Inclusion Criteria for HFpEF Clinical Trials: Making the Case for Precision Diagnosis and Greater Inclusivity

Heart failure (HF) with preserved ejection fraction (HFpEF) is a heterogeneous clinical syndrome that results in resting or exertional left atrial (LA) pressure elevation with concomitant symptoms of congestion. Despite its broad definition, the diagnosis of HFpEF remains challenging due to the absence of a singular non-invasive measurement that accurately estimates LA pressures. In certain cases, the diagnosis of HFpEF may only become apparent during invasive hemodynamic measurement, in which a pulmonary capillary wedge pressure (PCWP) measurement of ≥15 mmHg at rest or ≥25 mmHg with exertion (or a PCWP/cardiac output [CO] slope >2.0) has been considered the “gold-standard” for diagnosis.¹ Due to the perceived infeasibility of requiring invasive hemodynamic testing for enrollment in large-scale HFpEF clinical trials, such investigations have traditionally relied upon a combination of natriuretic peptides (NPs) and echocardiographic measurements as inclusion criteria. As our knowledge of HFpEF has grown, there have been several iterations of eligibility criteria using various cutoffs of NPs and echocardiographic measures. Trial definitions change over time to balance the maximization of enrollment of true HFpEF patients with achievement of sufficient power to detect a treatment effect by enrolling a HFpEF cohort that is enriched for events such as HF hospitalization. However, there remains ongoing concern that clinical trial definitions of HFpEF may exclude a substantial proportion of HFpEF patients encountered in clinical practice, limiting the enrollment of a large proportion of HFpEF patients.

In this issue, Vaishnav et al. highlight this concern by applying various trial inclusion criteria to a cohort of patients in which the HFpEF diagnosis was made through invasive hemodynamic testing.² The investigators utilized criteria from 4 HFpEF clinical trials: I-PRESERVE, RELAX, TOPCAT, and PARAGON-HF. A minority of the cohort met inclusion criteria for each trial, ranging from 13% (PARAGON-HF) to 38% (RELAX). The most common reasons for ineligibility were low NPs, obesity, uncontrolled hypertension, young age, and anemia. Over 2 years, there were no differences in clinical outcomes between those who met inclusion criteria for each trial compared with those who did not. The authors should be congratulated as this study offers insight into specific reasons driving the lack of enrollment of a substantial proportion of HFpEF patients. Nevertheless, there are important limitations that require consideration in the interpretation of their results. This was a single-center study involving a HFpEF referral center, which may limit generalizability. In addition, the rationale for evaluating these 4 HFpEF trials specifically is not clear, as 3 of the 4 trials are >7 years old, and several HFpEF trials have been completed that use updated inclusion criteria. Examples of such contemporary trials include PRESERVED-HF, EMPEROR-Preserved, and INDIE-HFpEF. It is likely that a large portion of patients in the study cohort may have met inclusion criteria for PRESERVED-HF given its lower N-terminal pro-B-type natriuretic peptide (NT-proBNP) requirement and lack of an upper limit of body mass index (BMI).³ EMPEROR-Preserved⁴ enrolled adults >18 years (as opposed to >60 years in I-PRESERVE) with a higher BMI cutoff, which would allowed for greater inclusion of the HFpEF patients studied by Vaishnav et al. Finally, it is likely that a large proportion of the study population would have been eligible for INDIE-HFpEF, as it allowed for inclusion by elevated PCWP at rest or with exertion (without need for elevated NT-proBNP).⁵ Additionally, the PARAGON-HF trial incorporated a lower inclusion criterion for NT-proBNP among those with prior HF hospitalization (>200 pg/mL).⁶ It does not appear the authors accounted for this modification, which likely underestimated the proportion of patients who would have been eligible. Despite these limitations, the study by Vaishnav et al. nicely demonstrates objective reasons for the under-enrollment of HFpEF patients in trials and offers the opportunity to reflect upon specific inclusion criteria, which are relevant for ongoing and future HFpEF clinical trials.

The Yin and Yang of Natriuretic Peptides

NT-proBNP levels have been incorporated into the inclusion criteria of the vast majority of large-scale HFpEF clinical trials. Due to the impaired production and accelerated clearance of BNP in the setting of adiposity, NT-proBNP levels are lower in obese HFpEF. This finding, in concert with the high prevalence of obesity in contemporary HFpEF populations, have created challenges with the use of NT-proBNP as an inclusion criterion.⁷ Furthermore, treatment with drugs such as mineralocorticoid antagonists and neprilysin inhibitors can lower NT-proBNP levels, thus in effect penalizing enrolling sites who use these evidence-based medications for the treatment of HFpEF.

Some have argued that the use of NPs for inclusion into HFpEF trials is supported by the findings of TOPCAT trial. TOPCAT allowed for enrollment based upon either elevation in NT-proBNP or HF hospitalization. On first glance, it appeared that patients enrolled in the HF hospitalization strata did not have a high event rate. However, after further analysis of the trial, it became apparent that this had more to do with region of enrollment than the enrollment criteria per se. Sites in Russia and the Republic of Georgia were much more likely than the sites in the Americas to use the HF hospitalization criteria, and patients enrolled in Russia and the Republic of Georgia had a much lower event rate than patients enrolled in the Americas. In the Americas, a history of HF hospitalization was associated with high risk for adverse outcomes.⁸

Higher NT-proBNP has been consistently associated with primary composite outcomes in HFpEF trials.^9,10 Notably, the strength of this association is consistent across the spectrum of BMI,¹⁰ suggesting that natriuretic peptide thresholds can vary based on BMI, which will allow for a higher proportion of obese HFpEF patients to enroll in trials. This type of BMI-adjusted natriuretic peptide threshold is currently being applied in trials involving HFpEF patients (RELIEVE-HF [V-wave atrial shunt device] and EMBARK-HFpEF [mavacamten]). Alternatively, several recent HFpEF trials have navigated challenges associated with enrollment of obese HFpEF through modification of NT-proBNP cutoffs. The PRESERVED-HF trial utilized an NT-proBNP cutoff of 225 pg/mL,³ significantly lower than cutoffs in the EMPEROR-Preserved and DELIVER trials (300 pg/mL) of SGLT2 inhibitors.

Finally, elevated NPs are not the only way to ensure enrollment of high-risk HFpEF patients who will have a high event rate necessary to adequately power pivotal trials. The MAGGIC risk score (which does not include NPs) can predict both HF hospitalizations and mortality and can be used as an alternative to NPs.¹¹

Invasive Exercise Hemodynamics: Making the Case for Precision Diagnosis of HFpEF for Inclusion into HFpEF Clinical Trials

There has been a shift toward incorporation of invasive hemodynamics as inclusion criteria for HFpEF trials, and several HFpEF trials serve as examples (RELAX, INDIE-HFpEF, REBALANCE-HF, REDUCE LAP-HF II). Despite the data that elevated NPs are associated with adverse outcomes, they may not be necessary in all HFpEF trials. In the REDUCE LAP-HF II trial¹² (n=626 randomized to an atrial shunt device vs. sham procedure), in which patients were not required to have an elevated NT-proBNP and diagnosis of HFpEF was confirmed by invasive exercise hemodynamic testing, the HF event rate was approximately 1.5 times higher, and KCCQ much lower, than in contemporary phase 3 HFpEF trials of pharmacotherapies.^13,14 Thus, in patients in whom the gold standard diagnosis of HFpEF is made based on invasive exercise hemodynamic testing, NP inclusion criteria for HFpEF trials may not be necessary.

Some may argue that while requirement of invasive exercise hemodynamics is viable for interventional HFpEF trials, it is not feasible for large-scale trials of pharmacotherapies with hard clinical endpoints. However, the REDUCE LAP-HF II trial proved that exercise invasive hemodynamic testing is feasible in a large-scale multi-center trial (89 sites) and ensured that all patients have an accurate HFpEF diagnosis. A provocative question for the field is whether it is better to enroll 6,000+ patients with possible HFpEF (diagnosed non-invasively, primarily by elevation of NPs) with a primary endpoint that only includes hard outcomes (e.g., cardiovascular death and HF hospitalization) vs. 600-800 patients with definite HFpEF (proven by invasive exercise hemodynamic testing) with a hierarchical primary endpoint that includes cardiovascular death, HF hospitalization, and health status or exercise capacity endpoints. We would argue that the latter may be more fruitful and is not any different than several trials of patients with epicardial coronary artery disease (CAD), which have required invasive angiographic evidence of significant coronary stenoses. In addition, exercise testing must become routine in the evaluation of HFpEF patients in the routine clinical and clinical trial settings. Exercise provocation is not questioned in the evaluation of ischemia in patients at risk for or who have CAD and should be viewed no differently in HFpEF patients, who carry a worse prognosis than CAD patients.

In summary, the state of HFpEF trials continues to evolve. Trial inclusion criteria may vary depending upon the HFpEF phenotype, therapy under investigation, and the primary outcome of the trial. We advocate for challenging the current dogma of NP-centric enrollment into HFpEF clinical trials and considering the gold-standard invasive exercise hemodynamic testing to ensure the correct diagnosis of HFpEF, which would simultaneously be more inclusive and could result in an improved track record for HFpEF trials in the future (Figure).

Figure. Challenging the Current Dogma of Natriuretic Peptide-centric Enrollment into HFpEF Clinical Trials.

The current dogma of phase 3 HFpEF clinical trials involves the inclusion of very large numbers of patients and an over-reliance on elevated natriuretic peptides, which can result in the inclusion of a large number of patients without HFpEF, symptoms driven by comorbidities rather than HFpEF, and/or alternative causes of elevated natriuretic peptides, which may be one reason for the poor track record of prior HFpEF trials. In the future state, HFpEF trials could be done with much fewer patients, use a hierarchical (e.g., Finkelstein-Schoenfeld) endpoint, include patient-centric outcomes (health status and/or exercise capacity), and allow for inclusion of a greater proportion of patients who actually have invasively-proven HFpEF. Screening committees, comprised of heart failure experts who review the medical history and results of screening tests (e.g., laboratory tests and echocardiography images) in enrolled patients could also increase the precision and specificity of the HFpEF diagnosis. These changes may result in an improved HFpEF clinical trial track record.

HFpEF = heart failure with preserved ejection fraction; NP = natriuretic peptide; PAH = pulmonary arterial hypertension; HF = heart failure; CV = cardiovascular; LVH = left ventricular hypertrophy; LA = left atrial; KCCQ = Kansas City Cardiomyopathy Questionnaire.