Echocardiographic phenotypes in sepsis: identifying subgroups using latent profile analysis.

Area of Science:

• Critical Care Medicine
• Cardiology
• Data Science

Background:

• Sepsis is a major cause of death, with treatment complicated by patient variability.
• Identifying cardiac phenotypes in sepsis can personalize treatment and improve resuscitation strategies.
• Current approaches often lack objective stratification of cardiac dysfunction.

Purpose of the Study:

• To classify sepsis patients into distinct cardiac phenotypes using unsupervised clustering of echocardiographic data.
• To identify potential differences in clinical characteristics and outcomes among these phenotypes.
• To establish objective subgroups for precision medicine in sepsis management.

Main Methods:

• Retrospective cohort study of 2,071 sepsis patients meeting Sepsis-3 criteria.
• Echocardiographic data analyzed using latent profile analysis to identify clusters.
• Exclusion of patients with pre-existing heart disease; comparison of clinical features and outcomes across identified clusters.

Main Results:

• Five distinct cardiac phenotypes were identified: diastolic dysfunction, low ejection fraction/cardiac index, right ventricular dilation, hyperdynamic function, and normal.
• The right ventricular dilation phenotype exhibited significantly higher mortality (36.9%) compared to the normal group (19.6%).
• Patients with right ventricular dilation had a 2.3-fold increased odds of mortality.

Conclusions:

• Unsupervised clustering of echocardiographic data reveals five distinct cardiac phenotypes in sepsis.
• Right ventricular dilation represents a high-mortality phenotype requiring further investigation.
• These identified phenotypes may guide future mechanistic research and targeted therapeutic interventions in sepsis.