Cardiology · Medical Literature
Implantable Cardioverter-Defibrillator Therapy in Contemporary Heart Failure Patients: An Analysis From the EMPEROR-Reduced Trial.
Abstract
The benefit of an implantable cardioverter-defibrillator (ICD) in patients with heart failure with reduced ejection fraction on foundational therapy that includes a sodium-glucose cotransporter 2 inhibitor is unknown. This study aims to evaluate the impact of an ICD among participants enrolled in EMPEROR-Reduced (Empagliflozin Outcome Trial in Patients with Chronic Heart Failure and a Reduced Ejection Fraction). Propensity score matching was used to compare outcomes in patients with and without an ICD in the placebo and empagliflozin arms of the EMPEROR-Reduced trial. Time-to-event curves and Cox regression analysis were used to evaluate the risk of the primary endpoint of sudden cardiac death (SCD). A total of 1,674 patients were included in the analysis of whom 837 (429 on placebo and 408 on empagliflozin) with an ICD at baseline were matched to 837 (424 on placebo and 413 on empagliflozin) without an ICD. Background therapy included an angiotensin-converting enzyme inhibitor/angiotensin receptor blocker/angiotensin receptor blocker-neprilysin inhibitor (≥88%); beta-blockers (≥94%), and mineralocorticoid receptor antagonists (≥65%). Cox regression analysis showed ICD vs non-ICD treatment was associated with a lower risk of SCD in both the empagliflozin (HR: 0.59; 95% CI: 0.26-1.3) and placebo arms of the trial (HR: 0.31; 95% CI: 0.13-0.74), treatment-by-ICD interaction P = 0.30. In patients with heart failure with reduced ejection fraction enrolled in EMPEROR-Reduced, those with an ICD (vs non-ICD) experienced a lower risk of SCD regardless of empagliflozin treatment.
Clinical relevance
Among patients with heart failure with reduced ejection fraction (HFrEF) enrolled in the EMPEROR-Reduced trial and receiving contemporary quadruple guideline-directed medical therapy including an SGLT2 inhibitor, does pre-existing ICD implantation reduce the risk of sudden cardiac death (SCD), and does this effect differ by empagliflozin versus placebo assignment? __SNAPSHOT__ {"studyDesign":"Post-hoc propensity score–matched analysis of a randomized controlled trial (EMPEROR-Reduced)","setting":"Multinational; academic and community heart failure centers participating in the EMPEROR-Reduced trial (NCT03057977)","sampleSize":"n=1,674 propensity-matched patients: 837 with ICD at baseline (429 placebo, 408 empagliflozin) matched to 837 without ICD (424 placebo, 413 empagliflozin)","population":"Adults with chronic HFrEF enrolled in EMPEROR-Reduced; high background GDMT use: RAAS blockade (ACEi/ARB/ARNi) ≥88%, beta-blockers ≥94%, mineralocorticoid receptor antagonists ≥65%","intervention":"Baseline ICD presence (pre-existing device at enrollment)","comparator":"No ICD at baseline; analysis performed separately within placebo and empagliflozin arms of the parent trial","primaryOutcome":"Time to sudden cardiac death (SCD), assessed by Cox regression and time-to-event analysis","followUp":"Not reported in the abstract; follows the EMPEROR-Reduced trial duration (median approximately 16 months in the parent trial)","fundingSource":"Not explicitly reported in the abstract; the parent EMPEROR-Reduced trial was funded by Boehringer Ingelheim and Eli Lilly"}
Key findings
In propensity score–matched Cox regression analyses, ICD use (vs. no ICD) was associated with a lower risk of SCD in both treatment arms. In the placebo arm, the hazard ratio for SCD with ICD vs. no ICD was 0.31 (95% CI: 0.13–0.74), indicating a statistically significant reduction. In the empagliflozin arm, the hazard ratio was 0.59 (95% CI: 0.26–1.30), a directionally consistent but statistically non-significant reduction whose wide confidence interval crosses unity. The interaction p-value for treatment-by-ICD was 0.30, indicating no statistically significant modification of ICD benefit by empagliflozin versus placebo assignment. Absolute risk differences, NNT, and NNH are not reported in the abstract and cannot be calculated without event rates. Whether ICD use translated to a reduction in cardiovascular death or all-cause mortality is not reported in the abstract.
Practice implications
This analysis establishes that, in HFrEF patients receiving contemporary four-pillar GDMT including an SGLT2 inhibitor, baseline ICD implantation remained associated with a lower risk of SCD — and this benefit was not statistically modified by empagliflozin therapy. The finding in the placebo arm (HR 0.31) appears clinically meaningful, consistent with prior trial-era data supporting ICD benefit. The point estimate in the empagliflozin arm (HR 0.59) is directionally supportive but the wide 95% CI (0.26–1.30) and non-significant p-value preclude a definitive conclusion about ICD benefit on top of SGLT2 inhibition. Critically, this analysis does not establish whether ICD use reduces cardiovascular or all-cause mortality in the SGLT2 inhibitor era, a key clinical question left unanswered. The analysis is post-hoc and observational in nature despite propensity matching, limiting causal inference. Results apply to patients similar to those enrolled in EMPEROR-Reduced (predominantly NYHA class II–III, high GDMT rates); patients with NYHA class IV, device-naïve lower-risk phenotypes, or underrepresented demographic groups (including women and non-White patients, who are known to be underrepresented in HF device trials) may not be adequately reflected. Only the abstract was available for this summary, which limits full assessment of covariate balance, event numbers, and secondary endpoints. __LIMITATIONS__ Post-hoc observational design: ICD assignment was not randomized, and propensity matching cannot fully eliminate residual confounding (e.g., HF etiology, device indication, center-level practice variation).,Small number of SCD events expected given the matched sample size of 1,674 and median follow-up of approximately 16 months, resulting in wide confidence intervals — particularly in the empagliflozin arm (HR 0.59; 95% CI: 0.26–1.30).,SCD is a surrogate endpoint for survival; whether ICD benefit translates to a reduction in cardiovascular or all-cause mortality in the contemporary SGLT2 inhibitor era is not reported.,The non-significant interaction p-value (0.30) does not confirm equivalence of ICD benefit across treatment arms; the study is likely underpowered to detect a true interaction.,Patients without an ICD may represent a systematically different risk profile (e.g., recently diagnosed HF not yet meeting ICD criteria, or contraindications to device therapy), limiting comparability even after matching.,Generalizability is constrained by the clinical trial population (high GDMT use, selected enrollment criteria); real-world patients may have lower rates of background therapy and higher comorbidity burden.,Funding source of the sub-analysis is not reported in the abstract; the parent trial was industry-sponsored (Boehringer Ingelheim/Eli Lilly), which may influence analytical priorities.,Only the abstract was available; full methods, covariate selection for propensity matching, event counts, and secondary endpoints cannot be independently assessed. __BOTTOM_LINE_CATEGORY__ potentially_practice_informing __BOTTOM_LINE__ In propensity-matched HFrEF patients from EMPEROR-Reduced, baseline ICD implantation was associated with a lower risk of SCD regardless of empagliflozin use, with a statistically significant reduction observed in the placebo arm (HR 0.31) and a consistent but non-significant trend in the empagliflozin arm (HR 0.59). The non-significant treatment-by-ICD interaction (p=0.30) does not support empagliflozin as a substitute for ICD therapy in guideline-eligible patients, though the analysis is underpowered to formally test this interaction. Current practice of offering ICD implantation to eligible HFrEF patients should not be withheld based on SGLT2 inhibitor use pending confirmatory prospective data.