Azathioprine Combined with Prednisone in the Treatment of Idiopathic Pulmonary Fibrosis: A Prospective Double-blind, Randomized, Placebo-controlled Clinical Trial 1 , 2
GANESH RAGHU, WILLIAM J. DEPASO, KEVIN CAIN, SAMUEL P. HAMMAR, CLAUDE E. WETZEL, DAVID F. DREIS, JOHN HUTCHINSON, NEELEY E. PARDEE, and RICHARD H. WINTERBAUER
Idiopathic pulmonary fibrosis (IPF) demonstrates substantial variability in histologic pattern, rate of disease progression, and response to corticosteroids (1-4). These variables interrelate as patients with disease of less than 1 yr duration and less fibrosis on biopsy have a more favorable response to therapy (1). Unfortunately, the latter group is a minority and, more commonly, patients with IPF have a slowlyprogressive illness with disease duration greater than 2 yr, a fibrotic histologic pattern, and no response to corticosteroids. A diagnosis of IPF carries a grave prognosis, as onethird to one-half of patients with IPF are dead at the end of 5 yr of follow-up (1-4). High-dose corticosteroids result in clinical improvement in 15to 300,10 of patients (4, 5). Corticosteroid therapy is considered the conventional treatment for patients with IPF, and adjunct immunosuppressive therapy is usually reserved for patients not responding to treatment (6, 7). In 1978 we reported 20 patients with progressiveIPF treated with combination prednisone and azathioprine with significant improvement in pulmonary function in 12and enhanced survival in those with measurable physiologic improvement (1). Based on this experience we initiated a prospective, double-blind, randomized, placebo-controlled study to compare the effect of combined prednisone/azathioprine with prednisone alone in the treatment of IPF. Methods
Study Population Symptomatic adult patients with diffuse pulmonary infiltrates presenting to the Virginia Mason Medical Center (VMMC) or the University of Washington Medical Center (UWMC) between 1982and 1987were evaluated for possible entry into the study. Patients underwent a standard evaluation that included a thorough history and physical examina-
SUMMARY 'TWenty-sevennewly diagnosed patients with Idiopathic pulmonary fibrosis (IPF) who were previously untreated for IPF were enrolled In a prospective, double-blind, randomized, placebocontrolled study to compare the therapeutic effect of combined prednlsoneJaz8thloprlne (n 14) with prednisone plus placebo (n = 13). Prednisone was started at 1.5 mgllcglday (not to exceed 100 mg/day) for the first 2 wk followed by a biweekly taper to a malntanance dose of 20 mglday. Azathioprine was administered at a dally dose of 3 mg/kg (not to exceed 200 mglday). The patients tolerated the use of IIZ8thloprine well with few associated side effects. Changes In lung function at 1 yr, as measured by resting alveolar-arterial oxygen difference P[A-aI02, FVC, and single breath diffusing capacity for carbon monoxide (DLcOSB)' were all somewhat better In the azathioprine/prednisone group compared with the plWdnlsone alone group, although none of the.. comparisons were statistically significant. Six of 14 (43%) patients randomized to prednisone plus azathioprine died during the 9-yr follow-up period, compared with 10 of 13 (77%) patients randomized to prednisone plus placebo. A Cox model survlvel analysis shows a nonsignificant but potentially large survlvel edventage for azathioprine/prednisone (hazard ratio 0.48,with 95% confidence Interval Increasing from 0.17 to 1.38). When adjusted for age, the survival advantage of azathioprine/prednisone becomes marginally significant (hazard ratio 0.26, with 95% confidence Interval Increasing from 0.08 to 0.88; P = 0.02 by large sample approXimation, p = 0.05 by randomization test). We conclude that combined prednisone and azathioprine Is a safe and possibly effective regimen for the treatment of IPF. A larger randomized study to confirm the results shown here Is clearly Indicated.•
AM REV RESPIR DIS 1911; 144:291-296
tion, screening blood chemistry profiles, antinuclear antibody (ANA), and rheumatoid factor. In addition, the patients' serum was screened by Oucterlony-gel diffusion techniques for precipitating antibodies to Micropolyspora faeni, Thermoactinomyces
vulgaris, Aureobasidiumpu//alus,AspergillusSPecies (A. flavus, A. fumigatus, A. niger), and pigeon serum and droppings. Patients with collagen vascular disease (CVD), hypersensitivity pneumonitis, pneumoconioses, drug-induced diffuse pulmonary injury, or irradiation fibrosis wereexcluded. A diagnosis of IPF was suspected in patients with diffuse reticulonodular infiltrates on chest roentgenogram, bibasilar crackles, finger clubbing, absence of fever, decrease in FVCand DLcOSB, and no evidence of extrathoracic disease (1). Only patients who were previously untreated and available for routine follow-up at either VMMC (RW) or UWMC (OR) were considered candidates for the study. The diagnosis of IPF was supported by lung biopsy in all patients. Tissue sections wereexamined in detail by an experienced pulmonary pathologist without knowledge of the clinical details. The histologic diagnosis of IPF was based on typical microscopic findings in the absence of infection, granuloma, vasculitis, or malig-
nancy (1). Atransbronchiallung biopsy (TBB) that demonstrated interstitial fibrosis in the absence of tumor, infection, or granuloma was accepted as evidence for IPF in only patients with typical clinical, roentgenographic, and physiologic features of IPF who either refused to have an open lung biopsy (OLB) or had a documented family history of familial IPF. All patients included in the study demonstrated progressive clinical disease by fulfilling one or more of the following criteria: (1) progressive dyspnea from day of onset; (2) progressive roentgenographic parenchymal abnormality; (3) 10070 or greater decrease in FVC or total lung capacity compared with previous values; or (4) 20% or greater reduction in DLCOSB compared with previous values. The duration of patients' respiratory (Received in original form October 15, 1990 and in revised form February 8, 1991) 1 From the Departments of Medicine, Biostatistics, and Pathology, the University of Washington Medical Center, and the Virginia Mason Medical Center, Seattle, Washington. 2 Supported by a grant from the Virginia Mason Research Center, Seattle, WA.
292 symptoms before lung biopsy was arbitrarily taken as the clinical duration of IPF.
Histologic Grading of Pulmonary Fibrosis Wedge biopsy lung specimens were obtained from at least two different areas from one lung. Sections for light microscopy were stained with hematoxylin and eosin, trichrome stain for collagen, methenamine silver stain for fungi and Pneumocystiscarinii, and the fluorochrome stain for acid-fast organisms. Quantitative grading of pulmonary fibrosis at light microscopy was performed by previously published criteria (I). Briefly, the degree of fibrosis was determined in 12randomly selected regions from each OLB. A region consisted of the amount of tissue in an area 2 mm square. Fibrosis was graded zero to 4 plus (0 = normal, 4 + = severe abnormality) through comparison with a previously defined series of reference slides (I). The rating score for fibrosis was obtained by calculating the arithmetic mean of the histologic gradings from the 12 regions. Fibrosis in transbronchiallung biopsy specimens was not graded. Measurement of Pulmonary Function Pulmonary function was measured by flowvolume curve, DLCOSB' and arterial blood gas at rest with calculation of the P[A-a]O]. The flow-volumecurve was measured with an Ohio spirometer (P.K. Morgan Instrument Inc, Andover, MA) and interpreted according to Schoenberg and coworkers (8). The DLCOSB was measured with a Medscience Model 572 Diffusion (Medscience Electronics, Inc., St. Louis, MO) corrected for hemoglobin concentration, and interpreted according to Ogilvie and associates (9) and Dinkara and colleagues (10). Both the FVC and DLCOSB are presented and analyzed as percent predicted normal. The P[A-a]O] was calculated assuming atmospheric pressure of 760 mm Hg and gas exchange ratio, R = 0.8. Treatment Protocol Patients were randomized by a method of block randomization (groups of 10) by a research pharmacist (CEW) to one of two double-blinded treatment groups. Informed consent was obtained from all patients in accordance with the Human Subjects Review Committees of the Virginia Mason Medical Center and the University of Washington. Treatment was not begun until at least 2 wk after OLB. All medications were dispensed in a double-blind manner by the research pharmacist. Patients in Group I receivedprednisone and a placebo. Patients in Group 2 received prednisone and azathioprine. All patients received oral prednisone according to an identical protocol. The initial dose was 1.5 mg/kg/day (not to exceed 100 mg/day) for the first 2 wk followed by a fortnightly decrease of 20 mg/day until a dose of40 mg/day was reached. The dosage was further decreased in 5 to 10 mg/day decrements every 2 wk according to patient tolerance in an effort to achieve a maintenance dose of 20 mg/day
RAGHU, DEPASO, CAIN, HAMMAR, WETZEL, DREIS, HUTCHINSON, PARDEE, AND WINTERBAUER
or less. Azathioprine was administered at a daily dose of 3 mg/kg/day (not to exceed 200 mg/day) to the nearest 25 mg dose increment for the duration of the trial. A similar number of placebo tablets were dispensed to the patients in Group I. None of the patients were taking concurrent medications such as cyclophosphamide, d-penicillamine, nonsteroidal antiinflammatory agents, colchicine, or any other agents that could potentially influence the course of IPF. This was particularly controlled because patients were treated and followed closely by only two physicians (RW or GR).
(3) 10% or greater improvement in resting P[A-a]O]. Patients who worsened had decrements of similar magnitude in one or more lung function tests. Lesser degrees of change in lung function were categorized as unchanged. The analyses of each of the three measurements were done separately to assess the outcome with the independent variables rather than overall patient improvement. Because IPF is a disorder with grave prognosis in the majority of patients, enhanced survival was considered the ultimate treatment success with IPF. Long-term survival refers to patients alive at the time of this report.
Adverse Effects All new symptoms and signs that manifested during the treatment period and unrelated to IPF were documented as possible adverse effects. Any undesirable pharmacologic effects on other organ systems, drug toxicity, allergic, and idiosyncratic reactions that manifested during the treatment protocol were noted. Patients were asked specific questions during each visit per a standardized form approved by the Human Subjects Review Committee of the University of Washington. Adverse effects were carefully monitored and categorized as subjectiveor objective. The treatment code was broken by the treating physician and the patient crossed over to the other treatment arm if any of the following occurred: (1) nausea, vomiting, or diarrhea unresponsive to symptomatic therapy; (2) white blood count less than 3,500/ml; (3) platelet count less than 80,000/ml; (4) respiratory failure requiring mechanical ventilation; (5) coma; (6) abnormal liver function tests; (7) rapid disease progression; (8) patient's request.
Statistical Methods Data are expressed as means ± standard error. Ordinal and interval variables were compared between the two groups using the Wilcoxon rank sum test. All p values are twosided. Fisher's exact test was used to compare proportions. Patients who died before I yr wereincluded in the analysis of change in lung function at I yr by assigning them the worst possible change. All analyses comparing the two treatment groups were done on an intent-to-treat basis. This meant that patients were analyzed according to the group into which they wereoriginally randomized, regardless of whether they later crossed over to the other treatment group. This method of analysis is conservative if the treatment is in fact effective but is preferable as it preserves the validity of the randomized study (11-13). Since patients who crossed over are likely to be very different from those who did not, this is the most appropriate statistical method for dealing with crossovers as it ensures that the actual type I error rate is not greater than 0.05. In addition, the intent-totreat analysis also addresses the question of most clinical relevance by comparing two treatment strategies: (1) initiate treatment with prednisone alone and continue until deterioration or adverse reactions, versus (2) initiate treatment with prednisone plus azathioprine and continue until deterioration or adverse reactions. Kaplan-Meier survival plots and the Cox proportional hazards regression model were used to analyze survival time (14). The Cox model is analogous to multiple linear regression in that one can test the effect of one variable, such as treatment assignment, while adjusting for other important variables, such as age. The Cox model measures the importance of a variable in terms ofthe hazard ratio, which can be interpreted as follows: if the hazard ratio for azathioprine/prednisone relative to placebo/prednisone is 0.5, this means that the probability that a patient on azathioprine/prednisone will die in a short time interval is one-half the probability that a patient on placebo/prednisone will die in the same short time interval. The usual p values produced by the Cox model are based on an approximation that is reasonable for very large sample sizes. However, this approximation can be very poor for small sample sizes. Wehave therefore also calculated p values for
Follow-up All patients were followed either at VMMC (RW) or at UWMC (GR). Patients wereevaluated after I, 3, 6, 9, and 12 months for response to therapy and evidence of adverse effects. Follow-up evaluation included: posteroanterior and lateral chest radiographs, flow~mumecurve,D~om,arterialbIDodgas
analysis at rest breathing room air, liver function tests, and fasting plasma glucose. A complete blood count with differential was performed weekly for the first month and then monthly for the duration of the trial. At 12 months patients were maintained on the same regimen with attempts to decrease prednisone to minimal dose and followed at intervals determined by their respective physicians. Patients' compliance with taking the medications was established by checking with patients, family members, and by patients' refills.
End Points and Measurements of Therapeutic Outcome The two end points examined for therapeutic outcome were (1) measureable change in lung function at I yr and (2) survival. Improvement in lung function was defined as anyone of: (1) 10% or greater improvement in FVC, (2) 20070 or greater improvement in DteOSB,
THERAPY OF IDIOPATHIC PULMONARY FIBROSIS
293 TABLE 1
STUDY POPULATION Group 1 Prednisone Plus Placebo (n = 13)
Group 2 Prednisone Plus Azathioprine (n = 14)
7 M, 6 F 54 ± 3 23 ± 6 65 ± 4
5 M, 9 F 58 ± 2 26 ± 6 70 ± 4
0.41 0.87 0.45
40 ± 4 35 ± 4
48 ± 5 36 ± 3
2.5 ± 0.3
2.8 ± 0.4
Sex Age, yr Clinical duration of illness. months FVC, % predicted OLCOSB' mllmin/mm Hg hemoglobin corrected. 0Al predicted P(A-al02 , mm Hg rest Severity of fibrosis on open lung biopsy
treatment effectiveness based on a randomi- tion tests at 1 yr, and surviving months zation test, which gives valid p values even' for all individual subjects are shown in in small samples (15).
The two patient groups were comparable in age, clinical duration of disease before randomization, pretreatment pulmonary function testing, and amount of fibrosis on OLB (table 1). All patients were available for serial follow-up. Clinical variables, change in pulmonary func-
table 2. IPF was confirmed histologically in 23 patients by OLB. One patient in Group 1 and three patients in Group 2 were diagnosed as IPF by transbronchiallung biopsies. Four patients with typical clinical and radiologic features of IPF and compatible TBB were also included in the study group. One of these patients had a documented family history for IPF by open lung biopsy in his siblings (Pa-
tient 4 in Group 2, table 2). This patient's autopsy subsequently confirmed IPF. The other three patients (two in Group 2 and one in Group 1) had no clinical evidence of collagen vascular disorders, hypersensitivity pneumonitis, or other disorders attributable to interstitial lung disorders. Twenty-seven patients were thus eligiblefor randomization. Thirteen patients (seven men and six women) were allocated to Group 1 (prednisone/placebo) and 14 (five men and nine women) to Group 2 (prednisone/azathioprine). Four patients from each group died before completing 1 yr treatment, and three patients were crossed to the opposite group because of clinical deterioration. In Group 1there werethree deaths caused by respiratory failure occurring at 2, 2, and 3 months of treatment. Group 2 also had three deaths caused by respiratory failure after 2, 6, and 10months of therapy. A single patient in each group died of acute myocardial infarction before completing 1 yr treatment. 1\vo patients in Group 1 showed rapid deterioration on prednisone alone and were switched to prednisone/azathioprine after 4 and
TABLE 2 CLINICAL VARIABLES OF STUDY SUBJECTS Age at Commencement of Treatment (yr)
Group 1 (n 1 2 3 4 5 6 711 8 9 1011 11 12 13
= 13) Prednisone F F F F M M M M M F M F M
28 42 42 50 52 55 57 59
60 61 61 65 70
Clinical Duration (months)
U U U
U U U U
0 I U
0 0 0 0 0
0 0 0 0 0
0 0 0 0 0 0
3 1 4 -§
1 1 4 3
Grade of Fibrosis
Alive (A) or Dead (D)
Plus Placebo 8 18
36 12 24 16 12
48 4 24 8 8
3 1 2 3 3 3 1 2 2 4 4 2 -§
90 65 53 69 57 30 56 48 5 32 1.7 2.5 1.5
0 A A
0 0 0 0 0 0 0 0
Group 2 (n 14) Prednisone Plus Azathioprine M 44 6 1 2 F 46 6 20 F 48 3 4 M 52 30 10 M 53 5 F 60 6 54 7 F 54 36 8 F 60 12 24 9 F 63 10 F 64 84 1111 F 24 67 5 12 M 68 13 M 3 68 14 F 48 71
• FVC '" % predicted.
t OLeo '"' OLeo (corrected for hemoglobin) mllmin/mm Hg (% predicted). :j:: I '"' improved. U • unchanged, 0 '"' deteriorated. § Degree of fibrosis not graded in TBB specimens.
80 53 104 1.5 79 93 5.4 13 41 10 19 99 1.5 85
A A A
0 A A 0 0 A 0 0 A
RAGHU, DEPASO, CAIN, HAMMAR, WETZEL, DREIS, HUTCHINSON, PARDEE, AND WINTERBAUER
ly within each of three age categories are also shown in table 5.
COMPARISON OF MEAN CHANGE IN PULMONARY FUNCTION AFTER ONE YEAR OF THERAPY Group 1 Prednisone Plus Placebo (n = 9)
Group 2 Prednisone Plus Azathioprine (n = 10)
+1.7 ± 7.4 +0.9 ± 5.7 -1.0 ± 2.0
+6.5 ± 5.3 +7.3 ± 5.3 -6.0 ± 4.0
0.87 0.70 0.12
Change in FVC, % predicted Change in DLCOSB' % predicted Change in rest P[A-a]02' mm Hg
TABLE 4 IMPROVED, UNCHANGED, OR DETERIORATING PULMONARY FUNCTION AFTER ONE YEAR'S TREATMENT Group 2 Prednisone Plus Azathioprine (n = 14)
Group 1 Prednisone Plus Placebo (n = 13)
FVC DLCOSB PIA-a]02 at rest
3 2 3
3 5 3
7 6 7
6 5 6
Adverse Effects During 12 Months of Therapy The total number of adverse effects was 25 in patients treated with prednisone alone and 28in patients who receivedazathioprine and prednisone (table 6). There were no episodes of pancytopenia in patients receiving azathioprine. Two patients manifested bacterial pneumonias in Group 2. These episodes resolved promptly with antibiotic therapy. There were a large number of adverse effects that appeared to be related to prednisone. The frequency of these eventsdid not differ between the two groups. The cumulative dose of prednisone during the first 12 months was the same for patients in both Groups 1 and 2.
• Deteriorated includes patients who died in the first yr of treatment.
3 months. One patient in Group 2 showed deterioration on prednisone/azathioprine, and azathioprine was discontinued at 6 months.
Frequency of Improvement in Lung Function The mean change in pulmonary function tests for the 19 patients who survived at 1 yr are shown in table 3. Patients in Group 2 had better improvement in lung function by all of the measures but not significantly so. Seven of 14 patients treated with prednisone/azathioprine had improved gas exchange at 1 yr compared with only three patients in the prednisone alone group (p = 0.24) (table 4). The FVC improved in five patients in Group 2 and in three patients in Group 1 (p = 0.68). Changes in the DLCOSB were the same in both groups with two patients improved in Group 1 and three patients in Group 2 (p = 1.00). Survival Sixteen of the 27 patients initially randomized died during the observation
~ :c ~
period including thirteen who died of respiratory failure from progressive IPF. Two deaths were from acute myocardial infarction, and one patient in Group 1 committed suicide after 69 months of follow-up. There were four deaths in the first year in both groups. The survival curves of Group 1 and Group 2 patients are compared in figure 1. Comparison of survival in the two treatment groups by Cox regression model was not statistically significant (P = 0.16, table 5). However, age was an important prognostic factor for these patients. The ages and surviving months of individual study patients are listed in table 2. All but one of the patients less than 50 yr of age are still alive, whereas all but three of the patients greater than 59 yr of age are dead. The three older survivors are all in the azathioprine group. A Cox model analysis that adjusts for age as a continuous variable shows a marginally significant survival advantage in the azathioprine group (P = 0.02 by large sample approximation, p = 0.05 by randomization test; table 5). Survival analyses done separate-
1'0~ 0.8 0.6
tL _ _
~-----I L __ ,
Fig. 1. Survival curves of patients in Group 1 (prednisone plus placebo) and Group 2 (prednisone plus azathioprine). Age adjusted analysis shows a significant survival advantage in the patients treated with combined prednisone plus 0.02, see text). Solid azathioprine (p line prednisone + azathioprine; dotprednisone + placebo. ted line
Based on our pilot experience (1), we initiated the present study to seeif combined prednisone and azathioprine therapy was superior to prednisone alone in the treatment of IPF. The present report is the first prospective, randomized, doubleblind, placebo-controlled treatment and long-term follow-up study of patients newly diagnosed with IPF. A major strength of this study is the homogeneous patient population. The study population truly consisted of patients with IPF as all patients with suspected hypersensitivity pneumonitis, CVD, or other known causes of interstitial fibrosis were very carefully excluded. Patients with CVD were excluded as we do not regard collagen vascular associated interstitial lung disease as equivalent to IPF. Rather, patients with scleroderma and rheumatoid interstitial lung disease in our experience have an immutable illness, whereas patients with systemic lupus erythematosus, polymyositis, dermatomyositis, and Sjogren's syndrome frequently have a prednisone responsive interstitial syndrome. The inclusion of either group thus could produce a significant distortion of the therapeutic response of patients with IPF. None of the patients had received previous treatment with medications known to possibly influence the outcome of IPF. None of the patients in either of the treatment groups were lost to follow-up during our study period. Weattribute the excellentcompliance to the study design, which was intentionally kept simple.Only patients who consented to be followed by one of us for a minimum of 1 yr were
THERAPY OF IDIOPATHIC PULMONARY FIBROSIS
295 TABLE 5
RESULTS OF COX MODEL SURVIVAL ANALYSIS OF TREATMENT GROUPS Hazard Ratio of Azathioprine Plus Prednisone Compared with Prednisone Alone
Model All subjects: Not adjusted for age Adjusted for age Subjects < 50 yr SUbjects 50 to 59 yr Subjects ~ 60 yr
a 0.68 0.30
95% Confidence Interval
p Value Based on Large Sample Approximation
p Value Based on Randomization Test
0.17,1.38 0.08, 0.88 0,6.36 0.11, 4.07 0.07,1.19
0.16 0.02 0.32 0.67 0.07
TABLE 6 ADVERSE EFFECTS DURING THERAPY
Subjective Gastrointestinal Neuropsychiatric Objective Elevated liver enzymes Vertebral fractures Acne Cushingoid features Hypertension Diabetes treatment Oral agent Insulin Congestive heart failure Myocardial infarction Urosepsis Bacterial pneumonia Herpes zoster Cataracts Myopathy Peptic ulcer disease Total
Group 1 Prednisone Plus Placebo
Group 2 Prednisone Plus Azathioprine
1 1 1 25
1 2 1 28
enrolled in the study. Because all patients ment compared with prednisone alone, were to receive the conventional medical (2) a marginally significant improvement treatment with prednisone, and azathio- in survival with combined therapy at up prine had been previously used success- to 9 yr follow-up, (3) excellent tolerance fully in some patients by ourselves and of azathioprine with very few associated others (1, 16-18), patients felt secure in side effects. All the patients enrolled in receiving prednisone as a known treat- the study had clinically progressive pulment and were willing to take the chance ' monary disease defined by either progresof receiving either azathioprine or place- sive dyspnea, worsening roentgenographbo in a blinded manner. Follow-up visits ic abnormality, and/or deterioration in and objective assessment were also sim- measured lung function. Because prople and essentially conformed to com- gressive IPF rarely has a spontaneous munity standard of care. Repeated inva- remission, it can be assumed that therasive procedures such as sequential bron- py interrupted a downhill course to crechoalveolar lavage were deliberately not ate stable function and enhanced survival included to increase the patients' willing- (19). Quantitative analysis of the. open ness to participate. Only two patients in lung biopsies showed that a fibrosis grade Group 1 were transferred to Group 2, and < 3.0 correlated with improvement in azathioprine was discontinued in one pa- lung function at 1 yr of therapy but not tient in Group 2 in accordance with study with survival at 1 yr or long-term surdesign. These three patients, however, re- vival (data not shown). A possible explamained in the study and continued to be nation is that patients with mild lung followed. fibrosis may respond to treatment with Our study demonstrates: (1) a trend for improved lung function. In contrast, the improved P[A-a]02 at rest after 1 yr of patients with severe fibrosis cannot be combined prednisone/azathioprine treat- reversed to measurably improved lung
function, but therapeutic interruption of a progressive decline can still result in improved survival. The clinical caveat is that no patient should be denied therapy based on the degree of fibrosis seen on open lung biopsy. Although there were no differences in survival between groups during the first 12 months of therapy, it is of interest that 80070 of the patients in Group 2 (combined prednisone and azathioprine) who survived the first 12 months are still alive and clinically stable at last follow-up (eight of 10), compared with 33070 (three of nine) of patients in Group 1 (prednisone plus placebo). It appears that the enhanced survival of patients in Group 1 is only seen after the first 3 yr of therapy (figure 1). One of the difficulties in studying therapeutic effect in IPF is establishing the criteria for improvement. Survival is obviously the indisputable marker of treatment success in the management of diseases with high mortality. We have found dyspnea too subjective an end point to use as a criteria for improvement. Patients commonly have an improved sense of well-being from the euphoric effects of prednisone, and this may decrease their perception of dyspnea in the absence of measurable improvement in lung function. Also, changes on chest roentgenogram are not a sensitive index of change in IPF. In our experience, improvement on the chest roentgenogram had been associated with measurable improvement in FVC and/or P[A-a]02 and not infrequently the latter measurements may be improved with an unchanged chest roentgenogram (1). There are only two randomized trials of prednisone plus adjuvant immunosuppressive therapy in the treatment of IPF; the current study and the cyclophosphamide study of Johnson and colleagues (20). Johnson and associates randomized 43 patients (including 10 with CVO) to either prednisolone alone (60 mg daily for 1 month, then decrease by 5 mg weekly to a maintenance dose of 20 mg every other day) or low dose prednisolone (20 mg every other day throughout the trial) plus 100 to 120 mg cyclophosphamide daily. Fifteen (68 OJo) of 22 patients receiving prednisolone alone and 15 (71070) of 21 patients receiving prednisolonecyclophosphamide died during a 5-yr follow-up period. The authors found no difference in the therapeutic response between the two groups and concluded both therapies equally effective. The equivalence of these two regimens suggests that cyclophosphamide may have a steroid
sparing effect. An alternative interpretation is that low-dose prednisolone is therapeutically equal to high-dose prednisolone, and cyclophosphamide has no influence on the clinical outcome. In our study the dosage of prednisone was the same in both groups with azathioprine the singletherapeutic variable thus avoiding a change in prednisone dose as a factor obscuring the effect of azathioprine. Cyclophosphamide had to be stopped in six patients after 1 to 18 months because of neutropenia, thrombocytopenia, or hemorrhagic cystitis. In the present study azathioprine was not stopped in any patient because of drug toxicity. In an uncontrolled, nonrandomized study, Meier-Sydow and coworkers found no significant differences in survival and/or improvement between patients treated with prednisone, prednisone plus azathioprine, and prednisone plus d-penicillamine (16, 17). Although our study design differed from theirs, the marginally significant survival advantage of combined prednisone and azathioprine in the present report is seen when analysis is adjusted for age. In summary, we have demonstrated a trend to improved survival in patients receiving combined azathioprine and prednisone that is marginally significant by a secondary analysis. However, the magnitude of the observed trend is quite large (hazard ratio approximately 0.5) and tolerance of azathioprine was excellent. Thus azathioprine is potentially a very important drug in the treatment of
RAGHU, DEPASO, CAIN, HAMMAR, WETZEL, DREIS, HUTCHINSON, PARDEE, AND WINTERBAUER
IPF. Since the evidence provided by this study is not strong enough to make clinical recommendations, a new, larger (multicenter) study should be conducted to verify the trends shown here. Our results also suggest that future intervention clinial trials should be long-term (more than 3 yr) and stratified by age because age seems to be a strong determining factor in survival. References 1. Winterbauer RH, Hammar SP, Hallman KO, et 01. Diffuse interstitial pneumonitis. Clinicopathologic correlations in 20 patients treated with prednisone/azathioprine. Am J Med 1978;65(4):661-72. 2. Carrington CB, Gaensler EA, Contu RE, Fitzgerald MX, Gupta RG. Natural history and treated course of usual and desquamative interstitial pneumonia. N Engl J Med 1978; 298(15):801-9. 3. Turner-Warwick M, Burrows B, Johnson A. Cryptogenic fibrosing alveolitis: clinical features and their influence on survival. Thorax 1980; 35:171-80. 4. Rudd RM, Haslam PL, Turner-Warwick M. Cryptogenic fibrosing alveolitis: relationships of pulmonary physiology and bronchoalveolar lavage to response to treatment and prognosis. Am Rev Respir Dis 1981; 124(1):1-8. 5. Stack HR, Choo-Kang YFJ, Heard BE. The prognosis of cryptogenic fibrosing alveolitis. Thorax 1972; 27:535. 6. Raghu G. Idiopathic pulmonary fibrosis: a rational clinical approach. Chest 1987; 92:148-154. 7. Turner-WarwickM. Approaches to therapy. Semin Respir Med 1984; 6:92-102. 8. Schoenberg JB, Beck GH, Bouhuys A. Growth and decay of pulmonary function in healthy blacks and whites. Respir Physiol 1978; 33:367-93. 9. Ogilvie CM, Forster RE, Blakemore WS, Morton JW. A standardized breath holding technique for the clinical measurement of the diffusing capacity of the lung for carbon monoxide. J Clin In-
vest 1957; 36:1-17. 10. Dinkara P, Blumenthal WS, Johnston RF, et 01. The effect of anemia on pulmonary diffusing capacity with derivation of a correction equation. Am Rev Respir Dis 1970; 102:965-9. 11. Peto R, Pike MC, Armitage P, et 01. Design and analysis of randomized clinical trials requiring prolonged observation of each patient. Br J Cancer 1976; 34:585-612. 12. Pocock SJ. Clinicaltrials: a practical approach. New York: Wiley, 1983. 13. Fisher LD, Dixon DO, Herson J, Frankowski RK, Hearron MS, Peace KE. Intention to treat in clinical trials. In: Peace KE ed. Statistical Issues in Drug Development. New York: Marcel Dekker, 1990. 14. Cox DR. Regression models and life tables (with discussion). J R Stat Soc, Series B 1972; 34:187-220. 15. Edgington ES. Randomization Tests. 2nd ed. New York: Marcel Dekker, 1987. 16. Meier-Sydow J, Rust M, Kronenberger H, et 01. Long term follow-upof lung function parameters in patients with idiopathic pulmonary fibrosis treated with prednisone and azathioprine or d-penicillamine. Prax Pneumol 1979; 33:680-99. 17. Meier-Sydow J, Rust M, Kronenberger H, KrempelS. Survival of patients with idiopathic pulmonary fibrosis following treatment with azathioprine, d-penicillamine or prednisone. Chest 1990; 98:18S (abstract). 18. Cegla VH. Behandlung der idiopathisch fibrosier-ender. Alveolitis. Schweiz Med Wochenschr 1977; 107:184-7. 19. Turner-Warwick M, Burrows B, Johnson A. Cryptogenic fibrosing alveolitis: response to corticosteroid treatment and its effect on survival. Thorax 1980; 35:593-9. 20. Johnson MA, Kwan S, Snell NJC, Nunn AJ, Darbyshire JH, Thrner-Warwick M. Randomized controlled trial comparing prednisolone alone with cyclophosphamide and low dose prednisolone in combination in cryptogenic fibrosing alveolitis. Thorax 1989; 44:280-8.