B-type natriuretic peptide identifies transition from atrial to ventricular remodelling in older adults: a non-linear association with left atrioventricular coupling index
D J Lim, L Y Teo, S Leng, X D Zhao, L Zhong, R S Tan, F Gao, A S KohAbstract
Background
Left atrioventricular coupling index (LACI) reflects the balance of left atrial (LA) and left ventricular (LV) remodelling at end-diastole. LA remodelling which often precedes LV dilatation in the progression to heart failure (HF), may be captured by LACI. B-type natriuretic peptide (BNP), a widely available marker of myocardial stretch, could serve as a surrogate of such remodelling transitions. We hypothesized that BNP may have a non-linear relationship with LACI, marking a shift from atrial to ventricular predominant cardiac remodelling in older adults.
Methods
Community older adults without known cardiovascular disease or HF were prospectively recruited to undergo cardiac magnetic resonance and serum sampling. LACI was determined as the ratio of LA to LV end-diastolic volume (%). BNPspline was modelled using restricted cubic splines with 3 degrees of freedom to detect threshold effects. Logistic regression evaluated LACI as continuous (per 1% increase) and elevated (≥25%), based on data-driven Youden’s index, consistent with prior external validation. Incremental discrimination was assessed by comparing area under the receiver operating characteristic curve (AUC) for a multivariable model of known cardiovascular risk factors (CVRF) with and without BNPspline, with bootstrap validation.
Results
In 234 participants (70.7±9.5 years; 55% male), median BNP levels were 46 pg/mL (IQR 23-88 pg/mL). BNPspline demonstrated a significant non-linear association with LACI (pnonlinearity<0.001). As BNP increased, LACI rose to an inflection point around 100pg/mL, after which higher BNP levels were associated with a decline in LACI. This pattern remained significant after multivariable adjustment for known CVRF (global p < 0.001). In models for elevated LACI (≥25%), adding BNPspline to the CVRF model significantly improved discrimination. The area under the curve (AUC) increased from 0.729 to 0.753 (∆AUC 0.024, 95% CI 0.002-0.073, p = 0.018), confirmed on bootstrap validation of 500 replications. Three-dimensional surface plots of LACI by age and BNPspline illustrated a clear transition from atrial-predominant to ventricular-predominant remodelling across BNP levels with increasing age.
Conclusion
BNP demonstrates a non-linear association with LACI and provides incremental value in discriminating elevated LACI beyond clinical risk factors in older adults. In older adults, where atrial stiffening and declining ventricular compliance coexist, BNP appears to index a transition from atrial- to ventricular-predominant remodelling, rather than simply reflecting myocardial stretch. These findings position BNP as a readily available biomarker of cardiac remodelling phases, providing mechanistic insights into HF development in older adults.Figure 1