IL-1 Blockade in Patients With Heart Failure With Preserved Ejection Fraction

Abstract

BACKGROUND:

Enhanced inflammation may lead to exercise intolerance in heart failure with preserved ejection fraction. The aim of the current study was to determine whether IL (interleukin)-l blockade with anakinra improved cardiorespiratory fitness in heart failure with preserved ejection fraction.

METHODS AND RESULTS:

Thirty-one patients with heart failure with preserved ejection fraction and CRP (C-reactive protein) >2 mg/L were randomized to anakinra (100 mg subcutaneously daily, N=21) or placebo (N=10) for 12 weeks. We measured peak oxygen consumption (Vo₂), ventilatory efficiency (V_E/Vco₂ slope), and high-sensitivity CRP and NT-proBNP (N-terminal pro-B-type natriuretic peptide) at 4, 12, and 24 weeks. Twenty-eight patients completed ≥2 visits, 18 women (64%), 27 (96%) obese. There were no differences in peak Vo₂ or V_E/Vco₂ slope between groups at baseline. Peak Vo₂ was not changed after 12 weeks of anakinra (from 13.6 [11.8–18.0] to 14.2 [11.2–18.5] mL-kg⁻¹-min⁻¹, P=0.89), or placebo (14.9 [11.7–17.2] to 15.0 [13.8–16.9] mL-kg⁻¹-min⁻¹, P=0.40), without significant between-group differences in changes at 12 weeks (−0.4 [95% CI, −2.2 to +1.4], P=0.64). V_E/Vco₂ slope was also unchanged with anakinra (from 28.3 [27.2–33.0] to 30.5 [26.3–32.8], P=0.97) or placebo (from 31.6 [27.3–36.9] to 31.2 [27.8–33.4], P=0.78), without significant between-group differences in changes at 12 weeks (+1.2 [95% CI, −1.8 to +4.3], P=0.97). Within the anakinra-treated patients, high-sensitivity CRP and NT-proBNP levels were lower at 4 weeks compared with baseline (P=0.026 and P=0.022 versus placebo [between-group analysis], respectively).

CONCLUSIONS:

Treatment with anakinra for 12 weeks failed to improve peak Vo₂ and V_E/Vco₂ slope in a group of obese heart failure with preserved ejection fraction patients. The favorable trends in high-sensitivity CRP and NT-proBNP with anakinra deserve exploration in future studies.

CLINICAL TRIAL REGISTRATION:

URL: https://www.clinicaltrials.gov. Unique identifier: NCT02173548.

Heart failure (HF) is a clinical syndrome of dyspnea and diminished exercise tolerance secondary to impaired cardiac function. In community-based studies, up to half of all patients with HF have preserved ejection fraction (HFpEF), in which cases impaired diastolic function and elevated filling pressures are the hallmarks of the disease.¹ CRP (C-reactive protein) identifies HFpEF patients at increased risk.² CRP is a surrogate of IL (interleukin)-1 activity, the prototypical inflammatory cytokine known to function as a soluble cardiodepressant factor.^3,4 IL-1 impairs diastolic relaxation and systolic contractility when administered to healthy animals (in vivo) and cardiomyocytes (in vitro).^5,6 In a crossover trial of 12 patients with HFpEF and elevated CRP, IL-1 blockade with anakinra (a recombinant IL-1 receptor antagonist) reduced CRP and improved exercise capacity.⁷ We herein present the results of a follow-up phase II clinical trial of IL-1 blockade with anakinra for 12 weeks in patients with stable HFpEF and elevated CRP, measuring the effects on peak oxygen consumption (peak Vo₂) as a primary end point (DHART2 [Diastolic Heart Failure Anakinra Response Trial 2]).

METHODS

Study Design

The study design of DHART2 has previously been presented.⁸ Approval was obtained by the Virginia Commonwealth University Institutional Review Board. Participants provided written consent. After screening, patients underwent a cardiopulmonary exercise test (CPX), transthoracic Doppler echocardiography, and testing for biomarkers at baseline, 4, 12, and 24 weeks. Patients were randomized to anakinra 100 mg daily or matching placebo in a 2:1 ratio for 12 weeks in a double-blinded fashion. This anakinra dosing follows the approved regimen and as used in the pilot DHART study⁸ and clinical trials of HF with reduced EF^5,9,10

Screening and Enrollment

A detailed list of inclusion and exclusion criteria has been previously presented⁹ and is included in Table I in the Data Supplement. We used the diagnostic criteria for HFpEF according to the European Society of Cardiology 2007 statement based on intracardiac filling pressures, BNP (B-type natriuretic peptide) levels, and Doppler echocardiography.¹¹ Eligible patients were also required to have CRP levels >2 mg/L (high-sensitivity assay).

Randomization and Allocation Concealment

A detailed description has been previously presented.⁸ A brief description is now included in the Data Supplement.

Data and Safety Monitoring Board

Table II in the Data Supplement shows the composition of the data and safety monitoring board.¹²

Cardiopulmonary Exercise Testing

A supervised maximal aerobic exercise test was administered using a conservative ramping treadmill protocol, as previously described.^7,8,13 Briefly, the peak respiratory exchange ratio (RER) was used to determine subject effort; a peak value ≥1.00 was considered the minimal acceptable effort threshold.^14,15 Patients who interrupted the CPX for reasons other than dyspnea and or fatigue (ie, myocardial ischemia, arrhythmias, hypertensive response, dizziness, and claudication) during the baseline CPX were excluded (screen failures). Patients who completed the baseline test but failed to attend any of the follow-up CPX visits were excluded from the paired efficacy analysis. However, all patients who received ≥1 doses of investigational treatment were included in the safety analysis. Additional details on the CPX analysis are included in the Data Supplement.

Doppler Echocardiography

All subjects underwent a transthoracic echocardiogram before initiation of treatment and at each subsequent visit. At baseline and week 12 visits, patients also underwent a limited stress Doppler echocardiography study immediately after peak exercise to measure interval changes in diastolic reserve (measured as changes in e’ velocity and E/e’ ratio) with exercise. Additional details on the Doppler echocardiography analysis are included in the Data Supplement.

Biomarkers

Blood samples were obtained at each visit to measure complete blood count with differential, comprehensive metabolic profile, and plasma levels of biomarkers such as high-sensitivity CRP (chosen as a surrogate for IL-1 activity) and NT-proBNP (N-terminal pro-BNP, surrogate for myocardial strain, as reductions in NT-proBNP levels with treatment, are generally considered a favorable prognostic indicator).⁸

Functional Capacity and Quality of Life Questionnaires

All patients completed 2 different questionnaires at each visit: the Duke Activity Status Index (with higher scores reflecting increased functional capacity)¹⁶ and the Minnesota Living With Heart Failure Questionnaire (MLWHFQ, with higher scores reflecting increased burden of HF symptoms—we calculated the Physical Domain of the MLWHFQ which better correlates with peak Vo₂).^17,18

Body Composition Assessment

Patients underwent bioelectrical impedance analysis to determine body composition. Details on the methods are provided in the Data Supplement.

Clinical Events

A dedicated blinded committee adjudicated all clinical events (Table III in the Data Supplement). The adjudicated events included death (cardiac and noncardiac), hospitalizations (cardiac [including HF-related and non-HF-related] or noncardiac), serious adverse events, and nonserious adverse events through study conclusion or 24 weeks, whichever occurred first.

Study End Points

The coprimary end points were the placebo-corrected changes in peak Vo₂ and V_E/Vco₂ slope after 12 weeks of treatment. Additional secondary end points included further data from CPX (interval changes in peak Vo₂ or the V_E/Vco₂ slope at 4 and 24 weeks), structural and functional parameters at Doppler echocardiography at each follow-up visit (including stress echocardiography at baseline and week 12), quality of life assessment questionnaire scores, interval changes in biomarkers (including CRP and NT-proBNP), and composite clinical outcomes of death and rehospitalization for HF (or hospitalization for any cause) up to 24 weeks.

Statistical Analysis

Descriptive summaries of continuous measurements are reported as median and interquartile ranges. Descriptive summaries of categorical measurements consist of frequencies, proportions, and 95% CIs, when applicable. All analyses were conducted after database locking on June 9, 2017, and based on the intention-to-treat principle (ie, analyzing groups as randomized and including all patients with outcome data available). The Statistical Package for Social Studies software 23.0 (IBM, New York, NY) was used. The difference in interval changes in peak Vo₂ or the V_E/Vco₂ slope at 12 weeks (primary end point) between anakinra versus placebo groups were compared using ANOVA for repeated measures assessing for the time×group interaction to compare the changes between groups. The difference between the means was calculated and reported with the 95% CI. Unadjusted P values are reported throughout, with statistical significance for the coprimary end points set at the 2-tailed 0.025 level. Secondary analyses included interval changes in biomarkers, exercise time, Doppler echocardiography parameters, and quality of life questionnaires at 4 and 12 weeks (ANOVA for repeated measures, timexgroup interaction).

The nonparametric Wilcoxon signed-rank test for repeated measures was used to determine the statistical significance of interval changes compared with baseline in each group (within-group analysis). Correlation between continuous variables was analyzed using the nonparametric Spearman rank test.

Considering that the main variable of interest (peak Vo₂) was not expected to significantly improve over time in patients with HFpEF without major changes in medications or other interventions such as weight loss or exercise training,¹⁹ we chose the last observation carried forward method to impute the missing data at 12 and 24 weeks in both the placebo and in the anakinra groups. Exclusion of patient with missing data would have inevitably led to survivorship bias.²⁰ We also conducted sensitivity analysis omitting the imputed data to determine the effects of the imputation. The data will be made available upon request to other researchers for the purpose of reproducing the results.

Sample Size and Power Calculations

Given an expected average peak Vo₂ of 15±3.5 mL-kg⁻¹-min⁻¹ for HFpEF patients, 40 subjects randomized to anakinra, and 20 to placebo (2:1 randomization) was initially expected to provide >99%, >99%, and 86% power to detect a difference in a peak Vo₂ of 3.5, 3.0, and 2.0 mL-kg^_1 min^_1, respectively.

In October 2016, because of a slower than expected enrollment rate—and in collaboration with the data and safety monitoring board—the investigators reviewed the baseline data and original power calculations: the preliminary data showed a smaller SD than originally expected from previously published studies (1.9 mL-kg⁻¹-min⁻¹ instead of 3.5 mL-kg⁻¹-min⁻¹), and a new calculation was made by which a sample size of 30 patients total (maintaining the same 2:1 ratio) was expected to retain >99%, 90%, and 78% power to detect a minimum difference in peak Vo₂ of 3.5, 3.0, and 2.0 mL-kg^_1-min^_1, respectively. In consideration of the single-center nature of the study, and the limited budget, the sample size was reduced to 30 patients.

RESULTS

Screening and Randomization

Screening started in August 2014 and closed in January 2017. Seventy-three patients were assessed, and 31 were randomized 2:1 to receive anakinra (N=21) or placebo (N=10), 1 patient was randomized to anakinra but never received treatment as the patient decided to have dental surgery before the study and never re-enrolled, and 2 patients (8%) were excluded before the 4-week follow-up because of unwillingness or inability to schedule any follow-up CPX. Of the 28 patients enrolled and with at least 1 follow-up visit, 20 (71%) were randomized to anakinra, and 8 (29%) to placebo and were considered on an intention-to-treat basis (Figure 1).

Figure 1. Screening and enrollment of patients.

CPX indicates cardiopulmonary exercise test; and hsCRP, high-sensitivity C-reactive protein.

Baseline Characteristics

The patient demographics and clinical characteristics are summarized in the Table. Of note, the patients were for the most part between 50 and 60 years of age, mostly women, and about evenly split between whites and blacks. Many patients had diabetes mellitus, all had hypertension, and all but 1 (96%) were obese and 19 (61%) with stage III obesity (body mass index ≥40 kg/m²). There were no significant differences between the anakinra and placebo groups.

Cardiorespiratory Fitness

Median RER was 1.1, and target RER>1.0 was achieved in all but 2 subjects (93%): 1 patient with an RER of 0.99 randomized to placebo and with RER of 0.98 randomized to anakinra were inadvertedly enrolled. Baseline peak was Vo₂ 13.8 (11.8–17.7) mL-kg⁻¹-min⁻¹ corresponding to 50% (42–65) of predicted, and the baseline V_E/Vco₂ slope was 30 (27–33).

Effects on Peak Vo₂ and V_E/Vco₂ Slope: Primary End Points

There were no differences in peak Vo₂ or the V_E/Vco₂ slope between groups at baseline. Peak Vo₂ was not changed after 12 weeks of anakinra (from 13.6 [11.818.0] to 14.2 [11.2–18.5] mL-kg⁻¹-min⁻¹, P=0.89), or of placebo (from 14.9 [11.7–17.2] to 15.0 [13.8–16.9] mL-kg⁻¹-min⁻¹, P=0.40), without any significant difference between mean values of changes between groups (−0.4 [95% CI, −2.2 to +1.4]; P=0.64; Figure 2). V_E/Vco₂ slope was also unchanged at 12 weeks with anakinra (from 28.3 [27.2–33.0] to 30.5 [26.3–32.8], P=0.97) or placebo (from 31.6 [27.3–36.9] to 31.2 [27.8–33.4], P=0.78), without any significant difference between mean values of changes between groups (+1.2 [95% CI, −1.8 to +4.3]; P=0.97; Figure 2).

Figure 2. Effects of treatment on peak oxygen consumption (Vo₂) and minute ventilation-carbon dioxide production slope (V_E/Vco₂ slope).

Treatment with anakinra was not associated with any significant changes, or differences in change vs placebo, in peak Vo₂ or the V_E/Vco₂ slope at 4 wk nor at 12 wk (coprimary end points) or after 12-wk washout (all P values >0.05 at timexgroup interaction analysis). There were no significant changes in peak Vo₂ or the V_E/Vc0₂ slope at 4 and 12 wk compared with baseline in either group (all P>0.05). Column height reflects median and bar reflects interquartile range.

There were no statistically significant differences in between-group changes in exercise time comparing anakinra and placebo (P=0.96). Within the anakinra-treated patients, treadmill exercise time was longer at 4 weeks (+48 seconds, P=0.004 for within-group) and 12 weeks (+27 seconds, P=0.009 for within-group) versus baseline. After 12-week of cessation of anakinra therapy, exercise time was not different than the baseline levels (+2 seconds, P=0.95; Figure 3). There were no statistically significant differences in exercise times at 4, 12, or 24 weeks compared with baseline in the placebo group (all P>0.05; Figure 3).

Figure 3. Effects of treatment on exercise time.

There were no statistically significant differences between-group changes in exercise time comparing anakinra and placebo (P=0.96). Within the anakinra-treated patients, treadmill exercise time was longer at 4 wk (+48 s, P=0.004 for within-group) and 12 wk (+27 s, P=0.009 for within-group) vs baseline. After 12 wk of cessation of anakinra therapy, exercise time was not different than the baseline levels (+2 s, P=0.95). Column height reflects median and bar reflects interquartile range.

Effects of Anakinra on CRP Levels

There were no significant differences in baseline CRP levels between the 2 groups (Figure 4). There was a significantly greater reduction in CRP levels with 4 and 12 weeks of anakinra than placebo (P=0.026 and P=0.033, respectively for 4 and 12 weeks [between-group analysis—timexgroup interaction]). Treatment with placebo for 12 weeks had indeed no significant effect on CRP levels at any time point (all P>0.05 versus baseline [within-group analysis]), whereas anakinra significantly reduced CRP levels from 6.1 [3.8–18.4] at baseline to 2.1 [1.0–4.0] mg/L at 4 weeks (−65% [95% CI, +6% to −137% of mean value], P=0.002 versus baseline [within-group analysis]) and to 2.9 [1.2–7.8] mg/L at 12 weeks (−52% [95% CI, +20% to −125% of mean value], P=0.004 versus baseline [within-group analysis]), with levels returning closer to baseline after 12-week washout (4.5 [2.6–12.9] mg/L, −26% [95% CI, +52% to −105% of mean value], P=0.048 versus baseline [within-group analysis]; Figure 4).

Figure 4. Changes in CRP (C-reactive protein) during the 12 wk of anakinra treatment.

CRP levels were lower after 4 and 12 wk of anakinra (P=0.026 and P=0.033, respectively, vs placebo [between-group analysis—timexgroup interaction]). No significant changes in CRP levels were seen at any time point in the placebo group (all P>0.05 vs baseline [within-group analysis]), whereas CRP levels were lower at 4 wk (−65%, P=0.009 vs baseline [within-group analysis]) and at 12 wk (−52%, P=0.009 vs baseline [within-group analysis]) in the anakinra group, with levels returning closer to baseline after 12 wk of stopping treatment (P=0.048 vs baseline [within-group analysis]). Column height reflects median and bar reflects interquartile range.

Effects on NT-proBNP Levels

There were no significant differences in baseline NT-proBNP levels between the 2 groups (Figure 5). There was a significantly greater reduction in NT-proBNP levels with 4 weeks of anakinra than placebo (P=0.022 [between-group analysis—timexgroup interaction]). There were no changes in NT-proBNP levels in the placebo group at any time point (all P>0.16 versus baseline [within-group analysis]), whereas NT-proBNP levels were lower at 12 weeks in the anakinra group (from a median of 166 pg/mL to a median of 81 pg/mL, −42% [95% CI from +73% to −163% of mean value] versus baseline, P=0.029 for within-group; however, P>0.05 for timexgroup interaction; Figure 5). Reductions in CRP in absolute values at 1 month significantly correlated with NT-proBNP changes (R=+0.41, P=0.032), with a greater reduction in CRP resulting in a greater reduction in NT-proBNP levels (Figure I in the Data Supplement). A sensitivity analysis restricted to those individuals with NT-proBNP >100 pg/mL (N=15) is presented in Figure II in the Data Supplement.

Figure 5. Changes in NT-proBNP (N-terminal pro-B-type natriuretic peptide).

There was a significantly greater reduction in NT-proBNP levels with 4 wk of anakinra than placebo (P=0.022 [between-group analysis—timexgroup interaction]). There were no changes in NT-proBNP levels in the placebo group at any time point (all P>0.16 vs baseline [within-group analysis]), whereas NT-proBNP levels were lower at 12 wk in the anakinra group (−42% vs baseline, P=0.029 for within-group; however, P>0.05 for timexgroup interaction). Column height reflects median and bar reflects interquartile range.

Doppler Echocardiography

Neither placebo nor anakinra had any significant effects on left ventricular ejection fraction, left atrial volume index, e’, or E/e’ measured at rest or at peak exercise (all P values >0.05; Figure III in the Data Supplement).

Functional Capacity and Quality of Life Questionnaires

There were no significant differences in baseline Duke Activity Status Index score or MLWHFQ score and physical domain scores between the groups, and there were no significant between-groups differences on the effects of treatments on any of the outcomes of interest (Figure 6). Patients randomized to treatment with anakinra reported a significant improvement in perceived functional capacity, as assessed by the Duke Activity Status Index questionnaire (+11 [95% from +2 to +20], P=0.044 for within-group), and of the MLWHFQ questionnaire score as a whole as well as the physical domain (−21% [95% CI from +17% to −136% of the mean value] and −66% [95% CI from −16% to −76% of the mean value], P=0.031 and P=0.001 at within-group analysis, respectively, at 12 weeks), whereas no significant changes were seen in the placebo group (all P values >0.05 for within-group in placebo), and none of the differences in anakinra versus placebo changes reached statistical significance for any of the changes in questionnaires over time (all P>0.05 for timexgroup interaction; Figure 6).

Figure 6. Changes in functional capacity and quality of life questionnaires.

There were no significant differences in baseline Duke Activity Status Index (DASI) score of Minnesota Living With Heart Failure Questionnaire (MLWHFQ) score and physical domain score between the groups at baseline, and there were no significant between-groups differences on the effects of treatments on any of the outcomes of interest. Patients randomized to treatment with anakinra reported significantly higher perceived functional capacity, as assessed by the DASI question-naire at follow-up (within-group analysis), and quality of life, as shown by the physical domain scores of the MLWHFQ and comprehensive MLWHFQ score (within-group analysis), whereas no placebo-treated patients did not (within-group analysis); however, none of the differences in anakinra vs placebo changes reached statistical significance (timexgroup interaction). Column height reflects median and bar reflects interquartile range.

Clinical Events

There was 1 hospitalization for HF in the placebo group (12%) and 2 additional patients who had an outpatient intensification of loop diuretics leading to a combined rate of worsening HF of 38%. There was 1 patient with HF hospitalization in the anakinra group (5%), and there were 3 additional patients requiring intensification of loop diuretics for a combined worsening HF rate of 20% in the anakinra group (P>0.05 between groups).

Adverse Events

Three patients in the placebo group (38%) were diagnosed with a nonserious infection during the study period: 2 patients had an upper respiratory infection during the treatment period, and 1 patient had a urinary tract infection in the washout period. In the anakinra group, 3 patients (15%) were diagnosed with a nonserious infection during the study period: 1 patient had a tooth abscess with culture negative sepsis during the treatment period, 1 patient had a sinus infection in the washout period, and 1 patient had an upper respiratory infection in the washout period. None of the infections were considered to be opportunistic in nature or to be secondary to the investigational drug by the investigators.

Sensitivity Analysis for Missing Data Imputation

Sensitivity analyses for missing data imputation at 4 and 12 weeks, as well as additional analysis including subjects excluded before visit 2 (N=2) or excluding subjects inadvertedly, included with RER<1.0 (N=2) are presented in the Data Supplement.

Exploratory Analysis on the Role of Adiposity

An exploratory analysis on the role of adiposity and the IL-1-mediated effects on peak Vo₂ are included in the Data Supplement (Figure IV in the Data Supplement).

DISCUSSION

Improving functional capacity, quality of life, and clinical outcomes in patients with HFpEF remains an unmet clinical need. In the current study, treatment with anakinra, a recombinant IL-1 receptor antagonist, for 12 weeks, while inhibiting the systemic inflammatory response, as shown by reduced CRP levels, failed to improve aerobic exercise capacity (peak Vo₂) or ventilator efficiency (V_E/Vco₂ slope) in patients with HFpEF (primary end points).

Enrollment in the study was, however, halted early, after only 31 of the 60 patients planned, thus leading to a significant reduction in the power. There were only 8 patients randomized to placebo, which made it even more limited in detecting between-group changes (timexgroup interaction) as prespecified, and thus with-in-group analysis for the anakinra group (N=20) were used as a measure of changes over time, with the limitation of not being corrected for changes within placebo.

The results of this study seem to be in contrast with a small pilot crossover trial in which 12 patients with HFpEF treated with anakinra had shown an absolute improvement in peak Vo₂ of 1.1 mL-kg^_1-min^_1.⁷ Several differences between these studies may have contributed to these findings. First, anakinra was associated with a reduction in CRP levels in both trials but possibly to a greater extent in the initial study.⁷ Second, there was a notably higher degree of obesity in the current trial (median body mass index=42 kg/m²) as compared with original pilot study (median body mass index=39 kg/m²), with twice as many patients having grade III or severe obesity (body mass index ≥40 kg/m², 60% versus 30%). This may have an implication for aerobic capacity, as adiposity is known to be one of the most important determinants of cardiorespiratory fitness in obese individuals, independent of cardiac function.^21–23 The association between fat mass index and reduced changes in peak Vo₂ with anakinra further support the concept that excess adiposity represented a barrier to improvement in this population, and therefore requires targeted weight loss interventions if cardiorespiratory fitness is to improve.^24,25 Despite the lack of improvement in aerobic capacity and ventilatory efficiency, anakinra was associated with favorable trends in the reduction in NT-proBNP levels, changes in exercise time and improvements in quality of life measures, a triad that has been linked with improved outcomes in HF trials.²⁶ NT-proBNP is a biomarker reflecting myocardial strain with an established clinical value in the diagnosis and management of patients with HF.²⁷ Other phase II clinical trials in HFpEF (eg, sacubitril-valsartan²⁸) have used NT-proBNP level reductions as primary end points. Despite the lower values of NT-proBNP in patients with HFpEF, as compared with HF with reduced EF, natriuretic peptides maintain their prognostic value.^29–31 NT-proBNP is also placebo-insensitive and free of assessment bias, making it a valuable end point in phase II clinical trials. Consistent with these findings, we have recently reported a significant reduction in NT-proBNP levels in patients with HF with reduced EF treated with anakinra for 12 weeks.¹⁰

Exercise time on a treadmill is a surrogate measure of cardiorespiratory fitness and an independent predictor of cardiovascular mortality, even after correction for peak Vo₂ and up to 35 other clinical variables.³² Although there were no statistically significant effects of anakinra on exercise time when compared with placebo, the exercise time was increased in the anakinra group while on treatment at 4 and 12 weeks, with a return to baseline 12 weeks after cessation was witnessed. It has been reported that becoming accustomed to the equipment can lead to a substantial increase in exercise time; however, the lack of significant improvement in placebo group, and the return to baseline after cessation of anakinra therapy, parallel to the changes in CRP, suggests a biological effect.

We had hypothesized that anakinra would increase exercise capacity while increasing peak Vo₂ and improving ventilatory efficiency, possibly through an effect on myocardial relaxation. Although the findings of increased exercise time without significant changes in peak Vo₂, the V_E/Vco₂ slope, left ventricular ejection fraction, or E/e’ undoubtedly challenge this hypothesis, it is worth discussing additional factors that may have influenced each of these variables. As mentioned above, peak Vo₂ in obese patients is inversely correlated with measures of adiposity and fat mass index; severe impairments in peak Vo₂ may reflect not only an impairment in cardiac output related to HF but also an impaired utilization of oxygen in the periphery at the skeletal muscle level.^33,34 If severe enough, skeletal muscle dysfunction may mask any measurable improvements in cardiac function as it would fail to translate in improved peak Vo₂. In selected patients with HFpEF, such as the elderly and the severely obese, reductions in peak Vo₂ may, therefore, be a reflection of skeletal myopathy more than of HF per se. As such, for each patient with HFpEF, a careful characterization of all components of exercise limitation should be addressed to guide treatment. Interventions aimed at improving skeletal muscle function (exercise training)^22,34–36 and reduce adiposity (low-calorie diet)²² in patients with HFpEF and severe obesity have been shown to lead to significant improvements in peak Vo₂, without measurable changes in cardiac or vascular function, or in natriuretic peptide levels. Despite significant improvements in peak Vo₂, exercise training seemed to provide rather small improvements in quality of life measures,^22,37 and in general changes in peak Vo₂ do not always correlate with changes in quality of life.³⁸ This dissociation between improvements in measures of aerobic capacity or cardiovascular parameters with those of quality of life, may suggest that this treatment approach is affecting one aspect of the clinical syndrome (ie, peripheral oxygen uptake by the skeletal muscle) and not other HF-specific abnormalities. Notably, the change in fat mass was a strong predictor of change in peak Vo₂ in a dietary-exercise intervention trial.²² By contrast, treatment with anakinra in this study seems to have favorable effects on NT-proBNP, a biomarker of myocardial strain, yet not affect peak Vo₂, thus suggesting that these effects are not mediated through action on the skeletal muscle. The lack of a direct assessment of skeletal muscle function is, however, a clear limitation of the current study.

The V_E/Vco₂ slope is a measure of ventilatory efficiency that is also significantly affected by obesity, with lower values of V_E at rest and with exercise, resulting in lower V_E/Vco₂ slopes,³⁹ thus limiting the sensitivity of these analyses.

Assessment of left ventricular ejection fraction and E/e’ at rest offer a limited assessment of the cardiac function in patients with HFpEF, and assessment with exercise is preferred.⁴⁰ However, obtaining images of sufficient quality in obese patients at peak exercise is challenging, and data in this study were available only in 60% of patients at peak exercise, thus further limiting the power of the assessment.

HFpEF is a syndrome of breathlessness and fatigue, associated with impaired quality of life. Although treatment with anakinra did not improve functional capacity or quality of life measures above what seen in placebo, the changes in the Duke Activity Status Index, addressing how functional an individual perceives to be, and the physical domain of the MLWHFQ which measures the burden of HF symptoms, at 4, 12, and 24 weeks seemed to be more pronounced in the anakinra group.

It is important, however, to note that the primary hypothesis was not confirmed and to highlight additional limitations of the current study. The first limitation is the limited power of study.⁸ This is likely because of the use of strict inclusion criteria of impaired diastolic function, and the result was a very small number of placebo-treated subjects displaying greater variability within that group, thus possibly interfering with timexgroup interaction analyses. Relying on within-group analysis has obvious disadvantages of not allowing comparisons between groups but provides greater power for detecting changes. The second limitation is the choice of the peak Vo₂ and the V_E/Vco₂ slope as primary end points, as these may be severely affected by noncardiac factors in HFpEF, that is, obesity, and not adequately positioned to capture the potential effects of anakinra. The inclusion of a majority of individuals with severe obesity may also limit the external validity of the study. Third, this study explored a single dose and a single duration of anakinra treatment, thus not allowing for exploration of any potential dose-response relationship. Data on response to IL-1 blockade would suggest that the greater the response greater is the benefit⁴; hence, a higher dose of anakinra or a more potent IL-1 blocker may be tested.

Any individual surrogate end points and efficacy markers by themselves in phase II studies might not be sufficient to inform chances of success in phase III trials; rather, a combination of markers provides better cumulative information to guide downstream research endeavors.^26,41 As such, the results seen with anakinra on NT-proBNP, exercise time, and quality of life are encouraging and warrant further exploration. IL-1 has been known to negatively impact myocardial systolic and diastolic function in experimental models, and therefore altogether, the data presented herein and the results of other pilot studies with anakinra in HF and ST-segment elevation acute myocardial infarction^{4,5,7,9,10,42–46} and the success of the large phase III trial with canakinumab in patients with prior myocardial infarction,⁴⁷ all point toward a beneficial effect of IL-1 blockade in heart disease.⁴ In conclusion, whether IL-1 blockade with anakinra will be clinically valuable in the treatment of patients with HFpEF remains undetermined and requires further clinical studies.

Table.

Demographic and Clinical Characteristics

	Placebo (N=18)	Anakinra (N=20)	P Value
Age, y	58 (51–64)	54 (45–61)	0.44
Sex, male	3 (37%)	7 (35%)	0.90
Race, black	3 (37%)	11 (55%)	0.40
Body mass index, kg/m2	41.0 (34.6–47.8)	42.4 (39.3–47.8)	0.41
Coronary artery disease	1 (12.5%)	2 (10%)	0.85
Atrial fibrillation	1 (12.5%)	2 (10%)	0.85
Arterial hypertension	8 (100%)	20 (100%)	1.00
Diabetes mellitus	7 (87.5%)	16 (80%)	0.64
Chronic obstructive pulmonary disease	1 (12.5%)	4 (20%)	0.64
CRP, mg/L	7.6 (3.4–12.2)	6.1 (3.9–18.4)	0.90
NT-proBNP, ng/mL	98 (48–209)	166 (39–268)	0.50
NYHA class			0.61
NYHA class II	2 (25%)	7 (35%)
NYHA class III	6 (75%)	13 (65%)
DASI score	22 (9–29)	20 (16–37)	0.60
MLWHF score	68 (45–89)	50 (27–72)	0.22
MLWHF physical domain score	27 (24–32)	25 (16–33)	0.53
Doppler echocardiography parameters
left ventricular ejection fraction, %	57 (50–62)	60 (58–63)	0.22
E/E′ ratio	11.1 (10.2–13.4)	9.8 (8.6–14.3)	0.35
Cardiopulmonary exercise test parameters
Exercise time, min	8.9 (7.2–9.9)	8.0 (5.6–10.2)	0.67
Respiratory exchange ratio	1.08 (1.04–1.12)	1.11 (1.04–1.17)	0.28
Peak Vo2, mL·kg−1·min−1	15.0 (11.7–17.2)	13.6 (11.8–18.0)	0.90
VE/Vco2 slope	31.5 (27.3–36.8)	28.3 (27.2–33.0)	0.41
OUES	1.95 (1.61–2.35)	2.14 (1.75–2.61)	0.44
Heart failure therapy, N (%)
Angiotensin blockers	4 (50%)	11 (55%)	0.81
β-adrenergic receptor blockers	16 (75%)	17 (85%)	0.53
Aldosterone blockers	3 (37.5%)	12 (60%)	0.28
Hydralazine/ isosorbide	1 (12.5%)	7 (35%)	0.23
Loop diuretics	8 (100%)	18 (90%)	0.35
Furosemide equivalent dose, mg	70 (37–128)	100 (83–189)	0.44

Median value (interquartile range) are shown. CRP indicates C-reactive protein; DASI, Duke Activity Status Index; E, early mitral pulsed-wave Doppler flow velocity; E’, early mitral annulus tissue pulsed-wave Doppler velocity; MLWHF, Minnesota Living With Heart Failure Questionnaire; NT-proBNP, N-terminal pro-B-type natriuretic peptide; NYHA, New York Heart Association; OUES, oxygen uptake efficiency slope; Vo₂, volume of oxygen consumption; and V_E/Vco₂ slope, minute ventilation-carbon dioxide production slope.

WHAT IS NEW?

• Heart failure with preserved ejection fraction is a clinical syndrome characterized by dyspnea and fatigue secondary, at least in part, to impaired diastolic function. The DHART2 (Diastolic Heart Failure Anakinra Response Trial 2) study tested whether an anti-inflammatory treatment with anakinra, IL (interleukin)-1 receptor antagonist, would be well tolerated, would quench systemic inflammation, as measured by reduced CRP (C-reactive protein) levels, and would result in improved cardiorespiratory fitness.

• Patients treated with anakinra demonstrated a greater decrease in CRP levels after 4 and 12 weeks of treatment; this, however, did not result in improved cardiorespiratory fitness (primary end point of the study). Anakinra was associated with greater reductions in NT-proBNP (N-terminal pro-B-type natriuretic peptide), a biomarker of myocardial strain versus placebo.

WHAT ARE THE CLINICAL IMPLICATIONS?

• Inflammation has been implicated in the pathophysiology of heart failure with preserved ejection fraction; the DHART2 study showed favorable trends on CRP and NT-proBNP, thus supporting the role of inflammation in heart failure with preserved ejection fraction and the planning for future larger, and adequately powered, studies of IL-1 blockade in heart failure with preserved ejection fraction.

• The lack of effects of anakinra on cardiorespiratory fitness may be related to the lower than expected power of the study or to the very high prevalence of obesity which affects cardiorespiratory fitness independently of cardiac function.