Assessing Cardiac Amyloidosis Subtypes by Unsupervised Phenotype Clustering Analysis

Area of Science:

• Cardiology
• Genetics
• Medical Diagnostics

Background:

• Cardiac amyloidosis (CA) encompasses light-chain (AL), hereditary transthyretin (ATTRv), and wild-type transthyretin (ATTRwt) amyloidosis, often presenting with cardiac symptoms.
• Phenotypic heterogeneity in CA contributes to diagnostic delays and poorer patient outcomes.

Purpose of the Study:

• To identify distinct clinical profiles within a large cohort of suspected cardiac amyloidosis patients using clustering analysis.
• To understand the varied presentations and prognostic implications across different CA subtypes.

Main Methods:

• Unsupervised clustering analysis, specifically artificial neural network-based self-organizing maps, was applied to a database of 1,394 patients with suspected CA.
• Clinicobiological phenotyping was used to identify patient profiles independent of final diagnosis and prognosis.

Main Results:

• Seven distinct patient clusters with contrasting profiles and prognoses were identified.
• Light-chain amyloidosis (AL) patients formed a distinct cluster.
• Hereditary transthyretin amyloidosis (ATTRv) patients were spread across four clusters, with one overlapping non-amyloidosis patients.
• Wild-type transthyretin amyloidosis (ATTRwt) patients distributed across three clusters, showing varied risk factors and prognoses.

Conclusions:

• Clustering analysis revealed 7 clinical profiles associated with varying characteristics, prognosis, and diagnostic associations in cardiac amyloidosis.
• These findings highlight opportunities for improved diagnosis and risk stratification, especially for ATTRwt amyloidosis, by considering associated risk factors.