COPD as a complex system

A complex system is a collection of linked individual elements with so-called emerging properties that cannot be attributed to each element considered separately.6 ,7 A plane (figure 1A) is a complex system since it is formed by numerous linked elements (engines, wings, fuselage, tires, others) and has one emerging property: flying. Yet, none of the individual elements of a plane can fly it on its own. Life, health and disease are emerging properties of an extremely complex system: the human body. In this context, the emerging discipline of systems/network medicine states that diseases should be viewed (diagnosed and treated) as the consequence of one or more biological networks in the relevant organ that become disease perturbed through genetic and/or environmental pathogenic changes.8 ,9

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

A plane (A) is a complex system and flying is an emerging property of the system. To fly the plane, the pilot in the cockpit (B) uses a control panel formed by different modules that inform them of the functioning of the system and the environment.

To fly the plane safely, pilots need a ‘control panel’ (figure 1B) that allows them to visualise the status of the relevant elements of the plane (instruments) and the environmental conditions that surround it to make the appropriate decisions. We propose that, similarly to pilots, doctors caring for patients with COPD (and likely other complex diseases9) should have a ‘COPD control panel’ that allows them to visualise the status of the relevant domains of the disease (and the environment) to make the appropriate therapeutic decisions. This is exactly what a good clinician does: to integrate information coming from diverse sources (clinical, biological, radiological, etc) to make a proper diagnosis and determine the best therapy in each patient. Thus, medicine has been ‘personalised’ since its very beginnings. The challenge today is that the volume and complexity of information that the clinician has to integrate has increased in proportion to the complexity of the disease and will increase exponentially in the future as a result of high throughput technologies that will provide data on proteomics, metabolomics and genomics, and others such as molecular imaging. Thus a ‘COPD control panel’ is likely to be important in the individualised management of this and other complex diseases.

The COPD control panel: a three-domain proposal

We propose that a COPD control panel could be constructed using three disease domains (severity, activity and impact). Each of these domains contains information on a number of elements of the system (COPD) that provide complementary and relevant information for the proper management of the individual patient, either because of its prognostic implications and/or requirement for specific therapeutic intervention (figure 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Proposal for a chronic obstructive pulmonary disease control panel. For further explanation, see text. 6MWD, 6 min walk distance; CAT, Chronic Obstructive Pulmonary Disease Assessment Test; FEV₁, forced expiratory volume in 1 s; IC/TLC, inspiratory to total lung capacity ratio; mMRC, modified Medical Research Council Dyspnea Scale; PaO₂, arterial oxygen pressure.

The severity of a given disease (including COPD) is inversely proportional to the functional reserve left in the target organ.10 In COPD, FEV₁ is a good estimate of that functional reserve, although other physiological measurements such as the inspiratory to total lung capacity ratio (IC/TLC),11 arterial blood gases5 and exercise capacity11 provide complementary information that also reflect the severity of COPD, and importantly, may require specific therapeutic interventions (bronchodilator treatment/lung volume reduction surgery, oxygen therapy or rehabilitation, respectively). Hence, we propose that the ‘severity’ module of the COPD control panel includes information on FEV₁, IC/TLC, arterial oxygenation and exercise capacity. We also propose to include the number and severity of the comorbidities because of their well known prognostic impact and need for specific therapy.3 The Charlson index12 may be a good indicator of comorbidities to be included in the severity domain, but individual comorbidities present in COPD (cardiovascular disease, metabolic syndrome, and depression, etc) would be an alternative.13

The activity of a disease reflects the intensity of the biological mechanisms that cause it.10 ,14 The concept of ‘activity of COPD’ has been ignored until now.15 Importantly activity and severity do not always run in parallel. For instance, activity may be high in the early stages of COPD but severity is mild. By contrast, advanced disease may be severe but may have low disease activity due to the spontaneous and/or therapeutically induced downregulation of the biological mechanisms that caused it. The most appropriate marker of activity in COPD is an unresolved issue15 but several clinical and biological candidates may be considered. Among the former, the rate of decline of FEV₁ is an obvious one since recent research has shown that the rate of change in FEV₁ among patients with COPD is highly variable.16–18 Given that smoking is the major pathogenic mechanism in the development of COPD, continued smoking may also be considered a marker of disease activity. Another potential clinical marker of disease activity may be the frequency of exacerbations since, although they tend to increase in patients with more severe disease, the frequent exacerbator phenotype can also occur in moderate disease.19 Among potential biological markers of disease activity, recent research has shown that the persistence of systemic inflammation (as indicated by circulating leukocytes, C-reactive protein, interleukin-6 and/or fibrinogen) is associated with increased mortality and exacerbation rate,20 and that the addition of similar biomarkers improves the ability of established clinical variables to predict mortality in COPD.21 Finally, it is well known that unintentional weight loss is also associated with poor prognosis in COPD,22 although the persistence of systemic inflammation alluded to above appears to be associated with an increased body mass index (BMI).20 Hence we propose that the activity domain of the COPD control panel includes information on smoking status, FEV₁ decline, annual rate of exacerbation, BMI and selected systemic biomarkers.

The impact of any disease depends on how the patient perceives the disease and modifies his/her activities of daily living. This perception is likely to vary substantially between individuals, as it is well established in asthma.23 In COPD, we have traditionally assumed that mild disease (as assessed by FEV₁) has a minor impact on the patient, whereas the impact is much greater in severe disease. Yet, instruments that measure such an impact, like the St George's Respiratory Questionnaire24 and the COPD Assessment Test (CAT),25 have shown that the relationship between FEV₁ and health status is poor and individual variability is enormous.1 In fact, as discussed above, the new GOLD 2011 recommendations recognise this variability and propose to determine the level of symptoms as a key component of the assessment of these patients.5 Similar arguments can be applied to other domains of the disease, such as exacerbations,19 since a number of patients suffer ‘unreported exacerbations’26 and lung function is severely impaired in a subgroup of patients who had never required hospitalisation because of exacerbation.27 These observations suggest that, as described in asthma,23 ,28 there may be poor symptom perceivers among patients with COPD. Finally, it is of note that daily activities are impaired in some, but not all, patients with COPD,29 and that objectively measured physical activity is the strongest predictor of all-cause mortality in patients with COPD.30 Hence, we propose that the impact domain of the COPD control panel includes some symptom measures (modified Medical Research Council Dyspnea Scale (mMRC) and/or CAT) as well as the level of daily activity.

In summary, we propose that such a ‘COPD control panel’ provides a way to visualise the complexity of COPD, and that the combined assessment of the severity, impact and activity can best inform the physician on the most appropriate management strategies for an individual patient.5 Yet, we acknowledge that our proposal has limitations. For instance, the type of measures included in each domain needs to be discussed and validated, and how to link this type of holistic information to specific therapeutic interventions needs to be determined.

Towards personalised medicine in COPD

The practice of medicine was originally based largely on personal experience. It was only in the late 1980s when evidence-based medicine (EBM) was introduced.31 EBM has facilitated the development and refinement of clinical practice guidelines32 but has well recognised limitations. The clearest one is that randomised clinical trials, the cornerstone of EBM, study a subset of the whole population of patients, and as a result, a significant degree of extrapolation is needed. Besides, they do not take into account the individual phenotypic variation that occurs inevitably in any disease. Thus only recommendations for the general population of patients can be formulated. As a result of these limitations, there is increasing interest in studying well identified subgroups, so-called ‘clinical phenotypes’ of patients who are associated with different outcomes and/or deserve specific therapeutic interventions.33 This is not yet ‘personalised’ medicine (perhaps ‘stratified’ medicine), but it is clearly a step forward in this direction.34 In this setting, two hypothetical patients may illustrate the potential practical use of the COPD control panel.

Patient A has mild disease (GOLD grade II, no hyperinflation, normal arterial oxygen pressure, normal exercise tolerance and no comorbidities), low impact (normal mMRC and CAT scores, normal daily activity) but a high level of disease activity (current smoker, increased FEV₁ decline with two exacerbations over the past 8 months plus raised levels of a potential biomarker of disease activity). Despite the ‘mild’ clinical presentation, the evidence of increased disease activity may indicate the need for a therapeutic intervention (anti-inflammatory therapy?) that prevents future progression of the disease. By contrast, patient B is characterised by severe disease (FEV₁<50% predicted, hyperinflation, low exercise capacity, comorbidities (cardiovascular disease and obesity)) and high impact (high mMRC, low CAT, house bound). Yet, there is little evidence of disease activity (ex-smoker, constant FEV₁, no frequent exacerbations, stable BMI and normal levels of potential biomarker(s) of disease activity). In this case, the physician may need to optimise bronchodilator treatment, treat comorbidities, and provide rehabilitation but may question the need for anti-inflammatory therapy.