Prediction of Long-Term Benefits of Inhaled Steroids by Phenotypic Markers in Moderate-to-Severe COPD: A Randomized Controlled Trial

Background The decline in lung function can be reduced by long-term inhaled corticosteroid (ICS) treatment in subsets of patients with chronic obstructive pulmonary disease (COPD). We aimed to identify which clinical, physiological and non-invasive inflammatory characteristics predict the benefits of ICS on lung function decline in COPD. Methods Analysis was performed in 50 steroid-naive compliant patients with moderate to severe COPD (postbronchodilator forced expiratory volume in one second (FEV1), 30–80% of predicted, compatible with GOLD stages II-III), age 45–75 years, >10 packyears smoking and without asthma. Patients were treated with fluticasone propionate (500 μg bid) or placebo for 2.5 years. Postbronchodilator FEV1, dyspnea and health status were measured every 3 months; lung volumes, airway hyperresponsiveness (PC20), and induced sputum at 0, 6 and 30 months. A linear mixed effect model was used for analysis of this hypothesis generating study. Results Significant predictors of attenuated FEV1-decline by fluticasone treatment compared to placebo were: fewer packyears smoking, preserved diffusion capacity, limited hyperinflation and lower inflammatory cell counts in induced sputum (p<0.04). Conclusions Long-term benefits of ICS on lung function decline in patients with moderate-to-severe COPD are most pronounced in patients with fewer packyears, and less severe emphysema and inflammation. These data generate novel hypotheses on phenotype-driven therapy in COPD. Trial Registration ClinicalTrials.gov NCT00158847


Introduction
Chronic obstructive pulmonary disease (COPD) is a major cause of morbidity and the fifth leading cause of mortality worldwide [Siafakas 1995]. The WHO organisation has foreseen that COPD mortality will be at the third ranking, after cancer and cardiovascular mortality, by the year 2020. This calls for adequate intervention in the progression of disease and, above all, a better understanding of the pathogenesis and pathophysiology of COPD.
Current knowledge shows that the disease is characterized by progressive and largely irreversible airways obstruction ]. This is demonstrated by an accelerated decline with age of FEV 1 (forced expiratory volume in 1 second), associated with symptoms of dyspnoea at rest or on exertion. The disease leads to impaired quality of life, disablement and eventually death. The pathogenesis of COPD is closely associated with cigarette smoking.
Risk factors for the development of COPD in smokers are continuation of smoking, low FEV 1 , and increased airway hyperresponsiveness [Tashkin 1996]. Furthermore, it has recently been reported that elevated serum IgE and peripheral blood eosinophilia are associated with a more rapid decline of FEV 1 in smokers [Villar 1995]. It appears that the susceptibility, development, clinical course and prognosis of COPD are at least partly determined by host characteristics.
The pathology of COPD is heterogeneous [Saetta 1994]. It can include (acute and chronic) inflammation within the small airways as well as features of emphysema.
The bronchiolar and peribronchiolar inflammation in smokers with COPD is characterized by e.g. increased numbers of epithelial mast cells and macrophages [Grashoff 1997], elevated numbers of neutrophils and of CD3 + T lymphocytes with an increase in the CD8 + subset [Saetta 1998], and increased expression of growth factors, such as TGFß (transforming growth factor beta) [de Boer 1998]. It is important to notice that some of these pathological features can be confirmed in bronchial biopsy specimens as obtained from the large airways ]. The current working hypothesis is that this inflammatory response not only leads to swelling and structural changes of the airway walls (fibrosis, smooth muscle growth) [Saetta 1998], but also to destruction of peribronchiolar alveolar attachments and parenchymal destruction by an imbalance of local proteases (such as neutrophil elastase) and their inhibitors (such as SLPI) [Thurlbeck 1990]. As this imbalance and the resulting destruction are not completely explaining the development of emphysema, it can be postulated that qualitative or quantitative defects in local tissue repair are also important in COPD (emphysema) patients. Tissue repair as a form of wound healing is regarded as an inflammatory process, in which inflammatory cells play a major regulating role. Consequently, long-term anti-inflammatory therapy can be expected to reduce not only damaging inflammation, but also the local tissue repair process [Timens 1997].
In addition to affecting tissue repair, steroids may also affect the local production of glutathione (GSH), a major anti-oxidant in the lung. Whereas cigarette smoke increases GSH both in vivo and in vitro, the steroid dexamethasone was found to decrease GSH production in cultured airway epithelial cells (Rahman 1998). Finally, steroids may also increase the antiinflammatory screen in the lung by increasing the production of the serine proteinase inhibitor secretory leukocyte proteinase inhibitor (SLPI), as has been shown in vitro (Abbinantie-Nissen, 1995). These observations have not yet been confirmed by in vivo studies, since treatment of patients with chronic bronchitis and emphysema by inhaled steroids did result in significant increases in sputum SLPI levels, just failing to reach statistical significance (Llewellyn -Jones 1996).
Based on the above, it may not be surprising that the benefits of currently available therapeutic interventions in COPD are limited [Siafakas 1995]. There is no doubt that smoking cessation remains the best option, so far being the only intervention that leads to a reduction in the annual decline in FEV 1 [Tashkin 1996]. Interestingly, such improvement is partly predicted by the level of airway hyperresponsiveness to methacholine [Tashkin 1996], which is thought to be determined by airway remodelling next to inflammatory cell activity [Reiss 1996].
The other intervention which has been postulated to reduce disease progression is antiinflammatory treatment by oral or inhaled corticosteroids [Postma 1988[Postma ,1991]. Indeed, there is evidence from prospective studies that regular therapy with inhaled corticosteroids leads to an improvement in FEV 1 within a 6-months period [Dompeling 1992]. However, such treatment does not seem to improve the annual decline in FEV 1 during long-term follow-up [Renkema 1996], although there seems to be clinical benefit in terms of symptoms, exacerbation rate and exercise capacity [Paggiaro 1997]. The transient benefits on FEV 1 have recently been confirmed in two large-scale multicentre European trials (still to be published: EUROSCOP, ISOLDE), in which inhaled steroid treatment in patients with COPD led to an initial 3-6 months of improvement in FEV 1 , which was maintained during the following 2.5 years of treatment. However, the subsequent course of FEV 1 (annual decline) after the first 3-6 months remained unaltered. This is indicative of transient benefits of inhaled steroids in COPD, predominantly during the first months of treatment. Using oral steroid therapy in COPD it has been shown that the greatest response occurs in those patients with eosinophilic inflammation within the airways. This suggests that steroids might reduce some of the features of acute airways inflammation in COPD, whereas they may not influence the structural changes within the airways. As indicated above, it may even be postulated that steroids can have deleterious effects on chronic inflammation [Timens 1997], e.g. by limiting neutrophil apoptosis or stimulation of collagen production [Gauldie 1997].
Taken together, it appears that inhaled steroid treatment in COPD has only transient effects on functional outcome. The reasons for this are unknown, but need to be examined at the level of airway pathology. Therefore, we aim to investigate the clinical, functional as well as pathological outcome of short-and long-term treatment with inhaled steroids in patients with COPD. This will clarify whether short-term inhaled steroid treatment will suffice, or whether long-term treatment is still required to preserve the clinical, functional and/or pathological benefits. Since add-on therapy by long-acting ß2-agonists in combination with inhaled steroids has been demonstrated to be highly effective in asthma [Woolcock 1996, Pauwels 1998], we will also examine the short-and long-term benefits of such combination therapy in COPD.
Short-term studies suggest that long-acting ß2-agonists may improve symptoms, quality of life, and exercise tolerance in COPD. However, this has not been formally addressed in longitudinal studies, nor in the presence of inhaled corticosteroids.

Hypothesis
We hypothesize that inhaled steroid treatment in patients with COPD leads to the following: 1. An initial improvement in symptoms, quality of life, and FEV 1 , associated with a reduction in features of airways inflammation.
2. 6 Months of treatment will be sufficient to maintain these initial effects; continuation of therapy beyond 6 months will lead to an unfavourable balance between destruction and repair within the airway wall.
3. After the initial improvement, the increased annual decline in FEV 1 continues, which is associated with on-going airway remodelling and parenchymal destruction.
4. Airway hyperresponsiveness, elevated serum IgE, Th2 cytokine profile, and eosinophil infiltration are predictive of the initial benefits on symptoms, FEV 1 , and airway inflammation.
5. Addition of a long-acting ß2-agonist to inhaled steroid treatment augments the benefit of therapy with regard to symptoms and quality of life, without changing the decline in FEV 1 or the pathological outcome.

Aim
The aim of the present study is to test the above hypotheses by comparing the clinical, functional and pathological benefits of: -short-term (6 months) inhaled fluticasone followed by placebo (24 months

Patient recruitment Groningen
Patients with COPD (n=135) will be selected from the practices of general practitioners and in the outpatient clinics of the Departments of Pulmonology from the University Hospital Groningen and the Leiden University Medical Center, and by advertisement through the media.
Participating general practitioners will be asked to select patients with COPD, who have not been treated with regular inhaled or oral corticosteroid during the last six months. In order to find these patients they will select patients between 45 and 75 years with a history of at least 10 pack years of smoking and at least one of the following symptoms: chronic cough, chronic sputum production, frequent exacerbations, or dyspnoea on exertion (see detailed inclusion and exclusion criteria, as listed below).
When selected the patient will be asked to cooperate in the study, and they will receive the patient information sheet. If in principle the patient is willing to cooperate, he or she will be referred to one of the study investigators. The study investigator will perform spirometry (FEV 1 ) in the investigational laboratory or at the home of the patient, and will provide detailed information to the patients about the study.
When the patient fulfils all the inclusion criteria and none of the exclusion criteria, informed consent must be signed by the patient and an appointment for visit one will be made.

Patient recruitment Leiden
In Leiden patients with COPD (n=135) will be recruited by advertisement through the media (appendix). Volunteers responding to this advertisement will first be screened by a telephone questionnaire taken by the investigators. When patients meet most of the inclusion criteria and if (in principle) they are willing to participate, they will be asked to visit the hospital to perform spirometry. Subsequently, when FEV 1 fulfills the inclusion criteria, the study investigator will provide information to the patient about the study, show a video demonstrating a bronchoscopy procedure and give the patient information form (appendix).
One week later the investigator will telephone the patient and ask if he or she is willing to participate. If so, the study investigator will make an appointment for the first visit.
On this first visit the patient and investigator or his/her designee have to sign and date the general informed consent for participation in the study and an informed consent concerning the three bronchoscopies (appendix).
After the first bronchoscopy the patient will be asked to complete a questionnaire about his or her experiences with this procedure. After having completed this questionnaire the patient will be asked specifically if he or she is still willing to participate in the study (appendix). If this is the case and the patient fulfills all the inclusion criteria and none of the exclusion criteria, the patient will be randomised and the next visit will be planned. 5. Other diseases likely to interfere with the purpose of the study.
6. Inability to keep diary and to understand written and oral instructions in Dutch.

Study design
In this prospective, longitudinal, double blind study COPD patients will be followed up for 2.5 years (see figure). The study begins as a 3 groups parallel design during 6 months, followed by a 4 groups parallel design during 24 months. The aim is to finish the study with 50 patients in each of the 4 parallel groups at 30 months.
Patients will be treated by the investigator-physician with inhaled corticosteroids (500 g b.i.d. fluticasone by Discus R ), or inhaled corticosteroids with long-acting ß 2 -agonist (+ 50 g b.i.d. salmeterol), or a matching placebo for the first 6 months. Thereafter, half of the corticosteroid group will continue and the other half will receive placebo for 2 years. The patients who will be treated with placebo or inhaled corticosteroids + salmeterol continue their treatment during the complete period of 2.5 years. Rescue treatment will be on demand usage of inhaled inpratropium bromide, which is a standard bronchodilator in COPD [Siafakas 1995].
Subjects who fulfil all inclusion criteria and none of the exclusion criteria (see patient selection) will be allocated to receive placebo, inhaled corticosteroids (equivalents of 500 g b.i.d. = 1000 g fluticasone) or inhaled corticosteroids + salmeterol at the beginning of the study using the minimization method. The use of minimization will provide treatment groups that are closely balanced for a number of variables (centre, gender, current smoker, FEV 1 /IVC < 60%, PC 20 methacholine < 2 mg/ml), which might influence the primary and secondary outcome variables [Altman 1995]. The first patient is given a treatment at random. For each subsequent patient we will determine which treatment will lead to better balance between the groups with respect to centre, gender, smoking, FEV 1 /IVC and PC 20 . The group with inhaled steroid will be randomly allocated to either the steroid group or the placebo group after 6 months of steroid treatment. Minimization will be performed using a computerized method [Pocock 1983].

Measurements
Measurements of symptoms, QOL, and spirometry (postbronchodilator) will be made every 3 months. In addition, exhaled nitric oxide (NO), bodyplethysmography and carbon-monoxide diffusion capacity will be measured at 0, 6 and 30 months. Bronchial biopsies, bronchoalveolar lavage, and sputum induction will also be performed at 0, 6 and 30 months. Peripheral blood eosinophils, IgE, and PC 20 methacholine at 0, 6 and 30 months. In bronchial biopsies, bronchoalveolar lavage and sputum special attention will be paid to the number of lymphocytes, neutrophils, eosinophils, macrophages, mast cells, and the state of activation of these cells and the epithelium.
Once during the study a spiral computed tomography (CT-scan) will be performed to quantify the presence of pulmonary emphysema in each patient [Gevenois 1996], and EDTAanti-coagulated whole blood will be collected for DNA-isolation, for future studies on gene polymorphisms possibly associated with glucocorticoid responsiveness or other study outcomes.
The effects of treatment will be analysed by relating the observed changes in clinical and pathophysiological outcome, to those in cellular and histological outcome (see analysis).

Number of patients
In order to achieve a total number of n=200 (n=50 per treatment arm) at the end of the study, we will initially include 270 patients.

Withdrawal of patients
The patients are free to discontinue their participation in the study at any time and without prejudice to further treatment. The patient's participation in the study can be discontinued at any time at the discretion of the investigator. Justifiable reasons to discontinue a patient from the double blind treatment of the study include: the development of a serious adverse event, non-compliance of the patient, erroneous inclusion of the patient in the trial, acute exacerbations requiring oral steroid treatment (see below) for > 3 periods during the past 12 months or > 2 periods during the past 3 months, any treatment not allowed in the study, development of a concomitant disease that interferes with the interpretation of the study, withdrawal of consent. Patients who discontinue their participation in the study will be contacted by the investigator to obtain information about the reason for discontinuation. The reasons for study discontinuation will be reported on a study-termination form.
An effort will be made to collect as many data as possible on the outcome variables after withdrawal, so that the reasons for withdrawal can be analysed to check for bias.

Postponing visits in case of exacerbations
Any period requiring a short course of oral steroids (30 mg prednison for 6 days), given at the discretion of the physician, will be labelled as an acute exacerbation. Study visits will have to be postponed if such oral course of steroids has been given < 8 weeks prior to the visit (12 weeks in case the visit will include a bronchoscopy. The next visit has to be scheduled at the next predefined time point according to the regular 3 monthly intervals.

Blinding
All study medication, active drugs and placebo will be of identical appearance throughout the study. The study inhalers will be labelled identically with centre code and subject number.

Treatment compliance
Patients are required to comply with the treatment regimens throughout participation in the study. Compliance will be checked by counting the doses on the Discus inhaler. A minimum of on average 70% of the treatment dose must have been taken in order to include a patient in the analysis.

Measurement of symptoms
Symptoms will be measured during the week before and the week after every visit with the Dutch version of the COPD Symptom Control Questionnaire (CSCQ). Additionally, the questionnaire will be administrated by the patients during the visit. Health related Quality of Life will be measured by generic and disease specific instruments.

Generic Health related QOL instruments
At the first visit, after six months and at the final visit patients will be asked to administer the

Disease specific instruments
At every visit patients will be asked to administer the St. George Respiratory Questionnaire (SGRQ) [Jones 1992].

Lung Function
Spirometry will be performed according to international guidelines [Quanjer 1993]. A weekly calibrated rolling-seal spirometer, or pneumotachograph (accuracy 50 ml) will be used throughout the study. First, 3 slow inspiratory vital capacity manoeuvres (IVC, use largest value) will be carried out. Second, maximally 5 forced expiratory vital capacity (FVC) manoeuvres will be performed to obtain at least 3 technically satisfactory expiratory flowvolume curves from which FVC does not deviate > 5% from the largest FVC. From these curves, we will use largest values of FVC, forced expiratory volume in 1 second (FEV 1 ), and instantaneous expiratory flows (MEF). Reference values will be obtained from Quanjer et al. [1993].
Reversibility of airways obstruction will be measured 20 min after administering 4 single puffs of 100 g salbutamol per metered dose-inhaler connected to a spacer (Volumatic®), following the same procedures. The response will be expressed as change in FEV 1 as percentage of predicted value [Brand 1992].
At 0, 6 and 30 months total lung capacity (TLC), residual volume (RV), functional residual capacity (FRC), airway resistance (Raw) and specific airway conductance (sGaw) will be measured using a constant volume bodyplethysmograph, according to a standardised method ]. Furthermore, the diffusion capacity (transfer factor) for carbon-monoxide (TL CO and K CO ) will be measured using the single breathholding method ]. The single breath nitrogen wash-out test (slope of phase 3) will be added in Leiden only, as a simple parameter of small airway function with predictive properties for the development and course of COPD [Stanescu 1998].

Exercise capacity
At 0, 6 and 30 months the 6 minutes walking distance will be determined, in a flat indoor corridor, after 3 praticing attempts [Paggiaro 1998].

Airway hyperresponsiveness
Airway hyperresponsiveness will be determined using standardized methods [Sterk 1992, Demedts 1998]. This includes challenge tests with methacholine by so-called tidal breathing method. Serial doubling concentrations of methacholine-bromide (0.035-312 mg/ml) will be nebulized by a DeVilbiss 646 jet-nebulizer (filled with 3 ml, connected to an in-and expiratory valve box with an expiratory aerosol filter, operated with pressurized air) with monthly calibrated output of 130 mg/min. The aerosols will be inhaled by tidal breathing during 2 min at 5-min intervals through the mouth with the nose clipped. The response will be measured by FEV 1 . After recording 3 reproducible (within 5% of largest) values of FEV 1 (to obtain mean baseline FEV1), the solvent will be used as first challenge. FEV 1 will be recorded at 30 and 90 s following each challenge, from which the lowest, technically satisfactory value will be used for further analysis. The first concentration of methacholine will be determined from recent guidelines [Demedts 1998]. The challenge will be discontinued if FEV 1 drops by > 20% from mean baseline value. Then the patient will receive 200 g inhaled salbutamol per MDI, in order to reverse bronchoconstriction. The response of the challenges will be expressed as provocative concentration causing a 20% fall in FEV 1 (PC 20 ) by log-linear interpolation.
Measurement of the fall in forced vital capacity (delta FVC) at the PC 20 level is optional [Gibbons 1996].

Exhaled nitric oxide (NO)
Exhaled NO levels will be determined at 0, 6 and 30 months using the latest guidelines [Kharitonov 1997, ATS 1999] by a chemiluminescence analyzer (Sievers NOA 270B or Ecophysics CLD 700 AL) [Olin 1999]. The patients will be connected to a closed system with NO-free inspired air (pressurized air < 2 parts per billion: ppb, using a bag-system connected to a 3-way valve), to avoid contamination of the measurements with ambient NO. The patients will perform a slow vital expiratory capacity manoeuvre with a constant expiratory flow of 100 ml/sec against an expiratory resistance of 5 cm H 2 O. Expiratory NO concentration is sampled continuously from the centre of the mouthpiece at a sample flow of 440 ml/min, and the average concentration will be determined during a period of 10 seconds. At each visit 3 successive recordings will be made at 30-s intervals, from which the mean values of exhaled NO will be used in the analysis.

Induced sputum
Sputum will be induced and processed according to a validated, and standardized technique [In 't Veen 1996], with some modifications. Prior to sputum induction, baseline FEV 1 will be recorded and, for safety reasons, 200 g salbutamol will be administered through a metered dose inhaler (plus Volumatic®). Hypertonic sodium chloride aerosols (4.5 w/v %) will be generated at room temperature by a DeVilbiss Ultraneb 2000 ultrasonic nebulizer with a calibrated particle size (MMAD 4.5 m) at maximal output (2.5 ml/min). The aerosols will be administered to the subjects through a 100 cm long tube with an internal diameter of 22 mm, and inhaled via the mouth through a two-way valve (No. 2700; Hans-Rudolph, Kansas City, MO, USA), with the nose clipped. Subsequently, the patients will inhale hypertonic saline aerosols during 3 periods of 5 min. After each inhalation, or as soon as the subjects experience cough, they are asked to blow their nose, to rinse their mouth and throat with water, and to expectorate sputum into a clean plastic container by coughing. Whenever the patients feels discomfort, and after completion of each period FEV 1 will be measured, and salbutamol will be administered if FEV 1 drops by > 20% from post-salbutamol baseline value.

Sputum processing
The volume of the induced sputum samples will be determined by weighing. The sample will then be mixed with an equal volume of 0.1% sputolysin (dithiotreitol, DTT, Calbiochem, USA).
To ensure complete homogenization, the samples are placed in a shaking water bath at 37ºC for 15 min, once interrupted by gently mixing the sample. The homogenized sputum will be centrifuged (350 Differential cell counts of eosinophils, neutrophils, lymphocytes, macro-phages, epithelial and squamous cells are performed on Diff-Quik (choice of staining to be discussed) stained cytospins by a qualified cytopathologist. To correct for the variable salivary contamination, differential leucocyte and cylindric epithelial cell counts will be expressed as a percentage of 500 nucleated cells excluding squamous cells. For each sample, differential cell counts will be performed twice by the same observer, and the mean data will be used in the analysis. A sputum sample will be considered adequate when the percentage squamous cells is less than 80%. To ensure a blind analysis of the sputum samples, all cytocentrifuge slides are coded before analysis by an investigator who is not involved in the counting.
Sputum supernatant will be used to measure soluble mediators (see elsewhere in this section).

Bronchoscopy
Bronchoscopy will be performed using established guidelines in general and those developed for asthma in particular [NHLBI 1985, Harrison 1993, Jarjour 1998]. Smokers are requested not to smoke their first cigarette on the day of the bronchoscopy before all procedures for investigation are completed. Patients are not allowed to drink or eat food from midnight onwards, except for taking their medication. On arrival for bronchoscopy they will receive 400 g salbutamol per Volumatic R spacer, 20 mg oral codeine phosphate and 0.5 mg of atropine s.c. [Harrison 1993]. Fifteen minutes later 2 ml of lidocain 1-4% will be instilled in the mouth 3-5 times, on the vocal cords and into the trachea to inhibit coughing. The total lidocain dose must not exceed 3 mg/kg [Ghio 1998]. A flexible bronchoscope will be introduced and bronchoalveolar lavage will be performed into the right middle lobe. 4 Times 50 ml PBS at 30ºC will be instilled and after 10 sec of dwelling, it will be removed by gentile suction at 20 cm water pressure or less. Thereafter, 3 times 50 ml of saline will be instilled with dwelling times of 10 sec and each portion of 50 ml of saline will be collected in a tube before the next 50 mls are instilled. All collected BAL will be immediately stored on ice and transported to the laboratory within 15 min after collection. Separate analysis of the first aliquot will be considered.
Bronchial biopsies will be taken from subsegmental carinae in the right or left lower lobe using cup forceps. When possible at least 6 biopsies will be taken . Four biopsies will be fixed in 4% neutral buffered formalin, processed and embedded in paraffin. The remaining biopsies will be embedded in Tissue Tek mounting medium, snap frozen in liquid isopentane and finally stored at -80C.

Processing of BAL fluid
BAL processing will be performed analogous to the methods used for sputum processing, with the major exception that no DTT/Sputolysin will be used for homogenization.
Immediately upon collection, the BAL fluid will be placed on ice and centrifuged (350 x g) The cell pellet will then be resuspended in phosphate-buffered saline (PBS) containing 1 % (w/v) human serum albumin (HSA), pH 7.4, to a final volume of appr. 1 ml, followed by filtration through a nylon gauze (pore-size approximately 48 m, Thompson, Ontario) to remove clumps (filter method to be standardized to a uniform method in Groningen and Leiden). Cell viability and total cell counts are performed by Trypan blue exclusion and using a hemacytometer (use of automated cell count using e.g. Coulter Counter to be discussed). Cytospin preparation and analysis will be performed as described in the Sputum Processing paragraph of this section.

Immunocytochemistry of sputum and BAL cells
Immunocytochemical staining for cytokines and SLPI as described in the biopsy paragraph will be performed on cytospin preparations of BAL and sputum. Staining will be performed on formalin-fixed, saponin-permeabilized preparations as described [Grunberg, 1997; details to be discussed between Leiden and Groningen].

Assessment of mediator concentration in sputum and BAL
For all mediators to be assessed in sputum, initial experiments will be performed to optimize these assays in this complex biological fluid [Grünberg 1997]. These experiments include assessment of the recovery of the mediator and the effect of the reducing agent dithiotreitol on the immunoassay for that specific mediator.

Histology
Frozen tissue: Four m thick frozen sections will be cut and stored at -20C until use. In frozen sections the haematoxylin-eosin (H-E) staining will be used for judging the biopsy quality. Two different three-step immunoperoxidase protocols are performed to enhance the signal of weakly staining antibodies. The first three-step protocol includes incubation with the second step reagent biotin-labelled secondary antibody followed by incubation with horseradish peroxidase-labelled streptavidin (SBA). The second three-step protocol consists of incubation with horseradish peroxidase-labelled rabbit anti-mouse immunoglobulin antiserum as the second step reagent, followed by horseradish peroxidase-labelled goat or swine anti-rabbit immunoglobulin antiserum as the third step reagent. Other incubation steps are performed as described in the two step immunoperoxidase protocol.

Paraffin tissue
Immunohistochemistry is performed on 3 m formalin fixed, paraffin embedded lung tissue sections using two or three step immunoperoxidase protocols as described in standard operation protocols. Both protocols are mainly as described for frozen sections. In short, to improve the staining quality of some of the antibodies, antigen retrieval is performed by either proteolytic treatment or by heating in either EDTA (pH 8.0) or citrate buffer (pH 6.0). Sections are then incubated with the primary antibody overnight. The remainder of the protocols is as described for frozen sections.

Biopsy parameters to be studied
In the following, an overview is given of the possible assessments in biopsies. The number of immunostainings (different antigens) to be performed is determined by the number of available biopsies and their quality.

Quantitative assessments in bronchial biopsies
All biopsy specimens will be coded and sections will be examined by a single investigator in a blinded fashion, by means of a semi-automated image analysis system using a 200x magnification. Images of areas with well preserved tissue structure and without bronchus associated lymphoid tissue will serve as basis for further computer-aided analysis. For subepithelial analysis, the widest possible zone of maximal 125 m deep beneath the epithelial basement membrane will be delineated excluding damaged tissue, mucus-secreting glands, and airway smooth muscle (Sont 1996Grashoff, 1997;Ten Hacken 1997. For epithelial analysis, the epithelium will be regarded using the same criteria (Grashoff, 1997).
The number of positively staining nucleated inflammatory cells will be counted by a validated full automated procedure

Computed tomography
At the start of the study the degree of pulmonary emphysema will be quantified in each COPD patient by CT-scan [Gevenois 1996]. Single slices will be obtained from lung apex to basis (120 kVp, 200 mA, scan time 1 s per slice, 1 mm thickness, 1 cm interval). The acquired images will be analysed to calculate mean lung density expressed in Hounsfield Units (HU) [Cheung 1997].

Determination of the sample size
The determination of the sample size is based on detectable differences in the primary outcome variables, the mean inflammatory cell counts in bronchial biopsies between the different treatment groups. Under the assumption that the distributions of (logtransformed) cell counts is approximately normal and the standard deviation in the different treatment groups is approximately equal, the sample size follows: A sample size of 8-25 patients is needed to detect at least one doubling difference in mean cell number per 0.1mm 2 for a particular inflammatory cell to detect changes within a group (alpha=0.05, power=0.80). A sample size of 13-48 subjects per group is required to detect at least 1 doubling difference in cell number between parallel groups ].
Based on an estimated standard deviation of repeated cell counts in bronchial biopsies of 2.34 fold change, a sample size of 50 per study arm allows us to detect at least a 1.62 fold different increase in number of inflammatory cells/0.1 mm 2 between two groups.

Statistical methods
Quality of data: The quality of the data entry will be checked in a random sample of 10% of the cases.
Safety: During the data entry process, all Case Report Forms will be reviewed for possible adverse events (AE). Adverse events are defined as: any unfavourable, unintended event (signs, symptoms, changes in laboratory data) temporarily associated with the administration of the study drug whether or not considered drug related. Examples are signs, symptoms or disorders reflecting adverse event found anywhere in the Case Report Form as notes or comments. All data regarding serious adverse events will be entered into a safety database and analysed at the department of Epidemiology & Statistics of the University of Groningen, 12 months after the start of the study and again after 24 months. The safety variables that will be analysed at those time points are: spontaneously reported adverse events, extended adverse events, question on diagnosis of conditions associated with glucocorticosteroid treatment (hypertension, bone fractures, subcapsular posterior cataract, myopathy and diabetes, and skin bruises on both forearms).

Statistical analyses
Prior to breaking the treatment code at the end of the study, all decisions on the evaluability of data from each individual patient for the statistical analyses are made and documented. The effects of treatment will be analysed according to treatment group. Primary analyses will be comparing changes in cell counts per treatment group in time. Secondary analyses will compare changes in clinical outcome variables such as lung function and respiratory symptoms with changes in cell counts.
Linear mixed effects models will be used to estimate the effects of treatment on changes in primary outcome variables (cell counts) and the secondary outcome variables (such as lung function, bronchial hyperresponsiveness, number of blood eosinophil count). Graphical visual inspection will be used to determine which model has the best fit (linear/non-linear model).
Descriptive statistics will be presented by visit. Confidence intervals will be computed using a 95% confidence interval level and p-values <0.05 are considered to be statistically significant.
In addition, the baseline values of lung function, bronchial biopsy, bronchoalveolar lavage and sputum induction will be evaluated as possible explanatory variables for the outcome.
Smoking habits will be evaluated as the duration of smoking (years), average daily consumption prior to the study, average daily consumption during the periods between visits, previous cigarette consumption expressed as packyears.
Time points will be calculated as time elapsed since baseline visit.

Goedkeuring
Het onderzoek is goedgekeurd door de directie en de Commissie Medische Ethiek van het LUMC te Leiden en het AZG van Groningen.