Blood neutrophil counts are associated with exacerbation frequency and mortality in COPD

Abstract

Background: Identifying patients with COPD at increased risk of poor outcomes is challenging due to disease heterogeneity. Potential biomarkers need to be readily available in real-life clinical practice. Blood eosinophil counts are widely studied but few studies have examined the prognostic value of blood neutrophil counts (BNC).

Methods: In a large population-based COPD registry in the East of Scotland (TARDIS: Tayside Allergic and Respiratory Disease Information System), BNC were compared to measures of disease severity and mortality for up to 15 years follow-up. Potential mechanisms of disease modification by BNC were explored in a nested microbiome substudy.

Results: 178,120 neutrophil counts were obtained from 7220 people (mean follow up 9 years) during stable disease periods. Median BNC was 5200cells/μL (IQR 4000-7000cells/μL). Mortality rates among the 34% of patients with elevated BNCs (defined as 6000-15000cells/μL) at the study start were 80% higher (14.0/100 person years v 7.8/100py, P < 0.001) than those with BNC in the normal range (2000-6000cells/μL). People with elevated BNC were more likely to be classified as GOLD D (46% v 33% P < 0.001), have more exacerbations (mean 2.3 v 1.3/year, P < 0.001), and were more likely to have severe exacerbations (13% vs. 5%, P < 0.001) in the following year. Eosinophil counts were much less predictive of these outcomes. In a sub-cohort (N = 276), patients with elevated BNC had increased relative abundance of Proteobacteria and reduced microbiome diversity.

Conclusion: High BNC may provide a useful indicator of risk of exacerbations and mortality in COPD patients.

Keywords: COPD; Eosinophil; Microbiome; Mortality; Neutrophil.

Fig. 1

Flow chart of numbers of patients and samples in TARDIS, and each set of analyses

Fig. 2

a: The proportion of individuals recorded as having died at the end of follow up (black) plotted against their index blood eosinophil count. The red symbols are proportions (out of all individuals) with death from a cause recorded as falling within ICD10 category J44 (COPD). b: Distribution of blood neutrophil counts (BNCs) showing all values (hollow bars); all measurements from individuals with stable and exacerbating results available (black bars) and measurements made during exacerbations (red bars). The black dots show the proportion of BNCs made outside of exacerbation periods. Three vertical dotted lines indicate the cut-offs applied to separate BNCs into normal (<6000cells/μL), elevated (6000–15,000cells/μL) and extreme (> 15,000cells/μL) c: The proportion of individuals recorded as having died by the end of follow up (black) plotted against their index BNC. The red symbols are proportions (out of all individuals) with death from a cause recorded as falling within ICD10 category J44. Three vertical dotted lines indicate the cut-offs applied to separate BNCs into low (<2000cells/μL), normal (2000–6000cells/μL) and elevated 6000–15,000cells/μL) d: as C, but using the last BNC recorded for each individual. Solid points are based on final stable BNC, while hollow ones include BNC recorded during exacerbations

Fig. 3

a: Kaplan Meier survival curves, with 95% confidence intervals, showing all-cause mortality for the normal (black) and elevated (red) BNC groups adjusted for age, sex, and smoking status. b: As A, but representing mortality recorded as ICD10 J44, with all other mortality as a competing risk. c: Ratios of the numbers of all, and severe, exacerbations over the year from the index date from both unadjusted models and models adjusted for age, sex and smoking status. All are relative to the normal BNC group. d: Rates of change of FEV₁, estimated by fitting mixed models to either all data or only that obtained in the three years from each individual’s index date. For clarity, the low and extreme BNC groups are not shown here, though equivalent plots showing all four BNC groups are shown in supplementary e-Figure 6

Fig. 4

Comparisons of BNCs to a: percentage Proteobacteria in the microbiome from n = 246 stable COPD patients. b: SWDI, a measure of the microbiome diversity. c: Bray Curtis dissimilarity index and Weighted UNIFRAC were used to assess the differences in the microbiome between pairs of patients with the same BNC (<6000BNCs/μL or > 6000BNCs/μL) and between those pairs of patients with different BNCs (<6000BNCs/μL v > 6000BNCs/μL). d: PCoA plot showing the relationship between samples according to BNC