Heterogeneity of recovery in HFrEF: biochemical and structural recovery phenotypes and prognosis

Abstract

Introduction and Aim

Reverse remodelling in HFrEF involves both biochemical improvements, reflecting neuro-hormonal inhibition, and structural changes in ventricular function, yet their relative and independent prognostic roles are unclear. This study evaluated baseline characteristics across recovery phenotypes during medical therapy (GDMT) optimization and assessed the prognostic impact of biochemical versus structural recovery.

Methods

We retrospectively analyzed 270 HFrEF patients followed for 3 years after GDMT optimization. Patients were categorized into three mutually exclusive groups according to predefined criteria: biochemical recovery only (≥50% NT-proBNP reduction and follow-up NT-proBNP <1000 pg/mL, without LVEF improvement), LVEF recovery (>10% absolute increase and follow-up LVEF >40%, irrespective of NT-proBNP changes) and no recovery (neither criterion fulfilled). Baseline characteristics were compared using ANOVA or χ² tests. Kaplan–Meier curves assessed 3-year HF event-free survival. Prognostic effects were quantified using univariable and multivariable Cox regression, and incremental value was evaluated by changes in chi-square model.

Results

Baseline characteristics were similar across groups. Only sex distribution and eGFR differed between groups. Three-year event-free survival varied markedly across phenotypes (log-rank p<0.001). In univariable Cox regression, LVEF recovery was associated with a significantly lower risk of HF events (HR 0.288, 95% CI 0.154–0.538; p<0.001), and biochemical recovery conferred a similarly strong protective effect (HR 0.244, 95% CI 0.112–0.532; p<0.001).

When both domains were included simultaneously, biochemical recovery remained independently associated with prognosis (HR 0.309, 95% CI 0.145–0.659; p=0.002), whereas LVEF recovery lost significance (HR 0.437, 95% CI 0.190–1.008; p=0.052). Biochemical recovery provided significant incremental prognostic information beyond LVEF recovery (Δχ²=22.4; p<0.001), whereas LVEF recovery added no meaningful improvement over biochemical recovery (Δχ²=3.77; p=0.052).

Conclusions

Biochemical reverse remodelling emerged as the only independent predictor of outcomes. Interestingly, these findings indicate that neuro-hormonal inhibition, and maybe, appropriate decongestion— captured biochemically — is the primary driver of prognostic benefit after GDMT optimization, whereas structural recovery appears later and may be partially mediated by upstream neuro-hormonal improvement.For image description, please refer to the figure legend and surrounding text.