NT-proBNP Testing for HFpEF Needs a BMI Makeover

TOPLINE:

Diagnostic thresholds of N-terminal prohormone of brain natriuretic peptide (NT-proBNP) for heart failure with preserved ejection fraction (HFpEF) in ambulatory patients with chronic dyspnearequire body mass index (BMI) stratification for optimal accuracy. The presence of atrial fibrillation alone strongly indicates HFpEF, making NT-proBNP testing less crucial in these cases.

METHODOLOGY:

• Researchers conducted a multicenter evaluation of NT-proBNP thresholds for diagnosing HFpEF in ambulatory patients with chronic unexplained dyspnea, aiming to improve diagnostic accuracy by stratifying the thresholds based on BMI and atrial fibrillation status.
• They established diagnostic thresholds for HFpEF or noncardiac dyspnea in a derivation cohort including 414 patients with HFpEF or definite noncardiac dyspnea who underwent invasive exercise catheterization.
• The derived thresholds were then validated in six independent validation cohorts including patients with definitive diagnoses of HFpEF.
• Optimal NT-proBNP thresholds for HFpEF rule out (sensitivity) and rule in (specificity) were derived, tested, and stratified by obesity and atrial fibrillation status.

TAKEAWAY:

• The currently recommended NT-proBNP threshold of less than 125 pg/mL for HFpEF rule out had an 82% sensitivity for those with a BMI less than 35 and a 67% sensitivity for those with a BMI ≥ 35, while a current age- and BMI-adjusted threshold for HFpEF rule in had a specificity of only 65%.
• In the derivation cohort, a rule-out NT-proBNP threshold (less than 50 pg/mL) lower than the currently recommended threshold demonstrated an overall sensitivity of 93%, with sensitivities of 97% (95% CI, 95%-99%) in those with a BMI less than 35 and 86% (95% CI, 79%-93%) in those with a BMI ≥ 35; this threshold demonstrated consistent results across the validation cohorts.
• A rule-in NT-proBNP threshold of ≥ 500 pg/mL achieved an overall specificity of 86%, with a specificity of 85% among those with a BMI less than 35, while a lower threshold of ≥ 220 pg/mL provided a specificity of 88% among those with a BMI ≥ 35; the results were consistent across the validation cohorts.
• HFpEF was prevalent in more than 98% patients with atrial fibrillation and dyspnea, suggesting that atrial fibrillation itself is a robust biomarker of HFpEF, with no incremental value added by NT-proBNP.

IN PRACTICE:

“In contrast to previous literature focused on acute decompensated HF or chronic HF defined imprecisely by investigator opinion, we found that currently recommended NT-proBNP thresholds have only modest accuracy among ambulatory patients with chronic dyspnea when applied in a binary fashion using a single diagnostic threshold,” the authors wrote.

“Use of separate diagnostic thresholds to optimize operational characteristics allows better diagnostic performance, particularly in context of intermediate pretest probability,” they added.

SOURCE:

This study was led by Yogesh N.V. Reddy, MBBS, MSc, and Atsushi Tada, MD, of the Mayo Clinic in Rochester, Minnesota. It was published online on January 22, 2025, in Circulation.

LIMITATIONS:

The cohorts included predominantly White and Asian participants, with limited representation of Black, South American, or African participants. The findings were based on single measurements of NT-proBNP thresholds. The identification of the derivation cohort was retrospective, which may have introduced selection bias. Additionally, the measurements of NT-proBNP thresholds were missing at random for 24% of participants in the derivation cohort.

DISCLOSURES:

Some authors received support through several grants and awards from agencies including the National Institutes of Health, US Department of Defense, and Mayo Clinic Cardiovascular Division, and one author received support through a research fellowship from the Uehara Memorial Foundation, Japan. Several authors reported having financial ties with multiple institutes, foundations, and companies outside the submitted work.

This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication.