검색
검색 팝업 닫기

Ex) Article Title, Author, Keywords

Article

Split Viewer

Brief Report

DTT 2024; 3(2): 177-184

Published online September 30, 2024

https://doi.org/10.58502/DTT.24.0001

Copyright © The Pharmaceutical Society of Korea.

Preventing Heart Failure Admission with Sodium-Glucose Cotransporter-2 Inhibitors versus Angiotensin Receptor-Neprilysin Inhibitor: A Target Trial Emulation Study

Sohee Park1* , Sungho Bea1* , Yunha Noh1,2 , Gregory Y. H. Lip3,4 , Seng Chan You5 , Eue-Keun Choi6,7 , Han Eol Jeong1,8 , Ju-Young Shin1,8,9

1School of Pharmacy, Sungkyunkwan University, Suwon, Korea
2Centre for Clinical Epidemiology, Lady Davis Institute, Jewish General Hospital, Montreal, Canada
3Liverpool Centre for Cardiovascular Science, University of Liverpool and Liverpool Heart & Chest Hospital, Liverpool, United Kingdom
4Department of Clinical Medicine, Aalborg University, Aalborg, Denmark
5Department of Biomedical Systems Informatics, Yonsei University College of Medicine, Seoul, Korea
6Department of Internal Medicine, Seoul National University Hospital, Seoul, Korea
7Department of Internal Medicine, Seoul National University College of Medicine, Seoul, Korea
8Department of Biohealth Regulatory Science, Sungkyunkwan University, Suwon, Korea
9Department of Clinical Research Design & Evaluation, Samsung Advanced Institute for Health Sciences & Technology, Sungkyunkwan University, Seoul, Korea

Correspondence to:Han Eol Jeong, haneoljeong@hotmail.com; Ju-Young Shin, shin.jy@skku.edu
*These authors contributed equally to this work.
Sohee Park’s current affiliation: Aston Pharmacy School, Aston University, Birmingham, United Kingdom
Sungho Bea’s current affiliation: Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA

Received: January 23, 2024; Revised: May 10, 2024; Accepted: August 14, 2024

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Little is known on the real-world comparative effectiveness of sodium-glucose cotransporter-2 inhibitors (SGLT2i) versus angiotensin receptor-neprilysin inhibitor (ARNi) used for heart failure (HF) management. This study used South Korea’s nationwide claims data from 2015 to 2020 to construct a population-based cohort of new users of SGLT2is or ARNi. Individuals were followed from the first prescription date of SGLT2is or ARNi until outcome occurrence, treatment switch or discontinuation, death, or end of the study period. Within the 1:1 propensity score-matched cohort, we estimated hazard ratios (HR) with 95% confidence intervals (CI) for the risk of HF admission with SGLT2is compared with ARNi using proportional subdistribution hazards model of Fine and Gray. We identified 496 propensity-score matched patient-pairs of SGLT2is and ARNi; with a mean age of 72.5 years and a male representation of 57.6%. Incidence rate of HF admission was 27.3 and 35.6 per 100 person-years in SGLT2is and ARNi group. When comparing the risk of HF admission associated with SGLT2is group with ARNi group, HR was 0.71 (95% CI 0.48-1.04). Effect modifications were observed by history of hospitalization for HF (p-for-interaction=0.002) and by recent use of renin-angiotensin-system inhibitors (p-for-interaction= 0.005). With future studies using more recent data warranted to corroborate our study results, these preliminary findings support current guideline recommendations for HF management and further, suggest similar effectiveness between SGLT2is and ARNi in routine care settings.

KeywordsSGLT2i, ARNi, heart failure, target trial emulation, real-world evidence

Sodium-glucose cotransporter-2 inhibitors (SGLT2is) are the recent addition to guideline-directed medical therapy in heart failure (HF) (McDonagh et al. 2021; Bayés-Genís et al. 2022; Heidenreich et al. 2022). SGLT2is are recommended for managing comorbid conditions and addressing the varying severity levels of HF alongside another relatively novel class of HF medication, angiotensin receptor-neprilysin inhibitor (ARNi), along with conventional HF medications such as renin-angiotensin-system inhibitors and beta-blockers (McDonagh et al. 2021; Bayés-Genís et al. 2022; Heidenreich et al. 2022). Although the absolute usage remains lower than other conventional medications, the utilization of SGLT2is and ARNis are increasing rapidly. Notably, SGLT2is primarily target renal reabsorption, reducing fluid accumulation and promoting beneficial cardiac structural changes, while ARNis aim to improve neurohormonal balance and alleviate symptoms. This divergence in mechanisms raises questions about potential differences in their effectiveness in preventing HF admission. However, despite these guideline updates and the distinct mechanisms of action, the direct comparative effectiveness of SGLT2is versus ARNi in routine clinical practice remains unexplored.

Previous data comparing their effectiveness are primarily based on indirect comparisons of randomized controlled trials through network meta-analysis, which have reported inconsistent results. Two studies showed non-differential risk of HF admission (hazard ratio [HR] 0.87, 95% CI 0.75-1.02) (Aimo et al. 2021) or composite of cardiovascular death or HF admission (HR 0.93, 95% CI 0.82-1.06) (Yan et al. 2021) associated with SGLT2is compared to ARNi, whereas one study found modest lower risk of composite of cardiovascular death or HF admission associated with ARNi compared to SGLT2is (HR 0.86, 95% CI 0.75-0.98) (Teo et al. 2022). Uncertainty remains on the comparative effectiveness of SGLT2is versus ARNi in preventing HF admission. Challenges also exist in generalizing and translating results from ideal trial settings to real-world settings (Dhruva and Redberg 2008). This study aimed to explore the potential effectiveness of SGLT2 inhibitors compared to ARNi in decreasing the risk of HF admission, providing insights into these new treatment options for managing HF in real-world settings. To focus our investigation while addressing potential confounding by indication, we have restricted our study population to patients with both type 2 diabetes and HF, as SGLT2is are indicated for both conditions, whereas ARNi is indicated exclusively for HF.

We emulated a hypothetical target trial using South Korea’s nationwide health insurance claims data between 1 January 2015 and 31 December 2020 (Supplementary Table 1) (Kim et al. 2017; Matthews et al. 2022). Individuals newly prescribed a SGLT2i or an ARNi from 1 January 2020 to 31 December 2020 were eligible for entry into the study cohort, given the publication of DAPA-HF (dapagliflozin and prevention of adverse outcomes in HF) trial results in late 2019; we assumed that SGLT2is were likely to have been considered for HF management after this landmark trial.

Cohort entry was defined as the date of the first prescription for either a SGLT2i (dapagliflozin, empagliflozin, ipragliflozin, ertugliflozin) or an ARNi (sacubitril/valsartan) during this period. Of these eligible patients, we excluded patients aged <18 years or prescribed both study drugs of interest at cohort entry, and those meeting any of the following criteria before cohort entry: not diagnosed with HF or type 2 diabetes; not prescribed conventional HF medications (e.g., renin-angiotensin-system inhibitors, beta-blockers, mineralocorticoid receptor antagonists); diagnosed with type 1 diabetes; prescribed insulin as monotherapy; had contraindication to SGLT2is, which is end-stage renal disease or received dialysis; or received cardiac surgery (Supplementary Fig. 1).

The outcome of interest was time to first HF admission, defined by hospital admission with primary or secondary diagnosis of HF, which has shown positive predictive value of 82.1%. Patient were followed from the date of cohort entry until the earliest of outcome occurrence, switch to a comparator drug, treatment discontinuation (no successive prescription within 30 days from the end of the day supply), death, or end of the study (31 December 2020).

To address the potential confounding between groups and obtain comparability, we conducted 1:1 greedy nearest neighborhood matching without replacement within caliper width of 0.05 in propensity scores. The propensity score was generated by including all baseline characteristics in the multivariable logistic regression model: age at cohort entry date, sex, history of hospitalization of HF, comorbidities, use of medications, baseline diabetes treatment (drug classes and the number of classes), baseline HF treatment (drug classes and the number of classes), and healthcare utilization (Table 1). Baseline comorbidities, use of medications, treatment for diabetes and HF, and healthcare utilization were assessed within a year before cohort entry. We also assessed recent diabetes and HF treatment within a month before and at the date of cohort entry and included in the propensity score model to minimize confounding by indication.

Table 1 Baseline characteristics of patients initiating a SGLT2 inhibitor or an ARN inhibitor in 2020

CharacteristicBefore PS* matchingAfter PS matching
SGLT2iARNiSMDSGLT2iARNiSMD
Total7,562793496496
Age, years
Mean (SD)67.0 (12.5)72.0 (11.1)–0.42673.1 (11.3)72.0 (11.3)0.098
Sex, n (%)
Male4,129 (54.6)483 (60.9)–0.128283 (57.1)288 (58.1)–0.020
Female3,433 (45.4)310 (39.1)0.128213 (42.9)208 (41.9)0.020
Comorbidities, n (%)
Hospitalization for heart failure551 (7.3)268 (33.8)–0.695150 (30.2)138 (27.8)0.053
Hypertension4,954 (65.5)420 (53.0)0.257273 (55.0)277 (55.8)–0.016
Dyslipidemia2,949 (39.0)327 (41.2)–0.046219 (44.2)199 (40.1)0.082
Atrial fibrillation or flutter1,055 (14.0)156 (19.7)–0.153112 (22.6)102 (20.6)0.049
Myocardial infarction412 (5.4)98 (12.4)–0.24449 (9.9)52 (10.5)–0.020
Coronary artery disease1,813 (24.0)224 (28.2)–0.097126 (25.4)134 (27.0)–0.037
Coronary revascularization241 (3.2)58 (7.3)–0.18638 (7.7)37 (7.5)0.008
Aortic valve surgery7 (0.1)3 (0.4)–0.0592 (0.4)3 (0.6)–0.028
Pacemaker or ICD19 (0.3)9 (1.1)–0.1073 (0.6)2 (0.4)0.028
Stroke591 (7.8)86 (10.8)–0.10451 (10.3)52 (10.5)–0.007
Transient ischemic attack101 (1.3)6 (0.8)0.0576 (1.2)4 (0.8)0.040
Peripheral arterial disease38 (0.5)8 (1.0)–0.05810 (2.0)7 (1.4)0.047
Cancer491 (6.5)79 (10.0)–0.12747 (9.5)48 (9.7)–0.007
Chronic kidney disease396 (5.2)136 (17.2)–0.38572 (14.5)70 (14.1)0.012
Chronic pulmonary disease1,334 (17.6)197 (24.8)–0.177128 (25.8)125 (25.2)0.014
Chronic liver disease1,070 (14.1)103 (13.0)0.03466 (13.3)68 (13.7)–0.012
Comedications, n (%)
Statin5,988 (79.2)655 (82.6)–0.087393 (79.2)401 (80.8)–0.040
Calcium-channel blocker4,481 (59.3)337 (42.5)0.340246 (49.6)239 (48.2)0.028
Nonsteroidal anti-inflammatory drug5,049 (66.8)482 (60.8)0.125319 (64.3)311 (62.7)0.034
Systemic corticosteroid4,169 (55.1)414 (52.2)0.059272 (54.8)269 (54.2)0.012
Antidepressants1,540 (20.4)154 (19.4)0.024104 (21.0)99 (20.0)0.025
Antipsychotics1,821 (24.1)214 (27.0)–0.067142 (28.6)143 (28.8)–0.004
Baseline diabetes treatment, n (%)
Metformin6,045 (79.9)556 (70.1)0.228374 (75.4)368 (74.2)0.028
Sulfonylurea3,541 (46.8)400 (50.4)–0.072252 (50.8)245 (49.4)0.028
Meglitinide34 (0.4)4 (0.5)–0.0082 (0.4)3 (0.6)–0.028
α-glucosidase inhibitor147 (1.9)21 (2.6)–0.0479 (1.8)9 (1.8)0.000
Thiazolidinedione952 (12.6)65 (8.2)0.14451 (10.3)52 (10.5)–0.007
Dipeptidyl peptidase-4 inhibitor5,036 (66.6)680 (85.8)–0.461417 (84.1)411 (82.9)0.033
GLP-1 receptor agonist95 (1.3)14 (1.8)–0.0428 (1.6)8 (1.6)0.000
Insulin1,377 (18.2)305 (38.5)–0.461188 (37.9)183 (36.9)0.021
No. of diabetes medications
Without medications819 (10.8)33 (4.2)0.25520 (4.0)26 (5.2)–0.058
1 class1,107 (14.6)85 (10.7)0.11847 (9.5)50 (10.1)–0.020
2 class2,018 (26.7)237 (29.9)–0.071148 (29.8)144 (29.0)0.018
3+ class3,618 (47.8)438 (55.2)–0.148281 (56.7)276 (55.6)0.020
Baseline heart failure treatment, n (%)
RAS inhibitor without neprilysin inhibitor6,232 (82.4)748 (94.3)–0.378447 (90.1)454 (91.5)–0.049
Beta-blocker4,706 (62.2)682 (86.0)–0.564386 (77.8)412 (83.1)–0.132
Mineralocorticoid antagonist1,492 (19.7)498 (62.8)–0.973290 (58.5)270 (54.4)0.081
Diuretic4,199 (55.5)700 (88.3)–0.782438 (88.3)424 (85.5)0.084
Digitalis721 (9.5)169 (21.3)–0.331113 (22.8)100 (20.2)0.064
No. of heart failure medications
Without medications175 (2.3)5 (0.6)0.1409 (1.8)5 (1.0)0.068
1 class1,730 (22.9)22 (2.8)0.63014 (2.8)18 (3.6)–0.046
2 class2,604 (34.4)91 (11.5)0.56870 (14.1)73 (14.7)–0.017
3+ class3,053 (40.4)675 (85.1)–1.044403 (81.3)400 (80.6)0.015
Recent diabetes treatment, n (%)
Metformin6,466 (85.5)363 (45.8)0.921277 (55.8)273 (55.0)0.016
Sulfonylurea3,156 (41.7)285 (35.9)0.119187 (37.7)183 (36.9)0.017
Meglitinide10 (0.1)2 (0.3)–0.0271 (0.2)1 (0.2)0.000
α-glucosidase inhibitor55 (0.7)7 (0.9)–0.0174 (0.8)5 (1.0)–0.021
Thiazolidinedione391 (5.2)23 (2.9)0.11618 (3.6)20 (4.0)–0.021
Dipeptidyl peptidase-4 inhibitor2,312 (30.6)547 (69.0)–0.832332 (66.9)330 (66.5)0.009
GLP-1 receptor agonist40 (0.5)10 (1.3)–0.0786 (1.2)7 (1.4)–0.018
Insulin1,111 (14.7)186 (23.5)–0.224132 (26.6)128 (25.8)0.018
No. of diabetes medications
Without medications570 (7.5)164 (20.7)–0.38489 (17.9)93 (18.8)–0.021
1 class2,675 (35.4)136 (17.2)0.42378 (15.7)78 (15.7)0.000
2 class2,553 (33.8)233 (29.4)0.094141 (28.4)142 (28.6)–0.004
3+ class1,764 (23.3)260 (32.8)–0.212188 (37.9)183 (36.9)0.021
Recent heart failure treatment, n (%)
RAS inhibitor without neprilysin inhibitor6,253 (82.7)364 (45.9)0.832310 (62.5)305 (61.5)0.021
Beta-blocker4,552 (60.2)703 (88.7)–0.690410 (82.7)423 (85.3)–0.071
Mineralocorticoid antagonist1,490 (19.7)483 (60.9)–0.926298 (60.1)288 (58.1)0.041
Diuretic3,656 (48.3)686 (86.5)–0.891428 (86.3)420 (84.7)0.046
Digitalis649 (8.6)125 (15.8)–0.22199 (20.0)81 (16.3)0.094
No. of heart failure medications
1 class2,266 (30.0)49 (6.2)0.65035 (7.1)33 (6.7)0.016
2 class2,675 (35.4)206 (26.0)0.205104 (21.0)117 (23.6)–0.063
3+ class2,621 (34.7)538 (67.8)–0.704357 (72.0)346 (69.8)0.049
Healthcare use, n (%)
Hospitalizations
04,611 (61.0)263 (33.2)0.580155 (31.3)174 (35.1)–0.081
1-22,397 (31.7)401 (50.6)–0.391247 (49.8)238 (48.0)0.036
3+554 (7.3)129 (16.3)–0.28094 (19.0)84 (16.9)0.053
Outpatient visits
0-273 (1.0)4 (0.5)0.0543 (0.6)4 (0.8)–0.024
3-5202 (2.7)16 (2.0)0.04310 (2.0)8 (1.6)0.030
6+7,287 (96.4)773 (97.5)–0.065483 (97.4)484 (97.6)–0.013

*Propensity score was estimated using a multivariable logistic regression model, which included all potential confounders shown in the Table 1 as independent variables. Assessed during the 365-day period before cohort entry. Assessed during the 30-day period before and at cohort entry. ARNi, angiotensin receptor/neprilysin inhibitor; ICD, intra-cardiac defibrillation; PS, propensity score; RAS, renin–angiotensin system; SD, standard deviation; SMD, standardized mean difference; SGLT2i, sodium-glucose cotransporter 2 inhibitor.


Within the propensity score matched cohort, we estimated the incidence rate of HF admission per 100 person-years, and HRs with 95% CIs using proportional subdistribution hazards model of Fine and Gray that treated death as a competing event (Fine and Gray 1999). Schoenfeld residual test was used to assess the proportional hazards assumptions before the survival analysis. As a sensitivity analysis, we adopted an intention-to-treat approach that did not censor follow-up at treatment interruption (switch or discontinuation). We performed subgroup analysis to investigate potential effect modification by sex, history of hospitalization for HF, chronic kidney disease, recent HF treatments. Interactions were tested using Wald test for heterogeneity. Propensity scores were re-estimated within each subgroup and the matching was reapplied for each comparison.

All analyses were conducted using SAS Enterprise Guide version 7.1 (SAS Institute Inc., USA). This study was approved by the institutional review board of Sungkyunkwan University, with a waiver of informed consent (SKKU 2022-04-015).

We identified 8,355 eligible new users of SGLT2is (7,562, 90.5%) and ARNi (793, 9.5%) in our study cohort (Supplementary Fig. 1). During the follow-up period, there were 17 deaths in the SGLT2is group and 3 in the ARNi group. After propensity score matching, 496 matched pairs were included in the main analysis, with a mean age of 72.5 years and a male representation of 57.6%. Among these, there were 6 deaths in the SGLT2is group and 1 in the ARNi group during the follow-up period. All baseline covariates, except for baseline use of beta-blockers, achieved balance between groups after matching, with absolute standardized mean differences <0.1 (Table 1).

In the propensity score matched cohort, incidence rate of HF admission was 27.3 and 35.6 per 100 person-years in SGLT2is and ARNi group. When comparing the risk of HF admission associated with SGLT2is group with ARNi group, HR was 0.71 (95% CI 0.48-1.04) In the intention-to-treat analysis, HR was 0.71 (95% CI 0.49-1.02); the point estimate was nearly identical with main analysis, indicating that the effect of treatment initiation was similar to the effect of adhering to the treatment in this study.

We observed effect modification by history of hospitalization for HF (p-for-interaction = 0.002) and recent use of renin-angiotensin-system inhibitors (p-for-interaction = 0.005). Among patients with a history of hospitalization for HF, the propensity score matched HR for the risk of HF admission with SGLT2is compared to ARNi was 1.37 (95% CI 0.67-2.77), while among those without such history, it was 0.58 (95% CI 0.39-0.91). Regarding recent use of renin-angiotensin-system inhibitors, the propensity score matched HR for the risk of HF admission with SGLT2is compared to ARNi was 0.55 (95% CI 0.33-0.92) among users and 0.77 (95% CI 0.37-1.63) among non-users (Fig. 1 and 2).

Figure 1.Risk of HF admission associated with SGLT2 inhibitor versus ARN inhibitor, using propensity score matching among subgroups population. *Assessed during 365 days before cohort entry. Assessed during 30 days before or at cohort entry. Significant interactions for history of hospitalization for HF (p = 0.002), and recent use of renin–angiotensin-system inhibitors without neprilysin inhibitors (p = 0.005). ARNi, angiotensin receptor-neprilysin inhibitor; CI, confidence interval; HF, heart failure; HR, hazard ratio; IR, incidence rate; PS, propensity score; PY, person years; SGLT2i, sodium-glucose cotransporter 2 inhibitor.
Figure 2.Cumulative incidence curve comparing risk of heart failure admission with SGLT2i versus ARNi. Propensity score was re-estimated and re-matched within each stratified subgroup. Assessed during 365 days before cohort entry. Assessed during 30 days before or at cohort entry. Significant interactions for history of hospitalization for HF (p = 0.002), and recent use of renin–angiotensin system inhibitors without neprilysin inhibitors (p = 0.005). ARNi, angiotensin receptor-neprilysin inhibitor; CI, confidence interval; HF, heart failure; HR, hazard ratio; SGLT2i, sodium-glucose cotransporter 2 inhibitor.

This is the nationwide population-based active comparator new-user cohort study that applied a target trial emulation framework to investigate the association between the risk of HF admission and the treatment with SGLT2is compared to ARNi. We have shown that there is no statistically significant difference in the risk of HF admission between SGLT2is- and ARNi-treated patients who have both type 2 diabetes and HF. Our findings from subgroup analysis suggests that the use of SGLT2is is associated with a reduced risk of HF admissions compared with ARNi, particularly in patients with a history of hospitalization for HF. Operating through mechanisms that involve the reduction of fluid accumulation and the promotion of favorable structural changes in the heart, SGLT2is may exhibit enhanced efficacy in advanced HF. Further research and clinical consideration are warranted to validate our findings.

This study has several limitations that should be considered when interpreting the findings. First, we acknowledge the substantial reduction in sample size resulting from our propensity score matching. This decision was driven by several considerations, including the relatively small comparator group of ARNi users within the larger cohort and significant differences in patient characteristics between treatment groups before matching. We applied matching to address these issues by creating a balanced cohort, enabling a more robust assessment of treatment effects while mitigating potential confounding. Second, it’s important to note that our study is based on data available up to December 31, 2020. This timeframe limits our ability to extrapolate findings beyond the guideline updates made around late 2021 or early 2022. Additionally, our findings are specific to patients with HF with type 2 diabetes and may not fully reflect the landscape of SGLT2i and ARNi use. Third, our study could not differentiate between specific types of HF as information on ejection fraction was unavailable for assessment in the data. Fourth, the potential for differences in censoring patterns between the treatment groups should be acknowledged. Last, despite our efforts in propensity score matching, the presence of more serious health conditions among ARNi group before matching indicates that imbalances may not have been fully addressed, raising the possibility of residual unmeasured confounding.

These preliminary real-world findings align with current guidelines for HF management (McDonagh et al. 2021; Bayés-Genís et al. 2022; Heidenreich et al. 2022), and previous network meta-analyses of trials (Aimo et al. 2021; Yan et al. 2021). However, a meta-analysis of 25 RCTs, including 47,275 individuals, reported a comparable risk of hospitalization for HF between SGLT2is and ARNi. Although these differences in point estimates are not statistically significant, they could be due to unmeasured confounding factors inherent in observational studies, which lack randomization (Teo et al. 2022). Future studies should utilize more recent data and improve control over residual confounders to validate our findings.

These results indicate that SGLT2is may provide comparable benefits to ARNi in the context of routine clinical care. This can potentially inform healthcare in clinical decision-making regarding HF management.

JYS received grants from the Ministry of Food and Drug Safety, the Ministry of Health and Welfare, the National Research Foundation of Korea, and pharmaceutical companies, including Pfizer, Celltrion, and SK Bioscience outside the submitted work. HEJ is employed by the Lunit Inc.

This study used data from the Health Insurance Review & Assessment Service of South Korea (M20210607316). This study was supported by a grant (No. 22213MFDS486) from the Ministry of Food and Drug Safety, South Korea, in 2022-2023. The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the manuscript. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.

  1. Aimo A, Pateras K, Stamatelopoulos K, Bayes-Genis A, Lombardi CM, Passino C, Emdin M, Georgiopoulos G (2021) Relative efficacy of sacubitril-valsartan, vericiguat, and SGLT2 inhibitors in heart failure with reduced ejection fraction: a systematic review and network meta-analysis. Cardiovasc Drugs Ther 35:1067-1076. doi: 10.1007/s10557-020-07099-2
    Pubmed CrossRef
  2. Bayés-Genís A, Aimo A, Metra M, Anker S, Seferovic P, Rapezzi C, Castiglione V, Núñez J, Emdin M, Rosano G, Coats AJS (2022) Head-to-head comparison between recommendations by the ESC and ACC/AHA/HFSA heart failure guidelines. Eur J Heart Fail 24:916-926. doi: 10.1002/ejhf.2542
    Pubmed CrossRef
  3. Dhruva SS, Redberg RF (2008) Variations between clinical trial participants and Medicare beneficiaries in evidence used for Medicare national coverage decisions. Arch Intern Med 168:136-140. doi: 10.1001/archinternmed.2007.56
    Pubmed CrossRef
  4. Fine JP, Gray RJ (1999) A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 94:496-509. doi: 10.1080/01621459.1999.10474144
    CrossRef
  5. Heidenreich PA, Bozkurt B, Aguilar D, Allen LA, Byun JJ, Colvin MM, Deswal A, Drazner MH, Dunlay SM, Evers LR, Fang JC, Fedson SE, Fonarow GC, Hayek SS, Hernandez AF, Khazanie P, Kittleson MM, Lee CS, Link MS, Milano CA, Nnacheta LC, Sandhu AT, Stevenson LW, Vardeny O, Vest AR, Yancy CW (2022) 2022 AHA/ACC/HFSA guideline for the management of heart failure: a report of the American College of Cardiology/American Heart Association Joint Committee on clinical practice guidelines. Circulation 145:e895-e1032. doi: 10.1161/CIR.0000000000001063
    Pubmed CrossRef
  6. Kim JA, Yoon S, Kim LY, Kim DS (2017) Towards actualizing the value potential of Korea Health Insurance Review and Assessment (HIRA) data as a resource for health research: strengths, limitations, applications, and strategies for optimal use of HIRA data. J Korean Med Sci 32:718-728. doi: 10.3346/jkms.2017.32.5.718
    Pubmed KoreaMed CrossRef
  7. Matthews AA, Danaei G, Islam N, Kurth T (2022) Target trial emulation: applying principles of randomised trials to observational studies. BMJ 378:e071108. doi: 10.1136/bmj-2022-071108
    Pubmed CrossRef
  8. McDonagh TA, Metra M, Adamo M, Gardner RS, Baumbach A, Böhm M, Burri H, Butler J, Čelutkienė J, Chioncel O, Cleland JGF, Coats AJS, Crespo-Leiro MG, Farmakis D, Gilard M, Heymans S, Hoes AW, Jaarsma T, Jankowska EA, Lainscak M, Lam CSP, Lyon AR, McMurray JJV, Mebazaa A, Mindham R, Muneretto C, Francesco Piepoli M, Price S, Rosano GMC, Ruschitzka F, Kathrine SA; ESC Scientific Document Group (2021) 2021 ESC guidelines for the diagnosis and treatment of acute and chronic heart failure. Eur Heart J 42:3599-3726. doi: 10.1093/eurheartj/ehab368
    Pubmed CrossRef
  9. Teo YN, Teo YH, Syn NL, Yoong CSY, Cheong AJY, Wee CF, Lim YC, Lee CH, Yeo TC, Chai P, Wong RCC, Lin W, Sia CH (2022) Comparing sacubitril/valsartan against sodium-glucose cotransporter 2 inhibitors in heart failure: a systematic review and network meta-analysis. Clin Drug Investig 42:1-16. doi: 10.1007/s40261-021-01098-3
    Pubmed CrossRef
  10. Yan Y, Liu B, Du J, Wang J, Jing X, Liu Y, Deng S, Du J, She Q (2021) SGLT2i versus ARNI in heart failure with reduced ejection fraction: a systematic review and meta-analysis. ESC Heart Fail 8:2210-2219. doi: 10.1002/ehf2.13313
    Pubmed KoreaMed CrossRef

Article

Brief Report

DTT 2024; 3(2): 177-184

Published online September 30, 2024 https://doi.org/10.58502/DTT.24.0001

Copyright © The Pharmaceutical Society of Korea.

Preventing Heart Failure Admission with Sodium-Glucose Cotransporter-2 Inhibitors versus Angiotensin Receptor-Neprilysin Inhibitor: A Target Trial Emulation Study

Sohee Park1* , Sungho Bea1* , Yunha Noh1,2 , Gregory Y. H. Lip3,4 , Seng Chan You5 , Eue-Keun Choi6,7 , Han Eol Jeong1,8 , Ju-Young Shin1,8,9

1School of Pharmacy, Sungkyunkwan University, Suwon, Korea
2Centre for Clinical Epidemiology, Lady Davis Institute, Jewish General Hospital, Montreal, Canada
3Liverpool Centre for Cardiovascular Science, University of Liverpool and Liverpool Heart & Chest Hospital, Liverpool, United Kingdom
4Department of Clinical Medicine, Aalborg University, Aalborg, Denmark
5Department of Biomedical Systems Informatics, Yonsei University College of Medicine, Seoul, Korea
6Department of Internal Medicine, Seoul National University Hospital, Seoul, Korea
7Department of Internal Medicine, Seoul National University College of Medicine, Seoul, Korea
8Department of Biohealth Regulatory Science, Sungkyunkwan University, Suwon, Korea
9Department of Clinical Research Design & Evaluation, Samsung Advanced Institute for Health Sciences & Technology, Sungkyunkwan University, Seoul, Korea

Correspondence to:Han Eol Jeong, haneoljeong@hotmail.com; Ju-Young Shin, shin.jy@skku.edu
*These authors contributed equally to this work.
Sohee Park’s current affiliation: Aston Pharmacy School, Aston University, Birmingham, United Kingdom
Sungho Bea’s current affiliation: Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA

Received: January 23, 2024; Revised: May 10, 2024; Accepted: August 14, 2024

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

Little is known on the real-world comparative effectiveness of sodium-glucose cotransporter-2 inhibitors (SGLT2i) versus angiotensin receptor-neprilysin inhibitor (ARNi) used for heart failure (HF) management. This study used South Korea’s nationwide claims data from 2015 to 2020 to construct a population-based cohort of new users of SGLT2is or ARNi. Individuals were followed from the first prescription date of SGLT2is or ARNi until outcome occurrence, treatment switch or discontinuation, death, or end of the study period. Within the 1:1 propensity score-matched cohort, we estimated hazard ratios (HR) with 95% confidence intervals (CI) for the risk of HF admission with SGLT2is compared with ARNi using proportional subdistribution hazards model of Fine and Gray. We identified 496 propensity-score matched patient-pairs of SGLT2is and ARNi; with a mean age of 72.5 years and a male representation of 57.6%. Incidence rate of HF admission was 27.3 and 35.6 per 100 person-years in SGLT2is and ARNi group. When comparing the risk of HF admission associated with SGLT2is group with ARNi group, HR was 0.71 (95% CI 0.48-1.04). Effect modifications were observed by history of hospitalization for HF (p-for-interaction=0.002) and by recent use of renin-angiotensin-system inhibitors (p-for-interaction= 0.005). With future studies using more recent data warranted to corroborate our study results, these preliminary findings support current guideline recommendations for HF management and further, suggest similar effectiveness between SGLT2is and ARNi in routine care settings.

Keywords: SGLT2i, ARNi, heart failure, target trial emulation, real-world evidence

Introduction

Sodium-glucose cotransporter-2 inhibitors (SGLT2is) are the recent addition to guideline-directed medical therapy in heart failure (HF) (McDonagh et al. 2021; Bayés-Genís et al. 2022; Heidenreich et al. 2022). SGLT2is are recommended for managing comorbid conditions and addressing the varying severity levels of HF alongside another relatively novel class of HF medication, angiotensin receptor-neprilysin inhibitor (ARNi), along with conventional HF medications such as renin-angiotensin-system inhibitors and beta-blockers (McDonagh et al. 2021; Bayés-Genís et al. 2022; Heidenreich et al. 2022). Although the absolute usage remains lower than other conventional medications, the utilization of SGLT2is and ARNis are increasing rapidly. Notably, SGLT2is primarily target renal reabsorption, reducing fluid accumulation and promoting beneficial cardiac structural changes, while ARNis aim to improve neurohormonal balance and alleviate symptoms. This divergence in mechanisms raises questions about potential differences in their effectiveness in preventing HF admission. However, despite these guideline updates and the distinct mechanisms of action, the direct comparative effectiveness of SGLT2is versus ARNi in routine clinical practice remains unexplored.

Previous data comparing their effectiveness are primarily based on indirect comparisons of randomized controlled trials through network meta-analysis, which have reported inconsistent results. Two studies showed non-differential risk of HF admission (hazard ratio [HR] 0.87, 95% CI 0.75-1.02) (Aimo et al. 2021) or composite of cardiovascular death or HF admission (HR 0.93, 95% CI 0.82-1.06) (Yan et al. 2021) associated with SGLT2is compared to ARNi, whereas one study found modest lower risk of composite of cardiovascular death or HF admission associated with ARNi compared to SGLT2is (HR 0.86, 95% CI 0.75-0.98) (Teo et al. 2022). Uncertainty remains on the comparative effectiveness of SGLT2is versus ARNi in preventing HF admission. Challenges also exist in generalizing and translating results from ideal trial settings to real-world settings (Dhruva and Redberg 2008). This study aimed to explore the potential effectiveness of SGLT2 inhibitors compared to ARNi in decreasing the risk of HF admission, providing insights into these new treatment options for managing HF in real-world settings. To focus our investigation while addressing potential confounding by indication, we have restricted our study population to patients with both type 2 diabetes and HF, as SGLT2is are indicated for both conditions, whereas ARNi is indicated exclusively for HF.

Materials|Methods

We emulated a hypothetical target trial using South Korea’s nationwide health insurance claims data between 1 January 2015 and 31 December 2020 (Supplementary Table 1) (Kim et al. 2017; Matthews et al. 2022). Individuals newly prescribed a SGLT2i or an ARNi from 1 January 2020 to 31 December 2020 were eligible for entry into the study cohort, given the publication of DAPA-HF (dapagliflozin and prevention of adverse outcomes in HF) trial results in late 2019; we assumed that SGLT2is were likely to have been considered for HF management after this landmark trial.

Cohort entry was defined as the date of the first prescription for either a SGLT2i (dapagliflozin, empagliflozin, ipragliflozin, ertugliflozin) or an ARNi (sacubitril/valsartan) during this period. Of these eligible patients, we excluded patients aged <18 years or prescribed both study drugs of interest at cohort entry, and those meeting any of the following criteria before cohort entry: not diagnosed with HF or type 2 diabetes; not prescribed conventional HF medications (e.g., renin-angiotensin-system inhibitors, beta-blockers, mineralocorticoid receptor antagonists); diagnosed with type 1 diabetes; prescribed insulin as monotherapy; had contraindication to SGLT2is, which is end-stage renal disease or received dialysis; or received cardiac surgery (Supplementary Fig. 1).

The outcome of interest was time to first HF admission, defined by hospital admission with primary or secondary diagnosis of HF, which has shown positive predictive value of 82.1%. Patient were followed from the date of cohort entry until the earliest of outcome occurrence, switch to a comparator drug, treatment discontinuation (no successive prescription within 30 days from the end of the day supply), death, or end of the study (31 December 2020).

To address the potential confounding between groups and obtain comparability, we conducted 1:1 greedy nearest neighborhood matching without replacement within caliper width of 0.05 in propensity scores. The propensity score was generated by including all baseline characteristics in the multivariable logistic regression model: age at cohort entry date, sex, history of hospitalization of HF, comorbidities, use of medications, baseline diabetes treatment (drug classes and the number of classes), baseline HF treatment (drug classes and the number of classes), and healthcare utilization (Table 1). Baseline comorbidities, use of medications, treatment for diabetes and HF, and healthcare utilization were assessed within a year before cohort entry. We also assessed recent diabetes and HF treatment within a month before and at the date of cohort entry and included in the propensity score model to minimize confounding by indication.

Table 1 . Baseline characteristics of patients initiating a SGLT2 inhibitor or an ARN inhibitor in 2020.

CharacteristicBefore PS* matchingAfter PS matching
SGLT2iARNiSMDSGLT2iARNiSMD
Total7,562793496496
Age, years
Mean (SD)67.0 (12.5)72.0 (11.1)–0.42673.1 (11.3)72.0 (11.3)0.098
Sex, n (%)
Male4,129 (54.6)483 (60.9)–0.128283 (57.1)288 (58.1)–0.020
Female3,433 (45.4)310 (39.1)0.128213 (42.9)208 (41.9)0.020
Comorbidities, n (%)
Hospitalization for heart failure551 (7.3)268 (33.8)–0.695150 (30.2)138 (27.8)0.053
Hypertension4,954 (65.5)420 (53.0)0.257273 (55.0)277 (55.8)–0.016
Dyslipidemia2,949 (39.0)327 (41.2)–0.046219 (44.2)199 (40.1)0.082
Atrial fibrillation or flutter1,055 (14.0)156 (19.7)–0.153112 (22.6)102 (20.6)0.049
Myocardial infarction412 (5.4)98 (12.4)–0.24449 (9.9)52 (10.5)–0.020
Coronary artery disease1,813 (24.0)224 (28.2)–0.097126 (25.4)134 (27.0)–0.037
Coronary revascularization241 (3.2)58 (7.3)–0.18638 (7.7)37 (7.5)0.008
Aortic valve surgery7 (0.1)3 (0.4)–0.0592 (0.4)3 (0.6)–0.028
Pacemaker or ICD19 (0.3)9 (1.1)–0.1073 (0.6)2 (0.4)0.028
Stroke591 (7.8)86 (10.8)–0.10451 (10.3)52 (10.5)–0.007
Transient ischemic attack101 (1.3)6 (0.8)0.0576 (1.2)4 (0.8)0.040
Peripheral arterial disease38 (0.5)8 (1.0)–0.05810 (2.0)7 (1.4)0.047
Cancer491 (6.5)79 (10.0)–0.12747 (9.5)48 (9.7)–0.007
Chronic kidney disease396 (5.2)136 (17.2)–0.38572 (14.5)70 (14.1)0.012
Chronic pulmonary disease1,334 (17.6)197 (24.8)–0.177128 (25.8)125 (25.2)0.014
Chronic liver disease1,070 (14.1)103 (13.0)0.03466 (13.3)68 (13.7)–0.012
Comedications, n (%)
Statin5,988 (79.2)655 (82.6)–0.087393 (79.2)401 (80.8)–0.040
Calcium-channel blocker4,481 (59.3)337 (42.5)0.340246 (49.6)239 (48.2)0.028
Nonsteroidal anti-inflammatory drug5,049 (66.8)482 (60.8)0.125319 (64.3)311 (62.7)0.034
Systemic corticosteroid4,169 (55.1)414 (52.2)0.059272 (54.8)269 (54.2)0.012
Antidepressants1,540 (20.4)154 (19.4)0.024104 (21.0)99 (20.0)0.025
Antipsychotics1,821 (24.1)214 (27.0)–0.067142 (28.6)143 (28.8)–0.004
Baseline diabetes treatment, n (%)
Metformin6,045 (79.9)556 (70.1)0.228374 (75.4)368 (74.2)0.028
Sulfonylurea3,541 (46.8)400 (50.4)–0.072252 (50.8)245 (49.4)0.028
Meglitinide34 (0.4)4 (0.5)–0.0082 (0.4)3 (0.6)–0.028
α-glucosidase inhibitor147 (1.9)21 (2.6)–0.0479 (1.8)9 (1.8)0.000
Thiazolidinedione952 (12.6)65 (8.2)0.14451 (10.3)52 (10.5)–0.007
Dipeptidyl peptidase-4 inhibitor5,036 (66.6)680 (85.8)–0.461417 (84.1)411 (82.9)0.033
GLP-1 receptor agonist95 (1.3)14 (1.8)–0.0428 (1.6)8 (1.6)0.000
Insulin1,377 (18.2)305 (38.5)–0.461188 (37.9)183 (36.9)0.021
No. of diabetes medications
Without medications819 (10.8)33 (4.2)0.25520 (4.0)26 (5.2)–0.058
1 class1,107 (14.6)85 (10.7)0.11847 (9.5)50 (10.1)–0.020
2 class2,018 (26.7)237 (29.9)–0.071148 (29.8)144 (29.0)0.018
3+ class3,618 (47.8)438 (55.2)–0.148281 (56.7)276 (55.6)0.020
Baseline heart failure treatment, n (%)
RAS inhibitor without neprilysin inhibitor6,232 (82.4)748 (94.3)–0.378447 (90.1)454 (91.5)–0.049
Beta-blocker4,706 (62.2)682 (86.0)–0.564386 (77.8)412 (83.1)–0.132
Mineralocorticoid antagonist1,492 (19.7)498 (62.8)–0.973290 (58.5)270 (54.4)0.081
Diuretic4,199 (55.5)700 (88.3)–0.782438 (88.3)424 (85.5)0.084
Digitalis721 (9.5)169 (21.3)–0.331113 (22.8)100 (20.2)0.064
No. of heart failure medications
Without medications175 (2.3)5 (0.6)0.1409 (1.8)5 (1.0)0.068
1 class1,730 (22.9)22 (2.8)0.63014 (2.8)18 (3.6)–0.046
2 class2,604 (34.4)91 (11.5)0.56870 (14.1)73 (14.7)–0.017
3+ class3,053 (40.4)675 (85.1)–1.044403 (81.3)400 (80.6)0.015
Recent diabetes treatment, n (%)
Metformin6,466 (85.5)363 (45.8)0.921277 (55.8)273 (55.0)0.016
Sulfonylurea3,156 (41.7)285 (35.9)0.119187 (37.7)183 (36.9)0.017
Meglitinide10 (0.1)2 (0.3)–0.0271 (0.2)1 (0.2)0.000
α-glucosidase inhibitor55 (0.7)7 (0.9)–0.0174 (0.8)5 (1.0)–0.021
Thiazolidinedione391 (5.2)23 (2.9)0.11618 (3.6)20 (4.0)–0.021
Dipeptidyl peptidase-4 inhibitor2,312 (30.6)547 (69.0)–0.832332 (66.9)330 (66.5)0.009
GLP-1 receptor agonist40 (0.5)10 (1.3)–0.0786 (1.2)7 (1.4)–0.018
Insulin1,111 (14.7)186 (23.5)–0.224132 (26.6)128 (25.8)0.018
No. of diabetes medications
Without medications570 (7.5)164 (20.7)–0.38489 (17.9)93 (18.8)–0.021
1 class2,675 (35.4)136 (17.2)0.42378 (15.7)78 (15.7)0.000
2 class2,553 (33.8)233 (29.4)0.094141 (28.4)142 (28.6)–0.004
3+ class1,764 (23.3)260 (32.8)–0.212188 (37.9)183 (36.9)0.021
Recent heart failure treatment, n (%)
RAS inhibitor without neprilysin inhibitor6,253 (82.7)364 (45.9)0.832310 (62.5)305 (61.5)0.021
Beta-blocker4,552 (60.2)703 (88.7)–0.690410 (82.7)423 (85.3)–0.071
Mineralocorticoid antagonist1,490 (19.7)483 (60.9)–0.926298 (60.1)288 (58.1)0.041
Diuretic3,656 (48.3)686 (86.5)–0.891428 (86.3)420 (84.7)0.046
Digitalis649 (8.6)125 (15.8)–0.22199 (20.0)81 (16.3)0.094
No. of heart failure medications
1 class2,266 (30.0)49 (6.2)0.65035 (7.1)33 (6.7)0.016
2 class2,675 (35.4)206 (26.0)0.205104 (21.0)117 (23.6)–0.063
3+ class2,621 (34.7)538 (67.8)–0.704357 (72.0)346 (69.8)0.049
Healthcare use, n (%)
Hospitalizations
04,611 (61.0)263 (33.2)0.580155 (31.3)174 (35.1)–0.081
1-22,397 (31.7)401 (50.6)–0.391247 (49.8)238 (48.0)0.036
3+554 (7.3)129 (16.3)–0.28094 (19.0)84 (16.9)0.053
Outpatient visits
0-273 (1.0)4 (0.5)0.0543 (0.6)4 (0.8)–0.024
3-5202 (2.7)16 (2.0)0.04310 (2.0)8 (1.6)0.030
6+7,287 (96.4)773 (97.5)–0.065483 (97.4)484 (97.6)–0.013

*Propensity score was estimated using a multivariable logistic regression model, which included all potential confounders shown in the Table 1 as independent variables. Assessed during the 365-day period before cohort entry. Assessed during the 30-day period before and at cohort entry. ARNi, angiotensin receptor/neprilysin inhibitor; ICD, intra-cardiac defibrillation; PS, propensity score; RAS, renin–angiotensin system; SD, standard deviation; SMD, standardized mean difference; SGLT2i, sodium-glucose cotransporter 2 inhibitor..



Within the propensity score matched cohort, we estimated the incidence rate of HF admission per 100 person-years, and HRs with 95% CIs using proportional subdistribution hazards model of Fine and Gray that treated death as a competing event (Fine and Gray 1999). Schoenfeld residual test was used to assess the proportional hazards assumptions before the survival analysis. As a sensitivity analysis, we adopted an intention-to-treat approach that did not censor follow-up at treatment interruption (switch or discontinuation). We performed subgroup analysis to investigate potential effect modification by sex, history of hospitalization for HF, chronic kidney disease, recent HF treatments. Interactions were tested using Wald test for heterogeneity. Propensity scores were re-estimated within each subgroup and the matching was reapplied for each comparison.

All analyses were conducted using SAS Enterprise Guide version 7.1 (SAS Institute Inc., USA). This study was approved by the institutional review board of Sungkyunkwan University, with a waiver of informed consent (SKKU 2022-04-015).

Results

We identified 8,355 eligible new users of SGLT2is (7,562, 90.5%) and ARNi (793, 9.5%) in our study cohort (Supplementary Fig. 1). During the follow-up period, there were 17 deaths in the SGLT2is group and 3 in the ARNi group. After propensity score matching, 496 matched pairs were included in the main analysis, with a mean age of 72.5 years and a male representation of 57.6%. Among these, there were 6 deaths in the SGLT2is group and 1 in the ARNi group during the follow-up period. All baseline covariates, except for baseline use of beta-blockers, achieved balance between groups after matching, with absolute standardized mean differences <0.1 (Table 1).

In the propensity score matched cohort, incidence rate of HF admission was 27.3 and 35.6 per 100 person-years in SGLT2is and ARNi group. When comparing the risk of HF admission associated with SGLT2is group with ARNi group, HR was 0.71 (95% CI 0.48-1.04) In the intention-to-treat analysis, HR was 0.71 (95% CI 0.49-1.02); the point estimate was nearly identical with main analysis, indicating that the effect of treatment initiation was similar to the effect of adhering to the treatment in this study.

We observed effect modification by history of hospitalization for HF (p-for-interaction = 0.002) and recent use of renin-angiotensin-system inhibitors (p-for-interaction = 0.005). Among patients with a history of hospitalization for HF, the propensity score matched HR for the risk of HF admission with SGLT2is compared to ARNi was 1.37 (95% CI 0.67-2.77), while among those without such history, it was 0.58 (95% CI 0.39-0.91). Regarding recent use of renin-angiotensin-system inhibitors, the propensity score matched HR for the risk of HF admission with SGLT2is compared to ARNi was 0.55 (95% CI 0.33-0.92) among users and 0.77 (95% CI 0.37-1.63) among non-users (Fig. 1 and 2).

Figure 1. Risk of HF admission associated with SGLT2 inhibitor versus ARN inhibitor, using propensity score matching among subgroups population. *Assessed during 365 days before cohort entry. Assessed during 30 days before or at cohort entry. Significant interactions for history of hospitalization for HF (p = 0.002), and recent use of renin–angiotensin-system inhibitors without neprilysin inhibitors (p = 0.005). ARNi, angiotensin receptor-neprilysin inhibitor; CI, confidence interval; HF, heart failure; HR, hazard ratio; IR, incidence rate; PS, propensity score; PY, person years; SGLT2i, sodium-glucose cotransporter 2 inhibitor.
Figure 2. Cumulative incidence curve comparing risk of heart failure admission with SGLT2i versus ARNi. Propensity score was re-estimated and re-matched within each stratified subgroup. Assessed during 365 days before cohort entry. Assessed during 30 days before or at cohort entry. Significant interactions for history of hospitalization for HF (p = 0.002), and recent use of renin–angiotensin system inhibitors without neprilysin inhibitors (p = 0.005). ARNi, angiotensin receptor-neprilysin inhibitor; CI, confidence interval; HF, heart failure; HR, hazard ratio; SGLT2i, sodium-glucose cotransporter 2 inhibitor.

Discussion

This is the nationwide population-based active comparator new-user cohort study that applied a target trial emulation framework to investigate the association between the risk of HF admission and the treatment with SGLT2is compared to ARNi. We have shown that there is no statistically significant difference in the risk of HF admission between SGLT2is- and ARNi-treated patients who have both type 2 diabetes and HF. Our findings from subgroup analysis suggests that the use of SGLT2is is associated with a reduced risk of HF admissions compared with ARNi, particularly in patients with a history of hospitalization for HF. Operating through mechanisms that involve the reduction of fluid accumulation and the promotion of favorable structural changes in the heart, SGLT2is may exhibit enhanced efficacy in advanced HF. Further research and clinical consideration are warranted to validate our findings.

This study has several limitations that should be considered when interpreting the findings. First, we acknowledge the substantial reduction in sample size resulting from our propensity score matching. This decision was driven by several considerations, including the relatively small comparator group of ARNi users within the larger cohort and significant differences in patient characteristics between treatment groups before matching. We applied matching to address these issues by creating a balanced cohort, enabling a more robust assessment of treatment effects while mitigating potential confounding. Second, it’s important to note that our study is based on data available up to December 31, 2020. This timeframe limits our ability to extrapolate findings beyond the guideline updates made around late 2021 or early 2022. Additionally, our findings are specific to patients with HF with type 2 diabetes and may not fully reflect the landscape of SGLT2i and ARNi use. Third, our study could not differentiate between specific types of HF as information on ejection fraction was unavailable for assessment in the data. Fourth, the potential for differences in censoring patterns between the treatment groups should be acknowledged. Last, despite our efforts in propensity score matching, the presence of more serious health conditions among ARNi group before matching indicates that imbalances may not have been fully addressed, raising the possibility of residual unmeasured confounding.

These preliminary real-world findings align with current guidelines for HF management (McDonagh et al. 2021; Bayés-Genís et al. 2022; Heidenreich et al. 2022), and previous network meta-analyses of trials (Aimo et al. 2021; Yan et al. 2021). However, a meta-analysis of 25 RCTs, including 47,275 individuals, reported a comparable risk of hospitalization for HF between SGLT2is and ARNi. Although these differences in point estimates are not statistically significant, they could be due to unmeasured confounding factors inherent in observational studies, which lack randomization (Teo et al. 2022). Future studies should utilize more recent data and improve control over residual confounders to validate our findings.

These results indicate that SGLT2is may provide comparable benefits to ARNi in the context of routine clinical care. This can potentially inform healthcare in clinical decision-making regarding HF management.

Supplementary Materials

Supplementary materials can be found via https://doi.org/10.58502/DTT.24.0001.

dtt-3-2-177-supple.pdf

Conflict of Interest

JYS received grants from the Ministry of Food and Drug Safety, the Ministry of Health and Welfare, the National Research Foundation of Korea, and pharmaceutical companies, including Pfizer, Celltrion, and SK Bioscience outside the submitted work. HEJ is employed by the Lunit Inc.

Acknowledgements

This study used data from the Health Insurance Review & Assessment Service of South Korea (M20210607316). This study was supported by a grant (No. 22213MFDS486) from the Ministry of Food and Drug Safety, South Korea, in 2022-2023. The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the manuscript. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.

Fig 1.

Figure 1.Risk of HF admission associated with SGLT2 inhibitor versus ARN inhibitor, using propensity score matching among subgroups population. *Assessed during 365 days before cohort entry. Assessed during 30 days before or at cohort entry. Significant interactions for history of hospitalization for HF (p = 0.002), and recent use of renin–angiotensin-system inhibitors without neprilysin inhibitors (p = 0.005). ARNi, angiotensin receptor-neprilysin inhibitor; CI, confidence interval; HF, heart failure; HR, hazard ratio; IR, incidence rate; PS, propensity score; PY, person years; SGLT2i, sodium-glucose cotransporter 2 inhibitor.
Drug Targets and Therapeutics 2024; 3: 177-184https://doi.org/10.58502/DTT.24.0001

Fig 2.

Figure 2.Cumulative incidence curve comparing risk of heart failure admission with SGLT2i versus ARNi. Propensity score was re-estimated and re-matched within each stratified subgroup. Assessed during 365 days before cohort entry. Assessed during 30 days before or at cohort entry. Significant interactions for history of hospitalization for HF (p = 0.002), and recent use of renin–angiotensin system inhibitors without neprilysin inhibitors (p = 0.005). ARNi, angiotensin receptor-neprilysin inhibitor; CI, confidence interval; HF, heart failure; HR, hazard ratio; SGLT2i, sodium-glucose cotransporter 2 inhibitor.
Drug Targets and Therapeutics 2024; 3: 177-184https://doi.org/10.58502/DTT.24.0001

Table 1 Baseline characteristics of patients initiating a SGLT2 inhibitor or an ARN inhibitor in 2020

CharacteristicBefore PS* matchingAfter PS matching
SGLT2iARNiSMDSGLT2iARNiSMD
Total7,562793496496
Age, years
Mean (SD)67.0 (12.5)72.0 (11.1)–0.42673.1 (11.3)72.0 (11.3)0.098
Sex, n (%)
Male4,129 (54.6)483 (60.9)–0.128283 (57.1)288 (58.1)–0.020
Female3,433 (45.4)310 (39.1)0.128213 (42.9)208 (41.9)0.020
Comorbidities, n (%)
Hospitalization for heart failure551 (7.3)268 (33.8)–0.695150 (30.2)138 (27.8)0.053
Hypertension4,954 (65.5)420 (53.0)0.257273 (55.0)277 (55.8)–0.016
Dyslipidemia2,949 (39.0)327 (41.2)–0.046219 (44.2)199 (40.1)0.082
Atrial fibrillation or flutter1,055 (14.0)156 (19.7)–0.153112 (22.6)102 (20.6)0.049
Myocardial infarction412 (5.4)98 (12.4)–0.24449 (9.9)52 (10.5)–0.020
Coronary artery disease1,813 (24.0)224 (28.2)–0.097126 (25.4)134 (27.0)–0.037
Coronary revascularization241 (3.2)58 (7.3)–0.18638 (7.7)37 (7.5)0.008
Aortic valve surgery7 (0.1)3 (0.4)–0.0592 (0.4)3 (0.6)–0.028
Pacemaker or ICD19 (0.3)9 (1.1)–0.1073 (0.6)2 (0.4)0.028
Stroke591 (7.8)86 (10.8)–0.10451 (10.3)52 (10.5)–0.007
Transient ischemic attack101 (1.3)6 (0.8)0.0576 (1.2)4 (0.8)0.040
Peripheral arterial disease38 (0.5)8 (1.0)–0.05810 (2.0)7 (1.4)0.047
Cancer491 (6.5)79 (10.0)–0.12747 (9.5)48 (9.7)–0.007
Chronic kidney disease396 (5.2)136 (17.2)–0.38572 (14.5)70 (14.1)0.012
Chronic pulmonary disease1,334 (17.6)197 (24.8)–0.177128 (25.8)125 (25.2)0.014
Chronic liver disease1,070 (14.1)103 (13.0)0.03466 (13.3)68 (13.7)–0.012
Comedications, n (%)
Statin5,988 (79.2)655 (82.6)–0.087393 (79.2)401 (80.8)–0.040
Calcium-channel blocker4,481 (59.3)337 (42.5)0.340246 (49.6)239 (48.2)0.028
Nonsteroidal anti-inflammatory drug5,049 (66.8)482 (60.8)0.125319 (64.3)311 (62.7)0.034
Systemic corticosteroid4,169 (55.1)414 (52.2)0.059272 (54.8)269 (54.2)0.012
Antidepressants1,540 (20.4)154 (19.4)0.024104 (21.0)99 (20.0)0.025
Antipsychotics1,821 (24.1)214 (27.0)–0.067142 (28.6)143 (28.8)–0.004
Baseline diabetes treatment, n (%)
Metformin6,045 (79.9)556 (70.1)0.228374 (75.4)368 (74.2)0.028
Sulfonylurea3,541 (46.8)400 (50.4)–0.072252 (50.8)245 (49.4)0.028
Meglitinide34 (0.4)4 (0.5)–0.0082 (0.4)3 (0.6)–0.028
α-glucosidase inhibitor147 (1.9)21 (2.6)–0.0479 (1.8)9 (1.8)0.000
Thiazolidinedione952 (12.6)65 (8.2)0.14451 (10.3)52 (10.5)–0.007
Dipeptidyl peptidase-4 inhibitor5,036 (66.6)680 (85.8)–0.461417 (84.1)411 (82.9)0.033
GLP-1 receptor agonist95 (1.3)14 (1.8)–0.0428 (1.6)8 (1.6)0.000
Insulin1,377 (18.2)305 (38.5)–0.461188 (37.9)183 (36.9)0.021
No. of diabetes medications
Without medications819 (10.8)33 (4.2)0.25520 (4.0)26 (5.2)–0.058
1 class1,107 (14.6)85 (10.7)0.11847 (9.5)50 (10.1)–0.020
2 class2,018 (26.7)237 (29.9)–0.071148 (29.8)144 (29.0)0.018
3+ class3,618 (47.8)438 (55.2)–0.148281 (56.7)276 (55.6)0.020
Baseline heart failure treatment, n (%)
RAS inhibitor without neprilysin inhibitor6,232 (82.4)748 (94.3)–0.378447 (90.1)454 (91.5)–0.049
Beta-blocker4,706 (62.2)682 (86.0)–0.564386 (77.8)412 (83.1)–0.132
Mineralocorticoid antagonist1,492 (19.7)498 (62.8)–0.973290 (58.5)270 (54.4)0.081
Diuretic4,199 (55.5)700 (88.3)–0.782438 (88.3)424 (85.5)0.084
Digitalis721 (9.5)169 (21.3)–0.331113 (22.8)100 (20.2)0.064
No. of heart failure medications
Without medications175 (2.3)5 (0.6)0.1409 (1.8)5 (1.0)0.068
1 class1,730 (22.9)22 (2.8)0.63014 (2.8)18 (3.6)–0.046
2 class2,604 (34.4)91 (11.5)0.56870 (14.1)73 (14.7)–0.017
3+ class3,053 (40.4)675 (85.1)–1.044403 (81.3)400 (80.6)0.015
Recent diabetes treatment, n (%)
Metformin6,466 (85.5)363 (45.8)0.921277 (55.8)273 (55.0)0.016
Sulfonylurea3,156 (41.7)285 (35.9)0.119187 (37.7)183 (36.9)0.017
Meglitinide10 (0.1)2 (0.3)–0.0271 (0.2)1 (0.2)0.000
α-glucosidase inhibitor55 (0.7)7 (0.9)–0.0174 (0.8)5 (1.0)–0.021
Thiazolidinedione391 (5.2)23 (2.9)0.11618 (3.6)20 (4.0)–0.021
Dipeptidyl peptidase-4 inhibitor2,312 (30.6)547 (69.0)–0.832332 (66.9)330 (66.5)0.009
GLP-1 receptor agonist40 (0.5)10 (1.3)–0.0786 (1.2)7 (1.4)–0.018
Insulin1,111 (14.7)186 (23.5)–0.224132 (26.6)128 (25.8)0.018
No. of diabetes medications
Without medications570 (7.5)164 (20.7)–0.38489 (17.9)93 (18.8)–0.021
1 class2,675 (35.4)136 (17.2)0.42378 (15.7)78 (15.7)0.000
2 class2,553 (33.8)233 (29.4)0.094141 (28.4)142 (28.6)–0.004
3+ class1,764 (23.3)260 (32.8)–0.212188 (37.9)183 (36.9)0.021
Recent heart failure treatment, n (%)
RAS inhibitor without neprilysin inhibitor6,253 (82.7)364 (45.9)0.832310 (62.5)305 (61.5)0.021
Beta-blocker4,552 (60.2)703 (88.7)–0.690410 (82.7)423 (85.3)–0.071
Mineralocorticoid antagonist1,490 (19.7)483 (60.9)–0.926298 (60.1)288 (58.1)0.041
Diuretic3,656 (48.3)686 (86.5)–0.891428 (86.3)420 (84.7)0.046
Digitalis649 (8.6)125 (15.8)–0.22199 (20.0)81 (16.3)0.094
No. of heart failure medications
1 class2,266 (30.0)49 (6.2)0.65035 (7.1)33 (6.7)0.016
2 class2,675 (35.4)206 (26.0)0.205104 (21.0)117 (23.6)–0.063
3+ class2,621 (34.7)538 (67.8)–0.704357 (72.0)346 (69.8)0.049
Healthcare use, n (%)
Hospitalizations
04,611 (61.0)263 (33.2)0.580155 (31.3)174 (35.1)–0.081
1-22,397 (31.7)401 (50.6)–0.391247 (49.8)238 (48.0)0.036
3+554 (7.3)129 (16.3)–0.28094 (19.0)84 (16.9)0.053
Outpatient visits
0-273 (1.0)4 (0.5)0.0543 (0.6)4 (0.8)–0.024
3-5202 (2.7)16 (2.0)0.04310 (2.0)8 (1.6)0.030
6+7,287 (96.4)773 (97.5)–0.065483 (97.4)484 (97.6)–0.013

*Propensity score was estimated using a multivariable logistic regression model, which included all potential confounders shown in the Table 1 as independent variables. Assessed during the 365-day period before cohort entry. Assessed during the 30-day period before and at cohort entry. ARNi, angiotensin receptor/neprilysin inhibitor; ICD, intra-cardiac defibrillation; PS, propensity score; RAS, renin–angiotensin system; SD, standard deviation; SMD, standardized mean difference; SGLT2i, sodium-glucose cotransporter 2 inhibitor.


References

  1. Aimo A, Pateras K, Stamatelopoulos K, Bayes-Genis A, Lombardi CM, Passino C, Emdin M, Georgiopoulos G (2021) Relative efficacy of sacubitril-valsartan, vericiguat, and SGLT2 inhibitors in heart failure with reduced ejection fraction: a systematic review and network meta-analysis. Cardiovasc Drugs Ther 35:1067-1076. doi: 10.1007/s10557-020-07099-2
    Pubmed CrossRef
  2. Bayés-Genís A, Aimo A, Metra M, Anker S, Seferovic P, Rapezzi C, Castiglione V, Núñez J, Emdin M, Rosano G, Coats AJS (2022) Head-to-head comparison between recommendations by the ESC and ACC/AHA/HFSA heart failure guidelines. Eur J Heart Fail 24:916-926. doi: 10.1002/ejhf.2542
    Pubmed CrossRef
  3. Dhruva SS, Redberg RF (2008) Variations between clinical trial participants and Medicare beneficiaries in evidence used for Medicare national coverage decisions. Arch Intern Med 168:136-140. doi: 10.1001/archinternmed.2007.56
    Pubmed CrossRef
  4. Fine JP, Gray RJ (1999) A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 94:496-509. doi: 10.1080/01621459.1999.10474144
    CrossRef
  5. Heidenreich PA, Bozkurt B, Aguilar D, Allen LA, Byun JJ, Colvin MM, Deswal A, Drazner MH, Dunlay SM, Evers LR, Fang JC, Fedson SE, Fonarow GC, Hayek SS, Hernandez AF, Khazanie P, Kittleson MM, Lee CS, Link MS, Milano CA, Nnacheta LC, Sandhu AT, Stevenson LW, Vardeny O, Vest AR, Yancy CW (2022) 2022 AHA/ACC/HFSA guideline for the management of heart failure: a report of the American College of Cardiology/American Heart Association Joint Committee on clinical practice guidelines. Circulation 145:e895-e1032. doi: 10.1161/CIR.0000000000001063
    Pubmed CrossRef
  6. Kim JA, Yoon S, Kim LY, Kim DS (2017) Towards actualizing the value potential of Korea Health Insurance Review and Assessment (HIRA) data as a resource for health research: strengths, limitations, applications, and strategies for optimal use of HIRA data. J Korean Med Sci 32:718-728. doi: 10.3346/jkms.2017.32.5.718
    Pubmed KoreaMed CrossRef
  7. Matthews AA, Danaei G, Islam N, Kurth T (2022) Target trial emulation: applying principles of randomised trials to observational studies. BMJ 378:e071108. doi: 10.1136/bmj-2022-071108
    Pubmed CrossRef
  8. McDonagh TA, Metra M, Adamo M, Gardner RS, Baumbach A, Böhm M, Burri H, Butler J, Čelutkienė J, Chioncel O, Cleland JGF, Coats AJS, Crespo-Leiro MG, Farmakis D, Gilard M, Heymans S, Hoes AW, Jaarsma T, Jankowska EA, Lainscak M, Lam CSP, Lyon AR, McMurray JJV, Mebazaa A, Mindham R, Muneretto C, Francesco Piepoli M, Price S, Rosano GMC, Ruschitzka F, Kathrine SA; ESC Scientific Document Group (2021) 2021 ESC guidelines for the diagnosis and treatment of acute and chronic heart failure. Eur Heart J 42:3599-3726. doi: 10.1093/eurheartj/ehab368
    Pubmed CrossRef
  9. Teo YN, Teo YH, Syn NL, Yoong CSY, Cheong AJY, Wee CF, Lim YC, Lee CH, Yeo TC, Chai P, Wong RCC, Lin W, Sia CH (2022) Comparing sacubitril/valsartan against sodium-glucose cotransporter 2 inhibitors in heart failure: a systematic review and network meta-analysis. Clin Drug Investig 42:1-16. doi: 10.1007/s40261-021-01098-3
    Pubmed CrossRef
  10. Yan Y, Liu B, Du J, Wang J, Jing X, Liu Y, Deng S, Du J, She Q (2021) SGLT2i versus ARNI in heart failure with reduced ejection fraction: a systematic review and meta-analysis. ESC Heart Fail 8:2210-2219. doi: 10.1002/ehf2.13313
    Pubmed KoreaMed CrossRef

Supplementary File

Stats or Metrics

Share this article on :

  • line