Introduction

Chronic obstructive pulmonary disease (COPD) is characterized by persistent airflow limitation that is usually progressive [1••]. For many years, the diagnosis, assessment of severity, and therapy of COPD has been guided primarily by the degree of airflow limitation, as assessed by post-bronchodilator forced expiratory volume in 1 s (FEV1) [2]. However, although the airflow limitation is an important characteristic of COPD, it is now widely recognized that COPD represents a spectrum of overlapping diseases that have important extra-pulmonary consequences [3]. Indeed, the clinical, physiologic and radiologic presentation of COPD varies significantly from patient to patient, even when the degree of airflow limitation is similar. Thus, COPD is a heterogeneous disease that can be characterized across multiple dimensions, which means that FEV1 alone is not sufficient for diagnosing, assessing, and managing this disease. As a result, recent guidelines have proposed that assessment of COPD should be based on the patient’s level of symptoms, future risk of exacerbations, and the identification of comorbidites, as well as the severity of the spirometric abnormality [1••].

Efficacy in the treatment of COPD can vary between patients due to the heterogeneity of the disease. However, at present, patients with COPD are still being treated with similar pharmacological strategies irrespective of heterogeneity. Identifying and classifying the clinically significant subgroups of COPD or “COPD phenotypes” may promote more personalized and effective treatment [4•]. Originally the term “phenotype” refers to the composite of the observable characteristics or traits of an organism. Phenotypes are the result of the expression of the genes of an organism, the influence of environmental factors and the interactions between these two factors. In the field of COPD, “the outward physical manifestation of patients with COPD and anything that is part of their observable structure, function, or behavior” have been described as a COPD phenotype [3]. However, not all phenotypes are clinically relevant. Therefore, from the clinical and research viewpoint, the term “COPD phenotype” needs to be refined. An international group of experts has defined the term “COPD phenotype” as “a single or combination of disease attributes that describe differences between individuals with COPD as they relate to clinically meaningful outcomes (symptoms, exacerbations, response to therapy, rate of disease progression, or death) [5••].” This definition provides a framework of categorizing unique characteristics of patients with COPD into distinct prognostic and therapeutic subgroups. Furthermore, identifying distinct subgroups of COPD may promote research into the etiological mechanisms behind the COPD phenotype, which in turn will provide information at the genotype, molecular, cellular, and phenotype levels that could be incorporated into our understanding and management of COPD.

A variety of methods have been used to explore the different phenotypes of COPD. The classic two extreme clinical phenotypes of “blue bloater” or “pink puffer” were based on rather subjective clinical assessment of patients and are not sufficient for categorizing various COPD phenotypes. Phenotypic classification based on inter-related characteristics would be theoretically attractive. In recent years, cluster analysis has been used for the classification of different clinical phenotypes of COPD. However, groups classified by cluster analysis do not prove that they represent distinct phenotypes that are clinically meaningful, and simple robust classification rules which accurately predict the phenotype for a particular patient have not been developed yet. Thus we classify the potential COPD phenotypes into one of three categories: clinical, physiologic and radiologic, and present the clinical relevance of those potential COPD phenotypes in this review.

Clinical phenotypes

Various clinical factors of COPD have been reported to be associated with disease presentations. Clinical phenotypes are determined by their symptomatic or epidemiologic significance. Individuals within a unique clinical phenotype would have a similar underlying pathophysiologic mechanism. Therefore, in many aspects, clinical phenotypes can overlap with physiologic phenotypes.

Sex

The prevalence of COPD in women is currently increasing and several recent studies have suggested that there are sex differences in the epidemiology and clinical presentation of COPD. Women may be at greater risk of smoking-induced lung function impairment for the same level of tobacco exposure. Women with COPD report more dyspnea and lower self-reported health status compared with men after adjusting for smoking burden and lung function [6]. Although they have lower mortality rates [7], rates of exacerbation of COPD are higher [8] and long-term oxygen therapy is less effective than men [9]. Higher prevalence of anxiety is also noted [10]. The biological mechanisms that explain these differences are not clear but may relate to susceptibility to the effects of cigarette smoke, decreased clearance of the toxins, and an exaggerated immune and hormonal response [11]. More research is needed to determine the implications of these differences with regard to therapy.

Body mass index

Unexplained weight loss in patients with COPD is frequent and the association between low body mass index (BMI) and poor prognosis is a common clinical observation. Weight loss is an important determinant of symptoms, disability and quality of life [12]. Furthermore, low BMI is an independent risk factor for mortality [13] and a profound decline in FEV1 over time in subjects with COPD [14]. However, BMI appears to have little impact on acute exacerbation of COPD [15]. Targeted therapy with nutritional support, pulmonary rehabilitation and subsequent reversal of weight loss for COPD patients with low BMI may promise improved outcomes, including increased muscle strength and exercise capacity, as well as increased survival [13]. However, the studies that have tested this notion have not been encouraging [16] and it appears that nutritional supplementation alone may not be sufficient. It should also be noted that weight loss is not necessarily due to inadequate nutrition. Indeed, it can arise from systemic inflammation, as discussed below. Further research investigating the role of systemic inflammation in the poor prognosis of patients with COPD who have undergone unexplained weight loss is warranted.

Chronic bronchitis

The early epidemiologic studies by Fletcher and coworkers suggested that chronic bronchitis was not associated with a decline in lung function, as measured by the annual decline of FEV1 [17]. However, subsequent studies found an association between mucus hypersecretion and a steeper decline in FEV1 [18]. In addition, chronic cough and sputum expectoration are associated with increased rates of mortality [19]. Moreover, patients with chronic bronchitis have worse respiratory symptoms, a poorer health status, greater physical activity limitation, and a higher risk of exacerbation [20]. Therefore, this group may need more directed and targeted therapy, such as anti-inflammatory treatment. This could include phosphodiesterase-4 (PDE-4) inhibitors such as roflumilast. This oral selective PDE-4 inhibitor was shown to reduce moderate to severe exacerbations treated with corticosteroids by 15–20 % and to improve pulmonary function in a subgroup of patients with chronic bronchitis, severe or very severe COPD, and a history of exacerbation [21•, 22]. Furthermore, the mode of action of roflumilast may provide a unique approach to targeting the inflammatory process underlying COPD compared with other currently available medication. Interestingly, recent studies have shown that long-term use of antibiotics can reduce the exacerbation rate [23]. These antibiotics are mostly macrolides, which may have anti-inflammatory effects in addition to their antibiotic effects. However, due to the unfavorable balance between the benefits and side effects of antibiotics, prophylactic and continuous use of these drugs is currently not recommended.

Dyspnea

Dyspnea is a cardinal symptom of COPD that can significantly impact health status of the patient. The level of dyspnea varies considerably for the same degree of airflow limitation [24]. It also correlates more significantly with survival than FEV1 [25]. Measurement of dyspnea in addition to degree of airflow limitation is now noted for predicting survival and assessing the complex systemic nature of COPD. The BODE method gives a composite score (Body mass index, Obstruction, Dyspnea, and Exercise) that is a better predictor of subsequent survival than any component singly [26].

Frequent exacerbators

Acute exacerbations of COPD (AECOPDs) are critical events in COPD and place significant socioeconomic burdens on the health care system [27]. An AECOPD is characterized by a worsening of the patient’s respiratory symptoms beyond normal day-to-day variations that necessitates a change in regular medication [1••]. AECOPDs episodes result in a significant deterioration in health status [28], accelerate the rate of decline of lung function [29], and are associated with significant mortality [30]. The incidence of AECOPDs does not have a normal distribution, and there are several reports that some patients appear to be particularly prone to suffer AECOPDs [31••]. These observations suggest the existence of an important subgroup of patients with COPD. Currently, patients with two or more AECOPDs per year are considered as “frequent exacerbators,” whose treatment may be differentiated [31••, 32]. Recent studies by Hurst et al. have shown that exacerbations become more frequent and more severe as the severity of underlying COPD increases and that the most important determinant of frequent exacerbations is a history of exacerbations [32]. Other risk factors associated with repeated exacerbations are older age, severe COPD (greater baseline dyspnea, low FEV1, and low PaO2), inflammation (greater airway and systemic inflammation), bacterial load, chronic bronchial hypersecretion and comorbidities (cardiovascular diseases, anxiety-depression, and myopathy) [31••]. The number of exacerbations and hospitalizations in these patients can be reduced by smoking cessation, influenza and pneumococcal vaccinations, and treatment with long-acting inhaled bronchodilators (with or without inhaled corticosteroids) and possibly also PDE-4 inhibitors [21•, 33, 34]. Identifying of the patients with COPD who predispose to exacerbations and closely following up them will help to decrease the exacerbations, the mortality rate, and the burden on the healthcare system.

Systemic inflammation

COPD is a chronic inflammatory condition of the airways and the lungs, and various inflammatory cells and mediators have long been considered to play a significant role in the pathogenesis of the disease. The roles of inflammation and pro-inflammatory cytokines may extend beyond the lungs and play a part in the systemic effects of the disease and associated comorbidities. Patients with clinically stable COPD exhibit low-grade systemic inflammation, such as leukocytosis and increased levels of serum C-reactive protein, interleukin 6 and tumor necrosis factor α [35], and the levels of biomarkers are further increased during exacerbations [36]. This systemic inflammation can be reduced by steroid therapy (both oral and inhaled). However, the impact of pharmacologically reducing systemic inflammation on clinically relevant outcomes of COPD remains unclear [37]. The systemic inflammation in patients with COPD varies and also appears in other chronic diseases. It is even observed in the normal process of aging. Thus, to date, it is unclear how systemic inflammation contributes to the pathophysiology of many systemic effects of COPD.

Comorbidities

COPD often coexists with comorbid diseases which may have a significant impact on prognosis [38]. Skeletal muscle wasting and dysfunction are seen in many patients with COPD. Several factors are likely to promote reduced muscle mass in COPD, including hypoxia, use of oral and inhaled corticosteroids, systemic inflammation, oxidative stress, inactivity, and poor diet. It can contribute to exercise intolerance and poor health status in patients with COPD [39]. Various methods for estimating skeletal muscle depletion can help to determine the prognosis of patients with COPD [40, 41]. For this phenotype, pulmonary rehabilitation may be helpful to improve the exercise capacity and health status in COPD [42]. Cardiovascular disease is a major comorbidity in COPD and is probably the most frequent and most important disease that coexists with COPD [43, 44]. Recently, several observational studies exploring the efficacy of statins and angiotensin converting enzyme inhibitors or angiotensin II receptor blockers revealed these drugs could reduce the overall cardiovascular risks, morbidity, and mortality of patients with COPD [45, 46]. Osteoporosis is also a major comorbidity in COPD. Along with the direct impact of osteoporosis on morbidity and mortality, vertebral fracture can result in a decreased forced vital capacity (FVC) [47]. Cigarette smoking, high corticosteroids use, low BMI, reduced mobility and muscle strength, and systemic inflammation all contribute to the development of osteoporosis in COPD. Thus, systemic corticosteroids should be avoided if possible, and bisphosphonates as well as pulmonary rehabilitation may be helpful. In addition, anxiety/depression, metabolic syndrome/diabetes and lung cancer are frequently seen in patients with COPD. All of these are associated with a poor prognosis [43, 48, 49]. Most of comorbidities correlate poorly with the degree of airflow limitation. Although these comorbidities could be caused by either smoking or aging, they may also develop as a result of the systemic inflammation, which is the extra-pulmonary effect of COPD [35]. However, in general, COPD and its comorbidities do not influence each other for treatment.

Overlap syndrome

It is necessary to distinguish between asthma and COPD to provide appropriate therapy for patients. However, within the spectrum of chronic airway obstruction, there are individuals who share features of both asthma and COPD, especially in older age group. A large population study showed that 19 % of patients with chronic airway obstruction had more than one obstructive lung disease condition, and the proportion of patients who had an overlapping diagnosis rose with age [50]. Indeed, Kim et al. demonstrated that there is an intermediate type between asthma and COPD whose clinical characteristics differ significantly [51]. Jo et al. also showed that elderly subjects with obstructive lung disease could be classified into three phenotypes that did not strictly meet the criteria of either asthma or COPD [52]. Overlap syndrome is characterized by the coexistence of increased variability of airflow in a patient with incompletely reversible airway obstruction [53]. Patients with overlap syndrome report worse health-related quality of life and experience more frequent and severe exacerbations [54]. Moreover, coexisting COPD and asthma have a higher risk of death than COPD or asthma singly [55]. Overlap syndrome can develop due to an accelerated decline in lung function or incomplete lung growth in children, or both: increasing age, bronchial hyper-responsiveness, tobacco smoke exposure, asthma and lower respiratory infections/exacerbations are all significant risk factors for both accelerated loss of lung function and incomplete lung growth [53]. Since patients with overlap syndrome are usually excluded from clinical pharmaceutical trials, it is necessary to extend them to clinical trial and to recognize the overlap syndrome as a part of the chronic airway obstructive disease spectrum.

Physiologic phenotypes

Physiologic phenotypes of COPD are defined by pulmonary function as well as functional capacity of patients with COPD. They may reflect unique biologic processes and consequently potential opportunities for targeted interventions. Various physiologic measurements such as spirometric indices, lung volume, diffusing capacity and hypoxemia are included in this category.

Airflow limitation and rapid decline in FEV1

The degree of airflow limitation, as measured by FEV1, is the most common way to categorize and stage patients with COPD. However, although FEV1 is useful for predicting health status [56], the development of exacerbations [32], and mortality [57], there is only a weak correlation between FEV1 and the symptoms of COPD and health-related quality of life [58]. It is also increasingly being realized that FEV1 alone is not the best predictor of mortality [26]. Although FEV1 declines with normal aging by approximately 30 mL per year, the rate of decline in certain susceptible smokers is greater than 60 mL per year. A rapid decline in FEV1 is associated with increased morbidity, mortality and hospitalization rates [59]. Along with smoking, airway hyper-responsiveness, repeated respiratory infections, concomitant respiratory diseases, emphysema extent on computed tomography and genetic factors may contribute to the accelerated decline of FEV1, and only smoking cessation has been shown to be effective in reducing the rate of FEV1 decline [60, 61].

Bronchodilator responsiveness and airway hyper-responsiveness

Although irreversible airflow limitation is an important characteristic of COPD, the majority of patients with COPD still have meaningful bronchodilator responsiveness (BDR) [62]. BDR is usually assessed by the improvement in FEV1 or FVC after inhalation of a short acting beta-2 agonist and/or anti-cholinergics [63]. In patients with COPD, different response patterns to bronchodilator exists, such that some patients show improvement principally in expiratory flow indices (flow responders), whereas others respond by improvement of lung volume (volume responders). These different response patterns to bronchodilator may be associated with the degree of emphysema and air trapping [64]. However, BDR is poorly reproducible in patients with COPD and variable significantly from patient to patient. It also cannot predict the rate of decline in FEV1, health status impairment, and exacerbations [65]. Thus, the significance of BDR remains unclear. In contrast to BDR, airway hyper-responsiveness is an important predictor for an accelerated decline in FEV1 [66]. Indeed, airway hyper-responsiveness is an independent risk factor for the development of smoking-induced irreversible airflow limitation, and is second only to cigarette smoking as the leading risk factor for COPD [67].

Hyperinflation, low diffusing capacity, and hypoxemia

Expiratory airflow limitation causes hyper-inflated lung in patients with COPD, which results in increased functional residual capacity with consequent reduction in inspiratory capacity. Hyperinflation correlates better with exercise impairment than with airflow limitation [68], and can be used to predict mortality when the inspiratory capacity to total lung capacity ratio is employed [69]. Long-acting bronchodilator can improve dyspnea and exercise capacity by decreasing hyperinflation and increasing inspiratory capacity. Single-breath diffusing capacity of the lung for carbon monoxide (DLCO) represents the best pulmonary function test to assess the integrity of the pulmonary capillary bed [70]. A low DLCO is related to the severity of emphysema, hypoxemia, and functional impairment [71-73]. In addition, it has shown to be a predictor of mortality after lung volume reduction surgery [74]. However, a low DLCO is not a specific phenotype of COPD but reflects the functional impact of emphysema, airflow limitation, and pulmonary capillary bed destruction. The ventilation and perfusion mismatch that is caused by airway obstruction and emphysema is the main cause of hypoxemia in patients with COPD. Chronic hypoxemia is associated with the development of pulmonary arterial hypertension, secondary polycythemia, systemic inflammation, and skeletal muscle dysfunction. Consequently, it leads to a poor quality of life, exercise intolerance, increased cardiovascular morbidity, and a higher mortality rate [75]. Long-term oxygen therapy is one of the few interventions that can prolong survival in hypoxemic patients with COPD [76].

Pulmonary arterial hypertension

The development of pulmonary arterial hypertension (PAH) is a poor prognostic sign in patients with COPD as it affects both mortality and quality of life [77]. The level of pulmonary arterial pressure (PAP) is one of the best prognostic factors in COPD patients who receive long-term oxygen therapy [78]. While PAH in COPD is thought to be due mainly to pulmonary hypoxic vasoconstriction, correction of the hypoxia alone is not sufficient to improve PAH. This may be because PAH involves extensive vascular remodeling within all layers of the pulmonary arterial wall rather than just medial hypertrophy [79]. In addition, lung function parameters correlate poorly with PAP, which suggests that factors other than airways obstruction and/or loss of alveolar surface may play a role in its etiology. Indeed, disproportionate PAH are observed in COPD patients with moderate airflow limitation. Low hemoglobin concentration may be associated with elevated PAP in this group [80]. Yet, there are few therapies that have been developed for PAH in COPD. Endothelin receptor antagonists, phosphodiesterase inhibitors, and simvastatin may be helpful for PAH in COPD as well as in idiopathic PAH and need to be evaluated through clinical trials [81, 82].

Radiologic phenotypes

Recent advances in computed tomography (CT) suggest that it might be a useful tool for evaluating both qualitatively and quantitatively the severity, extent, and distribution of the disease components of COPD such as emphysema and airway wall thickening [83]. CT scanning can differentiate these underlying pathological subtypes and has emerged as an important and noninvasive tool in phenotyping COPD. Recent studies reported that radiologic characterizations of COPD are associated with various clinical outcomes, as follows [84]. However, the role of CT imaging in phenotyping COPD beyond lung function is not fully determined.

Emphysema

CT can detect earlier emphysema that can be detected by spirometry or diffusing capacity and many studies have addressed the ability of CT to accurately quantify the extent and severity of pulmonary emphysema [85]. CT emphysema severity is associated with lower BMI, worse health status, BODE index and a rapid decline in FEV1 [71, 84, 86]. Furthermore, it was shown to be predictive of a poorer pulmonary function response to treatment with a long-acting beta-agonist and inhaled steroid [87]. Besides the extent of emphysema, the distribution of emphysema such as lower predominance can predict a high risk of mortality [88, 89].

Airway disease

Airway wall thickening can be assessed quantitatively by measuring wall area and lumen area on volumetric CT. Increased airway wall thickness is associated with a worse quality of life and a poorer response to treatment with inhaled corticosteroids [71, 84]. Moreover, greater airway wall thickness and emphysema percentage are associated with frequent COPD exacerbations, independent of the severity of airflow limitation [90]. This phenotype may be predicted by a genetic marker such as ADRB2 gene polymorphism, which may be associated with COPD susceptibility and the bronchodilator response [91].

Integration of physiologic and CT data

It may be useful to integrate physiologic and CT features of COPD patients in a multidimensional approach to phenotyping patients with COPD. For example, lung volume reduction surgery (LVRS) is more effective than medical therapy for patients with predominantly upper-lobe emphysema and low exercise capacity prior to treatment [92]. By contrast, LVRS results in more mortality than medical management when it is used to treat severe emphysema patients with lower FEV1 and either homogeneous emphysema on CT or reduced DLCO [74]. In addition, combined assessment of COPD with emphysema and FEV1 can predict the responses to therapy [93].

Future perspectives

In the past decades, COPD was classified on the basis of the degree of airflow limitation. However, it is now apparent that COPD is multidimensional and complex diseases that consist of several different phenotypes (clinical, physiologic and radiologic) with distinct clinical outcomes (symptoms or quality of life, frequency of exacerbations, rate of disease progression, and mortality) and with potential interventions that are specifically targeted to defined phenotypes (Table 1). Delineating heterogeneity and complexity of COPD by describing the relationships between clinical, physiologic and radiologic domains of the disease will advance therapeutic interventions for a specific phenotype. In this way, future treatments that are targeted towards specific phenotypes may improve clinically relevant outcomes.

Table 1 Potential phenotypes in COPD and their clinical relevance
Full size table

To develop novel biomarkers and targeted therapeutic interventions in COPD, it is necessary to improve our understanding of the interrelationships between phenotypes and their environmental, genetic, molecular and cellular basis. The interactions between genetic and environmental risk factors result in different phenotypes through several molecular and cellular pathophysiologic cascades, in other words, intermediate pathophenotypes (or endotypes). These cascades involve inflammation, oxidative stress, protease/antiprotease imbalance, apoptosis, senescence, innate and acquired immunity abnormalities, and abnormal repair (Fig. 1) [95]. While it remains challenging to understand precisely how this complex mixture of factors interact, comprehensive and integrated methods (namely systems biology approaches and network medicine) can be used to unravel the complex networks that exist between genes, proteins, RNA, and small molecules that interact at biochemical and physical levels [95]. Indeed, biomarkers such as Clara cell secretory protein-16 (CC-16), surfactant protein (SP)-D and serum amyloid A (SAA) have been reported as candidate biomarkers in COPD [96-98]. However, there are no well validated molecular biomarkers related to clinically meaningful outcomes in COPD. It is anticipated that the recent technological and analytical advances in the research strategies in human disease complexity will lead us to the identification of genetic markers for risk assessment, biomarkers for diagnosis and therapeutic targets. During this process, it may possible to develop simple validated classification criteria for COPD phenotypes that would allow clinicians to treat individual patients with Personalized, Predictive, Preventive, and Participatory (P4) medicine [99••]. Recent international guidelines for COPD have suggested individualized treatment for patients with COPD according to air flow limitation, dyspnea and exacerbation risk. We expect that more potential phenotypes discussed in this review will be validated and new treatment guidelines according to each phenotype will be emerged in the future.

Fig. 1
figure 1

COPD phenotypes and their potential clinically relevant outcomes. Systems biology approaches will aid our understanding of the interrelationships between clinical phenotypes and their environmental, genetic, molecular, and cellular basis. This information will allow COPD subtypes to be identified, which will make it possible for clinicians to treat individual patients with P4 medicine. (Modified from [94, 95, 99••].)

Full size image

Conclusions

COPD is a spectrum of various syndromes that share a defining physiologic feature (airflow limitation) but differ in many respects. By understanding COPD heterogeneity, it may become possible to detect the disease earlier and to treat specifically for defined phenotypic groups, rather than for COPD in general. Eventually, improved survival and delayed disease progression could be made possible.