Acute heart failure (AHF) is a complex syndrome with presentations ranging from hypotensive cardiogenic shock to hypertensive emergency with pulmonary edema. Most patients with AHF present with worsening of chronic HF signs and symptoms over days to weeks, and significant heterogeneity exists. It can, therefore, be challenging to characterize the overall population. The complexity of defining the AHF phenotype has been cited as a contributing cause for neutral results in most pharmacologic trials in patients with AHF. Dyspnea has been a routine inclusion criterion for AHF for over a decade, but the utility of current instruments for dyspnea assessment has been called into question. Furthermore, the threshold of clinical severity that prompts patient admission of an HF clinic visit may vary substantially across regions in global trials. Therefore, the inclusion of cardiac-specific biomarkers has been incorporated into AHF trials as 1 strategy to support inclusion of the target patient population and potentially enrich the population with patients at risk for clinical outcomes. In conclusion, we discuss strategies to support appropriate patient selection in AHF trials with an emphasis on using biomarker criteria that may improve the likelihood of success with future AHF clinical trials.
Heart failure (HF) is a progressive condition that includes acute and chronic phases with an overall heterogeneous patient population. In particular, acute heart failure (AHF) is a complex syndrome with presentations ranging from hypotensive cardiogenic shock to hypertensive emergency with pulmonary edema such that it has been challenging to characterize the population. As discussed in a recent consensus statement, many factors may have contributed to previous neutral trial results in AHF, including known and unknown pathophysiological pathways that likely differ among individual patients, challenges with the timing of patient enrollment, and enrolling an adequate number of acutely ill patients to detect a significant difference for the outcome of interest. Appropriate patient selection and defining the ideal end points are vital to the success of AHF clinical trials, and biomarkers could help to improve this process. This report focuses on patient selection with an emphasis on the utilization of natriuretic peptides in AHF. This review is based on discussions between scientists, clinical trialists, and regulatory representatives at the 10th Global Cardio Vascular Clinical Trialists Forum in Paris, France, from December 5 to December 7, 2013.
Defining AHF in Clinical Trials
AHF can be defined as gradual or rapidly worsening HF signs and symptoms necessitating immediate medical therapy/physician intervention. AHF has a substantial impact on the risk of repeat hospitalization. A recent Italian registry demonstrated that 1-year rehospitalization for AHF was 31%, which was higher than the rate of hospitalization in the stable chronic HF population (22%) and higher than the rate of rehospitalization after an acute coronary syndrome. The cumulative incidence of 30-day all-cause mortality or readmission for AHF in the Acute Decompensated Heart Failure Registry Emergency Module study was 27.4%. The Breathing Not Properly study helped to accelerate and validate the utilization of biomarkers as diagnostic tools as brain natriuretic peptide (BNP) levels in the emergency department were able to rapidly define the cause of dyspnea and diagnose patients with actual HF versus other causes of dyspnea. This has led to significant interest in biomarkers as inclusion criteria and surrogate end points in AHF.
Natriuretic Peptides
Background
Natriuretic peptides (plasma BNP or N-terminal pro-BNP [NT-proBNP]) are cardiac neurohormones dynamically synthesized and secreted from the ventricles in response to volume expansion, ventricular wall stress, and pressure overload. It has been well established in patients where the diagnosis is in question that rapid measurement of natriuretic peptides is useful in establishing or excluding the diagnosis of AHF in patients with acute dyspnea. Patients with a reduction in BNP levels for treatment of AHF had decreased 31 and 180 days mortality compared with those with little or no BNP reduction. These results suggest that response to treatment in BNP levels predicts both short- and long-term mortality risks.
Current limitations of natriuretic peptide assays
BNP and NT-proBNP have been used as an entry criterion in major clinical trials, but the shortcoming of this process is that all the studies differ in their respective end points, the number of patients enrolled, and mean BNP and NT-proBNP levels ( Table 1 ). Another fundamental issue is that a trend has been noted in multiple on-going studies that many centers are no longer performing “routine” measurements of natriuretic peptides, such that their availability as routine, non–protocol-specified measures is decreasing. This trend will not only delay enrollment in studies, since then measuring natriuretic peptides will have to wait until after informed consent is obtained, but also increases the cost. Future AHF trials should obtain BNP and NT-proBNP levels at the time of initial presentation or admission ( Table 2 ). The advantage of presentation is that it provides the best indicator of the initial pretreatment status. However, these types of assessments can only be obtained in clinical trials after informed consent is obtained if done solely for the purpose of the trial. There are several assays and methods available to measure BNP, and this leads to questions regarding assay purity. Although most methods are standardized to give a comparable value at 100 ng/L, agreement between methods more than and less than this point may vary substantially, and there are differences between manual and point-of-care assays versus central laboratory measurements of BNP. Using NT-proBNP as the biomarker of choice could minimize some of the variability seen with available BNP assays.
| Clinical Trial | Length of follow-up | Number of Patients (treated) | Study Drug | NT-proBNP entry criteria (median for study drug) | BNP entry criteria (median for study drug) | Death (time frame) | Death, Rehospitalization or worsening HF (time frame) | CV readmission or death |
|---|---|---|---|---|---|---|---|---|
| RELAX-AHF | 30 Days | 581 | Serelaxin | ≥1400 ng/L (5,125) | ≥350 ng/L (NA) | 2.1% (30 days) | 15.6% (30 days) | 7.5% (30 days) |
| ASTRONAUT | 6 months | 808 | Aliskiren | ≥1600 pg/mL (4278) ∗ | ≥400 pg/mL (474) ∗ | 9.5% (6 months) | 18.9% (6 months) | 24.9% (6 months) |
| PROTECT | 180 days | 1356 | Rolofylline | ≥2,000 pg/mL (3,000) | ≥500 pg/mL (1,290) | 17.9% (180 days) | (NA) | 31% (60 days) |
| ASCEND HF | 30 days | 3,496 | Nesiritide | ≥1,000 pg/mL (4,508) | ≥400 pg/mL (994) | 3.6% (30 days) | 6% (30 days) | 9.4% (30 days) |
| Clinical Trial | Estimated completion date | Study Design/Study Groups | Purpose of Study | BNP/NT-pro BNP Level For Inclusion |
|---|---|---|---|---|
| TRUE-AHF | 2015 | Randomized, Double-Blind, Placebo-Controlled Trial | Evaluate the efficacy and safety of ularitide infusion on the clinical status and outcome of patients with AHF | BNP >500 pg/mL or NT-pro BNP>2000 pg/mL |
| TUF | 2015 | Procedure: ultrafiltration (UF) for patients with AHF | Compare effectiveness of adding tolvaptan to usual care with UF as primary mode of therapy in AHF | BNP > 300 pg/ml |
| FUROPHARM-HF | 2015 | Drug: Oral vs. subcutaneous furosemide in patients with HF and volume overload | Investigate the pharmacodynamics and kinetics of subcutaneously administered furosemide | NT-proBNP > 300 ng/L |
| PreBNP | 2015 | Interventional/randomized single blind diagnostic study | Use BNP as a pre-hospital biomarker for risk stratification for patients who present with SOB prior to cardiac triage | NT-proBNP: <50 years: 450 pg/ml, 50-75 years: >900 pg/ml, >75 years: >1800 pg/ml |
| BLAST-HF | 2016 | Randomized, Double-Blind, Placebo-Controlled, Dose Ranging Study | Evaluate the safety and efficacy of TRV027 when administered in addition to standard of care in AHF | BNP > 400 pg/mL or NT-proBNP > 1600 pg/mL. |
| B-AHEF | 2016 | Prospective placebo-controlled double blind RCT | Assess effectiveness of Hydralazine and Isosorbide Dinitrate in African patients admitted with AHF and LV dysfunction | NT-proBNP >900 pg/ml/>1800 pg/ml for AF or >450 pg/ml if BMI >35 kg/m 2 |
| RESPECT-HF | 2016 | Interventional/randomized study of renal denervation (RDN) therapy in HFpEF | Compare effects of RDN versus standard medical therapy in HFpEF patients | ECHO with Diastolic dysfunction and/or NT-proBNP > 220 pg/ml |
| RELAX-AHF-2 | 2016 | Interventional/randomized, double-blind/placebo-controlled Phase III study | Evaluate efficacy, safety, and tolerability of IV infusion of serelaxin when added to standard therapy in AHF | BNP ≥ 350 pg/ml or NT-pro-BNP ≥ 1,400 pg/ml |
| GUIDE-IT | 2017 | Experimental group: serum NT-proBNP will guide medical therapy | Determine efficacy of biomarker-guided therapy vs. usual care in patients with systolic dysfunction | NT-proBNP >2000pg/ml or BNP > 400 pg/ml at any time 30 days prior to randomization |
Clinical outcomes based on natriuretic peptide levels
It is reasonable to conclude that natriuretic peptides could enhance patient selection in AHF trials, and their response to treatment serve as a possible predictor of outcomes. The RELAXin in Acute Heart Failure (RELAX-AHF) study was designed to evaluate the effects of serelaxin on dyspnea relief and postdischarge clinical efficacy outcomes and its safety and tolerability. In RELAX-AHF, both worsening HF during the hospital stay and a lack of decrease in NT-proBNP levels during the hospital stay were associated with increased 180-day all-cause mortality. Serelaxin administration significantly diminished signs and symptoms of congestion and reduced NT-proBNP levels 2 days after randomization and the rate of worsening HF. The Aliskiren Trial on Acute HF Outcomes (ASTRONAUT) study was associated with a statistically significant decrease in NT-proBNP level compared with placebo at each time point tested (1, 6, and 12 months). However, this decrease in NT-proBNP was not associated with an improvement in outcomes. The Acute Study of Clinical Effectiveness of Nesiritide in Decompensated Heart Failure trial enrolled patients with a BNP level ≥400 pg/ml or an NT-pro-BNP level ≥1,000 pg/ml. In these contemporary AHF trials, there was little variation with respect to BNP as an inclusion criterion, but there was marked variability in the NT-proBNP levels ( Table 1 ). Additionally, data from Treatment of Preserved Cardiac Function Heart Failure With an Aldosterone Antagonist (TOPCAT) suggest that spironolactone seemed to benefit patients who were enrolled in the natriuretic peptide stratum, which indicates that this biomarker was able to enrich the patient selection process and identify those at greater risk of experiencing the primary event.
Considerations for using natriuretic peptides in clinical trials
If BNP is going to be incorporated as a selection criterion, the timing of the AHF presentation, standardization of the biomarkers levels, and several clinical limitations of this biomarker need to be taken into consideration. First, BNP and NT-proBNP have very similar diagnostic performance characteristics and can be used to rule out HF as a cause of dyspnea in the acute clinical setting. However, there is no easily identifiable optimum cutpoint value for each peptide (the overall sensitivity for both biomarkers in AHF may be as high as 96%, but the specificity is only 69%). The timing of the laboratory measurement is also critical, as data suggest that a delay in obtaining BNP was independently associated with increased inhospital mortality and was associated with a delayed diagnosis of AHF. In RELAX-AHF, patients enrolled presented with AHF within the previous 16 hours with dyspnea at rest or with minimum exertion. In ASTRONAUT, patients were enrolled during acute HF hospitalization but after a stabilization period. This is reflected by the significant difference between BNP and NT-proBNP levels during admission versus randomization (admission BNP and NT-proBNP were 894 and 4,239 pg/ml, respectively, whereas randomized BNP and NT-proBNP were 447 and 2,718 pg/ml, respectively). Although the timing was different in the 2 clinical trials, the BNP cutoffs were similar, and this difference needs to be taken into consideration when analyzing the results.
The impact of natriuretic peptides on HFPEF versus HFREP
BNP levels are lower in heart failure with preserved ejection fraction (HFPEF) than in patients with heart failure with reduced ejection fraction (HFREF). Similarly, in the International Collaborative of NT-proBNP Study, it has been demonstrated that patients with AHF but HFPEF had lower median NT-proBNP concentrations (3,070 pg/ml) compared with those with HFREF (6,536 pg/ml). The overall sensitivity of NT-proBNP to detect AHF in subjects with HFPEF was 84% compared with 92% in those with HFREF. The natriuretic peptide differences based on ejection fraction were recently demonstrated in the RELAX-AHF study where the geometric mean NT-proBNP level for HFREF (LVEF <50%) was 5,535 pg/ml (5,194 and 5,897) versus 3,992 pg/ml (3,632 and 4,388) for HFPEF (LVEF ≥50%). According to the RELAX-AHF protocol, the recorded LVEF was the most recently available, including the one during the index hospitalization. Hence the cutoffs used in the recent clinical trials of AHF may not capture some of the HFPEF population, thus making these data less generalizable to this patient population. This exclusion may be why more of the recent trials are looking more favorable by specifically targeting a population more likely to benefit from this intervention ( Table 2 ).
Clinical Criteria That Can Confound Natriuretic Peptide Levels in AHF
Atrial fibrillation
NT-proBNP has been identified as a significant predictor of incident atrial fibrillation (AF), and BNP levels may correlate with AF burden with respect to chronicity and altered hemodynamics. In a prospective observational cohort study, patients without AF had a mean NT-proBNP of 93 versus 271 pg/ml for those with incident AF. In RELAX-AHF, half of the enrolled patients had a history AF or atrial flutter, and approximately 40% of these subjects were identified during screening. Similarly in ASTRONAUT, 40% of the enrolled patients had AF documented through an electrocardiogram. As the burden of AF may confound the natriuretic peptide levels, it may be reasonable to control or adjust for AF during statistical analysis or exclude these subjects when using BNP as an entry criterion.
Chronic kidney disease
With respect to renal function, the ProBNP Investigation of Dyspnea in the Emergency Department study demonstrated the utility and prognostic ability of NT-proBNP in patients with an estimated glomerular filtration rate (eGFR) ≤60 ml/min/1.73 m 2 . It was later demonstrated prospectively that BNP and NT-proBNP are equally dependent on renal function for their clearance, and there is no difference in the renal clearance between the 2 peptides. The mean eGFR in RELAX-AHF was 53 ml/min/1.73 m 2 , and in ASTRONAUT, it was 67 ml/min/1.73 m 2 . These differences highlight the importance of identifying and defining renal dysfunction (creatinine vs eGFR) during the planning phase of clinical trials of AHF and must be controlled for if BNP is to be used as a biomarker.
Age and gender
With respect to age and gender, a multivariable analysis demonstrated that the strongest predictors of higher natriuretic peptide levels were older age and female gender. Additionally, age and gender should be controlled for, as studies have shown that BNP increases significantly with age and is significantly higher in women than men.
Obesity
Preclinical studies suggest that there is a mechanistic link between obesity and an increased clearance of natriuretic peptide in adipose tissue. Several follow-up clinical studies clearly demonstrated that obese subjects have low circulating natriuretic peptide levels, and the results extend to patients of various ethnic backgrounds. Despite these findings, BNP has been shown to predict worse symptoms, impaired hemodynamics, and higher mortality at all levels of obesity ; thus, it can still be used as a biomarker, but obesity needs to be placed in the appropriate context when using BNP levels in clinical trials. Otherwise, congested obese patients with AHF may be erroneously excluded from AHF trials on the basis of their lower BNP levels. In RELAX-AHF, the body mass index was 29 kg/m 2 , and in ASTRONAUT, it was 27 kg/m 2 . The challenge is extrapolating these clinical trial data to the real-world populations where many of our patients with AHF are obese (body mass index ≥30 kg/m 2 ).
Natriuretic Peptides
Background
Natriuretic peptides (plasma BNP or N-terminal pro-BNP [NT-proBNP]) are cardiac neurohormones dynamically synthesized and secreted from the ventricles in response to volume expansion, ventricular wall stress, and pressure overload. It has been well established in patients where the diagnosis is in question that rapid measurement of natriuretic peptides is useful in establishing or excluding the diagnosis of AHF in patients with acute dyspnea. Patients with a reduction in BNP levels for treatment of AHF had decreased 31 and 180 days mortality compared with those with little or no BNP reduction. These results suggest that response to treatment in BNP levels predicts both short- and long-term mortality risks.
Current limitations of natriuretic peptide assays
BNP and NT-proBNP have been used as an entry criterion in major clinical trials, but the shortcoming of this process is that all the studies differ in their respective end points, the number of patients enrolled, and mean BNP and NT-proBNP levels ( Table 1 ). Another fundamental issue is that a trend has been noted in multiple on-going studies that many centers are no longer performing “routine” measurements of natriuretic peptides, such that their availability as routine, non–protocol-specified measures is decreasing. This trend will not only delay enrollment in studies, since then measuring natriuretic peptides will have to wait until after informed consent is obtained, but also increases the cost. Future AHF trials should obtain BNP and NT-proBNP levels at the time of initial presentation or admission ( Table 2 ). The advantage of presentation is that it provides the best indicator of the initial pretreatment status. However, these types of assessments can only be obtained in clinical trials after informed consent is obtained if done solely for the purpose of the trial. There are several assays and methods available to measure BNP, and this leads to questions regarding assay purity. Although most methods are standardized to give a comparable value at 100 ng/L, agreement between methods more than and less than this point may vary substantially, and there are differences between manual and point-of-care assays versus central laboratory measurements of BNP. Using NT-proBNP as the biomarker of choice could minimize some of the variability seen with available BNP assays.
