Abstract
Higher exacerbation risk in COPD ex-smokers could reflect higher prior exacerbation history http://ow.ly/TbwP30fJNGj
To the Editor:
We read with interest the findings of Kerkhof et al. [1], demonstrating an increased exacerbation risk in eosinophilic patients (eosinophil counts ≥0.45×109·L–1) with chronic obstructive pulmonary disease (COPD). When compared to a normal reference group (eosinophil counts of between 0.05 and 0.45×109·L–1), the rate ratios (RR) for exacerbations in the following year were lower in current smokers (RR=0.86) than in ex-smokers (RR=1.32). This finding is counter-intuitive and contrasts with data from large studies that demonstrate exacerbation reductions in COPD patients who cease smoking [2, 3]. A suppressive effect of active tobacco smoke on eosinophils is proposed as an explanatory mechanism by the authors; however, we feel the impact of potential confounders on this study finding should first be excluded prior to accepting this rationale.
The ECLIPSE study demonstrated that the strongest predictor of future exacerbations is prior exacerbations [4]. Despite this, data regarding the number of exacerbations in the preceding year were not reported or controlled for as a potential confounder in the predictive modelling. Why might this be important? Because the higher exacerbation rate in the ex-smoker subgroup could reflect a higher number of prior exacerbations (or a greater proportion of “frequent exacerbator” phenotypes). Why might this be relevant to ex-smokers? Because severe, potentially life-threatening COPD exacerbations can be potent stimuli for smoking cessation, thus increasing the likelihood of ex-smokers having a significant exacerbation history. It is therefore plausible that Kerkhof et al. [1] have reported upon eosinophilic ex-smokers with a history of prior exacerbations. If this is true, both prior exacerbations and high eosinophil counts may have predisposed subjects to the observed increase in exacerbation risk, with smoking status representing a simple proxy for prior exacerbations.
A number of other factors are potentially problematic. The data originated from a large UK-based national databank derived primarily from primary care. While findings from such “real life” data might differ from those derived from “typical” patients recruited into clinical trials (as acknowledged by the authors), one must also consider the inherent potential for error associated with retrofitting the authors' specific research question to such a data source. There is potential for unintended recruitment bias related to the availability and completeness of outcome data. There is also potential for eosinophil group misclassification due to the presence of unknown co-existing acute inflammatory illness or oral corticosteroid use (both likely to suppress eosinophil counts below subjects' usual levels). Finally, there is potential for inaccuracy in exacerbation detection due to inconsistent definitions (e.g. antibiotic/acute corticosteroid prescription versus hospital/emergency department presentations). This is particularly important as the criteria for defining exacerbations are known to vary greatly in terms of their sensitivity to detect exacerbations of differing severity [5]. Sensitivity analyses related to exacerbation definition would be helpful to address this concern.
Tailored therapies can improve clinical outcomes for COPD patients who have distinct clinical phenotypes, and blood eosinophils have emerged as a simple, attractive potential stratification metric for this purpose. The findings of Kerkhof et al. [1] offer interesting insights into the possible future priority areas for clinical research in this field. However, careful consideration and prospective evaluation are first required in order to confirm the interactions between eosinophils, COPD and other clinically impactful variables.
Disclosures
Supplementary Material
C.R. Osadnik ERJ-01820-2017_Osadnik
Footnotes
Conflict of interest: Disclosures can be found alongside this article at erj.ersjournals.com
- Received September 5, 2017.
- Accepted September 6, 2017.
- Copyright ©ERS 2017