Phenotypic Clustering in Non-Cystic Fibrosis Bronchiectasis Patients: The Role of Eosinophils in Disease Severity

Abstract

Whether high blood eosinophil counts may define a better phenotype in bronchiectasis patients, as shown in chronic obstructive pulmonary disease (COPD), remains to be investigated. Differential phenotypic characteristics according to eosinophil counts were assessed using a biostatistical approach in a large cohort study from the Spanish Online Bronchiectasis Registry (RIBRON). The 906 patients who met the inclusion criteria were clustered into two groups on the basis of their eosinophil levels. The potential differences according to the bronchiectasis severity index (BSI) score between two groups (Mann-Whitney U test and eosinophil count threshold: 100 cells/µL) showed the most balanced cluster sizes: above-threshold and below-threshold groups. Patients above the threshold exhibited significantly better clinical outcomes, lung function, and nutritional status, while showing lower systemic inflammation levels. The proportion of patients with mild disease was higher in the above-threshold group, while the below-threshold patients were more severe. Two distinct clinical phenotypes of stable patients with non-cystic fibrosis (CF) bronchiectasis of a wide range of disease severity were established on the basis of blood eosinophil counts using a biostatistical approach. Patients classified within the above-threshold cluster were those exhibiting a mild disease, significantly better clinical outcomes, lung function, and nutritional status while showing lower systemic inflammatory levels. These results will contribute to better characterizing bronchiectasis patients into phenotypic profiles with their clinical implications.

Keywords: biostatistical analyses; clinical outcomes; disease severity scores; eosinophil counts; multivariate analyses; non-cystic fibrosis bronchiectasis; phenotypic clusters.

Figure 1

Flowchart of the study patients.

Figure 2

Selection of the eosinophil threshold for patient clustering. (A) Number of patients in the below-threshold cluster for increasing the eosinophil thresholds. The point indicated by the black arrow corresponds to the percentage of the patients’ threshold in the below-threshold, which led to the most balanced cluster sizes. (B) Statistical significance resulting from the Mann–Whitley U test as a measure by the corresponding p-value for varying thresholds (in terms of the percentage of patients belonging to the below-threshold cluster). The horizontal solid red line shows the statistical significance threshold p = 0.05, while the vertical yellow dashed line shows the optimal threshold of the eosinophil counts corresponding to the p-value.

Figure 3

Histograms of the proportions of patients who were classified as mild–moderate–severe according to EFACED (A), BSI (B), and FACED (C) between the two clusters. EFACED: mild: 0–3, moderate: 4–6, and severe: 7–9; BSI: mild: 0–4, moderate: 5–8, and severe: ≥ 9; and FACED: mild: 0–2, moderate: 3 to 4, and severe: 5–7. Statistical significance: **, p < 0.01 between the patient groups.

Figure 4

Correlation matrix of the disease severity and analytical variables, in which the positive correlations are represented in blue, while the negative correlations are represented in red: (A) all the study patients, (B) patients in the below-threshold cluster, and (C) patients in the above-threshold cluster. The intersection within the circle represents a p-value > 0.05. The color intensity and the size of the circle are proportional to the correlation coefficients, as indicated in the Y-axis on the right-hand side of the graph.

Figure 5

The independent association of dichotomized eosinophils with severe BSI. A multivariate regression analysis was used to perform the analysis. The confidence intervals and statistical significance are represented in the figure panel. The odds ratio is represented as black dots. The model was adjusted by chronic colonization by the PA, Charlson Index, ESR, total number of leukocytes, total number of neutrophils, fibrinogen, hemoglobin levels, hematocrit levels, protein, albumin, and CRP levels.