563
CONTROLLED TRIAL OF CYCLOPHOSPHAMIDE IN RHEUMATOID ARTHRITIS ALEXANDER S. TOWNES, JAMES M. SOWA, and LAWRENCE E. SHULMAN
Twenty-four patients with severe progressive rheumatoid arthritis were randomly assigned to cyclophosphamide or placebo in a double-blind crossover trial. Eleven patients who completed 9 months on cycloplosphamide (average dose: 1.8 mg/kg/day) demonstrated significant decrease in painful joints, swollen joints, and moroing stiffness and increase in grip strength when compared to 11 patients on placebo. After crossover, significant improvement was observed in patients switched to cyclophosphamide, and deterioration within 2 months was observed in most patients changed from drug to placebo. Serum immunoglobulins and rheumatoid factor titers decreased with cyclophosphamide but antibody response to Vi antigen was unaffected. Primary delayed immune response to 2,4-dinitrochlorobenzene was markedly depressed. Adverse effects were troublesomehemorrhagic cystitis affected 4 patients and amenorrhea occurred in 3. Despite its striking beneficial effect, cyclophosphamide From the Department of Medicine, Johns Hopkins University School of Medicine, Baltimore, Maryland. Supported by USPHS Grant AM 12041 from the National Institute of Arthritis, Metabolism, and Digestive Diseases. Alexadder S. Townes, M.D.: Associate Professor of Medicine, Johns Hopkins University School of Medicine (presently Professor of Medicine, University of Tennessee College of Medicine, and Chief, Medical Service, Memphis VA Hospital); James M. Sowa, M.D.: Assistant in Medicine, Johns Hopkins University School of Medicine; Lawrence E. Shulman, M.D., Ph.D.: Associate Professor of Medicine, Johns Hopkins University School of Medicine. Address reprint requests to Alexander S. Townes, M.D., Chief, Medical Service, VA Hospital, 1030 Jefferson Avenue, Memphis, Tennessee 38104. Submitted for publication September 23. 1975; accepted November I I , 1975. Arthritis and Rheumatism, Vol. 19, No. 3 (May-June 1976)
should be prescribed cautiously and only in severe resistant cases of rheumatoid arthritis. Rheumatoid arthritis (RA) can be controlled satisfactorily in most patients by the appropriate use of salicylates, other antiinflammatory drugs, gold salts, physical therapy, and orthopedic surgery. However in some patients the disease proceeds aggressively at an intolerable level of activity, either despite these measures or because of limitations of their effectiveness as a result of concomitant adverse effects of therapy. Treatment of such patients with cytotoxic drugs had some early advocates (1-3). Additional trials were stimulated subsequently by the recognition of the importance of cellular and immunologic events in the inflammatory reactions in RA and of the potential suppressive effects of cytotoxic drugs on these mechanisms (4-9). However only two of these reports presented controlled studies of cyclophosphamide (8,9). This study was designed as a controlled trial to be carried out in a single arthritis clinical center to evaluate therapy with cyclophosphamide in patients with severe RA resistant to conventional therapy. Results confirm the efficacy of cyclophosphamide in controlling the activity of RA and provide additional data concerning the effect on immune responses and on the duration of clinical response after crossover to placebo. The study also demonstrates that a moderately t o markedly effective agent can be evaluated by means of a controlled therapeutic trial in a relatively small sample of patients with RA.
TOWNES ET AL
564
MATERIALS AND METHODS Selection of Patients
Drug Assignment and Treatment Program
Patients were considered for the trial if they were adults who fulfilled criteria for classic rheumatoid arthritis (10) and had active disease of at least 2 years duration. All patients exhibited evidence of progression of disease toward production of deformity or radiologic joint damage and showed persistent activity in multiple joints (as demonstrated by swelling, tenderness, synovial effusion, or hypertrophy) in spite of adequate conventional therapy with salicylates, physical therapy, and at least one other antirheumatic agent such as gold salts, antimalarials, or corticosteroids. Patients were excluded if a ) they had a serious complicating medical problem, b) they had ever previously received immunosuppressive therapy, c) they were taking more than 10 mg per day of prednisone or its equivalent, or d) they had received gold salts or antimalarial drugs within the 2 months before entry into the study. T o ensure adherence to these criteria, evaluation of patients was carried out on at least two visits over an interval of 4 to 6 weeks. Patients were also evaluated by a research social worker to identify and exclude those patients considered potentially unreliable in keeping their appointments or in taking medications. The social worker also ascertained whether the family supported participation in the study. Those tentatively admitted were then hospitalized for detailed medical evaluation and optimum adjustment in their treatment program of salicylates, physical therapy, and in some cases maintenance doses of corticosteroids. Those patients who improved during this pretreatment evaluation or showed substantial variations of disease activity were then excluded from the trial. All patients were given detailed information on the hazards and potential benefits of the study and on the available alternative methods of treatment.
Patients were randomly assigned t o drug or matching placebo tablet and crossed over t o the opposite regimen after 9 months. They were informed that they would receive both drug and placebo during the trial but they were not told the schedule or design of the crossover. Although unaware of drug assignment, physician observers did know the 9-month crossover design. In order to preserve the blindfold nature of the trial as much as possible, one of the authors, designated as the therapist, was responsible for all changes in dosage of cyclophosphamide but was not involved in clinical assessments. Results of WBC and differential counts were withheld from patients’ records so that they would not be known to the physicians responsible for clinical observations. Patients were started on a daily oral dose of 2 mg of cyclophosphamide per kilogram of body weight, given in 50mg tablets each morning before breakfast, or on an equivalent number of matching placebo tablets. After a review of a checklist of possible adverse drug effects, the therapist regulated the dose of cyclophosphamide for each patient at 2-week intervals according to the WBC, the goal being to maintain the WBC between 3000 and 4000/mm3 without toxicity. However the daily dose was not to exceed 3.5 mg/kg/day. Placebo doses were changed at intervals in a way designed to mimic changes of active drug. Patients were given a new supply every 2 weeks throughout the trial. Any unused tablets were returned and counted at each visit. Salicylates, physical therapy, and the small pre-established maintenance doses of corticosteroids (in 10 patients) were continued. All other antirheumatic drugs were prohibited. Decrease of corticosteroid dosage was permitted, but no increase over that at the beginning of the trial was allowed. Propoxyphene was allowed as a supplemental analgesic if necessary; the number taken was recorded.
Clinical Assessment
Immunologic Methods
At each pretreatment assessment and at monthly assessments throughout the study, patients were evaluated by the same rheumatologist. Presence or absence of tenderness to pressure, pain on passive motion, and swelling were recorded in a systematic manner according t o methods developed by the Cooperating Clinics Committee of the American Rheumatism Association ( I I). Grip strength was measured with a folded sphygmomanometer cuff. Circumference of the finger joints was measured with jeweler’s rings graded in half sizes from one to thirteen or with an expandable ring tape for sizes larger than thirteen. Circumference of wrists, elbows, knees, and ankles was measured at fixed bony landmarks with a tape measure. The time required to walk 50 feet, duration of morning stiffness, fatigue time, and the number of aspirins and other analgesic tablets ingested were also recorded. The Westergren method was used to determine erythrocyte sedimentation rate. Patients’ and physicians’ assessments of status were recorded on a scale ranging from “very good” to “very bad” and progress since the previous visit from “much better” to “much worse” ( I 1). All observations were recorded at each monthly visit. Radiographs of the hands obtained at onset and at each 9-month interval were evaluated by a radiologist who was unaware of drug assignments.
Humoral Antibody Response. Serum was obtained at 3month intervals for FII latex tests (12) and sheep cell agglutination test using 1/4 basic agglutination titer of a chromatographically isolated IgG fraction of rabbit amboceptor in a microtiter modification of the method of Heller et a1 (1 3). Serum immunoglobulin concentrations were measured by radial immunodiffusion (14) before treatment, a t 9 months, and a t 18 months. In order to determine the effect of cyclophosphamide on primary humoral immune response at the end of the first 9 months of the trial, all patients were subcutaneously injected with 50 pg of Vi antigen from E coli (kindly supplied by Dr. George Santos). Serum samples were taken 7, 14, 21, and 28 days after injection, and antibody titers were measured by hemagglutination according to the method of Landy and Lamb (15). Delayed Immune Response. Patients were skin tested with tuberculin, trichophyton, Candida, or mumps antigens before therapy and at 9 and 18 months. Induration and erythema were recorded at 24 and 48 hours. To test the effect of drug on primary delayed skin response, 2,4-dinitrochlorobenzene (DNCB) was applied as described by Brandriss5 mg followed in 4 days by a booster of 1 mg (16). Two weeks later patients in whom a delayed skin reaction was not observed a t the primary or booster sites were tested with
CYCLOPHOSPHAMIDE IN RA
100 pg of DNCB and the reaction was observed at 48 hours. At the edd of the 18-month trial, all patients were again tested
565
Table 1. Pretreatment Clinical Comparison
with 100 pg of DNCB.
To evaluate cell migration to an inflammatory stimulus, cells obtained at 4, 8, 12, and 24 hours on skin-window cover slips (17) were counted before therapy, at 9 months, and at 18 months. Two patients had technically unsatisfactory preparations at one or more intervals and these results were excluded.
Statistical Methods Statistical analyses of dichotomous variables were done with the chi square test with Yates correction, and continous variables were analyzed with the two sample ranks test of Wilcoxon as modified by White (18) or the matched pairs signed ranks test of Wilcoxon (19). Two-tailed probability levels were used throughout in testing the null hypothesis.
RESULTS Patients Entered Twenty-four patients were admitted to the study. By random assignment 13 received cyclophosphamide during the first 9 months (Period I ) and 11 received placebo. Baseline data on patients assigned to drug as compared to those assigned to placebo (Table 1) indicate that the groups were comparable. Measures of disease activity at the beginning of the trial suggest slightly more activity in those first assigned to drug (Table 3), although none of the observed differences was striking or approached statistical significance. Of the patients initially on drug, 2 had to be withdrawn from the study-one after only 2 months because of an unrecognized pre-existing illness (carcinoma of the lung), and another at 5 months because of hemorrhagic cystitis. Thus 11 patients on drug and 11 on placebo completed Period I and entered the second 9 months of the study, Period 11, in which drug and placebo assignments were switched. During Period 11, 2 patients were withdrawn while receiving placebo-1 patient because of increasing severity of arthritis requiring steroid therapy after 6 months, and another because of gross hematuria at 6 months. During Period 11, 1 patient on cyclophosphamide was withdrawn after 6 months because of severe hemorrhagic cystitis. Thus 9 patients on placebo and 10 patients on drug completed the 9 months after crossover. Overall, 21 patients received cyclophosphamide for a 9-month period.
Drug Dosage The dose of cyclophosphamide was varied with the intent of using leukopenia as an indicator of adequate drug dosage. None of the patients was leukopenic at the beginning of the study (median WBC: 7,600; 80%
Number of patients Classic RA Females Caucasians Functional Class I1 111
IV Anatomical Stage I I 111 Subcutaneous nodules On low-dose steroids Rheumatoid factor positive Median age (years and 80%range) Median duration of arthritis (years and 80%range)
Patients Assigned to Cyclophosphamide
Patients Assigned to Placebo
13 13
II II 7 6 8 2
8
I 10 3 0 3
10 5
1
3 8
5 6
4 12 52 (43-65)
55 (43-63)
10 (4-23)
13 (4-22)
10
range: 5,650-10,250). The drug dose required to maintain WBC in the prescribed range of 3,000 to 4,000 was variable, and frequent changes of dose were required. Table 2 records the average daily dose during 3-month intervals of the study. The median of the average dose of cyclophosphamide for the entire 9-month period for all patients was 1.83 mg/kg/day. This average dose is comparable to that described in previous trials ( 6 3 ) but was more variable and had higher peak levels in some patients because of the study design. The goal of maintaining the WBC within prescribed limits was accomplished in most patients (Table 2).
Measures of Disease Activity The crossover design of this study allowed comparisons between three groups. During the first 9 months before crossover, patients randomly assigned to drug could be compared to the control group assigned to placebo. After the crossover, patients could be compared to themselves as controls in the two trial periods. Response to drug as compared to placebo in patients who received placebo first might be considered somewhat analagous to a second controlled trial. Finally comparison between drug and placebo periods in patients who received drug first could demonstrate residual drug effect after crossover to placebo and could provide some additional evidence of efficacy of drug as compared to placebo. A summary of results of measures of disease activity is presented in Table 3. All results are expressed as change from the baseline measures immediately be-
TOWNES ET AL
566
Table 2. Drug Dose and M i r e Cell Count
Average Daily Dose 1-3 Months
Median mg/kg/day
Median WBC/mma
* Figures in
3-6 Months
1.81 (1.3-2.3)*
6-9 Months
I .73 (0.9-2.5)
1.85 (1.05-2.5)
1-9 Months 1.83 (0.98-2.2)
3 Months
6 Months
9 Months
3750 (3200-5400)
4000 (2500-5000)
3300 (2500-4500)
parentheses indicate 80% range.
fore the trial for Period I and a s change from the baseline before the crossover for Period 11. Because the means can be skewed by a few widely aberrant values, all results are expressed as median rather than mean change. The figures in parentheses indicate the range of distribution of 80% of the cases. At the end of Period I there was striking improvement in measures of disease activity in patients on cyclophosphamide as compared to those on placebo. The median decrease of 13 painful joints (as indicated by tenderness on pressure or pain on passive motion), me-
dian decrease of 13 swollen joints, median increase in grip strength of approximately 40 mm of mercury, and median decrease of 60 minutes of morning stiffness in patients on cyclophosphamide are statistically significhnt as compared to changes observed in patients on placebo. There was also a trend of reduction in walking time and erythrocyte sedimentation rate (ESR) in patients receiving drug, but these changes were not statistically significant. Patients treated with p l a e b o duriilg Period I and with cyclophosphamide during Period I1 also
Table 3i Measures of Disease Activity Median Baseline Measures
Number of painful Joints N um ber of swollen joints Grip strength (mm k g ) Right hand Left hand
50-foot walk time (sec) Morning stiffness (min) Erythrocyte sedimentation rate (mm/hr)
Median Change First 9 Months (Period I )
Drug
Placebo
29.5 (14-41)t 16 ( 1 1-24)
22 (13-40) 10 (6-29)
81.5 (59-156) 92 (46-158)
106 (43-163) 83 (56-148)
19.75 (12-31)
19 (12-51)
-2 (-I2t0-I)
90 (45- 120)
60 (30-18oj
(-120t00) (-3ot0+10)
60 (30-93)
54 (30-82)
Drug
Placebo
-13*' 0 (-31 to -6) (-5 t o 4 i ) - 13.' +I (- 19 to -4) ( - 5 to +6)
40*6 (8-75) 37*d (8-72)
-60*b
No. Patients Improved Period I Drug
Placebo
Drug
Placebo
Drug
Placebo
- 14.5.' (-31 to 10) -9.5*' (-24t07)
+ 13.5 (2-28) +10.5 (5-15)
IO/IO*'
0/9
10/IOZb
0/9
5/11
10 (2-77) 12 (7-48)
- 16 (-32 to + 2 ) -6 (-95t0+24)
7/11
(-31 too)
ll/ll*c
5/11
Il/ll*b
3/11
2 (-21 to +53) l l / l I * d 4 (-31 to481 10/11
6/11
-I (-3tO+6)
-1 10/1 I
15
-18 5 (-49t0 +15)(-17 t o + 2 5 )
No. Patients Improved Period I I
Median Change 9 Mohths After Crossover (Period 11)
+I (2.5 to 16)
-37.5.6
9/10
2/9
10/lO*c
5/9
5/10
I /9
22.5
9/11*b
2/11
(-180t0 15) (15-120)
9/10*b
0/9
7/1 I
4/1 I
- 18 -3 (-28 to +5) (-28 to 3)
6/10
6/9
* Significance of difference (Pboth tails) in cyclophospharnide and placebo groups is indi'cated as follows: *a 0.02, ** = P < 0.05. See text for details of statistical methods. t Numbers in parentheses indicate range of values of 80% of patients in each group.
=
P < 0.001, *b
=
P < 0.01, *= = P
CONTROLLED TRIAL OF CYCLOPHOSPHAMIDE IN RHEUMATOID ARTHRITIS ALEXANDER S. TOWNES, JAMES M. SOWA, and LAWRENCE E. SHULMAN
Twenty-four patients with severe progressive rheumatoid arthritis were randomly assigned to cyclophosphamide or placebo in a double-blind crossover trial. Eleven patients who completed 9 months on cycloplosphamide (average dose: 1.8 mg/kg/day) demonstrated significant decrease in painful joints, swollen joints, and moroing stiffness and increase in grip strength when compared to 11 patients on placebo. After crossover, significant improvement was observed in patients switched to cyclophosphamide, and deterioration within 2 months was observed in most patients changed from drug to placebo. Serum immunoglobulins and rheumatoid factor titers decreased with cyclophosphamide but antibody response to Vi antigen was unaffected. Primary delayed immune response to 2,4-dinitrochlorobenzene was markedly depressed. Adverse effects were troublesomehemorrhagic cystitis affected 4 patients and amenorrhea occurred in 3. Despite its striking beneficial effect, cyclophosphamide From the Department of Medicine, Johns Hopkins University School of Medicine, Baltimore, Maryland. Supported by USPHS Grant AM 12041 from the National Institute of Arthritis, Metabolism, and Digestive Diseases. Alexadder S. Townes, M.D.: Associate Professor of Medicine, Johns Hopkins University School of Medicine (presently Professor of Medicine, University of Tennessee College of Medicine, and Chief, Medical Service, Memphis VA Hospital); James M. Sowa, M.D.: Assistant in Medicine, Johns Hopkins University School of Medicine; Lawrence E. Shulman, M.D., Ph.D.: Associate Professor of Medicine, Johns Hopkins University School of Medicine. Address reprint requests to Alexander S. Townes, M.D., Chief, Medical Service, VA Hospital, 1030 Jefferson Avenue, Memphis, Tennessee 38104. Submitted for publication September 23. 1975; accepted November I I , 1975. Arthritis and Rheumatism, Vol. 19, No. 3 (May-June 1976)
should be prescribed cautiously and only in severe resistant cases of rheumatoid arthritis. Rheumatoid arthritis (RA) can be controlled satisfactorily in most patients by the appropriate use of salicylates, other antiinflammatory drugs, gold salts, physical therapy, and orthopedic surgery. However in some patients the disease proceeds aggressively at an intolerable level of activity, either despite these measures or because of limitations of their effectiveness as a result of concomitant adverse effects of therapy. Treatment of such patients with cytotoxic drugs had some early advocates (1-3). Additional trials were stimulated subsequently by the recognition of the importance of cellular and immunologic events in the inflammatory reactions in RA and of the potential suppressive effects of cytotoxic drugs on these mechanisms (4-9). However only two of these reports presented controlled studies of cyclophosphamide (8,9). This study was designed as a controlled trial to be carried out in a single arthritis clinical center to evaluate therapy with cyclophosphamide in patients with severe RA resistant to conventional therapy. Results confirm the efficacy of cyclophosphamide in controlling the activity of RA and provide additional data concerning the effect on immune responses and on the duration of clinical response after crossover to placebo. The study also demonstrates that a moderately t o markedly effective agent can be evaluated by means of a controlled therapeutic trial in a relatively small sample of patients with RA.
TOWNES ET AL
564
MATERIALS AND METHODS Selection of Patients
Drug Assignment and Treatment Program
Patients were considered for the trial if they were adults who fulfilled criteria for classic rheumatoid arthritis (10) and had active disease of at least 2 years duration. All patients exhibited evidence of progression of disease toward production of deformity or radiologic joint damage and showed persistent activity in multiple joints (as demonstrated by swelling, tenderness, synovial effusion, or hypertrophy) in spite of adequate conventional therapy with salicylates, physical therapy, and at least one other antirheumatic agent such as gold salts, antimalarials, or corticosteroids. Patients were excluded if a ) they had a serious complicating medical problem, b) they had ever previously received immunosuppressive therapy, c) they were taking more than 10 mg per day of prednisone or its equivalent, or d) they had received gold salts or antimalarial drugs within the 2 months before entry into the study. T o ensure adherence to these criteria, evaluation of patients was carried out on at least two visits over an interval of 4 to 6 weeks. Patients were also evaluated by a research social worker to identify and exclude those patients considered potentially unreliable in keeping their appointments or in taking medications. The social worker also ascertained whether the family supported participation in the study. Those tentatively admitted were then hospitalized for detailed medical evaluation and optimum adjustment in their treatment program of salicylates, physical therapy, and in some cases maintenance doses of corticosteroids. Those patients who improved during this pretreatment evaluation or showed substantial variations of disease activity were then excluded from the trial. All patients were given detailed information on the hazards and potential benefits of the study and on the available alternative methods of treatment.
Patients were randomly assigned t o drug or matching placebo tablet and crossed over t o the opposite regimen after 9 months. They were informed that they would receive both drug and placebo during the trial but they were not told the schedule or design of the crossover. Although unaware of drug assignment, physician observers did know the 9-month crossover design. In order to preserve the blindfold nature of the trial as much as possible, one of the authors, designated as the therapist, was responsible for all changes in dosage of cyclophosphamide but was not involved in clinical assessments. Results of WBC and differential counts were withheld from patients’ records so that they would not be known to the physicians responsible for clinical observations. Patients were started on a daily oral dose of 2 mg of cyclophosphamide per kilogram of body weight, given in 50mg tablets each morning before breakfast, or on an equivalent number of matching placebo tablets. After a review of a checklist of possible adverse drug effects, the therapist regulated the dose of cyclophosphamide for each patient at 2-week intervals according to the WBC, the goal being to maintain the WBC between 3000 and 4000/mm3 without toxicity. However the daily dose was not to exceed 3.5 mg/kg/day. Placebo doses were changed at intervals in a way designed to mimic changes of active drug. Patients were given a new supply every 2 weeks throughout the trial. Any unused tablets were returned and counted at each visit. Salicylates, physical therapy, and the small pre-established maintenance doses of corticosteroids (in 10 patients) were continued. All other antirheumatic drugs were prohibited. Decrease of corticosteroid dosage was permitted, but no increase over that at the beginning of the trial was allowed. Propoxyphene was allowed as a supplemental analgesic if necessary; the number taken was recorded.
Clinical Assessment
Immunologic Methods
At each pretreatment assessment and at monthly assessments throughout the study, patients were evaluated by the same rheumatologist. Presence or absence of tenderness to pressure, pain on passive motion, and swelling were recorded in a systematic manner according t o methods developed by the Cooperating Clinics Committee of the American Rheumatism Association ( I I). Grip strength was measured with a folded sphygmomanometer cuff. Circumference of the finger joints was measured with jeweler’s rings graded in half sizes from one to thirteen or with an expandable ring tape for sizes larger than thirteen. Circumference of wrists, elbows, knees, and ankles was measured at fixed bony landmarks with a tape measure. The time required to walk 50 feet, duration of morning stiffness, fatigue time, and the number of aspirins and other analgesic tablets ingested were also recorded. The Westergren method was used to determine erythrocyte sedimentation rate. Patients’ and physicians’ assessments of status were recorded on a scale ranging from “very good” to “very bad” and progress since the previous visit from “much better” to “much worse” ( I 1). All observations were recorded at each monthly visit. Radiographs of the hands obtained at onset and at each 9-month interval were evaluated by a radiologist who was unaware of drug assignments.
Humoral Antibody Response. Serum was obtained at 3month intervals for FII latex tests (12) and sheep cell agglutination test using 1/4 basic agglutination titer of a chromatographically isolated IgG fraction of rabbit amboceptor in a microtiter modification of the method of Heller et a1 (1 3). Serum immunoglobulin concentrations were measured by radial immunodiffusion (14) before treatment, a t 9 months, and a t 18 months. In order to determine the effect of cyclophosphamide on primary humoral immune response at the end of the first 9 months of the trial, all patients were subcutaneously injected with 50 pg of Vi antigen from E coli (kindly supplied by Dr. George Santos). Serum samples were taken 7, 14, 21, and 28 days after injection, and antibody titers were measured by hemagglutination according to the method of Landy and Lamb (15). Delayed Immune Response. Patients were skin tested with tuberculin, trichophyton, Candida, or mumps antigens before therapy and at 9 and 18 months. Induration and erythema were recorded at 24 and 48 hours. To test the effect of drug on primary delayed skin response, 2,4-dinitrochlorobenzene (DNCB) was applied as described by Brandriss5 mg followed in 4 days by a booster of 1 mg (16). Two weeks later patients in whom a delayed skin reaction was not observed a t the primary or booster sites were tested with
CYCLOPHOSPHAMIDE IN RA
100 pg of DNCB and the reaction was observed at 48 hours. At the edd of the 18-month trial, all patients were again tested
565
Table 1. Pretreatment Clinical Comparison
with 100 pg of DNCB.
To evaluate cell migration to an inflammatory stimulus, cells obtained at 4, 8, 12, and 24 hours on skin-window cover slips (17) were counted before therapy, at 9 months, and at 18 months. Two patients had technically unsatisfactory preparations at one or more intervals and these results were excluded.
Statistical Methods Statistical analyses of dichotomous variables were done with the chi square test with Yates correction, and continous variables were analyzed with the two sample ranks test of Wilcoxon as modified by White (18) or the matched pairs signed ranks test of Wilcoxon (19). Two-tailed probability levels were used throughout in testing the null hypothesis.
RESULTS Patients Entered Twenty-four patients were admitted to the study. By random assignment 13 received cyclophosphamide during the first 9 months (Period I ) and 11 received placebo. Baseline data on patients assigned to drug as compared to those assigned to placebo (Table 1) indicate that the groups were comparable. Measures of disease activity at the beginning of the trial suggest slightly more activity in those first assigned to drug (Table 3), although none of the observed differences was striking or approached statistical significance. Of the patients initially on drug, 2 had to be withdrawn from the study-one after only 2 months because of an unrecognized pre-existing illness (carcinoma of the lung), and another at 5 months because of hemorrhagic cystitis. Thus 11 patients on drug and 11 on placebo completed Period I and entered the second 9 months of the study, Period 11, in which drug and placebo assignments were switched. During Period 11, 2 patients were withdrawn while receiving placebo-1 patient because of increasing severity of arthritis requiring steroid therapy after 6 months, and another because of gross hematuria at 6 months. During Period 11, 1 patient on cyclophosphamide was withdrawn after 6 months because of severe hemorrhagic cystitis. Thus 9 patients on placebo and 10 patients on drug completed the 9 months after crossover. Overall, 21 patients received cyclophosphamide for a 9-month period.
Drug Dosage The dose of cyclophosphamide was varied with the intent of using leukopenia as an indicator of adequate drug dosage. None of the patients was leukopenic at the beginning of the study (median WBC: 7,600; 80%
Number of patients Classic RA Females Caucasians Functional Class I1 111
IV Anatomical Stage I I 111 Subcutaneous nodules On low-dose steroids Rheumatoid factor positive Median age (years and 80%range) Median duration of arthritis (years and 80%range)
Patients Assigned to Cyclophosphamide
Patients Assigned to Placebo
13 13
II II 7 6 8 2
8
I 10 3 0 3
10 5
1
3 8
5 6
4 12 52 (43-65)
55 (43-63)
10 (4-23)
13 (4-22)
10
range: 5,650-10,250). The drug dose required to maintain WBC in the prescribed range of 3,000 to 4,000 was variable, and frequent changes of dose were required. Table 2 records the average daily dose during 3-month intervals of the study. The median of the average dose of cyclophosphamide for the entire 9-month period for all patients was 1.83 mg/kg/day. This average dose is comparable to that described in previous trials ( 6 3 ) but was more variable and had higher peak levels in some patients because of the study design. The goal of maintaining the WBC within prescribed limits was accomplished in most patients (Table 2).
Measures of Disease Activity The crossover design of this study allowed comparisons between three groups. During the first 9 months before crossover, patients randomly assigned to drug could be compared to the control group assigned to placebo. After the crossover, patients could be compared to themselves as controls in the two trial periods. Response to drug as compared to placebo in patients who received placebo first might be considered somewhat analagous to a second controlled trial. Finally comparison between drug and placebo periods in patients who received drug first could demonstrate residual drug effect after crossover to placebo and could provide some additional evidence of efficacy of drug as compared to placebo. A summary of results of measures of disease activity is presented in Table 3. All results are expressed as change from the baseline measures immediately be-
TOWNES ET AL
566
Table 2. Drug Dose and M i r e Cell Count
Average Daily Dose 1-3 Months
Median mg/kg/day
Median WBC/mma
* Figures in
3-6 Months
1.81 (1.3-2.3)*
6-9 Months
I .73 (0.9-2.5)
1.85 (1.05-2.5)
1-9 Months 1.83 (0.98-2.2)
3 Months
6 Months
9 Months
3750 (3200-5400)
4000 (2500-5000)
3300 (2500-4500)
parentheses indicate 80% range.
fore the trial for Period I and a s change from the baseline before the crossover for Period 11. Because the means can be skewed by a few widely aberrant values, all results are expressed as median rather than mean change. The figures in parentheses indicate the range of distribution of 80% of the cases. At the end of Period I there was striking improvement in measures of disease activity in patients on cyclophosphamide as compared to those on placebo. The median decrease of 13 painful joints (as indicated by tenderness on pressure or pain on passive motion), me-
dian decrease of 13 swollen joints, median increase in grip strength of approximately 40 mm of mercury, and median decrease of 60 minutes of morning stiffness in patients on cyclophosphamide are statistically significhnt as compared to changes observed in patients on placebo. There was also a trend of reduction in walking time and erythrocyte sedimentation rate (ESR) in patients receiving drug, but these changes were not statistically significant. Patients treated with p l a e b o duriilg Period I and with cyclophosphamide during Period I1 also
Table 3i Measures of Disease Activity Median Baseline Measures
Number of painful Joints N um ber of swollen joints Grip strength (mm k g ) Right hand Left hand
50-foot walk time (sec) Morning stiffness (min) Erythrocyte sedimentation rate (mm/hr)
Median Change First 9 Months (Period I )
Drug
Placebo
29.5 (14-41)t 16 ( 1 1-24)
22 (13-40) 10 (6-29)
81.5 (59-156) 92 (46-158)
106 (43-163) 83 (56-148)
19.75 (12-31)
19 (12-51)
-2 (-I2t0-I)
90 (45- 120)
60 (30-18oj
(-120t00) (-3ot0+10)
60 (30-93)
54 (30-82)
Drug
Placebo
-13*' 0 (-31 to -6) (-5 t o 4 i ) - 13.' +I (- 19 to -4) ( - 5 to +6)
40*6 (8-75) 37*d (8-72)
-60*b
No. Patients Improved Period I Drug
Placebo
Drug
Placebo
Drug
Placebo
- 14.5.' (-31 to 10) -9.5*' (-24t07)
+ 13.5 (2-28) +10.5 (5-15)
IO/IO*'
0/9
10/IOZb
0/9
5/11
10 (2-77) 12 (7-48)
- 16 (-32 to + 2 ) -6 (-95t0+24)
7/11
(-31 too)
ll/ll*c
5/11
Il/ll*b
3/11
2 (-21 to +53) l l / l I * d 4 (-31 to481 10/11
6/11
-I (-3tO+6)
-1 10/1 I
15
-18 5 (-49t0 +15)(-17 t o + 2 5 )
No. Patients Improved Period I I
Median Change 9 Mohths After Crossover (Period 11)
+I (2.5 to 16)
-37.5.6
9/10
2/9
10/lO*c
5/9
5/10
I /9
22.5
9/11*b
2/11
(-180t0 15) (15-120)
9/10*b
0/9
7/1 I
4/1 I
- 18 -3 (-28 to +5) (-28 to 3)
6/10
6/9
* Significance of difference (Pboth tails) in cyclophospharnide and placebo groups is indi'cated as follows: *a 0.02, ** = P < 0.05. See text for details of statistical methods. t Numbers in parentheses indicate range of values of 80% of patients in each group.
=
P < 0.001, *b
=
P < 0.01, *= = P
