19
CORRELATION OF SEROLOGIC INDICATORS OF INFLAMMATION WITH EFFECTIVENESS O F NONSTEROIDAL ANTIINFLAMMATORY DRUG THERAPY IN RHEUMATOID ARTHRITIS JOHN J. CUSH, PETER E. LIPSKY, ARNOLD E. POSTLETHWAITE, RALPH E. SCHROHENLOHER, ANTHONY SAWAY, and WILLIAM J. KOOPMAN Forty-seven patients with rheumatoid arthritis (mean duration 5.7 years) who were receiving neither disease-modifying drugs nor corticosteroids were enrolled in a 12-week, multicenter study of the relationship between clinical and serologic measures of disease activity in patients taking nonsteroidal antiinflammatory drugs. After a 2-week drug washout period, patients received flurbiprofen (200 mg/day) or sustained-release ibuprofen (2,400 mg/day) for a 10-week trial. Clinical response was assessed biweekly using standard clinical parameters, including 50-foot walk time, tender joint score, duration of morning stitfness, and global assessment of disease activity and pain (by both the patient and the physician). Patients were classified as responders if there was 230% improvement in at least 3 of the 4 clinical measures of disease activity. Thirty patients completed at least 8 weeks of therapy; there were 12 responders and 18 nonresponders. Of the laboratory From the Harold C. Simmons Arthritis Research Center, University of Texas Southwestern Medical Center, Dallas; the University of Tennessee College of Medicine, Memphis; and the University of Alabama at Birmingham Medical Center and Veterans Administration Hospital, Birmingham, Alabama. Supported by USPHS grants AR-09989, AR-39169, and AR-03555, by the VA Research Program, and by The Upjohn Company. John J. Cush, MD: University of Texas Southwestern Medical Center; Peter E. Lipsky, M D University of Texas Southwestern Medical Center; Arnold E. Postlethwaite, M D University of Tennessee College of Medicine; Ralph E. Schrohenloher, PhD: University of Alabama at Birmingham Medical Center; Anthony Saway, MD: University of Alabama at Birmingham Medical Center and VA Hospital; William J. Koopman, MD: University of Alabama at Birmingham Medical Center and VA Hospital. Address reprint requests to Peter E. Lipsky, MD, Department of Internal Medicine, University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, TX 75235. Submitted for publication May 8, 1989; accepted in revised form August 28, 1989. Arthritis and Rheumatism, Vol. 33, No. 1 (January 1990)
parameters examined, the responders, but not the nonresponders, demonstrated significant reductions (from postwashout values) in levels of IgM rheumatoid factor and C-reactive protein (CRP), along with significant increases in the number of circulating lymphocytes and decreases in the number of circulating granulocytes (P5 0.05). In contrast, the nonresponders demonstrated either no change or worsening of the laboratory correlates of disease activity. The responders also appeared to have more aggressive disease at baseline, with significantly more painful joints, greater 50-foot walk times, elevated CRP values, and elevated erythrocyte sedimentation rates (P 5 0.05). These data suggest that there is a subset of rheumatoid arthritis patients in whom clinical improvement with nonsteroidal antiinflammatory drug therapy is associated with significant reductions in IgM rheumatoid factor and CRP levels.
The use of disease-modifying antirheumatic drugs (DMARDs) in the treatment of patients with rheumatoid arthritis (RA) is often associated with improvement or normalization of the serologic indicators of inflammatory activity, including the erythrocyte sedimentation rate (ESR), C-reactive protein (CRP) level, and titer of the IgM autoantibody, rheumatoid factor (RF) (1-3). These characteristic effects have been used to distinguish a DMARD from a nonsteroidal antiinflammatory drug (NSAID), which is considered to afford only symptomatic therapy and is not associated with improvement in serologic correlates of inflammation in RA (4-6).By inhibiting prostaglandin biosynthesis, NSAIDs have been shown to display a variety of antiinflammatory and immunomodulatory effects in vitro (7-13). Nonetheless, there is currently only limited evidence that the administra-
20
CUSH ET AL
Table 1. Clinical profile of rheumatoid arthritis study patients at enrollment and at baseline*
At study enrollment (n = 30) Maledfemales Disease duration (years) 50-foot walk time (seconds) Tender joint score Duration of morning stiffness (hours) Global assessment score C-reactive protein (mg/dl) ESR (mm/hour) RF Nephelometry (IU/ml) ELISA (&ml) Lymphocyte count (/mm3) Monocyte count (/mm3) Granulocyte count (/mm3) ~
At baseline (n = 30)
9/2 1 5.7 f 1.3 19.0 f 1.5
9/2 1 5.7 f 1.3 22.0 f 2 . l t
24.5 f 1.8 2.4 f 0.6
26.1 5.7
9.4 f 0.4 2.39 2 0.4 51.0f 6.1
10.2 IT 0.4 2.67 2 0.5 58.2 -C 6.9
1,354.1 f 319.7 244.4 ? 91.6 1,921.7 f 119.9 357.3 ? 45.1 5,329.8 f 331.7
1,774.2 ? 414.2 221.8 ? 76.8 1,971.5 2 117.5 407.8 f 49.9 5,651.2 ? 314.1
f
1.8
" I.lt
~~
* Baseline values represent values obtained after a washout period, which was begun at enrollment and continued for 2 weeks. Except for the number of males/females, values are the mean f SEM. The tender joint score represents the number of tender joints. See Patients and Methods for definition of the global assessment score. The erythrocyte sedimentation rate (ESR) was determined according to the Westergren method. Rheumatoid factor (RF) was assessed by rate nephelometry in 14 patients and by enzyme-linked immunosorbent assay (ELISA) in I5 patients ( I patient excluded because of incomplete data). The granulocyte count represents the total white blood cell count minus the number of lymphocytes plus monocytes. t P 5 0.05 versus enrollment.
tion of these agents alters the immunologically driven inflammation of RA, as evidenced b y a decrease in the previously elevated levels of acute-phase reactants (14,15) or R F (13,16,17). T h e purpose of this study w a s to determine whether conventional NSAID therapy might result in alterations in levels of acute-phase reactants, serum RF, and/or the numbers of circulating leukocytes or lymphocyte subpopulations, a n d to determine whether such alterations might correlate with clinical responses to NSAID therapy.
PATIENTS AND METHODS Patient seledion. Forty-seven patients with definite
or classic RA (18) of at least 3 months duration were enrolled in a 12-week, multicenter, open-label, study of flurbiprofen and sustained-release ibuprofen. All patients also met the American Rheumatism Association 1987 revised criteria for the diagnosis of RA (19). Criteria for inclusion in the study were the presence of active disease, synovitis of 2 or more joints, and a positive serum R F titer. Table 1 lists the clinical assessments of disease activity. Additionally, no patient had been treated with DMARDs (gold, penicillamine, etc.) or
glucocorticoids (oral or intraarticular) in the 3 months before study entry. The 30 patients who completed at least 8 weeks of NSAID therapy are described in this report. Seventeen patients dropped out: 8 because of lack of efficacy, 6 because of adverse effects (3 with cutaneous reactions, 3 with gastrointestinal reactions, and 1 with azotemia), and 3 for other reasons (1 had another illness, 1 had family problems, and 1 was lost to followup). No serious adverse effects were observed. Study design. Informed consent was obtained, and the patients were enrolled into a prospective, randomized, open-label study utilizing 200 mg of flurbiprofen per day (13 patients) or 2,400 mg of a sustained-release ibuprofen preparation per day (17 patients). At enrollment, all patients were receiving NSAID therapy only, and each patient underwent an initial clinical and laboratory assessment. After enrollment, a 2-week drug washout period was initiated. During the washout period, patients discontinued all NSAID therapy and utilized only analgesic agents (acetaminophen, propoxyphene, a n d o r acetaminophen plus codeine) as needed. After the 2-week washout period, baseline clinical and laboratory evaluations were carried out, and the 10week course of NSAID therapy was begun. Patients were evaluated every 2 weeks thereafter for 10 weeks. Clinical assessments, including duration of morning stiffness, global assessment of disease activity (by patient and physician), tender joint score, and time taken to walk 50 feet, were performed by the same physician in each center throughout the study period. The duration of morning stiffness was expressed in hours. The global assessment score was derived from the sum of the scores on 3 measures: the patient's assessment of pain, the patient's assessment of disease activity, and the physician's assessment of disease activity. Each measure was scored numerically, using a scale of 1-5, where 1 = inactivelvery good, 2 = mildlgood, 3 = moderatelfair, 4 = severe/poor, and 5 = incapacitating/ very poor. The potential range of the global assessment score was 3-15 points. The tender joint score was the total number of tender joints. The number of swollen joints was also determined, and the value correlated significantly with the number of tender joints (r = 0.61, P < 0.OOOl); however, only data concerning the number of tender joints are presented. The above clinical variables were ultimately used to define therapeutic efficacy. A clinical response occurred when there was 230% improvement in at least 3 of the 4 clinical parameters (morning stiffness, global assessment, tender joint score, and 50-foot walk time). Changes in clinical parameters were expressed as the mean percentage of change from baseline (postwashout) values. Laboratory assessment. Standard laboratory screening tests included a complete blood cell count with differential cell count, Westergren ESR, CRP (by rate nephelometry), and serum chemistry profile. Serum IgM-RF values were measured by different methods at different trial sites, with 47% of the sera assayed by rate nephelometry (20) and the remainder by enzyme-linked immunosorbent assay (2 1). (One responding patient was excluded from RF analysis because of incomplete data.) Methodologic differences in IgM-RF assays were normalized and merged by expressing the data as the mean percentage of change from baseline
21
NSAIDs AND SEROLOGIC MEASURES IN RA
Table 2. Clinical and laboratory measures of response to therapy with nonsteroidal antiinflammatory drugs in the entire rheumatoid arthritis patient population*
50-foot walk time (seconds) Tender joint score Duration of morning stiffness (hours) Global assessment score ESR (mmhour) CRP (mddl) RF
Mean change during therapy
Baseline values
Week 4
Week 8
Week 10
22.0 (29) 26.1 (30) 5.7 (27)
- 16.7 (29)t -11.5 (30)t -61.0 (27)t
- 19.4 (29)t -15.9 (30) -42.2 (27)t
0.46 (26) -16.4 (27)t 6.5 (21)
10.2 (30) 58.2 (30) 2.67 (30) - (30)
- 16.6 (30)t -3.2 (26) -0.83 (26)t 76.8 (30)
- 19.6 (30)t -3.4 (27) -0.89 (28)t -14.8 (28)
-25.3 (27)t -4.4 (26) - I .09 (23)t 89.6 (27)
* Values are the mean percentage of change from baseline, except for the erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) levels, which are the mean actual change from baseline. Numbers in parentheses are the numbers of patients. See Table 1 for baseline rheumatoid factor (RF) values and additional explanations. t P 5 0.05 versus baseline. (postwashout) values. At each center, monocytes and leukocyte subsets were enumerated with differential cell counts by light microscopy. There were comparable normal values at the different study sites. Cell preparation and lymphocyte subset analysis. Peripheral blood mononuclear cells (PBMC) were prepared from anticoagulated venous blood samples by density centrifugation on sodium diatrizoate/Ficoll gradients. The cells were washed 3 times in normal saline and stored in phosphate buffered saline (PBS) containing 2% normal human serum (NHS) and 0.1% sodium azide. Indirect immunofiuorescence staining, cell fixation, and analysis. PBMC (4-10 x 16 cells/sample) were stained for 30 minutes at 4°C with saturating concentrations of monoclonal antibodies OKT4 (anti-CD4) and OKT8 (anti-
CD8). After washing twice in PBS with 2% NHS and 0.1% sodium azide, cells were counterstained with fluorescein isothiocyanate-iabeled goat anti-mouse immunoglobulin. Serum and excess immunoglobulin were removed by washing, and the cells were fixed for 12 minutes at room temperature with 1% paraformaldehyde in PBS, pH 7.4. Cells were later analyzed for surface immunofluorescence using a fluorescence-activated cell sorter and analyzer (22). Lymphocytes were analyzed by selective gating, using the parameters of forward and orthogonal light scatter. Statistical analysis. Serial, paired laboratory values (CRP, ESR, and cell subsets) were analyzed as raw data using the Wilcoxon rank sum test. Significant changes in serum IgM-RFvalues were calculated as the percentage of change from baseline values and analyzed for statistical
Table 3. Changes in clinical efficacy variables during therapy with nonsteroidal antiinflammatory drugs in rheumatoid arthritis patients who responded and those who did not respond to treatment*
Enrollment values
Baseline values
Mean change during therapy Week4
Responders (n = 12) 50-foot walk time (seconds) Tender joint score Duration of morning stiffness (hours) Global assessment score Nonresponders (n = 18) 50-foot walk time (seconds) Tender joint score Duration of morning stiffness (hours) Global assessment score
Week8
____~
~~
_____
Week 10 ~
23.4t 27.8 3.3
30.5f 29.~1 5.1
-32.28 -i5.2# -77.5#
-38.38 -25.49 -76.5#
-43.38 -33.49 -76.6#
10.1
10.8
-29.38
-36.28
-41.69
16.0 22.2 1.8
17.5 20.5 6.0
-7.2 -9.1 -49.71
-7.9 -9.5 -18.6
30.5 -2.9 68.9
8.9
9.7
-8.1
-8.1
- 12.3
* Values are the mean percentage of change from baseline; enrollment and baseline values are the population mean. See Table 1 for definitions. t P = 0.04 versus nonresponders. f P = 0.02 versus nonresponders. 5 P 5 0.01 versus baseline. 7 P = 0.03 versus nonresponders. # P 5 0.05 versus baseline.
22
CUSH ET AL
significance using the Wilcoxon rank sum test. Significant changes in the measures of clinical efficacy were calculated as the percentage of change from baseline values and analyzed by Student's 1-test. Population means were analyzed using the 2-sample ?-test. Changes in the values of the clinical and laboratory variables between the baseline and final assessments were correlated using Spearman's rank correlation.
RESULTS Patient characteristics and clinical outcomes. The clinical characteristics of the 30 patients who completed at least 8 weeks of NSAID therapy are presented in Table 1 (17 of the 47 patients withdrew between weeks 2 and 8). The mean duration of disease was 5.7 years; 76% of the patients had had RA for 5 years or less at enrollment. Four patients had previously been treated with a DMARD. Although the sample sizes were small, no significant differences were noted between the group taking ibuprofen (2,400 mg/day) and the group taking flurbiprofen (200 mg/ day). No serious adverse effects from the NSAID therapy were reported by either group. All patients underwent a 2-week drug washout period, during which all NSAIDs were discontinued and only analgesic medication was taken. During this time, there was a significant increase in the 50-foot walk time and the duration of morning stiffness (Table 1).
After 10 weeks of NSAID therapy, the outcome was assessed. Twelve patients were classified as responders and 18 as nonresponders. At the completion of therapy, there was significant improvement in the global assessment score and tender joint score for the entire group of NSAID-treated patients (Table 2). With regard to laboratory assessments, there were no significant alterations in the ESR, serum RF value, or cell subset numbers in the entire patient population. By contrast, CRP values were significantly lower for the entire group after 4 weeks of NSAID therapy (P = 0.02), and remained significantly lower through week 10 of therapy (P = 0.03) (Table 2). Responders and nonresponders were individually assessed for clinical and serologic responses to therapy. In the responder group, significant improvement in all clinical parameters was noted after as little as 2 weeks of therapy (P I0.05) (results not shown), and the level of improvement was maintained or became more marked during the ensuing 8 weeks of therapy (Table 3). In the nonresponder group, only the duration of morning stiffness was significantly improved at week 4 (P I0.01). Three of the nonresponders withdrew between weeks 8 and 10, primarily because of lack of efficacy. As noted in Tables 3 and 4,the responder group exhibited significantly greater mean baseline 50-foot
Table 4. Changes in laboratory variables during therapy with nonsteroidal antiinflammatory drugs in rheumatoid arthritis patients who responded and those who did not respond to treatment*
Responders (n = 12) ESR (mm/hour) CRP (mg/dl) RF Nephelometry (IUlml) ELISA (pg/ml) Nonresponders (n = 18) ESR (mrdhour) CRP (mg/dl) RF Nephelometry (IU/ml) ELISA (pg/ml)
Mean change during therapy
Enrollment values
Baseline values
Week4
Week 8
Week 10
73.4t 3.95t
79.1$ 4.505
-3.6 -2.01
- 13.9 -1.4
-11.9 -2.4#
1,541.2 349.2
1,646.0
36.1 1.32
45.4 1.71
1,213.9 181.5
1,870.4 174.2
4.4
-23.6**
301.2 -2.9 -0.2 0.4
3.9 -0.6
1.6 -0.3
-7.8
-3.3
* Values are the mean absolute change from baseline, except for rheumatoid factor (RF), which is the mean percentage of change from baseline (determined by rate nephelometry in 6 responders and 8 nonresponders and by enzyme-linked immunosorbent assay [ELISA] in 6 responders and 9 nonresponders; 1 responder patient excluded because of incomplete data); enrollment and baseline values are the population mean. Erythrocyte sedimentation rate (ESR) was determined according to the Westergren method. Normal C-reactive protein (CRP) level 50.6 mg/dl, by rate nephelometry. t P 5 0.01 versus nonresponders. j: P = 0.04 versus nonresponders. 5 P = 0.02 versus nonresponders. B P = 0.01 versus baseline. # P = 0.02 versus baseline. ** P = 0.04 versus baseline.
NSAIDs AND SEROLOGIC MEASURES IN RA
walk times, tender joint scores, ESR, and CRP values at baseline compared with the nonresponder group (P < 0.05). Prior to the washout period, there were fewer differences between the responders and nonresponders, with only the 50-foot walk times, ESR, and CRP values observed to be significantly different (greater in the responders). Laboratory assessments. Results of the automated chemical studies of sera, by group, were unremarkable except for sporadic mild elevations in alkaline phosphatase levels without associated elevations in transaminase levels. At baseline, the responders manifested a higher mean CRP level than did the nonresponders. Moreover, significant decreases in this serologic correlate of inflammation occurred only among the responders and within 4 weeks of the beginning of NSAID therapy (Figure 1 and Table 4). The ESR did not change significantly during treatment in either group. The clinical responders, however, manifested a significant decrease in serum RF levels after 8 weeks and 10 weeks of therapy (P < 0.05), which was not observed in the nonresponders (Figure 2 and Table 4). Cell subsets. Comparisons of the 2 patient groups showed no demonstrable differences in monocyte numbers or percentages of CD4+ and CD8+ lymphocytes (Table 5). Moreover, no consistently significant changes in these cell subsets were observed during NSAID therapy, in either the responders or the nonresponders. At baseline, the responders exhibited significantly fewer circulating lymphocytes and significantly more granulocytes than the nonresponders ( P I0.05) (Table 5 ) . Only the responders demonstrated significant alterations in the total number of circulating lymphocytes a$d granulocytes during the NSAlD trial. After as little ;as 4 weeks of therapy, the absolute granulocyte count was significantly decreased (P 5 0.05), and it remained decreased through week 10 in the responders (Figure 3). And by the completion of the therapy, thf: responders also manifested a significant increase (P = 0.03) in the total lymphocyte count (Figure 4). Correlation between clinical efficacy and laboratory findings. Table 6 shows the relationship between improvement in the clinical variables and changes in serologic correlates of inflammation and in cell subset numbers. Improvement in these clinical variables was significantly correlated with reductions in CRP levels and inversely correlated with the number of circulating lymphocytes (P 5 0.05). There was also a significant correlation between the number of circulating granulocytes and global assessment of disease activity (P 5
23
't 1
Baseline
WK 4 0 NON-RESPONDER
WK8
WK10
0 RESPONDER
Figure 1. Mean absolute change from baseline C-reactive protein values for responder and nonresponder rheumatoid arthritis populations during treatment with nonsteroidal antiinflammatory drugs. Values were significant for only the responders, at weeks 4-10 (P5 0.05).
0.05). Significant correlations between laboratory results and other measures of clinical efficacy were not observed.
DISCUSSION These studies were undertaken to assess whether clinical response to NSAID treatment in patients with RA might be associated with changes in the laboratory correlates of systemic inflammation and/or immunologic activity. We used a measure of clinical response that relied on objective and subjective variables provided by patient and physician. This definition was constructed as a practical means whereby a significant antiinflammatory effect, rather than an analgesic effect, might be identified. Love11 et al (23) used a similar, multivariable definition of clinical response to study NSAID treatment in patients
24
2r
CUSH ET AL
10
-30 Baseline
WK 4 0
WK8
WK10
NON-RESPONDER 0 RESPONDER
Figure 2. Mean percentage of change from baseline serum IgM rheumatoid factor (IgM-RF) levels for responder and nonresponder rheumatoid arthritis populations during treatment with nonsteroidal antiinflammatory drugs. Values were significant for only the responders, at weeks 8 and 10 (P 0.05).
with juvenile RA. We used conventional clinical variables to assess clinical response (i.e., joint score, walk time, morning stiffness, and global assessments of disease activity). The patient’s response status was assessed independently of the laboratory data analysis. The capacity of this index to differentiate responders from nonresponders and the ability to correlate therapeutic response with laboratory evidence of diminished inflammatory activity suggest the practical utility of this method of assessing clinical outcome in therapeutic trials. Clinical and laboratory data presented in Tables 3 and 4 indicate that at the initiation of the study, the subset of responding patients exhibited more active disease than did the subset of nonresponding patients. In fact, compared with the nonresponders, the responders had significantly greater disease activity at baseline, as assessed by the %foot walk time, tender joint score, ESR, and CRP level ( P < 0.05). Although some patients with RA may improve spontaneously, the large percentage of patients demonstrating significant improvement according to the clinical criteria employed (40%) and the significant correlation between clinical efficacy variables and serologic correlates of inflammation strongly suggest that the improvements noted were neither random nor unrelated to therapy. However, in the absence of a placebotreated control group, it is impossible to conclude with certainty that the responses were related to the NS AID therapy. Since their introduction nearly 30 years ago,
Table 5. Profile of cell subsets during therapy with nonsteroidal antiinflammatory drugs in rheumatoid arthritis patients who responded and those who did not respond to treatment*
Responders (n = 12) Lymphocytes (/mm’) % CD4+ % CD8+ CD4:CD8 ratio Monocytes (/mm3) Granulocytes (/mm3) Nonresponders (n = 18) Lymphocytes (Imm’) % CD4+ % CD8+ CD4:CD8 ratio Monocytes (/mm3) Granulocytes (/mm3)
Mean change during therapy
Baseline values
Week 4
1,685.6t 37.2 22.8 1.94 466.8 6,435.2t
184.2 8.0$ 6.4 0.07 -6.9 - 1,174.0$
2,137.1 45.8 27.4 1.91 373.6 5,097.9
- 179. I
4.8 -1.1 0.3 1 $
-22.5 258.0
Week 8
Week 10
556. I $ 8.5$ 4.0 0.59 - 112.8 - 1,464.0f
496.8$ 6.0 3.8 0.30 - 195.8 -1,264.1$
-203.3 2.9 -0.3 0.21 -1.4 217.8
-76.3 1.8 -0.2 0.41 -90.1 - 17.7
* Values are the mean absolute change from baseline; baseline values are the population mean. The granulocyte value represents the total white blood cell count minus the number of lymphocytes plus monocytes. t P 5 0.05 versus nonresponders. $ P 5 0.05 versus baseline.
25
NSAIDs AND SEROLOGIC MEASURES IN RA
5L 600
300
500
0
-
400
cl
0
. 5 E E
v)
-
E E
. 300 In
-300
a,
0 v
2 200 c
c
C
3 0
3
0
2-
0 In
2 -600 0
0
0 -
c
C
E 1
Q
3
9
0 c
m
100
V x
%
O
c
-900
-100
r”
-
-200
-1200
-1 500
I
Baseline 0
I
I
I
WK 4
WK8
WK10
NON-RESPONDER 0RESPONDER
-400Baseline
WK8
WK 4
WK10
Figure 3. Mean absolute change from baseline granulocyte counts for responder and nonresponder rheumatoid arthritis populations during treatment with nonsteroidal antiinflammatory drugs. Values were significant for only the responders, at weeks 4-10 (PI0.05).
NSAIDs have served as a cornerstone of antirheumatic therapy because of their ability to reduce symptoms in a substantial number of patients (6,13,14,17). By virtue of their antiinflammatory properties, NSAIDs are capable of reducing joint swelling and duration of morning stiffness and improving many of the clinical parameters used to gauge disease activity in RA. In contrast to the DMARDs, the clinical response to an NSAID is thought to be rapid, usually within days or weeks. However, NSAIDs do not appear to correct the elevations in the levels of acutephase reactants and serum R F titers that are often seen in patients with active RA (2-6). These results have fostered the widely held opinion that NSAIDs have no capacity to modulate the chronic inflammatory response or the immune abnormalities associated with RA. We have reported here the favorable effects of NSAID therapy in a subset of RA patients. The early clinical responses (within 2 weeks) were ultimately
Correlation between changes in clinical variables and changes in immunologic variables during therapy with nonsteroidal antiinflammatory drugs in the entire rheumatoid arthritis patient population*
Table 6.
Clinical parameter Laboratory parameter CRP level
ESR RF level Lymphocyte count % cD4+ % CD8+ Monocyte count Granulocyte count
Tender joint score
Global assessment
0.44t 0.25 0.23 -0.32t -0.01 0.0s 0.09 0.21
0.421 0.29 0.19 -0.34t 0.01 -0.26 0.37 0.46t
* Values are Spearman’s rank correlation coefficient. See Table 4 for definitions and explanations of abbreviations. t Correlation significant at P 5 0.05.
26
associated with antiinflammatory and, possibly, immunomodulatory effects, as manifested by significant reductions in CRP levels, serum RF titers, and numbers of circulating granulocytes, and augmentation of the numbers of circulating lymphocytes. Our findings support the possibility that these agents may offer more than symptomatic therapy and suggest that they may have a more profound effect on RA. Discrepancies between the data presented here and those presented elsewhere may be partly explained by methodologic differences that include the presentation of average population data without consideration of individual clinical response status, the short duration ( 5 6 weeks) or low dosage used in many of the early NSAID studies, and possibly, variable effects of different pharmacologic compounds on rheumatoid inflammation (8). In contrast to earlier studies, our results were derived from a population of patients categorized according to clinical response status, who, for the most part, had not received DMARD therapy. Moreover, the use of newer and more sensitive quantitative assays for serum R F and CRP may have provided an advantage over traditional dilutional measurements, thereby permitting changes to be detected. The effect of NSAIDs on R F production has been observed by other investigators. Goodwin et a1 (13) studied 20 RA patients during a 10-week trial of piroxicam. After a 2-week drug washout period and the initiation of the NSAID, they noted clinical improvement that was associated with significant decreases in mean serum RF titers. Effects on the acute-phase reactants were not examined. Those investigators proposed that the in vivo inhibition of cyclooxygenase activity and prostaglandin E, (PGE,) production led to more effective suppression of RF synthesis. Earlier in vitro studies by the same investigators had demonstrated that PGE, inhibited the action of CD8+ suppressor cells, thereby augmenting the in vitro production of IgM-RF (7). NSAIDs prevented this contrasuppressive action of PGE,, and thereby facilitated the down-regulation of autoantibody production. Others have similarly claimed that therapy with piroxicam or fenclofenac might decrease serum RF titers (16,17). In the current studies, we noted a similar effect of NSAID administration on the serum RF titer, but only in patients who improved clinically. Whefher the reduction of RF titers in these patients reflects a direct immunomodulatory effect of the drugs or is secondary to amelioration of the inflammatory process (24) cannot be ascertained from these data.
CUSH ET AL
Numerous studies have demonstrated the inability of NSAID therapy to influence serologic correlates of inflammation, including the ESR and CRP (2-6). Elevations in the levels of acute-phase reactants are characteristic findings in patients with active RA (1-3). Although the factors contributing to an augmented acute-phase response are numerous, it has been proposed that local production of monocyte and lymphocyte-derived cytokines, such as interleukin-1 (IL-1), tumor necrosis factor (TNF), and IL-6 within the synovial tissue may be responsible for the acutephase response seen in RA (25-28). Reduction of the levels of acute-phase reactants might therefore be anticipated to correlate with reduction in disease activity since these same cytokines are thought to play a central role in the pathogenesis of joint inflammation and joint damage in RA (2,27,28). Whereas some of the DMARDs are thought to interfere with the production of monocyte-derived cytokines in vitro, it has been demonstrated that PGE, suppresses, and NSAIDs augment, IL-1 and TNF production in vitro (29,30). Moreover, it is believed that the action of these cytokines on hepatic production of acute-phase reactants is not mediated by prostaglandins. Thus, the finding that NSAIDs suppress the acute-phase response was unanticipated; it suggests that in a subset of patients, the decrease in CRP might be secondary to suppression of inflammation. The rapid decline in CRP levels, however, leaves open the possibility that NSAIDs have a direct effect on the production of, or the response to, the cytokines responsible for the induction of CRP synthesis. The finding that CRP levels decreased even in some patients who failed to respond clinically (8 of 18 patients) suggests the possibility that this decrease was an effect of the NSAIDs that were used and not a secondary effect related to the control of inflammation. In contrast with the findings by Goodwin and coworkers (13), we, like most other investigators, were unable to correlate clinical status or therapeutic response to alterations in the number of circulating CD4+ or CD8+ T cells in RA (31,32). Surprisingly, we observed decreased granulocyte numbers in patients who responded to NSAID therapy, and there was a significant correlation between the decrease in granulocyte numbers and improvement in global assessment scores. These findings are interesting because NSAIDs have been shown to alter a number of the functional activities of neutrophils (33). These data suggest that successful therapy may be associated with
NSAIDs AND SEROLOGIC MEASURES IN RA
decreases in the number of circulating neutrophils and, possibly, with inhibition of neutrophil function. Favorable clinical responses to either sustained-release ibuprofen or flurbiprofen were evident and significant as early as 2 weeks after initiation of treatment, and maximum benefit was generally achieved between weeks 8 and 10. Temporally, optimal clinical responses coincided with the late effects of therapy on the serologic correlates of inflammation, especially with the decline in RF levels. It has previously been suggested that late clinical effects of NSAID therapy occur in some patients (14.23). Our current findings support the conclusion that delayed therapeutic responses to NSAIDs can occur, and that such responses reflect a more profound antiinflammatory or immunomodulatory effect that is associated with decreases in levels of CRP and RF. The goals of therapy in patients with RA include the amelioration of symptoms and protection from the joint damage that is often observed in the disease. Status as a disease-modifying agent should be reserved for those drugs that not only improve clinical parameters and functional status, but also favorably alter serologic correlates of inflammation and RF titers, and possibly, even retard the development of articular erosions in patients with RA. In this study, therapy with NSAIDs alone resulted in improvements in clinical activity, levels of IgM-RF, and levels of CRP in a subset of patients with RA. Whether changes in these serologic indicators of inflammation will ultimately be associated with protection against damage to articular structures can neither be determined nor inferred from this short-term trial. Nonetheless, these data suggest that in a subset of RA patients, NSAIDs can provide clinical, antiinflammatory, and immunomodulatory effects hitherto ascribed only to DMARDs.
ACKNOWLEDGMENTS We acknowledge the cooperation and assistance of The Upjohn Company, in particular, Carol Daenzer, Dr. Derek Stubbs, K. W. Teoh. Rodney Beason, Rod Nunnelee, and Marv Berkowitz.
REFERENCES 1. Dawes F T , Fowler PD, Clarke S, Fisher J, Lawton A,
Shadforth MF: Rheumatoid arthritis: treatment which controls the C-reactive protein and erythrocyte sedimentation rate reduces radiological progression. Br J Rheumatol 25:4449, 1986
27
2. Wright V, Amos R: Do drugs change the course of rheumatoid arthritis? Br Med J 280:964-966, 1980 3. Dixon JS, Bird HA, Sitton NG, Pickup ME, Wright V: C-reactive protein in the serial assessment of disease activity in rheumatoid arthritis. Scand J Rheumatol 13:39-44, 1984 4. McConkey B, Crockson RA, Crockson AP, Wilkinson AR: The effects of some anti-inflammatory drugs on the acute phase proteins in rheumatoid arthritis. Q J Med 42:785-791, 1973 5. Aylward M, Maddock J, Wheeldon R, Parker RJ: A study of the influence of antirheumatic drug regimens on serum acute-phase proteins, plasma tryptophan, and erythrocyte sedimentation rate in rheumatoid arthritis. Rheumatol Rehabil 14:lOl-114, 1975 6. The Cooperating Clinics Committee of the American Rheumatism Association: A three month trial of indomethacin in rheumatoid arthritis, with special reference to analysis and inference. Clin Pharmacol Ther 8: 11-37, 1967 7 Ceuppens JL, Rodriguez MA, Goodwin JS: Nonsteroidal anti-inflammatory agents inhibit the synthesis of IgM rheumatoid factor in vitro. Lancet 1528-530, Lancet I, 1982 8. Spiers EM, Tavendale A, MacConnachie A, Swanson Beck J: The effect of non-steroidal anti-inflammatory drugs on phytohaemagglutinin stimulated lymphocytes: relevance to possible therapeutic immunosuppression by fenclonfenac. Int J Immunopharmacol 10:261-269, 1988 9. Levy J: Enhancement of deficient T-cell function in rheumatoid arthritis by tolmetin sodium. J Clin Pharmacol 23:324-328, 1983 10. Ceuppens JL, Robaeys G, Verdickt W, Vertessen S, Deckmyn H , Dequeker J: Immunomodulatory effects of treatment with naproxen in patients with rheumatic disease. Arthritis Rheum 29:305-311, 1986 11. Thoen J, Helgetveit K, Forre 0, Haile Y, Kass E: Effects of piroxicam and D-penicillamine on T lymphocyte subpopulations, natural killer cells, and rheumatoid factor production in rheumatoid arthritis. Scand J Rheumatol 17:91-102, 1988 12. Veys EM, Verbruggen G, Suykens S, Mielants H, Ackerman K, van Lerbeirghe J, Heynen G: Lymphocyte sub-populations after a single 40 mg. administration of piroxicam in 20 patients with rheumatoid arthritis: a placebo controlled study. Inflammation 8 (suppl): S1 K S 1 2 1 , 1984 13. Goodwin JS, Ceuppens JL, Rodriguez MA: Administration of nonsteroidal anti-inflammatory agents in patients with rheumatoid arthritis: effects on indexes of cellular immune status and serum rheumatoid factor levels. JAMA 250:2485-2488, 1983 14. Blechman WJ, Schmid FR, April PA, Wilson C, Brooks
28
15.
16.
17.
18.
19.
20.
21.
22.
23.
CUSH ET AL
CD: Ibuprofen or aspirin in rheumatoid arthritis therapy. JAMA 233:33&340, 1975 Waltham-Weeks CD: Etodolac versus naproxen in rheumatoid arthritis: a double-blind crossover study. Curr Med Res Opin 10540-547, 1987 Forre 0, Thoen J, Helgetveit K, Haile Y:Non-steroidal anti-inflammatory drugs in rheumatoid arthritis: effects on clinical parameters and cellular immunity. Inflammation 8 (suppI):SlWS113, 1984 Katona G: Fenclofenac: a new nonsteroidal antiinflammatory drug with an immunomodulatory effect in the treatment of active rheumatoid arthritis. Curr Ther Res 37:58-69, 1985 Ropes MW, Bennett GA, Cobb S, Jacox R, Jessar RA: 1958 revision of diagnostic criteria for rheumatoid arthritis. Bull Rheum Dis 9:175-176, 1958 Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF, Cooper NS, Healey LA, Kaplan SR, Liang MH, Luthra HS, Medsger TA Jr, Mitchell DM, Neustadt DH, Pinals RS, Schaller JG, Sharp JT, Wilder RL, Hunder GG: The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 31:315-324, 1988 Hicks MJ, Heick M, Finely P, Gall EP, Minnich L, Williams RJ: Rheumatoid factor activity by rate nephelometry correlated with clinical activity in rheumatoid arthritis. Am J Clin Pathol 78:342-345, 1982 Tomana M, Schrohenloher RE, Koopman WJ, Alarcon GS, Paul WA: Abnormal glycosylation of serum IgG from patients with chronic inflammatory diseases. Arthritis Rheum 31:333-338, 1988 Herzenberg LA, Herzenberg LA: Analysis and separation using the fluorescence-activated cell sorter, Handbook of Experimental Immunology. Edited by DM Weir. Oxford, Blackwell Scientific Publications, 1978 Lovell DJ, Giannini EH, Brewer EJ Jr: Time course of response to nonsteroidal antiinflammatory drugs in
juvenile rheumatoid arthritis. Arthritis Rheum 27: 1433-1437, 1984 24. Olsen NJ, Callahan LF, Pincus T: In vitro rheumatoid factor synthesis in patients taking second-line drugs for rheumatoid arthritis: independent associations with disease activity. Arthritis Rheum 31:1090-1096, 1988 25. Dinarello CA: An update on human interleukin-1: from molecular biology to clinical relevance. J Clin Immunol 51287-297, 1985 26. Beutler B, Cerami A: Cachectin: more than a tumor necrosis factor. N Engl J Med 316:379-385, 1987 27. Houssiau FA, Devogelaer J-P, van Damme J, Nagant de Deuxchaisnes C, van Snick J: Interleukin-6 in synovial fluid and serum of patients with rheumatoid arthritis and other inflammatory arthritides. Arthritis Rheum 31: 784-788, 1988 28. Lipsky PE, Davis LD, Cush JJ, Oppenheimer-Marks N: The role of cytokines in the pathogenesis of rheumatoid arthritis. Springer Semin Immunopathol 11: 123-162, 1989 29. Kunkel SL, Wiggins RE, Chensue SE, Lawick J: Regulation of macrophage tumor necrosis factor production by prostaglandin E,. Biochem Biophys Commun 137: 404410, 1986 30. Dinarello CA: Biology of interleukin 1 . FASEB J 2: 108-115, 1988 31. Cush JJ, Lipsky PE: Phenotypic analysis of synovial tissue and peripheral blood lymphocytes isolated from patients with rheumatoid arthritis. Arthritis Rheum 31: 123&1238, 1988 32. De Vries E, Meijer CJLM, Lafeber GJM, Cnossen J, Cats A: Lymphocyte subpopulations in rheumatoid arthritis: an immunological, enzyme histochemical and morphological study. Rheumatol Int 4:91-94, 1986 33. Perianin A, Giroud J-P, Hakim J: Differential in vivo effects of indomethacin, ibuprofen, and flurbiprofen on oxygen-dependent killing activities of neutrophils elicited by acute nonimmune inflammation in the rat. Inflammation 12:181-189, 1988
CORRELATION OF SEROLOGIC INDICATORS OF INFLAMMATION WITH EFFECTIVENESS O F NONSTEROIDAL ANTIINFLAMMATORY DRUG THERAPY IN RHEUMATOID ARTHRITIS JOHN J. CUSH, PETER E. LIPSKY, ARNOLD E. POSTLETHWAITE, RALPH E. SCHROHENLOHER, ANTHONY SAWAY, and WILLIAM J. KOOPMAN Forty-seven patients with rheumatoid arthritis (mean duration 5.7 years) who were receiving neither disease-modifying drugs nor corticosteroids were enrolled in a 12-week, multicenter study of the relationship between clinical and serologic measures of disease activity in patients taking nonsteroidal antiinflammatory drugs. After a 2-week drug washout period, patients received flurbiprofen (200 mg/day) or sustained-release ibuprofen (2,400 mg/day) for a 10-week trial. Clinical response was assessed biweekly using standard clinical parameters, including 50-foot walk time, tender joint score, duration of morning stitfness, and global assessment of disease activity and pain (by both the patient and the physician). Patients were classified as responders if there was 230% improvement in at least 3 of the 4 clinical measures of disease activity. Thirty patients completed at least 8 weeks of therapy; there were 12 responders and 18 nonresponders. Of the laboratory From the Harold C. Simmons Arthritis Research Center, University of Texas Southwestern Medical Center, Dallas; the University of Tennessee College of Medicine, Memphis; and the University of Alabama at Birmingham Medical Center and Veterans Administration Hospital, Birmingham, Alabama. Supported by USPHS grants AR-09989, AR-39169, and AR-03555, by the VA Research Program, and by The Upjohn Company. John J. Cush, MD: University of Texas Southwestern Medical Center; Peter E. Lipsky, M D University of Texas Southwestern Medical Center; Arnold E. Postlethwaite, M D University of Tennessee College of Medicine; Ralph E. Schrohenloher, PhD: University of Alabama at Birmingham Medical Center; Anthony Saway, MD: University of Alabama at Birmingham Medical Center and VA Hospital; William J. Koopman, MD: University of Alabama at Birmingham Medical Center and VA Hospital. Address reprint requests to Peter E. Lipsky, MD, Department of Internal Medicine, University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, TX 75235. Submitted for publication May 8, 1989; accepted in revised form August 28, 1989. Arthritis and Rheumatism, Vol. 33, No. 1 (January 1990)
parameters examined, the responders, but not the nonresponders, demonstrated significant reductions (from postwashout values) in levels of IgM rheumatoid factor and C-reactive protein (CRP), along with significant increases in the number of circulating lymphocytes and decreases in the number of circulating granulocytes (P5 0.05). In contrast, the nonresponders demonstrated either no change or worsening of the laboratory correlates of disease activity. The responders also appeared to have more aggressive disease at baseline, with significantly more painful joints, greater 50-foot walk times, elevated CRP values, and elevated erythrocyte sedimentation rates (P 5 0.05). These data suggest that there is a subset of rheumatoid arthritis patients in whom clinical improvement with nonsteroidal antiinflammatory drug therapy is associated with significant reductions in IgM rheumatoid factor and CRP levels.
The use of disease-modifying antirheumatic drugs (DMARDs) in the treatment of patients with rheumatoid arthritis (RA) is often associated with improvement or normalization of the serologic indicators of inflammatory activity, including the erythrocyte sedimentation rate (ESR), C-reactive protein (CRP) level, and titer of the IgM autoantibody, rheumatoid factor (RF) (1-3). These characteristic effects have been used to distinguish a DMARD from a nonsteroidal antiinflammatory drug (NSAID), which is considered to afford only symptomatic therapy and is not associated with improvement in serologic correlates of inflammation in RA (4-6).By inhibiting prostaglandin biosynthesis, NSAIDs have been shown to display a variety of antiinflammatory and immunomodulatory effects in vitro (7-13). Nonetheless, there is currently only limited evidence that the administra-
20
CUSH ET AL
Table 1. Clinical profile of rheumatoid arthritis study patients at enrollment and at baseline*
At study enrollment (n = 30) Maledfemales Disease duration (years) 50-foot walk time (seconds) Tender joint score Duration of morning stiffness (hours) Global assessment score C-reactive protein (mg/dl) ESR (mm/hour) RF Nephelometry (IU/ml) ELISA (&ml) Lymphocyte count (/mm3) Monocyte count (/mm3) Granulocyte count (/mm3) ~
At baseline (n = 30)
9/2 1 5.7 f 1.3 19.0 f 1.5
9/2 1 5.7 f 1.3 22.0 f 2 . l t
24.5 f 1.8 2.4 f 0.6
26.1 5.7
9.4 f 0.4 2.39 2 0.4 51.0f 6.1
10.2 IT 0.4 2.67 2 0.5 58.2 -C 6.9
1,354.1 f 319.7 244.4 ? 91.6 1,921.7 f 119.9 357.3 ? 45.1 5,329.8 f 331.7
1,774.2 ? 414.2 221.8 ? 76.8 1,971.5 2 117.5 407.8 f 49.9 5,651.2 ? 314.1
f
1.8
" I.lt
~~
* Baseline values represent values obtained after a washout period, which was begun at enrollment and continued for 2 weeks. Except for the number of males/females, values are the mean f SEM. The tender joint score represents the number of tender joints. See Patients and Methods for definition of the global assessment score. The erythrocyte sedimentation rate (ESR) was determined according to the Westergren method. Rheumatoid factor (RF) was assessed by rate nephelometry in 14 patients and by enzyme-linked immunosorbent assay (ELISA) in I5 patients ( I patient excluded because of incomplete data). The granulocyte count represents the total white blood cell count minus the number of lymphocytes plus monocytes. t P 5 0.05 versus enrollment.
tion of these agents alters the immunologically driven inflammation of RA, as evidenced b y a decrease in the previously elevated levels of acute-phase reactants (14,15) or R F (13,16,17). T h e purpose of this study w a s to determine whether conventional NSAID therapy might result in alterations in levels of acute-phase reactants, serum RF, and/or the numbers of circulating leukocytes or lymphocyte subpopulations, a n d to determine whether such alterations might correlate with clinical responses to NSAID therapy.
PATIENTS AND METHODS Patient seledion. Forty-seven patients with definite
or classic RA (18) of at least 3 months duration were enrolled in a 12-week, multicenter, open-label, study of flurbiprofen and sustained-release ibuprofen. All patients also met the American Rheumatism Association 1987 revised criteria for the diagnosis of RA (19). Criteria for inclusion in the study were the presence of active disease, synovitis of 2 or more joints, and a positive serum R F titer. Table 1 lists the clinical assessments of disease activity. Additionally, no patient had been treated with DMARDs (gold, penicillamine, etc.) or
glucocorticoids (oral or intraarticular) in the 3 months before study entry. The 30 patients who completed at least 8 weeks of NSAID therapy are described in this report. Seventeen patients dropped out: 8 because of lack of efficacy, 6 because of adverse effects (3 with cutaneous reactions, 3 with gastrointestinal reactions, and 1 with azotemia), and 3 for other reasons (1 had another illness, 1 had family problems, and 1 was lost to followup). No serious adverse effects were observed. Study design. Informed consent was obtained, and the patients were enrolled into a prospective, randomized, open-label study utilizing 200 mg of flurbiprofen per day (13 patients) or 2,400 mg of a sustained-release ibuprofen preparation per day (17 patients). At enrollment, all patients were receiving NSAID therapy only, and each patient underwent an initial clinical and laboratory assessment. After enrollment, a 2-week drug washout period was initiated. During the washout period, patients discontinued all NSAID therapy and utilized only analgesic agents (acetaminophen, propoxyphene, a n d o r acetaminophen plus codeine) as needed. After the 2-week washout period, baseline clinical and laboratory evaluations were carried out, and the 10week course of NSAID therapy was begun. Patients were evaluated every 2 weeks thereafter for 10 weeks. Clinical assessments, including duration of morning stiffness, global assessment of disease activity (by patient and physician), tender joint score, and time taken to walk 50 feet, were performed by the same physician in each center throughout the study period. The duration of morning stiffness was expressed in hours. The global assessment score was derived from the sum of the scores on 3 measures: the patient's assessment of pain, the patient's assessment of disease activity, and the physician's assessment of disease activity. Each measure was scored numerically, using a scale of 1-5, where 1 = inactivelvery good, 2 = mildlgood, 3 = moderatelfair, 4 = severe/poor, and 5 = incapacitating/ very poor. The potential range of the global assessment score was 3-15 points. The tender joint score was the total number of tender joints. The number of swollen joints was also determined, and the value correlated significantly with the number of tender joints (r = 0.61, P < 0.OOOl); however, only data concerning the number of tender joints are presented. The above clinical variables were ultimately used to define therapeutic efficacy. A clinical response occurred when there was 230% improvement in at least 3 of the 4 clinical parameters (morning stiffness, global assessment, tender joint score, and 50-foot walk time). Changes in clinical parameters were expressed as the mean percentage of change from baseline (postwashout) values. Laboratory assessment. Standard laboratory screening tests included a complete blood cell count with differential cell count, Westergren ESR, CRP (by rate nephelometry), and serum chemistry profile. Serum IgM-RF values were measured by different methods at different trial sites, with 47% of the sera assayed by rate nephelometry (20) and the remainder by enzyme-linked immunosorbent assay (2 1). (One responding patient was excluded from RF analysis because of incomplete data.) Methodologic differences in IgM-RF assays were normalized and merged by expressing the data as the mean percentage of change from baseline
21
NSAIDs AND SEROLOGIC MEASURES IN RA
Table 2. Clinical and laboratory measures of response to therapy with nonsteroidal antiinflammatory drugs in the entire rheumatoid arthritis patient population*
50-foot walk time (seconds) Tender joint score Duration of morning stiffness (hours) Global assessment score ESR (mmhour) CRP (mddl) RF
Mean change during therapy
Baseline values
Week 4
Week 8
Week 10
22.0 (29) 26.1 (30) 5.7 (27)
- 16.7 (29)t -11.5 (30)t -61.0 (27)t
- 19.4 (29)t -15.9 (30) -42.2 (27)t
0.46 (26) -16.4 (27)t 6.5 (21)
10.2 (30) 58.2 (30) 2.67 (30) - (30)
- 16.6 (30)t -3.2 (26) -0.83 (26)t 76.8 (30)
- 19.6 (30)t -3.4 (27) -0.89 (28)t -14.8 (28)
-25.3 (27)t -4.4 (26) - I .09 (23)t 89.6 (27)
* Values are the mean percentage of change from baseline, except for the erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) levels, which are the mean actual change from baseline. Numbers in parentheses are the numbers of patients. See Table 1 for baseline rheumatoid factor (RF) values and additional explanations. t P 5 0.05 versus baseline. (postwashout) values. At each center, monocytes and leukocyte subsets were enumerated with differential cell counts by light microscopy. There were comparable normal values at the different study sites. Cell preparation and lymphocyte subset analysis. Peripheral blood mononuclear cells (PBMC) were prepared from anticoagulated venous blood samples by density centrifugation on sodium diatrizoate/Ficoll gradients. The cells were washed 3 times in normal saline and stored in phosphate buffered saline (PBS) containing 2% normal human serum (NHS) and 0.1% sodium azide. Indirect immunofiuorescence staining, cell fixation, and analysis. PBMC (4-10 x 16 cells/sample) were stained for 30 minutes at 4°C with saturating concentrations of monoclonal antibodies OKT4 (anti-CD4) and OKT8 (anti-
CD8). After washing twice in PBS with 2% NHS and 0.1% sodium azide, cells were counterstained with fluorescein isothiocyanate-iabeled goat anti-mouse immunoglobulin. Serum and excess immunoglobulin were removed by washing, and the cells were fixed for 12 minutes at room temperature with 1% paraformaldehyde in PBS, pH 7.4. Cells were later analyzed for surface immunofluorescence using a fluorescence-activated cell sorter and analyzer (22). Lymphocytes were analyzed by selective gating, using the parameters of forward and orthogonal light scatter. Statistical analysis. Serial, paired laboratory values (CRP, ESR, and cell subsets) were analyzed as raw data using the Wilcoxon rank sum test. Significant changes in serum IgM-RFvalues were calculated as the percentage of change from baseline values and analyzed for statistical
Table 3. Changes in clinical efficacy variables during therapy with nonsteroidal antiinflammatory drugs in rheumatoid arthritis patients who responded and those who did not respond to treatment*
Enrollment values
Baseline values
Mean change during therapy Week4
Responders (n = 12) 50-foot walk time (seconds) Tender joint score Duration of morning stiffness (hours) Global assessment score Nonresponders (n = 18) 50-foot walk time (seconds) Tender joint score Duration of morning stiffness (hours) Global assessment score
Week8
____~
~~
_____
Week 10 ~
23.4t 27.8 3.3
30.5f 29.~1 5.1
-32.28 -i5.2# -77.5#
-38.38 -25.49 -76.5#
-43.38 -33.49 -76.6#
10.1
10.8
-29.38
-36.28
-41.69
16.0 22.2 1.8
17.5 20.5 6.0
-7.2 -9.1 -49.71
-7.9 -9.5 -18.6
30.5 -2.9 68.9
8.9
9.7
-8.1
-8.1
- 12.3
* Values are the mean percentage of change from baseline; enrollment and baseline values are the population mean. See Table 1 for definitions. t P = 0.04 versus nonresponders. f P = 0.02 versus nonresponders. 5 P 5 0.01 versus baseline. 7 P = 0.03 versus nonresponders. # P 5 0.05 versus baseline.
22
CUSH ET AL
significance using the Wilcoxon rank sum test. Significant changes in the measures of clinical efficacy were calculated as the percentage of change from baseline values and analyzed by Student's 1-test. Population means were analyzed using the 2-sample ?-test. Changes in the values of the clinical and laboratory variables between the baseline and final assessments were correlated using Spearman's rank correlation.
RESULTS Patient characteristics and clinical outcomes. The clinical characteristics of the 30 patients who completed at least 8 weeks of NSAID therapy are presented in Table 1 (17 of the 47 patients withdrew between weeks 2 and 8). The mean duration of disease was 5.7 years; 76% of the patients had had RA for 5 years or less at enrollment. Four patients had previously been treated with a DMARD. Although the sample sizes were small, no significant differences were noted between the group taking ibuprofen (2,400 mg/day) and the group taking flurbiprofen (200 mg/ day). No serious adverse effects from the NSAID therapy were reported by either group. All patients underwent a 2-week drug washout period, during which all NSAIDs were discontinued and only analgesic medication was taken. During this time, there was a significant increase in the 50-foot walk time and the duration of morning stiffness (Table 1).
After 10 weeks of NSAID therapy, the outcome was assessed. Twelve patients were classified as responders and 18 as nonresponders. At the completion of therapy, there was significant improvement in the global assessment score and tender joint score for the entire group of NSAID-treated patients (Table 2). With regard to laboratory assessments, there were no significant alterations in the ESR, serum RF value, or cell subset numbers in the entire patient population. By contrast, CRP values were significantly lower for the entire group after 4 weeks of NSAID therapy (P = 0.02), and remained significantly lower through week 10 of therapy (P = 0.03) (Table 2). Responders and nonresponders were individually assessed for clinical and serologic responses to therapy. In the responder group, significant improvement in all clinical parameters was noted after as little as 2 weeks of therapy (P I0.05) (results not shown), and the level of improvement was maintained or became more marked during the ensuing 8 weeks of therapy (Table 3). In the nonresponder group, only the duration of morning stiffness was significantly improved at week 4 (P I0.01). Three of the nonresponders withdrew between weeks 8 and 10, primarily because of lack of efficacy. As noted in Tables 3 and 4,the responder group exhibited significantly greater mean baseline 50-foot
Table 4. Changes in laboratory variables during therapy with nonsteroidal antiinflammatory drugs in rheumatoid arthritis patients who responded and those who did not respond to treatment*
Responders (n = 12) ESR (mm/hour) CRP (mg/dl) RF Nephelometry (IUlml) ELISA (pg/ml) Nonresponders (n = 18) ESR (mrdhour) CRP (mg/dl) RF Nephelometry (IU/ml) ELISA (pg/ml)
Mean change during therapy
Enrollment values
Baseline values
Week4
Week 8
Week 10
73.4t 3.95t
79.1$ 4.505
-3.6 -2.01
- 13.9 -1.4
-11.9 -2.4#
1,541.2 349.2
1,646.0
36.1 1.32
45.4 1.71
1,213.9 181.5
1,870.4 174.2
4.4
-23.6**
301.2 -2.9 -0.2 0.4
3.9 -0.6
1.6 -0.3
-7.8
-3.3
* Values are the mean absolute change from baseline, except for rheumatoid factor (RF), which is the mean percentage of change from baseline (determined by rate nephelometry in 6 responders and 8 nonresponders and by enzyme-linked immunosorbent assay [ELISA] in 6 responders and 9 nonresponders; 1 responder patient excluded because of incomplete data); enrollment and baseline values are the population mean. Erythrocyte sedimentation rate (ESR) was determined according to the Westergren method. Normal C-reactive protein (CRP) level 50.6 mg/dl, by rate nephelometry. t P 5 0.01 versus nonresponders. j: P = 0.04 versus nonresponders. 5 P = 0.02 versus nonresponders. B P = 0.01 versus baseline. # P = 0.02 versus baseline. ** P = 0.04 versus baseline.
NSAIDs AND SEROLOGIC MEASURES IN RA
walk times, tender joint scores, ESR, and CRP values at baseline compared with the nonresponder group (P < 0.05). Prior to the washout period, there were fewer differences between the responders and nonresponders, with only the 50-foot walk times, ESR, and CRP values observed to be significantly different (greater in the responders). Laboratory assessments. Results of the automated chemical studies of sera, by group, were unremarkable except for sporadic mild elevations in alkaline phosphatase levels without associated elevations in transaminase levels. At baseline, the responders manifested a higher mean CRP level than did the nonresponders. Moreover, significant decreases in this serologic correlate of inflammation occurred only among the responders and within 4 weeks of the beginning of NSAID therapy (Figure 1 and Table 4). The ESR did not change significantly during treatment in either group. The clinical responders, however, manifested a significant decrease in serum RF levels after 8 weeks and 10 weeks of therapy (P < 0.05), which was not observed in the nonresponders (Figure 2 and Table 4). Cell subsets. Comparisons of the 2 patient groups showed no demonstrable differences in monocyte numbers or percentages of CD4+ and CD8+ lymphocytes (Table 5). Moreover, no consistently significant changes in these cell subsets were observed during NSAID therapy, in either the responders or the nonresponders. At baseline, the responders exhibited significantly fewer circulating lymphocytes and significantly more granulocytes than the nonresponders ( P I0.05) (Table 5 ) . Only the responders demonstrated significant alterations in the total number of circulating lymphocytes a$d granulocytes during the NSAlD trial. After as little ;as 4 weeks of therapy, the absolute granulocyte count was significantly decreased (P 5 0.05), and it remained decreased through week 10 in the responders (Figure 3). And by the completion of the therapy, thf: responders also manifested a significant increase (P = 0.03) in the total lymphocyte count (Figure 4). Correlation between clinical efficacy and laboratory findings. Table 6 shows the relationship between improvement in the clinical variables and changes in serologic correlates of inflammation and in cell subset numbers. Improvement in these clinical variables was significantly correlated with reductions in CRP levels and inversely correlated with the number of circulating lymphocytes (P 5 0.05). There was also a significant correlation between the number of circulating granulocytes and global assessment of disease activity (P 5
23
't 1
Baseline
WK 4 0 NON-RESPONDER
WK8
WK10
0 RESPONDER
Figure 1. Mean absolute change from baseline C-reactive protein values for responder and nonresponder rheumatoid arthritis populations during treatment with nonsteroidal antiinflammatory drugs. Values were significant for only the responders, at weeks 4-10 (P5 0.05).
0.05). Significant correlations between laboratory results and other measures of clinical efficacy were not observed.
DISCUSSION These studies were undertaken to assess whether clinical response to NSAID treatment in patients with RA might be associated with changes in the laboratory correlates of systemic inflammation and/or immunologic activity. We used a measure of clinical response that relied on objective and subjective variables provided by patient and physician. This definition was constructed as a practical means whereby a significant antiinflammatory effect, rather than an analgesic effect, might be identified. Love11 et al (23) used a similar, multivariable definition of clinical response to study NSAID treatment in patients
24
2r
CUSH ET AL
10
-30 Baseline
WK 4 0
WK8
WK10
NON-RESPONDER 0 RESPONDER
Figure 2. Mean percentage of change from baseline serum IgM rheumatoid factor (IgM-RF) levels for responder and nonresponder rheumatoid arthritis populations during treatment with nonsteroidal antiinflammatory drugs. Values were significant for only the responders, at weeks 8 and 10 (P 0.05).
with juvenile RA. We used conventional clinical variables to assess clinical response (i.e., joint score, walk time, morning stiffness, and global assessments of disease activity). The patient’s response status was assessed independently of the laboratory data analysis. The capacity of this index to differentiate responders from nonresponders and the ability to correlate therapeutic response with laboratory evidence of diminished inflammatory activity suggest the practical utility of this method of assessing clinical outcome in therapeutic trials. Clinical and laboratory data presented in Tables 3 and 4 indicate that at the initiation of the study, the subset of responding patients exhibited more active disease than did the subset of nonresponding patients. In fact, compared with the nonresponders, the responders had significantly greater disease activity at baseline, as assessed by the %foot walk time, tender joint score, ESR, and CRP level ( P < 0.05). Although some patients with RA may improve spontaneously, the large percentage of patients demonstrating significant improvement according to the clinical criteria employed (40%) and the significant correlation between clinical efficacy variables and serologic correlates of inflammation strongly suggest that the improvements noted were neither random nor unrelated to therapy. However, in the absence of a placebotreated control group, it is impossible to conclude with certainty that the responses were related to the NS AID therapy. Since their introduction nearly 30 years ago,
Table 5. Profile of cell subsets during therapy with nonsteroidal antiinflammatory drugs in rheumatoid arthritis patients who responded and those who did not respond to treatment*
Responders (n = 12) Lymphocytes (/mm’) % CD4+ % CD8+ CD4:CD8 ratio Monocytes (/mm3) Granulocytes (/mm3) Nonresponders (n = 18) Lymphocytes (Imm’) % CD4+ % CD8+ CD4:CD8 ratio Monocytes (/mm3) Granulocytes (/mm3)
Mean change during therapy
Baseline values
Week 4
1,685.6t 37.2 22.8 1.94 466.8 6,435.2t
184.2 8.0$ 6.4 0.07 -6.9 - 1,174.0$
2,137.1 45.8 27.4 1.91 373.6 5,097.9
- 179. I
4.8 -1.1 0.3 1 $
-22.5 258.0
Week 8
Week 10
556. I $ 8.5$ 4.0 0.59 - 112.8 - 1,464.0f
496.8$ 6.0 3.8 0.30 - 195.8 -1,264.1$
-203.3 2.9 -0.3 0.21 -1.4 217.8
-76.3 1.8 -0.2 0.41 -90.1 - 17.7
* Values are the mean absolute change from baseline; baseline values are the population mean. The granulocyte value represents the total white blood cell count minus the number of lymphocytes plus monocytes. t P 5 0.05 versus nonresponders. $ P 5 0.05 versus baseline.
25
NSAIDs AND SEROLOGIC MEASURES IN RA
5L 600
300
500
0
-
400
cl
0
. 5 E E
v)
-
E E
. 300 In
-300
a,
0 v
2 200 c
c
C
3 0
3
0
2-
0 In
2 -600 0
0
0 -
c
C
E 1
Q
3
9
0 c
m
100
V x
%
O
c
-900
-100
r”
-
-200
-1200
-1 500
I
Baseline 0
I
I
I
WK 4
WK8
WK10
NON-RESPONDER 0RESPONDER
-400Baseline
WK8
WK 4
WK10
Figure 3. Mean absolute change from baseline granulocyte counts for responder and nonresponder rheumatoid arthritis populations during treatment with nonsteroidal antiinflammatory drugs. Values were significant for only the responders, at weeks 4-10 (PI0.05).
NSAIDs have served as a cornerstone of antirheumatic therapy because of their ability to reduce symptoms in a substantial number of patients (6,13,14,17). By virtue of their antiinflammatory properties, NSAIDs are capable of reducing joint swelling and duration of morning stiffness and improving many of the clinical parameters used to gauge disease activity in RA. In contrast to the DMARDs, the clinical response to an NSAID is thought to be rapid, usually within days or weeks. However, NSAIDs do not appear to correct the elevations in the levels of acutephase reactants and serum R F titers that are often seen in patients with active RA (2-6). These results have fostered the widely held opinion that NSAIDs have no capacity to modulate the chronic inflammatory response or the immune abnormalities associated with RA. We have reported here the favorable effects of NSAID therapy in a subset of RA patients. The early clinical responses (within 2 weeks) were ultimately
Correlation between changes in clinical variables and changes in immunologic variables during therapy with nonsteroidal antiinflammatory drugs in the entire rheumatoid arthritis patient population*
Table 6.
Clinical parameter Laboratory parameter CRP level
ESR RF level Lymphocyte count % cD4+ % CD8+ Monocyte count Granulocyte count
Tender joint score
Global assessment
0.44t 0.25 0.23 -0.32t -0.01 0.0s 0.09 0.21
0.421 0.29 0.19 -0.34t 0.01 -0.26 0.37 0.46t
* Values are Spearman’s rank correlation coefficient. See Table 4 for definitions and explanations of abbreviations. t Correlation significant at P 5 0.05.
26
associated with antiinflammatory and, possibly, immunomodulatory effects, as manifested by significant reductions in CRP levels, serum RF titers, and numbers of circulating granulocytes, and augmentation of the numbers of circulating lymphocytes. Our findings support the possibility that these agents may offer more than symptomatic therapy and suggest that they may have a more profound effect on RA. Discrepancies between the data presented here and those presented elsewhere may be partly explained by methodologic differences that include the presentation of average population data without consideration of individual clinical response status, the short duration ( 5 6 weeks) or low dosage used in many of the early NSAID studies, and possibly, variable effects of different pharmacologic compounds on rheumatoid inflammation (8). In contrast to earlier studies, our results were derived from a population of patients categorized according to clinical response status, who, for the most part, had not received DMARD therapy. Moreover, the use of newer and more sensitive quantitative assays for serum R F and CRP may have provided an advantage over traditional dilutional measurements, thereby permitting changes to be detected. The effect of NSAIDs on R F production has been observed by other investigators. Goodwin et a1 (13) studied 20 RA patients during a 10-week trial of piroxicam. After a 2-week drug washout period and the initiation of the NSAID, they noted clinical improvement that was associated with significant decreases in mean serum RF titers. Effects on the acute-phase reactants were not examined. Those investigators proposed that the in vivo inhibition of cyclooxygenase activity and prostaglandin E, (PGE,) production led to more effective suppression of RF synthesis. Earlier in vitro studies by the same investigators had demonstrated that PGE, inhibited the action of CD8+ suppressor cells, thereby augmenting the in vitro production of IgM-RF (7). NSAIDs prevented this contrasuppressive action of PGE,, and thereby facilitated the down-regulation of autoantibody production. Others have similarly claimed that therapy with piroxicam or fenclofenac might decrease serum RF titers (16,17). In the current studies, we noted a similar effect of NSAID administration on the serum RF titer, but only in patients who improved clinically. Whefher the reduction of RF titers in these patients reflects a direct immunomodulatory effect of the drugs or is secondary to amelioration of the inflammatory process (24) cannot be ascertained from these data.
CUSH ET AL
Numerous studies have demonstrated the inability of NSAID therapy to influence serologic correlates of inflammation, including the ESR and CRP (2-6). Elevations in the levels of acute-phase reactants are characteristic findings in patients with active RA (1-3). Although the factors contributing to an augmented acute-phase response are numerous, it has been proposed that local production of monocyte and lymphocyte-derived cytokines, such as interleukin-1 (IL-1), tumor necrosis factor (TNF), and IL-6 within the synovial tissue may be responsible for the acutephase response seen in RA (25-28). Reduction of the levels of acute-phase reactants might therefore be anticipated to correlate with reduction in disease activity since these same cytokines are thought to play a central role in the pathogenesis of joint inflammation and joint damage in RA (2,27,28). Whereas some of the DMARDs are thought to interfere with the production of monocyte-derived cytokines in vitro, it has been demonstrated that PGE, suppresses, and NSAIDs augment, IL-1 and TNF production in vitro (29,30). Moreover, it is believed that the action of these cytokines on hepatic production of acute-phase reactants is not mediated by prostaglandins. Thus, the finding that NSAIDs suppress the acute-phase response was unanticipated; it suggests that in a subset of patients, the decrease in CRP might be secondary to suppression of inflammation. The rapid decline in CRP levels, however, leaves open the possibility that NSAIDs have a direct effect on the production of, or the response to, the cytokines responsible for the induction of CRP synthesis. The finding that CRP levels decreased even in some patients who failed to respond clinically (8 of 18 patients) suggests the possibility that this decrease was an effect of the NSAIDs that were used and not a secondary effect related to the control of inflammation. In contrast with the findings by Goodwin and coworkers (13), we, like most other investigators, were unable to correlate clinical status or therapeutic response to alterations in the number of circulating CD4+ or CD8+ T cells in RA (31,32). Surprisingly, we observed decreased granulocyte numbers in patients who responded to NSAID therapy, and there was a significant correlation between the decrease in granulocyte numbers and improvement in global assessment scores. These findings are interesting because NSAIDs have been shown to alter a number of the functional activities of neutrophils (33). These data suggest that successful therapy may be associated with
NSAIDs AND SEROLOGIC MEASURES IN RA
decreases in the number of circulating neutrophils and, possibly, with inhibition of neutrophil function. Favorable clinical responses to either sustained-release ibuprofen or flurbiprofen were evident and significant as early as 2 weeks after initiation of treatment, and maximum benefit was generally achieved between weeks 8 and 10. Temporally, optimal clinical responses coincided with the late effects of therapy on the serologic correlates of inflammation, especially with the decline in RF levels. It has previously been suggested that late clinical effects of NSAID therapy occur in some patients (14.23). Our current findings support the conclusion that delayed therapeutic responses to NSAIDs can occur, and that such responses reflect a more profound antiinflammatory or immunomodulatory effect that is associated with decreases in levels of CRP and RF. The goals of therapy in patients with RA include the amelioration of symptoms and protection from the joint damage that is often observed in the disease. Status as a disease-modifying agent should be reserved for those drugs that not only improve clinical parameters and functional status, but also favorably alter serologic correlates of inflammation and RF titers, and possibly, even retard the development of articular erosions in patients with RA. In this study, therapy with NSAIDs alone resulted in improvements in clinical activity, levels of IgM-RF, and levels of CRP in a subset of patients with RA. Whether changes in these serologic indicators of inflammation will ultimately be associated with protection against damage to articular structures can neither be determined nor inferred from this short-term trial. Nonetheless, these data suggest that in a subset of RA patients, NSAIDs can provide clinical, antiinflammatory, and immunomodulatory effects hitherto ascribed only to DMARDs.
ACKNOWLEDGMENTS We acknowledge the cooperation and assistance of The Upjohn Company, in particular, Carol Daenzer, Dr. Derek Stubbs, K. W. Teoh. Rodney Beason, Rod Nunnelee, and Marv Berkowitz.
REFERENCES 1. Dawes F T , Fowler PD, Clarke S, Fisher J, Lawton A,
Shadforth MF: Rheumatoid arthritis: treatment which controls the C-reactive protein and erythrocyte sedimentation rate reduces radiological progression. Br J Rheumatol 25:4449, 1986
27
2. Wright V, Amos R: Do drugs change the course of rheumatoid arthritis? Br Med J 280:964-966, 1980 3. Dixon JS, Bird HA, Sitton NG, Pickup ME, Wright V: C-reactive protein in the serial assessment of disease activity in rheumatoid arthritis. Scand J Rheumatol 13:39-44, 1984 4. McConkey B, Crockson RA, Crockson AP, Wilkinson AR: The effects of some anti-inflammatory drugs on the acute phase proteins in rheumatoid arthritis. Q J Med 42:785-791, 1973 5. Aylward M, Maddock J, Wheeldon R, Parker RJ: A study of the influence of antirheumatic drug regimens on serum acute-phase proteins, plasma tryptophan, and erythrocyte sedimentation rate in rheumatoid arthritis. Rheumatol Rehabil 14:lOl-114, 1975 6. The Cooperating Clinics Committee of the American Rheumatism Association: A three month trial of indomethacin in rheumatoid arthritis, with special reference to analysis and inference. Clin Pharmacol Ther 8: 11-37, 1967 7 Ceuppens JL, Rodriguez MA, Goodwin JS: Nonsteroidal anti-inflammatory agents inhibit the synthesis of IgM rheumatoid factor in vitro. Lancet 1528-530, Lancet I, 1982 8. Spiers EM, Tavendale A, MacConnachie A, Swanson Beck J: The effect of non-steroidal anti-inflammatory drugs on phytohaemagglutinin stimulated lymphocytes: relevance to possible therapeutic immunosuppression by fenclonfenac. Int J Immunopharmacol 10:261-269, 1988 9. Levy J: Enhancement of deficient T-cell function in rheumatoid arthritis by tolmetin sodium. J Clin Pharmacol 23:324-328, 1983 10. Ceuppens JL, Robaeys G, Verdickt W, Vertessen S, Deckmyn H , Dequeker J: Immunomodulatory effects of treatment with naproxen in patients with rheumatic disease. Arthritis Rheum 29:305-311, 1986 11. Thoen J, Helgetveit K, Forre 0, Haile Y, Kass E: Effects of piroxicam and D-penicillamine on T lymphocyte subpopulations, natural killer cells, and rheumatoid factor production in rheumatoid arthritis. Scand J Rheumatol 17:91-102, 1988 12. Veys EM, Verbruggen G, Suykens S, Mielants H, Ackerman K, van Lerbeirghe J, Heynen G: Lymphocyte sub-populations after a single 40 mg. administration of piroxicam in 20 patients with rheumatoid arthritis: a placebo controlled study. Inflammation 8 (suppl): S1 K S 1 2 1 , 1984 13. Goodwin JS, Ceuppens JL, Rodriguez MA: Administration of nonsteroidal anti-inflammatory agents in patients with rheumatoid arthritis: effects on indexes of cellular immune status and serum rheumatoid factor levels. JAMA 250:2485-2488, 1983 14. Blechman WJ, Schmid FR, April PA, Wilson C, Brooks
28
15.
16.
17.
18.
19.
20.
21.
22.
23.
CUSH ET AL
CD: Ibuprofen or aspirin in rheumatoid arthritis therapy. JAMA 233:33&340, 1975 Waltham-Weeks CD: Etodolac versus naproxen in rheumatoid arthritis: a double-blind crossover study. Curr Med Res Opin 10540-547, 1987 Forre 0, Thoen J, Helgetveit K, Haile Y:Non-steroidal anti-inflammatory drugs in rheumatoid arthritis: effects on clinical parameters and cellular immunity. Inflammation 8 (suppI):SlWS113, 1984 Katona G: Fenclofenac: a new nonsteroidal antiinflammatory drug with an immunomodulatory effect in the treatment of active rheumatoid arthritis. Curr Ther Res 37:58-69, 1985 Ropes MW, Bennett GA, Cobb S, Jacox R, Jessar RA: 1958 revision of diagnostic criteria for rheumatoid arthritis. Bull Rheum Dis 9:175-176, 1958 Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF, Cooper NS, Healey LA, Kaplan SR, Liang MH, Luthra HS, Medsger TA Jr, Mitchell DM, Neustadt DH, Pinals RS, Schaller JG, Sharp JT, Wilder RL, Hunder GG: The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 31:315-324, 1988 Hicks MJ, Heick M, Finely P, Gall EP, Minnich L, Williams RJ: Rheumatoid factor activity by rate nephelometry correlated with clinical activity in rheumatoid arthritis. Am J Clin Pathol 78:342-345, 1982 Tomana M, Schrohenloher RE, Koopman WJ, Alarcon GS, Paul WA: Abnormal glycosylation of serum IgG from patients with chronic inflammatory diseases. Arthritis Rheum 31:333-338, 1988 Herzenberg LA, Herzenberg LA: Analysis and separation using the fluorescence-activated cell sorter, Handbook of Experimental Immunology. Edited by DM Weir. Oxford, Blackwell Scientific Publications, 1978 Lovell DJ, Giannini EH, Brewer EJ Jr: Time course of response to nonsteroidal antiinflammatory drugs in
juvenile rheumatoid arthritis. Arthritis Rheum 27: 1433-1437, 1984 24. Olsen NJ, Callahan LF, Pincus T: In vitro rheumatoid factor synthesis in patients taking second-line drugs for rheumatoid arthritis: independent associations with disease activity. Arthritis Rheum 31:1090-1096, 1988 25. Dinarello CA: An update on human interleukin-1: from molecular biology to clinical relevance. J Clin Immunol 51287-297, 1985 26. Beutler B, Cerami A: Cachectin: more than a tumor necrosis factor. N Engl J Med 316:379-385, 1987 27. Houssiau FA, Devogelaer J-P, van Damme J, Nagant de Deuxchaisnes C, van Snick J: Interleukin-6 in synovial fluid and serum of patients with rheumatoid arthritis and other inflammatory arthritides. Arthritis Rheum 31: 784-788, 1988 28. Lipsky PE, Davis LD, Cush JJ, Oppenheimer-Marks N: The role of cytokines in the pathogenesis of rheumatoid arthritis. Springer Semin Immunopathol 11: 123-162, 1989 29. Kunkel SL, Wiggins RE, Chensue SE, Lawick J: Regulation of macrophage tumor necrosis factor production by prostaglandin E,. Biochem Biophys Commun 137: 404410, 1986 30. Dinarello CA: Biology of interleukin 1 . FASEB J 2: 108-115, 1988 31. Cush JJ, Lipsky PE: Phenotypic analysis of synovial tissue and peripheral blood lymphocytes isolated from patients with rheumatoid arthritis. Arthritis Rheum 31: 123&1238, 1988 32. De Vries E, Meijer CJLM, Lafeber GJM, Cnossen J, Cats A: Lymphocyte subpopulations in rheumatoid arthritis: an immunological, enzyme histochemical and morphological study. Rheumatol Int 4:91-94, 1986 33. Perianin A, Giroud J-P, Hakim J: Differential in vivo effects of indomethacin, ibuprofen, and flurbiprofen on oxygen-dependent killing activities of neutrophils elicited by acute nonimmune inflammation in the rat. Inflammation 12:181-189, 1988
