There is currently no data to predict response to treatment by sacubitril/valsartan.

To assess whether unbiased clustering analysis, using dense phenotypic data, could identify phenotypically distinct HF-REF subtypes with good or no response after 6 months of sacubitril/valsartan administration.

A total of 78 patients treated by ACE inhibitor or AAR2, were prospectively assigned to equimolar sacubitril/valsartan replacement. We collected many continuous variables. Phenotypic domains were imputed with 5 eigenvectors for missing value, then filtered if the Pearson correlation coefficient was>0.6 and standardized to mean±SD of 0±1. Thereafter, we used agglomerative hierarchical clustering for grouping phenotypic variables and patients, then generate a heat map (Fig. 1). Subsequently, participants were categorized using Penalized Model-Based Clustering. P<0,05 was considered significant.

Mean age was 60.4±13.4 year old 79.0% patients were males. 16 phenotypic domains were isolated (Fig. 1) and 3 phenogroups were identified (Table 1). Phenogroup 1 was remarkable by left ventricular involvement with moderate diastolic dysfunction (DD), no mitral regurgitation (MR) and no pulmonary hypertension (PH). Phenogroups 2 and 3 corresponded to patients with severe PH, related to severe DD or MR. In both phenogroups, the left atrium was significantly enlarged and the right ventricle was remodelled. The echocardiographic response to sacubitril/valsartan was comparable in all groups with similar improvement of EF and reduction of cardiac chambers dimensions. The clinical response was even better in phenogroups 2 and 3.

HF-REF patients with severe diastolic dysfunction, significant mitral regurgitation and elevated pulmonary hypertension by echocardiographic had similar reverse remodelling but better clinical improvement than patients with isolated left ventricular systolic dysfunction.