Prediction of first acute exacerbation using COPD subtypes identified by cluster analysis

Area of Science:

• Pulmonology
• Respiratory Medicine
• Clinical Data Science

Background:

• Acute exacerbations (AE) significantly impact COPD prognosis and treatment.
• Identifying predictors for AE is crucial for effective COPD management.
• Current COPD classification may not fully capture AE risk.

Purpose of the Study:

• To evaluate the utility of COPD subtypes derived from cluster analysis in predicting the first AE.
• To identify distinct COPD patient groups based on clinical and functional parameters.
• To inform personalized treatment strategies for COPD patients at risk of AE.

Main Methods:

• Utilized K-means clustering on data from 1,195 COPD patients in the KOCOSS cohort.
• Included variables: age, BMI, smoking status, asthma history, CAT score, post-BD FEV1, and DLCO % predicted.
• Analyzed risk for first AE using Cox proportional hazards models.

Main Results:

• Four COPD subtypes were identified: putative asthma-COPD overlap (ACO), mild, moderate, and severe COPD.
• The ACO group exhibited better lung function and quality of life but a higher risk of first AE (HR 1.683) compared to mild COPD.
• Moderate and severe COPD groups also showed increased AE risk (HR 1.587 and 1.664, respectively).
• St. George's Respiratory Questionnaire score and GERD were independent predictors of first AE.

Conclusions:

• Cluster analysis identified a COPD exacerbator subtype independent of FEV1.
• This subtype is susceptible to AE, necessitating more aggressive therapeutic approaches.
• Subtyping COPD patients can refine risk stratification and guide treatment decisions.