European Journal of Heart Failure | 2021
Omics, machine learning, and personalized medicine in heart failure with preserved ejection fraction: promising future or false hope?
Abstract
Heart failure with preserved ejection (HFpEF) is a heterogeneous syndrome characterized by clinical signs and symptoms of exercise intolerance, volume overload, and impairment in functional status.1,2 HFpEF was initially thought to be a disease of impaired left ventricular relaxation.3 However, our understanding of HFpEF has evolved considerably over time. It is now considered a multisystem, geriatric syndrome perpetuated by systemic inflammation, cardiometabolic dysregulation, clustering of comorbidities, and accelerated functional decline.1,4–6 However, despite these advancements in our understanding of HFpEF, it remains refractory to most pharmacological therapies. The repeated failures of large-scale clinical trials evaluating the efficacy of well-established cardioprotective therapies in HFpEF highlight the need for a paradigm shift in our approach to its treatment. A major challenge in the treatment of HFpEF is the diagnosis and classification of this heterogeneous clinical syndrome. The one-size-fits-all approach to identifying patients with HFpEF based on the absence of reduced ejection fraction in the setting of clinical heart failure has proven inadequate for developing effective therapies for its management. Attempts to better understand the mechanistic heterogeneity in HFpEF have typically focused on identifying subgroups of patients with similar clinical characteristics. This ‘phenomapping’ approach uses big data-based unsupervised clustering methods to group patients in an unbiased manner. First pioneered by Shah et al.7 in 2015, the phenomapping efforts to identify biological signals and unique clinical characteristics in patients with HFpEF have seen substantial growth over the past decade. However, till recently, most of these phenomapping efforts