Heterogeneity in Kawasaki disease patients with coronary artery abnormalities investigated by data-driven cluster analysis.

BACKGROUND

To clarify the clinical characteristics associated with coronary artery abnormality (CAA) in the acute phase of Kawasaki disease (KD) cases, we performed a cluster analysis.

METHODS

A retrospective cohort study of 103 KD cases that developed CAA (maximum Z-score for any coronary arteries >2.5) within 30 days after disease onset was conducted among 726 consecutive KD cases. The hierarchical cluster analysis was performed based on 20 continuous variables before treatment.

RESULTS

101 out of 103 cases were clustered into 4 subgroups. Cluster 1 was a younger group characterized by leukocytosis and elevated serum triglyceride levels with frequent coronary artery dilation before treatment and good initial treatment responses. Cluster 2 was a younger group characterized by anemia and lower serum albumin and total cholesterol levels. Cluster 3 was an older group with frequent coronary artery dilation before treatment. Cluster 4 was an older group characterized by elevated serum aspartate aminotransferase, alanine aminotransferase, and total bilirubin levels, with poor treatment response and the lowest incidence of coronary artery dilatation before treatment.

CONCLUSION

We identified four subgroups with distinct features, suggesting different backgrounds for CAA development. Consideration of possible heterogeneity might be helpful for a better understanding of the pathophysiology and better treatment strategies.

IMPACT: To investigate possible heterogeneity in Kawasaki disease (KD) cases that developed coronary artery abnormality (CAA) in the acute phase and to evaluate associated clinical characteristics, we performed a cluster analysis in KD cases that developed CAA in the acute phase. Hierarchical cluster analysis in KD cases that developed CAA within 30 days after disease onset, based on clinical data before treatment, identified four subgroups with distinct features, suggesting different backgrounds for CAA development. Cluster analysis might be helpful for a better understanding of the pathophysiology and better treatment strategies in KD cases at high risk for developing CAAs.