The present review article is a summary of a. Exacerbations of chronic obstructive pulmonary disease REVIEW. B.R. Celli* and P.J.

Transcription

1 Eur Respir J 2007; 29: DOI: / CopyrightßERS Journas Ltd 2007 REVIEW Exacerbations of chronic obstructive pumonary disease B.R. Cei* and P.J. Barnes # ABSTRACT: Exacerbations of chronic obstructive pumonary disease are of major importance in terms of their proonged detrimenta effects on patients, the acceeration in disease progression and high heathcare costs. There is sti debate about how exacerbations shoud be defined and graded, and their mechanisms are poory understood. The major causa agents are either bacteria or vira infections, or a combination of the two. Noninfective causes incude air poution and pumonary embous but, in some patients, no cause is identified. Exacerbations represent an increase in the infammation that is present in the stabe state, with increased numbers of infammatory ces (particuary neutrophis), cytokines, chemokines and proteases in the airways, and increased concentrations of certain cytokines and C-reactive protein in the bood. There are presenty no reiabe biomarkers with which to predict exacerbations. Exacerbations have a ong-asting adverse infuence on heath status. High doses of bronchodiators are the mainstay of treatment and systemic corticosteroids have some benefit. The routine use of antibiotics remains controversia but they are of benefit with exacerbations of a bacteria origin. Noninvasive ventiation is beneficia in preventing the need for intubation and its important compications but it is not certain whether its use in stabe patients prevents exacerbations. Athough important advances have been made, more effective treatments are needed in the future for prevention and treatment of exacerbations. KEYWORDS: Bacteria infection, exacerbation, heath status, infammation, vira infection AFFILIATIONS *Caritas St. Eizabeth s Medica Center, Tufts University Schoo of Medicine, Boston, MA, USA. # Nationa Heart and Lung Institute, Imperia Coege, London, UK. CORRESPONDENCE P.J. Barnes Airway Disease Section Nationa Heart and Lung Institute Dovehouse Street London SW3 6LY UK Fax: E-mai: p.j.barnes@imperia.ac.uk Received: August Accepted after revision: February STATEMENT OF INTEREST A statement of interest for this review can be found at statements.shtm The present review artice is a summary of a meeting on the Exacerbations of COPD hed in Como, Itay, in Apri The meeting was the fourth in the series entited COPD: the Important Questions. The chairs of the meeting and its participants are isted in the Appendix. The aim of the meeting was to identify important questions reated to the exacerbations of chronic obstructive pumonary disease (COPD) and to discuss future approaches based on recent and evoving research. The present artice contains the resuts of those discussions. Exacerbations of COPD are of major goba importance. They have a profound and ongasting effect on patients, resuting in poor heath status; they may acceerate the progression of the disease; and they account for a arge proportion of the increasing heathcare spending on COPD. Yet controversies remain over the definition of exacerbations, how they shoud be monitored and their underying mechanisms. Exacerbations of COPD are now recognised as important events in the natura course of COPD and this fact is underined in major internationa guideines [1, 2]. Exacerbations are an important outcome, not ony because they pose a considerabe economic burden but more importanty because repeated exacerbations of COPD ead to deteriorating heath-reated quaity of ife [3, 4] and, when associated with ventiatory faiure, to premature death [5]. There is gathering evidence that exacerbations acceerate the progressive decine in ung function in COPD patients, making their prevention even more important [4]. In genera, exacerbation frequency increases with disease severity, as represented by airfow obstruction [6], but the reationship between exacerbation frequency and forced expiratory voume in one second (FEV1) is not particuary cose and new evidence indicates a possibe roe for extrapumonary factors in the genesis of European Respiratory Journa Print ISSN Onine ISSN VOLUME 29 NUMBER 6 EUROPEAN RESPIRATORY JOURNAL

2 B.R. CELLI AND P.J. BARNES COPD EXACERBATIONS exacerbation. Indeed, the BODE index, which incudes the body mass index (B), obstruction (O), dyspnoea (D) and exercise endurance measured by the 6-min wak distance (E), is a better predictor of hospitaisation from COPD in a cohort of patients than FEV1 [7]. Over time COPD exacerbations become more frequent and more severe, and this is associated with increasing functiona impairment. Thus it is the patients with more severe COPD who are prone to more severe exacerbations and are more ikey to need hospita admission, especiay in the winter months when respiratory vira infections are common [8]. Risk factors for exacerbation reapses incude ow pre-treatment FEV1, a need to increase bronchodiator or corticosteroid use, previous exacerbations (more than three in the ast 2 yrs), prior use of antibiotics and the presence of comorbid conditions (congestive heart faiure, coronary artery disease, chronic rena or iver faiure) [9 12]. Patients with exacerbations are at increased risk of dying compared with patients who do not exacerbate or those who do so but do not require hospita admission [13]. CAUSAL AGENTS Bacteria, viruses and environmenta agents account for the vast majority of episodes of exacerbation (tabe 1). In a recent study of patients admitted to hospita with severe exacerbations, 78% of patients had evidence of vira or bacteria infection (fig. 1) [14]. However, many patients suffer from exacerbations where no specific causes can be identified. Bacteria Severa ines of evidence now impicate bacteria as an important cause of exacerbation of COPD [15, 16]. Bronchoscopic samping of the dista airways of the ung has demonstrated the presence of pathogenic bacteria in 50% of exacerbations. Acquisition of new strains of bacteria pathogens has been associated with a more than two-fod increase in the risk of exacerbation [17]. Systemic and mucosa immune responses to nontypeabe Haemophius infuenzae, Moraxea catarrhais and Streptococcus pneumoniae deveop in the majority of exacerbations associated with isoation of these bacteria from sputum. Bacteria exacerbations are associated with increased numbers of activated neutrophis in sputum that decine with treatment of the exacerbations using antibiotics [18]. Antibiotic treatment of exacerbations appears to be beneficia, especiay in patients with more severe underying obstructive disease and mutisymptom exacerbations [19]. The host pathogen interaction that determines the consequence of acquisition of strains of bacteria pathogens in a patient with COPD is compex. Increased understanding of this interaction is required to deveop an effective means of treatment and prevention of bacteria exacerbations. Pathogen factors that may determine the outcome of the host pathogen interaction incude the abiity of bacteria strains to eicit cytokines from and to invade airway epitheia ces. Bacteria coonisation is frequenty found in patients with COPD but not in matched norma smokers, and this is associated with an increase in infammatory markers in induced sputum, suggesting that bacteria coonisation may be a factor increasing airway infammation [20]. There is an association between bacteria coonisation and increased markers of infammation in sputum [18] and with the frequency of exacerbations [12]. Host factors that may be important incude the deveopment of a mucosa antibody response, as we as ceuar immunity to bacteria pathogens. Aterations in innate immunity are ikey to be of substantia importance, athough these have not been investigated. There is evidence to suggest a defect in macrophage phagocytosis in COPD patients that may resut in defective cearance of infectious agents from the ower respiratory tract [21]. Infammation is recognised as an important pathoogica feature of COPD and exacerbations are now thought to represent the cinica manifestations of increased infammation. Litte information is avaiabe about the patterns of infammation and spectrum of mediators in exacerbations due to different causa mechanisms. Viruses There is considerabe evidence that upper respiratory tract viruses may precipitate exacerbations of COPD. Approximatey 50% of exacerbations are associated with upper respiratory tract virus infections and infection with rhinovirus, respiratory syncytia virus and infuenza have been associated with exacerbations [14, 22]. The presence of an upper respiratory TABLE 1 Causes of chronic obstructive pumonary disease Causa mechanism Common Less common Bacteria Haemophius infuenzae Pseudomonas Streptococcus pneumoniaie Moraxea catarrhais Virus Rhinovirus Infuenza A and B Respiratory syncytia virus Parainfuenza virus Coronavirus Adenovirus Atypica organisms Mycopasma pneumoniae Chamydia pneumoniae Noninfective Air poution (particuates, ozone) Pumonary embous Cod temperatures Congestive cardiac faiure c EUROPEAN RESPIRATORY JOURNAL VOLUME 29 NUMBER

3 COPD EXACERBATIONS B.R. CELLI AND P.J. BARNES (IL-8) [28]. This suggests that patients co-infected with a virus and bacteria may have more severe exacerbations. Bacteria 30% Other 22% Vira 23% Bacteria and vira 25% FIGURE 1. Causa mechanisms in chronic obstructive pumonary disease. Data are derived from [14]. tract infection eads to a more severe exacerbation and a onger symptom recovery time at exacerbation [23]. Increased symptoms induced by virus-associated exacerbations appear to ast onger than bacteria exacerbations [8]. More than 60% of exacerbations in COPD are associated with the symptoms of a common cod [23]. At east one virus is detected by PCR in 64% of exacerbations of COPD patients and these patients have a higher exacerbation frequency than patients in whom viruses are not detected [8]. COPD patients with a history of frequent exacerbations may be more susceptibe to respiratory vira infections, athough the nature of this susceptibiity has not yet been defined. It is possibe that upreguation of the interceuar adhesion moecue-1, which acts as a receptor for rhinoviruses, in airway epitheia ces is important [24] and is upreguated by rhinovirus infection via the activation of the transcription factor, nucear factor (NF)-kB [25]. Rhinovirus can be recovered from the sputum more easiy than from the upper airways, indicating that these viruses directy infect the ower respiratory tract [26, 27]. Virus infections, ike bacteria infections, are associated with increased numbers of neutrophis in the sputum, and aso with an increase in the numbers of eosinophis, suggesting that different infammatory mediators are invoved [14]. There is increasing recognition that many patients with exacerbations have concomitant vira and bacteria infection. Approximatey 25% of patients admitted to hospita with an exacerbation of COPD had co-infection with bacteria and viruses, and these patients had more severe exacerbations, as measured by ength of hospitaisation [14]. In a recent survey,,70% of exacerbations were associated with an increase in H. infuenzae and those patients who had concomitant rhinovirus infection (detected by PCR) had a greater fa in FEV1 and rise in serum intereukin (IL)-6 and sputum CXCL8 Noninfective mechanisms Epidemioogica studies have shown that hospita admissions with COPD exacerbations increase sighty with a rise in atmospheric eves of suphur dioxide, ozone, nitrogen dioxide and particuates [29]. There is convincing evidence that exposure to particuates with a 50% cut-off aerodynamic diameter of 10 mm is associated with increased hospita admissions in COPD patients [30]. Particuates induce oxidative stress and, in vitro, this eads to activation of NF-kB, histone acetyation and increased expression of CXCL8 [31]. This is enhanced by adenovira eary region 1A, suggesting that there may be an interaction between virus infection and air poution in triggering exacerbations [32]. Low temperatures may aso be associated with exacerbations of COPD. Reduced temperatures in the bedroom and outside air have been associated with fas in the ung function of COPD patients and an increased frequency of exacerbations [33]. The mechanisms are not yet understood but may reate in part to increased susceptibiity to upper respiratory tract virus infections in cod weather. In patients admitted to hospita with severe COPD exacerbations of unknown cause, 25% had pumonary emboism confirmed by spira computerised tomography [34]. Heart faiure may aso ead to a symptomatic exacerbation of COPD, athough it may be difficut to differentiate the symptoms of increased heart faiure from those of a COPD exacerbation [35]. MECHANISMS Pathoogica and ceuar changes There is an increase in sputum neutrophi numbers in exacerbations of COPD and a recent biopsy study demonstrated an increase in neutrophis in bronchia biopsies, athough they are rarey seen in the stabe state [36]. In bronchia biopsies of patients with mid exacerbation of chronic bronchitis, an eosinophiia exists that is associated with increased expression of CCL5 (aso known as RANTES) [37 39]. Viray induced exacerbations are aso associated with increased eves of eosinophis in sputum, as discussed previousy [14]. Vira infections induce the expression of CCL5 in airway epitheia ces [40]. CCL5 may act synergisticay with CD8+ ces to enhance the apoptosis of viray infected ces, thus eading to increased tissue destruction [41]. There is aso an increase in the concentration of the eastoytic enzyme matrix metaoproteinase-9 and a decrease in its major inhibitor, tissue inhibitor of metaoproteinase-1, in sputum during exacerbations [42]. This is consistent with an increase in urinary desmosine, which is an indicator of eastoysis [43]. This may provide a causa ink between exacerbations and acceerated decine in ung function. Moecuar mechanisms The ceuar and moecuar mechanisms of exacerbations are sti not we understood but most evidence suggests that they are due to a further ampification of the infammatory process triggered by bacteria, viruses and noninfective stimui, such as air poution (fig. 2). There is a marked increase in neutrophi numbers (accounting for the change in sputum coour) and cytokines that are increased in stabe COPD (tumour necrosis 1226 VOLUME 29 NUMBER 6 EUROPEAN RESPIRATORY JOURNAL

4 B.R. CELLI AND P.J. BARNES COPD EXACERBATIONS Bacteria Viruses Noninfective further reduced by the increased oxidative stress and NO production during an exacerbation [62]. This may expain why high doses of corticosteroids have reativey ow efficacy in the treatment of COPD exacerbations [63]. Macrophages NK-kB NK-kB TNF-a CXCL8 IL-6 Neutrophis Oxidative stress Epitheia ces BIOMARKERS OF EXACERBATIONS A biomarker refers to the measurement of any moecue or materia (ces and tissue) that refects the disease process. In COPD, severa types of biomarker have been measured that are reated to disease pathophysioogy and the infammatory and destructive process in the ung, but there are few measurements during exacerbations [64]. Biomarkers have been measured in bood, urine, sputum, bronchoaveoar avage and exhaed breath. So far there is itte information about whether biomarkers can predict exacerbations or distinguish between different causa mechanisms for exacerbations, thus providing a means of guiding the therapy. FIGURE 2. Pathogenic mechanisms in chronic obstructive pumonary disease. Bacteria, viruses and other factors, such as air poution, may activate the transcription factor nucear factor (NF)-kB in airway epitheia ces and macrophages, which then reease infammatory cytokines, incuding intereukin (IL)-8 (CXCL8), which in turn attracts neutrophis, tumour necrosis factor (TNF)-a and IL-6, both of which ampify infammation. Increased numbers of neutrophis generate oxidative stress, which further ampifies the infammatory process. factor (TNF)-a and IL-8) are further increased during the exacerbation [44, 45]. There is aso increased expression of the chemokine CCL5, which attracts T-ces and eosinophis, and CXCL5 (previousy known as epitheia neutrophi-activating peptide (ENA)-78), which attracts neutrophis [36, 46]. IL-6 is increased in sputum, breath and pasma of COPD patients during exacerbations [8, 47, 48]. These mediators are reguated primariy by NF-kB and there is evidence for its activation in macrophages in COPD exacerbations [49]. Rhinoviruses and bacteria, such as H. infuenzae, may activate NF-kB through To-ike receptors (TLR), which recognise surface pathogens (predominanty TLR2, TLR3 and TLR4) [50]. Nontypeabe H. infuenzae activates NF-kB synergisticay with TNF-a through the activating enzyme inhibitor of NF-kB kinase-2 (IKK2) and activation of a p38 mitogen-activated protein kinase (MAPK) pathway [51]. There is an increase in eukotriene (LT)B 4 (chemotactic for neutrophis and invoved in T-ymphocyte activation) in the sputum and exhaed breath [52, 53]. Oxidative stress is aso increased, as evidenced by an increase in exhaed 8-isoprostane and hydrogen peroxide and this persists over severa weeks [53 55]. Nitrative stress may aso be increased in exacerbations, as exhaed nitric oxide (NO) eves are higher during exacerbations [56 58]. The increased formation of peroxynitrite may be important in ampifying infammation during an exacerbation. The proonged increase in COPD exacerbations may refect the perpetuation of infammation through an interaction between oxidative stress and NF-kB activation, and the estabishment of ampification oops. COPD exacerbations respond reativey poory to corticosteroids indicating a degree of steroid resistance, which may refect a reduction in histone deacetyase (HDAC)-2 activity in the airways [59, 60]. The reduction in HDAC in COPD provides a moecuar mechanism for ampification of infammatory gene expression in COPD [61] and may be Bood Various infammatory biomarkers, incuding TNF-a, IL-6 and C-reactive protein (CRP) are increased in the pasma of patients with stabe COPD, but it is uncertain whether or not they predict exacerbations. There is an increase in pasma concentrations of infammatory markers during acute exacerbations, incuding TNF-a and IL-6, and this may represent overspi from the ung [65, 66]. A recent study measured 36 pasma biomarkers during acute exacerbations in 90 patients with COPD and compared concentrations with the baseine state [67]. The most seective biomarker turned out to be CRP, athough it was not specific for an exacerbation. The interreationships between the various biomarkers suggested that there was an increase in monocyte and ymphocyte activation during an exacerbation. None of the biomarkers proved to be usefu in predicting the cinica severity of an exacerbation. Pasma eptin concentrations are increased during an exacerbation and this may indicate negative energy baance during an acute exacerbation [65, 68]. Systemic oxidative stress is aso increased during exacerbations, with increased concentrations of markers of oxidative stress and reduced antioxidants [69]. Sputum Severa infammatory markers are increased in induced sputum of COPD patients during acute exacerbations and fa during recovery. Increased sputum concentrations of TNF-a, IL-6, CXCL8, LTB 4 and endothein-1 have been reported [45, 47, 70 72]. The increased puruence and coour change of sputum during exacerbations refects the increased numbers of neutrophis containing the green pigment myeoperoxidase [73]. The coour may be usefu in guiding whether antibiotic therapy is ikey to be effective [18]. Exhaed NO The eves of exhaed NO measured at the mouth are usuay norma in patients with COPD [74] but when exhaed NO is partitioned by the mutipe fow technique, periphera NO (incuding sma airways and ung parenchyma) is increased, whereas bronchia NO is norma [75]. However, exhaed NO is increased during exacerbations of COPD [56 58]. This may refect increased nitrative stress during exacerbations and this hypothesis is supported by the demonstration of increased c EUROPEAN RESPIRATORY JOURNAL VOLUME 29 NUMBER

5 COPD EXACERBATIONS B.R. CELLI AND P.J. BARNES Exhaed 8-isoprostane pg ml *** Exacerbation 2 weeks 8 weeks numbers of nitrotyrosine-positive ces (as a resut of increased nitrite and peroxynitrite formation) in induced sputum during an exacerbation compared to the stabe state [72]. Exhaed breath condensate Severa infammatory mediators and markers of oxidative stress have been measured in exhaed breath condensate. There is a high variabiity in the measurement and diution resuts in very ow concentrations of many mediators, making this a difficut measurement [76]. However, this is a noninvasive measurement and is suited to seria measurements during an exacerbation of COPD. An increased concentration of hydrogen peroxide in exhaed breath condensate has been reported during exacerbations of COPD, suggesting increased oxidative stress [54, 55]. In a genera practice-based study, an increase in LTB 4 and 8-isoprostane, a marker of oxidative stress, were found in exacerbations of COPD and these abnormaities took severa weeks to normaise (fig. 3) [53]. Acute exacerbations of COPD are associated with an increase in severa cytokines in exhaed breath condensate, incuding TNF-a, IL-1b, IL-6 and CXCL8 [77]. CLINICAL CONSEQUENCE OF EXACERBATIONS Heath status Athough it is recognised that exacerbations are associated with considerabe symptomatic and physioogica deterioration, the burden imposed on patients may be underestimated. Interestingy, patients are not famiiar with the term exacerbation as used by heathcare professionas and, if given a choice, they may use their own words to describe the worsening of symptoms. The terminoogy used by patients is extremey varied but consistent for each patient. The way in which patients perceive exacerbations and how exacerbations affect patients have been greaty heped by the systematic scoring of heath status questionnaires. The appication of such toos to study exacerbations has provided major new insights, athough it sti remains a hypothesis-generating exercise. It has been shown that the frequency of exacerbations accounts for some of the differences in heath status between patients [3] # FIGURE 3. Increased oxidative stress in a chronic obstructive pumonary disease (COPD) exacerbation resuts in increased 8-isoprostane in exhaed breath condensate of COPD patients treated with antibiotics. 8-Isoprotane concentrations fa to stabe eves over 8 weeks. ***: p,0.001; # :p and for some of the deterioration in heath status scores within patients over time [78]. Reducing exacerbation rate appears to reduce this deterioration. The mechanism for this is not cear but is ikey to be associated with the arge and sustained effect of a singe exacerbation on heath [78]. Even if the patient fais to improve by a very sma degree foowing each exacerbation, the accumuated effect coud account for the progressive deterioration attributabe to COPD exacerbations [4]. During an exacerbation, poorer heath status is associated with an increased ikeihood of further exacerbations within 6 months [78]. Thus, the composite data from patient-centred studies and, mosty, pharmacoogica trias indicate an important roe for repeated exacerbations on the heath status of patients with COPD. Furthermore, in one ong-term randomised tria comparing inhaed corticosteroids with pacebo, there was no significant change in the rate of decine of FEV1 whereas, in the group receiving active medication, there was a sower deterioration in the rate of worsening of the heath status scores over time that appeared to be reated to a decrease in exacerbation frequency [79]. Physioogica consequences There is imited information describing the physioogica changes that occur during exacerbations of COPD that do not require mechanica ventiation. The data from patients requiring mechanica ventiation indicates the presence of increased centra drive, dyspnoea, tachypnoea, reduced tida voume and deveopment of hypercapnoeic respiratory faiure, whie ventiation/perfusion matching seems to be reativey preserved [80, 81]. Two recent studies have increased present knowedge about the changes in ung mechanics and its reation to dyspnoea during exacerbation of COPD [82, 83]. In both studies, ung mechanics, incuding spirometry, inspiratory capacity and dyspnoea, were recorded during recovery from an exacerbation. Consistent reduction in dyspnoea was seen as resoution of the exacerbation. In both studies, the FEV1/forced vita capacity (FVC) ratio and expiratory fow imitation changed reativey itte throughout the study period. In contrast, both studies demonstrated that changes in ung voume rather than airfow resistance predominated. During hospitaisation in one of the studies, there was rapid and significant improvement in dyspnoea, respiratory rate, inspiratory capacity, puse and FVC, and fewer changes in FEV1. The FEV1 and ung voume improved over time. An exacerbation of COPD appears to be characterised by increased centra drive, decreased inspiratory capacity and decreased inspiratory musce force, perhaps secondary to dynamic hyperinfation. Few studies have attempted to reate the changes in physioogy to changes in the infammatory process that are thought to occur during the episodes. There is an association between increased serum eves of IL-6 and LTB 4 and the magnitude of dyspnoea, respiratory rate and inspiratory capacity [84], suggesting that it may be possibe to detect serum changes that refect the infammatory burden of the exacerbation. This finding has been confirmed in sputum where changes in concentrations of IL-6, CXCL8, TNF-a, myeoperoxidase, neutrophi eastase and LTB 4 suggest an infammatory burst [45, 47, 70]. Once patients deveop ventiatory insufficiency, the prognosis is very poor, with,50% mortaity at 2 yrs [5] VOLUME 29 NUMBER 6 EUROPEAN RESPIRATORY JOURNAL

6 B.R. CELLI AND P.J. BARNES COPD EXACERBATIONS Economic consequences Some studies have determined that hospitaisation costs represent 40 57% of tota direct costs generated by patients with COPD, and this percentage may be as high as 63% in severe patients [85]. The average cost of hospitaisation for COPD in a cohort of severe patients was estimated to be,us$7,000. Since acute exacerbations are the main cause of hospitaisation among COPD patients, it is evident that the economic burden of acute exacerbations is considerabe. Observationa studies performed in primary care centres observed that 16 22% of patients having exacerbations were admitted during 1 yr [86]. The costs of exacerbations that require hospitaisation increase dramaticay compared with those that can be treated in an ambuatory setting. An anaysis derived from a cinica tria in patients with COPD demonstrated that the 15% of exacerbations requiring hospita admission generated 90% of the costs associated with exacerbations [87]. In a recent study in primary care in Spain, the mean tota cost of an acute exacerbation of COPD was estimated to be US$159, with the main part being due to hospitaisations, which represented 58% of the tota cost, foowed by the drug costs amounting to 32% of the tota [88]. However, these costs may not be appicabe to other countries because of the differences in reference prices, management practices and heathcare systems. Faiure impies a cost that is three times higher than the cost of management of the exacerbation, particuary due to the high cost of hospitaisation. If the percentage of reapses coud be reduced, especiay in severe cases, or if switching a patient from parentera to ora therapy coud reduce the ength of hospita stay, vauabe resources coud be saved. This is particuary important considering that a recent study demonstrated that patients with stage IV COPD (FEV1,35% predicted) had a significanty greater percentage of faiures than successfu exacerbations, with 52% of faiures requiring hospitaisation [87]. The costs of managing acute exacerbations of chronic bronchitis are high, particuary because of the high costs associated with reapse [89, 90]. Strategies to improve the outcome of ambuatory treatment of exacerbations shoud be very cost-effective, especiay in more severe patients who are at increased risk of being admitted to hospita as a consequence of therapeutic faiure. COPD comorbidities and exacerbations COPD is a condition that becomes cinicay apparent in midto-ate ife. Comorbidity is reativey common in patients with COPD and this raises the issue as to whether such a comorbidity is age-reated, reated to a common factor, such as smoking and cardiovascuar disease, due to the effect of drugs ike corticosteroids and the deveopment of diabetes, or a refection of the increase in systemic infammatory cytokine concentrations, which are a feature of COPD with systemic invovement. Data are emerging that the same infammatory mediators are centra to the pathogenesis of other diseases and, to iustrate this, focus was directed towards type 2 diabetes and cardiovascuar diseases. Comorbidities are reevant to exacerbations of COPD as there may be an interaction between the comorbid condition and the severity of the exacerbation. For exampe, exacerbations may worsen heart faiure, whereas heart faiure may in turn increase the cinica severity of an exacerbation by increasing the degree of dyspnoea. There is increasing evidence that the cytokines systemicay reeased in COPD, such as CRP, IL-6 and TNF-a, pay a key roe in the pathogenesis of insuin resistance and type 2 diabetes [91]. Eevated eves of acute-phase proteins and pasminogen activator inhibitor-1 predict deveopment of type 2 diabetes [92]. Indeed, the odds ratio for deveoping type 2 diabetes in femaes with COPD is 1.8 [93]. The mechanism may occur via a common pathway. For exampe, in COPD, there is a genera metaboic shift toward cataboism and an increase in resting energy expenditure [94]. The pasma concentration of TNF-a in patients with COPD and manutrition is increased [95] and TNFa decreases periphera insuin action [96]. In addition, oxidative stress is increased in COPD, particuary during exacerbations, and has aso been impicated in insuin resistance [97]. Cardiovascuar disease Recent studies have shown that there is a 3.14 and 2.57 incidence ratio of myocardia infarction or first stroke during the first 3 days after a systemic respiratory infection, suggesting a cose association [98]. Systemic respiratory infections are associated with periphera acute-phase responses, incuding the production and reease of TNF-a, IL-6 and CRP. A subsequent study indicated that the concentrations of IL-6 and CRP in pasma were significanty reated to an increased risk of coronary heart disease in maes and femaes, and that an increase in soube TNF-a receptors was associated with an increased risk in femaes [99]. Cardiac troponin-1 is often raised during acute exacerbations that require hospitaisation but are without other evidence of an acute coronary syndrome, indicating that an acute exacerbation has detrimenta effects on cardiac musce [100]. Eevation of the CRP is an independent predictor of myocardia infarction. The reative risk for subjects with CRP concentrations.3 mg?l -1 is 1.79 compared with those in whom the CRP is,1 mg?l -1 [101]. CRP increases the expression of interceuar adhesion moecues, induces monocyte chemoattractant production, activates compement and mediates ow density ipoprotein uptake by macrophages. In addition, CRP may deposit directy into the arteria wa during atherogenesis to create foam ces, which are the buiding bocks of atheroscerotic paques. Furthermore, there is an increased pro-thrombotic state in patients with COPD, as shown by increased circuating fibrinogen eves during acute exacerbations [66]. The interaction between comorbidities and COPD is ony just beginning to be expored but it is cear that surviva in COPD is better predicted by variabes other than simpy the degree of airfow imitation [102]. How these factors reate to the deveopment and perpetuation of exacerbations remains to be eucidated but they are ikey to be very important in patients with COPD. THERAPY Bronchodiators Bronchodiators have important roes in both the prevention and treatment of acute exacerbations of COPD. In exacerbations, bronchodiators are the primary treatment modaity designed to aeviate patient symptoms, improve physioogica state and prevent or reverse respiratory faiure. Unike noninvasive ventiation, the use of bronchodiators has not c EUROPEAN RESPIRATORY JOURNAL VOLUME 29 NUMBER

7 COPD EXACERBATIONS B.R. CELLI AND P.J. BARNES been shown to proong surviva. Nevertheess, severa important questions remain about bronchodiators in the setting of acute exacerbations. There is no difference in outcome between nebuised short-acting b 2 -agonists and ipratropium bromide in acute exacerbations and no evidence that the combination of these two drugs is any more effective [103]. This contrasts with the greater efficacy of these combinations in stabe disease [104]; it is not known whether sequentia administration differs from concurrent deivery. Patients admitted for acute exacerbations are usuay aready taking chronic bronchodiator treatment but there is no convincing evidence for the deveopment of toerance. The question of how effective rapid-acting bronchodiators are when added to ong-acting agents needs to be evauated in future studies. Severa studies suggest that b-agonists and antichoinergic bronchodiators can prevent exacerbations [ ]. These studies show that ong-acting agents appear to be more effective than shorter-acting agents in reducing exacerbation frequency [110]. The mechanism by which this benefit is achieved has not yet been fuy deineated. It is possibe that bronchodiators reduce exacerbation frequency by a common physioogica effect through defating the ungs and reducing dynamic hyperinfation. Why short-acting agents are ess effective, however, is not cear, athough this may be due to ess effective bronchodiatation and fuctuating airway function as the drug effects wear off. An effect mediated through nonbronchodiator mechanisms, such as an anti-infammatory effect, is aso pausibe for ong-acting b 2 -agonists [111] and possiby for tiotropium [112]. Corticosteroids The infammation in COPD is argey resistant to the antiinfammatory effects of corticosteroids and this appears to be due to a reduction in HDAC-2, which is required for corticosteroids to switch off infammatory genes [62]. However, systemic corticosteroids reduce both systemic infammation and certain types of airway infammation. Prednisoone (30 mg for 2 weeks) reduces sputum eosinophi counts by six-fod in patients diagnosed as having COPD but does not have a significant effect on other infammatory ces [113]. A simiar treatment protoco reduces pasma CRP eves by.60% in COPD patients [114]. A very high dose of corticosteroids (prednisone mg q.d.) reduced pasma CRP by 80% during an acute exacerbation after 24 h of therapy and this effect was maintained for a week [115]. Ora corticosteroids have beneficia, abeit sma, effects in the management of acute exacerbations. In one tria, prednisoone (30 mg q.d. p.o. for 14 days) shortened the ength of hospitaisation by 2 days (29% reduction compared with pacebo), improved FEV1 by 60 ml per day (or 7% pred overa) and acceerated recovery from symptoms [116]. In another study, systemic corticosteroids improved oxygenation and dyspnoea during acute exacerbations [117]. Treatment with systemic corticosteroids for 2 weeks resuted in higher FEV1 vaues and ower treatment faiures than pacebo over 6 months in patients admitted to hospita with COPD exacerbations [118]. Therapy for 8 weeks produced no incrementa benefits above those achieved by a 2-week course. These data are consistent with the observation that 60 80% of COPD patients recover from their exacerbation by day 15 [23]. The majority of patients with COPD probaby ony require 2 weeks of ora corticosteroids, but a few with more severe exacerbations may require a onger duration of exposure. Systemic corticosteroids are aso usefu in preventing hospitaisations and reapses in those who visit emergency departments ony during fare-ups. Prednisone (40 mg q.d. p.o. for 10 days) reduced reapses (either an unschedued physician visit or repeat emergency visit) by 37% and the risk for hospitaisation by 47% (athough the atter did not reach statistica significance due to insufficient power of the study) [119]. Interestingy, the ony subgroup in which prednisone reduced reapses was the group of patients that were aso receiving inhaed corticosteroids (reative risk of 0.44). In one study, nebuised budesonide (2 mg, 6 houry) had a simiar cinica effect to ora prednisone (30 mg q.d.) in the management of more acute exacerbations [120]. One concern about the use of systemic corticosteroids in exacerbations is the possibiity of diagnostic confusion with community-acquired pneumonia. However, there is no evidence that systemic corticosteroids worsen heath outcomes in community-acquired pneumonia if appropriate antibiotics are used. Indeed, a recent study has suggested that systemic corticosteroids may even reduce morbidity and mortaity in community-acquired pneumonia [121]. In summary, there is convincing evidence that systemic corticosteroids improve heath outcomes during COPD exacerbations. Their use improves heath status, reduces dyspnoea, acceerates recovery of ung function, reduces ength of hospitaisations and prevents reapses, which are very common in moderate-tosevere COPD. Whether the addition of inhaed corticosteroids at the time of hospita or emergency department discharge can provide incrementa benefits is unknown. Severa recenty reported arge muticentre trias evauated the roe of inhaed corticosteroids in preventing or sowing the progressive course of symptomatic COPD [79, ]. In a of these trias, whie there was no evidence for any reduction in disease progression, exacerbations were reduced by 12 25% depending on the severity and the definition used. These findings have been confirmed in studies of fixed combination inhaers containing acorticosteroids and a ong-acting b 2 - agonist. Both futicasone/sametero and budesonide/formotero combinations reduce exacerbation frequency to a greater extent than using a corticosteroid or ong-acting b 2 -agonist aone [ ]. Retrospective anayses of arge databases suggest a possibe effect of inhaed corticosteroids on reducing a-cause mortaity in COPD patients and, by impication, some effect in reducing exacerbations [ ]. This prompted the initiation of a arge prospective tria to expore the effect of inhaed corticosteroids and a combination inhaer on mortaity. Preiminary resuts of this tria indicate a 17.5% reduction in death over 3 yrs in patients receiving the futicasone/sametero combination compared with pacebo, athough this did not quite reach significance (p50.52) [131]. Patients treated with inhaed futicasone aone showed no such reduction in mortaity. In the same tria, there was a 25% decrease in moderate-to-severe exacerbations compared with futicasone/ sametero with pacebo regardess of the airfow severity. Interestingy, there were more investigator-reported pneumonias in the two arms of treatment with inhaed corticosteroids. This finding deserves further study to estabish the nature of the reationship VOLUME 29 NUMBER 6 EUROPEAN RESPIRATORY JOURNAL

8 B.R. CELLI AND P.J. BARNES COPD EXACERBATIONS Antibiotics The debate about whether or not to give an antibiotic has been driven by studies of the bacterioogy of COPD both during and between exacerbations [12, ] and by severa pacebocontroed antibiotic trias. The resuts of these trias are not concordant so there is continuing uncertainty on this topic [ ]. The recent avaiabe data support the prescription of antibiotics in the presence of puruent sputum, athough a proportion of the atter wi not benefit [ ]. Specific properties of new antimicrobias, the demonstration of different bacteria isoates in those with different ung function, identification of a subgroup of patients who are more ikey to fai with conventiona antibiotic therapy and the resuts of comparative antibiotic studies have been used to support different antibiotic strategies [135, 143, 144]. The guiding principe for the use of antibiotics remain that of knowedge of the oca prevaence of bacteria in the popuation of patients being evauated, as has been suggested by the American Thoracic Society (ATS)/European Respiratory Society (ERS) and Goba Initiative for Chronic Obstructive Lung Disease (GOLD) guideines [1, 2]. Mechanica ventiation Noninvasive or invasive (via endotrachea intubation) mechanica ventiation is a form of ife support to be instituted unti the cause underying the acute respiratory faiure is reversed with medica therapy [145]. Noninvasive ventiation shoud be used whenever possibe as it has been shown to be an effective treatment for respiratory faiure during acute exacerbations of COPD [146, 147]. Indeed, 1-yr mortaity is ower in patients receiving noninvasive ventiation for exacerbations of COPD compared with both conventiona mechanica ventiation and optima medica therapy aone [148, 149]. The institution of mechanica ventiation shoud be considered when, despite optima medica therapy and oxygen administration, one of the foowing persists: 1) moderate-to-severe dyspnoea with evident use of accessory musces and abdomina paradox; 2) moderate-to-severe acidosis (ph,7.36) and hypercapnia (carbon dioxide arteria tension (Pa,CO 2 ).6 8 kpa); and 3) respiratory frequency.24 breaths?min -1. Compared with invasive mechanica ventiation, noninvasive ventiation owers infectious compications and reduces hospitaisation time, resuting in considerabe reductions in heathcare costs [150, 151]. It is not yet certain whether the use of noninvasive ventiation at home may prevent exacerbations. FUTURE THERAPIES Current pharmacoogica treatments have a modest effect in treating and preventing exacerbations, indicating the need for new therapies. Since exacerbations, from whatever cause, appear to represent an increase in ongoing infammation, more effective anti-infammatory treatments may not ony treat exacerbations but may aso prevent them, which woud be very cost-effective. Severa potentia anti-infammatory treatments are now in cinica deveopment [152]. Targeting the causa mechanisms of COPD is a ogica approach but it is difficut to differentiate between vira and bacteria infections based on cinica features aone. More rapid diagnostic approaches are therefore needed; for exampe, the use of PCR to diagnose specific viruses and bacteria in sputum sampes, thus aowing more ogica treatment. This may then guide the use of appropriate antibiotics and, in the future, antivira treatments. The future use of biomarkers in the bood or breath to predict the evoution of an exacerbation may aow earier intervention to prevent the deveopment of a severe exacerbation. Athough severa new casses of drug are in deveopment for COPD, there are few cinica trias and itte information about whether they prevent exacerbations. Unfortunatey, no anima modes of COPD exacerbations exist that coud be used to expore this in pre-cinica studies and this is an area for future research using acute infections in addition to cigarette smoke exposure. The most cinicay advanced new anti-infammatory treatments are phosphodiesterase (PDE)4 inhibitors; these have a broad spectrum of anti-infammatory effects that are reevant to COPD [153] and reduce emphysema in an anima mode of COPD [154]. Athough the dose of PDE4 inhibitor that can be used is imited by side-effects, such as nausea, vomiting and diarrhoea, PDE4 inhibitors not ony improve ung function but aso reduce the frequency of exacerbations in COPD patients. One such PDE4 inhibitor, rofumiast (500 mg q.d. p.o.), was found to significanty decrease the frequency of exacerbations by 34% over a 6-month period [155]. This effect was argey due to a reduction in mid exacerbations, defined by increased rescue bronchodiator use over 2 days, without any change in the use of ora steroids or hospitaisation. Another study using a different PDE4 inhibitor, ciomiast (15 mg b.i.d. p.o.) showed that a significanty greater percentage of patients were free of exacerbations over a 6-month period in the ciomiast (74%) compared with the pacebo group (62%) [156]. Additiona studies are now needed over onger periods of time in more severe patients who have frequent exacerbations. There is a striking increase in sputum neutrophis during acute exacerbations, suggesting that any drug that inhibits neutrophi recruitment or activation may be of cinica benefit. Chemotactic factors for neutrophis incude LTB 4 and CXCL8, and the eves of both of these were increased during exacerbations [45, 52]. Specific LTB 4 inhibitors and anti-il-8 antibodies reduce the neutrophi chemotactic activity of COPD sputum in the stabe state [157]. LTB 4 antagonists may therefore be used to prevent and treat exacerbations but no cinica studies have been reported. CXCL8 and reated CXC chemokines, incuding CXCL1 and CXCL5, which are eevated in COPD, act through a common receptor, CXCR2, for which sma moecue inhibitors have now been deveoped [158]. CXCR2 antagonists are currenty entering into cinica trias in COPD patients and it is predicted that they may be usefu in reducing and treating exacerbations. Oxidative stress is increased during COPD exacerbations and acts as an ampifying mechanism for infammation. This suggests that antioxidants may be usefu in preventing and treating exacerbations. Athough a meta-anaysis of cinica studies with N-acetycysteine, which has antioxidant properties, showed a significant reduction in exacerbation frequency of,25% [159], a arge pacebo-controed tria did not show any significant reduction in the number of exacerbations [160]. However, when patients not treated with inhaed corticosteroids were anaysed, a significant reduction was noted. It is cear that more potent antioxidants are needed in the future. c EUROPEAN RESPIRATORY JOURNAL VOLUME 29 NUMBER

9 COPD EXACERBATIONS B.R. CELLI AND P.J. BARNES Other approaches that are currenty being expored are the use of inhibitors of NF-kB (IKK2 inhibitors) and p38 MAPK inhibitors; both these kinases may mediate the increased infammation during exacerbations. TOWARDS A NEW DEFINITION OF EXACERBATIONS There is currenty no genera agreement on the definition of a COPD exacerbation. A standardised definition coud provide benefits to patients, physicians, researchers and other heathcare payers (e.g. Primary Care Trusts, insurance companies, etc.) and decision-makers. It woud aso hep patients to optimise their primary and emergency management and woud guide physicians in seecting appropriate pharmacoogica and nonpharmacoogica therapeutic interventions. An agreed definition woud hep in the design of randomised cinica trias and aow the resuts to be accuratey evauated and compared with other trias. COPD exacerbations have been defined according to the presence of specific signs and symptoms, changes in symptoms and the need for medica intervention, and each of these approaches has positive and negative connotations. Possiby the most semina definition of COPD exacerbation is the one provided by ANTHONISEN et a. [136], based on the presence of three specific symptoms in a patient with COPD, namey increased shortness of breath, increased sputum voume and increased puruence. Moreover, three subtypes were aso proposed (types 1, 2, and 3) according to the occurrence of a or some of the symptoms [136]. In the year 2000, an internationa pane of chest physicians proposed a second definition: a sustained worsening of the patient s condition, from the stabe state and beyond norma day-to-day variations, that is acute in onset and necessitates a change in treatment in patients with underying COPD [161]. Exacerbations were defined in terms of mid, moderate and severe according to the use of heathcare resources. The first definition is essentiay symptom based and was designed for the study of the effectiveness of therapies, such as antibiotics. The second definition is action or event-driven, which dictates a specific medica intervention. Here, instead of specifying particuar symptom criteria, the definition incudes a deterioration of respiratory symptoms that requires the use of systemic corticosteroids either with or without antibiotics and/or hospitaisation due to symptoms. These action-driven definitions may vary among regions and countries, and are heaviy dependent on the oca avaiabiity of heathcare. In addition, they are of itte vaue to the cinician seeing an individua patient. It may be time to earn from the advances provided by the cumuative experience seen in the evauation and cassification of the natura history of coronary artery disease; a parae approach is summarised in tabe 2. Coronary artery disease is simiar to COPD in that it has a ong asymptomatic phase where preventive therapy is highy effective. Once detected, the disease may be stabe and treated as such. In patients with COPD, at a certain point in time the disease may become more symptomatic and unstabe; this cinica syndrome coud be named unstabe COPD akin to the simiar picture in the patient with coronary artery disease where the diagnosis of unstabe angina prompts modifications in therapy and more carefu foow-up. If the decompensation is associated with refractory dyspnoea (.4 on a 0 10 scae), worse cough and sputum, manifestations of systemic invovement, such as tachypnoea (.24 breaths?min -1 ), fever, eevated white ce count (.9,000 ces?dl -1 ) and CRP (.10 mg?dl -1 ), without evidence of TABLE 2 Syndrome Proposed cinica definitions of chronic obstructive pumonary disease (COPD) syndromes by anaogy to syndromes in coronary artery disease Eements COPD Unstabe COPD Exacerbation Ventiatory insufficiency Coronary artery disease Unstabe angina Myocardia infarction Cardiogenic shock Worsening of dyspnoea cough or sputum RR,24 Dyspnoea,4 on 0 10 scae Norma aboratory investigations Worsening of dyspnoea, cough or sputum Dyspnoea o4 on a 0 10 scae Norma chest radiograph WBC count.9000 ces?dl -1 or CRP.10 mg?dl -1 Same pus eevation of Pa,CO2 in arteria bood gases Worsening of chest pain No ECG changes Norma aboratory resuts Chest pain Abnorma ECG Abnorma serum enzyme pattern Same syndrome pus shock RR: reative risk; WBC: white bood ce; CRP: C-reactive protein; Pa,CO2: carbon dioxide arteria tension VOLUME 29 NUMBER 6 EUROPEAN RESPIRATORY JOURNAL