P348 - UNSUPERVISED PHENOTYPIC CLUSTERING IN DILATED AND HYPOKINETIC NON-DILATED FORMS OF NON-ISCHEMIC CARDIOMYOPATHY AND ASSOCIATIONS WITH FUTURE CLINICAL OUTCOMES

Background: Despite significant phenotypic heterogeneity of idiopathic non-ischemic cardiomyopathy (NICM), management remains generalized and dominantly guided by left ventricular (LV) ejection fraction (EF) and New York Heart Association (NYHA) class. Phenotypic clustering holds promise for identifying subgroups who may benefit from personalized therapy. Only a single European study has explored phenotype clustering in NICM, studying only the dilated cardiomyopathy (DCM) subtype with limited patient-reported features. We sought to characterize both DCM and hypokinetic non-dilated cardiomyopathy (HNDC) NICM subtypes through phenomics-based clustering, inclusive of comprehensive clinical and CMR features, and identify cluster associations with future adverse outcomes.

METHODS AND RESULTS: A total of 1,142 patients with idiopathic NICM were identified from the Cardiovascular Imaging Registry of Calgary (CIROC), defined as a CMR LV EF < 50% in the absence of any identifiable ischaemic or non-ischaemic aetiology. Patients were sub-classified as DCM (n=653) and HNDC (n=489) cohorts based on sex-matched, BSA-defined reference ranges. All patients underwent baseline health questionnaires and standardized reporting at time of CMR and were followed for a minimum of 6 months for the composite outcome of all-cause mortality, survived cardiac arrest, ventricular tachycardia, or hospital admission for heart failure. The Kamila clustering algorithm was applied to a total of 56 features, including patient-reported (e.g., NYHA class, quality of life, and socio-demographics), CMR-derived (e.g., chamber volumes and fibrosis burden) and electronic health record-derived (e.g., labs and medications). Baseline clinical and imaging characteristics, as well as future clinical outcomes, were compared between the resulting clusters.

Three clusters were identified for both DCM and HNDC sub-cohorts, each demonstrating unique clinical and CMR characteristics (Table 1). In both sub-cohorts, Cluster 1 was younger with mildly reduced LV EF. Cluster 2 exhibited worse clinical symptoms and severe LV systolic dysfunction with high fibrosis burden. Cluster 3 was older with higher prevalence of comorbidities and moderate systolic dysfunction. In HNDC, Cluster 2 experienced a 4.7-fold increased risk of the composite outcome versus Cluster 1 (Figure 1; HR [95% CI] 4.7 [2.4-9.2]; p< 0.001), where-as a trend towards worse outcomes was observed for cluster 3 (HR 2.0 [0.9-4.3]; p=0.07). In DCM, both Clusters 2 and 3 experienced a 2.8-fold increased risk of the outcome versus Cluster 1 (HR for each 2.8 [1.8-4.4]; p< 0.001).

Conclusion: Unsupervised phenotype clustering identifies high risk patients across both DCM and HNDC forms of NICM. This novel approach has the potential to improve capacity for the delivery of personalized care in patients with NICM.