Abstract
In patients with combined pulmonary fibrosis and emphysema, emphysema and fibrosis do not have a synergistic effect that results in worsened survival when compared to IPF patients without emphysema https://bit.ly/35EJMo6
To the Editor:
Emphysema is one of the most common pulmonary comorbidities of idiopathic pulmonary fibrosis (IPF), presenting in about one-third of IPF patients [1]. The term combined pulmonary fibrosis and emphysema (CPFE) has been used to describe a potential phenotype characterised by the coexistence of upper lobe-predominant emphysema, lower lobe-predominant fibrosis and relative preservation of lung volumes (forced vital capacity; FVC) in the context of a disproportionately reduced gas transfer (diffusing capacity of the lung for carbon monoxide; DLCO) [1–3]. With regard to patient survival, it remains unclear whether mortality in patients with CPFE reflects the cumulative effects of two disease processes (emphysema and fibrosis), or whether CPFE represents a distinct disease phenotype where outcome is worse than the sum of disease parts (emphysema and fibrosis).
In a previous single centre study [4], we demonstrated that the CPFE phenotype (defined as the presence of emphysema on computed tomography (CT)) in IPF patients did not independently predict mortality once summed visual lobar CT extents of emphysema and interstitial lung disease (ILD) had been considered. Put another way, survival in CPFE patients was the same as for IPF patients without emphysema, once the total extents of emphysema and fibrosis on CT were considered. The findings suggested that there was no additional synergistic impact on mortality when both disease patterns (emphysema and ILD) co-existed. Past analyses of CPFE populations have shown conflicting results regarding the impact of CPFE on mortality [3, 5–11] and may relate to heterogeneous study populations, varied CPFE inclusion criteria, and inconsistent adjustment for disease severity in mortality models [12]. The aim of the current study was to confirm our earlier study findings [4] that a CPFE phenotype has no independent mortality effect beyond that described by emphysema and ILD extent. The question was evaluated using various definitions of CPFE that have been considered in the literature, with results run on independent validation datasets.
We evaluated two separate cohorts of IPF patients diagnosed by a multidisciplinary team. Cohort 1: 220 patients from two centres in Turkey and Italy, 102 deaths observed; cohort 2: 310 patients from two centres in the Netherlands and England, and from the Australian IPF Registry, 169 deaths observed. CT extents of emphysema and ILD were separately scored, averaged across the lobes and then summed to develop a total lung percentage for both patterns as previously described [4]. We also performed a sub analysis in IPF patients that fulfilled drug trial inclusion criteria (DLCO >30% predicted and FVC >50% predicted) in cohort 1 (n=150, 57 deaths observed) and cohort 2 (n=239, 117 deaths observed). The median and interquartile ranges of emphysema extent were 4.17% and 11.67% in cohort 1; 2.92% and 8.33% in cohort 2. In the populations qualifying for drug trials: median and interquartile ranges of emphysema extent were 3.33% and 10.00% in cohort 1; 2.50% and 7.50% in cohort 2.
In each cohort, multivariable mixed-effects Cox regression models were used to evaluate whether the CPFE phenotype had any impact on outcome after considering the sum of visual CT extents of ILD and emphysema: VILDemph. To ensure that VILDemph and the various expressions of CPFE could be applied in the same model, we tested for collinearity using univariable linear regression. No strong collinearity was shown between VILDemph and the various expressions of CPFE (maximum R2=0.36). All mortality models were adjusted for patient age, sex, smoking status (never versus ever) and antifibrotic use (never versus ever). Models were repeated evaluating DLCO instead of VILDemph as a distinct functional measure of disease severity, thereby complimenting the models where a morphological measure of disease severity had been used (VILDemph). Different centres/countries within each cohort were modelled as multilevel with random effects between centres/countries (with a random intercept per centre/country). To encompass the breadth of published definitions of the CPFE phenotype [12], the CPFE phenotype was separately characterised as a binary emphysema variable in multivariable mixed-effects Cox regression models using four different emphysema thresholds (0%, 5%, 10% or 15% emphysema). The concordance index (C-index) was used to compare the predictive performance of the Cox models. Bootstrapping with 500 replications was used in the estimation of the C-index. p-values <0.05 were regarded as statistically significant. All analyses were implemented by R Studio.
Our results demonstrated that in both IPF cohorts, the CPFE phenotype did not independently predict mortality once summed extents of ILD and emphysema were considered in multivariable models (table 1). The results were maintained when patients fulfilling drug trial entry criteria were sub analysed in both cohorts. Results remained unchanged when the models examined baseline DLCO instead of VILDemph to adjust for disease severity. 181 (82%) out of 220 patients in cohort 1, and 266 (86%) out of 310 patients in cohort 2 had baseline DLCO values, whilst all patients in the drug trial population had baseline DLCO values.
Multivariable mixed-effects Cox proportional hazards regression models in two cohorts of idiopathic pulmonary fibrosis (IPF) patients
Our study confirms that mortality in patients with CPFE is explained by the sum of its two disease processes: the extents of fibrosis and emphysema. CPFE does not appear to manifest a malignant phenotype where survival is worse than that expected from the combination of two bad disease processes. Accordingly, once you consider emphysema and ILD patterns on CT, survival in CPFE is no different to survival in IPF patients without emphysema. The results were maintained when all of the different definitions of CPFE were separately analysed in both study cohorts and the smaller subsets of patients that would be included in drug trials.
A limitation of the study by Jacob et al. [4] was that the extent of emphysema in the cohort was relatively limited, with 11% (30 out of 272) patients having >15% emphysema extent (a threshold above which emphysema has been associated with significantly reduced FVC decline [13]). The proportion of patients with emphysema >15% was higher in the current study populations, 41 (19%) out of 220 patients in cohort 1 and 39 (13%) out of 310 patients in cohort 2. There were also very few patients in whom emphysema extent was greater than fibrosis extent. Only five (2%) out of 220 patients in cohort 1 and 13 (4%) out of 310 patients had more emphysema than fibrosis. A recent CPFE study considered patients in whom emphysema was more extensive than ILD on CT [14]. Repeating our analyses with a CPFE population defined in this way would be important to confirm our findings. Yet powering such a study in IPF patients will be extremely challenging.
In summary, we have validated findings across independent datasets confirming that in CPFE patients mortality is explained by the sum of emphysema and fibrosis extents. We have demonstrated that in CPFE patients, emphysema and fibrosis do not have a synergistic effect resulting in a malignant disease phenotype. CPFE patients and IPF patients without emphysema have indistinguishable mortality once the extents of emphysema and ILD on CT have been considered.
Footnotes
Submitted article, peer reviewed.
Conflict of interest: A. Zhao has nothing to disclose.
Conflict of interest: E. Gudmundsson has nothing to disclose.
Conflict of interest: N. Mogulkoc has nothing to disclose.
Conflict of interest: M.G. Jones has nothing to disclose.
Conflict of interest: C. van Moorsel has nothing to disclose.
Conflict of interest: T.J. Corte reports personal fees from Ad Alta, grants and personal fees from Boehringer Ingelheim and Bristol Myers Squibb, personal fees from Promedior, grants and personal fees from Roche, and grants from Actelion, Avalyn Pharma, Biogen and Galapagos, outside the submitted work.
Conflict of interest: C. Romei has nothing to disclose.
Conflict of interest: R. Savas has nothing to disclose.
Conflict of interest: C.J. Brereton has nothing to disclose.
Conflict of interest: H.W. van Es has nothing to disclose.
Conflict of interest: H. Jo has nothing to disclose.
Conflict of interest: A. De Liperi has nothing to disclose.
Conflict of interest: O. Unat has nothing to disclose.
Conflict of interest: K. Pontoppidan has nothing to disclose.
Conflict of interest: F. van Beek has nothing to disclose.
Conflict of interest: M. Veltkamp has nothing to disclose.
Conflict of interest: P. Hopkins has nothing to disclose.
Conflict of interest: Y. Moodley has nothing to disclose.
Conflict of interest: A. Taliani has nothing to disclose.
Conflict of interest: L. Tavanti has nothing to disclose.
Conflict of interest: B. Gholipour has nothing to disclose.
Conflict of interest: A. Nair reports a proportion of their permanent employment at UCL Hospital is funded by the Biomedical Research Centre, and nonfinancial support for an advisory board from Aidence BV, the Netherlands, outside the submitted work.
Conflict of interest: S. Janes reports personal fees and nonfinancial support from AstraZeneca, personal fees from Bard1 Bioscience, Achilles Therapeutics and Jansen, nonfinancial support from Takeda, and grants from GRAIL Inc., GlaxoSmithKline plc and from Owlstone, outside the submitted work.
Conflict of interest: I. Stewart has nothing to disclose.
Conflict of interest: D. Barber has nothing to disclose.
Conflict of interest: D.C. Alexander has nothing to disclose.
Conflict of interest: A.U. Wells reports personal fees and nonfinancial support from Boehringer Ingelheim, Bayer and Roche Pharmaceuticals, and personal fees from Blade, outside the submitted work.
Conflict of interest: J. Jacob reports personal fees from Boehringer Ingelheim and Roche, grants and personal fees from GlaxoSmithKline, personal fees from NHSX, and grants from the Wellcome Trust, outside the submitted work.
Support statement: This study was supported by Wellcome Trust grant 209553/Z/17/Z and the UCLH Biomedical Research Centre. Funding information for this article has been deposited with the Crossref Funder Registry.
- Received May 5, 2021.
- Accepted June 15, 2021.
- Copyright ©The authors 2021
This version is distributed under the terms of the Creative Commons Attribution Non-Commercial Licence 4.0. For commercial reproduction rights and permissions contact permissions{at}ersnet.org