Introduction
Chronic obstructive pulmonary disease (COPD) is currently the third leading cause of death and disability worldwide [
1‐
3]. Exacerbation is the major reason for hospital admission of patients with COPD and related to a significantly worse survival outcome [
4‐
6]. Inhaled corticosteroids (ICSs) or combined with long-acting bronchodilators have been recommended to treat COPD patients with repeated exacerbations [
1].
Although ICSs are generally considered to be relatively safe and well tolerated in patients, some adverse effects associated with ICSs have also been observed, such as the development of oropharyngeal candidiasis [
6], adrenal suppression [
7], diabetes [
8], and pneumonia [
9]. However, the association between ICSs and risk of upper respiratory tract infection (URTI) remains unclear, though URTI is the most common respiratory infection and also an important cause of exacerbation of COPD [
10]. The large prospective study Toward a Revolution in COPD Health (TORCH) trial reported ICSs might increase the morbidity of URTI in COPD patients [
11]. Moreover, other randomized controlled trials (RCTs) reported different or even contrary outcomes, and most of these studies were inadequate to detect significant difference between ICSs treatment groups and control groups [
12‐
17].
Whether ICSs increase the risk of URTI in COPD patients may depend on duration, dosage and type of ICSs. Lack of safety evidence may result in insufficient use or over use of ICSs. Therefore, we conducted this meta-analysis of RCTs to assess the association between ICSs use and the risk of URTI in patients with COPD. We also aimed to clarify the contributions of medication details for the association, including duration, dosage level and type of ICSs.
Discussion
This meta-analysis of 17 multicenter RCTs (including 20,478 patients) suggested that short-term use of ICSs increased the risk of URTI but not for long-term use. Futher subgroup analyses revealed that only short-term use of high-dose fluticasone increased the risk of URTI but not for long-term use of high-dose fluticasone. Medium-dose and low-dose fluticasone did not increase the risk of URTI regardless of duration. Neither mometasone nor budesonide increased the risk of URTI, regardless of dosage or duration.
Exacerbation is common in patients with COPD affecting about 20% of patients with 40–45% of predicted FEV
1 (1.3 events per year). Repeated exacerbations lead to worse survival outcome of patients [
5]. Daily use of ICSs has been proved to decrease the frequency of exacerbations and improve quality of life in patients with FEV
1 less than 50% predicted [
1,
4]. However, daily use of ICSs may cause drug-related adverse events, such as increased risks of fracture and infections [
32,
33], but may not increase the risk of cardiovascular events [
34]. URTI is the most common respiratory infections and also an important cause of exacerbation of COPD. Moreover, URTI can significantly reduce the quality of life in patients [
10]. However, the association between ICSs and the risk of URTI remains unclear.
Our results suggested that use of ICSs was associated with a significantly increased the risk of URTI in COPD patients. However, further subgroup analyses suggested that only short-term use of ICSs significantly increased the risk of URTI but not for long-term use of ICSs in COPD patients. The result was unexpected. We speculated that short-term use of ICSs could not effectively control the airway inflammation, whereas the immunosuppression effects of corticosteroids due to local high concentration might cause susceptibility to URTI in patients with COPD. On the contrary, long-term use of ICSs may improve health status of COPD patients by decreasing the frequency of exacerbations and improving clinical symptoms. In addition, the immune compensation mechanism of patients may play a role in resisting the immunosuppression of corticosteroid in the long term. A previous RCT conducted by Eichenhorn et al. may support our findings [
35]. In their study, Eichenhorn et al. evaluated the effects of inhaled triamcinolone on adrenal function in 221 patients who were recruited from patients already enrolled in the Lung Health Study II, and found that use of triamcinolone (1200 μg daily) for 3 years did not have suppression effects on adrenal function [
35,
36].
Subgroup analyses were performed according to different dosage of ICSs. The results suggested that high-dose ICSs significantly increased the risk of URTI but not for medium- and low-dose ICSs. These findings may be explained by the possible dose-response effect of ICSs [
7,
37]. In the further analysis, we found that only short-term use of high-dose ICSs was associated with a significantly increased the risk of URTI but not for long-term use of high-dose ICSs. The results further supported that long-term use of ICSs did not increase ICS-related risk of URTI.
We also conducted subgroup analyses according to type of ICSs. The results suggested that high-dose fluticasone was associated with a significantly increased risk of URTI but not for mometasone, budesonide, and medium- and low-dose fluticasone. Previous studies have reported that fluticasone may effectively suppress innate immunoresponse of alveolar macrophages, and the immunosuppressive effects of fluticasone might be tenfold greater than those of budesonide in human airways [
38,
39]. Moreover, the included RCTs mainly explored medium- or low-dose budesonid rather than high-dose. Similarly, we also found that short-term use of high-dose fluticasone was associated with a significantly increased the risk of URTI but not for long-term use of high-dose fluticasone. The results were consistent with the above results of the merged different types of ICSs.
Our results were not consistent with a previous meta-analysis conducted by Yang et al. in 2016 [
40], which first reported that long-term use of ICSs may increase the risk of URTI in patients with COPD. However, that meta-analysis had limitations mainly because it did not include several large multicenter RCTs (including 3507 patients) published recently [
14,
15,
17,
28,
31]. The smaller sample size of the previous meta-analysis may weaken the reliability and generalizability of the conclusion. Moreover, the previous meta-analysis failed to provide implication for clinical practice due to lack of subgroup analyses according to medication details including duration, dosage level and type of ICSs.
The strengths of our study were that we used a comprehensive search strategy and explicit inclusion criteria including 17 multicenter RCTs (20,478 patients), which met the requirements of sequential analysis. In addition, we conducted multiple subgroup analyses according to medication details including duration, dosage, and type of ICSs, that minimized the heterogeneity of pooled analyses and provided implications for clinical practice.
This study also had some limitations which mainly stemed from the quality of reported data. First, since some adverse events (such as URTI) were not the predefined outcomes and there were no homogeneous definitions among the clinical trials, these adverse events may be misclassified. This inherent methodological defect of clinical trials is one of the factors limiting the results of meta-analyses on drug safety. However, as most of the included trials in this meta-analysis were double blind, such misclassification would not have a substantial impact on the results of the meta-analysis, because the direction of bias may be toward the null. Second, the possibility of asthma-COPD overlap (ACO) in the study population may be one of the confounding factor of results. However, as a result of randomization, this confounding factor could not substantially affect the results of the meta-analysis as ACO may be approximately evenly distributed between the ICSs treatment group and the control group. Third, further subgroup analyses could not be performed due to lack of some patient related information, such as eosinophil count, exacerbations in the past year, comorbidities, smoking status and preexisting ICSs use and this may be a confounding factor of results.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.