and absence of a reliable measurement despite the patient's presence (most often technical, from acoustic window quality). Technical missingness for exercise E/e' is limited by the 20 W rescue rule. Analyses use methods appropriate for data missing at random (multiple imputation), and the proportions of both types of missingness are reported transparently as feasibility metrics. Safety Despite the observational design, all treatment-related adverse events and serious adverse events are recorded and reported to the pharmacovigilance system per applicable regulations, with particular attention to GLP-1/GIP-characteristic gastrointestinal symptoms (nausea, vomiting, diarrhea), dehydration, hypoglycemia, pancreatitis, gallbladder disease, hospitalizations, and serious adverse events. Exercise echocardiography follows standard stress-test safety practice (continuous monitoring, physician presence, ability to stop immediately). Ethics and data protection The registry is conducted in accordance with the Declaration of Helsinki, good clinical practice appropriate to non-interventional studies, and applicable personal data protection law. Enrollment requires written informed consent covering data processing and the imaging and laboratory assessments within the registry panel. Participation is fully voluntary, does not affect access to treatment or the treating physician's therapeutic decisions, and may be withdrawn at any stage without giving a reason and without consequences for further care. Because the scope of assessment is identical across all enrolled patients and lies within a deepened clinical evaluation of an obesity-related HFpEF patient, participation does not entail exposure to risk beyond that arising from routine care and diagnostic testing. Data used for analysis are pseudonymized and stored in a secured center database; echocardiographic images submitted for reading are anonymized. The protocol will obtain the opinion of the competent bioethics committee before enrollment of the first patient; registry assumptions will be recorded in a public study registry before recruitment for scientific transparency. Results will be published in peer-reviewed journals and presented at conferences.