Background

Chronic obstructive pulmonary disease (COPD) is a progressive inflammatory airway disease characterized by reduced airflow, and is a leading cause of mortality worldwide [1].

COPD is regarded as heterogeneous condition and chronic airway inflammation in COPD is thought to be driven by neutrophils and lymphocytes [2]. This type of airway inflammation is a prominent feature in most COPD patients unlike in asthmatic patients, and related to the degree of airflow limitation [3].

Airway inflammation in asthma is mediated by type-2 T helper (Th2) cells and eosinophils, associated with an exacerbation risk and poor disease control following inhaled corticosteroid (ICS) withdrawal [4, 5]. The presence of eosinophils has been reported in the blood or sputum of 30–40% COPD patients [6]. Excluding patients with clinical features suggestive of asthma, such as bronchial hyperresponsiveness, history of asthma, atopy, and reversible airflow limitation, a subset of COPD patients demonstrates eosinophilic airway inflammation and responds to corticosteroid therapy [7,8,9].

Several studies have reported a greater risk for COPD exacerbation with higher blood counts of eosinophils, although the relationship remains controversial. In the general population, spirometry-based COPD patients experience more exacerbations with high blood eosinophil counts [10, 11]. Although higher eosinophil levels increase the risk for COPD exacerbation, they are also associated with a greater risk reduction after the use of ICS [12,13,14].

However, few studies have investigated the effects of blood levels of eosinophils on lung function. In this study, we evaluated longitudinal changes in forced expiratory volume in 1 s (FEV1) in patients with eosinophilic COPD, and analyzed the effects of ICS on this parameter.

Methods

Study population

Data were obtained from the Korean COPD Subgroup Study (KOCOSS; 2012–2019), an ongoing prospective multicenter observational cohort, which has recruited COPD patients from 48 referral hospitals in the Republic of Korea [15]. The inclusion criteria were (1) age ≥ 40 years; and (2) COPD diagnosed by pulmonologists based on respiratory symptoms and spirometry-confirmed fixed airflow limitation (post-bronchodilator FEV1/forced vital capacity (FVC) < 0.70). Only patients with blood eosinophil count data at baseline enrollment during stable state and followed up at least 3 years from enrollment were included in the analyses.

Clinical data

KOCOSS cohort contained detailed information regarding sociodemographics (including age, sex, smoking history, and education level), symptoms (chronic cough or persistent phlegm ˃ 3 months), dyspnea severity based on the Modified Medical Research Council Dyspnea Scale (mMRC), and quality of life based on the COPD assessment test (CAT) and the St George’s Respiratory Questionnaire (SGRQ). Data on comorbidities, including cardiovascular disease, tuberculosis, and asthma, were also collected. The treatment of the subjects depended on their attending pulmonologist.

The KOCOSS cohort patients were followed-up at least every 6 months, and moderate and severe exacerbations were recorded at each visit. Moderate exacerbation was defined as an exacerbation leading to an outpatient-clinic visit earlier than scheduled and prescribed systemic steroids and/or antibiotics, whereas severe exacerbation was defined as leading to an emergency department visit or hospitalization.

Pulmonary function test

Spirometry was performed based on American Thoracic Society and European Respiratory Society guidelines [16]. Absolute FEV1 values were obtained, and the predicted percentage values (% pred) for FEV1 were calculated using an equation developed for the Korean population [17]. Positive bronchodilator response (BDR) was defined as post-bronchodilator FEV1 increase in 12% or more, and 200 mL from baseline. Subjects were followed-up at least every 6 months, and spirometry was performed annually.

Statistical analysis

Baseline characteristics were expressed as means ± standard deviations and absolute numbers with percentages. Continuous and categorical variables were compared using t-test and chi-square test, respectively.

In this study, we assess the longitudinal changes in post-bronchodilator FEV1 (mL). To exclude the immediate effects of bronchodilator treatment, lung function data used for longitudinal analysis were collected one year after enrollment.

Although variability on serial blood eosinophil counts may be concernable, over 300 cells/µL is regarded as threshold for predicting increased risk of exacerbation and high likelihood of benefit with ICS [18, 19]. We compared the longitudinal FEV1 change between eosinophilic (blood eosinophil count ≥ 300 cells/μL) and non-eosinophilic (blood eosinophil count < 300 cells/μL) COPD, using a linear mixed-effects model with random intercepts and slopes. We analyzed mean annual rates of FEV1 change based on multiple covariates: age, sex, smoking (pack-years), and body mass index (BMI). Then, we analyzed FEV1 changes based on ICS usage.

All tests were two-sided and a p-value less than.05 was considered statistically significant. Statistical analyses were performed using Stata (v. 16; StataCorp, College Station, TX, USA).

Results

Study subjects

A total of 2,181 COPD patients were enrolled between January 2012 and December 2019. FEV1 change analysis based on blood counts of eosinophils was available for 627 patients (Fig. 1). Comparison of clinical features between eosinophilic and non-eosinophilic COPD is shown in Table 1. Eosinophilic COPD was more prevalent in males (98.7% and 91.2%, for eosinophilic and non-eosinophilic groups, respectively). There were no significant differences in age, smoking status, symptoms, and quality of life. Past exacerbations were more common in eosinophilic COPD group, compared to non-eaosinophilic group (25.3% and 18.5%, respectively), but the difference was not statistically significant.

Fig. 1
figure 1

Study flow. COPD, chronic obstructive pulmonary disease; KOCOSS, Korean COPD subgroup study; FEV1, forced expiratory volume in 1 s; FVC, forced vital capacity;

Full size image
Table 1 Baseline characteristics of eosinophilic and non-eosinophilic COPD
Full size table

Baseline pulmonary function test results are shown in Table 2. More than 60% of the patients in both groups were GOLD stage 1 or 2 (FEV1 ≥ 50% of predicted value), and there were no significant differences in the severity of airflow limitation and exercise capacity. Mean blood counts of eosinophils were higher in the eosinophilic COPD group (546.1 vs. 141.8 cells/μL).

Table 2 Lung function parameters and inhaler prescription status
Full size table

ICS therapy, either alone or in combination with long-acting β2-agonists (LABA) and/or long-acting muscarinic antagonists (LAMA), was prescribed in 41.1% and 40.7% in non-eosinophilic and eosinophilic groups, respectively.

Exacerbation risk

Exacerbations were identified in 414 of the 627 patients. Moderate to severe exacerbations occurred in 71.0% of the patients (294/414), and 20.8% of the patients (86/414) experienced frequent exacerbations (≥ 2 events/year).

Exacerbations were occurred in 76.6% (82/107) of eosinophilic COPD group, while in 69.1% (212/307) of non-eosinophilic group. The exacerbation risk was significantly higher in the eosinophilic group compared to the non-eosinophilic group (adjusted odds ratio [aOR]: 1.49; 95% confidence interval [CI]: 1.10–2.03; p < 0.05). However, the statistical significance was lost with adding ICS therapy as covariate (aOR: 1.63; 95% CI: 0.89–2.99) (Table 3).

Table 3 The risk of exacerbation in patients with eosinophilic COPD compared with non-eosinophilic COPD
Full size table

Longitudinal FEV 1 change

A greater FEV1 decline rate was observed in the non-eosinophilic group, compared to the eosinophilic group (− 19.6 mL/y and − 12.9 mL/y, respectively) (Fig. 2). The adjusted annual FEV1 declined significantly in the non-eosinophilic COPD group (− 19.4 mL/y). Although the adjusted annual FEV1 also decreased in the eosinophilic COPD group, the rate of decline was not significant (− 12.2 mL/y) (Table 4). In the eosinophilic COPD group, the adjusted annual FEV1 decline rate was not significant regardless of ICS therapy, but the decline rate was greater in ICS users (− 19.2 mL/y and − 4.5 mL/y, with and without ICS therapy, respectively) (Table 5). The exact values of FEV1 changes during two years of follow-up between eosinophilic and non-eosinophilic COPD are shown in Additional file 1: Figure S1.

Fig. 2
figure 2

Changes of FEV1 over follow-up period in patients with eosinophilic and non-eosinophilic COPD. COPD, chronic obstructive pulmonary disease; FEV1, forced expiratory volume in 1 s

Full size image
Table 4 Annual FEV1 decline rate in eosinophilic and non-eosinophilic COPD
Full size table
Table 5 Annual FEV1 decline rate in eosinophilic COPD according to the use of ICS
Full size table

Discussion

There is limited data regarding the effects of ICS on FEV1 change in COPD. We hypothesized that the rate of FEV1 change differed based on the baseline blood eosinophil counts and ICS therapy. The adjusted annual rate of FEV1 change declined significantly in the low-blood-eosinophil group, compared to the high-blood-eosinophil group, although the difference was not statistically significant. Analyses of ICS effects on FEV1 change in high-blood eosinophil group showed no statistically significant differences.

It is widely accepted that COPD develops because of a genetic susceptibility for progressive airway inflammation, triggered by complex environmental factors over time, such as smoking, pollutants, and allergens, which leads to irreversible damage and airflow obstruction [3, 20]. Typically, airways of COPD patients exhibit increased CD8 + T cells and neutrophils. Although Th2 inflammatory pathway is considered a feature of asthma rather than COPD, 30–40% of COPD patients demonstrated eosinophilic-inflammation [6]. These heterogeneity of disease are now accepted as various clinical phenotypes with different pathophysiologies and endotypes. Therefore, response to therapies might also be variable. The ECLIPSE (Evaluation of COPD Longitudinally to Identify Predictive Surrogate End points) cohort study reported that 37.4% of COPD patients had persistently high (≥ 2%), 13.6% had persistently low (< 2%), and 49% had variable eosinophil counts over a 3-year follow-up period [6]. A subset of COPD patients demonstrate eosinophilic inflammation either in stable period or during acute exacerbations. Blood eosinophilia was associated with increased exacerbation risk, and this phenotype could benefit from corticosteroid therapy directed against eosinophilic inflammation [13, 14, 21,22,23]. A recent trial targeted eosinophilic inflammation (IL-5) to reduce exacerbation risk in eosinophilic COPD patients, suggesting that selectively eosinophilic-pathway blockers may be effective for such patients [24]. Elevated peripheral blood eosinophil counts in COPD may be used to identify patients who are expected to have favorable response to ICS therapy or biologics targeting Th2 inflammatory pathway. Conversely, ICS withdrawal in eosinophilic COPD leads to increases exacerbation risk [25]. Following ICS withdrawal, trough FEV1 decreased and exacerbation risk increased only in the high eosinophil group [26].

Most trials assess the impact of ICS on COPD using exacerbation risk as the outcome, therefore data regarding ICS effects on FEV1 change is limited. A post hoc study reported favorable effects of extrafine beclomethasone dipropionate and formoterol fumarate, compared to formoterol fumarate alone, on FEV1 change in the highest blood eosinophil quartile (≥ 279.8 cells/μL). In this trial, severe COPD patients with FEV1 of 30–50% of predicted value, and at least one exacerbation in the previous year were included [27].

A study on effects of ICS monotherapy on disease progression in moderate to severe COPD reported no differences in the rate of decline of post-bronchodilator FEV1 between fluticasone propionate and placebo over 3 years. However, when analyzed using baseline blood eosinophil levels, the rate of FEV1 decline was significantly lower in the fluticasone propionate group compared to placebo in the ≥ 2% eosinophil group (− 40.6 mL/y and − 74.5 mL/y, respectively; p = 0.003)[22].

In the present study, annual rate of FEV1 change declined significantly in the lower blood eosinophil group. Analyses of ICS effects on longitudinal FEV1 change in high-blood-eosinophil group revealed a greater decline rate in ICS users. Although we demonstrated increased exacerbation risk in the eosinophilic COPD group, consistent with previous studies [10,11,12], the annual FEV1 decline rate was low in eosinophilic COPD compared to non-eosinophilic COPD, and ICS provided no benefits. Unlike previous trials [22, 27], only 20% of subjects had previous exacerbations and approximately 66% had mild to moderate FEV1 severities in the KOCOSS cohort. Different airflow obstruction severities and previous exacerbation history may have caused the differences in FEV1 change.

In this study, we used blood eosinophil count as a marker of treatment response to ICS in COPD cohort. In consideration of patients with asthma-COPD overlap (ACO) high probability, we defined ACO by eosinophil count of cells/μL and/or extreme BDR criteria (> 15% and 400 mL). More than 98% were classified according to the eosinophil criteria. Absence of unified diagnostic criteria for ACO resulted in inconsistent clinical manifestations and outcomes. Moreover, in our previous study, we showed that blood eosinophil was the single most important biomarker to predict the decrease of exacerbation by ICS [28]. Recently, interests in precision medicine according to the subtype of COPD, rather than ACO is increasing. Blood eosinophil count is considered as the reliable biomarker in identifying subtype that might be benefitted by ICS. Although we did not find meaningful relation between FEV1 change and ICS use in eosinophilic COPD in real world data, additional prospective studies will be needed.

There were several limitations in this study. First, we analyzed subsequent 2 years of FEV1 data, which is a relatively short follow-up period. Second, we defined eosinophilic COPD using blood eosinophil counts at a stable period, and variation over time were not reflected. Third, COPD patients in our study were enrolled irrespective of treatment- naïve status. Whether blood eosinophil count less than or over 300cell/μL, the prescription rate for ICS containing inhalers was similar, which implies considerable numbers of subjects who have already been exposed to ICS are included, and this might be related to attenuation of ICS effects on eosinophilic COPD. Fourth, this was not a randomized clinical trial. Although the prescription rate for ICS containing inhalers was similar in both groups (41.1% and 40.7% in non-eosinophilic and eosinophilic groups, respectively.), ICS containing inhalers was not randomly assigned. ICS prescription was decided by pulmonologists. Thus, there can be a bias for the prescription of ICS. Caution is needed to interpret the impact of ICS on eosinophilic COPD. Fifth, changes in FEV1 according to GOLD classification could not be performed because approximately 90% of subjects were in GOLD group A or B. Lastly, we aimed to assess the rate of decline of FEV1 according to blood eosinophil count and whether FEV1 decline differs between eosinophilic COPD with and without ICS. Whittaker et al. [29] reported ICS use is associated with slower rates of FEV1 decline in COPD regardless of blood eosinophil level in a cohort of 26,675 COPD patients. On the other hand, Celli et al. [30] reported ICS plus LABA combination or the component, reduces the rate of decline of FEV1 in moderate to severe COPD. This study included more than 1,000 patients in each of subgroups. However, our study is underpowered to address intended questions about rate of decline of FEV1 due to small number of subjects.

Conclusion

In conclusion, we found that the favorable longitudinal FEV1 change in eosinophilic COPD patients compared to non-eosinophilic COPD over 3 years in a Korean COPD cohort. ICS use did not have any beneficial effects on FEV1 change in eosinophilic COPD, but FEV1 decreased more rapidly with ICS use. Further studies with longer follow-up periods in larger number of patients are required.