Input From the Clinical Experts Consulted by CADTH

According to the clinical experts consulted by CADTH, many patients with HFrEF are not being assessed by specialists in Canada, and assistance from other specialists is needed, given the growing number of patients. The clinical experts further noted that the use of goal-directed guideline-recommended pharmacological therapy and medical devices in patients with HFrEF remains suboptimal. The clinical experts highlighted that current treatment strategies in HFpEF are limited to supportive therapies focusing on symptom control rather than morbidity or mortality benefit, including ARBs, MRAs, and ARNIs, while data on SGLT2 inhibitors show clear benefit in this population. The clinical experts indicated that empagliflozin can be used as an alternative to dapagliflozin in combination with other goal-directed guideline-recommended pharmacotherapy in patients with HFrEF, and it is likely to be a first-line therapy for patients with HFpEF, given the ease of use, strength of evidence, safety profile, and familiarity with the use of empagliflozin in patients with type 2 diabetes mellitus. The population least likely to benefit from empagliflozin treatment are patients with low N-terminal pro-brain natriuretic peptide (NT-proBNP) levels and those with NYHA classes I and IV due to limited clinical evidence. The clinical experts indicated that the response to therapy in clinical practice is assessed based on the frequency of hospitalizations for HF, which in turn may lead to a reduction in mortality, improved quality of life, and a slower decline in kidney function. The clinical experts further noted that admission for HF is a major cost burden in the health care system. The clinical experts identified the following factors to consider when deciding to discontinue treatment with empagliflozin:

• the development of severe kidney dysfunction (estimated glomerular filtration rate [eGFR] < 15 mL/min/1.73 m²) and euglycemic diabetic ketoacidosis in patients with diabetes as important AEs
• NYHA functional class IV.

The clinical experts highlighted that empagliflozin is already widely used by primary care providers and endocrinologists for the management of diabetes, by nephrologists to reduce decline in kidney function, and by cardiologists.

Clinician Group Input

No clinician group input was received for this review.

Description of Studies

Two phase III, double-blind, placebo-controlled randomized controlled trials (EMPEROR-Reduced and EMPEROR-Preserved) were pivotal trials and included in the systematic review. Both trials were multinational and multicentre and included Canadian sites. The EMPEROR-Reduced trial (N = 3,730) was designed to assess the superiority of empagliflozin at 10 mg compared with matched placebo as an adjunct to SOC treatment in patients with HFrEF (LVEF ≤ 40%). In EMPEROR-Reduced, patients had a mean age of 66.8 years (standard deviation [SD] = 11.0 years), 76.1% were male, and the mean LVEF was 27.5% (SD = 6.0%), and most patients (75.1%) had an NYHA functional class of II. The EMPEROR-Preserved trial (N = 5,988) was designed to assess the superiority of empagliflozin at 10 mg compared with matched placebo as an adjunct to SOC treatment in patients with HFpEF (LVEF > 40%). In EMPEROR-Preserved, patients had a mean age of 71.9 years (SD = 9.4 years), 55.3% were male, the mean LVEF was 54.3% (SD = 8.8%), and most patients (81.5%) had an NYHA functional class of II.

In both EMPEROR trials, the primary efficacy end point was the time to first event of adjudicated cardiovascular (CV) death or hospitalization for heart failure (HHF). The key secondary end points were occurrence of adjudicated HHF (first and recurrent) and eGFR (calculated using Chronic Kidney Disease Epidemiology Collaboration creatinine [CKD-EPIcr] equation) slope of change from baseline. Other secondary and further exploratory outcomes in either trial that were important to the CADTH review included other hospitalization-related and mortality outcomes, as well as patient-reported outcomes such as health-related quality of life (HRQoL) and HF symptoms assessed by the Kansas City Cardiomyopathy Questionnaire (KCCQ) and 5-Level EQ-5D (EQ-5D-5L) questionnaires, and functional ability. Harms and notable harms were assessed.

Efficacy Results

A summary of the results for the main efficacy and safety outcomes of both EMPEROR trials is presented in Table 2. Statistical testing in both pivotal trials was conducted based on a hierarchical testing procedure. The following outcomes were controlled for multiplicity in both EMPEROR trials: time to first event of adjudicated CV death or HHF, occurrence of HHF (fist and recurrent), and eGFR (CKD-EPIcr equation) equation slope of change from baseline. The clinical experts consulted by CADTH for this review indicated that both HHF and CV death are the most important outcomes to assess the treatment response in patients with HF, while change in eGFR is not commonly used in clinical practice. Other secondary and further end points were tested in a non-hierarchical fashion without adjustments for multiplicity.

Adverse Events

In EMPEROR-Reduced, 1,420 (76.2%) patients in the empagliflozin group and 1,463 (78.5%) patients in the placebo group experienced at least 1 adverse event (AE). Patients in the empagliflozin and placebo groups experienced treatment-emergent AEs (TEAEs) at a similar frequency (15.2% and 12.2%, respectively). The most common TEAEs occurring in at least 0.5% of patients in the empagliflozin and placebo groups were hypotension (2.3% versus 1.8%, respectively), renal impairment (1.4% and 1.1%, respectively), urinary tract infection (1.4% in each group), and ||||||||||||| ||||| ||| ||||| |||||||||||||.

In EMPEROR-Preserved, 2,574 patients (85.9%) in the empagliflozin group and 2,585 patients (86.5%) in the placebo group experienced at least 1 AE. |||||||| || ||| ||||||||||||| ||| ||||||| |||||| ||||||||||| ||||| || | ||||||| ||||||||| |||||| ||| |||||| |||||||||||||| ||| |||| |||||| ||||| ||||||||| || || ||||| |||| || |||||||| || ||| ||||||||||||| || ||||||| |||||| |||| ||||||| ||||| ||||||||| ||||| ||| ||||| |||||||||||||| ||||||||||| ||||| ||| ||||| |||||||||||||| ||||| |||||||||| ||||| ||| ||||| |||||||||||||| ||| ||||||||||||| ||||| || |||| ||||||.

In EMPEROR-Reduced, 772 patients (41.4%) in the empagliflozin group and 896 patients (48.1%) in the placebo group experienced 1 or more serious AEs (SAEs). In EMPEROR-Preserved, 1,436 patients (47.9%) in the empagliflozin group and 1,543 patients (51.6%) in the placebo group experienced 1 or more SAEs.

Withdrawals Due to Adverse Events

In EMPEROR-Reduced, the overall frequency of AEs leading to treatment discontinuation was similar between the treatment groups in both pivotal trials (17.3% and 17.6% in the empagliflozin and placebo groups in EMPEROR-Reduced, ||| ||||| ||| ||||| || |||||||||||||||||| |||||||||||||. The most frequently reported types of withdrawals due to AEs in both trials were cardiac failure, death, acute myocardial infarction, renal impairment, and urinary tract infection.

Mortality

|| ||| ||||||||||||||| |||||| ||| |||||||||| || ||||| ||| |||||| ||| |||||||||||| ||||||||| ||||| ||| ||||||| |||||| ||| ||||||||| |||||| ||||||| || |||||||||||||||||| |||| || ||| || ||| ||||||||||||| ||||| ||| |||| || ||| ||||||| |||||| |||||||| || ||||||

Notable Harms

The frequency of notable harms identified in the protocol were comparable between the treatment groups.

In both EMPEROR trials, acute renal failure was the most commonly reported notable AE (9.4% versus 10.3%, and 12.1% versus 12.8% in the empagliflozin and placebo groups in EMPEROR-Reduced and EMPEROR-Preserved, respectively), followed by hypotension (9.4% versus 8.7%, and 10.4% versus 8.6% in the empagliflozin and placebo groups in EMPEROR-Reduced and EMPEROR-Preserved, respectively), urinary tract infection (4.9% versus 4.5%, and 9.9 versus 8.1% in the empagliflozin and placebo groups in EMPEROR-Reduced and EMPEROR-Preserved, respectively), and bone fracture (2.4% versus 2.3%, and 4.5% versus 4.2% in the empagliflozin and placebo groups in EMPEROR-Reduced and EMPEROR-Preserved, respectively). No new safety concerns were identified.

Table 2

Summary of Key Results From Pivotal and Protocol-Selected Studies.

Internal Validity

Both the EMPEROR-Reduced and EMPEROR-Preserved trials appeared to have used accepted methods for blinding, allocation concealment, and randomization with stratification. For both EMPEROR trials, a computer-generated block randomization scheme was used, and randomization with stratifications was performed centrally, which typically has a low risk of bias. The demographic and baseline patient characteristics appeared to be generally balanced between the treatment groups in both trials, so randomization was maintained. Both EMPEROR trials included only patients with elevated NT-proBNP, as high concentrations of NT-proBNP can confirm HF in patients who present with dyspnea when the clinical diagnosis remains uncertain.² However, the clinical experts consulted by CADTH highlighted that physicians only need to perform NT-proBNP tests in 10% to 20% of cases, when they are unsure of the diagnosis of HF. A relatively high proportion of patients prematurely discontinued the trial medication (26.7% and 31.5% in EMPEROR-Reduced and EMPEROR-Preserved, respectively, including fatal events), while the cause of discontinuations occurred at a similar frequency between the treatment groups. The clinical experts noted that a high proportion of AEs leading to treatment discontinuation were fatal, which reflects the natural history of the HF more than intolerance to the drug under review. An independent blinded committee of clinical experts performed a central adjudication of the primary and key secondary outcomes based on criteria defined a priori. The clinical experts consulted indicated that CV death and HHF are the main outcomes used in clinical practice to assess the response to HF treatment. While improvement in HRQoL, HF symptoms, and functional ability were of primary importance to patients with HF according to the patient group input, these were exploratory outcomes and were outside the statistical testing hierarchy; thus, the results should be viewed as supportive evidence for the overall effect of empagliflozin. The symptoms associated with HF and HRQoL were assessed using KCCQ and EQ-5D-5L instruments. The clinical experts indicated that these tools are not used in clinical practice but are used in multiple studies, allowing comparisons between different treatments. Since treatment discontinuation rates were relatively high across both treatment groups, and many patients did not complete the KCCQ or EQ-5D-5L at baseline or follow-up, there is a high risk of bias, as the patients who completed the questionnaires may be fundamentally different from those who did not complete (e.g., differences in treatment response, AEs). Assessment of functional ability was based on the change in NYHA functional class from baseline at week 52 using descriptive statistics. The evidence of empagliflozin in patients with chronic HF was limited by 2 placebo-controlled pivotal trials, and no head-to-head evidence for empagliflozin compared against other comparators, including dapagliflozin or sacubitril-valsartan in the HFrEF population, were available for this review.

External Validity

In general, the clinical experts consulted by CADTH for this review confirmed that the populations of both the EMPEROR-Reduced and EMPEROR-Preserved trials were similar to the patients seen in Canadian clinics, and the study results would be generalizable to patients with HF in Canada, with some limitations. While empagliflozin has been approved by Health Canada for use as an adjunct to SOC therapy in patients with chronic HF regardless of NYHA class, CADTH was unable to draw conclusions related to patients with NYHA functional classes I and IV, since both trials excluded patients who had NYHA class I, and there was a very small number of patients who had NYHA class IV. One of the clinical experts consulted highlighted that the benefit of empagliflozin in patients with NYHA class IV is unclear due to limited clinical data and high mortality, while another clinical expert indicated that he would prescribe empagliflozin to patients with NYHA class IV. In addition, the clinical experts indicated they would not prescribe empagliflozin to patients with chronic HF with NYHA class I, as they are asymptomatic, which is consistent with the reimbursement request. About 48% of patients in both trials did not pass the screening, most commonly because of NT-proBNP levels below the pre-specified thresholds at screening, which further reduces the generalizability of the results. The clinical experts consulted indicated that NT-proBNP testing is not widely available in Canada, as some jurisdictions have limited access to it; thus, this patient selection criterion would be difficult to implement in clinical practice. The clinical experts further noted that this inclusion criterion likely created an enriched patient population in both trials; the patients with elevated NT-proBNP appeared to be sicker and could benefit more from treatment with empagliflozin than the population in the real-world setting. In the EMPEROR-Preserved trial, about 33% of patients had mid-range LVEF (41% to 49%); however, the clinical experts do not expect this to be a major issue with the generalizability of the trial results, as the LVEF definition is arbitrary, and estimates of LVEF may vary depending on the patient or technical factors as well as on clinical deterioration. The clinical experts consulted noted the patients included in both EMPEROR trials were younger, as the median age of the population with HF in the real-world setting is approximately 75 years. The generalizability of the EMPEROR-Reduced trial results may be compromised by the high proportion of males (more than 75%) who were enrolled, as half of the population with HFrEF in Canada is female. Nonetheless, the clinical experts consulted noted that they would treat both male and female patients with chronic HF with empagliflozin. The majority of patients in both EMPEROR trials were receiving guideline-recommended treatment of HF; thus, they represented patients who were optimally managed, while the clinical experts noted that a goal-directed treatment of HF is suboptimal in clinical settings. Lastly, although the recommended dose of empagliflozin for the treatment of HF is 10 mg, the clinical experts indicated that both the 10 mg and 25 mg doses of empagliflozin are used in clinical practice.

Description of Studies

In the absence of direct comparative evidence from trials, the aim of the indirect treatment comparison (ITC) conducted according to the methodology described by Bucher et al. (1997)⁹ was to compare the efficacy of empagliflozin plus SOC versus dapagliflozin plus SOC in patients with HFrEF. ||| ||||||| ||||| || |||||||| ||| ||| || ||| ||| ||||||| ||||| | |||||| ||| |||| ||||| ||||| ||||||||||| ||| ||||||| |||| |||| ||||||||| ||| ||| |||||||| || |||||||||| ||||||||| || |||| || ||||| ||| ||| ||||||| |||| ||| ||||||| |||||| |||| ||| |||| ||||| ||| ||||||||||||||| ||| ||||||| ||||||| |||| ||||||||||||||| ||| ||||||| |||| ||||| || |||||| |||||| |||||||||||| ||||||| ||||||||| ||| || |||||| ||||||||||||| || |||||||||||||| ||||||||||||| ||| |||| |||||||| || ||||||||| ||| |||||||| |||||| || ||||||| ||||| |||||||| |||||||| || ||||||||||||||| ||| |||||||| ||| ||||||||| |||||||| |||| ||||||| |||| ||| ||||||||| || |||||||||||||||| ||||| ||||||| ||| |||||| |||||||| ||||||||| ||| |||||| |||| |||||||||| ||||| ||||||||| || ||| ||||| ||||||||| ||||||||| ||| ||||||| |||||||| || ||||||||||||||| ||| ||||||||||||| ||||||||||||||| ||| ||||| ||||||| || || |||||| ||||||| || |||||||| |||| ||||||||| |||||||| |||| |||||||| |||||| ||||| ||||||| ||||||| ||| ||| |||||||| || ||| ||||||||| ||||| ||||||||| |||| |||||||||| ||||||||||| |||| ||||||| |||| |||||||| || ||| |||||||| ||||||||| || ||| |||||||| |||| |||| || |||| |||||||||| || |||| ||| |||| || |||||

Efficacy Results

||| |||||| ||| |||||||| ||||||||||||| ||||||| ||||||||||||| || |||||||| |||| ||||| || |||| || ||||| ||||| || ||||||||||| || ||||| || ||||||||||| |||| |||| || ||||| ||||||||||| |||| |||| || ||||||||||| || |||||| |||| || ||||||||| |||||||||| ||| |||| || ||||||||| ||||| ||||||||| ||||| |||| || ||||||||| |||| |||||||| || | || ||||||||| | ||| ||| |||||||||| || ||||||||||||| |||||| |||||||||||||| ||| ||||||| |||||||| |||| |||| |||||||| |||| || ||||| ||||| || ||||||||||| || ||||| || ||||||||||| |||| |||||| || |||||||||| ||||||| ||||||||||||| ||| ||||||||||||| |||| | || |||| ||| || |||| ||||| || ||||||

Critical Appraisal

The sponsor conducted a Bucher ITC comparing empagliflozin against dapagliflozin in patients with HFrEF. Studies were identified from a systematic review; however, the included studies from the systematic review were further refined on an ad hoc basis to arrive at the 2 pivotal trials for each drug to be analyzed in the ITC, potentially introducing selection bias. The Bucher methodology for ITC assumes all differences in patient characteristics or study design have no impact on treatment effects, estimating relative treatment effects using the common comparator arm of 2 treatments that have not been investigated in a head-to-head study. Important differences between the EMPEROR-Reduced and DAPA-HF trials included the broader primary composite end point in DAPA-HF (the impact of which is uncertain), baseline characteristics indicating sicker patients in EMPEROR-Reduced, potentially biasing the results in favour of empagliflozin, and the more effective basket of background SOC therapies used in EMPEROR-Reduced, potentially biasing results against empagliflozin | ||||||| ||| ||||||||| ||||||||||| ||||||| ||| | |||||||| ||||| |||| ||| |||||| || || |||||||| ||| | |||||||||| || ||||||||| ||||||| ||||||| ||||||||||||| ||| |||||||||||||| |||||||| |||| ||| ||||||| || |||||||| ||||||| ||||||||||. Two additional ITCs were identified from the literature search conducted by CADTH. Given the lack of details provided, the results were highly uncertain; however, the results indicating no difference between empagliflozin and dapagliflozin were consistent with the ||||||| ||||||||| ||| ||| the opinion of the clinical experts consulted.

Description of Studies

Efficacy Results

The primary end point was change from baseline in 6-minute walk test distance (6MWTD) at week 12. Key secondary end points were change from baseline in KCCQ total symptom score (KCCQ-TSS) and Chronic Heart Failure Questionnaire Self-Administered Standardized Format (CHQ-SAS) dyspnea score at week 12. Results for the KCCQ-TSS and CHQ-SAS dyspnea score are presented in accordance with the protocol for the CADTH review. The median difference from baseline to week 12, empagliflozin versus placebo, in KCCQ-TSS was 3.13 (95% CI, 0.00 to 7.29) and 2.08 (95% CI, −2.08 to 6.25) in EMPERIAL-Reduced and EMPERIAL-Preserved, respectively. The median difference, empagliflozin versus placebo, in CHQ-SAS dyspnea score was 0.10 (95% CI, −0.20 to 0.40) and −0.07 (95% CI, −0.35 to 0.20) in EMPERIAL-Reduced and EMPERIAL-Preserved, respectively.¹¹ The results for other symptom outcomes are presented in the Other Relevant Evidence section.

Harms Results

There was no notable difference for empagliflozin versus placebo regarding the overall frequencies of any AE or any AE leading to treatment discontinuation in both trials. SAEs were reported less frequently with empagliflozin than with placebo in EMPERIAL-Reduced (12.7% for empagliflozin versus 18.4% for placebo) and EMPERIAL-Preserved (13.5% for empagliflozin versus 17.3% for placebo). Decreased kidney function was reported with similar frequencies in both groups. No ketoacidosis or confirmed hypoglycemic events occurred in participants without type 2 diabetes. No new safety concerns were identified.¹¹

Critical Appraisal

The following limitations were identified:

• HF is a chronic condition, which means the progression of HF is generally slow, thus the assessment of change in outcomes may require a long-term follow-up period.
• The follow-up period for the EMPERIAL-Reduced and EMPERIAL-Preserved trials was 12 weeks, which may not be sufficient to assess meaningful changes in the outcome measures.
• The EMPERIAL trials were powered to detect an improvement of 30 m in 6MWTD; however, the study sample size may not be sufficient to detect any between-group changes of less than 30 m.
• As the primary end point (change from baseline in the 6MWTD at week 12) was not met, the analyses of all secondary outcomes, such as the KCCQ-TSS and CHQ-SAS dyspnea score, were considered exploratory.
• While the changes in the KCCQ-TSS and CHQ-SAS dyspnea score may suggest a possible favourable effect of empagliflozin in patients with HFrEF, these results are considered exploratory.
• The baseline demographic and baseline characteristics (sex and 6MWTD) were suggestive of an over-representation of male patients with lower functioning status, which may compromise the representativeness of the study sample compared with the general population of adult patients with HF.

Although the EMPERIAL studies provide additional data on the effectiveness and safety of empagliflozin in patients with HF, the limitations identified introduce uncertainty.

Unmet Needs

According to the clinical experts consulted by CADTH, many patients with HFrEF are not being assessed by a specialist in Canada, and assistance from other specialists and primary care providers will be required, given the growing number of patients. The clinical experts further noted that the use of goal-directed guideline-recommended drug therapy and medical devices in patients with HFrEF remains suboptimal. The clinical experts consulted highlighted that current treatment strategies in HFpEF are limited to supportive therapies focusing on symptom control rather than morbidity or mortality benefit, including ARBs, MRAs, and ARNIs, while data on SGLT2 inhibitors show clear benefit in this population.

Place in Therapy

The clinical experts indicated that empagliflozin can be used as an alternative to dapagliflozin in combination with other foundational guideline-recommended pharmacotherapy, including beta blockers, ACEIs and ARBs, MRAs, and ARNIs in patients with HFrEF, and it is likely to be a first-line therapy for patients with HFpEF, given the ease of use, strength of evidence, safety profile, and familiarity with the use of empagliflozin in patients with diabetes. In addition, the clinical experts believe that empagliflozin is likely to be better tolerated than other classes of medications.

Patient Population

The clinical reviewers indicated there is no evidence of benefit for empagliflozin in patients with HF with low NT-proBNP levels, as the existing trials were limited to including patients with elevated NT-proBNP levels. The clinical experts also highlighted that NT-proBNP testing is not widely available in Canada, as some jurisdictions have limited access to it; thus, this patient selection criterion would be difficult to implement in clinical practice. The clinical experts consulted by CADTH highlighted that the benefit of empagliflozin in patients with NYHA classes I and IV is unclear due to limited clinical data and high mortality rate in patients with NYHA class IV.

Assessing Response to Treatment

The clinical experts indicated that the response to therapy in clinical practice is assessed based on the frequency of hospitalizations for HF, which in turn may lead to a reduction in mortality, improved quality of life, and a slower decline in kidney function.

Discontinuing Treatment

The clinical experts identified the following factors to consider when deciding to discontinue treatment with empagliflozin:

• the development of severe kidney dysfunction (eGFR < 15 mL/min/1.73 m²), and euglycemic diabetic ketoacidosis in patients with diabetes as important AEs
• NYHA functional class IV.

Prescribing Conditions

The clinical experts indicated that empagliflozin is already widely used by primary care providers and endocrinologists for the management of diabetes, by nephrologists to reduce decline in kidney function, and by cardiologists.

EMPEROR-Reduced

The EMPEROR-Reduced trial was a phase III, randomized, double-blind, multinational, parallel-group trial that aimed to assess the superiority of empagliflozin 10 mg once daily compared with matched placebo as an adjunct to SOC treatment in patients with chronic HFrEF (LVEF ≤ 40%). A total of 3,730 patients were enrolled across 520 sites from 20 countries in North America (including 35 sites in Canada [136 patients]), Europe, Asia Pacific, South America, South Africa, and other. After a screening period of 4 to 28 days, patients were randomized at visit 2 in a 1:1 ratio to receive either empagliflozin at a dose of 10 mg once daily (N = 1,863) or matching placebo (N = 1,867) in a double-blind manner.

EMPEROR-Preserved

The EMPEROR-Preserved trial was a phase III, randomized, double-blind, multinational, parallel-group trial that aimed to assess the superiority of empagliflozin 10 mg once daily compared with matched placebo as an adjunct to SOC treatment in patients with chronic HFpEF (LVEF > 40%). A total of 5,988 patients were enrolled across 622 sites from 23 countries in North America (including 42 sites in Canada [199 patients]), Europe, Asia Pacific, South America, South Africa, and other. After a screening period of 4 to 28 days, patients were randomized at visit 2 in a 1:1 ratio to receive either empagliflozin at a dose of 10 mg once daily (N = 2,997) or matching placebo (N = 2,991) in a double-blind manner.

In both EMPEROR-Reduced and EMPEROR-Preserved trials, randomization in blocks was conducted centrally via interactive response technology. In EMPEROR-Reduced, randomization was stratified by geographical region (North America, Lain America, Europe, Asia, and other), history of diabetes (diabetes, pre-diabetes, and no diabetes), and eGFR (CKD-EPIcr equation) at screening (< 60 mL/min/1.73 m², and ≥ 60 mL/min/1.73 m²). In EMPEROR-Preserved, randomization was stratified by geographical region (North America, Lain America, Europe, Asia, and other), history of diabetes (diabetes, pre-diabetes, and no diabetes), eGFR (CKD-EPIcr equation) at screening (< 60 mL/min/1.73 m², and ≥ 60 mL/min/1.73 m²), and LVEF (< 50%, and ≥ 50%). Treatment allocation was determined by a computer-generated random sequence. Given that empagliflozin is a diabetes drug, there was a possibility that more patients with diabetes would be recruited in these trials. Therefore, interactive response technology was used to ensure similar proportions of patients with diabetes, pre-diabetes, and no diabetes at the regional level in both trials.

Both EMPEROR-Reduced and EMPEROR-Preserved trials were event-driven trials and were to stop once 841 adjudicated primary end point events (CV death or HHF) were reached. Thus, the duration of double-blind treatment was different for each patient. In both EMPEROR-Reduced and EMPEROR-Preserved trials, the number of confirmed primary end points was continuously monitored in a blinded manner. In both trials, onsite visits were scheduled at 4, 12, 32, and 52 weeks after randomization during the first year, and then every 24 weeks throughout the trial. During the onsite visits, safety and efficacy end points, treatment compliance, and concomitant treatment or intervention were assessed. In addition, follow-up phone calls were scheduled 10 to 12 weeks after each onsite visit, starting at visit 4 and continuing throughout the trial. End of treatment in both trials was defined as reaching the required number of primary end points (841 events), or when the patient permanently discontinued study medication. All patients were required to complete a follow-up visit within 30 days of the regular or premature termination of the treatment period. A schematic of the EMPEROR-Reduced and EMPEROR-Preserved trials is presented in Figure 2.

An interim analysis was performed by the independent data-monitoring committee after 544 and 494 primary end point events (about 60% of information) in the EMPEROR-Reduced and EMPEROR-Preserved trials, respectively, after which it was recommended to continue the trials as planned. The database lock was performed after the completion of the blinded treatment period in the EMPEROR-Reduced and EMPEROR-Preserved trials. In EMPEROR-Reduced, an interim database lock was executed on October 11, 2019, and the final database lock was executed on July 14, 2020. In EMPEROR-Preserved, an interim database lock was executed on January 27, 2020, and the final database lock was executed on June 1, 2021.

In both EMPEROR trials, the primary efficacy end point was the time to first event of adjudicated CV death or adjudicated HHF, and the key secondary end points were occurrence of adjudicated HHF (first and recurrent), and eGFR (CKD-EPIcr equation) slope of change from baseline. HRQoL was assessed using the KCCQ and EQ-5D instruments. Overall, baseline characteristics were well balanced between treatment groups in both pivotal trials.

Figure 2

Study Schema for the EMPEROR-Reduced and EMPEROR-Preserved Studies.

Inclusion and Exclusion Criteria

The key inclusion and exclusion criteria applied to the EMPEROR-Reduced and EMPEROR-Preserved trials are summarized in Table 6. Briefly, patients eligible for enrolment in the EMPEROR-Reduced trial were adults with chronic HF diagnosed for at least 3 months with NYHA functional class II to IV, reduced ejection fraction (LVEF ≤ 40%), and elevated NT-proBNP (i.e., > 2,500 pg/mL for patients without atrial fibrillation). Patients were also required to have received appropriate doses of HF therapy (i.e., ACEI, ARB, beta blocker, oral diuretic, MRA, ARNI, ivabradine), and appropriate use of medical devices. Patients eligible for enrolment in the EMPEROR-Preserved trial were adults with chronic HF diagnosed for at least 3 months with NYHA functional class II to IV, LVEF greater than 40%, elevated NT-proBNP (i.e., > 300 pg/mL without atrial fibrillation), and evidence of structural heart disease or an HHF within 12 months before the trial. Patients were excluded from both EMPEROR trials if they had a diagnosis of myocardial infarction, stroke, transient ischemic attack, acute decompensated HF, major CV surgery, or any major surgery within the last 90 days. Patients with a history of renal impairment or ketoacidosis, as well as patients with an implanted cardioverter-defibrillator or implanted cardiac resynchronization therapy within the last 3 months, those with current or prior use of an SGLT2 inhibitor, or use of combined sodium-glucose cotransporter-1 (SGLT1) and SGLT2 inhibitors within the last 12 weeks were also excluded.

Baseline Characteristics

EMPEROR-Preserved Study

Baseline characteristics were well balanced between the treatment arms. The mean age of all randomized patients in the EMPEROR-Preserved study was 71.9 years (SD = 9.4 years), nearly half of the patients were male (55.3%), and most patients were White (75.9%). More patients were aged 70 years and older (64.1%) compared with those under the age of 70 years (35.9%). The mean LVEF was 54.3% (SD = 8.8%), with 33.1% of patients having an LVEF of less than 50%. Most patients had NYHA functional class II (81.5%) at baseline. Almost half of the patients were diagnosed with diabetes mellitus (49.1%) at baseline, and about 52.4% had a history of atrial fibrillation or flutter. The median NT-proBNP was 974 pg/mL (Q1 to Q3, 499 pg/mL to 1,731 pg/mL), while the mean eGFR (CKD-EPIcr equation) was 60.6 mL/min/1.73 m² (SD = 19.8). The mean time of HF diagnosis to enrolment was 4.4 years (SD = 5.1 years), and about 22.9% of patients had a prior HHF. Patients received the following previous HF medications at the start of the EMPEROR-Preserved trial: ACE inhibitors or ARBs (78.6%), ARNIs (2.2%), beta blockers (86.3%), MRAs (37.5%), ||| |||||||||| ||||||.

Table 7

Summary of Baseline Characteristics — EMPEROR-Reduced and EMPEROR-Preserved, RS.

Efficacy Outcomes

In both EMPEROR trials, an independent group of clinical experts performed a central adjudication of the following outcomes occurring after randomization in a consistent and blinded fashion:

• all fatal events
• HF hospitalization
• myocardial infarction
• stroke and transient ischemic attack
• ketoacidosis
• hepatic events.

The primary composite end point for both the EMPEROR-Reduced and EMPEROR-Preserved trials was the time to first event of adjudicated CV death or adjudicated HHF.

CV death included death due to:

• acute myocardial infarction, which refers to a death from any CV mechanism that occurs within 30 days after an MI
• sudden cardiac death, which refers to a death that occurred unexpectedly and not following an acute MI
• other causes:
  • HF, which refers to a death in association with clinically worsening symptoms and/or signs of HF regardless of HF etiology
  • stroke, which refers to a death after a stroke that is either a direct consequence or a complication of the stroke
  • CV procedures, which refers to a death caused by the immediate complication of a cardiac procedure
  • CV hemorrhage, which refers to a death related to a hemorrhage such as non-stroke intracranial hemorrhage, non-procedural or non-traumatic vascular rupture, or hemorrhage causing cardiac tamponade
  • other CV causes, which refers to a CV death not included in the previous categories, but with a specific, known cause (i.e., pulmonary embolism or peripheral arterial disease).

Non-CV death was defined as any death with a specific cause that is not thought to be CV in nature.

HHF was defined as an event that met all the following criteria:

• The adjudicated primary diagnosis is admission to hospital for HF.
• The patient’s length of stay in hospital extends for at least 12 hours.
• The patient exhibits documented new or worsening symptoms due to HF on presentation (i.e., dyspnea, fatigue).
• The patient has objective evidence of new or worsening HF, consisting of at least 2 physical examination findings or 1 physical examination finding and at least 1 laboratory criterion, including increased brain natriuretic peptide (BNP) and NT-proBNP levels, radiological evidence of pulmonary congestion, or non-invasive diagnostic evidence of clinically significant elevated left- or right-sided ventricular filling pressure.
• The patient receives initiation or intensification of treatment specifically for HF, including at least 1 of the following: augmentation in oral diuretic therapy, IV diuretic or vasoactive drug, or mechanical or surgical intervention.
• The documented changes in physical signs or laboratory tests, whenever available, were considered to be supportive.

The key secondary end points for both EMPEROR trials included:

• occurrence of adjudicated HHF (first and recurrent)
• eGFR (CKD-EPIcr equation) slope of change from baseline.

In addition to the clinical end points, 2 patient-valued outcomes, such as HRQoL and symptoms associated with HF, were measured in the EMPEROR-Reduced and EMPEROR-Preserved trials using the KCCQ and the EQ-5D instrument.

Harms

The primary safety outcomes assessed in the EMPEROR-Reduced and EMPEROR-Preserved trials were:

• AEs
• TEAEs
• SAEs
• AEs leading to the discontinuation of the investigational drug
• AEs of special interests, including hepatic injury, decreased renal function, ketoacidosis, events leading to lower-limb amputation, hypoglycemia
• worsening of underlying condition
• changes from baseline for clinical laboratory measure and vital signs.

An independent group of medical experts performed a central, blinded adjudication of ketoacidosis and certain hepatic events.

Sample Size Determination

In the EMPEROR-REDUCED trial, at least 841 primary end point events in the intention-to-treat population were required to achieve 90% power at a 2-sided significance level of 0.05. A true HR of 0.8 between empagliflozin and placebo was chosen based on the HF outcomes in the EMPA-REG-OUTCOME trial.⁴⁰ The sponsor estimated that at least 2,825 patients needed to be enrolled to achieve the required number of events, assuming a yearly event rate in the placebo group of 15%,^41-44 an accrual period of 18 months, and an average follow-up period of 20 months. The yearly dropout rate was assumed to be below 1% and was not considered for the sample size determination. There was 1 planned interim analysis for the primary outcome, which was conducted when 544 (approximately 60%) of the events had occurred. The EMPEROR-Reduced trial used the Hwang, Shin, and De Cani alpha-spending function to control for type I error, yielding a 2-sided significance level of 0.0496 for the final primary end point analysis. Following the interim analysis, the study was recommended to continue as planned. During the trial, based on the actual accrual over time of the primary outcome events, the number of randomized patients was adjusted to 3,600.

In the EMPEROR-Preserved trial, at least 841 primary end point events in the intention-to-treat population were required to achieve a 90% power at a 2-sided significance level of 0.05. A true HR of 0.8 between empagliflozin and placebo was chosen based on the HF outcomes in the EMPA-REG-OUTCOME study.⁴⁰ Based on the previously mentioned assumption, the sponsor estimated that at least 4,126 patients were needed, assuming a yearly event rate in the placebo group of 10%,^45,46 an accrual period of 18 months, and an average follow-up period of 20 months. The yearly dropout rate was assumed to be below 1% and was not considered for the sample size determination. There was 1 planned interim analysis for the primary outcome, which was conducted when 494 (approximately 60%) of the events had occurred. The EMPEROR-Preserved trial used the Hwang, Shin, and De Cani alpha-spending function to control the type I error, yielding a 2-sided significance level of 0.0497 for the final primary end point analysis. Following the interim analysis, the study was recommended to continue as planned. During the trial, based on the actual accrual over time of the primary outcome events, the number of randomized patients was adjusted to 5,750.

Primary Efficacy Analysis

In both EMPEROR trials, the composite end point of time to first event of adjudicated CV death or adjudicated HHF was analyzed using a Cox proportional hazards model adjusted for age, geographical region (Asia, Europe, Latin America, North America, and other), diabetes status (diabetes, pre-diabetes, and no diabetes), sex, LVEF, and eGFR (CKD-EPIcr equation). The proportional hazards assumption was evaluated by plotting Schoenfeld residuals for each covariate and treatment against time and log (time). Each component of the composite primary end point was also summarized separately, but was not formally tested for significance. The primary end point was displayed using a cumulative incidence function, considering non-CV death as competing risk and expressed using HR and 95% CI. Incidence rate was calculated as the number of patients with events per 100 person-years at risk. The time to event was derived from the date of randomization. This analysis was based on the RS (described subsequently), using all data available until completion of the planned treatment phase, including the data after the end of treatment for patients who did not complete the treatment phase as planned. A patient with at least 1 event (CV death or HHF) was considered to have an event and the date of the first event was used for the composite end point analysis. Only the adjudicated and confirmed events were included in the primary analysis. Patients without a specific end point event were censored at the last date the patient was known to be free of the event at the end of the planned treatment period, whichever was earliest.

The following sensitivity analyses (exploratory) were conducted for the primary outcome:

• a Cox proportional hazards model including only treatment, not adjusted for any other factors
• a Cox proportional hazards model with the same covariates as per the primary analysis performed on the treated set (TS), with observation period up to 30 days after treatment discontinuation
• a Cox proportional hazards model with multiple imputations for patients without primary end point events and lost to follow-up before the trial completion
• analysis of investigator-defined events
• a competing risk model by Fine-Gray, including the same set of covariates as in the primary analysis
• a Cox proportional hazards model of all events in the trial (including the follow-up period)
• a Cox proportional hazards model similar to the primary analysis but excluding those missing a physical sign or laboratory test, or both
• a Cox proportional hazards model similar to the primary analysis but including the following additional prognostic covariates: Log(NT-proBNP) and HHF in the last 12 months
• a Cox proportional hazards model similar to the primary analysis but including only events up to cut-off dates before a COVID-19 outbreak.

Sensitivity analyses were performed based on the TS (described subsequently).

Of the subgroups listed in the CADTH review protocol, the following subgroups were pre-specified in the EMPEROR-Reduced trial:

• HF physiology (LVEF ≤ 30% and NT-proBNP < median, LVEF ≤ 30%, NT-proBNP ≥ median, or LVEF > 30%)
• NYHA class (classes II or III/IV)
• history of diabetes (yes or no)
• renal function (eGFR [CKD-EPIcr equation]) < 60 or ≥ 60)
• prior use of ARNI (yes or no)
• prior use of MRA (yes or no).

The following subgroups were pre-specified in the EMPEROR-Preserved trial:

• LVEF (< 50%, 50% to 59%, or ≥ 60%)
• NYHA class (classes II or III/IV)
• History of diabetes (yes or no)
• Renal function (eGFR [CKD-EPIcr equation]) < 60 or ≥ 60)
• History of atrial fibrillation or atrial flutter (yes or no)
• Prior use of ACE inhibitor, ARB, or ARNI (yes or no)
• Prior use of MRA (yes or no).

In EMPEROR-Reduced, the randomization of patients was stratified by geographical region, history of diabetes, and eGFR (CKD-EPIcr equation) at screening. In EMPEROR-Preserved, the randomization of patients was stratified by geographical region, history of diabetes, LVEF, and eGFR (CKD-EPIcr equation) at screening. The subgroup analyses were performed using a Cox proportional hazards model as per the primary end point analysis. There were no adjustments made for multiplicity; thus, all subgroup analyses are exploratory in nature. The between-group treatment effect with a nominal 95% CI for these end points was estimated within each category. Forest plots were created, including interaction P values for treatment by subgroup interactions.

Key Secondary Efficacy Analysis

EMPEROR-Reduced and EMPEROR-Preserved Trials

In both EMPEROR trials, the key secondary end points, including occurrence of adjudicated HHF (first or recurrent) and the eGFR (CKD-EPIcr equation) slope of change from baseline, were tested in order using a hierarchical testing procedure to control the overall type I error rate for multiple end points (Table 9). If the primary end point was statistically significant, the overall type I error was preserved for the test in the next step. This testing procedure continued through each of the key secondary end points until the end point failed to reach statistical significance, after which subsequent key secondary end points were considered exploratory.

In both EMPEROR trials, the occurrence of HHF (first and recurrent) was analyzed using a joint frailty model that accounts for the dependence between recurrent HHF and CV death, with factors of treatment (empagliflozin, placebo), geographical region, baseline status of diabetes, age, sex, LVEF, and eGFR (CKD-EPIcr equation) at baseline as covariates. All data from all randomized patients until the end of the planned treatment period were used. The number of HHF events per patient was summarized descriptively. The number of HHF events was analyzed descriptively. Negative binomial models were additionally fitted for recurrent HHF events. The mean cumulative incidence was displayed for adjudicated first and recurrent HHF. Subgroups analyses were carried out; however, there were no adjustments made for multiplicity.

The following sensitivity analyses of adjudicated HHF (first and recurrent) were conducted:

• a Cox model, including only treatment as covariate, not adjusting for any other factors
• a Cox regression with the same covariates as the primary analysis performed on the TS, including only events up to 30 days after treatment discontinuation
• a Cox regression with multiple imputations for patients without primary end point events and lost to follow-up before trial completion
• analysis of investigator-defined events
• a Fine-Gray competing risk model (considering non-CV death as a competing risk)
• a Cox regression of all events in the trial (including the follow-up period)
• a Cox regression similar to the primary analysis but excluding events without documented physical signs or symptoms
• a Cox regression similar to the primary analysis but including the following additional prognostic covariates: Log(NT-proBNP) and HHF in the last 12 months
• a Cox regression similar to the primary analysis but including only events that were thought to occur before the COVID-19 pandemic.

In both EMPEROR trials, the slope in change from baseline of eGFR (CKD-EPIcr equation) was analyzed using a random coefficient model allowing for random intercept and random slope per patient. The model was adjusted for sex, geographical region, status of diabetes as fixed effects, and eGFR (CKD-EPIcr equation) at baseline, LVEF, age, time, and interaction of treatment by time and interaction of eGFR at baseline by time as linear covariates. Only data from treated patients (based on TS, i.e., measurement up to 1 day after the last intake of study medication) were used. Subgroup analyses were carried out; however, there were no adjustments made for multiplicity.

Table 9

Summary of Hierarchical Testing — EMPEROR-Reduced and EMPEROR-Preserved.

Other Secondary Efficacy Analysis

In both pivotal trials, end points listed as other secondary end points were tested in a non-hierarchical fashion without adjustments for multiplicity. The following time-to-event end points were analyzed using a Cox proportional hazards model similar to the primary outcome analysis: time to the first event in the composite renal end point (chronic dialysis, renal transplant, or sustained reduction in eGFR), time to first adjudicated HHF, time to adjudicated CV death, and time to all-cause mortality. The models were adjusted for sex, geographical region, baseline status of diabetes, eGFR (CKD-EPIcr equation) at baseline, LVEF, and age. If the end point did not include any cause of death, a cumulative incidence function curve was displayed with all-cause of death as the competing risk. A Kaplan-Meier survival curve was displayed for all-cause mortality.

In both pivotal trials, change from baseline KCCQ clinical summary, total symptom, and overall summary scores, and KCCQ domain scores at week 52 were analyzed using a mixed-model for repeated measures analysis, including age, eGFR (CKD-EPIcr equation) as linear covariates, and baseline score by visit, visit by treatment, sex, geographical region, LVEF, and baseline diabetes status as fixed effects. The analysis was carried out for both the RS, and the TS, which included all observed cases with on-treatment data up to week 52. Responder analyses using logistic regression were conducted to evaluate the improvement and deterioration of the KCCQ clinical summary, total symptom, and overall summary scores at week 52.

Further Efficacy End Points

In both pivotal trials, further end points were tested in a non-hierarchical fashion without adjustments for multiplicity. The following end points were analyzed using a mixed-model for repeated measures: change from baseline in KCCQ overall summary score and clinical summary score, and total symptom score at week 52. Furthermore, responders for clinically meaningful improvement (an increase in score of at least 5 points at week 52 from baseline) or deterioration (a reduction of at least 5 points) were analyzed using logistic regression. The following further efficacy end points were analyzed using a Cox proportional hazards model:

• composite of time to first event of all-cause mortality and all-cause hospitalization
• time from first to second adjudicated HHF
• time to first all-cause hospitalization
• time to all-cause hospitalization or all-cause mortality
• time to first investigator-defined CV hospitalization.

Change in NYHA class at week 52 and EQ-5D-5L scores were analyzed using descriptive statistics.

Missing Data

In both pivotal trials, the analyses of primary and key secondary end points were performed based on all available data in the RS (description follows). No data were imputed for time-to-event or safety end points. All efforts were made to follow all patients until the end of the trial for their survival status and for any other end points, including the primary and key secondary end points. With regard to KCCQ scores, for patients who died, the worst score was assigned to all scores scheduled to be assessed after the date of death.

Harms

In both pivotal trials, all safety end points were reported using descriptive statistics and were carried out on the safety population (TS). Separate summaries were provided for most notable safety end points: decreased renal function, ketoacidosis, events leading to lower-limb amputation, hypoglycemic events, urinary tract and genital infections, hypotension, and bone fracture events. The incidence of these end points was analyzed by treatment as well as by subgroups. Safety analyses were based on the TS and included patients who had received at least 1 dose of the trial medication.

Table 10

Statistical Analysis of Efficacy End Points — EMPEROR-Reduced and EMPEROR-Preserved.

Analysis Populations

All patient populations were defined and documented before database lock. The following analysis populations were used in the statistical analysis: RS, TS, and TS with follow-up (TSFU).

The randomized set, also known as the full analysis set (N = 3,730 and N = 5,988 in EMPEROR-Reduced and EMPEROR-Preserved, respectively), consisted of all randomized patients. Patients were analyzed according to their randomized group. Unless otherwise specified, all efficacy end points were summarized and analyzed using the randomized set.

The TS, also known as the safety set (N = 3,726 and N = 5,985 in EMPEROR-Reduced and EMPEROR-Preserved, respectively) consisted of all randomized patients who received at least 1 dose of the study drug. The summary for the safety analysis set was based on patients “as treated.”

The TS with follow-up, also known as the per-protocol set (N = 1,062 and N = 3,269 in EMPEROR-Reduced and EMPEROR-Preserved, respectively) consisted of all randomized patients who received at least 1 dose of the study drug and who performed the follow-up visit.

EMPEROR-Reduced Trial

In EMPEROR-Reduced, 7,220 individuals were screened, of whom 3,490 (48.3%) did not pass screening. The main reasons were not meeting eligibility criteria (95.0%), most commonly because the patients’ NT-proBNP levels were below the pre-specified thresholds at screening, and consent withdrawal (2.3%). In total, 3,730 patients were randomized in the treatment period. Overall, 3,688 patients (98.9%) completed the study or died, with similar completion rates across treatment groups. Vital status was known for 1,857 patients (99.5%) in the empagliflozin group and 1,852 patients (99.4%) in the placebo group. Of the 3,726 patients treated with the study medication, 25.9% of patients in the empagliflozin group and 27.4% of patients in the placebo group discontinued treatment. The most frequently reported reasons for discontinuation were AEs (18.3%), including 8.7% with non-fatal events and 9.5% with fatal events, and patient choice (5.8%). ||| || |||||||| ||||||||| ||||| ||| || ||| ||| || ||||||||| |||||| ||| ||||| ||| || ||||||||| |||||| |||| ||||||||| || ||||| ||||||| ||||||| |||||||||| |||||| |||||||||||

EMPEROR-Preserved Trial

In EMPEROR-Preserved, 11,583 individuals were screened, of whom 5,595 (48.3%) did not pass screening. The main reasons for this were not meeting eligibility criteria (95.4%), most commonly because the patients’ NT-proBNP levels were below the pre-specified thresholds at screening, and consent withdrawal (3.2%). In total, 5,988 patients were randomized in the treatment period. Overall, 5,816 patients (97.1%) completed the study or died, with similar completion rates across treatment groups. Vital status was known for 2,980 patients (99.4%) in the empagliflozin group and 2,972 patients (99.4%) in the placebo group. Of the 5,985 patients treated with the study medication, 31.5% of patients in both the empagliflozin and placebo groups discontinued treatment. The most frequently reported reasons for discontinuation were AEs (18.8%), including 10.6% of non-fatal events and 8.2% of fatal events, and patient choice (9.8%). ||| || |||||||| ||||||||| ||||| ||||| ||| || ||||| ||| ||||| |||||||| || ||||||||| ||| |||||||| || ||||| |||||||| ||||||| ||| || ||||| ||| ||||| ||||||||| || ||||| ||||||| ||||||| |||||||||| |||||| |||||||||| ||| |||| ||| || ||||| ||| ||||| ||||||||| ||||||| ||| |||||||||||||||| ||||||| | ||||| || |||| || |||||||| ||| |||||||||||| || ||||| |||||||||| ||| |||| |||| | |||| ||| || |||||||| ||| |||| ||| ||||||||| ||||||||| ||||||||||||||| ||| || |||||||| ||||||||||

Table 11

Patient Disposition: EMPEROR-Reduced and EMPEROR-Preserved.

Time to First Event of Adjudicated CV Death or HHF

The results of the primary efficacy end point for both pivotal trials are presented in Table 15, Figure 3, and Figure 4.

Subgroup Analysis

The primary end point subgroup analysis in both pivotal trials is presented in (Appendix 3, Table 36). The analyses may not have been powered to detect a treatment difference and there were no adjustments made for multiplicity. As such, all subgroup analyses are exploratory in nature.

In EMPEROR-Reduced, while the effect of empagliflozin on the primary end point events was generally consistent across pre-specified subgroups, potential differences were noted depending on LVEF (P value for interaction < 0.05). In EMPEROR-Preserved, the effect of empagliflozin on the primary end point events was generally consistent across pre-specified subgroups.

In EMPEROR-Reduced, the sensitivity analyses (Appendix 3, Figure 24, and Figure 25) for the primary end point, including those assessing missing data, were exploratory and were generally consistent with the primary analysis. In EMPEROR-Preserved, the sensitivity analyses for the primary end point, including those assessing missing data, were generally consistent with the primary analysis, aside from the treatment effect associated with the COVID-19 outbreak.

Table 15

Time to First Event of Adjudicated CV Death or HHF^a — EMPEROR-Reduced and EMPEROR-Preserved, RS.

Figure 3

Redacted.

Figure 4

Redacted.

Occurrence of Adjudicated HHF (First and Recurrent)

The results of the first key secondary outcome for both pivotal trials are presented in Table 16, Figure 5, and Figure 6.

EMPEROR-Preserved

The number of patients with adjudicated HHF was 246 (8.6%) in the empagliflozin group and 352 (11.8%) in the placebo group. The total number of HHF events was lower in patients who received empagliflozin compared with those who received placebo (407 versus 541, respectively). The total proportion of CV deaths was similar across the treatment groups (7.3% and 8.2% in the empagliflozin and placebo groups, respectively), with an HR of 0.89 (95% CI, 0.71 to 1.12). The hazard rate of recurrent HHF was significantly reduced in the empagliflozin group compared with placebo, with an HR of 0.73 (95% CI, 0.61 to 0.88; P = 0.0009).

A secondary end point subgroup analysis for both pivotal trials is presented in Appendix 3 (Table 38). The analyses may not have been powered to detect a treatment difference and there were no adjustments made for multiplicity. As such, all subgroup analyses are exploratory in nature. In EMPEROR-Preserved, while the effect of empagliflozin on the occurrence of adjudicated HHF was generally consistent across pre-specified subgroups, potential differences were noted among LVEF and prior MRA use subgroups (P value for interaction < 0.05). In EMPEROR-Reduced, the effect of empagliflozin on the occurrence of adjudicated HHF was consistent across pre-specified subgroups.

The results of the sensitivity analyses, including those assessing missing data, were consistent with the results of the primary analysis for the occurrence of adjudicated first and recurrent HHF (Appendix 3, Figure 30 and Figure 31).

Table 16

Occurrence of HHF (First and Recurrent) — EMPEROR-Reduced and EMPEROR-Preserved, RS.

Figure 5

Redacted.

Figure 6

Redacted.

eGFR (CKD-EPIcr Equation) Slope of Change From Baseline

The results of the second key secondary outcome for both pivotal trials are presented in Table 17, Figure 7, and Figure 8. The analysis was performed based on the TS and using observed “on treatment” data.

In EMPEROR-Reduced, over the double-blind treatment period, the rate of decline in the eGFR (CKD-EPIcr equation) per year was slower in the empagliflozin group (−0.55 mL/min/1.73 m² per year; 95% CI, −0.99 to −0.10) than in the placebo group (−2.28 mL/min/1.73 m² per year; 95% CI, −2.73 to −1.83), with a between-group difference in slope of 1.73 per year (95% CI, 1.10 to 2.37).

In EMPEROR-Preserved, over the double-blind treatment period, the rate of decline in the eGFR (CKD-EPIcr equation) per year was slower in the empagliflozin group (−1.25 mL/min/1.73 m² per year; ||| ||| ||||| || |||||) than in the placebo group (−2.62 mL/min/1.73 m² per year; ||| ||| ||||| || |||||), with a between-group difference in slope of 1.36 per year (95% CI, 1.06 to 1.66).

A secondary end point subgroup analysis for both pivotal trials is presented in Appendix 3 (Table 39). The analyses may not have been powered to detect a treatment difference and there were no adjustments made for multiplicity. As such, all subgroup analyses are exploratory in nature | || |||||||||||||||| ||||| ||| |||||| || ||||||||||||| || ||| |||| ||||| || |||||| |||| |||||||| ||| ||||||||| |||||||||| |||||| ||||||||||||| |||||||||| ||||||||| ||||||||||| |||| ||||| ||||||||| || |||||||| |||| ||||||||| || ||| ||||||||||| | ||||||. In EMPEROR-Preserved, potential differences in the benefit of empagliflozin on the eGFR slope of change from baseline were noted, depending on baseline diabetes status.

Table 17

eGFR (CKD-EPIcr Equation) Slope of Change From Baseline — EMPEROR-Reduced and EMPEROR-Preserved, TS.

Figure 7

Redacted.

Figure 8

Redacted.

Other Secondary and Further End Points (Exploratory)

Other Mortality Outcomes

A summary of other mortality-related outcomes for both EMPEROR trials is presented in Table 18. These mortality outcomes were tested in a non-hierarchical sequence without adjustments for multiplicity and were exploratory in nature. In EMPEROR-Reduced, a total of 249 patients (13.4%) in the empagliflozin group and 266 patients (14.2%) in the placebo group died from any cause (HR = 0.92; 95% CI, 0.77 to 1.10). In EMPEROR-Preserved, a total of 422 patients (14.1%) in the empagliflozin group and 427 patients (14.3%) in the placebo group died from any cause (HR = 1.00; 95% CI, 0.87 to 1.15) (Figure 9 and Figure 10). The majority of deaths (75.5%) in the EMPEROR-Reduced trial and nearly half of deaths (54.5%) in the EMPEROR-Preserved trial were due to CV causes (Table 37). The analyses showed no differences between treatment groups in the time to adjudicated CV death (HR = 0.92; 95% CI, 0.75 to 1.12; and HR = 0.91; 95% CI, 0.76 to 1.09, in EMPEROR-Reduced and EMPEROR-Preserved, respectively) (Figure 11 and Figure 12) ||| |||| || ||||||||||| |||||| ||||| ||| | ||||| ||| ||| |||| || ||||| ||| || | ||||| |||||| |||| || ||||| || ||||||||||||||| ||| |||||||||||||||||| |||||||||||||| ||||| || |||||||||| |||| ||| ||||||| |||||||| |||||||||

Table 18

Time to All-Cause Mortality, Time to Adjudicated CV Death, and Time to Adjudicated Non-CV Death — EMPEROR-Reduced and EMPEROR-Preserved, RS.

Figure 9

Redacted.

Figure 10

Redacted.

Figure 11

Redacted.

Figure 12

Redacted.

Other Hospitalization-Related Outcomes

The results of other hospitalization-related outcomes for both pivotal trials are presented in Table 19. These hospitalization-related outcomes were tested in a non-hierarchical sequence without adjustments for multiplicity and were exploratory in nature.

The proportion of patients with 1 or more adjudicated HHFs was lower in the empagliflozin group compared with placebo in both EMPEROR trials. A Cox proportional hazards regression analysis was used to assess the time to first adjudicated HHF (Figure 13 and Figure 14). The hazard of first adjudicated HHF was lower in the empagliflozin group compared with placebo, with an HR of 0.69 (95% CI, 0.59 to 0.81) in EMPEROR-Reduced and 0.71 (95% CI, 0.60 to 0.83) in EMPEROR-Preserved. A joint frailty model was used to examine the occurrence of all-cause hospitalization (first and recurrent) that accounts for the dependence between recurrent all-cause hospitalization and all-cause mortality in both EMPEROR trials. The total number of all-cause hospitalizations was lower in the empagliflozin group compared with placebo (1,364 versus 1,570 in EMPEROR-Reduced, and 2,566 versus 2,769 in EMPEROR-Preserved, respectively). The hazard of recurrent all-cause hospitalization was lower in the empagliflozin group compared with placebo (HR = 0.85; 95% CI, 0.75 to 0.95, and HR = 0.93; 95% CI, 0.85 to 1.01, in EMPEROR-Reduced and EMPEROR-Preserved, respectively), and it was positively correlated with all-cause mortality in both trials. A Cox proportional hazards regression analysis was used to assess the time to all-cause hospitalization or all-cause mortality, and time to investigator-defined CV hospitalization. The incidence rate of first all-cause hospitalization or all-cause mortality was lower in the empagliflozin group compared with placebo in both trials (35.58 versus 43.82 per 100 person-years at risk, and 26.85 versus 29.18 per 100 person-years at risk in EMPEROR-Reduced and EMPEROR-Preserved, respectively), with an HR of 0.81 (95% CI, 0.74 to 0.90), and 0.92 (95% CI, 0.85 to 0.99) in the EMPEROR-Reduced and EMPEROR-Preserved trials, respectively (Figure 15 and Figure 16). The hazard rate of investigator-defined CV hospitalization was lower in the empagliflozin group compared with placebo in both trials, with an HR of 0.75 (95% CI, 0.67 to 0.85) and 0.85 (95% CI, 0.77 to 0.94) in the EMPEROR-Reduced and EMPEROR-Preserved trials, respectively. Since the analysis of the time from the first to the second adjudicated HHF included only patients with 1 or more HHF events, it was not a randomized treatment comparison and is therefore affected by selection bias.

Table 19

Time to First Adjudicated HHF, Time From First to Second HHF, Time to First All-Cause Hospitalization, and Occurrence of All-Cause Hospitalization — EMPEROR-Reduced and EMPEROR-Preserved, RS.

Figure 13

Redacted.

Figure 14

Redacted.

Figure 15

Redacted.

Figure 16

Redacted.

Composite Renal Outcome (Exploratory)

The composite renal end point included the following events: chronic dialysis, renal transplant, or sustained reduction in eGFR (CKD-EPIcr equation) (Figure 17 and Figure 18). Sustained reduction of eGFR was determined by 2 or more consecutive post-baseline central laboratory measurements separated by at least 30 days. In EMPEROR-Reduced, the composite renal end point occurred in 30 patients (1.6%) in the empagliflozin group and 58 patients (3.1%) in the placebo group. The incidence rate was 1.56 per 100 person-years at risk in the empagliflozin group versus 3.07 per 100 person-years at risk in the placebo, with an HR of 0.50 (95% CI, 0.32 to 0.77). In EMPEROR-Preserved, the composite renal end point occurred in 108 patients (3.6%) in the empagliflozin group and 112 (3.7%) in the placebo group. The incidence rate was similar between the empagliflozin and placebo groups (2.13 and 2.23 per 100 person-years at risk, respectively), with an HR of 0.95 (95% CI, 0.73 to 1.24).

Table 20

Time to the First Event in the Composite Renal End Point — EMPEROR-Reduced and EMPEROR-Preserved, RS.

Figure 17

Redacted.

Figure 18

Redacted.

HRQoL, Symptoms of HF, and Functional Status (Exploratory)

KCCQ Clinical Summary Score

The KCCQ-CSS incorporates the physical limitation and total symptom domain into a single score that is transformed into a range of 0 to 100 (higher scores represent better outcomes).

In EMPEROR-Reduced, the analysis based on the RS, including both on- and off-treatment values (Table 21 and Figure 19), showed a smaller decline from baseline of −1.30 points (SE = 0.69) in the empagliflozin group than in the placebo group (−3.36 points; SE = 0.69) in the KCCQ-CSS at week 52, with an adjusted mean difference of 2.06 (95% CI, 0.16 to 3.96) favouring empagliflozin. For patients who died, the worst score (score of 0) was imputed for all subsequent scheduled visits after the date of death. Responder analyses for an improvement (an increase in score of at least 5 points at week 52 from baseline) or a deterioration (a decrease in at least 5 points) were also conducted, although these outcomes were not part of the statistical testing hierarchy. At week 52, 40.0% of patients in the empagliflozin group reported at least a 5-point increase in the KCCQ-CSS compared with 35.9% of patients in the placebo group (OR = 1.23; 95% CI, 1.05 to 1.45).

|| |||||||||||||||||| ||| |||||||| ||||| || ||| ||| ||||||||| |||| ||| ||| ||||||||||||| |||||| |||||||||||| ||| |||||||||||| |||||| | |||||| |||||||| || ||| |||| |||||||| ||||||| ||||| |||| |||||||| || |||| |||||| ||| | ||||| || ||| ||||||||||||| ||||| |||||||| || ||||||| |||||| ||||||| || | ||||| || |||| ||| |||| || |||||||| |||| |||||||||| || |||| |||| ||| |||| || ||||| |||||||| |||||||||||||| ||||||||| |||||||| |||||| |||| || |||| ||| ||||| || |||||||| || ||| ||||||||||||| ||||| |||||||| || ||||| | ||||||| |||||||| || |||| |||||||| ||||||| |||||| |||||||| |||| ||||| || |||||||| || ||| ||||||| |||||| |||| |||| ||||| || |||| |||| ||| |||| || ||||||

Table 21

Change From Baseline in KCCQ Clinical Summary Score — EMPEROR-Reduced and EMPEROR-Preserved, RS.

Figure 19

Redacted.

Figure 20

Redacted.

KCCQ Total Symptom Score

Symptoms were measured using the KCCQ-TSS, which incorporates symptom burden and frequency into a single score that is transformed into a range of 0 to 100 (higher scores represent better outcomes).

|| |||||||||||||||| ||| |||||||| ||||| || ||| ||| ||||||||| |||| ||| ||| ||||||||||||| |||||| ||||||||||| ||| ||||||||||| |||||| | ||||||| ||||||| |||| |||||||| || ||||| |||||| ||| | ||||| || ||| ||||||||||||| |||| || ||| ||||||| ||||| |||||| ||||||| || | ||||| || ||| |||| ||||| ||||||| ||||| || |||| ||| |||| || |||||||| |||| |||||||||| || |||| |||| ||| |||| || ||||| |||||||| |||||||||||||| ||||||||| |||||||| |||||| |||| || |||| ||| ||||| || |||||||| || ||| ||||||||||||| ||||| |||||||| || ||||| | ||||||| |||||||| || |||| |||||||| ||||||| |||||| |||||||| |||| ||||| || |||||||| || ||| ||||||| |||||| |||| || || || |||| |||| ||| |||| || ||||||

|| |||||||||||||||||| ||| |||||||| ||||| || ||| ||| ||||||||| |||| ||| ||| ||||||||||||| |||||| |||||||||||| ||| |||||||||||| |||||| || |||||||| |||| |||||||| || |||| |||||| ||| | ||||| || ||| ||||||||||||| |||||||| || ||||||| |||||| ||||||| || | ||||| || ||| |||| ||||| ||||||| ||||| || |||| ||| |||| || |||||||| |||| |||||||||| || |||| |||| ||| |||| || ||||| |||||||| |||||||||||||| ||||||||| |||||||| |||||| |||| || |||| ||| ||||| || |||||||| || ||| ||||||||||||| ||||| |||||||| || ||||| | ||||||| |||||||| || |||| ||||| ||||||| |||||| |||||||| |||| ||||| || |||||||| || ||| ||||||| |||||| |||| || || || |||| |||| ||| |||| || ||||||

Table 22

Redacted.

Figure 21

Redacted.

Figure 22

Redacted.

KCCQ Overall Summary Score

The KCCQ-OSS incorporates the physical limitation, total symptom, social limitation, and HRQoL into a single score that is transformed into a range of 0 to 100 (higher scores represent better outcomes).

|| |||||||||||||||| ||| |||||||| ||||| || ||| ||| ||||||||| |||| ||| ||| ||||||||||||| |||||| |||||||||| |||||| | ||||||| ||||||| |||| |||||||| || ||||| |||||| ||| | ||||| || ||| ||||||||||||| ||||| |||| || ||| ||||||| ||||| |||||| ||||||| || | ||||| || ||| |||| ||||||| ||||||| ||||| || |||| ||| |||| || |||||||| |||| |||||||||| || |||| |||| ||| |||| || ||||| |||||||| ||||||||||||||||| |||||||||||||||||| ||| |||||||| ||||| || ||| ||| ||||||||| |||| ||| ||| ||||||||||||| |||||| ||||||||||| |||||| | |||||| ||||||||||| |||| |||||||| || |||| |||||| ||| | ||||| || ||| ||||||||||||| ||||| |||||||| || ||| ||||||| ||||| |||||| ||||||| || | ||||| || ||| |||| ||||||| ||||||| ||||| || |||| ||| |||| || |||||||| |||| |||||||||| || |||| ||||| || ||||| |||||||| ||||||||||||||

Table 23

Redacted.

Other KCCQ data reported included the change from baseline to week 52 in the KCCQ quality of life and patient-preferred outcomes, as well as the results of the analyses based on the TS, including only observed on-treatment data (Appendix 3).

|||||||||||||| ||||||| || |||| |||||||| || |||||||||||| ||| |||||||| || ||| ||||| |||||||||| ||||| || ||| || ||| ||| ||||||||||||||| |||||| |||| || |||| |||| ||||||| ||| ||||| ||| ||||| || |||||||| || ||| ||||||||||||| ||| ||||||| ||||||| ||||||||||||| ||||| |||| ||| |||| |||| ||||||| ||| ||||| || |||||||| || |||| |||||| || ||| ||||||||||||||||| |||||| |||| || |||| |||| ||||||| ||| ||||| ||| ||||| || |||||||| || ||| ||||||||||||| ||| ||||||| ||||||| ||||||||||||| ||||| |||| ||| |||| |||| ||||||| ||| ||||| ||| ||||| || ||||||||| ||||||||||||| ||| |||| ||||||| ||||||| ||| |||||| |||| |||||||| || ||| |||||||| ||| |||||||| ||||| ||||||||||| ||||||||||| ||| ||||| |||| || ||||||||||| |||||| ||| ||||||||| |||||| |||||||||||| ||||||||||| |||| ||| ||| |||||| |||||| |||| | |||||| |||||||||| || |||||||| |||||||| ||||||||||| || ||| |||| |||||| || ||||| || ||| ||| || ||| ||||||||||||| ||||| |||||||| || ||||||| || ||| ||||||||||||||| ||| ||||||||||||||||| |||||| ||||||||||||

Table 24

Redacted.

Functional Ability

Descriptive data for the NYHA functional class showed (Table 25) that more patients had improvement in their NYHA functional class in the empagliflozin group compared with placebo (26.8% versus 22.8%, and 22.6% versus 18.3% in EMPEROR-Reduced and EMPEROR-Preserved, respectively) at week 52.

Table 25

Change in NYHA Functional Class From Baseline at Week 52 — EMPEROR-Reduced and EMPEROR-Preserved, RS^a.

Adverse Events

In EMPEROR-Reduced, 1,420 patients (76.2%) in the empagliflozin group and 1,463 patients (78.5%) in the placebo group experienced at least 1 AE. Patients in the empagliflozin and placebo groups experienced TEAEs in a similar frequency (15.2% and 12.2%, respectively). The most common TEAEs occurring in at least 0.5% of patients in the empagliflozin or placebo groups were hypotension (2.3% versus 1.8%, respectively), renal impairment (1.4% and 1.1%, respectively), urinary tract infection (1.4% in each group), and ||||||||||||| ||||| ||| ||||| ||||||||||||||.

In EMPEROR-Preserved, 2,574 patients (85.9%) in the empagliflozin group and 2,585 patients (86.5%) in the placebo group experienced at least 1 AE. |||||||| || ||| ||||||||||||| ||| ||||||| |||||| ||||||||||| ||||| || | ||||||| ||||||||| |||||| ||| |||||| |||||||||||||| ||| |||| |||||| ||||| ||||||||| || || ||||| |||| || |||||||| || ||| ||||||||||||| || ||||||| |||||| |||| ||||||| ||||| ||||||||| ||||| ||| ||||| |||||||||||||| ||||||||||| ||||| ||| ||||| |||||||||||||| ||||| |||||||||| ||||| ||| ||||| |||||||||||||| ||| ||||||||||||| ||||| || |||| |||||||

Serious Adverse Events

In EMPEROR-Reduced, 772 patients (41.4%) in the empagliflozin group and 896 patients (48.1%) in the placebo group experienced 1 or more SAEs. Cardiac failure was the most frequently reported SAE (17.8% and 23.8% in the empagliflozin and placebo groups, respectively), followed by pneumonia (2.8% and 3.3% in the empagliflozin and placebo groups, respectively), acute kidney injury (1.9% and 3.0% in the empagliflozin and placebo groups, respectively), and atrial fibrillation (1.3% and 2.4% in the empagliflozin and placebo groups, respectively).

In EMPEROR-Preserved, 1,436 patients (47.9%) in the empagliflozin group and 1,543 patients (51.6%) in the placebo group experienced 1 or more SAEs. Cardiac failure was the most frequently reported SAE (15.0% and 19.9% in the empagliflozin and placebo groups, respectively), followed by pneumonia (3.3% and 4.0% in the empagliflozin and placebo groups, respectively), atrial fibrillation (3.1% and 2.7% in the empagliflozin and placebo groups, respectively), acute kidney injury (2.7% and 3.6% in the empagliflozin and placebo groups, respectively) and COVID-19 (1.6% in each treatment group).

Withdrawals Due to Adverse Events

In EMPEROR-Reduced, the overall frequency of AEs leading to treatment discontinuation was similar in the 2 treatment groups (17.3% and 17.6% in the empagliflozin and placebo groups, respectively). The most frequently reported types of withdrawals due to AEs were cardiac failure (3.5% and 3.7% in the empagliflozin and placebo groups, respectively), death (0.9% and 1.4% in the empagliflozin and placebo groups, respectively), acute myocardial infarction (0.6% and 0.3% in the empagliflozin and placebo groups, respectively), and renal impairment (0.5% and 0.2% in the empagliflozin and placebo groups, respectively).

In EMPEROR-Preserved, the overall frequency of AEs leading to treatment discontinuation was similar in the 2 treatment groups |||||| ||| ||||| || ||| ||||||||||||| ||| ||||||| ||||||| |||||||||||||. The most frequently reported types of withdrawals due to AEs were cardiac failure (1.5% and 2.1% in the empagliflozin and placebo groups, respectively), death (1.8% and 1.0% in the empagliflozin and placebo groups, respectively), renal impairment (0.7% in each group), and urinary tract infection (0.6% and 0.3% in the empagliflozin and placebo groups, respectively).

Mortality

In the EMPEROR-Reduced trial, the proportion of fatal AEs during the double-blind treatment phase was similar across the treatment groups ||||||. In EMPEROR-Preserved, |||| of AEs in the empagliflozin group and |||| in the placebo groups resulted in death.

Notable Harms

The frequency of notable harms identified in the protocol was comparable between the treatment groups.

In EMPEROR-Reduced, acute renal failure was the most commonly reported notable AE (9.4% and 10.3% in the empagliflozin and placebo groups, respectively), followed by hypotension (9.4% and 8.7% in the empagliflozin and placebo groups, respectively), urinary tract infection (4.9% and 4.5% in the empagliflozin and placebo groups, respectively), and bone fracture (2.4% and 2.3% in the empagliflozin and placebo groups, respectively). No new safety concerns were identified.

In EMPEROR-Preserved, acute renal failure was the most commonly reported notable AE (12.1% and 12.8% in the empagliflozin and placebo groups, respectively), followed by hypotension (10.4% and 8.6% in the empagliflozin and placebo groups, respectively), urinary tract infection (9.9% and 8.1% in the empagliflozin and placebo groups, respectively), and bone fracture (4.5% and 4.2% in the empagliflozin and placebo groups, respectively). No new safety concerns were identified.

Table 26

Summary of Harms — EMPEROR-Reduced and EMPEROR-Preserved, TS.

Table 27

Notable Harms — EMPEROR-Reduced and EMPEROR-Preserved, TS.

Internal Validity

Both the EMPEROR-Reduced and EMPEROR-Preserved trials appeared to have used accepted methods for blinding, allocation concealment, and randomization with stratification. For both EMPEROR trials, a computer-generated block randomization scheme was used, and randomization with stratifications was performed centrally, which typically has a low risk of bias. While both EMPEROR trials were double-blinded and the investigators were blinded to treatment assignment, risk of bias cannot be ruled out. In the empagliflozin group, about 78% of patients in EMPEROR-Reduced and 86% of patients in EMPEROR-Preserved experienced at least 1 AE, including TEAEs, which may have possibly made the investigator aware of the patient’s treatment assignment; however, AEs were similar between groups, so the risk of unblinding is likely low. The demographic and baseline patient characteristics appeared to be generally balanced between the treatment groups in both trials, so randomization was maintained. Both EMPEROR trials included only patients with elevated NT-proBNP, as high concentrations of NT-proBNP can confirm HF in patients who present with dyspnea when the clinical diagnosis remains uncertain.² However, the clinical experts consulted by CADTH for this review highlighted that clinicians only need to perform NT-proBNP tests in 10% to 20% of cases when they are unsure of the diagnosis of HF.

Only 1% of patients in EMPEROR-Reduced and 3% of patients in EMPEROR-Preserved were lost to follow-up for the primary end point, and vital status was known for more than 99% of patients. The results of the sensitivity analyses assessing missing data did not change the conclusions for the primary and key secondary outcomes. A relatively high proportion of patients prematurely discontinued the trial medication (26.7% and 31.5% in EMPEROR-Reduced and EMPEROR-Preserved, respectively), while the cause of discontinuations occurred at a similar frequency between the treatment groups. The clinical experts consulted noted that a high proportion of AEs leading to treatment discontinuation were fatal, which reflects the natural history of HF more than intolerance to the drug under review. In addition, patients who withdrew from the study and those who discontinued study medication early continued to be followed up and were included in outcome analyses. ||| |||||||| |||||||||| |||| |||||||| || ||||||||||||| | || || || |||||||| |||||| ||| | ||||||||| |||||| || |||| ||||||| ||||||| ||| ||| |||||||||| || |||||||| |||||||||| |||| |||||||||| ||||||| ||||||| ||| ||||||| || ||| ||||||||||| |||||||| |||||| |||| ||| |||||||| |||||||| ||| || |||||||||| |||||| || ||| |||||||| || ||| ||||||| ||| ||| ||||||||| ||||||||| || |||| |||||||

An independent clinical expert committee performed a central adjudication of the primary and key secondary outcomes based on criteria defined a priori in a blinded manner. The primary composite outcome in both EMPEROR trials was the time to first event of adjudicated CV death or adjudicated HHF. The clinical experts consulted indicated that this outcome was appropriate. However, the list of criteria used to define CV death in both trials appeared to be too comprehensive, as it includes death due to CV procedures and cardiac hemorrhage, which could have resulted in a similar proportion of CV deaths across the treatment groups in both trials. The key secondary outcomes in both trials were occurrence of HHF (first and recurrent) and eGFR (CKD-EPIcr equation) slope of change from baseline. The clinical experts indicated that reduction in the number of HHF events is 1 of the main outcomes used in clinical practice to assess the response to HF treatment, and the eGFR (CKD-EPIcr equation) slope of change is not usually used in clinical practice, although there is a strong relationship between kidney disease and HF. The statistical analysis methods appear to be acceptable. The interim and final analyses were planned a priori and adequately described. The results were robust to a number of different sensitivity analyses for the primary and key secondary outcomes. Subgroup analyses by LVEF, NYHA class, baseline diabetes status, renal function, prior use of HF medications, and history of atrial fibrillation were pre-specified in both EMPEROR trials and considered exploratory. The analyses may not have been powered to detect a treatment difference and there were no adjustments made for multiplicity, and the results should be viewed as supportive evidence only for the overall effect of empagliflozin. The interim analysis applied the Hwang, Shin, and De Cani alpha-spending function, which is deemed conservative in controlling type I error across the primary and 2 key secondary outcomes tested. While improvement in HRQoL, HF symptoms, and functional ability were of primary importance to patients with HF according to patient group input, these were exploratory outcomes and were outside the statistical testing hierarchy; thus, the results should be viewed as supportive evidence only for the overall effect of empagliflozin. The HRQoL and symptoms associated with HF were assessed using the KCCQ and EQ-5D-5L instruments ||||| ||||||||| ||||||||||||||| ||||| |||| |||||||||| |||| |||||| |||| ||||||||| ||||||| ||| |||| |||||||| ||| ||| |||||||| ||| |||| || |||||||| || |||||||| || |||||||||| ||||| || | |||| |||| || |||| || |||||||| ||| ||||||||| ||| |||||||||||||| ||| || ||||||||||||| ||||||||| |||| ||||| ||| ||| ||| |||||||| |||||| ||||||||||| || ||||||||| ||||||||| |||| ||||||. The KCCQ is a generally valid, reliable, and responsive questionnaire for CV diseases, including HF, and a 5-point difference in the KCCQ scores using in both EMPEROR trials was within the reported MID range (4.5 to 6). The literature search completed by CADTH did not find any evidence of the validity, reliability, responsiveness, and MID of the EQ-5D-5L instrument in patients with HF. The clinical experts consulted indicated that these tools are not used in clinical practice but are used in multiple studies, allowing comparisons between different treatments. Assessment of functional ability was based on the change in NYHA functional class from baseline at week 52 using descriptive statistics. The evidence of empagliflozin in patients with chronic HF was limited by 2 placebo-controlled pivotal trials, and no head-to-head evidence of empagliflozin compared against other comparators, including dapagliflozin, or sacubitril-valsartan in the HFrEF population, were available for this review.

External Validity

In general, the clinical experts consulted by CADTH for this review confirmed that the populations of both the EMPEROR-Reduced and EMPEROR-Preserved trials were similar to patients seen in Canadian clinics, and the study results would be generalizable to patients with HF in Canada, with some limitations. While empagliflozin has been approved by Health Canada for use as an adjunct to SOC therapy in patients with chronic HF, regardless of NYHA class, CADTH was unable to draw conclusions related to patients with NYHA functional classes of I and IV, since both trials excluded patients who had an NYHA class of I, and only a very small proportion of patients had an NYHA class of IV (0.3% to 0.5%). The clinical experts indicated they would not prescribe empagliflozin to patients with chronic HF with an NYHA class of I, as they are asymptomatic, which is consistent with the reimbursement request. One of the clinical experts consulted highlighted that the benefit of empagliflozin in patients with NYHA functional class IV is unclear due to limited clinical data and high mortality, while another clinical expert indicated that he would prescribe empagliflozin to patients with an NYHA class of IV.

About 48% of patients in both trials did not pass screening, most commonly because their NT-proBNP levels were below the pre-specified thresholds at screening, which further reduces the generalizability of the results. The clinical experts consulted indicated that NT-proBNP testing is not widely available in Canada, as some jurisdictions have limited access to it; thus, this patient selection criterion would be difficult to implement in clinical practice. In addition, the clinical experts consulted noted that this inclusion criterion likely created an enriched patient population in both trials, and patients with elevated NT-proBNP appeared to be sicker and could benefit more from treatment with empagliflozin than the population in the real-world setting. The clinical experts noted the majority of patients (about 64%) in the EMPEROR-Reduced trial had LVEF in the range of 20% to 30%; as such, they tend to be sicker. In the EMPEROR-Preserved trial, about 33% of patients had mid-range LVEF (41% to 49%); however, the clinical experts consulted do not expect this to be a major issue with the generalizability of the trial results to patients with HFpEF, as the LVEF definition is arbitrary, and estimates of LVEF may vary depending on the patient or technical factors as well as on clinical deterioration.

The clinical experts consulted noted that patients included in both EMPEROR trials were younger, as the median age of the population with HF in the real-world setting is approximately 75 years. The generalizability of the EMPEROR-Reduced trial results may be compromised by the high proportion of males (more than 75%) who were enrolled, as the clinical experts indicated that half of the population with HFrEF in Canada is female. Nonetheless, the clinical experts noted they would treat both male and female patients with chronic HF with empagliflozin. The majority of patients in both EMPEROR trials were receiving guideline-recommended treatment of HF; thus, they represented patients who were optimally managed, while the clinical experts noted that a goal-directed HF treatment is suboptimal in clinical settings. In addition, only about 3% of the patients in both EMPEROR trials were recruited from Canada. However, the clinical experts noted that the lack of representation of patients from Canada does not reduce the generalizability of the results to Canadian clinical practice. Lastly, although the recommended dose of empagliflozin for the treatment of HF is 10 mg, clinical experts indicated that both doses of empagliflozin, at 10 mg and 25 mg, are used in clinical practice.

Objectives

In the absence of direct comparative evidence from trials, the aim of this analysis was to compare the efficacy of empagliflozin plus SOC in patients with HFrEF versus dapagliflozin plus SOC in patients with HFrEF.

||||| ||||||||| |||||||||| |||||||| ||||||| || ||||||||| | |||||||||| |||||| ||| ||||||||| || ||| |||||||| ||| ||||||| || ||||| ||| |||||||||| || ||||||||| |||||||| |||||||||| |||| ||||| || ||| ||| |||| |||| ||||||| |||||||| || ||||||| || ||||| |||||||| |||||||| |||| |||| ||| ||||||| ||| ||||||| ||||||| |||||||||| ||| ||| |||||||| |||||||| ||||| |||||||||| ||||||||||| |||| ||| |||||||| ||||| |||||||||||| |||||||| ||||||| || |||||||||| |||||||| ||| |||||||| ||||||| || |||||||||| |||| |||| |||||||| || |||||||| ||||||||||| |||||||| ||||||||| ||| |||||||| ||||||||| |||| |||||||| ||| ||||||||| || ||| ||||||||||| ||||||||||

Table 28

Redacted.

| ||||| || || ||||||| |||| |||||||||| |||| ||| |||||||||||| |||| |||||||| ||| ||||||||| || ||| |||||||||| ||||||| |||| ||||| |||||||||| || ||||||| || ||| ||||| ||||||||||| ||| ||||||| ||||||| ||||||| ||| |||||||| ||||||| ||| ||| |||||||| || ||| |||| ||| ||||||| ||||| |||| ||||||||||||| || ||| |||| ||||| ||||| ||||||||| ||||||||| |||||||| ||| |||| || |||||| ||| ||||||||| ||| |||||||| |||| ||| ||||||| || | ||||||| |||| ||||||||||| ||||||||||||| ||| | ||||||| |||| |||||||||||| |||||||||||||| |||||||| ||| ||||||| ||||| || |||||||| ||| ||| || ||| ||| ||||||| ||||| | |||||| ||| |||| ||||| ||||| ||||||||||| ||| ||||||| |||| |||| ||||||||| ||| ||| |||||||| || |||||||||| ||||||||| || |||| || ||||| ||| ||| ||||||| |||| ||| ||||||| |||||| |||| ||| |||| ||||| ||| ||||||||||||||| ||| ||||||| |||||||

||| ||||||||||| ||||||| ||||| ||| |||||||||||| || ||||| |||||| ||| ||||||| ||||||||||||||| ||||||| ||| ||||||||||||||| ||||| ||| ||| ||||||| |||||| ||||| ||| ||||||||||||| ||| ||||||| ||||| |||||||||||| || ||||||||| || |||||| || ||| |||||||||| ||||||| || ||| || |||||||| |||||||| ||||||||| ||||||| ||||||| ||||||||||||| ||| |||||||||||||| |||| ||| ||||||||||| |||| ||| |||||| |||||||||| ||| ||||||| ||| |||||| || |||||||| ||| ||||||||||||| || ||| |||||| |||||||| || ||| ||||||| ||| |||||||| ||| |||||||| ||||||||| |||||||| ||||||| ||| ||||| |||| |||| |||||||||| ||| |||| |||||||||||| || | |||||||||||| |||||| |||| ||||||||||| ||||||| ||||||||| ||||||| ||| |||||||||| ||||||| |||||| ||| || ||||||| ||||| ||||||||| || ||||||||| |||||| ||||||||| || |||||||||| |||||||| ||||||||||||| |||||||||| ||| ||||||||||| || ||||| ||| |||||||| ||||||||| ||||||| ||| | ||||| ||||||| |||||||||||| |||||||| || |||||||||| ||| |||||| |||||||||| ||| || ||| |||| ||||||| || ||| ||||||||||||||| ||| |||||||| ||||||||||| || ||| |||||| | || |||||||||||| ||| || |||| || |||||||| ||| |||||||| |||||||| || ||||||||||||| ||||||| |||||||||||||| ||||||||||| || ||| ||||||| | ||| | || ||||||||||||| ||| |||||||| || ||| || |||||||| |||||||| |||||||| || ||| ||| || ||| |||||||| || ||| |||||| || ||| || ||||||||||||

Figure 23

Redacted.

||||||| || ||| ||||||| ||||||||| ||||| ||||||| || ||| |||||||| |||||| || ||| ||||||| ||||||||| ||| || |||||||| || ||||||||| |||| ||||||||||||||| ||| ||||||| |||| ||||| || |||||| |||||| |||||||||||| ||||||| ||||||||| ||| || |||||| ||||||||||||| || |||||||||||||| ||||||||||||| ||| |||| |||||||| || ||||||||| ||| |||||| || ||||||| ||||| |||||||| |||||||| || ||||||||||||||| ||| |||||||| ||| ||||||||| |||||||| |||| ||||||| |||| ||| ||||||||| || ||||||||||||||| ||||||||| | |||||| |||||||| ||||||||| ||| |||||| |||| |||||||||| ||||| ||| ||||||| |||||||| || ||||||||||||||| ||| ||||||||||||| ||||||||||||||| ||| ||||| ||||||| || || |||||| ||||||| || |||||||| |||| ||||||||| |||||||| |||| |||||||| |||||| ||||| ||||||| ||||||| ||| ||| |||||||| || ||| ||||||||| ||||| ||||||||| |||| |||||||||| ||||||||||| |||| ||||||| |||| |||||||| || ||| |||||||| ||||||||| || ||| |||||||| |||| |||| || |||| |||||||||| || |||| ||| |||| || |||||

Table 29

Redacted.

|| | |||||| ||||||||||||| || | |||||||| ||||||||| |||| | ||||||||| |||||||||| |||||||||| ||||| || | ||||| |||||||| ||| | ||||||||||||||| ||| ||||| |||||||| ||||||||| | |||||||||| ||||||||| ||||||||||| |||||||| |||| | ||| |||| ||||| |||||||||||| || | |||||||| |||||||||| ||| | |||||||| || ||||| || | |||||||| |||||||| ||||||| ||| ||||||||||||||||| ||||||||||||||| ||| ||| | |||||||| ||||||| ||| |||||||||| || ||||||||| ||||| |||| | |||||| |||||||||| || |||||||| |||| |||| ||||| ||| ||||||| || ||||||| |||||| ||| ||||| || |||| |||| |||||||| || ||||||||||||||| |||||| || |||| |||||| ||| |||||||| |||||||||||| ||| |||||||| ||||||| ||| ||||||| |||| ||||| || ||||| || |||||||| || ||| ||||||||||||||| ||||| |||| ||||| || ||||| ||||||| |||||||| || ||||| || ||||| || |||||||| |||||||| |||||||| || |||||||| ||| ||||| || ||||||||||||||| |||||| || |||||| |||||||| || ||||||| |||||| || ||||||| ||||||||| ||| |||||| || ||||||||||||||| |||||| ||||| || ||||| |||||| |||||||| || ||||||| |||||| ||||| || ||||| |||||||

Table 30

Redacted.

||| | |||| |||| |||||| |||| | ||||||||| |||||||||| |||||||||| ||||| || | ||||| |||||||| ||| | ||||||||||||||| ||| ||||| |||||||| ||| | ||||||||||||| |||||| |||| | |||| ||||||||||| |||||||| ||||||||| ||||||||| | |||||||||| ||||||||| ||||||||||| |||||||| |||| | ||| |||| ||||| |||||||||||| || | |||||||| |||||||||| |||||||| ||||||| ||| ||||||||||||| ||||||||||| || ||| || |||| ||||||| || |||||||||| || ||||||| ||| ||| ||||||| || |||| ||||||||||||||| ||| ||||||| |||| ||| ||||||||| ||||||||||| ||||| ||| | |||||| ||| || |||||||||| | ||||||||| || ||||||||||||||| |||||| || ||| |||||| ||| ||| ||||| || ||| ||||||| |||| |||||||| || ||||||| |||||| || ||| |||||| ||| ||| ||||| || ||| ||||||| ||||| ||||| ||| | |||||| ||| || ||| |||||||||| || |||| || ||| ||||||| |||||| || ||| |||||| ||| ||| ||||| || ||| ||||||| |||| ||||||| || ||||||||||||||| |||||| || ||| |||||| ||| ||| ||||| || ||| ||||||| |||||

Table 31

Redacted.

|||||||| ||||||| || ||| ||||||| ||||||||| ||||||| || ||| ||| |||||||| ||||||||| |||| ||| |||||||||| |||||||| ||||| |||| ||| |||||||| |||||||| ||||||||| ||| ||||||||||||| |||||| ||||||||||||| |||| ||| ||| ||| ||||||||| || |||||||| ||||| |||| || ||||||||| |||| |||||||| || | || ||||||||| | ||| ||| |||||||||| || ||||||||||||| |||||| |||||||||||||| ||| ||||||| |||||||| |||| |||| |||||||| |||| || ||||| ||||| || ||||||||||| || ||||| || ||||||||||| |||| |||||| || |||||||||| ||||||| ||||||||||||| ||| ||||||||||||| |||| | || |||| ||| || |||| ||||| || ||||||

Table 32

Redacted.

EMPERIAL-Reduced

The EMPERIAL-Reduced (effect of empagliflozin on exercise ability and HF symptoms in patients with chronic heart failure with reduced ejection fraction) trial was a phase III, multicentre, randomized, double-blind, placebo-controlled trial that aimed to evaluate the effect of empagliflozin (10 mg once daily) on exercise capacity and patient-reported outcomes compared with placebo in patients with HFrEF (LVEF ≤ 40%), with or without type 2 diabetes mellitus. A total of 312 patients were enrolled across 109 sites from 11 countries (Australia, Canada, Germany, Greece, Italy, Norway, Poland, Portugal, Spain, Sweden, and the US). Patients were randomized in a 1:1 ratio to receive either empagliflozin at a dosage of 10 mg once daily (n = 156) or matching placebo (n = 156) in a double-blind manner. Among these 312 patients, the mean age was 69.0 years (SD = 10.2 years) and the majority of patients were male (74.4%) and White (84.3%). The cause of HF was ischemic in 50.6% (n = 158) of participants, the mean LVEF was 30.3% (SD = 6.7%), and diabetes was present in 59.9% (n = 187) of participants. Beta blockers (94.6%), loop or high-ceiling diuretics (87.8%), lipid-lowering drugs (79.2%), MRAs (58.3%), ACEIs, and ARBs (55.4%) were the major medications for treating patients, and about 36.5% (n = 114) of participants were treated with ARNIs at baseline. The median baseline 6MWTD was 309.0 m (IQR, 248.5 to 332.0) with placebo and 306.0 m (IQR, 260.0 to 333.5) with empagliflozin. The number of participants who discontinued the trial medication prematurely for any reason was 13 (8.3%) with placebo and 15 (9.7%) with empagliflozin. The study was funded by Boehringer Ingelheim.^10,11

EMPERIAL-Preserved

The EMPERIAL-Preserved (effect of empagliflozin on exercise ability and HF symptoms in patients with chronic HF with preserved ejection fraction) trial was a phase III, multicentre, randomized, double-blind, placebo-controlled study that aimed to evaluate the effect of empagliflozin (10 mg once daily) on exercise capacity and patient-reported outcomes as compared with placebo in patients with HFpEF (defined as LVEF > 40%), with or without type 2 diabetes mellitus. A total of 315 patients were enrolled across 108 sites in 11 countries (Australia, Canada, Germany, Greece, Italy, Norway, Poland, Portugal, Spain, Sweden, and the US). Patients were randomized in a 1:1 ratio to receive either empagliflozin at a dose of 10 mg once daily (n = 157) or matching placebo (n = 158) in a double-blind manner. Among these 315 patients, the mean age was 73.5 years (SD = 8.8 years) and the majority of patients were male (56.8%) and White (87.3%). The cause of HF was ischemic in 50.6% (n = 158) of participants, the mean LVEF was 53.1% (SD = 8.0%), and diabetes was present in 51.1% of participants (n = 161). Beta blockers (89.2%), ACEIs, and ARBs (74.6%), lipid-lowering drugs (74.0%), and loop or high-ceiling diuretics (71.7%) were the major medications for treating patients, and about 3.5% (n = 11) of participants were treated with ARNIs at baseline. The median 6MWTD was 299.5 m (Q1 to Q3, 245.0 to 331.0) with placebo and 297.0 m (Q1 to Q3, 246.0 to 326.0) with empagliflozin. The number of participants who discontinued the trial medication prematurely for any reason was 11 (7.0%) with placebo and 13 (8.3%) with empagliflozin. The study was funded by Boehringer Ingelheim.^10,11

The primary end point in the EMPERIAL-Preserved and EMPERIAL-Reduced trials was the change from baseline in 6MWTD at week 12. The key secondary end points in both trials were the change from baseline in KCCQ-TSS at week 12 and the change from baseline in CHQ-SAS dyspnea score at week 12. Other secondary end points were: change from baseline in 6MWTD at week 6, change from baseline in Clinical Congestion Score at week 12, change from baseline in Patient Global Impression of Severity of HF symptoms at week 12, change from baseline in Patient Global Impression of Severity of dyspnea at week 12, Patient Global Impression of Change in HF symptoms at week 12, Patient Global Impression of Change in dyspnea at week 12, and change from baseline in NT-proBNP at week 12.^10,11

Table 33

Details of Other Relevant Studies — EMPERIAL-Reduced and EMPERIAL-Preserved.

Efficacy

Only results for KCCQ-TSS, CHQ-SAS dyspnea score, Clinical Congestion Score (summary score of orthopnea, jugular venous distension, and edema), Patient Global Impression of Severity, and Patient Global Impression of Change are presented in accordance with the protocol for the CADTH review. The median difference from baseline to week 12, empagliflozin versus placebo, in KCCQ-TSS was 3.13 (95% CI, 0.00 to 7.29) and 2.08 (95% CI, −2.08 to 6.25) in EMPERIAL-Reduced and EMPERIAL-Preserved, respectively. The median difference, empagliflozin versus placebo, in the CHQ-SAS dyspnea score was 0.10 (95% CI, −0.20 to 0.40) and −0.07 (95% CI, −0.35 to 0.20) in EMPERIAL-Reduced and EMPERIAL-Preserved, respectively. More participants taking empagliflozin versus placebo showed improvements in KCCQ-TSS in pre-specified clinically meaningful thresholds (5 points or greater and 8 points or greater), with adjusted ORs of 1.83 (95% CI, 1.12 to 2.98) and 1.66 (95% CI, 1.02 to 2.72), respectively, in the EMPERIAL-Reduced trial. Analyses assessing the same cut-offs did not suggest any treatment difference in the EMPERIAL-Preserved trial. Reduction in Clinical Congestion Score at week 12 for empagliflozin versus placebo was −0.31 (95% CI, −0.53 to −0.09) in EMPERIAL-Reduced and −0.09 (95% CI, −0.31 to 0.14) in EMPERIAL-Preserved. No significant changes in Patient Global Impression of Severity or Patient Global Impression of Change in HF symptoms or dyspnea were observed in either study. Seven participants (4.5%) in the empagliflozin group versus 25 (16.1%) in the placebo group required intensification of diuretic therapy in EMPERIAL-Reduced, and 17 (11.0%) versus 24 (15.4%), respectively, required intensification of diuretic therapy in EMPERIAL-Preserved.¹¹

Harms

In terms of AEs, there was no notable difference between the 2 trials for empagliflozin versus placebo regarding the overall frequency of any AE or any AE leading to treatment discontinuation. SAEs were reported less frequently with empagliflozin compared with placebo in EMPERIAL-Reduced (12.7% with empagliflozin versus 18.4% with placebo) and EMPERIAL-Preserved (13.5% with empagliflozin versus 17.3% with placebo). Decreased kidney function was reported with similar frequencies in both groups. No ketoacidosis or confirmed hypoglycemic events occurred in participants without type 2 diabetes. No new safety concerns were identified.¹¹

Critical Appraisal

The following limitations were identified:

• HF is a chronic condition, which means the progression of HF is generally slow; thus, the assessment of change in outcomes may require a long-term follow-up period.
• The follow-up period for the EMPERIAL-Reduced and EMPERIAL-Preserved trials is 12 weeks, which may not be sufficient to assess meaningful changes in the outcome measures.
• The EMPERIAL trials were powered to detect an improvement of 30 m in the 6MWTD; however, the study sample size may not be sufficient enough to detect any between-group changes of less than 30 m.
• As the primary end point (change from baseline in 6MWTD at week 12) was not met, the analyses of all secondary outcomes, such as the KCCQ-TSS and CHQ-SAS dyspnea score, were considered exploratory.
• While the changes in the KCCQ-TSS and CHQ-SAS dyspnea score may suggest a possible favourable effect of empagliflozin in patients with HFrEF, these results are considered exploratory.
• The baseline demographic and baseline characteristics (sex and 6MWTD) were suggestive of an over-representation of male patients with lower functioning status, which may compromise the representativeness of the study sample compared with the general population of adult patients with HF.

Although the EMPERIAL studies provide additional data on the effectiveness and safety of empagliflozin in patients with HF, the limitations identified introduce uncertainty.

ClinicalTrials.gov

Produced by the US National Library of Medicine. Targeted search used to capture registered clinical trials.

• [Search -- (empagliflozin OR jardiance OR “BI 10773” OR BI10773) AND (“heart failure” OR “cardiac failure” OR “cardiac insufficiency” OR “myocardial failure” OR “heart insufficiency”)]

WHO ICTRP

International Clinical Trials Registry Platform, produced by the WHO. Targeted search used to capture registered clinical trials.

• [Search terms -- (empagliflozin OR jardiance OR “BI 10773” OR BI10773) AND (“heart failure” OR “cardiac failure” OR “cardiac insufficiency” OR “myocardial failure” OR “heart insufficiency”)]

Health Canada’s Clinical Trials Database

Produced by Health Canada. Targeted search used to capture registered clinical trials.

• [Search terms -- (empagliflozin OR jardiance OR “BI 10773” OR BI10773) AND (“heart failure” OR “cardiac failure” OR “cardiac insufficiency” OR “myocardial failure” OR “heart insufficiency”)]

EU Clinical Trials Register

European Union Clinical Trials Register, produced by the European Union. Targeted search used to capture registered clinical trials.

• [Search terms -- (empagliflozin OR jardiance OR “BI 10773” OR BI10773) AND (“heart failure” OR “cardiac failure” OR “cardiac insufficiency” OR “myocardial failure” OR “heart insufficiency”)]

Description and Scoring

The KCCQ is a self-administered, 23-item, disease-specific HRQoL questionnaire that was originally developed in 2000 to measure the patient’s perception of their health status within a 2-week recall period.^27-29 The items of the KCCQ can be categorized into the following domains: physical limitation, symptoms (frequency, severity, and recent change over time), social limitation, self-efficacy, and HRQoL. Responses are scored using ordinal values, beginning with 1 for the response that implies the lowest level of functioning. Domain scores are transformed to a 0 to 100 range by subtracting the lowest possible scale score, dividing by the range of the scale, and multiplying by 100. Missing values within each domain are assigned the average of the answered items within the same domain.^27,29Various combinations of the KCCQ domains create 3 KCCQ summary scores including the KCCQ-TSS, KCCQ-CSS, and KCCQ-OSS. The KCCS-TSS combines the symptom burden and symptom frequency domains and evaluates patient-reported swelling in feet, ankles, or legs; fatigue; shortness of breath; and disturbed sleep.³⁰ The KCCQ-CSS includes the physical limitation and total symptom domains, and the KCCQ-OSS combines the physical limitation, total symptom, social limitation, and HRQoL domains into a single score. Summary scores are then transformed to a 0 to 100 range, where larger scores represent a better outcome: 0 to 24: very poor to poor; 25 to 49: poor to fair; 50 to 74: fair to good; and 75 to 100: good to excellent.^27,29

Validity

The KCCQ was originally validated in patients with a clinical diagnosis of congestive HFrEF (LVEF < 40%).²⁷ A cohort of patients (N = 39; mean age 64 years; 69% male; mean NYHA = 2.0 ± 0.59) with stable disease was used to assess the validity of the KCCQ. Convergent validity was demonstrated through a strong correlation of the KCCQ physical limitation domain with NYHA classification (Spearman’s correlation coefficient r = −0.65) and Minnesota Living with Heart Failure Questionnaire (MLHFQ)Physical (r = 0.65), and a moderate correlation with the 6-minute walk distance (6MWD) (r = 0.48). The quality of life domain were strongly correlated with NYHA classification (r = −0.64). The social limitation domain was strongly correlated with NYHA classification and the Short Form (36) Health Survey social limitation scale (r = 0.62). No adequate criterion standard was available for the self-efficacy domain.²⁷

Convergent validity has also been assessed in a variety of other publications.^31-35 Napier et al.,³¹ assessed convergent validity in patients with HFpEF (n = 110). The KCCQ-OSS, KCCQ-CSS and KCCQ physical limitation score (KCCQ-PLS) showed moderate correlations with NYHA class (I to IV) and the 6MWD (range of Spearman rank order correlation coefficient, r = −0.38 to 0.47; P < 0.001, for each). The KCCQ quality of life (KCCQ QoL) score was weakly correlated with NYHA functional class (r = −0.28; P = 0.003) and 6MWT (r = 0.19; P = 0.04). The KCCQ self-efficacy score was not correlated with NYHA functional class (r = −0.10; P = 0.30) or 6MWT (r = −0.02; P = 0.87). These findings were corroborated in patients with HFrEF with regard to the convergent validly of KCCQ-OSS in the FAIR-HF trial (N = 459). There were moderate correlations between the Patient Global Assessment (PGA) and the KCCQ-OSS (r = 0.31; P < 0.001, and r = 0.42; P < 0.001), the KCCQ-CSS (r = 0.36; P < 0.001, and r = 0.42; P < 0.001), and the KCCQ-PLS (r = 0.31; P < 0.001, and r = 0.39; P < 0.001) at 4 and 24 weeks, respectively.³⁵ Similar findings were observed in a publication assessing the convergent validity of the KCCQ-PLS in a population of patients with HFpEF in the VITALITY-HFpEF trial (N = 698). There were moderate correlations between the Patient Global Impression of Change and the KCCQ-PLS (r = 0.28, and r = 0.31, at week 12, and 24, respectively).⁷¹ Convergent validity was further analyzed in a cohort of patients with stable compensated HF (N = 41; mean age = 68 ± 12 years; 100% male). The KCCQ-TSS moderately correlated (r = 0.30) with peak VO₂.³³ This evidence bundle presented supports the presence of convergent validity of the KCCQ-OSS, and the total symptom score. However, in a publication by Tucker et al.,³⁴ the authors assessed the presence of convergent validity in a population of patients hospitalized with chronic HF (N = 233). The authors found no evidence of convergent validity, when the KCCQ domain scores and summary scores (KCCQ-OSS and KCCQ-CSS) were correlated with NYHA class (either class III or IV), BNP levels, and the Charlson Comorbidity Index scores. The authors explain that this may be due to the presence of an alternate population in the current study compared with previous studies analyzing the convergent validity of the KCCQ.³⁴ Nevertheless, these findings taken together support the presence of convergent validity for the KCCQ-OSS, KCCQ-PLS, and the total symptom score.

Concurrent validity of the KCCQ was assessed by administration of the KCCQ and the Minnesota Living with MLHFQ to patients with HFpEF (N = 110) at baseline, 6 weeks, and 12 weeks in the Nitrate Effect on Activity Tolerance in Heart Failure (NEAT) trial. The level of agreement of change was moderate (Cohen kappa statistic = 0.36; 95% CI, 0.2 to 0.52), supporting the presence of concurrent validity.³¹

Reliability

The internal consistency reliability of the KCCQ domains and summary scores (KCCQ-OSS and KCCQ-CSS) has been assessed in several studies and has demonstrated consistent results across all studies.^27,30-32,34 In a number of publications, the KCCQ domains, with the exception of the self-efficacy domain has consistently presented Cronbach alpha values > 0.7.^27,30,31,34 The KCCQ self-efficacy domain has been evaluated in a number of studies, and has demonstrated Cronbach alpha values in the range of 0.61 to 0.63,^27,30 with 1 publication calculating the Cronbach alpha value > 0.7 for this domain.³⁴ The KCCQ-CSS, KCCQ-OSS, and KCCQ-TSS have demonstrated Cronbach alpha values greater than 0.7, 0.93 to 0.95, and 0.8, respectively.^27,31 Lastly, these findings were confirmed in a meta-analysis performed by Garin et al., where Cronbach alpha values were > 0.7 for all KCCQ domains, with the exception of the self-efficacy domain (Cronbach alpha = 0.62 to 0.66).³²

Test-retest reliability of the KCCQ has been evaluated in multiple studies.^27,32,37 In the original paper evaluating the KCCQ, among those with stable HF who remained stable (N = 39), mean changes in KCCQ domains and summary scores (KCCQ-OSS and KCCQ-CSS) over the 3 months of observation were 0.8 to 4.0 points, none of which were statistically significant.²⁷ A meta-analysis which summarized the test-retest reliability of the KCCQ domains found an acceptable ICC (> 0.7) for the KCCQ symptom domain, the physical limitation domain, and the social limitation domain, but an ICC < 0.7 for the KCCQ self-efficacy, and the quality of life domains.³² Furthermore, in a cohort of 280 patients with chronic stage-C HF, test-retest reliability was assessed at baseline and at 6 months, and ICC > 0.7 were demonstrated for the physical limitation domain, and the symptom domain, but not for the self-efficacy domain.³⁰ Taken together, these findings suggest that the KCCQ symptom, physical limitation, and social limitation domains have acceptable test-retest reliability, while the KCCQ self-efficacy and quality of life domains do not demonstrate acceptable test-retest reliability.

Responsiveness

In the original study validating the KCCQ, a cohort of patients with HF, which were admitted to the hospital for HF exacerbations were used to assess the responsiveness of the KCCQ. The KCCQ exhibited high responsiveness, with Guyatt’s responsiveness statistics ranging from 0.62 for the social limitation domain to 3.19 for the symptoms domain, and was specifically 2.77 for the KCCQ-CSS and 1.74 for the KCCQ-OSS.²⁷ Another study evaluated the responsiveness of the KCCQ in patients with stable chronic HFpEF (N = 110). None of the KCCQ domains were responsive to changes in NT-proBNP. Of the KCCQ scores evaluated, the KCCQ-OSS and the KCCQ-CSS were ranked as the most responsive to improvement, and deterioration in distance walked in the 6MWD, respectively.³¹ These findings were corroborated in a study completed by Eurich et al. which evaluated the responsiveness of the KCCQ-CSS and the KCCQ-OSS sin a cohort of patients with HF (N = 298). Irrespective of the responsiveness index used, the KCCQ-CSS and the KCCQ-OSS were consistently ranked as the most responsive measures.³⁶ Furthermore, a meta-analysis which evaluated the responsiveness of 5 domains of the KCCQ (physical limitation, social limitation, symptom, HRQoL, and self-efficacy) produced very large effect sizes (from 0.6 to 3.2), indicating high responsiveness of the KCCQ domains.³² Taken together these findings indicate that the KCCQ domains and the KCCQ summary scores exhibit evidence of high responsiveness to change.

Minimal Important Difference

Baseline data from a large randomized controlled trial (HF-ACTION; N = 2,331; mean age = 59.1 years; 71.6% male; 63.4% NYHA class II, 35.7% class III, and 1% class IV) were used to examine associations between the KCCQ domain and summary scores, and clinical indicators of disease severity, including the 6MWD and peak VO₂.³⁷ In this study, a 1-SD difference in 6MWD and peak VO₂ was found to be associated with an approximately 5-point difference in the KCCQ-OSS, a 6-point difference in the KCCQ-CSS, and a 5-point difference in the KCCQ-TSS. The authors considered a 1-SD difference in 6MWD and peak VO₂ to represent a meaningful difference in patients with HF, citing that it is a more stringent criterion used for these indicators than previous studies.³⁷ This finding was corroborated when the KCCQ-OSS was associated with clinical change as assessed by a cardiologist (15-point Likert scale, from extremely worse to extremely better and grouped into categories of change) in a study (N = 476; mean age = 61 years; 75% male; 11% NYHA class I, 41% class II, 44% class III, and 5% class IV) in patients with HF and an ejection fraction < 40%.⁷⁰ When the KCCQ-OSS was administered at baseline and at 6 weeks, a mean improvement of 5.7 points in the KCCQ-OSS was associated with a small improvement in HF. A mean decrease of 5.4 points in the KCCQ-OSS was associated with a small deterioration in HF.⁷⁰ Furthermore, the minimal clinically important difference (MCID) for various KCCQ domain scores was evaluated in the FAIR-HF trial (N = 459) in patients with HFrEF, using PGA scale as an anchor at 4 and 24 weeks.³⁵ At week 4, all of the KCCQ domains had less than a 5-point MID based on “little improvement” in PGA. At week 4 and 24, the MCID estimates for improvement were 3.6 (95% CI, 1.0 to 6.2) and 4.3 (95% CI, 0.2 to 0.4) for the KCCQ-OSS, 4.5 (95% CI, 1.8 to 7.2) and 4.5 = ; (95% CI, 0.2 to 8.4)) for the KCCQ-CSS, and 4.7 (95% CI, 1.4 to 8.0) and 4.9 (95% CI, −0.9 to 9.0) for the KCCQ-PLS, respectively.³⁵ With regards to patients who reported a slight worsening in their condition, MCID estimates for deterioration were −0.4 (95% CI, −8.6 to 7.7) and −5.0 (95% CI, −15.5 to 5.6) for the KCCQ-OSS, 1.4 (95% CI, −7.1 to 10.0) and −1.1 (95% CI, −11.7 to 9.4) for the KCCQ-CSS, 1.8 (95% CI, −9.1 to 12.7) and −1.7 (95% CI, −14.8 to 11.2) for the KCCQ-PLS, at week 4 and 24, respectively.³⁵ In patients with HFpEF, the MID for KCCQ-PLS for improvement or worsening were estimated in the VITALITY-HFpEF trial. The study used an anchor-based approach using Patient Global Impression of Change as an anchor and reported that a median change in KCCQ-PLS of more or equal to 8.33 points (corresponding to an improvement in ≥ 2 response categories of KCCQ-PLS) may represent the MID for improvement and a median change of ≤ −4.17 points (corresponding to a worsening in ≥ 1 response category of KCCQ-PLS) may suggest deterioration in patients with HFpEF.⁷¹

Description and Scoring

The EQ-5D-5L is a generic self-reported HRQoL outcome measure that may be applied to a variety of health conditions and treatments. The EQ-5D-5L was developed by the EuroQol Group as an improvement to the EQ-5D-3L to measure small and medium health changes and reduce ceiling effects.^38,39 The instrument comprises 5 dimensions: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. Each dimension is rated on 5 levels: level 1 “no problems,” level 2 “slight problems,” level 3 “moderate problems,” level 4 “severe problems,” and level 5 “extreme problems” or “unable to perform.”^38,39 A total of 3,125 unique health states are possible, with 55555 representing the worst health state and 11111 representing the best state. The corresponding scoring of EQ-5D-5L health states is based on a scoring algorithm that is derived from preference data obtained from interviews using choice-based techniques (e.g., time trade-off) and discrete choice experiment tasks.^38,39 The lowest and highest score vary depending on the scoring algorithm used. Scores less than 0 represent health states that are valued by society as being worse than dead, while scores of 0 and 1.00 are assigned to the health states “dead” and “perfect health,” respectively. As an example, a Canadian scoring algorithm results in a score of −0.148 for health state 55555 (worst health state) and a score of 0.949 for health state 11111 (best health state).^38,39 Another component of the EQ-5D-5L is a visual analogue scale (EQ VAS), which asks respondents to rate their health on a visual scale from 0 (worst health imaginable) to 100 (best health imaginable).^38,39

Assessment of Validity, Reliability, and Responsiveness

The literature search completed by CADTH did not find any evidence on the validity, reliability, responsiveness, and MID of the EQ-5D-5L questionnaire in patients with HF. However, there is evidence for these metrics for the EQ-5D-3L questionnaire and the EQ VAS in patients with HF. Since this is an exploratory outcome for the EMPEROR-Reduced and EMPEROR-Preserved trials under review, CADTH will provide a high-level summary of the EQ-5D-3L and the EQ VAS in an HF population.

The discriminant validity of the EQ-5D-3L was determined in a North American cohort study (N = 476) in patients with HF and an ejection fraction less than 40%.⁷⁰ The EQ-5D index and VAS c-statistic ranged from 0.56 and 0.58 for small clinical improvements, to 0.69 and 0.76 for moderate to large improvements.⁷⁰ From this study, the EQ-5D-3L was found to show less discriminative abilities than the KCCQ and NYHA class, but similar discriminative abilities to the 12-item Short Form Survey (SF-12). In addition, the EQ-5D and SF-12 did not exhibit much sensitivity to the magnitude of observed clinical change.⁷⁰

The responsiveness of the EQ-5D-3L was compared with the KCCQ and SF-12 in patients with HF and an ejection fraction less than 40% (N = 298).³⁶ Patients were administered questionnaires at baseline and 6 weeks in addition to a 6MWD. Overall, the EQ-5D index and VAS were less responsive than the KCCQ, but showed similar responsiveness to the SF-12.³⁶

A systematic review of studies looking at the validity and reliability of the EQ-5D-3L in patients with CV disease identified 10 studies.⁷² When EQ-5D-3L scores were stratified by disease severity in the HF studies, the mean EQ-5D index scores decreased from 0.78 (SD 0.18) for mild states to 0.51 (SD 0.21) for moderate/severe health states.⁷²

Minimal Important Difference

Baseline data from a large randomized controlled trial (HF-ACTION trial; N = 2,331) were used to examine associations between the EQ VAS and clinical indicators of disease severity, including the 6MWD and peak VO₂.³⁷ In this study, a 1 SD difference in 6MWD and peak VO₂ was found to be associated with an approximate 3-point difference in the EQ VAS. The 1 SD change in 6MWD and peak VO₂ used in the present study is considered a clinically meaningful difference to patients with HF, and is a more stringent criterium than typically used in previous studies.³⁷ Moreover, a Canadian-specific MID of 0.037 has been reported for the EQ-5D-5L.^38,39