COPD Biomarkers and their interpretation; a Personal
perspective
Robert A Stockley1, David Halpin2, Bartolome R Celli3 and
Dave Singh4.
Affiliations
1.Lung Investigation Unit, Medicine - University Hospitals Birmingham NHS Foundation
Trust, Queen Elizabeth Hospital Birmingham, Mindelsohn Way, Edgbaston, Birmingham, B15
2GW,UK. 2. Department of Respiratory Medicine, Royal Devon & Exeter Hospital, Barrack
Road, Exeter, EX2 5DW, UK. 3.Pulmonary and Critical Care Dept. Brigham and Women’s
Hospital, Harvard Medical School, Boston, MA,USA. 4. University of Manchester, Medicines
Evaluation Unit, Manchester University NHS Foundation Hospital Trust, Southmoor Road,
Manchester, M23 9QZ, UK.
Corresponding Author
Robert A Stockley Lung Investigation Unit, Medicine - University Hospitals Birmingham NHS
Foundation Trust, Queen Elizabeth Hospital Birmingham, Mindelsohn Way, Edgbaston,
Birmingham, B15. Telephone +44 121 3716808: E mail [email protected]
Authors contribution
RAS conceived of the project and wrote the initial draft. DH BC and DS critically appraised
and added to the content and approved the final submission.
Running head. COPD Biomarkers
Descriptor COPD Emphysema
Total word count 5083.
At a Glance
The pathology and impact of COPD results from an abnormal inflammatory process
resulting in tissue damage with ineffective repair in response to toxic inhalants (especially
cigarette smoke). Identification of mechanisms provides the opportunity to develop new
therapies and a personalised approach to management. The collection of multiple genetic
and detailed biochemical data from small and large patient cohorts has led to an explosion
of studies investigating biomarkers, to achieve these aims.
Despite widespread enthusiasm and many statistically significant associations, the
interpretation of COPD biomarker results requires thought and leaves many questions
unanswered. The current review assesses the importance of these associations, whether
they represent cause or effect, reflect disease severity or activity, the complexity of the
pathway to the final pathogenic and hence interventional step, and problems interpreting
cross sectional studies without knowing individual disease trajectories. The complexity of
biomarker specificity without sufficient clinical phenotype and endotype information
contributes to problems of interpretation. A strategic change is needed to develop useful
COPD biomarkers; this includes focusing on endotype biomarkers within specific clinical
phenotypes, biomarkers in the early phases of COPD development that may differ from
those in established disease , exacerbation subtype biomarkers, and biomarkers to predict
or measure drug effects. Generating these data are critical for the future and multiple
potential use/s of biomarkers. At present only blood eosinophils is close to implementation
in clinical practice.
Introduction
Biomarkers are defined as “characteristics that are objectively measured and evaluated as
an indicator of normal biological processes, pathogenic processes, or pharmacologic
responses to a therapeutic intervention” 1 and are increasingly explored in COPD for several
important reasons. These include diagnosis, facilitating patient characterisation, quantifying
disease severity, determining prognosis, predicting and detecting response to treatment as
well as investigating the pathophysiology of the disease/diseases 1. Clinical phenotyping
categorises patients using observable characteristics 2, while endotyping uses the presence
of a biological mechanism to define a patient subgroup2. The development of biomarkers
that identify endotypes most likely to respond to targeted pharmacological treatments is
integral to precision medicine2,3.
COPD is the result of an abnormal and persistent inflammatory process that damages the
lung architecture, resulting in progressive physiological deterioration (at least in some
patients)4. Understanding the processes involved can lead to the rational development of
new molecular biomarkers which can be used to (a) monitor disease ACTIVITY to identify
individuals at high risk of rapid disease progression (b) identify endotypes most likely to
respond to a pharmacological intervention and (c) measure the pharmacological modulation
of disease pathophysiology in clinical trials, particularly early phase 1/2 studies where proof
of pharmacological effect provides confidence to progress to larger phase 3 studies focused
on clinical efficacy. Biomarker development, validation and interpretation varies according
to the potential future clinical application as a biomarker of disease activity, endotype or
pharmacological effect.
COPD biomarker studies have investigated various lung media including sputum,
bronchoalveolar lavage and exhaled breath condensate5. All have major problems of
reproducibility1,6, and may rely on assays not developed for / validated in the medium being
studied. For practical purposes, most of the published studies have involved blood
biomarkers because of easy access and assay reproducibility, and therefore the current
article focusses primarily on blood biomarkers. It is not the purpose of the current article to
document the extensive COPD biomarker literature, but rather highlight problems that
complicate data interpretation. We also propose that the historic approach to biomarker
identification and validation needs to be modified, due to the clinical and biological
complexity of COPD. More careful consideration should be given to the study designs
required to develop biomarkers that measure disease activity, identify endotypes (that
might predict response to treatment) or monitor response to pharmacological treatment.
COPD vs controls.
Many COPD biomarker studies have evaluated disease versus healthy controls in the hope
of finding a difference between groups coupled with little or no overlap between group
individual measurements. As COPD is an inflammatory disease it is understandable that
many markers such as CRP, TNFα, IL6, IL8, fibrinogen, neutrophil numbers and activation are
raised in COPD patients7,8. While statistically significant differences between groups for such
biomarkers have been reported once all available confounding factors have been taken into
account, the individual subjects usually show marked overlap that reduces both the
sensitivity and specificity of the biomarker as a diagnostic tool for COPD. This could reflect
time dependant variations in disease activity in only a proportion of patients or different
clinical phenotypes as well as assay variability, particularly in individuals with low grade
inflammation where the disease noise to signal ratio may be difficult to distinguish . This
uncertainty can only be addressed by assay characterisation, extensive phenotyping, and/or
longitudinal follow up.
The analysis of plasma cytokine and chemokine biomarkers in the COPD gene (n=2123) and
SPIROMICS (n=1117) cohorts7 illustrates these problems; the data ranges for COPD patients
in both cohorts were wide with a large overlap between COPD patients and controls.
Significant p values were obtained for comparisons between groups for some biomarkers
such as eotaxin and IL-6, but the considerable overlap prevents use as a diagnostic
biomarker in clinical practice. In addition, even though the systemic levels of biomarkers
may relate to outcomes, they provide little prognostic value above easily obtainable clinical
variables9.
In smokers with symptoms, the diagnosis of COPD is based on the FEV 1/FVC ratio of ≤0.7 as
recommended by GOLD10. This has been frequently challenged as this cut-off is too low in
younger adults and too high in more elderly subjects. This becomes most important in the
presence of a wide range of values for those classified as “normal”, especially in subjects
with “early” phases of disease when significant pathological change can exist in the small
airways prior to influencing forced expiratory values detected by spirometry11. In addition,
because of the wide normal range, subjects can have either excessive or minimal
physiological decline before the “disease threshold” is crossed12 giving a false impression of
disease ACTIVITY. Furthermore, recent studies have also identified subjects with regular
symptoms and periodic worsenings that resemble COPD exacerbations and those with
significant emphysema on CT scans who do not meet the spirometric criterion for COPD
diagnosis13,14. A proposal to identify this ‘early phase of COPD’ in younger (<50 years age)
individuals with normal FEV1, has suggested the criteria required; FEV1/FVC ratio <0.7 or CT
scan evidence of emphysema or documented rapid FEV1 decline15. This proposal moves
beyond the usual spirometric criteria in order to identify individuals with high levels of
disease activity at a relatively early age that precedes crossing the spirometric threshold.
While these criteria need validation, biomarker studies in such individuals that focus on
disease ACTIVITY prior to the development of significant tissue damage would support early
effective clinical management and indeed may differ from the activity processes as the
disease becomes more established (see later).
Cross-sectional studies; Cause or effect
Cross-sectional studies have frequently been used to assess the association between
biomarkers and COPD severity, usually expressed as the FEV1 percent predicted. A
statistically significant correlation is usually taken as a positive finding, but then raises the
question of “cause or effect”. The biomarker may reflect a pathophysiological process that
drives or fails to dampen the ongoing development of lung damage (“cause”), and therefore
be related to disease activity levels. For this to be a distinct possibility all subjects studied
should be age matched suggesting they have reached different degrees of severity in the
same time frame. For the same reason, all should have similar smoking history and
treatment. However, the relationship may still reflect disease severity (“effect”) as the
biomarker is measuring lung damage itself either directly or indirectly, or may be an
epiphenomenon reflecting a shared aetiological factor, or even be a cohort related chance
finding.
The use of the statistical p value will determine the likelihood of a type 2 error (chance
finding) of the study. However, it has been suggested that such analyses should use a lower
threshold p value to provide additional reassurance that the findings do not simply reflect
chance16 especially when multiple marker analysis is undertaken. In addition the larger the
sample size the more likely that minimally important differences achieve statistical support.
Also important is the strength of the association (rho value), as statistically significant
associations (judged by p value, usually set at 0.05 and lower) with a low rho value are likely
to have little overall clinical implication (see below). For these reasons such correlations
need careful analysis and interpretation.
Temporal associations may help interpretation with careful matching of lung function and
age. At the same level of lung function impairment, in general, those with a rapid
progression/higher disease activity should be younger than those with slow progression,
assuming the disease started at a similar age. Similarly with patients matched for age those
with rapid progression/higher disease activity should have worse lung function, again
assuming the disease started at a similar age. The interpretation of cross-sectional data
from cohorts based on levels of lung function impairment even when matched for age is
further complicated by the observation that the normal range is wide and in some people
low levels of baseline lung function result from poor lung development in early life 4. Indeed,
the diagnosis of COPD in some individuals may be mainly attributable to poor lung
development, associated only with a relatively slow subsequent rate of FEV1 decline in later
life. None of this can be understood from cross-sectional data, as different levels of disease
ACTIVITY can be present in individuals with the same FEV1 percent predicted, even after
careful matching for age and smoking history as the starting point is unknown. Ideally,
longitudinal data-collection is needed to determine on going progression rate and properly
develop biomarkers that reflect disease ACTIVITY.
The study by Campos and colleagues 17 in patients with alpha-1 antitrypsin deficiency (AATD)
illustrates these problems. Fast or slow FEV1 decliners were identified by studying different
age cohorts matched for smoking history and FEV1. The oldest cohort was therefore defined
as a group with “slower” decline in lung function compared to the younger cohorts,
consistent with differences in long term disease ACTIVITY. However, there are 2 alternative
explanations; firstly that each cohort had progressed at the same rate but the older cohort
started at a higher baseline lung function than the younger cohort. Secondly, and potentially
more likely, each cohort had developed a similar decline until they stopped smoking (all
groups stopped in their 40s) and thereafter disease progression stabilised. Without baseline
lung function and longitudinal FEV1 measurements, the true explanation remains uncertain.
This is consistent with recent longitudinal data in AATD where ex and never smokers had a
similar rate of decline despite differing lung function at baseline 18 . In this scenario, any
biomarker associated with lung function using a simple cross-sectional analysis may reflect
disease SEVERITY rather than ACTIVITY (Figure2). Longitudinal analysis with careful
consideration of smoking history is the only way to overcome this conundrum.
Figure1 The figure demonstrates the path of idealised FEV1 decline with time, derived from
longitudinal progression in AATD 17. If the factors responsible for the decline persist, the
cross sectional assessment at the age of 60 will reflect the ongoing cause (ACTIVITY marker).
However if the progression has been amplified by smoking (the dotted line) cessation of
smoking will return the disease progression to its natural course. In this instance any marker
of disease ACTIVITY will no longer relate cross sectionally to the SEVERITY at the age of 60
and could be dismissed as irrelevant. The non-smoking and smoking pathological process
can therefore only be explored prospectively in cohorts of healthy never smokers, current or
ex-smokers with and without COPD.
Longitudinal studies of disease progression
The rate of progression of general COPD and that due to AATD is highly variable 4,18-20. The
term “rapid” decliners is often used, although a standard definition is lacking because lung
function normally declines with age and depends on height. Perhaps it is best to define
decline of FEV1 as % predicted to overcome this natural deterioration.
Linking baseline biomarkers to the pathophysiological processes causing lung damage
requires the concept that this is a continuous and homogenous process. This may be true, at
least for some clinical phenotypes and thus the baseline marker predicts this low grade
activity and hence subsequent progression regardless of whether the marker relates cross
sectionally to SEVERITY. Since many other factors can also influence decline (such as
smoking status, exacerbations and emphysema phenotype etc.20), it appears unlikely that a
single biomarker would identify all of the mechanisms associated with these different
factors21. Multiple biomarkers will likely be needed to identify these different disease
ACTIVITY mechanisms (endotypes) within well-defined and monitored clinical phenotypes.
In recent years the ability to make multiple measurements in cohort studies has become
possible by the “omics” explosion and multiplex assays. The relationship between plasma
cytokine and chemokine biomarkers and clinical parameters or outcomes including severity
of airflow obstruction, FEV1 decline, emphysema progression and exacerbations was studied
in the COPDgene and SPIROMICS cohorts7. However, a lack of replication between cohorts
was observed, and positive findings only gave minor additional value to clinical variables.
The heterogeneous clinical and pathophysiological nature of COPD likely explains some of
the discordance.
Various biomarkers have been studied in several COPD cohorts and the reported
relationship to patient outcomes are summarised and appraised for the most frequently
studied biomarkers in table 1.8 These include surfactant protein D (SP-D) and club cell
protein 16 (CC-16). SP-D is involved in pulmonary innate immune defence, while CC-16 is an
immunosuppressant protein secreted by Clara cells that protects against excessive lung
injury22. Increased blood SP-D levels and decreased CC-16 levels were reported in COPD
patients compared to controls8,23, although in both cases there was a large data overlap
between the groups. SP-D levels were statistically associated with emphysema progression
and mortality over 3 years24, while low CC16 levels were related to FEV1 decline, but not
mortality. The findings for SP-D were also seen in COPDgene but the overlap and influence
of other factors on SP-D levels including age, BMI, gender and current smoking makes it
difficult to use on an individual basis for any treatment or prognostic outcome22.
A major study of over 4000 subjects over 10 years also analysed both the correlation of CC-
16 with lung function and its progression25. The statistical p values confirmed that the
relationships were highly unlikely to have occurred by chance and the authors concluded
that the measurement was informative of the pathophysiological process and could be used
to identify rapid decliners for future interventional studies of potential disease modifying
therapies. However, examination of the individual data (Figure 3) indicates the wide range
of values for the patient group. This is reflected in the r value for the correlation which,
though significant, is low. Indeed here the r2 value is helpful in that it indicates that less than
5% of the decline variability can be predicted in a given patient by measuring CC16. This
marker would therefore be of little use in enriching populations for studies of rapid
decliners or targeting patients with disease modifying therapy. Whether CC16 would be
more discerning in patients characterised for COPD features other than FEV1 remains to be
determined. The future development of biomarkers of disease progression will be improved
by studying distinct clinical subgroups (e.g. frequent exacerbators, emphysema etc).
Figure 2. The rate of decline in FEV1 is shown compared to baseline CC16 level. The relationship was statistically significant as shown by the p value and the linear relationship is shown by the regression line with its 95% confidence interval. However the scatter of the data is shown and for 2 patients highlighted by the circles with the same baseline level of CC16, subsequent decline is clearly vastly different emphasising it is not a suitable biomarker for predicting FEV1 decline or selection of rapid decliners for future studies despite the p value. Park HY, Churg A, Wright JL, et al. Club cell protein 16 and disease progression in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2013; 188(12): 1413-9. Figure 2 Reprinted with permission of the American Thoracic Society. Copyright © 2018 American Thoracic Society.
This concept of the variability of the relationship between biomarker and outcome such as
the FEV1 is important to note even when statistically significant. Where such data can be
derived the correlations between FEV1 and individual biomarkers indicate that little of the
variance (<1% to a maximum of 6%) can be attributed to the value 7,8,23. The use of multiple
biomarkers may improve the predictive value, though in general not greatly. However, a
combination of CC16, SP-D, CRP, sRAGE, and fibrinogen increases the sensitivity for FEV 1
cross sectionally to about 24% and combining SP-D, CRP, sRAGE and fibrinogen increases the
sensitivity for emphysema to approximately 30%26.
Emphysema; complexity of biomarker development
The role of neutrophil elastase (NE) has long had a prominent role in the pathophysiology of
COPD in general and emphysema in particular. AAT is the major serum inhibitor of serine
proteinases and especially NE. AAT deficiency impairs the ability to inhibit NE and since this
enzyme can cause all of the pathological features of COPD 27, it seems a clear direct path
from cause to effect consistent with susceptibility of patients with genetic deficiency to
develop emphysema even in the absence of smoking. NE is a potent elastolytic enzyme and
destruction28 or failure of elastin repair29 is certainly associated with emphysema like
changes in animal models. However, other potential elastolytic enzymes such as the
MMPs30 and cysteine proteinases31 have also been implicated via animal models suggesting
NE is not necessarily the sole and key mediator of the degradation of lung elastin. Increases
in the presence of these enzymes in COPD32,33 adds weight to their potential role especially
as these other enzyme classes also inactivate AAT function, thereby potentially facilitating
NE activity indirectly. On the other hand NE itself activates MMPs and cysteine proteinases
by cleavage of the pro-enzymes and inactivates their cognate inhibitors34. This leads to an
interactive cycle where the role of any enzyme group could be the initial step towards tissue
damage and alternative pathways may determine the eventual pathology (Figure 3). Both
pathways have complex interactions which lead to emphysema11,35. The interpretation of
biomarkers which are part of a pathophysiological network, rather than a distinct linear
chain is therefore complex as homeostatic mechanisms in a network may counteract
changes in a single biomarker and the relationship between that biomarker and disease
activity or tissue damage may or may not be modified.
Figure 3. The direct pathway of Neutrophil Elastase (NE) to Tissue damage leading to
Emphysema has long been described. At the same time NE activates the proenzymes of
Cathepsin B and MMPs which may also contribute to tissue damage.NE also inactivates the
inhibitors of Cathepsins and MMPs to facilitate activity of these groups of enzymes indirectly
leading to tissue damage whilst also damaging the inactivators of NE itself enhancing the
direct NE pathway. Emphysema leads to airflow obstruction (AO) reflected in the FEV1.
However there is also the alternative pathway involving the activation of growth factors by
MMPs leading to small airways disease (SAD) and hence AO by a different pathological
process35. Thus the biomarkers central to the disease process may also change as the
pathological phases of the disease change.
Measuring these enzymes in lung secretions may provide information of their presence in
the airway and role in epithelial damage, but not necessarily of their role in the interstitium
(where elastin degradation is thought to be central to the development of emphysema) or
fibrotic changes leading to loss of the small airways. Even measuring elastin breakdown and
its products in the circulation does not provide lung specific evidence of the enzyme /
enzymes involved, as skin36 and the cardiovascular system37 are major repositories for
elastin and degradation products may simply reflect cell degranulation/turnover or aging at
these sites and involve multiple enzymes and pathways not reflective of pulmonary tissue
breakdown.
At the point of release NE concentration exceeds the normal AAT concentration resulting in
an area of obligate proteolysis before dilution by diffusion to enable total enzyme
inhibition38. This area of obligate proteolysis is understandably greater in AATD39. This
means that migrating neutrophils produce protein degradation in the immediate vicinity
that can be tracked by specific cleavage products such as that for fibrinogen40 or elastin41.
These biomarkers may provide more direct evidence of enzyme activity in the lung (or other
tissues with ongoing neutrophilic inflammation) and hence reflect the immediate efficacy of
specific NE inhibitors. Indeed the fibrinogen elastase biomarker decreases in AATD patients
on AAT augmentation therapy40. The development of such markers specific for the
emphysema phenotype of COPD may provide information on disease ACTIVITY and also
indicate early response to therapeutic intervention.
Soluble receptror for advanced glycation end-products (sRAGE); Plausible or not ?
sRAGE is a biomarker that demonstrates pathophysiologic credibility, genetic associations
and a relationship to emphysema. Inflammatory mediators associated with COPD and
emphysema generate RAGE ligands, which interact with cell membrane RAGE, activating
intracellular pathways leading to gene transcription of many inflammatory proteins which
then interact with RAGE forming an amplification loop. RAGE cleavage by
metalloproteinases release soluble RAGE (sRAGE) which acts as a decoy interrupting cell
RAGE binding by its ligands and reduces cell associated RAGE receptors, both with putative
beneficial effects on the inflammatory cascade. RAGE overexpression in mice increases
susceptibility to developing air space enlargement42 and KO mice have reduced
inflammation in response to cigarette smoke43. Furthermore, wild type mice treated with
sRAGE display less inflammation as it acts as a decoy receptor44.
Human studies have demonstrated a relationship of COPD and emphysema to blood sRAGE
levels (Table 1). There are lower sRAGE levels in patients with emphysema compared to
controls45, and the level of sRAGE relates to GOLD severity stage46,47. Cheng et al48
demonstrated that lower sRAGE levels related to both gas transfer and lung density (as a
marker of emphysema), and a SNP was identified that was related to sRAGE levels. Iwanato
and colleagues demonstrated lower sRAGE levels in COPD and a relationship to lung
function decline49.
Superficially these data indicate that not only is sRAGE a viable mediator of the
pathophysiological processes implicated in COPD but also a marker of disease ACTIVITY and
likely has a genetic cause. However, careful examination uncovers uncertainties in the
findings. Firstly the individual values in COPD are wide with significant overlap with controls.
Only Cockayne et al46 showed all the individual data points (rather than mean / SD or
median / interquartile range), and only a minority of COPD patients were different to
controls even though the group data is statistically different. The relationship to gas
transfer accounted for up to 21% of the variability. Iwanato and colleagues 49 confirmed the
wide range of sRAGE values, but also showed that levels for both smoking controls and
smokers with COPD were lower compared to non-smokers. The similarity between both
smoking groups suggested that sRAGE is modulated by smoking and not COPD specific.
Baseline sRAGE levels did not predict FEV1 decline, although in a multivariate model both
current smoking and sRAGE levels were independent predictors of the rate of FEV1/FVC ratio
decline.
Where does this data leave us? Although sRAGE remains the best associated marker of the
emphysema phenotype of COPD24 the associations are mainly weak (though statistically
significant). It remains to be seen how sRAGE responds to factors that slow or stop
emphysema progression (such as smoking cessation). sRAGE regulation by other
inflammatory processes in COPD such as exacerbations, bacterial colonisation and
bronchiectasis should be studied to understand if this represents a physiological response to
inflammatory processes in general. The impact of the genetic polymorphism also need to
be determined. So even this biomarker has a long way to go before it can be used to
understand the pathophysiology or as part of a precision medicine approach.
Exacerbation Biomarkers
Multiplex methodology was used to evaluate the potential of 90 protein biomarkers to
predict exacerbation frequency in the COPDgene (n=1544) and SPIROMICS (n=602)
cohorts50. Although biomarkers predictive of exacerbations were found in each cohort, there
was poor replication between the studies. This likely reflects the fact that COPD
exacerbations have many causes and the definition of an exacerbation remains the
“subjective” clinical assessment of the need for increased pharmacological treatment, which
adds to variability. Given the heterogeneous nature of exacerbations, it is unlikely that one
biomarker could reliably predict such a range of clinical events better than past history. The
blood biomarkers studied in the COPDgene and SPIROMICS cohorts provided only a
marginal additional improvement to the ability of clinical characteristics to predict
exacerbations50. A combined analysis of 5 cohorts including the ECLIPSE cohort showed that
plasma fibrinogen used with a clinical history of exacerbations increased the ability to
predict future events51, leading to FDA qualification of this biomarker in the context of use
as an enrichment tool for recruitment into clinical trials with exacerbations as an outcome.
However, the additional value of adding plasma fibrinogen in the context of clinical trial use
is relatively small and lacks the sensitivity and specificity to be used on an individual basis in
clinical practice50. Furthermore, fibrinogen failed to emerge in the COPDgene and
SPIROMICS cohorts, albeit using a different analytical method, which highlights the general
need to also standardise biomarker analytical methodology. Ultimately, the utility of
predictive exacerbation biomarkers will only be improved by studying exacerbation
subtypes defined by clinical criteria combined with objective biological criteria.
Pharmacological biomarkers
There is increasing interest in whether biomarkers which reflect underlying
pathophysiological processes can be used to predict the effects of therapeutic interventions.
There are many novel immunomodulatory drugs in clinical development that selectively
target components of the host immune system, and are unlikely to be effective in all COPD
patients. Combining clinical information with biomarkers to improve the benefit versus risk
ratio is essential for the future clinical use of these drugs.
There is growing evidence that blood eosinophil counts can be used to predict the effects of
certain drug classes in COPD patients. A subgroup of COPD patients have increased
eosinophil numbers in the lungs52, and there is a relationship between blood and lung
eosinophil numbers52-54. Much attention has been focused on the relationship between
blood eosinophil counts and exacerbation risk, which has been found to be positive in some
COPD studies, but not others55-57. The reasons for negative findings may reflect the inclusion
of patients with no exacerbation history, the confounding effects of concurrent ICS use and
insufficient sample size; the larger COPDgene and ECLIPSE cohorts have shown blood
eosinophil counts add predictive value in the subgroup of individuals with a history of
exacerbations57. Regardless, the main interest in blood eosinophil counts has arisen from
post-hoc analyses of clinical trials comparing ICS/LABA versus LABA indicating that blood
eosinophils predict the ICS effect on exacerbation prevention, with a continuous “blood
eosinophil concentration - response” relationship and a treatment effect of approximately
50% observed at > 300 cells/µL58,59. A large post-hoc analysis of 3 clinical trials (n=4528)
comparing ICS/LABA with LABA showed a significant effect of ICS at > 100 cells/µL, with the
effect size increasing with higher blood eosinophil counts60. The predictive nature of blood
eosinophils with regard to ICS effects has now been confirmed in prospective clinical trials61-
63.
The predictive ability of blood eosinophils with regards to ICS response is not an “all or
nothing” phenomenon64, and a single dichotomous threshold does not fully describe this
relationship. Nevertheless, the threshold of approximately 100 cells/µL is emerging as a
negative predictor for ICS response, and could be used as part of a precision medicine
strategy to optimise the benefit versus risk of ICS use at an individual level. There seem to
be underlying biological differences between the airways of eosinophilic and non-
eosinophilic COPD patients with the former having higher levels of T2 cytokines and greater
reticular basement membrane thickening54,65, drawing a similarity to airway inflammation in
asthma. Furthermore, there is less airway colonisation with pathogenic bacteria in
eosinophilic COPD54. These biological differences could explain the variations in ICS
response.
Blood eosinophil counts seem to predict the clinical response to monoclonal antibody
targeting of the cytokine IL-5 which lowers eosinophil counts 66, and post-hoc analysis of
clinical trials of the PDE4 inhibitor roflumilast suggest that patients with higher blood
eosinophil counts experience a greater clinical benefit67. Interestingly, in the mepoluzimab
trial, reduction of exacerbations were minimal compared with placebo, even though the
blood eosinophil counts were reduced to less than 50 cells/µl for the 1 year duration of the
trial. The use of blood eosinophil counts may be affected by the reproducibility of the
measurement56,68, although this appears to be more stable at lower levels69. It should be
noted that this measurement is reduced during sepsis and by oral corticosteroid use52.
Understanding biomarker data
Studies of biomarkers will continue to be published especially when statistically significant
data is reported suggesting it is unlikely to reflect a type 2 error. Whereas the p value
indicates the likelihood of an association having occurred by chance its threshold needs
adjustment when multiple assays are undertaken and the separation of data from
appropriate controls providing sensitivity and specificity as well as positive (or negative)
predictive values and the r2 value of correlations will indicate the strength of the message.
Studies with multiple biomarkers are suitable for a systems biology approach to gain a
better understanding of the interactions of components into defined networks; this may
offer a new dimension to the reductionist approach to understanding pathophysiology.
Whether this results in the emergence of more or less potential biomarkers or realistic
targets for drug development remains to be seen. In terms of a practical biomarker for daily
use, the centre of the network may prove less important than downstream components.
The development of new COPD biomarkers and interpretation of data therefore requires
more extensive study and careful interpretation than is currently undertaken. We
summarise key issues below, and propose areas of importance to focus on regarding the
further development of COPD biomarkers
1.Comparisons of COPD versus controls: Given the inevitable overlap between COPD and
controls, it is important to perform comprehensive clinical characterisation with multiple
groups including matched healthy smokers and ex smokers, with current and ex smoking
COPD patients. In addition it is important to understand if the outliers are a distinct clinical
phenotype and / or whether specific biological mechanisms reflecting disease activity are
the cause. The investigation of younger patients with “early stages of COPD” is important in
this context.
2.Endotype biomarker: Dissecting COPD into component mechanisms is essential for the
development of biomarkers. Endotype biomarkers that reflect these mechanisms may have
multiple applications in COPD, ranging from monitoring disease activity to identifying
individuals likely to respond to a therapeutic intervention or measuring the pharmacological
effects of drugs. AAT deficiency is an example of a COPD endotype that is detected by a
specific biomarker. A number of endotypes have been proposed (T2/ eosinophilic COPD,
bacterial colonisation, systemic inflammation)2; further biomarker studies of these
endotypes should be carried out in specific clinical phenotypes (e.g. frequent exacerbators,
rapid decliners) rather than poorly characterised and broad populations.
3.COPD ACTIVITY: The discovery of biomarkers related to the process/es that cause direct
damage or block repair can be used to monitor disease activity, particularly after
therapeutic interventions. Longitudinal studies in discrete clinical phenotypes, rather than
general COPD populations, are needed to identify disease activity biomarkers correctly
related to specific mechanisms. The influence of factors such as active smoking need to be
taken into account, and changes in such biomarkers over time in both stable and
progressive disease need to be determined. The biomarker needs to have or reflect a
plausible mechanism of progression.
4.Pharmacological biomarker: Precision medicine involves the use of clinical and biological
information to predict therapeutic response. The use of blood eosinophil counts in COPD
patients at high risk of exacerbation to predict ICS effect is an example of such an approach.
Biomarkers may also be used to measure drug responses, and it should be determined
whether it is robust enough to be used clinically in an individual patient and / or as an
endpoint especially in early phase clinical trials.
5.Exacerbation biomarkers: Biomarkers should be developed that add extra information in
addition to the history of exacerbations, such as the cause/nature of the exacerbation/s and
the need/likely response to specific therapies.
6. Assay validation. Many studies include the use of multiple assay systems not necessarily
developed in the biological medium to be assessed. This together with the assay variability,
lower limit of detection and lower limit of quantification are essential information and
critical in the assessment of low grade inflammation in COPD patients without
exacerbations. Such information would be enhanced by day to day variation in the values
obtained as well as monthly or annual confirmation of stability to explore relationships with
long term disease progression.
Conclusion
Future biomarker studies need to move away from the simplistic approach of comparing
COPD patients versus controls, for all the reasons already outlined, and focus on more
targeted questions in specific patient subgroups / phenotypes. The concepts of disease
endotypes and disease activity are crucial to this future approach. The positive data
regarding the ability of blood eosinophil counts (a biomarker of an endotype) in patients at
increased exacerbation risk (clinical phenotype) to predict response to ICS integrates some
of these concepts, providing an opportunity to use this biomarker to practice precision
medicine in COPD patients. We advocate more focussed and detailed studies in the future
or we run the continued risk of data overload and missing the initiating and mediating
factors responsible required for new therapy design and delivery in COPD.
References
1. Hollander Z, DeMarco ML, Sadatsafavi M, McManus BM, Ng RT, Sin DD. Biomarker Development in COPD: Moving From P Values to Products to Impact Patient Care. Chest 2017;151:455-67.2. Woodruff PG, Agusti A, Roche N, Singh D, Martinez FJ. Current concepts in targeting chronic obstructive pulmonary disease pharmacotherapy: making progress towards personalised management. Lancet 2015;385:1789-98.3. Singh D, Roche N, Halpin D, Agusti A, Wedzicha JA, Martinez FJ. Current Controversies in the Pharmacological Treatment of Chronic Obstructive Pulmonary Disease. Am J Respir Crit Care Med 2016;194:541-9.4. Lange P, Celli B, Agusti A, et al. Lung-Function Trajectories Leading to Chronic Obstructive Pulmonary Disease. N Engl J Med 2015;373:111-22.5. Barnes PJ, Chowdhury B, Kharitonov SA, et al. Pulmonary biomarkers in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2006;174:6-14.6. Dickens JA, Miller BE, Edwards LD, et al. COPD association and repeatability of blood biomarkers in the ECLIPSE cohort. Respir Res 2011;12:146.7. Bradford E, Jacobson S, Varasteh J, et al. The value of blood cytokines and chemokines in assessing COPD. Respir Res 2017;18:180.8. Faner R, Tal-Singer R, Riley JH, et al. Lessons from ECLIPSE: a review of COPD biomarkers. Thorax 2014;69:666-72.9. Celli BR, Locantore N, Yates J, et al. Inflammatory biomarkers improve clinical prediction of mortality in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2012;185:1065-72.10. Vogelmeier CF, Criner GJ, Martinez FJ, et al. Global Strategy for the Diagnosis, Management, and Prevention of Chronic Obstructive Lung Disease 2017 Report: GOLD Executive Summary. Am J Respir Crit Care Med 2017.11. Hogg JC, Pare PD, Hackett TL. The Contribution of Small Airway Obstruction to the Pathogenesis of Chronic Obstructive Pulmonary Disease. Physiol Rev 2017;97:529-52.12. Miller MR, Quanjer PH, Swanney MP, Ruppel G, Enright PL. Interpreting lung function data using 80% predicted and fixed thresholds misclassifies more than 20% of patients. Chest 2011;139:52-9.13. Woodruff PG, Barr RG, Bleecker E, et al. Clinical Significance of Symptoms in Smokers with Preserved Pulmonary Function. N Engl J Med 2016;374:1811-21.14. Regan EA, Lynch DA, Curran-Everett D, et al. Clinical and Radiologic Disease in Smokers With Normal Spirometry. JAMA Intern Med 2015;175:1539-49.15. Martinez FJ, Han MK, Allinson JP, et al. At the Root: Defining and Halting Progression of Early Chronic Obstructive Pulmonary Disease. Am J Respir Crit Care Med 2018.16. Daniel J. Benjamin JOB, Magnus Johannesson, Brian A. Nosek, E.-J. Wagenmakers,, Richard Berk KAB, Björn Brembs, Lawrence Brown, Colin Camerer, David Cesarini,, Christopher D. Chambers MC, Thomas D. Cook, Paul De Boeck, Zoltan Dienes, Anna Dreber,, et al. Redefine statistical significance. Nature 2018;2:6 - 10.17. Campos MA, Alazemi S, Zhang G, et al. Clinical characteristics of subjects with symptoms of alpha1-antitrypsin deficiency older than 60 years. Chest 2009;135:600-8.18. Stockley RA, Edgar RG, Pillai A, Turner AM. Individualized lung function trends in alpha-1-antitrypsin deficiency: a need for patience in order to provide patient centered management? Int J Chron Obstruct Pulmon Dis 2016;11:1745-56.19. Silverman EK, Sandhaus RA. Clinical practice. Alpha1-antitrypsin deficiency. N Engl J Med 2009;360:2749-57.20. Vestbo J, Edwards LD, Scanlon PD, et al. Changes in forced expiratory volume in 1 second over time in COPD. N Engl J Med 2011;365:1184-92.
21. Pinto-Plata V, Toso J, Lee K, et al. Use of proteomic patterns of serum biomarkers in patients with chronic obstructive pulmonary disease: correlation with clinical parameters. Proc Am Thorac Soc 2006;3:465-6.22. Lock-Johansson S, Vestbo J, Sorensen GL. Surfactant protein D, Club cell protein 16, Pulmonary and activation-regulated chemokine, C-reactive protein, and Fibrinogen biomarker variation in chronic obstructive lung disease. Respir Res 2014;15:147.23. Lomas DA, Silverman EK, Edwards LD, et al. Evaluation of serum CC-16 as a biomarker for COPD in the ECLIPSE cohort. Thorax 2008;63:1058-63.24. Coxson HO, Dirksen A, Edwards LD, et al. The presence and progression of emphysema in COPD as determined by CT scanning and biomarker expression: a prospective analysis from the ECLIPSE study. Lancet Respir Med 2013;1:129-36.25. Park HY, Churg A, Wright JL, et al. Club cell protein 16 and disease progression in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2013;188:1413-9.26. Zemans RL, Jacobson S, Keene J, et al. Multiple biomarkers predict disease severity, progression and mortality in COPD. Respir Res 2017;18:117.27. Stockley RA. Proteases and antiproteases. Novartis Found Symp 2001;234:189-99; discussion 99-204.28. Kuhn C, Yu SY, Chraplyvy M, Linder HE, Senior RM. The induction of emphysema with elastase. II. Changes in connective tissue. Lab Invest 1976;34:372-80.29. Kuhn C, 3rd, Starcher BC. The effect of lathyrogens on the evolution of elastase-induced emphysema. Am Rev Respir Dis 1980;122:453-60.30. Foronjy R, Nkyimbeng T, Wallace A, et al. Transgenic expression of matrix metalloproteinase-9 causes adult-onset emphysema in mice associated with the loss of alveolar elastin. Am J Physiol Lung Cell Mol Physiol 2008;294:L1149-57.31. Lesser M, Padilla ML, Cardozo C. Induction of emphysema in hamsters by intratracheal instillation of cathepsin B. Am Rev Respir Dis 1992;145:661-8.32. Demedts IK, Morel-Montero A, Lebecque S, et al. Elevated MMP-12 protein levels in induced sputum from patients with COPD. Thorax 2006;61:196-201.33. Burnett D, Crocker J, Stockley RA. Cathepsin B-like cysteine proteinase activity in sputum and immunohistologic identification of cathepsin B in alveolar macrophages. Am Rev Respir Dis 1983;128:915-9.34. R.A.Stockley ALSa. Proteinases and COPD. In Chronic Obstructive Pulmonary Disease. Blackwell Publishing Ltd 2007:p349-66.35. Churg A, Zhou S, Wright JL. Series "matrix metalloproteinases in lung health and disease": Matrix metalloproteinases in COPD. Eur Respir J 2012;39:197-209.36. Maclay JD, McAllister DA, Rabinovich R, et al. Systemic elastin degradation in chronic obstructive pulmonary disease. Thorax 2012;67:606-12.37. Rabinovich RA, Miller BE, Wrobel K, et al. Circulating desmosine levels do not predict emphysema progression but are associated with cardiovascular risk and mortality in COPD. Eur Respir J 2016;47:1365-73.38. Liou TG, Campbell EJ. Nonisotropic enzyme--inhibitor interactions: a novel nonoxidative mechanism for quantum proteolysis by human neutrophils. Biochemistry 1995;34:16171-7.39. Sinden NJ, Baker MJ, Smith DJ, Kreft JU, Dafforn TR, Stockley RA. alpha-1-antitrypsin variants and the proteinase/antiproteinase imbalance in chronic obstructive pulmonary disease. Am J Physiol Lung Cell Mol Physiol 2015;308:L179-90.40. Carter RI, Ungurs MJ, Pillai A, Mumford RA, Stockley RA. The Relationship of the Fibrinogen Cleavage Biomarker Aalpha-Val360 With Disease Severity and Activity in alpha1-Antitrypsin Deficiency. Chest 2015;148:382-8.41. Kristensen JH, Karsdal MA, Sand JM, et al. Serological assessment of neutrophil elastase activity on elastin during lung ECM remodeling. BMC Pulm Med 2015;15:53.
42. Stogsdill MP, Stogsdill JA, Bodine BG, et al. Conditional overexpression of receptors for advanced glycation end-products in the adult murine lung causes airspace enlargement and induces inflammation. Am J Respir Cell Mol Biol 2013;49:128-34.43. Reynolds PR, Kasteler SD, Schmitt RE, Hoidal JR. Receptor for advanced glycation end-products signals through Ras during tobacco smoke-induced pulmonary inflammation. Am J Respir Cell Mol Biol 2011;45:411-8.44. Jeong SJ, Lim BJ, Park S, et al. The effect of sRAGE-Fc fusion protein attenuates inflammation and decreases mortality in a murine cecal ligation and puncture model. Inflamm Res 2012;61:1211-8.45. Miniati M, Monti S, Basta G, Cocci F, Fornai E, Bottai M. Soluble receptor for advanced glycation end products in COPD: relationship with emphysema and chronic cor pulmonale: a case-control study. Respir Res 2011;12:37.46. Cockayne DA, Cheng DT, Waschki B, et al. Systemic biomarkers of neutrophilic inflammation, tissue injury and repair in COPD patients with differing levels of disease severity. PLoS One 2012;7:e38629.47. Polverino F, Laucho-Contreras ME, Petersen H, et al. A Pilot Study Linking Endothelial Injury in Lungs and Kidneys in COPD. Am J Respir Crit Care Med 2017.48. Cheng DT, Kim DK, Cockayne DA, et al. Systemic soluble receptor for advanced glycation endproducts is a biomarker of emphysema and associated with AGER genetic variants in patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2013;188:948-57.49. Iwamoto H, Gao J, Pulkkinen V, Toljamo T, Nieminen P, Mazur W. Soluble receptor for advanced glycation end-products and progression of airway disease. BMC Pulm Med 2014;14:68.50. Keene JD, Jacobson S, Kechris K, et al. Biomarkers Predictive of Exacerbations in the SPIROMICS and COPDGene Cohorts. Am J Respir Crit Care Med 2017;195:473-81.51. Miller BE, Tal-Singer R, Rennard SI, et al. Plasma Fibrinogen Qualification as a Drug Development Tool in Chronic Obstructive Pulmonary Disease. Perspective of the Chronic Obstructive Pulmonary Disease Biomarker Qualification Consortium. Am J Respir Crit Care Med 2016;193:607-13.52. Bafadhel M, Pavord ID, Russell REK. Eosinophils in COPD: just another biomarker? Lancet Respir Med 2017;5:747-59.53. Singh D, Kolsum U, Brightling CE, et al. Eosinophilic inflammation in COPD: prevalence and clinical characteristics. Eur Respir J 2014;44:1697-700.54. Kolsum U, Donaldson GC, Singh R, et al. Blood and sputum eosinophils in COPD; relationship with bacterial load. Respir Res 2017;18:88.55. Zysman M, Deslee G, Caillaud D, et al. Relationship between blood eosinophils, clinical characteristics, and mortality in patients with COPD. Int J Chron Obstruct Pulmon Dis 2017;12:1819-24.56. Casanova C, Celli BR, de-Torres JP, et al. Prevalence of persistent blood eosinophilia: relation to outcomes in patients with COPD. Eur Respir J 2017;50.57. Yun JH, Lamb A, Chase R, et al. Blood eosinophil count thresholds and exacerbations in patients with chronic obstructive pulmonary disease. J Allergy Clin Immunol 2018;141:2037-47 e10.58. Pascoe S, Locantore N, Dransfield M, Barnes NC, Pavord ID. Blood eosiophil counts, exacerbations, and response to the addition of inhaled fluticasone furoate to vilanterol in patients with chronic obstructive pulmonary disease: a secondary analysis of data from two parrallel randomised controlled trials. Lancet Respiratory Medicine 2015;3:435-42.59. Siddiqui SH, Guasconi A, Vestbo J, et al. Blood Eosinophils: A Biomarker of Response to Extrafine Beclomethasone/Formoterol in Chronic Obstructive Pulmonary Disease. Am J Respir Crit Care Med 2015;192:523-5.60. Bafadhel M, Peterson S, De Blas MA, et al. Predictors of exacerbation risk and response to budesonide in patients with chronic obstructive pulmonary disease: a post-hoc analysis of three randomised trials. Lancet Respir Med 2018;6:117-26.
61. Vestbo J, Papi A, Corradi M, et al. Single inhaler extrafine triple therapy versus long-acting muscarinic antagonist therapy for chronic obstructive pulmonary disease (TRINITY): a double-blind, parallel group, randomised controlled trial. Lancet 2017;389:1919-29.62. Papi A, Vestbo J, Fabbri L, et al. Extrafine inhaled triple therapy versus dual bronchodilator therapy in chronic obstructive pulmonary disease (TRIBUTE): a double-blind, parallel group, randomised controlled trial. Lancet 2018;391:1076-84.63. Lipson DA, Barnhart F, Brealey N, et al. Once-Daily Single-Inhaler Triple versus Dual Therapy in Patients with COPD. N Engl J Med 2018;378:1671-80.64. Singh D. Predicting Corticosteroid Response in Chronic Obstructive Pulmonary Disease. Blood Eosinophils Gain Momentum. Am J Respir Crit Care Med 2017;196:1098-100.65. Kolsum U, Damera G, Pham TH, et al. Pulmonary inflammation in patients with chronic obstructive pulmonary disease with higher blood eosinophil counts. J Allergy Clin Immunol 2017;140:1181-4 e7.66. Pavord ID, Chanez P, Criner GJ, et al. Mepolizumab for Eosinophilic Chronic Obstructive Pulmonary Disease. N Engl J Med 2017;377:1613-29.67. Martinez FJ, Rabe KF, Calverley PMA, et al. Determinants of Response to Roflumilast in Severe COPD: Pooled Analysis of Two Randomized Trials. Am J Respir Crit Care Med 2018.68. Landis SH, Suruki R, Hilton E, Compton C, Galwey NW. Stability of Blood Eosinophil Count in Patients with COPD in the UK Clinical Practice Research Datalink. COPD 2017;14:382-8.69. Southworth T, Beech G, Foden P, Kolsum U, Singh D. The reproducibility of COPD blood eosinophil counts. Eur Respir J 2018.
Table 1
Biomarker FEV1 FEV1
declineEmphysema Emphysema
progressionMortality Comment
34% 53% 49% 31% Variance explained by using clinical markers
1.CC16 2% 3% 5% 0.5% 2% StableAnti-inflammatory Marked variance
2.SP-D 3% 8% 4% 0.8% 3% StableAnti-inflammatoryAnti-oxidant
3.sRAGE 10% 4% 18% 4% 3% Stability: UnknownInflammation decoy
4.CRP 9% 0% 4% 0.6% 1% UnstableNon-specific acute phase response
5.Fibrinogen 11% 2% 13% 0.6% 2% UnstableCoagulation Acute phase response
6.IL-6 4-5% low UnstablePro and anti-inflammatory.Acute phase response. Chemokine signalling but complete overlap in COPD
Combinedmarkers
1-524%
1,3,57%
2-530%
1,3,54.3%
2,4,56%
The clinical outcomes are shown together with selected biomarkers numbered 1 to 6. The variance of the outcome explained by the biomarker is shown as a % and combination of biomarkers are shown in red in each column. Data is summarised where available from the ECLIPSE study and references 6, 7 and 26.
The conclusion from Reference 26 was “Therefore, the field must acknowledge that statistically significant associations between biomarkers and outcomes that can be observed in large cohorts may be largely inadequate to explain remaining variance after strong clinical covariates are included in the models.”