Stable COPD Treatment: Where are We?

ABSTRACT

The number of pharmacological treatments available for COPD has increased markedly in the last years, mostly corresponding to new agents, combinations and devices within know pharmacological classes. Hierarchizing these options is not straightforward since expected effects are limited by the intrinsically fixed character of the underlying lung damage. In addition, all options have not been directly compared face-to-face. Therefore, guidelines derive from some level of subjective interpretation of the available evidence. Determining which magnitude of change can be taken as clinically relevant is complex although crucial to define long-term strategies. Similarly, estimating not only the possible benefits but also the risks of treatments at the individual level is of major importance to guide choices. In the future biomarkers may be of help in that respect. They will hopefully emerge from progresses in systems biology and medicine. Before then, prescriptions should be restricted to the appropriate treatment indications, as established by high level studies and formalized by guidelines.

KEYWORDS:

• COPD
• Treatment
• Bronchodilators
• Corticosteroids

Introduction

Significant changes in the conceptual framework of guidelines on COPD treatment have occurred during the last decade, following the results of large therapeutic trials (1–4). The purpose of this opinion paper is not to report an extensive analysis of this literature, but to provide the reader with some clues to understand and interpret current trends in COPD management. As a result, a limited number of references has been chosen arbitrarily to illustrate the messages, without any attempt at being exhaustive. Before considering any therapy, every caregiver shall keep in mind that the first and most universal measures to implement when initiating COPD management are smoking cessation interventions, physical activity counselling and vaccination against influenza and Streptococcus pneumoniae. Risk reduction also includes measures against occupational or domestic exposures, when indicated (3). With this in mind, the present article will focus on pharmacological treatments, which should not prevent clinicians from considering the important roles of rehabilitation (indicated in ALL patients in whom some exercise limitation remains despite optimal pharmacological therapy) and, in very specific subgroups, instrumental support (oxygen therapy and noninvasive ventilation) and interventional lung volume reduction (3). Of note, the Cochrane collaboration decided to stop performing systematic reviews and meta-analyses on the beneficial effects of rehabilitation versus conventional care in terms of quality of life (5). This illustrates that clinical research on this topic has already given the best possible level of evidence, although the most optimal modalities remain to be explored in more details.

Goals of COPD management and target mechanisms

A major question to address when designing treatment guidelines relates to the definition of goals of care. In the COPD area, these have been clearly stated, e.g., in the GOLD document, and include symptoms relief (i.e., decreased dyspnea, improved exercise tolerance and quality of life) and prevention of future risks (i.e., risks of exacerbation, disease progression and premature death). Reaching these goals could be considered as some form of disease control. However, although this term (“control”) has a precise significance in asthma, it does not correspond to a well-defined notion in COPD yet (6).

Once treatment goals are defined, knowledge of the mechanisms underlying the impact of disease will allow identifying potential treatment targets. For instance, relieving dyspnea requires to reduce the level of hyperinflation through bronchodilation while fighting muscle deconditioning, thereby acting on the two most important components of the vicious circles that link lung alterations to impaired quality of life and handicap (Figure 1) (7). Reducing exacerbations requires to target inflammation and infection (8), but more detailed insights in underlying^{Figure 1, Figure 2} targets remain necessary. For instance, the role of the microbiome and its modulation represent an area of intensive research (9).

Figure 1. Mechanisms linking chronic airflow limitation to quality of life impairment. Adapted from (7) with permission.

A benefit-risk approach to select the best pharmacological options

Pharmacological approaches comprise short-acting (SA) and long-acting (LA) bronchodilators and agents with anti-inflammatory properties. Bronchodilators belong to two classes, i.e., beta2 agonists (BA) and anticholinergic (AC) agents. Anti-inflammatory drugs include inhaled corticosteroids (ICS), phosphodiesterase-4 inhibitors (PDE4i), macrolides (azithromycin), mucoactive agents with antioxidant effects. Theophylline has both bronchodilator and anti-inflammatory properties, but its therapeutic index is narrow, limiting its use due to side-effects and multiple interactions with food, smoking and other treatments.

The choice between these various agents needs to rely on a benefit-risk approach based on the individual patient‘s characteristics (10): possible benefits will derive from the patient's disease manifestations (mainly dyspnea, exacerbations and their correlates in terms of exercise tolerance, handicap, quality of life) and the possible treatment effects as documented through randomized clinical trials and eventually supported by observational evidence. Possible risks will be deduced from the patient's individual risk factors and comorbidities (Figure 2).

Figure 2. Components of the benefit-risk balance of current COPD treatments. Adapted from (10) with permission.

Inhaled treatments have long been considered as completely safe since they aim at delivering high amounts of drugs locally with minimal systemic exposure. However, there is evidence from randomized controlled trials and/or observational studies that inhaled corticosteroids administered at high doses (those authorized for most products in COPD) during many years have the potential to induce lung infections (pneumonia, tuberculosis, nontuberculous mycobacteria, aspergillosis) and systemic side-effects (11–14). Regarding long-acting bronchodilators, a recent large database study has suggested an increase of cardiovascular events following their initiation (15). Even if these analyses do not correspond to the highest possible level of evidence, they are sufficient to induce caution, underlining that the prescription of these agents should be restricted to their recognized indications.

Determining the best strategy, from levels of evidence to clinical judgment

Current guideline documents rely mostly on symptoms (dyspnea) and exacerbations to define treatment indications. Some variations do exist between guidelines although all are based on the same evidence. These differences illustrate the weight of interpretation in the development of recommendations. A major question when considering treatment effects is to determine their clinical relevance. The magnitude of the effects (i.e., of the differences between the considered treatment and placebo on the one hand, active comparators on the other) must be put in the perspective of the impact of the disease and its intrinsically limited reversibility. Other aspects to consider include the availability, cost and practicality of treatments as well as patients' preferences and abilities, which are crucial in the field of inhaled therapy: in COPD as in asthma, adherence to treatment and inhalation technique are poor and tightly associated with “disease control” (16,17) which encompasses symptoms burden and exacerbation risk (6).

With all the advantages and limitations of each treatment option in mind, the challenge is to determine whether therapeutic strategies should systematically follow a step-by-step approach, or whether some categories of patients deserve maximal treatment from the beginning when the disease is diagnosed at a relatively advanced stage: in other words, should a newly diagnosed patient with a mMRC dyspnea grade of 3, three acute respiratory episodes in the past year and a FEV₁ at 42% predicted (i.e., GOLD D category and GOLD 3 level of airflow limitation) be treated with a long-acting bronchodilator, a dual combination (i.e., a LABA+LAMA or a LABA+ICS) or triple therapy (ICS+LABA+LAMA) right from the beginning? No study has assessed the respective benefits of these approaches in initially naïve patients, making it impossible to provide strictly evidence-based answers to this question. Therefore, the choice will result from the estimation of the added value^{Figure 3} of one strategy over the other, based on differences between available options: e.g., is it worth initiating treatment with a combination of agents that provide a 10% absolute increase in the proportion of responders in terms of dyspnea, from 40% to 50%, when compared to a single agent? Or should the prescriber keep in mind that 80% of those who would respond to the combination would also respond to the single agent? Similarly, is a 15% decrease in exacerbation frequency with a dual approach compared to mono-component sufficient to justify dual therapy as first-line option? This issue is complexified by the variability of exacerbation frequency at an individual level (18) despite the statistically significant value of past exacerbation rate for prediction of future exacerbations at a group level (19). This is where subjectivity plays a major role, explaining why hours of discussions are often required within guidelines committees to produce an apparently simple (although often challenged subsequently) recommendation based on apparently strong randomized trials. To help clinical decision-making, many studies have been performed to define the minimal clinically important differences for variables used to assess treatment efficacy (20). Various methods have been used, based on anchoring strategies and distribution analyses. Interpreting their results is complex since they can be influenced by many factors including disease severity, comparator (active agent or placebo), poor relations between clinical and lung function variables, concomitant therapy and comorbidities, context of care… The relatively limited responsiveness to treatments overall (as opposed to asthma, for instance) is another point to consider.

Network analyses where direct comparisons are lacking

Another challenge is the appraisal of possible differences between agents within a given family, when face-to-face comparisons are mostly performed with lung function rather than clinical outcomes as a primary endpoint. Extrapolation from these studies is made difficult by the poor relation between lung function and clinical improvements. As a result, it is also difficult to determine whether results obtained with a given agent or combination of agents can be extrapolated to other drugs of the same family(ies). Thus, network meta-analyses have been performed to allow indirect comparisons between agents that have not been directly compared in terms of clinical endpoints, through the combination of studies with a common comparator (21). However, although the methodology of these analyses is now better standardized, it seems difficult to ensure that patients from different studies are sufficiently similar to compare treatment effects with full confidence: subtle differences in respiratory characteristics, comorbidities, previous and concomitant treatments, disease history may induce masked residual confounding.

Guidelines paradigms and a selection of the underlying evidence

Current guidelines are more explicitly based on patients' categorization than their ancestors. Decision trees are extensively used, some guidelines aiming at presenting one single algorithm for all clinical situations (Figure 3) (22–25) while others^{Table 1} propose specific algorithms for specific patients' populations (Figure 4) (3). Altogether, differences between these documents are rather minimal, although their presentations vary (26). The main results on which these recommendations are based are the following: for exacerbation prevention, LAMAs do better than LABAs; (27) ICS+LABA combinations are superior to their mono-components in that respect (28–30), but less effective than LABA+LAMA combinations, at least in patients with no history of asthma and less than 600 eosinophils/µL (31). Bronchodilator combinations are more efficacious than mono-components in terms of lung function and clinical endpoints (32), which could relate to not only additive but also synergistic effects resulting from receptor cross-talk (33). However, this may not translate in improved exercise capacity, due to peripheral limitations related to skeletal muscle dysfunction (34). More recent studies have explored the potential of triple combinations of LABA+LAMA+ICS, especially regarding exacerbation prevention (35–37). Superiority of triple therapy has been demonstrated versus LAMA alone and LABA+ICS. Results addressing the comparison with LABA+LAMA should be published soon. In the WISDOM trial, patients have been treated with triple therapy for a few weeks before being randomly allocated between ICS withdrawal (leaving them on LABA+LAMA) and ICS continuation (leaving them on triple therapy) (38). No increase in exacerbations occurred in the withdrawal arm, although some lung function was lost. However, since this was a step-down study, it does not allow to drive conclusions as to the differences between LABA+LAMA and triple therapy when prescribed in a step-up approach.

Figure 3. Propositions of the French-Language Respiratory Society for COPD pharmacological treatment. Adapted from (22) with permission.

Figure 4. GOLD recommendations for COPD pharmacological treatment. Adapted from (3) with permission.

Importantly, interpreting the results of randomized trials requires evaluating how they can be extrapolated to the general COPD population. Unfortunately, there is still a paucity of high-level real-life research in the field of COPD pharmacological treatment. Only very few pragmatic trials have been performed (39) and rigorous observational research remains limited. Although these study designs can certainly not replace randomized controlled trials they may be of help to generate hypotheses especially regarding the best target populations and the most appropriate strategies regarding device choices (40). In that sense, they can be of use to inform guidelines (41), especially considering that patients recruited in randomized clinical trials represent only a minority of the COPD patients population, although this proportion has improved over time (42,43).

Towards (more) precision medicine

Large cohort studies have been used to identify clinical COPD phenotypes (44) associated with relevant outcomes and, possibly, specific treatment effects (19,45,46). However, in most therapeutic studies subgroup analyses aiming at identifying responders have provided disappointing results, preventing from identifying specific target subgroups. One exception is the PDE4i roflumilast, which is specifically indicated in patients with severe airflow obstruction, chronic bronchitis and exacerbations despite long-acting bronchodilators (47). Long-term treatment with azithromycin may be more effective at preventing exacerbations in older former smokers with mild airflow obstruction, but this requires prospective testing (48).

The lack of sufficient predictive ability of clinical characteristics to identify treatment responders has triggered increasing interest in the possible role of biomarkers to guide treatment choices. In particular, increased blood eosinophils (although remaining within the normal range) appear associated with a greater chance of response to inhaled corticosteroids in terms of exacerbation reduction (49). Indeed, at least in some patients (identified through their blood eosinophil count), the role of eosinophils in COPD pathophysiology is strongly suggested by the efficacy of anti-IL-5 agents for exacerbation prevention (50). However, more studies remain necessary to confirm prospectively that eosinophil-based strategies can be recommended and determine the best thresholds and the most adequate target populations, especially since blood eosinophil counts are somehow variable within individuals and merely reflect airway eosinophilia (51). Progresses in systems/network biology and medicine, made possible by advances in bioinformatics (52) and the development of biological and clinical databases, should open new avenues for translational research and therapeutic developments (53).

To conclude, although at its early stages, precision COPD care is already there. It will hopefully progress in the coming years, allowing more targeted therapeutic choices with more favorable benefit-risk profiles at the individual patient level. However, the lung damages observed in COPD prevent lung function abnormalities from being markedly reversible with current treatment options. Regenerative medicine may contribute to reverse these damages (54). Awaiting such advances, guidelines implementation needs to be improved to avoid current deviations that appear frequent (55,56) and may be associated with suboptimal patients' outcomes. This requires to better understand the factors involved in the way guidelines affect real-life prescriptions (57). Another important component of guidelines on pharmacological therapy should be a check-list of what needs to be done before changing prescribed medicines (Table 1) (22).

Table 1. Check list of questions to address before modifying pharmacological treatment for COPD. Adapted from (22) with permission.

Domain	Question
Disease assessment	Differential diagnosis
	Lung function
	Comorbid conditions
Patient's assessment	Smoking cessation
	Adherence to treatment
	Inhaler technique
Associated treatments' assessment	Therapeutic education
	Rehabilitation

Disclosure

Dr. Roche reports grants and personal fees from Boehringer Ingelheim, Pfizer and Novartis, personal fees from Teva, GSK, AstraZeneca, Chiesi, Mundipharma, Cipla, Sanofi, Sandoz, 3M, Zambon, outside the submitted work.