Lean BMI shifts the mortality landscape toward pump failure in japanese heart failure patients with primary-prevention high-voltage devices: a post hoc analysis of the HINODE Registry
K Nakamura, Y E Enomoto, J Y Yamamomo, T S Shimizu, N S Sahara, M A Asami, T I Ikeda, H H HaraAbstract
Background
Low body mass index (BMI) paradoxically associates with worse outcomes in heart failure (HF), yet whether lean BMI shifts cause-specific mortality patterns and thereby diminishes the net benefit of implantable cardioverter-defibrillator (ICD) therapy has not been examined in Asian populations.
Purpose
To determine whether lean BMI modifies the balance between ICD-unpreventable competing deaths and ICD-treatable arrhythmic events in primary-prevention ICD/CRT-D recipients from the prospective Japanese HINODE registry.
Methods
Among 171 high-voltage device (ICD/CRT-D) recipients in HINODE, 166 with BMI data were analyzed (mean follow-up 20.2 months). Lean BMI was defined as <20 kg/m² (near the first quartile; sensitivity analysis at <18.5). Deaths were classified by adjudicated organ cause into arrhythmic, pump failure/other cardiac, and non-cardiovascular. Kaplan-Meier and Cox proportional hazards analyses assessed all-cause mortality and ventricular arrhythmia (VA) events. A 2×2 interaction analysis (lean BMI × age >70 years) identified the highest-risk phenotype.
Results
Lean patients (n=34, 20.5%) had significantly higher all-cause mortality than non-lean patients (n=132): 26.5% vs 9.1% (log-rank P=0.033; HR 2.49, 95% CI 1.04–5.94, P=0.040; 24-month survival 76.5% vs 92.7%). This excess mortality was driven predominantly by pump failure deaths (14.7% vs 4.5%; 3.3-fold), which accounted for 56% of the mortality difference. In contrast, VA event rates were comparable (11.8% vs 9.8%; P=0.81). Among classifiable lean deaths, 75% were non-arrhythmic — deaths that ICDs cannot prevent. BMI quartile analysis demonstrated a dose-response relationship (mortality: Q1 21%, Q2 12%, Q3 12%, Q4 5%). Continuous BMI was also a significant predictor (HR 0.85 per 1 kg/m² increase, P=0.013). Sensitivity analysis using BMI <18.5 yielded even stronger separation (36.4% vs 9.0%; P=0.003). The 2×2 interaction revealed that elderly lean patients (n=17) had strikingly poor outcomes: 41.2% mortality and 58.8% composite events versus 5.8% and 29.0% in young non-lean patients. This persisted despite comparable or higher guideline-directed medical therapy utilization (MRA 65% vs 42%).
Conclusions
In Japanese primary-prevention ICD/CRT-D recipients, lean BMI shifted the mortality landscape toward pump failure — deaths that ICDs cannot prevent — while arrhythmic risk remained comparable. This mechanistically explains why ICD net benefit diminishes in lean elderly patients: not because arrhythmic risk decreases, but because competing pump failure mortality overwhelms device benefit. These findings support incorporating BMI into pre-implantation risk–benefit assessment alongside LVEF and NYHA class.Figure