Inspecting the risk of heart failure in general population using metabolic health status and obesity profiles

Background: Obesity and metabolic unhealth don’t always co-exist as the risk factors of heart failure (HF). Phenotypes derived from obesity and metabolic unhealth have promising clinical relevance. Their predictive effect for different subtypes of HF is to be investigated. Methods and findings: Totally 8018 participants from the ARIC study were classified into four phenotypes: metabolic healthy non-obesity (MHNO), metabolic healthy obesity (MHO), metabolic unhealthy non-obesity (MUNO) and metabolic unhealthy obesity (MUO). Cox models were applied to explore the relationship between these phenotypes and the risk of HF with preserved ejection fraction (HFpEF, left ventricular ejection fraction [LVEF] ≥50%) or HF with reduced or mildly reduced LVEF (HFrEF/HFmrEF, LVEF <50%) in total population and subgroups. Association between phenotypes transition and HF was further analyzed. Compared with MHNO, participants with MHO (hazard ratio and 95% confidence interval, 2.04 [1.61-2.59]), MUNO (1.80 [1.40-2.32]) and MUO (2.50 [1.95-3.20]) were related to higher HFpEF risk, MUNO (1.74 [1.36-2.22]) and MUO (1.92 [1.49-2.49]) were associated with higher HFrEF/HFmrEF risks. Subgroup analyses revealed that the associations between the phenotypes and HF risk were more distinct ( P _-interaction < 0.009) in participants < 55 years. Serum lipid might impact the relationship of the phenotypes with HFrEF/HFmrEF ( P _-interaction =0.033). From a dynamic aspect, persistent MHO, MUNO or MUO was associated with increased HFpEF risk, whereas progression from MHNO to MHO didn’t exhibit higher HFrEF/HFmrEF risk. Conclusions: Both metabolic unhealth and obesity independently and cumulatively contributed to HFpEF risk, while metabolic unhealth rather than obesity are more influential in HFmrEF/HFrEF risk.