Downloaded from
`
`http://www.jimmunol.org/
`
` by guest on October 5, 2017
`
`Regulation of Cytokines, Cytokine Inhibitors, and Acute-Phase
`Proteins Following Anti-TNF-a Therapy in Rheumatoid
`Arthritis1
`
`Peter Charles,* Michael J. Elliott,† Diana Davis,† Alison Potter,† Joachim R. Kalden,‡
`Christian Antoni,‡ Ferdinand C. Breedveld,§ Josef S. Smolen,¶ Gabriele Eberl,‡ Kim deWoody,i
`Marc Feldmann,† and Ravinder N. Maini2†
`
`Treatment with a chimeric mAb to TNF-a has been shown to suppress inflammation and improve patient well-being in rheu-
`matoid arthritis (RA), but the mechanisms of action of such treatment have not been fully explored. Here we show that in vivo
`administration of anti-TNF-a Ab, using a longitudinal analysis, results in the rapid down-regulation of a spectrum of cytokines,
`cytokine inhibitors, and acute-phase proteins. Marked diurnal variation in the serum levels of some of these were detected. These
`results were consistent with the concept of a cytokine-dependent cytokine cascade, and the degree of clinical benefit noted after
`anti-TNF-atherapy is probably due to the reduction in many proinflammatory mediators apart from TNF-a, such as IL-6, which
`reached normal levels within 24 h. Serum levels of cytokine inhibitors such as soluble p75 and p55 TNFR were reduced as was
`IL-1 receptor antagonist. Reductions in acute-phase proteins occurred after serum IL-6 fell and included serum amyloid A,
`haptoglobin, and fibrinogen. The latter reduction could be of importance, as it is a risk factor for atherosclerosis, which is
`augmented in RA patients. The Journal of Immunology, 1999, 163: 1521–1528.
`
`A wide range of cytokines and other inflammatory medi-
`
`ators are expressed in the joint in rheumatoid arthritis
`(RA)3 (1). IL-1, IL-6, GM-CSF, and the growing family
`of chemokines have all been proposed to play a role in disease
`pathogenesis (1, 2). However, the arguments that place TNF-a at
`the heart of the inflammatory process in RA are particularly com-
`pelling. TNF-a and its two receptors (p55 and p75 TNFR) are
`expressed at many sites within the synovial membrane, including
`the cartilage/pannus junction (3, 4). Tissue expression of these
`molecules is reflected in the synovial fluid, where elevated levels
`of TNF-a and soluble forms of the receptors (sTNFR) are seen
`(5–7). Using an RA synovial cell culture system in which there is
`spontaneous production of many cytokines (8), we showed that
`neutralizing TNF-a down-regulates the production of IL-1, IL-6,
`IL-8, and GM-CSF (9 –11), and these findings led us to propose
`TNF-a as a therapeutic target in RA.
`Direct demonstrations of the importance of TNF-a in inflam-
`matory arthritis in vivo was first provided by animal studies. In
`experiments aiming to block the endogenous TNF production in
`arthritis, we and others showed that systemic administration of
`
`*Department of Rheumatology, Charing Cross Hospital, London, United Kingdom;
`†The Kennedy Institute of Rheumatology, London, United Kingdom; ‡Institute of
`Clinical Immunology and Rheumatology, Erlangen, Germany; §Department of Rheu-
`matology, University Hospital, Leiden, The Netherlands; ¶University Klinik fur In-
`nere Medizin III, Vienna, Austria; and iCentocor, Malvern, PA 19335
`Received for publication October 30, 1998. Accepted for publication May 11, 1999.
`
`The costs of publication of this article were defrayed in part by the payment of page
`charges. This article must therefore be hereby marked advertisement in accordance
`with 18 U.S.C. Section 1734 solely to indicate this fact.
`1 This work was supported in part by a core grant from the Arthritis Research Cam-
`paign of Great Britain. The clinical trial was supported by Centocor.
`2 Address correspondence and reprint requests to Dr. Ravinder N. Maini at his current
`address: The Kennedy Institute of Rheumatology, 1 Aspenlea Road, Hammersmith,
`London W6 8LH, U.K. E-mail address: r.maini@ic.ac.uk
`3 Abbreviations used in this paper: RA, rheumatoid arthritis; sTNFR, soluble TNFR;
`IL-1ra, IL-1 receptor antagonist; CRP, C-reactive protein; SAA, serum amyloid A.
`
`blocking Abs to TNF or of sTNFR:Fc fusion proteins after disease
`onset led to amelioration of joint disease in murine collagen-induced
`arthritis (12–14). Transgenic mice carrying a 39-modified human
`TNF-atransgene showed dysregulated TNF-aexpression and the de-
`velopment of a chronic inflammatory polyarthritis, preventable with
`monoclonal anti-human-TNF-a(15). Evidence that TNF-aexerted its
`pathogenic effect via IL-1 in this model was provided by the thera-
`peutic benefit observed with an anti-IL-1 receptor Ab (16), reflecting
`the hierarchy we observed in RA cultures.
`These findings led to the conclusion that TNF-a was a prime
`therapeutic target in RA (17, 18) and provided the rationale for
`clinical trials of a specific, TNF-blocking chimeric (human/mouse)
`mAb (cA2) in patients with RA (19 –21). In the most definitive of
`these studies, cA2 was compared with placebo in a multicenter,
`randomized, double-blind trial. The outcome was unequivocal,
`with the induction of large and highly significant improvements in
`a wide range of clinical and laboratory measures of disease activity
`in active treatment groups (20).
`There are not many opportunities to investigate the pathogenesis
`of human disease in vivo. The clear outcomes of cA2 therapy
`provides us with an opportunity to probe deeper into the role of
`TNF-a in the pathogenesis of RA. We have previously described
`the effect of cA2 administration on leukocyte migration (22), doc-
`umenting a reduction in the expression of adhesion molecules and
`relating these changes to circulating leukocyte numbers. Here, we
`explore the effects of cA2 on circulating cytokines and cytokine
`inhibitors, and test the relationship between these mediators and
`their major recognized systemic effect in RA, the production of
`hepatic acute-phase proteins. The results confirm the central role of
`TNF-a in the pathogenesis of RA.
`
`Materials and Methods
`Trial procedures
`
`The selection of patients for this study and their treatment have been
`described in detail elsewhere (20). In brief, 73 patients meeting the revised
`
`Copyright © 1999 by The American Association of Immunologists
`
`0022-1767/99/$02.00
`
`

`

`Downloaded from
`
`http://www.jimmunol.org/
`
` by guest on October 5, 2017
`
`1522
`
`CYTOKINE REGULATION BY ANTI-TNF THERAPY
`
`American College of Rheumatology (ACR) criteria for the diagnosis of RA
`(23) were recruited from the clinics of four cooperating trial centers. All
`patients had active RA and evidence of erosive disease on x-rays of hands
`or feet. Patients taking disease modifying antirheumatic drugs were with-
`drawn from their therapy at least 4 wk before study entry, but were per-
`mitted to continue taking low dose oral corticosteroids or nonsteroidal
`antiinflammatory drugs at stable dosage.
`cA2 (Remicade, Infliximab; Centocor, Malvern, PA) is a human/murine
`chimeric mAb of IgG1k isotype, with specificity for recombinant and nat-
`ural human TNF-a(24). At entry to the study, patients were randomized to
`receive a single 2 h infusion of either placebo (0.1% human serum albumin,
`24 patients), low-dose cA2 (1 mg/kg, 25 patients), or high-dose cA2 (10
`mg/kg, 24 patients) as an outpatient procedure. Patients were then followed
`using clinical and laboratory parameters for a period of 4 wk. Patients,
`investigators, and laboratory personnel were blinded as to the treatment
`administered.
`Blood samples for laboratory measurements were drawn before the in-
`fusion on day 0 and at the following times after completion of the infusion:
`1 and 8 h, 1 and 3 days, 1, 2, 3, and 4 wk. This resulted in the following
`mean collection times: 0830, 1300, 2000 (on day 0); 1200, 1100 (on days
`1 and 3); 1030, 1100, 1045, 1045 (on weeks 1– 4). Blood was collected into
`sterile tubes, allowed to clot for 30 min, and spun at room temperature for
`20 min at 2500 rpm. Serum was aliquoted into plastic tubes and stored at
`270°C until assayed. Plasma was prepared from EDTA blood and handled
`similarly.
`
`Laboratory measurements
`
`Laboratory measurements were made using commercially available assays
`where available according to the manufacturers’ directions. All samples
`from a given patient were assayed together to reduce interassay variability.
`Cytokines and inhibitors. TNF-a, IL-6, and IL-10 were measured using
`enzyme-amplified sensitivity immunoassays, each based on an oligoclonal
`detection system (Biosource, Fleurus, Belgium). The sensitivity of the as-
`says was 10 pg/ml (TNF-a and IL-6) and 4 pg/ml (IL-10). IL-1 receptor
`antagonist (IL-1ra) was measured using a quantitative sandwich enzyme
`immunoassay (R&D Systems Europe, Abingdon, U.K.). The sensitivity of
`the assay was 94 pg/ml.
`IL-1b is difficult to measure in serum, but we have previously investi-
`gated a number of different assays for its measurement. IL-1b was mea-
`sured using a solid phase ELISA. Monoclonal anti-IL-1b 58.121.08 (Bio-
`source) was coated onto a microtiter plate overnight at 4°C. Free binding
`sites were blocked using 3% BSA in PBS (0.15 M, pH 7.2). Excess block-
`ing solution was washed away using 0.1% Tween in 0.8% saline. then, 200
`ml of samples or standards were added in duplicate to the wells together
`with 50 ml of biotin conjugated mAb 58.121.03 (Biosource). The plates
`were then incubated fo r 2 h atroom temperature. Unbound material was
`washed away using 0.1% Tween in 0.8% saline. Avidin-peroxidase
`(Boehringer Mannheim, Lewes, U.K.) was added to the wells and incu-
`bated at room temperature for 30 min. Unbound conjugate was then
`washed away. The reaction sites were amplified using Biotynal tyramide
`(DuPont, Stevenage, U.K.) at room temperature for 15 min, and then un-
`bound material was washed away. Avidin-peroxidase (Boehringer Mann-
`heim) was added to the wells and incubated at room temperature for 30
`min. Unbound conjugate was then washed away, tetramethylbenzidine
`(Sigma, Poole, U.K.) was added to visualise the reaction, and the resulting
`color following addition of 50 ml 0.1 M sulfuric acid to each well was read
`at 450 nm. The OD obtained were compared with those obtained from a
`standard curve made from dilutions of rIL-1b (Biosource), and the con-
`centration of IL-1b in each sample calculated. The sensitivity of this assay
`was found to be 0.5 pg/ml. Analysis was restricted to the placebo and 10
`mg/kg cA2-treated groups to conserve on samples and reagents.
`Soluble p55 and p75 TNFR were measured using an in house immu-
`noassay, according to the method previously described (7, 25). Briefly,
`mAb to either p55 or p75 receptor were coated onto microtiter plates and
`incubated overnight at 4°C. Free binding sites were blocked with 3% BSA/
`PBS. Diluted standard or serum was added to each well and incubated for
`2 h at room temperature. Unbound material was then washed away. Bound
`receptor was detected using complementary mAbs conjugated to biotin.
`After washing, incubation with streptavidin-peroxidase, further washing,
`and incubation with tetramethylbenzidine, the reaction was stopped with
`sulfuric acid. OD values obtained at 450 nm were compared with those
`obtained from a standard curve constructed using recombinant p55 and p75
`TNFR. The standard curve covered a range of values from 200 pg/ml to
`12.5 ng/ml. The sensitivity of each assay was 50 pg/ml. The normal values
`obtained from a panel of blood donors was ,1000 pg/ml for p55 TNFR
`and ,2000 pg/ml for p75 TNFR.
`
`FIGURE 1. Effect of cA2 on circulating immunoreactive TNF-a. Pa-
`tients were treated on day 0 with a single, 2-h infusion of either placebo
`(E), 1 mg/kg cA2 ((cid:140)), or 10 mg/kg cA2 (f). Each point represents the
`median circulating immunoreactive TNF-a level in up to 24 patients, with
`interquartile ranges omitted for clarity. pp, p , 0.01; ppp, p , 0.001
`compared with placebo, by ANOVA.
`
`Acute-phase proteins. C-reactive protein (CRP) was measured by fluo-
`rescent polarization immunoassay using the TDX system (Abbot Diagnos-
`tics, Maidenhead, U.K.). The system works by comparing the polarization
`value obtained for a given sample to a precalibrated standard value. Each
`assay was validated by the inclusion of control sera containing known
`quantities of CRP. Serum amyloid A (SAA) was measured by a solid phase
`ELISA (Biosource).
`Haptoglobin and fibrinogen were measured using radial immunodiffu-
`sion (Behring, Hounslow, U.K.). EDTA plasma was placed into a well cut
`into a gel containing Abs to haptoglobin or to fibrinogen. After 48 h (hap-
`toglobin) or 18 h (fibrinogen), the diameter of the resulting precipitin rings
`was measured and the concentration compared with a predetermined con-
`centration table. The assays were validated by the inclusion of control sera
`of known concentration. These assays were confined to high-dose (10
`mg/kg cA2) and placebo-treated groups to conserve samples and reagents.
`CRP results for all treated groups have been reported previously (20).
`
`Statistics
`
`Data are expressed as median, interquartile range. Diurnal variation in
`sTNFR, IL-1ra, and IL-6 was assessed in the placebo group using the
`Wilcoxon signed rank test. ANOVA on the van der Waerden normal scores
`was used to compare baseline values of TNF-a, IL-6, IL-1b, sTNFR, IL-
`1ra, CRP, and SAA as well as for comparison of changes from baseline at
`each posttreatment point. The model included terms for both investiga-
`tional site and treatment group. Significant differences were further tested
`by Dunnett’s comparison to the placebo group. The Mann-Whitney U test
`was used to compare the IL-10, haptoglobin, and fibrinogen data for the
`placebo and high-dose cA2 groups. Comparison between the percent re-
`ductions in IL-6, CRP, and SAA in the high-dose cA2 group was made
`using the Kruskal-Wallis test. Associations between parameters were de-
`fined using Spearman’s rank correlation coefficient (r). No adjustment was
`made for multiplicity of time points or laboratory parameters. Analyses
`were performed on a VAX computer using SAS (SAS Software, NC) and
`on a PowerMacintosh computer using Minitab (Minitab, PA).
`
`Results
`Effect of cA2 on TNF-a and sTNFR
`Our first aim was to examine the effect of cA2 on the systemic
`homeostasis of TNF-a. We could detect circulating immunoreac-
`tive TNF-a in baseline serum samples in 39 of 72 patients tested
`(54%). Median, interquartile range baseline circulating TNF-alev-
`els in the placebo group were 15.5, 5–23 pg/ml (n 5 24), for the
`1 mg/kg-treated group were 5, 5–19 pg/ml (n 5 24), and for 10
`mg/kg-treated group were 12, 5–28 pg/ml (n 5 24). The normal
`range is 10 pg/ml. Patients treated with placebo showed no
`change in circulating immunoreactive TNF-a over the 28-day
`
`

`

`Downloaded from
`
`http://www.jimmunol.org/
`
` by guest on October 5, 2017
`
`The Journal of Immunology
`
`1523
`
`FIGURE 2. Effect of cA2 on circulating p55 sTNFR. Patients were
`treated on day 0 with a single, 2-h infusion of either placebo (E), 1 mg/kg
`cA2 ((cid:140)), or 10 mg/kg cA2 (f). Each point represents the mean circulating
`p55 sTNFR with ranges omitted for clarity. p, p , 0.05 compared with
`placebo by ANOVA.
`
`FIGURE 3. Effect of cA2 on circulating p75 sTNFR. Patients were
`treated on day 0 with a single, 2-h infusion of either placebo (E), 1 mg/kg
`cA2 ((cid:140)), or 10 mg/kg cA2 (f). Each point represents the mean circulating
`p75 sTNFR with ranges omitted for clarity. There were no significant dif-
`ferences between cA2- and placebo-treated groups at any point by
`ANOVA.
`
`course of this study (Fig. 1). In contrast, patients treated with cA2
`showed a rapid and highly significant increase in median circulat-
`ing TNF-a, with evidence of a clear dose response relationship
`(Fig. 1). In patients treated with low-dose cA2, median circulating
`TNF-a levels peaked at day 3 and thereafter showed a gradual
`decline, with a return to baseline values by day 21. Peak median
`values in the high-dose cA2 group were seen at day 7 and showed
`a more gradual decline, with continuing significant elevations rel-
`ative to baseline at day 28. To determine whether the circulating
`TNF-a was biologically active, we tested samples from several
`patients with high circulating immunoreactive TNF-ain the WEHI
`164 TNF-a bioassay. No patient showed biologically active
`TNF-aeither before or after treatment with cA2 (data not shown).
`Normal sera contains sTNFR, both p55 and p75, and these are
`elevated in RA patients. It was of interest to evaluate the effect of
`therapy with anti-TNF-a on levels of these TNF inhibitors. p55
`sTNFR was detectable in baseline serum samples from all 73
`patients, with 67 (92%) showing values above the normal range
`(,1000 pg/ml). Median,
`interquartile range circulating p55
`sTNFR levels at baseline for the placebo, low-dose, and high-dose
`cA2 groups were 2050, 1360 –3683 pg/ml (n 5 24), 2050, 1710 –
`3130 pg/ml (n 5 25), and 1910, 1345–2670 pg/ml (n 5 24), re-
`spectively. The changes in circulating p55 sTNFR are shown in
`Fig. 2. p55 levels fell in the placebo group during the course of day
`0, with trough values reported at 1300 h ( p 5 0.038 compared
`with baseline). By 1200 h on day 1, p55 levels in the placebo group
`had returned to baseline values. Median p55 levels in the two
`cA2-treated groups showed a similar rapid decline over the morn-
`ing of day 0 but in contrast to the placebo group, remained below
`baseline values on day 1. Overall, patients treated with placebo
`showed no significant change over this period, while patients
`treated with low- or high-dose cA2 showed modest reductions in
`circulating p55, which were maximal by day 3 and showed sig-
`nificance ( p , 0.05) compared with placebo at day 14 and 28.
`However, median values remained above the upper limit of normal
`in all treatment groups.
`p75 sTNFR was also detectable in all 73 patients at baseline,
`with 54 (74%) showing baseline values above the normal range
`(,2000 pg/ml). Median,
`interquartile range circulating p75
`sTNFR at baseline for the placebo, low-, and high-dose cA2
`groups were 3313, 1988 –5075 pg/ml (n 5 24), 2560, 1770 –3315
`pg/ml (n 5 25), and 2812, 2049 –3778 pg/ml (n 5 24), respec-
`tively. Although median values were higher for the placebo group
`than for the two cA2 groups, this difference was not statistically
`
`significant. Changes in p75 sTNFR during day 0 and day 1 were
`similar to those seen for p55, with reduction in all three treatment
`groups at 1300 h and 2000 h on day 0, recovery to near baseline
`values in the placebo group by 1200 h on day 1, but continuing
`suppression of values for the two cA2 groups at this time. The
`median p75 sTNFR values are similar over the 28 days of the study
`(Fig. 3) in the placebo group, but show a reduction in both cA2
`treatment groups, although the changes failed to reach statistical
`significance compared with placebo.
`
`Effect of cA2 on circulating IL-1 and IL-1ra
`While IL-1 is clearly produced in the joints in RA, there is not
`much present in the serum (26). To establish the normal range for
`serum IL-1b using our assay, we tested serum samples obtained
`from 36 normal individuals. IL-1bwas detectable (.0.5 pg/ml) in
`11 of 36 samples (31%), with values ranging from 1 to 128 pg/ml.
`The median value and interquartile range were ,0.5, ,0.5– 48.0
`pg/ml.
`Using the same assay, we measured IL-1b in serum samples
`from placebo-treated patients (n 5 24) and patients treated with
`high-dose cA2 (n 5 22) only. IL-1b was detectable in 33 of 46
`patients at baseline (72%) with median, interquartile range values
`of 5.8, ,0.5–23.1 pg/ml and 8.4, ,0.5–28.1 pg/ml in the placebo
`and high-dose cA2 groups, respectively. Changes in serum IL-1b
`levels in the two RA treatment groups during the course of the
`study are shown in Fig. 4. The median values throughout the study
`remained well within the range established earlier for normal in-
`dividuals. However, there was no statistically significant difference
`between the two groups, and no clear cut reduction over the 28
`days analyzed.
`Circulating IL-1ra was detectable in all 68 pretreatment sera
`tested, with most showing values above the manufacturers quoted
`normal range (,375 pg/ml). Median, interquartile range pretreat-
`ment values were not significantly higher in the placebo group
`(613, 485– 855 pg/ml; n 5 24) than in the two active treatment
`groups (525, 235–760 pg/ml, n 5 23; 470, 275– 620 pg/ml, n 5
`21; low- and high-dose cA2, respectively).
`The changes in circulating IL-1ra occurring during day 0 and
`day 1 are shown in Fig. 5. IL-1ra levels fell in the placebo group
`during the course of day 0, with trough values recorded at 2000 h,
`( p 5 0.026 compared with pretreatment). By 1200 h on day 1,
`IL-1ra levels had recovered in the placebo group, with median
`values similar to those observed pretreatment. The early changes
`
`

`

`Downloaded from
`
`http://www.jimmunol.org/
`
` by guest on October 5, 2017
`
`1524
`
`CYTOKINE REGULATION BY ANTI-TNF THERAPY
`
`tested had elevated pretreatment IL-1ra levels, and, following
`treatment, 10 achieved normal values.
`
`Effect of cA2 on IL-6
`IL-6 was detectable in all but 4 of the 72 pretreatment sera tested.
`The median, interquartile range pretreatment circulating IL-6 lev-
`els for the three treatment groups were 125, 56 –209 pg/ml, n 5 24;
`130, 57–225 pg/ml, n 5 24; 114, 78 –188 pg/ml, n 5 24 (normal
`range, ,10 pg/ml) (placebo, low-, and high-dose cA2, respec-
`tively, p , 0.05 for each group v/s normal).
`The changes in circulating IL-6 following treatment are shown
`in Fig. 6. IL-6 levels showed significant reductions in the placebo
`group at the 1300 and 2000 h time points on day 0 ( p , 0.001, p 5
`0.002, respectively), with partial recovery by day 1 (Fig. 6). Pa-
`tients treated with cA2 showed even more marked reductions in
`circulating IL-6 at 1300 h on day 0 and continuing decline there-
`after, reaching significance compared with placebo by day 1 ( p ,
`0.01, p , 0.001, low- and high-dose cA2, respectively). In Fig. 6,
`changes in serum IL-6 over the longer term are displayed. The
`highly significant falls in serum IL-6 seen at day 1 were main-
`tained for the duration of the study in patients receiving high-dose
`cA2, but there was a partial loss of effect in patients treated with
`low-dose cA2 by week 4.
`The changes in median IL-6 values were reflected in the indi-
`vidual patient responses for IL-6. Of the 24 low-dose cA2 patients
`tested, 22 had elevated IL-6 values pretreatment and 14 of the 22
`(64%) had values within normal limits by day 1. Similarly, 23 of
`24 high-dose cA2 patients had elevated circulating IL-6 pretreat-
`ment, of whom 16 (70%) had reverted to normal values by day 1.
`
`Effect of cA2 on IL-10
`While IL-10 is abundant in synovium, the levels reported in serum
`are lower (27). IL-10 was measured in patients receiving placebo,
`(n 5 18) and in those receiving high-dose cA2 (n 5 18) only. Of
`these, nine patients (24%) had detectable IL-10 at baseline (.4
`pg/ml). Median, interquartile range circulating IL-10 levels at
`baseline were 2, 2–10.5 pg/ml and 2, 2–3.5 pg/ml in the placebo
`and high-dose cA2 groups, respectively (normal range, ,8.8 pg/
`ml). There was no significant change in values in either treatment
`group during the course of the study (Table I).
`
`Effect of cA2 on acute-phase proteins
`The changes in circulating CRP values in this study have been
`reported previously (20). In brief, patients treated with placebo
`
`FIGURE 4. Effect of cA2 on circulating IL-1b. Patients were treated on
`day 0 with a single, 2-h infusion of either placebo (E) or 10 mg/kg cA2
`(f). Each point represents the median circulating immunoreactive IL-1b,
`with interquartile ranges omitted for clarity. There were no significant dif-
`ferences between cA2- and placebo-treated groups at any point by
`ANOVA.
`
`seen in the two cA2 groups were similar in kinetics, but greater in
`magnitude than in the placebo group and failed to show recovery
`on day 1. The reduction from pretreatment values in the high-dose
`cA2 group was statistically significant compared with the reduc-
`tion in placebo patients at this time ( p , 0.01).
`Changes in circulating IL-1ra in the same treatment groups over
`a 4-wk period are shown in Fig. 5. Overall, patients treated with
`placebo showed no significant change over this period, while pa-
`tients treated with low- or high-dose cA2 showed rapid and sub-
`stantial reductions in circulating IL-1ra, which were maximal by
`day 3 and highly significant compared with placebo ( p , 0.001).
`The reduction in IL-1ra remained significant until day 14 in both
`groups. Although the maximal fall seen with the treatment dose
`groups was similar, patients treated with low-dose cA2 showed a
`more rapid loss of effect, with a return to pretreatment values by
`week 4.
`The changes in median IL-1ra values were reflected in the in-
`dividual patient responses for IL-1ra. Of the 23 low-dose cA2 pa-
`tients tested, 16 had elevated circulating IL-1ra levels before treat-
`ment, and, following treatment, 7 of these achieved a serum value
`the normal range. Similarly, 14 of the 21 high-dose cA2 patients
`
`FIGURE 5. Effect of cA2 on circulating IL-1ra.
`Patients were treated on day 0 with a single, 2-h
`infusion of either placebo (E), 1 mg/kg cA2 ((cid:140)), or
`10 mg/kg cA2 (f). A detailed time/response profile
`on day 0 and 1 is shown in A, with the mean sam-
`pling times indicated on the figure. Changes in cir-
`culating IL-1ra in the same three patient groups
`over the longer term are shown in B. Each point
`represents the median circulating immunoreactive
`TNF-a level in up to 24 patients, with interquartile
`ranges omitted for clarity. p, p , 0.05; pp, p ,
`0.01; ppp, p , 0.001 compared with placebo, by
`ANOVA.
`
`

`

`Downloaded from
`
`http://www.jimmunol.org/
`
` by guest on October 5, 2017
`
`The Journal of Immunology
`
`1525
`
`FIGURE 6. Effect of cA2 on circulating IL-6.
`Patients were treated on day 0 with a single, 2-h
`infusion of either placebo (E), 1 mg/kg cA2 ((cid:140)),
`or 10 mg/kg cA2 (f). A detailed time/response
`profile on day 0 and 1 is shown on the left, with
`the mean sampling times indicated on the figure.
`Changes in circulating IL-6 in the same three pa-
`tient groups over the longer term are shown on the
`right. Each point represents the median circulat-
`ing IL-6 level in up to 24 patients, with interquar-
`tile ranges omitted for clarity. p, p , 0.05; pp,
`p , 0.01; ppp, p , 0.001 compared with placebo,
`by ANOVA.
`
`showed no significant change in CRP values, while those treated
`with high-dose cA2 showed a large and highly significant reduc-
`tion, evident as early as day 1 posttreatment and reaching maximal
`extent by day 7 (Table II).
`Pretreatment circulating SAA were similar in both placebo and
`high-dose treatment groups (335, 62–750, mg/L, n 5 24; 378,
`180 –935 mg/L, n 5 24; placebo and high-dose cA2, respectively;
`normal range, ,10 mg/L). No significant changes in SAA levels
`were seen following treatment in the placebo group (Table I),
`while those patients treated with high-dose cA2 showed a large
`and highly significant reduction.
`Pretreatment haptoglobin and fibrinogen levels were similar in
`the placebo and high-dose cA2 treatment groups (Table II). Pla-
`cebo-treated patients showed no significant change in either mea-
`sure, while the high-dose cA2 group showed a significant reduc-
`tion in both measures by week 4 ( p , 0.001, p , 0.002,
`haptoglobin and fibrinogen, respectively).
`
`Association between circulating IL-6 and acute-phase proteins
`Because IL-6 is the principal regulator of hepatic acute-phase pro-
`tein synthesis in vitro (28, 29) and in vivo in IL-6 gene-targeted
`mice (30), we tested the association between IL-6, CRP, and SAA
`in our patients. A comparison of the kinetics of change in each of
`these mediators following treatment with high-dose cA2 showed
`falls in IL-6 preceding those of CRP or SAA. By day 1, median
`circulating IL-6 values had fallen compared with pretreatment val-
`ues by 95% and exceed the reductions in CRP (20%) and SAA
`(5%) ( p , 0.001). Although the median values for the two acute-
`phase proteins had fallen further by day 3, thereby narrowing the
`gap with IL-6, the reduction in IL-6 at this time was still signifi-
`cantly greater than for the acute-phase proteins ( p , 0.001). A
`
`Table I.
`
`IL-10a
`
`Day
`
`0
`7
`14
`28
`
`Placebo
`
`2,2–10
`2,2–15
`2,2–9
`2,2–8
`
`10 mg/kg cA2
`
`p Value
`
`2,2–8
`2,2–3
`2,2–3
`2,2–3
`
`NS
`NS
`NS
`NS
`
`a The data are expressed as median, interquartile range. There were no significant
`differences between the two treatment groups for any measure prior to or following
`treatment. Values of p assess significance of the change from pretreatment values in
`the cA2 group compared with change in the placebo group by ANOVA or Mann-
`Whitney U test.
`
`scatter graph comparing pretreatment IL-6 and CRP levels in all 73
`patients is shown in Fig. 7A, indicating a moderate association
`between these variables (r 5 0.55, p , 0.002). A similar associ-
`ation was found when comparing the reduction in circulating IL-6
`by day 3 with the reduction in CRP over the same time period (Fig.
`7B; r 5 0.59, p , 0.002). Less impressive, but still statistically
`significant, associations were seen between circulating IL-6 and
`SAA (pretreatment comparison: r5 0.44, p , 0.002; reduction by
`day 3 comparison: r 5 0.48, p , 0.002). The strongest associa-
`tions observed were between CRP and SAA (pretreatment com-
`parison: r5 0.73, p , 0.002; reduction by day 3 comparison: r5
`0.76, p , 0.002).
`
`Discussion
`The analysis of cytokine expression and regulation in rheumatoid
`arthritis has led to the definition of new therapeutic targets for this
`
`Table II. Acute-phase proteinsa
`
`Day
`
`Placebo
`
`10 mg/kg cA2
`
`p Value
`
`CRP (mg/L)
`0
`1
`3
`7
`14
`28
`SAA (mg/ml)
`0
`1
`3
`7
`14
`28
`Haptoglobin (g/L)
`0
`28
`Fibrinogen (g/L)
`0
`28
`
`n 5 24
`56, 33–70
`52, 27–63
`47, 31–74
`56, 31–72
`49, 29–72
`60, 29–71
`n 5 24
`335, 62–750
`420, 127–755
`441, 115–890
`378, 136–810
`493, 228–853
`720, 176–887
`n 5 20
`3.4, 3.0–4.3
`3.4, 3.0–4.3
`n 5 20
`3.8, 3.4–5.0
`4.0, 3.3–5.4
`
`n 5 24
`65, 28–94
`55, 22–71
`22, 13–31
`18, 13–35
`19, 10–44
`24, 13–50
`n 5 24
`378, 180–935
`380, 121–908
`50, 19–163
`62, 26–206
`56, 26–272
`69, 22–345
`n 5 15
`3.3, 2.7–3.7
`2.5, 1.5–3.2
`n 5 14
`4.1, 3.2–4.7
`2.8, 2.2–3.1
`
`0.05
`0.001
`0.001
`0.01
`0.001
`
`NS
`0.01
`0.01
`0.01
`0.01
`
`0.001
`
`0.002
`
`a The data are expressed as median, interquartile range. There were no significant
`differences between the two treatment groups for any measure prior to treatment.
`Values of p assess significance of the change from pretreatment values in the cA2
`group compared with change in the placebo group by ANOVA (CRP, SAA) or Mann-
`Whitney U test (haptoglobin, fibrinogen). Normal ranges: CRP, ,10 mg/L; SAA,
`,10 mg/L; haptoglobin, 0.7 –3.8 g/L; fibrinogen, 1.8 –3.5 g/L.
`
`

`

`Downloaded from
`
`http://www.jimmunol.org/
`
` by guest on October 5, 2017
`
`1526
`
`CYTOKINE REGULATION BY ANTI-TNF THERAPY
`
`ment difficult and the assays that can be used in synovium such
`as immunohistochemistry or PCR are not very quantitative.
`Here, we report an analysis of the effects of cA2 on cytokines
`within the circulation and relate changes in their levels to a key
`cytokine-dependant biological effect, the production of acute-
`phase proteins (31).
`Our first aim in undertaking this study was to analyze changes
`in systemic TNF homeostasis following infusion of cA2. Approx-
`imately half of the patients had detectable circulating TNF-a at
`baseline, but levels were generally low. The changes that occurred
`following infusion of cA2 were remarkable, with a rapid and dose-
`dependent increase in immunoreactive, but not biologically active,
`TNF-a, evident as early as 8 –24 h and peaking by day 7. While the
`mechanism of this increase in immunoreactive TNF-ais still under
`investigation, preliminary evidence suggest that the TNF-a is
`present in the form of a high m.w. complex, presumably with cA2
`(our unpublished observations). Similar increases in circulating
`immunoreactive IL-6 were noted following treatment with a mAb
`to IL-6 in patients with RA (32) and plasma cell leukaemia (33).
`Therefore, the rapid, but transiently delayed (not present at 4 h)
`rise in immunoreactive TNF-ain our patients is likely to represent
`trapping of the TNF-a overproduced in the disease.
`Changes in sTNFR following cA2 administration were in the
`opposite direction of those seen for TNF-a itself. Both p55 and
`p75 sTNFR were rapidly diminished following infusion of the
`mAb, reaching statistical significance compared with placebo for
`p55 sTNFR at several time points. Interestingly, although levels of
`sTNFR were markedly reduced by 1300 h on day 0 in both low-
`and high-dose cA2 groups, a similar reduction was observed for
`the placebo group throughout day 0. These findings strongly sug-
`gest a diurnal variation for sTNFR, although the changes might
`also have resulted from some unknown factor inherent in the in-
`fusion procedure with either cA2 or placebo, and further studies
`are warranted. The observed reductions in sTNFR following a