Short versus conventional term glucocorticoid therapy in acute exacerbation of chronic obstructive pulmonary disease

a Clinic for Internal Medicine, b Clinic for Endocrinology, Diabetology, and Clinical Nutrition, c Clinic for Pneumology, d Basel Institute for Clinical Epidemiology and Biostatistics, e Emergency Department, University Hospital Basel; f Institute of Social and Preventive Medicine, University of Basel; Switzerland; g Department of Clinical Epidemiology and Biostatistics, McMaster University, Hamilton, Canada; h Clinic of Internal Medicine, Hospital Centre, Biel-Bienne, Switzerland


Introduction
Chronic obstructive pulmonary disease (COPD) is the fourth commonest cause of death worldwide, posing a large socioeconomic burden [1,2] .In Switzerland, it has been estimated that 5-7% of the population and 28% of the current smokers suffer from COPD [3].The clinical course of the disease is characterised by progressive, irreversible airflow obstruction associated with chronic inflammation of the respiratory tract.Acute exacerbations are triggered mainly by respiratory tract infections.According to evidence-based reviews and current guidelines, systemic glucocorticoid therapy
As patients with COPD frequently experience exacerbations, many will be exposed to large cumulative doses of glucocorticoids during the course of their disease.A significant number of patients remain steroid-dependent beyond the acute exacerbation.Thus, approximately 50% of eligible patients had to be excluded from the SCCOPE study because of systemic glucocorticoid use in the 30 days prior to enrolment.
The side-effects and dangers of treatment with systemic glucocorticoids are well known.Their use in patients with COPD is independently associated with increased mortality in models adjusting for multiple confounders [18].Moreover, these patients are at enhanced risk for osteoporosis, in part due to decreased muscle mass, physical inactivity, and cigarette smoking.In a cohort of individuals with chronic pulmonary diseases, including COPD, the cumulative dose of glucocorticoids strongly correlated with vertebral fracture risk due to loss of bone mineral density [19] .Glucocorticoids contribute to muscle catabolism in COPD [20], and the cumulative dose has been found to correlate with muscle weakness [21].In the SCCOPE study, hyperglycaemia was seen significantly more often in glucocorticoid-as compared to placebo-treated patients, and rehospitalisation rates due to serious infections, particularly pneumonia, were high in the long-term active treatment group [15] .Moderate to high-dose (i.e., ≥25 mg/d prednisone equivalent) systemic glucocorticoid treatment, even if prescribed for a relatively short period (i.e., 5-30 days), may suppress the hypothalamicpituitary-adrenal (HPA) axis as assessed by the low-dose (1 µg) ACTH stimulation test in almost half the patients [22].Although the clinical significance of this finding is unclear, a suppressed HPA axis may be potentially hazardous in stressful situations such as an acute exacerbation of COPD.
Thus, in order to alleviate these multiple adverse effects of glucocorticoid therapy, a strong effort should be made to minimise its cumulative dose in patients with COPD.

Patients and methods
The main objective of this trial is to test the hypothesis that in acute exacerbations of COPD, a short course (5 days) of glucocorticoid treatment will not result in an inferior clinical outcome compared to a conventional course (14 days), but will significantly decrease steroid exposure and steroid-associated side effects in patients with exacerbation of COPD.The study was approved by the institutional review boards of all participating centres and is conducted in accordance with the Declaration of Helsinki.

Study design
This is a multicentre, randomised, controlled study aiming to demonstrate non-inferiority of a five-day as compared to a 14-day course of systemic glucocorticoids with respect to COPD outcome during a follow-up of 180 days.The primary endpoint will be time to next COPD exacerbation, defined as acute-onset worsening of the patient's condition beyond day-today variations requiring interaction with a healthcare provider [23], which may occur during the index exacerbation or during follow-up.
Secondary endpoints will be cumulative steroid dose; time to open-label standard-dose glucocorticoid therapy during the index exacerbation; need for invasive or non-invasive mechanical ventilation; change in FEV 1 ; clinical outcome at discharge and during follow-up as assessed according to the ATS consensus statement for dyspnoea [24] and a standardised questionnaire [25]; duration of hospital stay; all-cause mortality; steroid-associated side-effects and complications, defined as follows: development or exacerbation of hyperglycaemia (fasting plasma glucose ≥5.6 mmol/L or random plasma glucose ≥7.8 mmol/L or rise ≥20% in daily doses of insulin or any increase in oral antidiabetic drugs or initiation of one or more antidiabetic therapeutic principle), development or worsening of hypertension (blood pressure ≥140 mm Hg systolic and/or ≥90 mm Hg diastolic or the addition of one or more antihypertensive drugs to previous treatment regimens), integrity of the HPA axis at study entry and during follow-up as assessed with the low dose (1 µg) ACTH stimulation test and with a clinical questionnaire on signs and symptoms of hypocortisolism [26]; effect on boneturnover as assessed by specific biochemical markers; secondary infections; and other potentially steroid-related adverse events (e.g., gastrointestinal bleeding or psychiatric disease).

Methods
Six hospitals in the northern and central part of Switzerland participate in this trial.Patients admitted to the local emergency department because of exacerbated COPD will be consecutively assessed for eligibility.The diagnosis of exacerbated COPD is defined clinically by the presence of at least 2 of the following signs and symptoms: change in baseline dyspnoea, cough, and sputum quantity or purulence [8,27] .Irreversible airway obstruction will be documented by a bedside post-bronchodilator spirometric evaluation performed at inclusion, and the severity of COPD will be categorised according to GOLD criteria [1].Patients must be 40 years of age or older, with a smoking history of ≥20 pack-years (past or present smokers) in order to minimise contamination of the study cohort with asthmatics.Exclusion criteria are inability to give written informed consent, diagnosis of asthma, radiological diagnosis of pneumonia, severe co-existing disease making survival of >6 months unlikely, and pregnancy or lactation.
At study entry, patients will be randomly assigned to one of two treatment regimens using a centralised, computerbased randomization procedure with fixed blocks of 6. Study subjects will be randomised after stratification according to age, presence or absence of systemic glucocorticoid treatment prior to the study period, and severity of COPD according to the GOLD classification [1].Each study subject will be randomly assigned to receive a prepared set of study medication, consisting of two drug vials.The drugs will be prepared prior to the initiation of the study and packed by the Pharmacology Department, University Hospital, Basel, according to a randomization list.All patients will receive methylprednisolone 40 mg i.v. on day 1, followed by prednisone 40 mg daily dispensed from vial 1, containing the tablets for days 2 through 5. We chose the i.v.route for the first dose in order to facilitate administration upon hospital admission, when patients are often severely distressed.On day 6, vial 2 will be started which contains either prednisone or matching placebo for nine days.Thus, the standard group will receive a 14-day course of 40 mg prednisone equivalent q.d., and the intervention group will receive 40 mg prednisone equivalent q.d. for five days.The decision to administer open-label systemic glucocorticoid therapy at any time during the trial or to increase the dose during the initial 5-days treatment phase will be entirely at the discretion of the treating physicians, without interference by the study team.The treating physicians may be hospital-based (during the in-patient phase) or the general practitioner or chest physician following the patients after discharge.
All data relevant to our primary endpoint and the secondary endpoints will be assessed daily during the hospitalisation period, on day 6, on the day of discharge, as well as on days 15, 30, 90, and 180 of follow-up.Follow-up on days 15 and 90 will be by telephone interview, and on days 30 and 180 by clinical visit.Patients who have reached the primary endpoint are followed up for the entire 180 day as specified in the protocol.Apart from the systemic glucocorticoid regimen, COPD-related treatment during hospitalisation and the 180 day follow-up will be standardised; all patients will be treated during their hospital stay with nebulised inhaled short-acting bronchodilators 4 to 6 times daily as needed, an inhaled topic steroid (fluticasone 250 µg or budenoside 400 µg b.i.d.) combined with an inhaled long-acting beta agonist, (formoterol 12 µg or salmeterol 50 µg b.i.d.), plus tiotropium 18 µg once daily.In addition, to avoid any treatment bias between groups, all patients will receive a broad spectrum antibiotic for the first 7 days of the index exacerbation.Management during the hospital phase will include physical therapy, supplemental oxygen, and ventilatory support, as indicated clinically and recommended by current ATS / ERS guidelines [8].The inhaled topic steroid, long-acting beta agonists and tiotropium will be maintained throughout the follow-up period.
Adverse events within 180 days after inclusion into the study will be monitored by the investigators and the data safety and monitoring board.We will assess hypertension, hyperglycaemia, infections and suppression of the hypothalamic-pituitary-adrenal axis as potential adverse events of glucocorticoid therapy and document these as secondary outcomes.To assess the function of the adrenal glands, the low-dose (1 µg) ACTH (tetracosactide; Synacthen ® ) test will be performed prior to the administration of study drugs at entry, on day 6, and on the day of discharge.The test will be repeated during follow-up until results have normalized.Per study protocol, glucocorticoids are not tapered.Patients discharged from the hospital with adrenal insufficiency (i.e., plasma cortisol level <500 nmol/L after stimulation with 1 µg tetracosactide) will receive instructions about stress prophylaxis therapy with hydrocortisone by means of a tutorial developed by the Clinic for Endocrinology, Diabetes and Clinical Nutrition at the University Hospital Basel.

Statistical considerations, sample size, and power calculation
This study is designed to show non-inferiority of a short-course (5 days) as compared to a standard course (14 days) glucocorticoid treatment.In order to use a reliable treatment protocol, we chose a dose of 40 mg prednisone for 14 days for the standard treatment arm, which has unequivocally shown a benefit (albeit in non-hospitalised patients) over placebo for the primary endpoint of treatment failure [17].
We chose time to next COPD exacerbation as primary endpoint because of its clinical relevance and its potential to show differences between the two groups.Other clinically relevant outcomes as well as improvement of FEV 1 achieved by glucocorticoid treatment [16], will be assessed as secondary endpoints.To estimate the frequency of our primary endpoint, we used data published in the SCCOPE study [15] , and own data from the "ProResp" study [28] and the ProCOLD trial on acute exacerbations of COPD [29] .Based on this information, we estimated that the primary endpoint would occur in ~50% of patients within 6 months after randomisation.
We used a modified Delphi approach to define non-inferiority with regard to the primary endpoint.Briefly, 11 board certified specialists for internal medicine, endocrinology, and pulmonology were individually asked to define the maximum increase of therapeutic failures they would be willing to accept for the benefit of shortening the glucocorticoid treatment duration.According to the results of this survey, the planning committee judged that a 15% absolute difference in the percentage of patients with an exacerbation during the 6 months of follow-up would be a clinically tolerable upper limit.Assuming the proportion of patients under standard treatment with an exacerbation during follow-up equals 50%, this non-inferiority definition implies that the proportion of patients under experimental treatment with exacerbations during follow-up must not exceed 65%.We based our power calculation on an exponential survival model with proportional hazards between groups tuned to the above non-inferiority definition.The choice of the simplest possible survival model can be justified on the grounds that the log rank test and the proportional hazard assumption are both

Original article
Swiss Med Wkly.2010;140:w13109 • www.smw.chinvariant to order-preserving transformations of the time scale if censored observations only occur after maximum followup.
Translating the above non-inferiority scenario into a survival-analytical setting with proportional hazards and timeindependent hazard provides survival functions S s (t) = 0.5 t / 180 (standard arm) and S e (t) = 0.35 t / 180 (experimental arm), where S(t) = probability for event-free survival until time t and t = time from inclusion in days.
Here, 50% of subjects in the standard arm and 35% of the subjects in the experimental arm are assumed to have been without exacerbation during the entire follow-up of 180 days.
Under the constant hazard assumption characterising the exponential survival model, the critical hazard ratio corresponding to our non-inferiority scenario is given by HR = ln(0.35)/ ln(0.5)= 1.515When restricted to subjects with an event during 180-day follow-up, the mean and median event times under this model are 80 and 75 days, respectively, in the standard treatment arm; and 74 and 67 days, respectively, in the experimental arm.
Providing for the possibility of loss to follow-up, the necessary group sizes under a non-inferiority scenario with α = 0.05 and ß = 0.1 are as follows (we used the algorithm based on reference [30]): 177 patients if there is no loss to follow-up; 186 patients if loss to follow-up is 10%; 195 patients if loss to follow-up is 20%; 206 patients if loss to followup is 30%.
To corroborate these numbers, we performed a Monte Carlo simulation study under the same scenarios (i.e., assuming exponential survival distributions with a) 195 observations per arm, b) equal exacerbation rates of 50% during follow-up under the equality hypothesis, c) 65% vs. 50% exacerbations under the inferiority hypothesis, and d) with a drop-out rate of 20% during follow-up).In each of the simulations, the decision was based on the one-sided 95%-confidence interval for the hazard ratio between the experimental and standard arm and non-inferiority was concluded if this interval did not include 1.515 (see equation above).Among 5000 simulations under the inferiority hypothesis, non-inferiority was falsely concluded in 4.5% of cases, whereas the non-inferiority hypothesis was falsely rejected in 14% of 5000 simulations conducted under the equality hypothesis.
We therefore originally planned to include a total of 390 patients in the trial.Monitoring of patient recruitment and follow-up of included patients over the first years of the trial showed that the actual recruitment was considerably slower and the drop-out rate much lower (<4%) than anticipated.Given these observations and our limited resources, we recalculated the necessary sample size using a drop-out rate of 5%, an error of 5% and a power of 85% in order to complete the trial within a realistic time-frame.Under these assumptions, 150 patients need to be recruited per arm.Reassessment using a Monte-Carlo simulation indicated that the true power might be slightly lower, but at least 80%, with our non-inferiority definition.We now plan to include a total of 300 patients in the study.

Statistical analysis
Because in non-inferiority trials intention to treat and per protocol analyses may both falsely favour the conclusion of equivalence [31,32] , we will analyse our data in both ways.To test non-inferiority of the experimental treatment, a simple Cox regression analysis will be performed.Non-inferiority will be concluded if the two-sided 90%-confidence interval for the hazard ratio between experimental and standard arm in the intention-to-treat and per protocol analysis lies below ln(0.35)/ln(0.5)= 1.515, i.e., the critical hazard ratio under the chosen non-inferiority scenario.
Since differences between the two survival curves might appear mainly at the beginning of follow-up (which would be in contradiction to the proportional hazard assumption), we will also consider the area between survival curves equalling the difference in average event-free survival time during follow-up between the two arms.

Results
Recruitment started in March 2006.So far, 211 patients have been randomised, and a total of 127 have completed the 180 days follow-up.Their mean (± SD) age was 70 ± 11 years.Patients had smoked 50 ± 23 (mean ± SD) pack-years.58% of the study subjects had very severe (GOLD stage IV) and 30% had severe (GOLD stage III) COPD; the remainder (12%) had mild or moderate airway obstruction (GOLD stage I and II, respectively).20% of patients had received systemic glucocorticoids before entering the study, either for the index exacerbation or as part of their chronic COPD therapy.
The data safety and monitoring board carried out a pre-planned interim safety analysis after about 50% of the originally planned number of patients had completed the study.Based on the results of this analysis, the board recommended continuation of the trial without safety concerns.

Discussion
A 10 to 14 day course of systemic corticosteroids in patients with acute COPD exacerbations is an evidence-based treatment recommendation and has become common clinical practice.Systemic glucocorticoids improve lung function, reduce hospital stay, and decrease treatment failure.However, randomised clinical trials addressing the use of systemic glucocorticoids are relatively sparse and heterogeneous in their designs and settings [7,10] .Notably, the optimal dose and duration of treatment remains unknown.The benefits of glucocorticoids on recovery of lung function appear to level off after 3 to 5 days [15,16] and are compromised by complications, notably a higher rate of hyperglycaemia requiring treatment, osteoporotic bone disease, muscle catabolism, and suppression of the endogenous cortisol production.Most importantly, prolonged use of glucocorticoids raises mortality in patients with COPD [18].This therapeutic dilemma prompted the initiation of the present multicentre trial aiming to demonstrate non-inferiority of a short course of systemic glucocorticoids as compared to a standard course, thereby significantly reducing steroid exposure.Some considerations made during the planning phase of this trial warrant additional comment.
Previous studies examined the effects of systemic corticosteroids in acute exacerbations of COPD and used spirometric improvement (FEV 1 ) as a main outcome parameter [11,12,16] .However, FEV 1 is not closely associated with the perception of dyspnoea or with exercise limitation [33] and is therefore of limited patient relevance.In the SCCOPE study [15] , treatment failure rates in the placebo and glucocorticoid groups were identical after 6 months of follow-up, and the primary outcome of treatment failure was defined as a composite of multiple clinical events of varying significance.We therefore decided to use "time to next exacerbation" as our primary endpoint and assess other clinical data, including FEV 1 , as secondary endpoints.
In our trial, the cumulative prednisone dose in the standard arm will be 560 mg.The cumulative dose used in the largest available study for hospitalised patients (the SCCOPE study [15]) amounts to ~2000 mg prednisone-equivalent in the short (2 weeks) and ~2600 mg in the long (8 week) treatment arm.We did not see a rationale to use such high dosage in our trial other than the attempt to rigorously copy the study design used in that trial.Thus, we decided to use a more practical and ethically more justifiable dose, although this will result in a treatment difference in the standard arm as compared to the SCCOPE study.
Apart from the intervention, we decided to rigorously standardise COPD medication, particularly inhaled steroids and bronchodilators, and the use of antibiotics during the initial treatment phase.At the time our study was designed, the fixed combined use of long-acting beta agonists, inhaled steroids, and tiotropium was beyond generally accepted treatment recommendations.However, we wanted to minimise any treatment bias that might have lead to falsification of the effects of systemic glucocorticoids.After our study had been initiated, the TORCH trial was published showing that, compared to placebo, inhaled long-acting beta-agonists combined with inhaled glucocorticoids reduce exacerbation rate, improve health status, and preserve lung function [34] .Similarly, in the UPLIFT study, tiotropium was shown to improve lung function and quality of life as well as reduce COPD exacerbations [35] .These results are reassuring in view of the treatment plan chosen in our study protocol.As for antimicrobial therapy, broad spectrum antibiotics have been shown to improve clinical outcome and, in hospitalised patients, reduce mortality [36].
Three previous trials conducted at the University Hospital Basel advocate procalcitonin-guided antibiotic treatment in patients with lower respiratory tract infections and exacerbated COPD [28,29,37] .The introduction of a procalcitoninbased treatment algorithm would have complicated our study protocol substantially.Also, published guidelines recommend antibiotic use in patients with altered sputum characteristics [2,8] , but there is a large scope for discretion as to the choice of the antibiotic and the duration of treatment [7,36] .In the present study, an unbalanced distribution of antibiotic use in the two treatment arms has to be avoided, especially as physicians are likely to prescribe antibiotics more frequently in patients with slow improvement, which could be the case in the short treatment arm.Thus, antibiotic treatment bias could compromise our non-inferiority design and obscure the true treatment effects from systemic corticosteroids.
We were confronted with doubts whether in the emergency room setting, the execution of an ACTH test prior to the initiation of glucocorticoid therapy is feasible.However, in a smaller trial studying the HPA axis in nine patients with exacerbated COPD, we were able to demonstrate the feasibility of this approach [38].The low-dose (1 µg) ACTH test measures plasma cortisol at baseline and 30 min.after stimulation, omitting the value 60 min.after stimulation, which is usually measured in the conventional (250 µg) ACTH test.Thus, the low-dose ACTH test does not cause a relevant delay in treatment.
In conclusion, the REDUCE study aims to clarify whether a 5-day course of systemic glucocorticoid therapy is noninferior to a 14-day course for COPD patients with an acute exacerbation.The trial will add significant evidence regarding the treatment of exacerbated COPD with systemic glucocorticoids.Furthermore, the study addresses other important therapeutic issues, such as the practice of glucocorticoid withdrawal and adverse effects of systemic glucocorticoids.

Funding / potential competing interests
Department of Internal Medicine, University Hospital Basel, Switzerland; Clinic of Internal Medicine, Biel-Bienne, Switzerland; Freie Akademische Gesellschaft Basel, Switzerland; Fonds für Lehre und Forschung, Basel, Switzerland; Astra-Zeneca (Switzerland); Viollier Laboratories, Allschwil, Switzerland; Gottfried und Julia Bangerter-Rhyner-Stiftung, Switzerland.None of the sponsors had a role in the development of the trial design, in the analysis of the results, or in the writing of this report.