Diagnostic performance of guideline-recommended natriuretic peptide thresholds to detect echocardiographic preclinical heart failure
H Ismahel, M C Petrie, A Bayes-Genis, P L Myhre, M Vaduganathan, A M Shah, S D Solomon, J J V Mcmurray, K F DochertyAbstract
Background
Periodic measurement of natriuretic peptides in patients with heart failure (HF) risk factors, followed by echocardiography in those with elevated levels, has been proposed to identify individuals with preclinical HF (Stage B HF) at risk of progression to symptomatic HF, enabeling optimisation of primary prevention. Elevated natriuretic peptide levels in older adults are frequently multifactorial and may not reliably indicate asymptomatic adverse cardiac remodelling.
Purpose
To evaluate the diagnostic performance of guideline-recommended NT-proBNP thresholds for detecting echocardiographic preclinical HF in community-based older adults at risk of HF.
Methods
We analysed participants from the Atherosclerosis Risk in Communities (ARIC) study at visit 5 with no HF history and at least one of hypertension, coronary heart disease, diabetes, or obesity. Echocardiographic stage B HF was defined as any of: left ventricular (LV) mass index >95/115 g/m² (females/males), left atrial volume index >34 mL/m² (SR) or >40 mL/m² (AF), or LV ejection fraction <50%. The performance of three NT-proBNP strategies was evaluated: (1) AHA/ACC rule-in threshold (≥125 pg/mL); (2) ESC HFA age-adjusted rule-in thresholds (≥150 pg/mL if <75 years, ≥300 pg/mL if ≥75 years); and (3) ESC HFA rule-out threshold (≤50 pg/mL). Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated for the detection of echocardiographic Stage B HF overall, and in subgroups of interest (sex, age, race, eGFR and BMI).
Results
Among 3374 participants (median age 75yrs, 57% female), median NT-proBNP was 130 pg/mL (IQR 66-250 pg/mL), with 52% ≥125 pg/mL. Echocardiographic stage B HF was present in 765 (23%) individuals, who had higher median NT-proBNP than those without remodelling (228 vs 112 pg/mL). The HFA rule-out threshold (≤50 pg/mL) was met by 558 (16.5%) participants and had high sensitivity (94.9%) and NPV (93.0%) (Fig. 1) with minimal subgroup variation (Fig. 2A). The fixed ≥125 pg/mL rule-in had a sensitivity of 70.2% and NPV of 86.1%, but limited specificity (53.9%) with a low PPV (30.9%). The HFA age-adjusted rule-in thresholds had a higher specificity of 75.6%, but PPV improved only modestly to 36.6% (Fig. 1), with lower sensitivity (48%). Both rule-in strategies had lower PPV in women, white individuals, those <75yrs, and those with lower eGFR (Fig. 2B, 2C).
Conclusion
In older adults with HF risk factors, the HFA NT-proBNP rule-out threshold (≤50 pg/mL) had high negative predictive value for excluding asymptomatic adverse cardiac remodelling, albeit with low yield. Current rule-in thresholds had limited diagnostic performance to detect echocardiographic structural abnormalities, with a high false-positive rate. These findings suggest that use of current NT-proBNP thresholds is unlikely to be feasible in clinical practice to identify individuals with both biochemical and echocardiographic features of preclinical HF.Performance of NT-proBNP cutoffsFor image description, please refer to the figure legend and surrounding text.Performance of cutoffs across subgroupsFor image description, please refer to the figure legend and surrounding text.