Usefulness of routine blood test-driven clusters for predicting acute exacerbation in patients with asthma

Abstract

Aim: Acute exacerbation (AE) is a significant burden in the management of asthma. In this study we aimed to investigate whether routine blood test results predicted AE in asthmatics.

Methods: We applied k-means cluster to routine blood test results which included eosinophil counts, total calcium, phosphorus, uric acid (UA), total cholesterol, total protein, albumin, total bilirubin, alkaline phosphatase, aspartate transaminase (AST), alanine transferase (ALT), gamma-glutamyltransferase, blood urea nitrogen, creatinine, and high-sensitive C-reactive protein (hsCRP) obtained from 590 asthmatics. AEs collected over the prospective follow-up of one-year were used to evaluate clinical trajectories of these clusters.

Results: Three blood clusters were identified. The essential features of each cluster can be characterized as follows: (i) high eosinophil count, UA, total cholesterol, AST, ALT, and hsCRP levels (Cluster 1); (ii) intermediate features (Cluster 2); (iii) low UA, total cholesterol and total bilirubin levels (Cluster 3). Kaplan-Meier analysis confirmed that clusters were strongly predictive of time to the first AE (log-rank P = 0.001). Hazard ratio for each group was as follows: Cluster 2 = 1, Cluster 1 = 2.67 (1 vs. 2, P = 4.68 × 10^-4), and Cluster 3 = 1.69 (2 vs. 3, P = 0.021).

Conclusions: We defined three blood clusters in asthmatics. These blood clusters are easily identifiable from routine test results and may help clinicians to predict the future risk of AE in asthmatics.

Keywords: Asthma; Cluster analysis; Hematologic tests; Symptom flare up.