O4.2.1 Eosinophil count and inhaled corticosteroids

There is significant interest in the use of blood eosinophil count as both a prognostic marker and to guide the use of inhaled corticosteroids in COPD.

In a US cohort study (Zeiger 2018), elevated blood eosinophils at baseline were independently associated with COPD exacerbations and COPD-related ED visits or hospitalisations during a year of follow-up. After adjusting for confounders, rate of future exacerbations were 25%, 48% and 76% greater for patients with eosinophils ≥ 300 cells/ mm³  ≥ 400 cells/ mm³ and  ≥ 500 cells/ mm³, respectively. Analysis of data from the COPD Gene and ECLIPSE longitudinal studies (Yun 2018) also found baseline blood eosinophils ≥ 300 cells/ mm³ to be associated with increased exacerbation frequency. In a large group of patients (n=7,180) from the Danish Copenhagen General Population Study (Vedel-Krogh 2018), blood eosinophils  ≥ 0.34 x109 cells/L in people whose FEV1 was < 50% predicted were associated with a higher risk of hospitalisation for pneumonia compared with those with the same degree of airflow obstruction but  a lower eosinophil count. In the Korean Obstructive Lung Disease cohort study, patients with COPD who had persistently high blood eosinophils (≥ 300 cells/ mm³) had a better survival rate and improved symptoms and quality of life than those with persistently low eosinophil counts (<300 cells/ mm³) while those with variable eosinophil counts had survival rates similar to those with persistently low counts (Shin 2018). In an Australian study by MacDonald et al (2019), low blood eosinophil counts (<50/uL) during admission for acute exacerbation of COPD were associated with bacterial infection, increased length of stay and a higher 12-month mortality, while just over half of exacerbations associated with higher eosinophil counts (>150/uL) also demonstrated evidence of infection, likely requiring antibiotic therapy (MacDonald 2019). A retrospective study from a single centre in China found no association between in hospital eosinophil count and in hospital mortality or length of stay, or exacerbation within one year of discharge (Yu 2021) [evidence level III-B].

Higher eosinophil counts have also been shown to be associated with a higher rate of lung function decline in individuals with and without COPD in the Canadian CANCOLD study, a prospective cohort study based on the Canadian COPD prevalence study (COLD). CANCOLD evaluated 6000 males and females ≥40 years, recruited through random sampling. The study included all subjects with COPD from the original COLD study and an equal number of age and sex-matched peers without COPD. A total of 1285 individuals had bloods drawn for eosinophil counts at 0 and 18 months as well as lung function tests, and high-resolution CT scans. Baseline eosinophil count of ≥ 300 cells/ul was an independent risk factor for accelerated decline in lung function in those with and without COPD, independent of exacerbations, and was related to the presence of gas trapping, airway wall thickening and reduction of total airway count base on CT (Tan 2021) [evidence level III-2].

In a post-hoc analysis of the FORWARD study, a double blind randomised controlled study which compared 48 weeks of treatment with extra fine beclomethasone dipropionate (BDP) plus formoterol furoate 100/6 ug two puffs bd with formoterol furoate (FF) 12 ug one puff bd in patients with COPD, patients with eosinophil counts ≥ 279.8 cells/μl experienced the highest exacerbation rate with FF and the greatest benefit from the BDP/FF combination (Siddiqui 2015). In a post-hoc review of data from WISDOM, patients with higher blood eosinophil counts were more likely to develop exacerbations after withdrawal of inhaled corticosteroids, with a significant treatment-by-subgroup interaction above an eosinophil count of 4% or greater or above 300 cells/μL (Watz 2016). Bafadhel et al used negative binomial regression analysis using splines to examine data from RCTs of budesonide/formoterol in patients with COPD, a history of exacerbations and available eosinophil counts (n=4,528) (Bafadhel 2018).  They found a treatment effect interaction between the budesonide-formoterol combination as compared with formoterol alone and eosinophil count, with respect to exacerbations, lung function and health status. At eosinophil counts of 100/μl or more, a significant treatment effect was found for exacerbation reduction with budesonide/formoterol compared with formoterol alone (RR 0.75, CI 0.57 to 0.99); p interaction =0.015).

Casanova et al examined the prevalence and stability of the finding of a blood eosinophil count ≥ 300 cells/μl and its relationship to outcomes over two years using Cox hazard analysis in patients from the CHAIN (patients with COPD and smokers without COPD) and BODE (patients with COPD only) cohorts (Casanova 2017).  15.8% of COPD patients in CHAIN and 12.3% of those in BODE had persistently elevated eosinophils during the period of follow-up (at least 3 measurements over two years). A similar eosinophil blood pattern was observed in controls. Exacerbation rates did not differ in patients with and without eosinophilia. All-cause mortality was lower in patients with high eosinophils compared with those with values <300 cells/μL–1 (15.8% versus 33.7%; p=0.026). In the SPIROMICS database of patients with COPD, smokers without COPD and 7% non-smokers, blood eosinophil count alone was not a reliable biomarker for COPD severity or exacerbations (Hastie 2017). Although there was a statistically significant relationship between blood and sputum eosinophils, blood eosinophil count did not reliably predict the level of sputum eosinophilia. Sputum eosinophils were available in a subset of just on 1,000 patients. The authors found that high sputum eosinophils, but not blood eosinophils, identified a subset of patients with more severe airflow obstruction, worse quality of life, more emphysema and gas trapping and more exacerbations. However, there were no differences in COPD Assessment Test (CAT) scores noted with either blood or sputum eosinophil stratification.  In the prospective GLUCOLD study of patients with COPD using ICS or placebo during 30 months of follow up, neither baseline blood eosinophil levels nor baseline eosinophil levels in sputum, bronchoalveolar lavage (BAL) or bronchial biopsy predicted longitudinal changes in FEV1 with or without ICS (Hartjes 2018).

Prospective studies that randomise patients based on eosinophil count are required to confirm these associations.

A meta-analysis by You et al (2020) compared outcomes of acute exacerbations of COPD (AECOPD) with and without eosinophilia (defined as  an eosinophil count  ≥2% or an absolute eosinophil count ≥0.34×109). Outcomes were better overall for eosinophilic AECOPD, with decreased hospital mortality (OR0.59, 95% CI 0.31-0.95,p=0.03), decreased length of stay ( OR=0.72,95% CI-1.44 to -0.00,p=0.05), higher FEV1 (mean difference =0.14, 95% CI 0.08-0.2, p<0.00001) and a lower risk of arrhythmias 9 (OR=1.5, 95% CI 1.01-2.21, p=0.04). It was noted that there were more males among the non-eosinophilic group (OR=1.34, 95% CI 1.15-1.56, p=0.0002), but that steroid use did not differ between the groups. The majority of studies in this meta-analysis were single centre and retrospective in design.