Meta-Analysis of SGLT2 Inhibitors in Transthyretin Cardiac Amyloidosis

Abstract

Background

Sodium glucose co-transporter 2 inhibitors (SGLT2i) improve outcomes in heart failure (HF), but evidence in transthyretin cardiac amyloidosis (ATTR-CM)—particularly in patients not receiving disease-modifying therapy (DMT)—remains limited.

Objectives

This study aimed to evaluate the efficacy and safety of SGLT2i in ATTR-CM, stratified by DMT.

Methods

PubMed, Embase, and Cochrane were systematically searched for studies evaluating SGLT2i in ATTR-CM. Primary outcomes were all-cause mortality and HF hospitalization; secondary outcomes included safety events. Effect estimates were reported as HR or risk ratio with 95% CIs, with heterogeneity assessed by I².

Results

Fourteen observational studies (10.852 patients) were analyzed; 5.453 patients received SGLT2i and 43.6% had diabetes. Of 7.708 patients (k = 12) with known DMT status, 34.6% received DMT. Follow-up ranged from 3 months to 5.5 years. Treatment with SGLT2i was associated with a lower risk of all-cause mortality (k = 6; n = 5.135; risk ratio: 0.44; P < 0.001; I² = 53.9%), although 2 studies reported zero events (k = 4; n = 5.038). Similarly, the risk of HF hospitalization was reduced (k = 2; n = 1.438; HR: 0.62; P < 0.001; I² = 0%). Subgroup analyses stratified by DMT and diabetes status showed a consistent direction of association, although these findings require further validation. The pooled prevalence of adverse events (k = 7; n = 535) was 7.28% (I² = 76.6%).

Conclusions

In observational studies, SGLT2i therapy in ATTR-CM is associated with lower all-cause mortality and HF hospitalizations. These findings are clinically encouraging, particularly for patients not receiving DMT, in whom therapeutic options remain limited. Prospective randomized trials are warranted to confirm these findings and better define SGLT2i benefit according to DMT.

Central Illustration

Sodium glucose co-transporter 2 (SGLT2) inhibitors are established therapies in several clinical settings, including heart failure (HF), type 2 diabetes, and chronic kidney disease.1, 2, 3 In the context of HF, randomized clinical trials have evaluated SGLT2 inhibitors across the full spectrum of left ventricular ejection fraction, consistently demonstrating a reduction in the composite risk of HF hospitalization and cardiovascular mortality.⁴ However, certain specific etiologies of HF, including infiltrative cardiomyopathies, were underrepresented or excluded from pivotal SGLT2 inhibitor trials.^5,6 To date, the efficacy of SGLT2 inhibitors in patients with transthyretin cardiac amyloidosis (ATTR-CM) remains largely based on retrospective data: the larger retrospective study evaluating their effect in patients with ATTR-CM demonstrated a reduced risk of HF hospitalization and cardiovascular mortality, regardless of the ejection fraction, in patients with ATTR-CM.⁷ One recent meta-analysis reported excellent tolerability and an overall reduction in all-cause mortality, albeit with a neutral effect on hospitalization for worsening HF in patients with ATTR-CM treated with SGLT2 inhibitors.⁸ However, the effect of these agents across the whole spectrum of ATTR-CM patients, including those without background disease-modifying therapy (DMT) remains poorly defined.

To address this knowledge gap, we conducted a systematic review and meta-analysis to evaluate the efficacy and safety of SGLT2 inhibitors in patients with ATTR-CM, stratified by the use of concomitant DMT and by diabetes status.

Methods

Study design

This systematic review and meta-analysis was conducted according to the Meta-analysis of Observational Studies in Epidemiology guidelines and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.^9,10 The study protocol was registered in the PROSPERO (CRD420251060714).

Ethics statement

As this meta-analysis is based solely on published data, no ethical approval or informed consent from participants was required.

Data sources and search strategy

We systematically searched PubMed, Embase, and the Cochrane Library from inception to May 15, 2025. The search strategy combined keywords related to ATTR-CM and SGLT2 inhibitors using Boolean operators and keywords including terms for transthyretin amyloidosis (eg, ATTR cardiomyopathy, cardiac ATTR, ATTR-CM, amyloid cardiomyopathy) and SGLT2 inhibitors (eg, dapagliflozin, empagliflozin, canagliflozin). The full electronic search strategy is detailed in Supplemental Methods 1.

Eligibility criteria

Inclusion criteria consisted of observational studies and clinical trials involving adult patients (≥18 years) with ATTR-CM treated with SGLT2 inhibitors.

Studies were eligible if they reported at least one of the following outcomes:

• •All-cause mortality or HF hospitalization reported in both the treatment and control groups;
• •Cumulative prevalence of adverse events within the SGLT2 inhibitor treatment group.

Exclusion criteria for comparative efficacy analyses encompassed conference abstracts, case reports and case series, narrative or systematic reviews, editorials, commentaries, and expert opinions, clinical guidelines, protocol and dissertations, preclinical studies, and non-English publications.

For the safety synthesis (adverse event prevalence), case series providing extractable safety data were considered eligible and additionally included, whereas they were not considered in comparative efficacy analyses to ensure methodological consistency for effect size estimation.

The full list of inclusion and exclusion criteria is provided in Supplemental Methods 2.

Study selection and data collection process

Search results were screened using Rayyan reference manager software (Qatar Computing Research Institute, Doha, Qatar). After removal of duplicate records, 2 authors (A.M. and G.C.) independently screened titles and abstracts to identify potentially relevant studies. The full-text review provided to confirm the included studies based on predefined eligibility criteria. Discrepancies in study selection were resolved by a third author (L.B.). Data extraction was performed using a standardized worksheet. The extracted data included study characteristics (name of the first author, year of publication, total number of participants, intervention, design study), baseline demographic and clinical characteristics of the patients (age, gender, genetic background), cardiovascular comorbidities (diabetes status), DMT, and duration of follow-up.

Outcomes and data items

The two primary efficacy endpoints were all-cause mortality and HF hospitalization.

Subgroup analyses assessed these same efficacy outcomes in patients with and without concomitant DMT for transthyretin amyloidosis, and by diabetic status.

Secondary outcomes included safety endpoints, specifically the cumulative prevalence of adverse events potentially related to SGLT2 inhibitors, stratified by the concomitant use of DMT for ATTR-CM, and by diabetic status.

Statistical analysis

Categorical variables are presented as counts and percentages; continuous variables as mean ± SD, or as median and IQR, as appropriate. If continuous data were not directly available as mean values and standard deviations, in case of non-skewed distribution, conversion was performed from median and IQR according to the evidence from the literature.¹¹ If outcome event data were not directly reported, numerical values were extracted from measures of association (risk ratios [RRs]) following Cochrane guidelines.¹² The longest available follow-up data were extracted for each study. To minimize the risk of overlapping populations, when multiple studies derived from the same registry evaluated the same outcome, only the study with the largest cohort was retained. HRs for time-to-event outcomes or RRs with 95% CIs were used to summarize categorical outcomes, and all effect estimates were unadjusted. When available, unadjusted HRs were extracted as reported in the original studies. When HRs were not available, unadjusted RRs were calculated from event counts and sample size using standard Cochrane methods implemented in Review Manager (RevMan, Computer program, version 5.4, The Cochrane Collaboration, 2020). Given heterogeneity in follow-up duration, RRs should be interpreted as cumulative risk at the longest reported follow-up rather than time-to-event effects. Binary outcomes were pooled using the inverse variance method (metabin function from the R meta package). In this approach, studies with more precise estimates (smaller variance) are given greater weight in the overall effect estimate. 95% CI were calculated, and a random-effects model was applied to account for between-study heterogeneity, with tau² estimated using the restricted maximum likelihood method, implemented using the rma() function from the “metafor” R package. The pooled prevalence of cumulative adverse events was calculated using a random-effects model, after applying appropriate transformations. The choice of transformation was based on the literature¹² and was implemented with the metaprop() function from the “meta” R package. The Freeman–Tukey double arcsine transformation (function “PFT”) was used by default, allowing variance stabilization of studies with extreme proportions, including 0 or 1; this approach ensured robust and reproducible pooled estimates.¹³ Subgroup analyses were conducted to assess the consistency of results according to DMTs and diabetic status. When subgroup data were directly available, these were used; otherwise, studies with more than 50% of patients treated were classified as “on-treatment,” and those with <50% as “not-on-treatment.” The same threshold was applied to classify studies according to diabetic status, acknowledging that this ecological approximation may introduce misclassification bias. These subgroups were also used for the DMT-stratified estimates presented in the Central Illustration. In case of at least moderate heterogeneity, sensitivity analyses using a leave-one-out strategy were performed with the “metainf()” function to assess the influence of individual studies on the overall effect estimate. Meta-regression analysis was performed with the “metareg()” function from the meta package, referencing the meta-analysis object and the covariate of interest, using event rates and population size, corresponding to a cumulative risk analysis (unadjusted RR analysis). Bubble plots were generated using the “bubble()” function to visualize the association between covariates and treatment effect. Statistical heterogeneity across studies was assessed with the I² statistic; values of I² >30% were considered as indicating moderate to substantial heterogeneity. Publication bias was assessed by visual examination of funnel plots (function “funnel()”). A 2-tailed P value <0.05 was considered statistically significant. Statistical analyses were performed using Review Manager (RevMan, Computer program, version 5.4. The Cochrane Collaboration, 2020) and R (The R Foundation for Statistical Computing, 2024, R version 4.4.1, 2024-06-14). A summary of the statistical methods is provided in Supplemental Table 1.

Central Illustration.

SGLT2 Inhibitors and Disease-Modifying Therapy in Patients With ATTR-CM

Impact of SGLT2 inhibitors in ATTR-CM population stratified by presence or absence of disease-modifying therapy. Created using icons taken from BioRender (https://BioRender.com) and Freepik (https://www.freepik.com). ATTR-CM = transthyretin cardiac amyloidosis; DMT = disease-modifying therapy; TTR = Transthyretin; other abbreviation as in Figure 1.

Study risk for bias assessment

The risk of bias was assessed using the ROBINS-I tool (Risk Of Bias In Non-randomized Studies of Interventions) by 2 independent reviewers (F.G. and G.C.). Discrepancies in judgment were resolved by a third reviewer (L.B.). Each study was evaluated across 7 domains: bias due to confounding, selection of participants, classification of interventions, deviations from intended interventions, missing data, outcome measurement, and selection of reported results. An overall risk of bias rating—low, moderate, or serious—was assigned to each study.

Results

Study selection

The initial literature search identified 304 records. After removing 33 duplicates, 271 records were screened by title and abstract, leading to the exclusion of 244 studies. The full-text review of the remaining 27 articles was assessed for eligibility, resulting in the inclusion of 14 studies in the meta-analysis. A Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram illustrating the study selection process is presented in the Supplemental Figure 1, while a summary of the study extraction, describing characteristics of each study and main outcomes, is provided in Table 1.

Table 1.

Study Extraction and Key Results of the Included Studies

First Author (Year)	Study Design	Study Population (n)	Sample of the Groups SGLT2i vs No-SGLT2i	Outcomes	Events	Estimates of the Outcomes
Dobner 202314	RC	57	17 vs 40	All-cause mortality	0%	Absence of events
				Safety adverse events	0%	Absence of events
Chi 202515	RC	998	499 vs 499	All-cause mortality	NA	HR 0.76 (95% CI: 0.54-1.06); P = 0.108
				HF hospitalization	NA	↓ HR: 0.65 (95% CI: 0.46-0.92); P = 0.015
Porcari 20247	RC	440	220 vs 220	All-cause mortality	13.2% vs 37.3%	↓ HR 0.57 (95% CI: 0.37-0.89); P = 0.01 ↓ RR 0.35 (95% CI: 0.24-0.52); P < 0.001
				HF hospitalization	11.8% vs 33.6%	↓ HR 0.57 (95% CI: 0.36-0.91); P = 0.014
				Safety adverse events	4.5%	Prevalence SGLT2i group: 10 (4.5%)
Lang 202416	RC	182	87 vs 95	Safety adverse events	24.1%	Prevalence SGLT2i group: 21 (24.1%)
Jaiswal 202417	RC	4,306	2,153 vs 2,153	All-cause mortality	13.4% vs 25.1%	↓ RR 0.54 (95% CI: 0.47-0.61); P < 0.001
Byer 202518	RC	1,246	623 vs 623	All-cause mortality	NA	HR 0.83 (95% CI: 0.63-1.08); P = 0.17
				HF exacerbations	NA	↓ HR 0.74 (95% CI: 0.58-0.96); P = 0.021
Augusto 202519	RC	1878	939 vs 939	All-cause mortality	NA	↓ HR 0.64 (95% CI: 0.52-0.79)
				HF exacerbations	NA	HR 1.35 (95% CI: 0.95-1.92)
Sivamurugan 202520	RC	1,266	633 vs 633	All-cause mortality	NA	↓ OR 0.28 (95% CI: 0.20-0.39)
				HF exacerbations	NA	OR 1.07 (95% CI 0.85-1.34)
Ermolaev 202521	RC	40	20 vs 20	All-cause mortality	0%	Absence of events
Zampieri 202222	PS	15	15	Safety adverse events	6.7%	Prevalence SGLT2i group: 1 (6.7%)
Steinhardt 202423	RC	79	79	Safety adverse events	5.1%	Prevalence SGLT2i group: 4 (5.1%)
Pillai 202424	RC	53	53	Safety adverse events	9.4%	Prevalence SGLT2i group: 5 (9.4%)
Schwegel 202425	RC	116	51 vs 65	All-cause mortality	21.6% vs 41.5%	↓ RR 0.52 (95% CI: 0.29-0.94); P = 0.032
González-Montes 202426 202,426	RC	176	64 vs 112	All-cause mortality	9.4% vs 34.8%	↓ RR 0.27 (95% CI: 0.12-0.60); P = 0.001
				Safety adverse events	6.3%	Prevalence SGLT2i group: 4 (6.3%)

HF = heart failure; PS = patient series; RC = retrospective cohort; RR = risk ratio; SGLT2i = sodium glucose co-transporter 2 inhibitors.

Qualitative values are expressed as absolute numbers and percentages (%). Quantitative values are expressed as mean ± SD or as median and IQR. If the SD was not available, continuous data are presented as mean (or median if indicated with ∗). Omitted P values are considered nonapplicable. Additional data were retrieved via correspondence with the study authors.

Baseline characteristics

The 14 studies included a total of 10.852 patients with ATTR-CM. Among them, 5.453 patients (50.2%) were treated with SGLT2 inhibitors, while 5.399 patients (49.8%) did not receive this treatment and were considered as “controls.” Twelve studies (n = 7.708) reported baseline characteristics specifically describing age, sex, and diabetic status. The mean age across these cohorts was 75.9 (±11.2) years; 5.564 patients (72.2%) were male, and 3.359 (43.6%) had diabetes. Genetic background was assessed in 11 studies (n = 4.552). Of these patients, 3.531 (77.6%) were diagnosed with wild-type ATTR-CM, 165 (3.6%) with hereditary ATTR-CM variants, while in 856 (18.8%) genetic test results were not reported. Data on DMT were available from 12 studies (n = 7.708). Overall, 2.666 patients (34.6%) received such treatments: 2.598 (33.7%) were treated exclusively with TTR stabilizers (tafamidis or acoramidis), 7 (0.1%) with TTR silencers (patisiran), while in 61 patients (0.8%) the specific class of therapy was not reported. The duration of follow-up—reported as median or mean, depending on the study—ranged from 3 months to 5.5 years. A comprehensive summary of baseline characteristics of the population is provided in Table 1, while Supplemental Table 2 provides an overview of the studies included in the pooled analyses, presented separately for each outcome and subgroup.

Pooled analyses of all included studies: efficacy and safety outcomes

Nine studies reported data on all-cause mortality in patients with ATTR-CM. After exclusion of potentially overlapping cohorts, 6 studies (k = 6; n = 5.135) provided effect estimates with a common measure of association. As 2 studies did not present events in both arms, accounting 0% in weight, they were not included in the pooled analysis.^14,21 Accordingly, 4 studies (k = 4; n = 5.038) were included in the pooled analysis. SGLT2 inhibitor therapy was associated with reduced all-cause mortality (RR: 0.44; 95% CI: 0.33-0.59; P < 0.001; I² = 53.9%) (Figure 1A). Heterogeneity was substantial; however, leave-one-out sensitivity analysis confirmed the robustness of the findings, as no single study significantly altered the overall effect estimate (Supplemental Figure 2). The omission of Jaiswal et al, reduced the heterogeneity to 0%, therefore suggesting that this study largely contributed to the observed heterogeneity between studies.¹⁷

Figure 1.

Sodium Glucose Co-Transporter 2 Inhibitors on Efficacy and Safety Outcomes in Patients With Transthyretin Cardiac Amyloidosis

Forest plots for all-cause mortality (A) and heart failure hospitalization (B). (C) The pooled prevalence of safety adverse events related to SGLT2 inhibitor therapy. For pairwise comparisons, a random-effects meta-analysis using REML method was performed. REML = restricted maximum likelihood; RR = risk ratio; SGLT2i = sodium glucose co-transporter 2 inhibitors.

Two studies (k = 2; n = 1.438) reported outcomes related to HF hospitalizations, showing a lower risk of HF hospitalizations in patients treated with SGLT2 inhibitors compared with untreated individuals (HR: 0.62; 95% CI: 0.47-0.82; P < 0.001; I² = 0%) (Figure 1B). Given the limited number of studies, these findings should be interpreted cautiously. However, the same pooled estimate was observed when used a fixed-effects model instead of a random-effects model (HR: 0.62; 95% CI: 0.47-0.82; P < 0.001; I² = 0%) (Supplemental Figure 3).

Adverse events associated with SGLT2 inhibitor therapy were reported in 7 studies (k = 7; n = 535), with a pooled prevalence of 7.28% (Figure 1C). However, substantial heterogeneity was observed (I² = 76.6%), and the relatively small sample size limits generalizability of these findings. The most frequently observed events included genitourinary infections or symptoms, mild hypotension, and episodes of acute kidney injury, although definitions and reporting methods often lack consistency across studies. Overall, 45 patients experienced adverse events. Of these, 11 developed acute kidney injury or a reduction in renal function, classified as serious adverse events, whereas 34 were considered nonserious adverse events, and included hypotension, constipation, and genitourinary infections. The full details of safety adverse events are reported in Supplemental Table 3. Leave-one-out sensitivity analysis showed that omitting Lang et al¹⁶ reduced heterogeneity to 0% (Supplemental Figure 4), suggesting that this study largely contributed to the observed heterogeneity between studies.

Main subgroup analysis: impact of SGLT2i on efficacy and safety by DMT

Subgroup analyses stratified patients by use of concomitant DMT. Among patients receiving such therapies (k = 2, n = 1.112), including tafamidis and patisiran, treatment with SGLT2 inhibitors was associated with a reduction in all-cause mortality (HR: 0.72; 95% CI: 0.52-0.99; P = 0.04; I² = 0%) (Figure 2A). The pooled estimate remained the same using a fixed-effects model instead of a random-effects model (HR: 0.72; 95% CI: 0.52-0.99; P = 0.04; I² = 0%) (Supplemental Figure 5). However, given the limited number of studies, these findings should be interpreted cautiously.

Figure 2.

Sodium Glucose Co-Transporter 2 Inhibitors on All-Cause Mortality Outcomes Stratified by Disease-Modifying Therapy

Forest plots for all-cause mortality in patients treated with disease-modifying therapy (A) and in those not receiving disease-modifying therapy (B). Meta-regression analysis (C). For pairwise comparisons, a random-effects meta-analysis using REML method was performed. Abbreviations as in Figure 1.

In patients not treated with DMT (k = 3, n = 4.598), SGLT2 inhibitors resulted associated with a reduction in all-cause mortality (RR: 0.52; 95% CI: 0.44-0.61; P < 0.001; I² = 27.1%) (Figure 2B). The denominators reported in the forest plots reflect the subset of patients with available time-to-event data, expressed as HR for each subgroup (Figure 2A), or as RR when time-to-event outcomes were not reported (Figure 2B). These denominators may differ from the total population reported in Tables 1 and 2 because only patients with available information on DMT and clinical outcomes were included in the specific subcohort For example, in Porcari et al, 2024, this difference arises because some patients may have started the DMT after baseline, so the forest plot includes only those contributing to the time-to-event analyses.⁷

Table 2.

Baseline Characteristics of the Population

First Author (Year)	Study Design	Mean Follow-Up (Months)	Number of Patients		Mean Age		Gender Male		Diabetes		ATTR-CM WT		ATTR-CM Variants		Disease-Modifying Therapy
			SGLT2i	Control	SGLT2i	Control	SGLT2i	Control	SGLT2i	Control	SGLT2i	Control	SGLT2i	Control	SGLT2i	Control
Dobner 202314	RC	3	17	40	79.8 ± 4	78.9 ± 5.8	17 (100%)	38 (95%)	5 (29.4%)	2 (6.3%)	15 (88.2%)	38 (97.4%)	2 (11.8%)	1 (2.6%)	17 (100%)	40 (100%)
Chi 202515	RC	SGLT2i + 12∗ SGLT2i − 45∗	499	499	76.1 ± 10.3	76.3 ± 11	403 (80.8%)	393 (78.8%)	136 (27.3%)	131 (26.3%)	69 (13.8%)	75 (15%)	10 (2%)	15 (3%)	499 (100%)	499 (100%)
Porcari 20247	PO	28 (18-45)	220	220	77.2 ± 7.5	76.7 ± 7.7	199 (90.5%)	196 (89.1%)	89 (40.5%)	94 (42.7%)	181 (82.3%)	174 (79.1%)	39 (17.7%)	46 (20.9%)	46 (20.9%)	48 (21.8%)
Lang 202416	RC	5.6 (2.9-8.1)	87	95	80 ± 10	79 ± 10	78 (89.7%)	83 (87.4%)	10 (11.5%)	4 (4.2%)	76 (87.4%)	76 (80%)	11 (12.6%)	19 (20%)	87 (100%)	95 (100%)
Jaiswal 202417	RC	36	2,153	2,153	74.2 ± 10.8	74.4 ± 14.1	1,392 (64.7%)	1,401 (65.1%)	1,149 (53.4%)	1172 (54.4%)	NA	NA	NA	NA	281 (13.1%)	241 (11.2%)
Byer 202518	RC	60	623	623	77.9 ± 7.6	78.3 ± 8.5	491 (78.8%)	502 (80.6%)	220 (35.3%)	223 (35.8%)	623 (100%)	623 (100%)	0 (0%)	0 (0%)	297 (47.7%)	293 (47.0%)
Augusto 202519	RC	60	939	939	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA
Sivamurugan 202520	RC	60	633	633	NA	NA	NA	NA	NA	NA	633 (100%)	633 (100%)	0 (0%)	0 (0%)	NA	NA
Ermolaev 202521	PR	6 (4-8)	20	20	76.2 ± 7.1	80.8 ± 4.7	14 (70%)	17 (85%)	3 (15%)	3 (15%)	17 (85%)	19 (95%)	3 (15%)	1 (5%)	20 (100%)	20 (100%)
Zampieri 202222	PS	8 ± 3	15	0	80.4 ± 5.7		14 (93%)		15 (100%)		13 (86%)		2 (14%)		0 (0%)
Steinhardt 202423	RC	8.2 (6.8-13)	79	0	79.6 ± 6.8		57 (72%)		17 (21.5%)		66 (86%)		NA		60 (75.9%)
Pillai 202424	RC	19 ± 9	53	0	81 ± 8		38 (72%)		17 (32%)		39 (74.6%)		14 (26%)		45 (84.9%)
Schwegel 202425	PR	31.2∗	51	65	79.3 ± 4.6	80 ± 4.5	46 (90%)	53 (82%)	10 (19.6%)	13 (20%)	NA	NA	NA	NA	8 (15.7%)	9 (13.8%)
González-Montes 202426	PR	65.8∗	64	112	80.5 ± 4.5	82.9 ± 5.6	53 (82%)	79 (70%)	25 (39.1%)	21 (18.8%)	63 (98.4%)	98 (87.5%)	0 (0%)	2 (1.8%)	26 (40.6%)	35 (31.3%)

ATTR-CM = transthyretin cardiac amyloidosis; PO = prospective observational; PR = prospective registry; WT = wild-type; other abbreviations as in Table 1.

Qualitative values are expressed as absolute numbers and percentages (%). Quantitative values are expressed as mean ± SD or as median and IQR. If the SD was not available, continuous data are presented as mean (or median if indicated with ∗). Additional data were retrieved via correspondence with the study authors.

The exploratory mixed-effects meta-regression on all-cause mortality (k = 4; n = 5.038) (Figure 2C), for each study based on events counts, sample size populations, and percentage of patients on DMT, suggested a negative association between the proportion of patients receiving DMT and the effect size of SGLT2i (coefficient −3.59; P = 0.0129), with no residual heterogeneity observed (I² = 0%). This finding suggests that higher proportions of DMT were associated to a reduced effect size of SGLT2i. However, given the very small number of studies included and the exploratory nature of the analysis, these findings should be interpreted with caution and considered hypothesis-generating rather than confirmatory.

In those receiving DMT (k = 2; n = 1.112), although limited in size, SGLT2 inhibitor use was associated to a lower risk of HF hospitalizations (HR: 0.58; 95% CI: 0.35-0.93; P = 0.02; I² = 23%) (Figure 3). A similar association of benefit was observed using a fixed-effects model instead of a random-effects model (HR: 0.61; 95% CI: 0.44-0.84; P = 0.003; I² = 23%) (Supplemental Figure 6).

Figure 3.

Sodium Glucose Co-Transporter 2 Inhibitors on Heart Failure Hospitalization in Patients With Transthyretin Cardiac Amyloidosis

Effect of SGLT2 inhibitors on heart failure hospitalization in ATTR-CM patients treated with disease-modifying therapy. For pairwise comparisons, a random-effects meta-analysis using REML method was performed. Abbreviations as in Figure 1.

Adverse events associated with SGLT2 inhibitor therapy were reported in 7 studies (k = 7; n = 535). The cumulative prevalence of adverse events was 7.28% and did not differ significantly between patients treated with DMT and those not treated (8.66% and 4.18%, respectively; P = 0.345 for subgroup differences) (Figure 4). However, this analysis is limited by small sample size and heterogeneous reporting, that may limit representativeness of the pooled estimate.

Figure 4.

Adverse Events Associated With Sodium Glucose Co-Transporter 2 Inhibitors in Patients With Transthyretin Cardiac Amyloidosis

Prevalence of adverse events associated with SGLT2 inhibitors in ATTR-CM patients stratified by presence or absence of background disease-modifying therapy. Abbreviation as in Figure 1.

Subgroup analysis: impact of SGLT2i on efficacy and safety by diabetic status

A subgroup analysis of all-cause mortality according to diabetic status was conducted (Supplemental Figure 7).

One study was excluded from the analysis because of the absence of events in both groups.¹⁴ The nondiabetic subgroup included 3 studies (k = 3; n = 732), whereas the diabetic subgroup was represented by a single study (k = 1; n = 4.306), thus limiting interpretability of subgroup comparisons. Effect estimates show a reduction in all-cause mortality in both subgroups (nondiabetic: RR: 0.37; 95% CI: 0.28-0.50; I² = 0%; diabetic subgroup: RR: 0.54; 95% CI: 0.47-0.61). A significant difference between the subgroups was detected (P = 0.032); however, this finding is exploratory, given the presence of a single study for the diabetic subgroup, and should therefore be interpreted with caution given the structure of the evidence.

The exploratory meta-regression (k = 4; n = 5.038) showed a trend linking diabetes to treatment effect, with wide uncertainty and substantial risk of overinterpretation. The model did not explain the observed heterogeneity (R² = 0%) and the bubble plot illustrates the uncertainty, showing a nonsignificant positive trend without evidence of a consistent effect (Supplemental Figure 8).

Risk of bias and publication bias assessment

The risk of bias in the 11 nonrandomized studies comparing patients treated with SGLT2i to those not receiving SGLT2i was assessed using the ROBINS-I tool. Eight studies were judged moderate risk and 3 serious risks.^14,16,21 Confounding domain was the most frequent serious source of bias, mainly due to inadequate adjustment for baseline differences in comorbidities, concomitant medications, and follow-up duration. Selection bias was commonly rated moderate, reflecting the observational design. Other domains were low risk, except outcome measurement, moderately affected in some studies due to semisubjective endpoints, influenced by clinician judgment and rarely blinded. A summary of domain- and overall-level assessments is provided in Supplemental Figure 9. In addition, assessment of publication bias was limited for both efficacy and safety outcomes. For all-cause mortality, the funnel plot was underpowered due to the small number of included studies (k = 4), precluding any meaningful assessment of publication bias (Supplemental Figure 10). The funnel plot for adverse events (k = 7) showed a slight visual asymmetry (Supplemental Figure 11); however, the small number of studies and events substantially limits the reliability of visual inspection, and no robust conclusions regarding publication bias can be drawn.

Discussion

SGLT2 inhibitors have become a cornerstone therapy in HF across the spectrum of left ventricular ejection fraction, with robust evidence from randomized clinical trials demonstrating reductions in HF hospitalizations and cardiovascular mortality.27, 28, 29 However, patients with ATTR-CM have traditionally been excluded from these trials, resulting in a substantial knowledge gap regarding the efficacy and safety of SGLT2 inhibitors in this population.³⁰ In contrast to conventional HF therapies—which are often poorly tolerated in ATTR-CM due to restrictive pathophysiology, autonomic dysfunction, and associated comorbidities³¹—SGLT2 inhibitors appear to have a favorable tolerability profile.¹⁶ These molecules exert beneficial natriuretic effects without causing significant reductions in blood pressure, enhance myocardial energetics, and exhibit nephroprotective, antioxidant, and antifibrotic properties.³² Collectively, these mechanisms provide a strong pathophysiological rationale for their use in patients with ATTR-CM. The present meta-analysis provides an updated real-world synthesis on the effectiveness and safety of SGLT2 inhibitors in ATTR-CM patients, specifically exploring for the first time these outcomes in presence and absence of concomitant DMT.

Impact of SGLT2 inhibitors on efficacy outcomes

Our pooled analysis shows that SGLT2 inhibitors use in ATTR-CM patients is associated with a lower risk of all-cause mortality, corresponding to a 56% lower relative risk reduction compared to untreated controls. These findings are consistent with a recent meta-analysis by Laborante et al which reported excellent tolerability and an overall reduction in all-cause mortality, albeit with a neutral effect on hospitalization for worsening HF.⁸ Importantly, in our study, SGLT2 inhibitor therapy was associated with a 38% relative risk reduction in HF hospitalizations. This result, although relatively limited in population size (k = 2, n = 1.438), may reflect an improved volume status, mitigation of diuretic resistance, and enhanced myocardial energetics, as suggested by several included studies.^7,16

Impact of concomitant DMT on the effectiveness and safety of SGLT2 inhibitors

A key objective of this study was to explore whether the association between SGLT2 inhibitors and clinical outcomes differs according to concomitant DMT use. Although these analyses are limited by the small number of studies and methodological heterogeneity, an important observation is that the direction of association remained favorable both in patients receiving and not receiving DMT. Among ATTR-CM patients not treated with DMT, SGLT2 inhibitor use was associated with a 48% lower unadjusted relative risk of mortality (RR: 0.52), while a 28% unadjusted relative risk reduction was appreciated in DMT-treated patients (HR: 0.72). While formal comparisons between subgroups are not methodologically robust—given differences in effect measures, sample size, and study design—the absence of a signal suggesting harm or lack of benefit in either subgroup is clinically relevant. In particular, the consistent direction of association in patients not treated with DMT is noteworthy, as this subgroup represents a substantial proportion of real-world ATTR-CM patients who remain reliant solely on supportive HF therapies. Meta-regression analyses exploring the relationship between DMT use and treatment effect were limited by the small number of included studies and should be considered exploratory. Therefore, no definitive conclusions can be drawn regarding effect modification by DMT. Subgroup analysis revealed a positive effect of SGLT2 inhibitors on HF hospitalization in DMT-treated patients, although this observation is limited to the available evidence, which precluded exploratory analysis for patients not receiving DMT. Similarly, the analysis according to diabetic status was constrained by the structure of the available data, with the diabetic subgroup largely represented by a single study. Although no clear signal of differential effect emerged, the limited evidence base and the use of study-level classification preclude definitive conclusions.

No safety signals or pharmacological interactions between SGLT2 inhibitors and disease-specific agents were observed in the available data. The subgroup analysis showed that the cumulative prevalence of adverse events did not differ significantly between patients treated and untreated with DMT (8.66% vs 4.18%, respectively).

To our knowledge, this study represents the first meta-analysis specifically evaluating the effect of SGLT2 inhibitors in ATTR-CM patients stratified by the presence or absence of concomitant DMT. This distinction is particularly relevant given the limited HF treatment options available for patients already on transthyretin-targeted therapies, as well as the substantial proportion of ATTR-CM patients who remain untreated due to factors such as advanced age, frailty, socioeconomic barriers, or advanced NYHA functional class. In this latter group, in which the supportive HF care is the only therapeutic option, further defining the potential clinical benefits SGLT2 inhibitors is of a major importance to address the current lack of prognostic treatments in this population.

Methodological considerations and study limitations

The evidence synthesized in this analysis must be interpreted within the context of the methodological limitations of the included studies. All available data derive from observational research, consisting of 10 retrospective analyses and 4 prospective registry or cohort studies. By design, these studies cannot definitively establish causality and remain susceptible to various biases, such as confounding by indication and selection bias. Regarding safety, data were inconsistently reported across studies, with several large cohorts not providing systematic adverse event ascertainment; therefore, the available data may not fully capture the true adverse-event burden in the broader observational literature and the variability in adverse-event definitions limits comparability across studies. A consistent heterogeneity in study design, analytic methods, patient populations, and the use of different effect unadjusted effect measures (HRs for time-to-event outcomes and RRs for cumulative outcomes) preclude direct comparability across studies. Given the heterogeneity in follow-up duration (3 months to 5.5 years), RRs should be interpreted as cumulative event risk at the longest available follow-up rather than as time-to-event estimates. Variability was particularly evident for HF hospitalizations, likely influenced by differences in outcome definitions, adjudication methods, and follow-up duration. Population heterogeneity may also represent a significant limitation, with hereditary ATTR-CM underrepresented in all analysis. A methodological limitation arises from our use of a 50% threshold to classify studies as either “on-treatment” or “not-on-treatment” when subgroup-specific data were unavailable. This arbitrary cutoff may have introduced misclassification bias. Finally, some of the analyses were based on a limited number of studies, which may affect the robustness of the findings. In particular, heterogeneity estimates should be interpreted with caution when ≤5 studies are available, as they are likely to be imprecise. Prospective, adequately powered trials with standardized endpoints and longer follow-up are warranted to confirm these findings and more accurately define the therapeutic role of SGLT2 inhibitors in this population.

Conclusions

In this comprehensive meta-analysis of real-world evidence, SGLT2i therapy in patients with ATTR-CM was associated with lower all-cause mortality and reduced risk of HF hospitalizations. This association, although exploratory and derived from limited data, appeared consistent across subgroups of patients treated and not treated with DMT. These findings are clinically encouraging, particularly for patients not receiving DMT due to unmet therapeutic need. Prospective randomized trials are warranted to confirm these associations and define the optimal therapeutic integration of SGLT2 inhibitors in ATTR-CM care.

Perspectives.

COMPETENCY IN PATIENT CARE: SGLT2i might be considered as adjunctive therapy to disease-modifying treatments or as standalone therapy in ATTR-CM. Randomized clinical trials are needed to define their role, particularly in patients not receiving DMT, a population with substantial unmet clinical need.

TRANSLATIONAL OUTLOOK: The observed association between SGLT2i therapy and improved clinical outcomes in ATTR-CM warrants mechanistic studies to better understand their effects and optimize therapeutic strategies.

Funding support and author disclosures

The authors have reported that they have no relationships relevant to the contents of this paper to disclose.