Glucagon-like peptide-1 receptor agonists (GLP-1RA) may play a role in improving quality of life and improving long-term outcomes in heart failure with preserved ejection fraction (HFpEF) patients. We conducted a systematic review and meta-analysis evaluating the safety and efficacy of GLP-1RA in HFpEF patients.
MethodsA systematic search strategy was developed using databases, namely PubMed, Web of Science, Cochrane, and Clinicaltrials.gov. The primary outcomes analyzed were hospitalisation events, MACE, and quality of life parameters, i.e., Kansas City Cardiomyopathy Questionnaire Clinical Summary Score (KCCQ-CSS), 6-minute walk distance (6MWD). Review Manager and RStudio were employed for the statistical analysis, utilizing the random effects model with pre-defined effect measures.
ResultsWe identified six randomized controlled trials (RCTs) and four observational studies in our meta-analysis, with a total of 108,634 patients. Hospitalization due to heart failure was reduced notably by 37% (HR 0.63; 95% CI: 0.54–0.72), and MACE events favoured GLP-1RA by HR of 0.73 (95% CI: 0.61–0.88; p=0.0008). We found a significant improvement with GLP-1RA therapy in KCCQ-CSS scores (MD 8.55, 95% CI: 6.78–10.29; p<0.00001) and 6MWD (MD 15.90, 95% CI: 15.35–16.46; p<0.00001). Regarding safety outcomes, a significant increase in GI effects (RR 1.56; 95% CI: 1.39–1.75; p<0.00001), and a lower risk of arrhythmias (RR 0.84; 95% CI: 0.79–0.90; p<0.0001).
ConclusionGLP-1RA yields significant advantages in HFpEF patients, including enhanced quality of life, valuable weight loss, reduced heart biomarkers, and a decrease in heart failure hospitalizations. Regardless of elevated GI side effects, the GLP-1RAs act as encouraging therapeutic alternatives in HFpEF.
Los agonistas del receptor del péptido similar al glucagón-1 (AR GLP-1) pueden desempeñar un papel en la mejora de la calidad de vida y los resultados a largo plazo en pacientes con insuficiencia cardíaca con fracción de eyección preservada (IC-FEp). Realizamos una revisión sistemática y un metaanálisis para evaluar la seguridad y eficacia de los AR GLP-1 en pacientes con IC-FEp.
MétodosSe desarrolló una estrategia de búsqueda sistemática utilizando las bases de datos PubMed, Web of Science, Cochrane y Clinicaltrials.gov. Los resultados primarios analizados fueron los eventos de hospitalización, los eventos cardiovasculares adversos mayores (MACE, por sus siglas en inglés), y los parámetros de calidad de vida, es decir, la Puntuación del Resumen Clínico del Cuestionario de Miocardiopatía de Kansas City (KCCQ-CSS) y la distancia recorrida en la prueba de caminata de 6 minutos (PC6M). Se utilizaron Review Manager y RStudio para el análisis estadístico, empleando el modelo de efectos aleatorios con medidas de efecto predefinidas.
ResultadosIdentificamos seis ensayos controlados aleatorizados (ECA) y cuatro estudios observacionales en nuestro metaanálisis, con un total de 108,634 pacientes. La hospitalización por insuficiencia cardíaca se redujo notablemente en un 37% (HR 0.63; IC 95%: 0.54 a 0.72), y los eventos MACE favorecieron a los AR GLP-1 con un HR de 0.73 (IC 95%: 0.61 a 0.88; P=0.0008). Encontramos una mejora significativa con la terapia de AR GLP-1 en las puntuaciones KCCQ-CSS (DM 8.55, IC 95%: 6.78 a 10.29; P<0.00001) y en la PC6M (DM 15.90, IC 95%: 15.35 a 16.46; P<0.00001). En cuanto a los resultados de seguridad, hubo un aumento significativo de los efectos gastrointestinales (RR 1.56; IC 95%: 1.39-1.75; P<0.00001), y un menor riesgo de arritmias (RR 0.84; IC 95%: 0.79-0.90; P=<0.0001).
ConclusiónLos AR GLP-1 ofrecen ventajas significativas en pacientes con IC-FEp, incluida una mejor calidad de vida, una pérdida de peso valiosa, una reducción de los biomarcadores cardíacos y una disminución de las hospitalizaciones por insuficiencia cardíaca. A pesar de los elevados efectos secundarios gastrointestinales, los AR GLP-1 actúan como alternativas terapéuticas prometedoras en la IC-FEp.
In the past, heart failure with preserved ejection fraction (HFpEF) was considered less common than heart failure with reduced ejection fraction (HFrEF). However, data from the Framingham Heart Study and other research suggests that HFpEF's incidence and prevalence are increasing and may even be surpassing HFrEF in some populations.1 HFpEF constitute more than half of all heart failure patients and are on the rise, especially because more people are suffering with the underlying risk factors of hypertension, obesity and type 2 diabetes mellitus (T2DM).2,3 Lifestyle modification and caloric reduction can perhaps offer modest benefit but are frequently challenging to maintain and inadequate for the provision of significant clinical benefit in this group.2
Although the increasing burden of HFpEF has initiated a search for effective therapy, few have been discovered to target directly the underlying the disease.4 Currently, recommended management largely mirrors that of general HF populations with use of diuretics to reduce congestion and improve symptoms.5 This highlights the imperative need for the development of targeted pharmacologic interventions that address HFpEF and its associated comorbidities.6–8
Glucagon-like peptide-1 receptor agonists (GLP-1RAs) have already demonstrated significant body weight reductions, improvement of cardiovascular risk factors, and improved physical function in patients with overweight.7,9 One of the GLP-1RAs, semaglutide in the STEP-HFpEF and STEP-HFpEF DM trials has led to dramatic improvements in the Kansas City Cardiomyopathy Questionnaire Clinical Summary Score (KCCQ-CSS), enhanced 6-minute walk distance, reduced inflammation, and attained clinically significant weight reduction.12
We aimed to conduct a systemic review and meta-analysis to analyse the safety and efficacy of GLP-1RAs in the treatment of HFpEF to look for future considerations as a standard therapy in this population.
MethodsThe review protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO) under the identifier CRD420250653153. This systematic review and meta-analysis were conducted per the Cochrane Handbook for Systematic Reviews of Interventions and reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines.10
Search strategy and information sourcesThe literature search was performed systematically in PubMed, Cochrane Library Database, Clinicaltrials.gov, and Web of Science databases by two independent authors for articles related to the safety and efficacy of GLP-1RAs in patients with HFpEF. The following keywords were used to search the relevant studies: “Heart failure or cardiovascular death”, “Glucagon-Like Peptide-1 Receptor Agonists or GLP-1 Agonists or Semaglutide or Liraglutide or Tirzepatide”, “diabetes mellitus or hypertension or obesity”. The search, conducted using appropriate keywords, included studies published up until February 06th, 2025. The line-by-line search strategy for each database is presented in Supplementary Table S1.
Inclusion and exclusion criteriaStudies were included in this systematic review and meta-analysis if they met the following criteria: (1) patients diagnosed with heart failure and having a preserved left ventricular ejection fraction (LVEF ≥45%); (2) age ≥18 years; (3) no hospitalization due to HF between screening and randomization; (4) studies that compared at least two treatment strategies; (5) survival outcomes of grouped patients should be presented.
Exclusion criteria were as follows: (1) patients with uncontrolled DM or history of Type-1 DM; (2) conference abstracts, reviews, case reports, studies without usable data, and studies published before 2010.
Study selection processThe retrieved references were screened based on their titles, abstracts, and relevant outcomes, and assessed for eligibility according to pre-defined inclusion criteria by two independent authors. Any discrepancies were resolved by a third independent author. The conference presentations and grey literature were not included in the literature search.
Screening and data extractionTwo reviewers, BA and HAC, extracted data from the included studies, focusing on data characteristics, interventions, comparisons, and outcomes. Key outcomes of interest included NT-ProBNP level, change in body weight, change in KCCQ-CSS score, change in 6MWD, change in glycemic control or HbA1c, GI side effects, change in SBP (systolic blood pressure), major adverse cardiovascular events (MACE), HF hospitalization, cardiovascular mortality, arrhythmias and GI events. The extracted data were compiled into a standardized Excel sheet. A third independent reviewer was involved in resolving any discrepancies.
Risk of bias assessment and quality assessmentThe risk of bias in the included studies was evaluated by two authors, i.e., BA and HAC using version 2 of the revised Cochrane risk-of-bias 2.0 (RoB 2.0) tool for randomized controlled trials. The NewCastle-Ottawa scale (NOS) was employed by the same authors for the quality assessment of observational studies.11 RoB 2.0 comprises five domains which appraise the bias and quality of the study. These domains include: (1) randomization process, (2) deviation from intended interventions, (3) missing outcome data, (4) measurement of outcome, and (5) selection of the reported results. The results of bias evaluation were graded as low, some concern, and high. To gauge the quality of observation studies, NOS encompasses three domains which are: (1) selection, (2) comparability, and (3) outcomes (exposure). NOS awards a maximum of nine stars depending upon the quality of studies. The stars were assigned as follows: high quality studies: 7–9 stars, moderate quality studies: 4–6 stars, and low-quality studies: less than 4 stars. Bias in the bias studies was presented as a traffic light plot and summary table. The risk of bias in the merged studies was revealed blindly by the two authors, i.e., BA and HAC, and the disparities were addressed by the third senior author, RS.
Statistical analysis of outcomesWe employed Review Manager (RevMan) version 5.4.1 and RStudio version 2024.04.2 for the conduction of statistical analysis. Assessment of risk ratios (RR) and mean differences (MD) for dichotomous and continuous outcomes was done using the Mantel–Haenszel and inverse variance methods, respectively. Further, hazard ratios (HR) and ratios of geometric means were used for time-to-event outcomes and skewed continuous outcomes, respectively. The confidence interval (CI) was set at 95%, with results depicted as forest plots. A p-value of <0.05 was considered statistically significant. We used I2 statistics to assess heterogeneity among studies and conducted sensitivity analyses to scrutinize the robustness of results and reinforce the conclusions of the review. In addition, subgroup analyses were performed to examine the effectiveness of the intervention across different population subgroups. Subgroup analysis by BMI and intervention drug was also done to identify if the BMI difference or using different intervention drugs of GLP-1RA class (i.e., tirzepatide, semaglutide and exenatide) among studies was responsible for heterogeneity. For assessment of BMI effect, the studies were divided into low, moderate, and high BMI groups based on the mean BMI at baseline, with studies having higher numbers of patients in either the <30 (low BMI), 30–35 (high BMI), or >35 (obese) categories. Further, meta-regression analysis was also done to estimate the covariates responsible for heterogeneity among the studies using estimation by β and R2 metrics. Funnel plots were created to assess the publication bias in this meta-analysis.
ResultsStudy selection and populationOur initial search identified 1978 potential studies from three databases (i.e., PubMed, Cochrane Central, ClinicalTrials.gov, and Web of Science). After a meticulous process that involved the removal of duplicates (n=472) and exclusion of articles based on titles and abstract, followed by a detailed assessment of the text of the remaining 64 articles, we ended up with 10 studies for this review – six of those were RCTs, and four were observational studies.3,6,7,9,12–15 For the FLOW trial, pre-specified analysis study was included rather than original study for data extraction as results from the FLOW trial study were stratified by LVEF (HFpEF ≥50%, HFmrEF 40–50%, HFrEF <40%) in the secondary analysis study. Together, the study population size comprised of 108,634 patients from 10 included studies. Moreover, 8768 came from the RCTs, and 99,866 were included from observational studies. The PRISMA flowchart summarizing the studies selection process is shown in Fig. 1. Screening of the articles was done based on the PICO framework (population, intervention, comparator and outcome), alongside inclusion and exclusion criteria defined as above in “Methods” section. The baseline characteristics and details about the included studies are reported in Table 1.
Baseline characteristics of included studies.
| Study | Follow-up | Sample size | Male(%) | Systolic BP(mmHg) | HbA1c(%) | NT-proBNP(pg/mL) | CRP(mg/L) | 6MWT(m) | KCCQ-CSS | Ejection fraction, % | ||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Intervention | Comparator | Intervention | Comparator | Intervention | Comparator | Intervention | Comparator | |||||||
| FLOW | 40.83 months | 325 | 176 (54.2) | 138.0 (129.0–147.0) | 8.1 (7.2–8.9) | 196 (56–488) | 169 (64–476) | 3.1 (1.6–7.5) | NA | NA | NA | <40% 69 (10.2)40%–<50% 66 (9.7) | <40% 14 (0.5)40%–<50% 38 (1.3) | |
| SUMMIT | 23.93 months | 731 | 164 (45.1%) | 127.9 (±13.1) | NA | 196 (56–488) | 169 (64–476) | 5.8±8.5 | 5.8±8.4 | 305.0 (±80.0) | 53.9±17.9 | 53.2±19.0 | 61.0±6.5 | 60.6±6.2 |
| STEP-HFpEF | 11.97 months | 529 | 232 (43.9%) | 133 (121–144) | 450.8 (218.2–1015.0) | 3.8 (1.9–7.0) | 3.9 (2.0–8.4) | 320.0 (240.0–389.0) | 58.9 (41.7–72.9) | 57.0 (50.0–60.0) | 57.0 (50.0–60.0) | |||
| STEP-HFpEF DM | 11.97 months | 616 | 343 (55.7%) | 6.8 (6.2–7.6) | 493 (245–1041.9) | 3.7 (1.8–8.4) | 3.3 (1.6–8.4) | 280 (202.6–351.3) | 60.4 (44.8–72.9) | 58.3 (41.1–70.8) | 57.0 (50.0–61.0) | 55.0 (50.0–60.0) | ||
| SELECT | 39.79 months | 2273 | 1567 (68.9%) | 130.7 (13.6) | 5.80 (0.33) | NA | NA | 2.0 (0.9–4.4) | NA | N/A | NA | 57.0 (50.0–60.0) | 57.0 (50.0–60.0) | |
| EXSCEL trial | 38.4 months | 4294 | 1440 (68) | 135±16 | 8.1±1.0 | N/A | N/A | N/A | N/A | N/A | ||||
| EHC Reterospective Cohort | 11.97 months | 318 | 146 (45.9) | 133 (121–144) | N/A | 450.8 (218.2–1015.0) | 3.8 (1.9–7.7) | 320.0 (240.0–389.0) | N/A | NA | N/A | |||
| TriNetX research network Cohort | N/A | 84,990 | 43,327 (51) | NA | 27,734 (65.3%) | NA | NA | >5mg/dl | NA | NA | NA | N/A | ||
| LinYu et al. | N/A | 14,154 | 3023 (42.7) | NA | 7.4 (1.8) | NA | NA | NA | NA | NA | NA | NA | N/A | |
| Trenas et al. | 18 months | 404 | 84 (41.6) | 132.5±12.4 | 7.8±1.2 | 1328.5±652.0 | 1255.5±631.0 | NA | NA | NA | 53.5±22.0 | 54.9±23.1 | 65.4±13.8 | 66.4±14.0 |
Analyzing results of risk of bias assessed by using RoB 2.0 across 6 RCTs, we determined that the STEP-HFpEF trial has some concerns due to problems with randomization and allocation sequence concealment, while other remaining RCTs have overall low risk of bias. However, FLOW trial has some concerns over missing outcome data. While the EXSCEL and SUMMIT trials have shown some concerns over selective outcome reporting. Also, SELECT trials have some concerns about allocation sequencing. The assessment of individual RCTs is presented in Fig. S1. For publication bias, funnel plot or the primary outcome i.e., hospitalization and cardiovascular mortality can be seen in Figs. S2 and S3, respectively.
Pooled analysesHF hospitalization eventHF hospitalization (hospitalization due to worsening of heart failure) was one of our primary outcomes. The pooled analysis as shown in Fig. 2 confirmed GLP-1RAs significantly lowered HF hospitalization risk by HR: 0.63 (95% CI: 0.54–0.72; p-value=0.02). There was some variation in the results, though, with an overall I2 of 54%, indicating that the studies did not all agree perfectly. GLP-1RA showed a significant reduction in HF hospitalization events in the subgroup of the RCT population compared to placebo (HR 0.60, 95% CI: 0.51–0.72; p=0.001). While observational studies trended toward reduced HF hospitalization events, the difference was not statistically significant, as indicated by a p-value of 0.05. Subgroup analysis by BMI category as shown in Fig. S4, no significant change in heterogeneity was seen with 46% in obese population studies, 51% in moderate BMI and 61% in low BMI studies. Subgroup analysis by intervention drug as shown in Fig. S5 showed 60.3% heterogeneity in studies using semaglutide as target intervention with similar effect size of 0.63 (95% CI: 0.50–0.79). Tirzepatide subgroup showed negligible heterogeneity (HR 0.57, 95% CI: 0.46–0.69).
Leave-one-out sensitivity analysis showed no improvement in heterogeneity as shown in Fig. S6. To further analyze the primary outcome and understand the possible cause of variation among studies due to multiple covariates, we also performed meta-regression analysis of the HF hospitalization event with the follow-up period and percentage of T2DM. Regression analysis for follow-up shown in Fig. S7 indicated some of treatment results variability at high follow-up durations, as indicated by the bubble plot and a coefficient of +0.014. However, the p-value was insignificant (0.0785), and R2 was 0.00%. Further, regression analysis with T2DM (Fig. S8) has shown a flat regression line, indicating no relationship with the covariate (β=0.0007, p=0.7478, R2=0%). Leave-one-out sensitivity analysis based on intervention drug showed no significant improvement in heterogeneity as shown in Fig. S9, with heterogeneity reduced to 42.9% by omitting semaglutide group and increased further to 56.9% by omitting tirzepatide.
Cardiac outcomesCardiovascular mortalityGLP-1RAs significantly reduced cardiovascular mortality with a risk ratio of 0.81 (95% CI: 0.77–0.84; p<0.00001). Heterogeneity among RCTs was negligible (I2=0%), and these studies contributed most of the weight. Observational studies showed no significant effect (Fig. 3(A)).
Major adverse cardiac events (MACE)
When we pooled the data as shown in Fig. 3(B), GLP-1RAs cut the risk of serious cardiac events, with a HR of 0.73 (95% CI: 0.61–0.88; p=0.0008). The studies varied somewhat (I2=49%), and a leave-one-out sensitivity analysis showed 0% heterogeneity after excluding the EXSCEL trial (Fig. S10).
Functional outcomes and quality of lifeIn patients with HFpEF, the primary outcome, i.e., worsening heart failure, was assessed by assessing functional outcomes using three scales (change in KCCQ-CSS, change in 6MWD, and change in body weight), as explained in “Methods” section.
Kansas City Cardiomyopathy Questionnaire Clinical Summary Score (KCCQ-CSS)In analyzing the change in KCCQ-CSS from baseline, GLP-1RAs showed an improvement compared to placebo, with a MD of 8.55 points (95% CI: 6.78–10.29) as shown in Fig. 4(A). This magnitude of improvement is clinically relevant (p-value <0.0001). The studies showed slightly high heterogeneity (I2=53%) but leave-one-out sensitivity analysis indicated that the variation was due to Trenas et al. (Fig. S11).
Six-minute walking test distance (6MWD)
Similarly, analyzing change in 6MWD as shown in Fig. 4(B), GLP-1RA therapy offered a significant improvement with a pooled MD of 15.90 meters on average (95% CI: 15.35–16.46; p<0.00001), suggesting better results against the comparator. The heterogeneity was 0% in the analysis.
Systolic blood pressureThe GLP-1RA population had reduced SBP by MD of −2.98 (95% CI: −4.44 to −1.59; p<0.0001) as shown in Fig. 5(A). The studies mostly agreed (I2=28%), and leave-one-out sensitivity analysis improved heterogeneity to 0% by excluding either the SUMMIT trial or the EHR Cohort (Fig. S12).
Forest plot showing change in quality-of-life outcomes. (A) Forest plot showing change in systolic BP (SBP) comparing GLP-1 RA with placebo. (B) Forest plot showing change in body weight comparing GLP-1 RA with placebo. (C) Forest plot showing change in glycemic control/HbA1c comparing GLP-1 RA with placebo. (D) Forest plot showing change in NT-pro BNP levels comparing GLP-1 RA with placebo.
Furthermore, the change in body weight is shown in Fig. 5(B), indicated reduction in overall weight using GLP-1RA compared with placebo, with M.D of −7.86 (95% CI: −4.84 to −10.88; p<0.00001). The overall heterogeneity was high (I2=99%). We also conducted a subgroup analysis based on the type of study (i.e., observational vs. RCTs) and the intervention drug used in the studies i.e., semaglutide and tirzepatide. On the basis of type of the study, the heterogeneity between the subgroups and within each subgroup was quite high, i.e., 98% among observational studies and 97% among RCT's and a leave-one-out sensitivity analysis showed no significant change in heterogeneity by omitting any study as shown in Fig. S13. By the intervention drug the heterogeneity among the tirzepatide intervention studies were 99.5% and among the semaglutide group were 93.1%. The outcome in the tirzepatide group although favored intervention but was insignificant (MD: −7.32; 95% CI: −15.94 to 1.30, I2=99.5%), but very high heterogeneity made the finding inconclusive (Fig. S14).
Change in glycemic control (HbA1c)The studies that include the diabetes population indicate that GLP-1Ras reduced HbA1c as shown in Fig. 5(C), by −0.71 MD. (95% CI: −0.97 to −0.45; p<0.00001). The studies were found to be highly varied with I2=100%. Leave-one-out sensitivity analysis showed no significant change upon the exclusion of any of the studies (Fig. S15).
BiomarkersChange in NT-ProBNP levelThe analysis reported a significant improvement in NT-ProBNP level among GLP-1RA users, with a ratio of geometric means of 0.85 (95% CI: 0.78–0.91; p<0.0001), and heterogeneity was zero as shown in Fig. 5(D).
Adverse effectsArrhythmiaArrhythmia was a key adverse event assessed across studies as shown in Fig. S15(A). In observational studies, GLP-1RAs were associated with a significantly lower risk of arrhythmia compared to placebo (RR 0.84; 95% CI: 0.79–0.90; p<0.0001). However, this association was not significant in the RCT subgroup (RR 0.45; 95% CI: 0.18–1.17), and the overall pooled analysis also showed no significant reduction (RR 0.53; 95% CI: 0.26–1.11; p=0.10). Heterogeneity among studies was high (I2=88%).
GI disordersThe most prevalent reported adverse outcome in most studies is GI disorder. Compared to the placebo, GLP-1RAs increased the risk of experiencing GI side effects (RR 1.56, 95% CI: 1.39–1.75; p<0.00001). Additionally, a low level of heterogeneity was observed among the studies (I2=2%) as shown by Fig. S15(B).
DiscussionIn this systematic review and meta-analysis, we included 10 studies. Out of them, six were RCTs, and four were observational cohort studies that assessed the impact of GLP-1RA in patients with HFpEF with having a large population size of 108,634. The study population comprise of multiple ethnicities including blacks, whites, Asians and South Asians. Also, the study population comprised of multiple regions thus indicating wider implications of the findings. Our analysis focused on critical outcomes for HFpEF patients, such as quality of life, physical ability, and safety. Quality of life was assessed by HF hospitalizations, MACE, cardiovascular mortality, changes in SBP, and levels of NT-ProBNP. Further, to analyze physical function, changes in BMI, changes in KCCQ-CSS, and 6MWD scores were used. We also evaluated the safety of GLP-1RA by looking for adverse events like arrhythmias and GI issues. Our study proved that the GLP-1RA therapy in HFpEF patients leads to significant improvements in quality of life and functional capacity, as indicated by increased KCCQ-CSS scores and improvement in 6MWD. Patients also experienced meaningful improvement in BMI and reductions in NT-ProBNP levels, leading to decreased cardiac stress. These advantages were achieved throughout studies with negligible variation, indicating constant effectiveness.
One of our primary outcomes was hospitalization for worsening of heart failure. It indicates that the advancement of the disease and a worse prognosis are major drivers of healthcare costs and morbidity in HFpEF.16 A 37% relative reduction in HF hospitalizations compared to placebo suggests that it may reduce healthcare expenses and improve prognosis in HFpEF. This reduction is clinically meaningful, suggesting that GLP-1RAs may prevent acute decompensations, improve patient stability, and lower the use of medical resources.17 Overall results of HF hospitalization have shown significant heterogeneity, i.e., 54%. To address the variation in our results, we conducted a subgroup analysis based on the study type, specifically RCTs and observational studies. However, the results indicate that although GLP-1RA is associated with a reduction in HF hospitalization in both study types. However, there is still variability among studies in both study types. This aligns with other studies that have analyzed the efficacy of pharmacological treatments for HFpEF, which have shown significant heterogeneity in HF hospitalization events among the studies.18
Other than HF hospitalization, the primary outcomes include cardiovascular mortality. GLP-1RA showed a decline in cardiovascular mortality compared to placebo. The results have shown no variability in the studies when assessed for CV mortality. Subgroup analysis based on the type of study indicated that results were insignificant for observational studies. However, RCTs have shown significant results. Also, neither of the subgroups has shown variability among the studies. Our analysis also showed a favourable impact on MACE, thus suggesting that GLP-1RA can play a protective role in preventing serious cardiac complications. Although some variability existed between studies, the overall direction of benefit was consistent, highlighting their potential role in improving long-term outcomes among HFpEF patients. Nikolaus et al., in a review, have elaborated the role of GLP-1RA in improving cardiovascular outcomes, including MACE and cardiovascular mortality.19
Furthermore, patients on GLP-1RA tended to have lower NT-ProBNP levels compared to those receiving a placebo, suggesting that treatment helped reduce heart strain. Importantly, findings were consistent across all studies, making the results more reliable. Our findings align with the previous meta-analysis done by Avogaro et al., which specifies improvement in NT-ProBNP level by GLP-1RA drugs.20 Also, our findings strengthen the findings in secondary analysis of STEP-HFpEF and STEP-HFpEF DM trials by Petrie et al., where they concluded similar improvements.21
The KCCQ-CSS score assesses patients’ health perceptions, encompassing physical restrictions, symptoms, social limitations, and quality of life, and is crucial for HFpEF patients.22 A change of 5 points is considered clinically meaningful.23 Our meta-analysis found that GLP-1RA therapy considerably improves quality of life, as shown by higher KCCQ-CSS scores. Our analysis had shown moderate heterogeneity among studies for this outcome. Trenas et al. were identified as the major culprits behind variability among studies, indicating that real-world outcomes can differ from those derived in controlled settings. Another important measure of exercise tolerance and functional status in heart failure is the 6MWD. Improved 6MWD also correlates with better quality of life and lower HF hospitalization risk in HFpEF, which reinforces the therapeutic potential of GLP-1RAs beyond metabolic control. The lack of heterogeneity supports the reliability of the results and the generalizability of the study to a vast population group.
The reduction in SBP observed across studies by the analysis highlighted the potential antihypertensive role of GLP-1RA drugs. This is particularly important in HFpEF patients, where elevated blood pressure is key to disease progression.24 The consistency of the blood pressure-lowering effect across studies with different designs and population groups suggests that this benefit of GLP-1RAs is likely applicable to the real-world population outside of experimental studies or controlled patient groups. Clinically, this is a valuable observation, as hypertension plays a substantial role in the development and progression of HFpEF. Improved blood pressure control in this population could contribute meaningfully to symptom relief and better long-term outcomes.25
Improvement in glycemic control was evident among diabetic subgroups treated with GLP-1RAs, aligning with their established role in lowering blood glucose.26 However, the high heterogeneity observed among the studies indicates that the magnitude of this effect may be compromised by variability in the baseline characteristics such as HbA1c levels, duration of diabetes, different GLP-1RAs and their doses used. Thus, it suggests that patient-specific factors play a substantial role in determining the glycemic response. Despite this variability, the direction of benefit remains consistent among all the studies, supporting the continuation of GLP-1RAs for metabolic and cardiac optimization in HFpEF patients with coexisting diabetes.
When compared to placebo, GLP-1RA drastically reduces body weight, which in turn improves cardiorespiratory fitness and ventricular compliance. Obesity exacerbates diastolic dysfunction and ventricular stiffness, which are the primary contributors to HFpEF.27 Nonetheless, this result showed significant heterogeneity, suggesting variation throughout the study. Furthermore, subgroup analysis and excluding one study do not alter the heterogeneity. This indicates it could be due to differences in baseline BMI, diabetes status, and study duration. However, this implies that GLP-1RA may be able to treat metabolic risks in HFpEF and therapeutically modulate important pathophysiological drivers of HFpEF. Recently, Popoviciu et al., in a comprehensive review of multiple trials, concluded the role of GLP-1RA as an anti-obesity drug.28
Our analysis examining the safety of GLP-1RA for the HFpEF population revealed a significantly higher risk of GI disorders associated with GLP-1RA, as indicated by the increased risk ratios. Although they provide benefits in symptom management and functional capacity in HFpEF, clinicians should watch out for the impact of GI adverse events to ensure safe and balanced treatment decisions. Further, this analysis set up the directions for future trials to look for the standard dose achieving maximal efficacy with the lowest possible side effects. In contrast to GI adverse events, arrhythmia has shown a better outcome compared to placebo, which further signifies the recommendation of GLP-1RA for future consideration. The observed reduction in arrhythmia incidence among patients receiving GLP-1RA may be attributable to their capacity to decrease levels of inflammatory markers, such as C-reactive protein.29,30
As an analysis of multiple outcomes had shown significant heterogeneity. So, we performed both meta-regression and subgroup analysis of major outcomes, focused primarily on important baseline clinical factors, to investigate the causes of heterogeneity. The significant baseline causes that we sought to identify as potential sources of variation among studies were BMI, Follow-up period, and T2DM population. Upon meta-regression analysis, there was no apparent association between treatment effect and percentage of T2DM patients in each study. Further, a potential tendency toward treatment effect variability at longer follow-up periods was revealed by a meta-regression of HF hospitalization outcomes against follow-up time. However, this did not reach statistical significance. These results suggest that the follow-up duration of the study or diabetes status at baseline are not individual culprits behind the variation among the studies. Furthermore, subgroup analysis of HF hospitalization events by BMI level at baseline was also done. BMI was categorized into low, moderate, and high groups, but it failed to reduce heterogeneity, with variation persisting nearly similar among studies across all BMI groups. However, the low BMI group has slightly higher heterogeneity. Additionally, subgroup analysis based on different intervention drugs of GLP-1RA class was also performed to understand the influence of dual action of tirzepatide as GLP-1RA and GIP compared to traditional GLP-1RA on overall analysis. It showed studies having semaglutide drug as intervention showed significant heterogeneity among each other with the ones having tirzepatide showed negligible heterogeneity between each other. Moreover, leave-one-out sensitivity analysis showed slight improvement in heterogeneity by omitting either of the exenatide or semaglutide group. Also, in one of the two outcomes having high heterogeneity i.e., change in body weight, subgroup analysis based on the intervention drug showed both tirzepatide and semaglutide group had very high heterogeneity. While, leave-one-out sensitivity analysis of change in glycemic control, showed the study using tirzepatide i.e., SUMMIT trial leads to no improvement in heterogeneity. These findings suggested the lesser role of the intervention drugs towards high heterogeneity among the studies. Together, these analyses highlighted the fact that the observed heterogeneity is probably caused by the complex interactions of these analysed and several other different factors rather than the impact of a single variable. This emphasizes the challenge of interpreting pooled results across diverse study populations and highlights the need for patient-level data to understand treatment response in HFpEF better.
Our research, in contrast to previous meta-analyses that have mainly concentrated on the broader picture of HF or patient groups, such as those with diabetes, our study focuses solely on the effects of GLP-1RA in HFpEF patients, regardless of their diabetic status. This is important because HFpEF is a diverse condition with different underlying causes than HFrEF, and therapies effective in HFrEF have not always been successful in HFpEF.8,31,32
While prior reviews may have given importance to conventional endpoints like mortality and HF hospitalization, our study emphasis the significance of symptom burden and functional status, which are crucial for living with HFpEF.33 Prior analyses of GLP-1RAs in HFpEF have yielded insightful information, however, they have some limitations. Some focused primarily on cardiovascular outcomes, such as CV death and worsening heart failure events, without exploring into patient-centred outcomes like quality of life or functional capacity.34 Others explored the effects of GLP-1RAs across various HFpEF phenotypes but lacked quantitative synthesis of data and did not assess the impact on specific functional or quality-of-life measures, certain articles focused on mechanistic aspects of GLP-1RAs in endothelial function without providing clinical outcome data or performing systematic reviews or meta-analyses.35,36 In contrast, our study addresses these gaps by providing a focused, quantitative synthesis of data specifically targeting the HFpEF population. It provides patient-centred outcomes, including the KCCQ-CSs and 6-minute walk distance, offering a comprehensive view of patient well-being.
LimitationsThere are few limitations in our research that should be considered. Significant variability may have been introduced by the limited number of trials and their varying designs, as well as variations in baseline features like BMI and comorbidities profile like diabetes and hypertension, in the included populations of the studies. GLP-1RA dosage schedules which may have affected the accuracy of calculations. Differences in the definition of cardiovascular death and HF hospitalization might have skewed the results. Due to variations in follow-up times, it was difficult to evaluate long-term outcomes, and few cardiovascular death events reduced statistical power. Publication bias and inclusion of observational studies may have affected generalizability. Future studies should use more standardized criteria and consistent dosing to better define the role of GLP-1RA in HFpEF.
ConclusionThis research makes notable contribution in the current understanding of GLP-1RA in the management of HFpEF by providing a comprehensive evaluation of their impact on clinically important outcomes in HFpEF with having large study population of 108,634 of multiple ethnicities and regions. It demonstrated that use of GLP-1RAs results in enhancement in quality of life and functional ability as indicated by KCCQ-CSS score and distances covered in 6MWT. In addition to that these agents contribute to substantial weight loss, decreased NT-ProBNP levels and reduction in HF hospitalization, MACE and cardiovascular mortality, SBP and arrhythmias, indicating reduced cardiac stress and decreased healthcare expenses with improved patient prognosis. However, clinicians should be aware of the increased risk of GI side effects associated with this therapy. Finally, our analysis supports the incorporation of GLP-1RAs in management of HFpEF, particularly in patients with comorbid obesity or diabetes as it offers a promising avenue for improving outcomes in this challenging patient population. However, due to the existence of high heterogeneity in the change in body weight and HbA1c the magnitude of their effect in HF associated comorbid management can vary.
Ethical approvalThis review is exempted from the Ethical Review Committee.
FundingAuthors received no funding.
Conflict of interestAuthors have no conflict of Interest.
The following are the supplementary data to this article:
