PROVE-HF Biomarker Improvement Algorithm (PROVE-HF Calculator)

PROVE-HF Biomarker Improvement Algorithm — Explanation and Clinical Context The PROVE-HF (Prospective Study of Biomarkers, Symptom Improvement, and Ventricular Remodeling During Sacubitril/Valsartan Therapy for Heart Failure) program examined whether changes in NT-proBNP after initiation of sacubitril/valsartan are associated with reverse cardiac remodeling and improvements in LV function. In the published cohort (Januzzi JL et al., JAMA 2019), median NT-proBNP decreased from 816 pg/mL at baseline to 455 pg/mL at 12 months, and mean LVEF increased by 9.4 percentage points. The study reported significant but modest correlations between the change in log2 NT-proBNP and changes in LVEF and LV volumes (e.g., r ≈ -0.38 for LVEF).

The calculator above performs three basic steps: (1) computes the percentage and log2 fold change in NT-proBNP between two timepoints; (2) applies an empirical, cohort-scale slope (≈ -11.15 percentage points change in LVEF per 1 unit change in log2 NT-proBNP) derived from the PROVE-HF group average decrease and observed mean LVEF rise; and (3) projects a predicted LVEF by adding that estimated change to the supplied baseline LVEF. This provides a quick, transparent, cohort-level expectation of cardiac functional change associated with the observed biomarker trajectory in PROVE-HF-like patients. The model is intentionally simple and exploratory: it assumes linearity on the log2 NT-proBNP scale and is not adjusted for confounders, medical therapy differences, baseline comorbidities, or measurement timing.

Use this tool as an educational aid and not as a replacement for individualized clinical assessment. Limitations: The slope used here is an approximation derived from group medians/means in the PROVE-HF publication and not an individual-level regression model. Individual response may differ; biomarker changes can be influenced by renal function, body habitus, measurement timing, and concurrent therapies.

The original PROVE-HF study was an open-label, single-arm cohort investigating associations rather than building a calibrated prognostic algorithm. If you require an individualized predictive model, patient-level regression coefficients and validation are needed.