Interleukin-1 Blockade in Recently Decompensated Systolic Heart Failure: Results from the REcently Decompensated Heart failure Anakinra Response Trial (REDHART)

Study Design

The study design of the REcently Decompensated Heart failure Anakinra Response Trial has been previously presented⁸. Approval was obtained from the Virginia Commonwealth University Institutional Review Board and all participants provided written informed consent. After an initial screening and evaluation, patients underwent a cardiopulmonary exercise test (CPX), a transthoracic echocardio-Doppler study, testing for biomarkers, at baseline, 2, 4, 12, and 24 weeks⁸. Patients were randomized in a 1:1:1 ratio to one of three treatment arms exploring 2 different durations of anakinra treatment: 1) anakinra 100 mg daily for 2 weeks, followed by placebo (equal volume of 0.67 mL) daily for the remaining 10 weeks; 2) anakinra 100 mg daily for 12 weeks; or 3) placebo daily for 12 weeks (Supplemental Figure 1). The anakinra dosing follows the approved regimen for rheumatoid arthritis and is the same regimen utilized in the 2-week proof-of-concept studies^5,6.

Screening and Enrollment

Consecutive patients were screened for enrollment during the initial hospital admission or within 2 weeks of discharge. The inclusion and exclusion criteria are presented in the Supplemental Table 1, and previously⁸. Briefly, patients were considered for enrollment if they had been admitted with a diagnosis of acute (or acute-on-chronic) systolic HF based on symptoms, signs, biomarkers, or non-invasive or invasive measures of elevated cardiac filling pressures and reduced left ventricular ejection fraction (<50%)⁸. Prior to enrollment, patients needed to be considered clinically stable and ready for hospital discharge, able and willing to follow study instruction and complete study tests, but with evidence of enhanced systemic inflammation, as shown by CRP plasma levels >2 mg/L (measured by high-sensitivity assay). Patients needed to be free of malignancy or systemic inflammatory illnesses⁸.

Randomization and allocation concealment

An independent investigator (GBZ) created a randomization sheet that was then provided to the Investigational Pharmacy in Richmond, Virginia, USA. Anakinra or placebo (vehicle) were provided by SOBI (Stockholm, Sweden) in 0.67 mL syringes identifiable by lot number, but otherwise indistinguishable. To further ensure concealment of group allocation, the investigators were blinded to all CRP levels throughout the study other than the screening level.

Data and Safety Monitoring Board

A Data and Safety Monitoring Board (DSMB) was formed according to USA Food and Drug Administration recommendations⁹. Supplemental Table 2 shows the composition of the DSMB.

Cardiopulmonary exercise testing (CPX)

A supervised maximal aerobic exercise test was administered using a metabolic cart adapted to a treadmill using a conservative ramping treadmill protocol, as described^5,6,8,10. The peak respiratory exchange ratio (RER) was used to determine subject effort and a peak value ≥1.00 was considered a minimal acceptable threshold^11,12. Patients who interrupted the CPX for reasons other than dyspnea (i.e. ischemia, arrhythmias, leg pain) were excluded and considered screen failures if occurring at time of the baseline test or the first follow up at 2 weeks, or kept in the analysis of clinical event, but the results of the CPX were not considered for that visit. The de-identified data from each CPX were transferred to Dr. Ross Arena, PhD, at the Core Lab at the University of Illinois at Chicago, Chicago, Illinois, USA, for analysis.

Doppler echocardiography

All subjects underwent a transthoracic Doppler echocardiogram prior to initiation of treatment and at each additional visit to measure left ventricular dimension and function, as previously described^13,14. All measurements occurred off-line at the end of the study by two separate operators blinded to group allocations. The average of the two measurements was used for measures differing ≤10%, whereas those with >10% were re-reviewed and discussed in order to achieve consensus. We used the non-invasive hemodynamic monitor ccNexfin^® (BMEYE, Amsterdam, The Netherlands; and Edwards Lifesciences, Irvine, CA, United States) ¹⁵ coupled with transthoracic echocardiography to calculate arterial elastance (E_a) ¹⁶, end-systolic elastance (E_es)¹⁷, and ventriculo-arterial coupling was defined as the E_a/E_es ratio (Supplemental data).

Laboratory analysis and biomarkers

Blood samples were used for a complete blood count with differential, comprehensive metabolic profile, and plasma levels of biomarkers (e.g. high sensitivity C-reactive protein [CRP], brain natriuretic peptide [BNP], inflammatory cytokines).

Quality of life assessments

All patients completed 2 different quality of life questionnaires at each visit: the Duke Activity Status Index (DASI) ¹⁸ and the Minnesota Living with Heart Failure (MLWHF) ¹⁹. These questionnaires were selected to explore different aspects of HF symptoms: a lower DASI score reflects impaired functional capacity, whereas a higher MLWHF score reflects greater symptom burden.

Clinical events

A dedicated committee adjudicated all clinical events (Supplemental Table 3). The adjudicated events included: 1) Death; 2) Cardiac and 3) Non-Cardiac death (in which a direct cause attributable to cardiac disease is not present); 4) Hospitalization for HF (in which the primary diagnosis for hospitalization is decompensated HF established as the finding at admission of both of the following conditions listed: a. dyspnea or respiratory distress or tachypnea at rest or with minimal exertion; b. evidence of elevated cardiac filling pressure or pulmonary congestion [pulmonary congestion/edema at physical exam or chest X-ray; plasma BNP levels ≥200 pg/mL; or invasive measurement of LV end-diastolic pressure >18 mmHg or of pulmonary artery occluding pressure >16 mmHg]²⁰; 5) Acute myocardial infarction²¹; 6) Acute renal failure (defined as in increase in plasma creatinine levels of 100%); 6) Sepsis (or other serious infection requiring antibiotic therapy); and 7) Acute stroke²².

Study end-points

The co-primary endpoints were the placebo-corrected changes in peak VO₂ or the VE/VCO₂ slope after 2 weeks of treatment, by comparing patients treated with anakinra (pooled anakinra group, N=40) vs. placebo (n=20)⁸. The rationale for using the 2-week endpoint as primary endpoint derives from the improvement in peak VO2 after 2 weeks of treatment, documented in 2 pilot trials in stable HF patients^5,6. Additional endpoints included: 1) changes in peak VO₂ or the VE/VCO₂ slope at 4, 12 and 24 weeks in each of the 3 groups; 2) structural and functional parameters at Doppler echocardiography at each follow up visit in each group to provide mechanistic insight; 3) Quality of life assessment: the MLWHF and the DASI at 2, 4, 12 and 24 weeks in each of the 3 groups; 4) Biomarkers; and 5) Clinical outcomes: such as death (cardiac and non-cardiac), re-hospitalization for HF or for other causes up to 24 weeks.

Statistical analysis

Descriptive summaries of continuous measurements are reported as median and interquartile ranges due to potential deviation from Gaussian distribution. Descriptive summaries of categorical measurements consist of frequencies, proportions and 95% confidence intervals, when applicable. All analyses were conducted after database locking on November 12, 2016 and based on the intention-to-treat principle (i.e., analyzing groups as randomized and including all patients with outcome data available). The Statistical Package for Social Studies (SPSS) software 22.0 (IBM, New York, NY, US) was used. The difference in interval changes in peak VO₂ or the VE/VCO₂ slope at 2 weeks between the pooled anakinra vs. placebo groups were compared using random-effect analysis of variance for repeated measures to analyze the effects of treatment within each group and the effect of time_x_group allocation. Unadjusted p-values are reported throughout, with statistical significance for the co-primary endpoints set at the 2-tailed 0.025 level. Cases with missing data for the primary endpoint were omitted from the analysis. To evaluate the group differences in the secondary endpoints, data was compared across all groups using the random-effect analysis of variance for repeated measures as indicated above for paired analyses, or Kaplan-Meier curves with Log-rank testing for event rates. Plasma levels of CRP were compared across groups after log10 transformation:

Given an expected average peak VO₂ of 15±3.5 mL•kg⁻¹•min⁻¹ for HF patients, 40 subjects randomized to anakinra and 20 to placebo (2:1 randomization) provided >99%, >99%, 99% and 86% power to detect a difference in peak VO₂ of 3.5, 3.0, 2.5 and 2.0 mL•kg⁻¹•min⁻¹.

Considering that the main variable of interest (peak VO2) is expected not to improve over time with placebo in patients with systolic HF^23–25, we chose the last observation carried forward (LOCF) method to impute the missing data at 4, 12 and 24 weeks. Exclusion of patient with missing data would have inevitably lead to a survivorship bias ²⁶. We also performed the Little’s Missing Completely At Random (MCAR) test to determine whether the appropriateness of imputing missing values (Supplemental Data), and performed a Multivariate Imputation by Chained Equation (MICE) ²⁷ to validate the finding of peak VO2 using a different method for data inference (Supplemental Data).

Screening and randomization

Screening started on 23 January 2014 and closed on 23 March 2016: 180 patients were assessed and 60 were randomized 1:1:1 to the 3 different groups. Eight patients (13%) were lost to follow up prior to the visit 2 at 2 weeks. Of the 52 patients who were enrolled and had at least 1 follow up visit, 18 (35%) were randomized to Placebo for 12 weeks, 16 (30%) to Anakinra for 2 weeks, and 18 (35%) to Anakinra for 12 weeks (Figure 1). Fifty patients were uninterested in being screened for CRP or participating in the study but agreed to be followed clinically in an observational arm of the study, without any investigational treatment. Seventy patients were excluded for other reasons listed in Figure 1.

Characteristics of the patients

Table 1 shows a summary of the clinical and demographic characteristics of the patients. Guideline-directed HF therapies at baseline and at the end of the study are shown in Table 1 and Supplemental Table 4. With the exception of a lower use of hydralazine/isosorbide in the placebo group at baseline, there were no statistically significant differences between groups. There was a trend toward higher NTproBNP plasma levels at baseline in the placebo group, although it did not reach statistical significance (P=0.07).

Effects on peak aerobic capacity (VO₂) and ventilatory efficiency (VE/VCO₂ slope)

RER ranged from 1.0 to 1.39 with median values between 1.08 and 1.18, without any significant difference between the 3 groups. When compared with placebo (N=18), treatment with anakinra (N=36, combining 2-week and 12-week treatment) failed to improve peak VO₂ or the VE/VCO₂ slope at 2 weeks. Figure 2 shows the changes in peak VO₂ or the VE/VCO₂ slope at 2, 4, 12 and 24 weeks in the 3 different groups. No significant changes in peak VO2 were seen in the group treated with placebo or with anakinra for 2 weeks. However, treatment with anakinra in the 12-week group was associated with a significant improvement in peak VO2 from 14.5 [10.5–16.6] at baseline to 15.7 [11.8–17.3] at 4 weeks (P=0.005) and to 16.1 [13.2–18.6] mL•kg⁻¹•min⁻¹ at 12 weeks (within group P=0.009; median increase from baseline +2.1 mL•kg⁻¹•min⁻¹, 95% confidence interval [CI] +0.6/+3.6). We validated the statistical significance of the increase in peak VO2 from baseline to 12 weeks in the 12-week anakinra group using a Multivariate Imputation by Chained Equation (MICE) instead of the LOCF method for the missing data inference (multivariate P value 0.007). When directly compared, the changes in the peak VO₂ in the anakinra 12-week treatment group were not significantly different between treatment groups.

Treatment with anakinra for 12 weeks was also associated with a nominal improvement in VE/VCO₂ slope from 34.9 [29.4–41.4] to 31.7 [27.3–34.2](within group P=0.037) at 12 weeks, however it did not reaching the predefined statistical significance set at P=0.025. Normalization of the VE/VCO₂ by peak VO₂ (VE/VCO₂-to-VO₂ ratio) provides a stronger prognostic indicator than the individual parameters ²⁸: anakinra for 12 weeks provided a significant improvement (reduction) at 4, 12 and 24 weeks (P=0.005, P=0.002 and P=0.018, respectively), whereas no significant changes were seen with placebo or Anakinra 2-week (Supplemental Figure 2).

Effects on CRP plasma levels

CRP plasma levels were used as a measure of systemic inflammation. Anakinra given for 2 weeks significantly reduced CRP levels, from 7.0 [2.9–14.6] to 1.5 [0.7–4.1] mg/l, in the anakinra treated patients [N=34] vs from 5.9 [2.2–13.5] to 3.0 [1.8–11.8] in the placebo (P=0.001 for between group comparison). Figure 3 shows the changes in plasma CRP at 2, 4, 12 and 24 weeks in the 3 different groups. CRP at 12 weeks was reduced by a median of 66% in the 12-week anakinra group (P=0.011), whereas by 12 weeks the CRP levels in anakinra 2-week treatment group were no longer significantly different from baseline. Changes in CRP levels at 12 weeks showed a correlation with changes in peak VO2 (R=−0.57, P=0.001)(Supplemental Figure 3).

Doppler echocardiography

There were no significant changes in LVEF in the placebo group (from 28% [23–40] at baseline to 28% [20–44] at 12 weeks). A trend toward improved LVEF was noted in the anakinra treated groups (from 33% [24–40] to 43% [33–50], P=0.002 in the 2-week treatment group; and from 32% [28–38] to 38% [30–45], P=0.18 in the 12-week treatment group)(Figure 4). The changes in LVEF at 12 weeks showed no significant correlation with changes in peak VO₂ (R=+0.22, P=0.12). A modest reduction in LVEDV over the 12 weeks was noted in all 3 groups without significant differences between groups (data not shown). The E/E′ ratio, an indirect measure of elevated LV filling pressures, was unchanged over time in the placebo-group, while it showed a trend toward improvement over time in the anakinra treated groups, with a significant reduction (from 18 [11–24] to 12 [9–17], P=0.006) at 12 weeks in the anakinra 12-week group (Figure 4). Treatment with anakinra for 12 weeks improved ventriculo-arterial coupling, Ea/Ees ratio, and stroke work efficiency, without significant changes in arterial blood pressure or heart rate (Supplemental Figure 4)(Supplemental Table 5).

Quality of life measures

Patients treated with anakinra for 12 weeks reported a significant improvement in the DASI over time, reflecting an improved perceived functional capacity, while those treated with placebo or with anakinra for 2 weeks did not (Figure 5). The improvement in DASI at 12 weeks showed a modest correlation with the improvement in peak VO₂ (R=+0.30, P=0.034), reduction in CRP levels (R=−0.29, P=0.039), as well as improved Ea/Ees (R=+0.33, P=0.023) and stroke work efficiency (R=+0.30, P=0.043). HF symptoms burden measured with as MLWHF scores significantly improved in all 3 treatment groups, although the improvement was numerically greatest in the anakinra groups, and the improvement in MLWHF at 12 weeks also showed a correlation with peak VO₂ (R=−0.33 and P=0.018).

Effects on cardiac events

There was 1 death (6%) due to cardiac cause (worsening HF) in the placebo group and 1 death (6%) due to a cardiac cause (acute myocardial infarction) in the anakinra 2-week group (occurring after treatment with anakinra had been completed) and no deaths (0%) in the anakinra 12-week group. For comparison, the 6-month mortality in the observational (non-interventional) arm was 8% (4 of 50 patients)(Supplemental Figure 5).

The incidence of the composite endpoint of death due to any cause and re-admission for HF was 5 patients (28%) in the placebo group, 5 patients (31%) in the anakinra 2-week treatment group, and 1 patient (6%) in the anakinra 12-week group (Log-Rank test P=0.10)(Figure 6). There were a total of 11 HF admissions in the placebo group (1 patient had 4 admissions, 1 patient had 3 admissions, 1 patient had 2 admission, and 2 patients had 1 admission each), 8 HF admissions in the anakinra 2-week treatment group (1 patient had 4 admissions and 4 patients had 1 admission each), and 3 HF admissions in 1 patient in the anakinra 12-week group. The incidence of the combined endpoint of death and HF hospitalization in the observational arm was 18 (36%)(Supplemental Results)(Supplemental Figure 5). There was 1 admission for acute myocardial infarction (fatal) in the anakinra 2-week group (6%) and none in the other 2 groups. There were no episodes of unstable angina or urgent revascularization. Two patients in the placebo group (11%) and 1 (6%) in the anakinra 2-week group had an admission prompted by a ventricular or supraventricular arrhythmia (Supplemental Table 6).

Non-cardiac adverse events

Treatment with anakinra was well tolerated, with no serious unanticipated adverse events. Two patients in the placebo group (11%), 1 patient in the anakinra 2-week group (6%) and 1 patient in the anakinra 12-week group (6%) experienced a serious infection requiring prescription of an antimicrobial drug, considered to be unrelated. One patient in each group (6%) experienced an ischemic or hemorrhagic stroke, considered to be unrelated. One patient in the placebo (6%) and three patients in the anakinra 2-week group (19%) experienced acute kidney injury (doubling of serum creatinine): all 4 of these patients had complete recovery of renal function by week 24. A complete list of clinical events is available in Supplemental Table 6.

Additional sensitivity analyses

The effects of anakinra treatment on NTproBNP levels and stratifications according to NTproBNP levels are shown in Supplemental Data and Supplemental Tables 7 and 8. We also stratified patients according to LVEF≤35% (Supplemental Table 9), and use of hydralazine/isosorbide at baseline (Supplemental Table 10) to account for potential imbalances in the baseline characteristics. None of the these parameters showed a significant interaction with anakinra effect.