Miller PE; Department of Biostatistics, Boston University School of Public Health, Boston, MA, USA.
Gajjar P; Mitchell GF; Khan SS; Vasan RS; Larson MG; Lewis GD; Shah RV;Nayor M;
ESC heart failure [ESC Heart Fail] 2024 Jun 28.
Date of Electronic Publication: 2024 Jun 28.
Aims: New tools are needed to identify heart failure (HF) risk earlier in its course. We evaluated the association of multidimensional cardiopulmonary exercise testing (CPET) phenotypes with subclinical risk markers and predicted long-term HF risk in a large community-based cohort.
Methods and Results: We studied 2532 Framingham Heart Study participants [age 53 ± 9 years, 52% women, body mass index (BMI) 28.0 ± 5.3 kg/m 2 , peak oxygen uptake (VO 2 ) 21.1 ± 5.9 kg/m 2 in women, 26.4 ± 6.7 kg/m 2 in men] who underwent maximum effort CPET and were not taking atrioventricular nodal blocking agents. Higher peak VO 2 was associated with a lower estimated HF risk score (Spearman correlation r: -0.60 in men and -0.55 in women, P < 0.0001), with an observed overlap of estimated risk across peak VO 2 categories. Hierarchical clustering of 26 separate CPET phenotypes (values residualized on age, sex, and BMI to provide uniformity across these variables) identified three clusters with distinct exercise physiologies: Cluster 1-impaired oxygen kinetics; Cluster 2-impaired vascular; and Cluster 3-favourable exercise response. These clusters were similar in age, sex distribution, and BMI but displayed distinct associations with relevant subclinical phenotypes [Cluster 1-higher subcutaneous and visceral fat and lower pulmonary function; Cluster 2-higher carotid-femoral pulse wave velocity (CFPWV); and Cluster 3-lower CFPWV, C-reactive protein, fat volumes, and higher lung function; all false discovery rate < 5%]. Cluster membership provided incremental variance explained (adjusted R 2 increment of 0.10 in women and men, P < 0.0001 for both) when compared with peak VO 2 alone in association with predicted HF risk.
Conclusions: Integrated CPET response patterns identify physiologically relevant profiles with distinct associations to subclinical phenotypes that are largely independent of standard risk factor-based assessment, which may suggest alternate pathways for prevention.