Mutations of bone morphogenetic protein receptor type II are not found in patients with pulmonary hypertension and underlying connective tissue diseases.код для вставкиСкачать
ARTHRITIS & RHEUMATISM Vol. 46, No. 10, October 2002, pp 2817–2830 © 2002, American College of Rheumatology LETTERS damage, it is not surprising that some patients can fail to have ACR20 improvement yet experience protection against radiographic damage. The Disease Activity Score (DAS) is an index that combines tender and swollen joint counts, with a minor contribution from 1 or 2 other measures (5). Because the swollen joint count predominantly determines the DAS, and because the swollen joint count is modestly correlated with radiographic damage, the DAS is sometimes better correlated with damage than is the ACR20. The analytic method used by Boers to test this is an area under the curve approach, which improves DAS precision. Because there is no similar approach for the ACR20, it is more likely that DAS will correlate better than the ACR20 with radiographic damage. Unfortunately, Boers did not present correlations between DAS and radiographic progression for the Combinatietherapie Bij Reumatoide Artritis (COBRA) trial. A prognostic measure for joint damage could be useful. Indeed, because the swollen joint count (even though its correlation with radiographs is poor) is the best method known, it might be preferable to use that instead of the DAS. The swollen joint count would be simpler to use compared with the DAS, which is a complex combination of variables and has several versions. Although improvement that does not meet ACR20 is possible in a clinical trial, there is a question of how clinically meaningful such improvement might be. One of the questions addressed by ATTRACT was whether the combination of infliximab and methotrexate inhibited radiographic progression of damage in patients who failed to have a meaningful clinical response. For this purpose, the ACR20 is the most appropriate means for classifying responders. The issue of whether patients who failed to achieve ACR20 were nonresponders or suboptimal responders is more semantic than clinically relevant. An intriguing question is whether a treatment would be chosen if it had no measurable effect on symptoms and signs of disease but slowed radiographic progression. Several subanalyses using the ATTRACT data were performed to validate the conclusion that inhibition of radiographic progression occurred in infliximab- and methotrexatetreated patients regardless of whether they had a meaningful clinical response. In this regard, infliximab- and methotrexatetreated patients were divided into groups according to those who did and those who did not experience a 20% improvement in the number of swollen joints, the number of tender joints, or the C-reactive protein level. In each of these groups, similar and significant inhibition of radiographic progression, compared with that in patients treated with methotrexate alone, was noted. Finally, patients treated with the combination of infliximab and methotrexate were divided into groups according to those who failed to improve by ACR20 criteria at the end of 54 weeks and those who failed to meet ACR20 criteria at 13 or more of the 14 evaluations during the 54 weeks of the ATTRACT trial. Both groups of “nonresponders” had significant and similar inhibition of radiographic progression compared with patients receiving methotrexate alone. All of these results are consistent with the conclusion that there is an important dichotomy between meaningful DOI 10.1002/art.10446 Blocking tumor necrosis factor inhibits radiographic damage even in patients who show minimal or no clinical improvement: comment on the concise communication by Boers To the Editor: We would like to comment on several aspects of the recent concise communication by Boers (1) regarding results of the Anti–Tumor Necrosis Factor Trial in Rheumatoid Arthritis with Concomitant Therapy (ATTRACT). In this trial, inhibition of radiographic progression was noted in patients treated with the combination of infliximab and methotrexate regardless of whether they met American College of Rheumatology (ACR) 20% criteria for improvement in signs and symptoms (2). Boers describes this result as being “counter to intuition” and believes that the interpretation resulted from inappropriate application of the ACR criteria. To support his view, he provides data from a different, unrelated trial. We would like to correct the possible misperceptions created by these claims. The first point relates to the use of the ACR criteria to classify patients as responders. Indeed, the ACR preliminary definition of improvement (the so-called ACR20) was created and validated to identify patients with rheumatoid arthritis (RA) enrolled in clinical trials who experienced a response in symptoms and signs that approximated a clinician’s impression of patient improvement (3). The ACR criteria comprised a core set of measures evaluating different aspects of disease activity, including joint counts, disability, acute-phase response, and measures of patient pain and global assessments. Elements of the ACR20 were selected, in part, because results of previous studies indicated that they predicted long-term outcomes in RA, including radiographic joint damage (4). In fact, the strongest such predictor, the swollen joint count, was included in the core set (and highlighted in the ACR improvement criteria) for just this reason. Other measures in the core set, such as the tender joint count and patient’s assessment of pain (4), did not correlate with radiographic change, but were included in the core set and definition of improvement, because they were thought to be important indicators of disease activity and in previous trials (3) showed evidence of responsiveness to change. The ACR20 (and ACR50) are now extensively used in RA trials and have rationalized such trials, allowing us to evaluate treatment efficacy in a manner that is more standardized and efficient than that used before these definitions existed. Surely Boers, who was prominently involved in developing the core set and the ACR definition of improvement, would not advocate returning to these earlier times, when individual trials used their own outcome measures (and often used many). Such an approach frequently led to problems in data interpretation resulting from multiple comparisons. One comment by Boers focuses on radiographic change. Clearly, preventing radiographic damage is an important goal of RA therapy. Because many elements of the ACR core set do not correlate well (or at all) with radiographic 2817 2818 LETTERS Table 1. Relationship between clinical improvement and inhibition of radiographic progression in ATTRACT patients treated with infliximab and methotrexate* radiographic damage and clinical activity can be largely independent variables in patients with RA treated with the combination of infliximab and methotrexate. Pearson’s correlation coefficients Length of treatment, weeks Measure of disease activity Change in radiographic score P ACR-N P 54 DAS 28 ACR-N DAS 28 ACR-N 0.25 ⫺0.14 0.18 ⫺0.21 ⬍0.001 0.011 0.003 ⬍0.001 ⫺0.71 ⬍0.001 ⫺0.76 ⬍0.001 102 * ATTRACT ⫽ Anti–Tumor Necrosis Factor Trial in Rheumatoid Arthritis with Concomitant Therapy; ACR-N ⫽ numeric American College of Rheumatology measure of improvement; DAS ⫽ Disease Activity Score. clinical response to the combination of infliximab and methotrexate and inhibition of progressive radiographic damage. The results suggest that one important impact of the combination of infliximab and methotrexate is protection against progressive cartilage and bone damage even in patients who fail to respond clinically to this combination. These findings imply that tumor necrosis factor (TNF) has a critical role in progressive joint damage, even in patients in whom it plays only a modest role in causing signs and symptoms of RA. Consistent with this finding is the observation that patients who did not achieve ACR20 response with the combination of infliximab and methotrexate exhibited significantly less radiographic progression than did patients treated with methotrexate alone, regardless of whether or not they improved. Boers suggests that use of continuous measures of disease activity might be a more appropriate way to establish a relationship between radiographic progression and response to therapeutic agents. To determine whether this would affect the interpretation of the ATTRACT data, we analyzed clinical activity using the DAS and the recently described numeric ACR (ACR-N) (5,6). As shown in Table 1, there was a modest but significant correlation between disease activity assessed in either of these ways and inhibition of radiographic progression (r ⫽ ⫺0.14 to ⫺0.21 for ACR-N and r ⫽ 0.25 to 0.18 for DAS). Importantly, however, the correlation was very modest in that only 2–6% of the inhibition of radiographic progression (r2) could be explained by changes in either continuous measure of disease activity. Therefore, we believe it is safe to conclude that there is only a very modest relationship between clinical benefit caused by treatment with infliximab and methotrexate and inhibition of damage to articular structures. Notably, the correlation between disease activity assessed by the DAS and the ACR-N was also not complete (r ⫽ ⫺0.71 to ⫺0.76), perhaps because the former is weighted toward swollen joints, whereas the latter provides a broader view of disease activity (3). In summary, the data from ATTRACT demonstrate that blocking TNF inhibits radiographic damage even in patients who show minimal or no clinical benefit. This dichotomy in response was apparent when clinical activity was analyzed using continuous variables or the ACR20. Finally, and perhaps of greatest interest, the data clearly show that Peter E. Lipsky, MD National Institute of Arthritis and Musculoskeletal and Skin Diseases National Institutes of Health Bethesda, MD David T. Felson, MD, MPH Boston University School of Medicine Boston, MA Ravinder N. Maini, MD, FRCP Kennedy Institute of Rheumatology Division Imperial College of Science, Technology and Medicine London, UK 1. Boers M. Demonstration of response in rheumatoid arthritis patients who are nonresponders according to the American College of Rheumatology 20% criteria: the paradox of beneficial treatment effects in nonresponders in the ATTRACT trial. Arthritis Rheum 2001;44:2703–4. 2. Lipsky PE, van der Heijde DM, St Clair EW, Furst DE, Breedveld FC, Kalden JR, et al, for the Anti-Tumor Necrosis Factor Trial in Rheumatoid Arthritis with Concomitant Therapy Study Group. Infliximab and methotrexate in the treatment of rheumatoid arthritis. N Engl J Med 2000;343:1594–1602. 3. Felson DT, Anderson JJ, Boers M, Bombardier C, Furst D, Goldsmith C, et al. American College of Rheumatology preliminary definition of improvement in rheumatoid arthritis. Arthritis Rheum 1995;6:727–35. 4. Felson DT, Anderson JJ, Boers M, Bombardier C, Chernoff M, Fried B, et al. The American College of Rheumatology preliminary core set of disease activity measures for rheumatoid arthritis clinical trials. Arthritis Rheum 1993;36:729–40. 5. Van der Heijde DMFM, van’t Hof MA, van Riel PLCM, Theunisse LM, Lubberts EW, van Leeuwen MA, et al. Judging disease activity in clinical practice in rheumatoid arthritis: first step in the development of a disease activity score. Ann Rheum Dis 1990;49:916–20. 6. Van Riel PLCM, van Gestel AM. Area under the curve for the American College of Rheumatology improvement criteria: a valid addition to existing criteria in rheumatoid arthritis? [letter] Arthritis Rheum 2001;44:1719. DOI 10.1002/art.10447 Reply To the Editor: I appreciate the opportunity to reply to the comments by Lipsky and colleagues, and I am concerned that they may have misinterpreted the point I was trying to make. Space limitations may have obstructed clarity, for which I offer my apologies. To expand on my main point, I argue that use of any measure of disease activity that bisects outcome (e.g., response–nonresponse) is likely to result in loss of information when compared with use of continuous measures. The usefulness of such binary measures as prognostic indicators is, therefore, suboptimal. In other words, when studying the relationship (or lack thereof) between disease activity and LETTERS 2819 Table 1. Analysis of indices and change in single disease activity measures in the COBRA trial, by responder status* ACR nonresponders Disease activity score AUC† Pooled index, change after 6 months Disease activity score, change after 6 months Erythrocyte sedimentation rate Physician global assessment, 0–10 Patient global assessment, 0–10 Painful joint count Health assessment questionnaire Pain on visual analog scale, 1–10 Swollen joint count MACTAR Grip strength ACR responders Sulfasalazine (n ⫽ 37) COBRA (n ⫽ 21) P Sulfasalazine (n ⫽ 38) 116 0.34 101 1.06 0.05 ⬍0.0001 91 1.29 ⫺0.37 ⫺1.05 0.001 ⫺17 ⫺0.3 ⫺0.8 0 ⫺0.26 ⫺1.3 ⫺1 3.7 4 ⫺38 ⫺2.3 ⫺1.4 ⫺4 ⫺0.88 ⫺2.4 ⫺3 7.7 21 0.02 0.001 0.39 0.18 0.004 0.17 0.22 0.01 ⬍0.0001 COBRA (n ⫽ 56) P Overall P 85 1.59 0.26 0.02 0.003 ⬍0.0001 ⫺2.18 ⫺2.54 0.08 ⬍0.0001 ⫺36 ⫺3.1 ⫺3.1 ⫺16 ⫺0.73 ⫺2.8 ⫺10 9.8 16 ⫺41 ⫺3.6 ⫺2.8 ⫺21 ⫺1.19 ⫺3.8 ⫺12 10.6 26 0.38 0.002 0.52 0.06 0.04 0.05 0.15 0.4 0.008 0.002 0.001 0.25 0.0004 0.0001 0.002 0.001 0.0007 ⬍0.0001 * COBRA ⫽ Combinatietherapie Bij Reumatoide Artritis; ACR ⫽ American College of Rheumatology; AUC ⫽ area under the curve; MACTAR ⫽ McMaster Toronto Arthritis patient preference questionnaire. † AUC of absolute (not change) scores, by parametric analysis; P value slightly different from original nonparametric analysis. radiographic progression of damage, ACR responder status is a less-than-ideal indicator of disease activity. In support of this view, I demonstrated that COBRA therapy (a combination of step-down prednisolone, methotrexate, and sulfasalazine), like infliximab, slowed radiographic progression of damage even in ACR20 nonresponders. However, in contrast to Lipsky et al, I did not interpret this as an indication that COBRA appears to have a “critical role” in the progression of bone and cartilage damage, even in patients in whom clinical manifestations of RA are not apparently influenced by therapy (1). Also, I chose not to conclude that “COBRA inhibits radiographic damage even in patients who show minimal or no clinical benefit,” let alone that “damage and clinical activity can be largely independent variables.” Instead, I suspected that this finding was an artifact caused by bisection of disease activity through use of the ACR response criteria. This is by no means a fault of these criteria and would most likely occur with other bisections (as Lipsky et al, indeed, showed for the 20% response in the swollen joint count and the C-reactive protein concentration ). The artifact is revealed by my subsequent analysis. Using the COBRA data set, I demonstrated that in the subgroup of ACR20 nonresponders, disease activity (expressed as a continuous measure, i.e., the area under the curve of the DAS) was lower in the COBRA group than in the sulfasalazine group. Lipsky and colleagues have chosen not to replicate my analysis using ATTRACT data, although an abstract presented at the 2001 ACR meeting (2) suggests that the same phenomenon was demonstrated in that data set also. Instead, they take issue with my choice of the DAS as a measure of disease activity and then interpret my analyses in light of the relationship between DAS and swollen joints, as well as between DAS and radiographic damage. However, and with all due respect, this issue may not be relevant. I probably could have used any other measure of disease activity, expressed as either area under the curve or as 6-month change score, and shown the same effect: that, in the subgroup of ACR20 nonresponders, disease activity was lower among those using COBRA therapy than among those treated with sulfasalazine. To prove this, I repeated my previous analysis for the change after 6 months in the pooled index (the primary trial outcome measure, a weighted change score for erythrocyte sedimentation rate, swollen and tender joint counts, physician’s global assessment, and McMaster Toronto Arthritis functional index) and for change scores in all core set measures, grip strength, and DAS (Table 1). Among ACR nonresponders, for every measure, the response (change in disease activity) was numerically greater in the COBRA group. The difference between the COBRA group and the sulfasalazine group was often statistically significant and clinically relevant. I am also concerned about other aspects of the letter from Lipsky et al. The implication that I would discount the clinical usefulness of the ACR response criteria is without ground, and such a conclusion cannot be reached based on my concise communication. As I tried to make clear, I do oppose use of the ACR criteria for purposes for which they were not intended. The presentation by Lipsky et al of correlations between DAS, ACR-N, and radiographic progression of damage is not very useful if the skewness of radiographic data is not taken into consideration. Furthermore, these correlations are, in my opinion, not relevant to the argument. References regarding the ACR-N are problematic for several reasons. The article by van der Heijde et al (3) describing the DAS was originally published in 1990, which was well before publication of any ACR response criteria, and the letter from van Riel and van Gestel (4) was a rather blistering critique of the ACR-N. In my recollection, the ACR-N was never described, let alone formally validated, but suddenly appeared “out of the blue” in the report of the trial of etanercept versus high-dose methotrexate in early RA (5). Finally, I’m no great fan of the DAS. Lipsky et al, however, use reasoning I cannot follow to conclude that this measure is less useful than the core set (or the ACR response) despite its better correlation with damage. 2820 LETTERS In summary, although a damage mechanism independent of disease activity is entirely possible (in fact, our group will be presenting new data to support this theory), I maintain that the current analyses of the COBRA data and probably the ATTRACT data are incapable of showing such a mechanism. As an aside, if Lipsky et al adhere to their reasoning, they should at least admit that COBRA therapy is roughly equivalent to infliximab even in this respect. For now, I believe that current observations are best explained by the hypothesis that the threshold of disease activity above which damage progression occurs is highly individualized, with some RA patients needing only a little “clinically irrelevant” disease activity for progression, and others not progressing despite significant and persisting disease activity (6). Such heterogeneity hinders the study of this relationship unless groups are large or followup is long. Without doubt, such studies are further impeded by the application of measures designed for other purposes. I continue to assert that the ACR criteria are best used to describe and interpret trial results on disease activity at the individual patient level, not as prognostic indicators. Maarten Boers, MSc, MD, PhD VU University Medical Center Amsterdam, The Netherlands 1. Lipsky PE, van der Heijde DM, St Clair EW, Furst DE, Breedveld FC, Kalden JR, et al, for the Anti-Tumor Necrosis Factor Trial in Rheumatoid Arthritis with Concomitant Therapy Study Group. Infliximab and methotrexate in the treatment of rheumatoid arthritis. N Engl J Med 2000;343:1594–1602. 2. Furst D, Yocum D, Han C, Keenan G, Bala M, Weisman W, et al. Clinical benefits of infliximab ⫹ methotrexate treatment for patients with rheumatoid arthritis who were nonresponders by ACR criteria [abstract]. Arthritis Rheum 2001;44:S80. 3. Van der Heijde DMFM, van’t Hof MA, van Riel PLCM, Theunisse LM, Lubberts EW, van Leeuwen MA, et al. Judging disease activity in clinical practice in rheumatoid arthritis: first step in the development of a disease activity score. Ann Rheum Dis 1990;49:916–20. 4. Van Riel PLCM, van Gestel AM. Area under the curve for the American College of Rheumatology improvement criteria: a valid addition to existing criteria in rheumatoid arthritis? [letter] Arthritis Rheum 2001;44:1719. 5. Bathon JM, Martin RW, Fleischmann RM, Tesser JR, Schiff MH, Keystone EC, et al. A comparison of etanercept and methotrexate in patients with early rheumatoid arthritis. N Engl J Med 2000;343: 1586–93. 6. Van Leeuwen MA, van Rijswijk MH, Sluiter WJ, van Riel PL, Kuper IH, van de Putte LB, et al. Individual relationship between progression of radiological damage and the acute phase response in early rheumatoid arthritis: towards development of a decision support system. J Rheumatol 1997;24:20–7. DOI 10.1002/art.10689 Comment on the reply by Boers To the Editor: We are pleased that the series of letters initiated by Dr. Boers’s concise communication has begun to shed light on the issue of the relationship between clinical improvement and radiographic outcome in the ATTRACT and COBRA data sets. We agree with Dr. Boers that from a purely metrologic point of view, dividing patients into responders and nonresponders can result in loss of information compared with the use of continuous measures. For that reason, we provided information in our response to Dr. Boers concerning the significant but rather modest correlation between 2 continuous measures of clinical activity, DAS and ACR-N, and radiographic outcome. Use of the ACR responder criteria addresses the issue that is in the minds of patients and practicing rheumatologists, namely, whether there is a relationship between clinically relevant disease improvement and radiographic outcome, and in the mind of the clinical investigator, namely, whether inflammation, as manifested by a clinically meaningful degree of signs and symptoms, is associated with progressive radiographic damage. For clinicians, patients, and investigators, the answer appears to be that protection from radiographic damage can be seen even in the absence of meaningful improvement in signs and symptoms of inflammation. Of course, there may be evidence of minimal change in the degrees of inflammation that is less than that appreciated reproducibly by a clinician as improvement that still could contribute to progressive radiographic damage. However, the hypothesis that “the threshold of disease activity above which damage progression occurs is highly individualized” is nearly impossible to test, given the realities of clinical trials and the relative imprecision of clinical and radiographic tools. We performed the analysis that Dr. Boers suggested, but our interpretation of the results is somewhat different from his. Indeed, in ATTRACT, infliximab-plus-methotrexate nonresponders assessed by ACR20 achieved significant improvement in mean ACR-N relative to methotrexate-only nonresponders at weeks 30, 54, and 102 (P ⬍ 0.05). In addition, infliximab-plus-methotrexate nonresponders, as assessed by ACR20 at 102 weeks, manifested significant improvement in 5 of 7 ACR components (swollen joint count, tender joint count, C-reactive protein, physician global assessment and patient pain assessment) relative to methotrexate-only nonresponders (P ⬍ 0.05). The infliximab-plus-methotrexate nonresponders also demonstrated significant improvement in the physical component summary score of the Short Form 36 relative to methotrexate-only nonresponders (P ⫽ 0.006). The results indicate that in ATTRACT, as in COBRA, nonresponders to the more effective therapy fared better clinically compared with nonresponders receiving the control therapy. Despite this, the correlation between improvement in disease activity using continuous variables (DAS, ACR-N) and radiographic progression was very modest, implying that changes in signs and symptoms could account for only a small proportion of the radiographic protection. As opposed to the contention of Dr. Boers, we believe that the lack of a more important correlation between improvement in continuous measures of disease activity and radiographic benefit is very telling in the interpretation of the results, and strongly suggests that radiographic damage and clinical activity can be largely independent variables. In summary, we agree with Dr. Boers that “a damage mechanism independent of disease activity is entirely possible.” We await with interest the new data from the Boers group that will apparently support this conclusion. We disagree, however, with Dr. Boers, in that we believe that the LETTERS 2821 ATTRACT data strongly support this contention. In addition, we cannot agree that COBRA and infliximab plus methotrexate are “roughly equivalent,” because no direct comparisons have been performed. Finally, we agree with Dr. Boers that ACR criteria are not useful as a prognostic indicator to predict radiographic outcome in patients treated with infliximab and methotrexate, because both responders and nonresponders manifest protection from structural damage. Peter E. Lipsky, MD National Institute of Arthritis and Musculoskeletal and Skin Diseases National Institutes of Health Bethesda, MD Ravinder N. Maini, MD, FRCP Kennedy Institute of Rheumatology Division Imperial College of Science, Technology and Medicine London, UK DOI 10.1002/art.10469 Case series of selective anti–tumor necrosis factor ␣ therapy using infliximab in patients with nonresponsive chronic HLA–B27–associated anterior uveitis: comment on the articles by Brandt et al To the Editor: Recent reports by Brandt et al (1,2) described the effectiveness of selective anti–tumor necrosis factor ␣ (TNF␣) therapy in ankylosing spondylitis. Because ⬃25–30% of patients with spondylarthropathies are affected by a concurrent eye disease such as anterior uveitis (3), we have been looking at the effectiveness of infliximab for HLA–B27–associated recurrent or chronic uveitis that is nonresponsive to conventional therapy with systemic or local corticosteroids and methotrexate (MTX). We now report on 3 patients with chronic uveitis who were treated with infliximab. Patient 1 had a 2-year history of recurrent uveitis flares. She experienced an average of 2–3 episodes per year and was treated with local and, if necessary, systemic corticosteroids for each episode. One year ago she was treated with a single dose of infliximab (10 mg/kg body weight) after an acute flare, and her uveitis resolved within 1 week. She experienced another flare 6 months later and from then on had repeated episodes of uveitis despite high doses of local as well as systemic corticosteroids, in addition to MTX (17.5 mg/week). Her visual acuity worsened, from 20/20 to 20/30, because of the development of cystoid macular edema. Infliximab was given again, at 5 mg/kg body weight. Systemic and local corticosteroid therapy was stopped, but MTX was continued at 12.5 mg/week. One week after the first infliximab infusion, the uveitis was under control. She received a second infusion 2 weeks later and a third after another 4 weeks. The uveitis remained in remission for 5 months, her visual acuity returned to 15/20, and the dosage of MTX was reduced to 7.5 mg/week. Unfortunately, 5 months after her last infusion of infliximab, she had another relapse. Patient 2, with an 8-year history of recurrent episodes of anterior uveitis, developed iridocyclitis while being treated with MTX (12.5 mg/week) for ankylosing spondylitis. Despite intensive administration of corticosteroid drops (every hour), the uveitis could not be brought to full remission. Seven weeks after onset of uveitis, 1⫹ anterior chamber cells were still visible. Her visual acuity decreased, from 20/25 to 20/50, due to cystoid macular edema, and infliximab therapy was started. Five days after administering infliximab, 5 mg/kg body weight, her visual acuity returned to 20/25, and only rare anterior chamber cells were seen. Additional infusions of infliximab (5 Table 1. Disease activity in patients with chronic uveitis treated with infliximab* Patient 1 Activity Cells Flare Visual acuity Therapy Patient 2 Activity Cells Flare Visual acuity Therapy Patient 3 Activity Cells Flare Visual acuity Therapy Baseline Week 2 Week 8 2⫹ 1⫹ 20/25 CS drops every 2 hours; oral pred 32.5 mg/day; MTX 17.5 mg/week 0 0 15/20 MTX 12.5 mg/week 0 0 15/20 MTX 7.5 mg/week 1⫹ 0 20/50 CS drops every hour; MTX 12.5 mg/week 0 0 20/25 MTX 12.5 mg/week 0 0 15/20 MTX 12.5 mg/week 0 0 20/100 Oral pred 5 mg/day; MTX 12.5 mg/week 0 0 20/30 Oral pred 5 mg or 2.5 mg/day (alternate days); MTX 12.5 mg/week 2⫹ 1⫹ 20/100 CS drops every 2 hours; oral pred 20 mg/day; MTX 20 mg/week * Cells ⫽ anterior chamber cells; CS ⫽ corticosteroids; pred ⫽ prednisone; MTX ⫽ methotrexate. 2822 mg/kg body weight) were given at weeks 2 and 6. Two weeks after initiation of infliximab therapy, not only had the uveitis improved, but her active ankylosing spondylitis had subsided as well. Three weeks after the first infusion of infliximab, her visual acuity improved to 20/20, with only residual subtle pigment epithelial changes present in the macula. Fluorescence angiography was negative for edema. At her last visit, 5 months after initiation of infliximab therapy, the uveitis as well as the ankylosing spondylitis remained in remission, and her visual acuity was 15/20. Patient 3, with a known history of ankylosing spondylitis, had experienced recurrent flares of uveitis for the past 7 years. Within the last 3 years, the uveitis could not be controlled with systemic steroids (17.5 mg/day) plus MTX (20 mg/week). Because of longstanding inflammation that worsened every time steroid therapy was tapered below 17.5 mg/day, she developed chronic cystoid macular edema, which resulted in a decrease in her visual acuity to 20/100. Two weeks after receiving only 1 infusion of infliximab (5 mg/kg body weight), her uveitis subsided completely. The systemic steroid therapy was reduced to 5 mg/day, and the dosage of MTX was reduced to 12.5 mg/week. Because she developed flu-like symptoms following infliximab therapy, she refused any further treatment with infliximab. Nevertheless, uveitis has not recurred within the last 3 months. In the meantime, the chronic macular edema resolved almost completely, and her visual acuity improved, from 20/100 to 20/30. In our case series (Table 1), all patients had developed chronic uveitis before receiving infliximab therapy, despite combined therapy with MTX and corticosteroids. Because infliximab was shown to be very effective in the treatment of various HLA–B27–associated conditions such as ankylosing spondylitis (1,2) and the fact that uveitis went into remission in all of our patients, it seems that this group of patients responds especially well to infliximab therapy. One patient had a recurrence of uveitis at month 5 of followup, a time at which infliximab may no longer be detectable in the body. Reiff and colleagues recently described treatment with etanercept (Immunex, Seattle, WA), a soluble p75 TNF receptor, in children with juvenile rheumatoid arthritis (JRA) who had concurrent chronic corticosteroid- and MTX-resistant uveitis (4). After 6 months of treatment, uveitis was in remission in 3 of 14 eyes. Five of 14 eyes showed improvement, 5 remained unchanged, and 1 deteriorated. The fact that uveitis went into remission in only 20% of the cases—a much smaller percentage than in our study—might be attributable to the difference in the patients treated. It must be considered that clinically, anterior uveitis is a diverse collection of diseases (5), and that JRA and HLA–B27–associated intraocular inflammatory diseases are distinct entities. In addition, we used infliximab rather than etanercept in our study. This might be of some relevance, because in a second study (6) of 16 patients treated with either etanercept (n ⫽ 14) or infliximab (n ⫽ 2) for inflammatory eye diseases or associated joint disease, only 4 of the 14 etanercept patients showed an improvement in the ocular inflammation, as opposed to improvement in both infliximab patients. We would like to stress the fact that the number of patients in our case series is small because one of our goals was to keep the group of patients as homogenous as possible. Thus, our group included only patients with HLA–B27–associated LETTERS chronic uveitis, most of whom usually respond adequately to corticosteroids and, if necessary, systemic MTX (7). The good clinical response to infliximab of other HLA–B27–associated inflammatory diseases, such as ankylosing spondylitis and psoriatic arthritis (1,2,8), supports our finding that this group of patients seems to be especially responsive to selective anti-TNF␣ therapy. In conclusion, in our case series, infliximab was highly effective in treating patients with HLA–B27–associated uveitis resistant to current antiinflammatory and immunosuppressive therapy. The series was small, however, and additional larger studies will be needed to further evaluate the effectiveness of infliximab therapy in this entity. Yosuf El-Shabrawi, MD Josef Hermann, MD Karl-Franzens University Graz, Austria 1. Brandt J, Haibel H, Cornely D, Golder W, Gonzalez J, Reddig J, et al. Successful treatment of active ankylosing spondylitis with the anti–tumor necrosis factor ␣ monoclonal antibody infliximab. Arthritis Rheum 2000;43:1346–52. 2. Brandt J, Haibel H, Sieper J, Reddig J, Braun J. Infliximab treatment of severe ankylosing spondylitis: one-year followup. Arthritis Rheum 2001;44:2936–7. 3. Kahn M. An overview of clinical spectrum and heterogeneity of spondyloarthropathies. Rheum Dis Clin North Am 1992;18:1–10. 4. Reiff A, Takei S, Sadeghi S, Stout A, Shaham B, Bernstein B, et al. Etanercept therapy in children with treatment-resistant uveitis. Arthritis Rheum 2001;44:1411–5. 5. Wakefield D. Report on the 4th International Symposium of Ocular Inflammation. Ocul Immunol Inflamm 1997;5:1–9. 6. Smith JR, Levinson RD, Holland GN, Jabs DA, Robinson MR, Whitcup SM, et al. Differential efficacy of tumor necrosis factor inhibition in the management of inflammatory eye disease and associated rheumatic disease. Arthritis Rheum (Arthritis Care Res) 2001;45:252–7. 7. Samson C, Waheed N, Baltatzis S, Foster CS. Methotrexate therapy for chronic noninfectious uveitis: analysis of a case series of 160 patients. Ophthalmology 2001;108:1134–9. 8. Ogilvie AL, Antoni C, Dechant C, Manger B, Kalden JR, Schuler G, et al. Treatment of psoriatic arthritis with antitumor necrosis factor-alpha antibody clears skin lesions of psoriasis resistant to treatment with methotrexate. Br J Dermatol 2001;144:587–9. DOI 10.1002/art.10470 Reply To the Editor: Eye involvement is quite common in inflammatory rheumatic conditions. Among 407 patients with uveitis, Behcet’s disease, spondylarthritides (SpA), and sarcoidosis were the most frequent diagnoses possibly associated with rheumatologic symptoms (1). In other rheumatic diseases such as rheumatoid arthritis (RA), different anatomic parts of the eye, such as the sclera, are affected. Although the pathophysiologic and genetic backgrounds of these diseases are thought to be rather different, common denominators such as the proinflammatory cytokine TNF␣ seem to be involved in some, but LETTERS clearly not all, immunologically mediated diseases. As an outstanding example, anti-TNF therapy is not effective in multiple sclerosis (2). In inflammatory eye diseases, some positive but also controversial results on the role of TNF␣ have been reported, from experimental studies (3–5) and more recently from early pilot studies (6–8), such as that described in the letter from our Austrian colleagues, to which we now respond. In an early study (3), mice with endotoxin-induced uveitis (EIU) were pretreated with either repeated injections of TNF␣ or with a single injection of antibody against it. Although both pretreatments conferred protection against the systemic toxic effects of lipopolysaccharides (LPS), TNFresistant mice and those treated with anti-TNF␣ antibody demonstrated an exacerbation of EIU when compared with control animals. The authors concluded, “Unlike its apparent participation in the systemic effect of endotoxin, TNF␣ is not directly involved in the pathogenesis of EIU and may even protect against the inflammatory processes of this disease.” In an investigation of TNF␣ in rabbits with uveitis induced by intravitreal injection of LPS (4), peak concentrations of aqueous TNF␣ preceded peak aqueous leukocyte counts. Administration of anti-TNF␣ monoclonal antibody (10 g) suppressed the number of infiltrating cells by 50% and suppressed protein leakage by 40%. TNF␣ seems also to have an important proinflammatory role in experimental autoimmune uveoretinitis (EAU), which serves as a model for several immune-mediated diseases that affect the eye in humans (5). Mice treated with 100 l or 300 l of rabbit antiserum or polyclonal antibodies to human TNF␣ significantly improved when the treatment was given during the afferent but not during the efferent stage of EAU. In patients with Behçet’s disease, permanent loss of vision resulting from relapsing ocular inflammation occurs frequently. As recently reported (9), a single infusion of infliximab to 5 patients with relapsing panuveitis, given at the immediate onset of last relapse, led to remission of ocular inflammation within the first 24 hours, and complete suppression was seen 7 days after treatment in all patients. Similarly, other investigators reported beneficial effects in Behçetassociated uveitis (10). If confirmed in prospective studies, this can be regarded as a significant success of medicine because of the bad prognosis of uveitis in Behçet’s disease despite intensive immunosuppressive therapy. In an earlier, retrospective study (6) of 16 patients between the ages of 7 and 78 (8 with RA, 3 with JRA, 1 with ankylosing spondylitis [AS], 1 with psoriatic SpA, and 3 with uveitis with no concomitant rheumatic disease), 14 received etanercept and 2 received infliximab for either inflammatory eye disease (9 uveitis, 7 scleritis) or associated joint disease. Although 12 (100%) of 12 patients with active articular inflammation had improvement of joint disease, only 6 (38%) of 16 with ocular inflammation experienced improvement of eye disease. In 5 patients, first episodes of inflammatory eye disease occurred while they were taking a TNF inhibitor. From these data, it is not clear whether anti-TNF therapy is beneficial for inflammatory eye disease in patients with rheumatic conditions. More recently, treatment with infliximab was reported to be beneficial in patients with uveitis associated with Crohn’s disease (7). In a prospective study (8) involving 10 children with 2823 chronic active uveitis, 7 had uveitis associated with pauciarticular JRA, and 5 were antinuclear antibody (ANA) positive. All patients had failed previous therapy with topical steroids and methotrexate and/or cyclosporine. All were treated with etanercept at a dosage of 0.4 mg/kg twice weekly for the first 3 months, and then, if eyes did not improve, with 25 mg twice weekly (mean 1.1 mg/kg) for at least 3 additional months. Within 3 months, 10 (63%) of 16 affected eyes showed a rapid decrease in anterior chamber cell density (P ⫽ 0.017), including remission in 4 eyes. An exacerbation of uveitis during etanercept therapy occurred in only 1 child (7%). Following a dose increase, to an average of 1.1 mg/kg after 3 months in 7 children, no further improvement was noted. The natural course of uveitis in patients who are HLA–B27 positive versus those who are ANA positive is known to be rather different. The authors concluded that treatment of uveitis with etanercept in systemic and/or topical form (which has not been studied so far) needs further study. In their letter, El-Shabrawi and Hermann reported a beneficial effect of infliximab with HLA–B27–associated uveitis in 3 patients, and these authors recently published their 1-year experience with this treatment in abstract form (11). Thus, the results from these uncontrolled observations are basically positive. In addition, our own experience with infliximab in a randomized trial with AS patients (12) is also suggestive of a beneficial effect, because 3 patients in the placebo group, compared with 1 patient in the infliximab group, developed uveitis over 3 months. Nevertheless, there seems to be a need to contribute some points for discussion from the perspective of rheumatologists. In the vast majority of patients with SpA, the natural course of anterior uveitis seems rather benign. Thus, we think that anti-TNF therapy should be considered only in severe refractory cases. In this regard, there is a need to mention that some evidence suggests that sulfasalazine is effective in preventing episodes of uveitis in SpA (13). In contrast to methotrexate, sulfasalazine was also shown to be effective for peripheral arthritis in AS and other SpA (14). El-Shabrawi and Hermann used the one-shot approach, administering a high dose (10 mg) of infliximab in some patients. Because this was partly successful in uveitis and apparently also in Crohn’s disease, it raises the questions of whether this approach should be tried initially in SpA in general, and whether the high dosage is associated with more severe side effects. The only trial on dosage is a small, open study (15). However, it seems clear that there is a need for a new approach using individual dosing regimens (which are difficult to study). Concerning side effects, there is no evidence for a major dose effect to date, but this issue is increasingly relevant because severe and even fatal side effects have been reported with use of both anti-TNF agents (16). Taken together, these findings provide increasing evidence that anti-TNF therapy is efficacious in SpA—possibly even more than in RA. It is not completely clear whether all types of organ involvement and all possible targets of SpA (i.e., spinal inflammation, peripheral arthritis, enthesitis, psoriasis, colitis, uveitis) are equally affected, and whether both currently available anti-TNF agents are equally effective. Currently, there is some evidence that both infliximab and etanercept are beneficial for spinal inflammation, peripheral arthritis, en- 2824 LETTERS thesitis, and psoriasis, but there is no evidence for efficacy of etanercept in Crohn’s disease (17). As already discussed, there are no randomized studies on the efficacy of anti-TNF therapy in uveitis. Based on the open observations published, there is hope that both infliximab and etanercept might be effective in some, but clearly not all, cases, which could be very interesting for scientific purposes, especially in relation to the pathogenesis of uveitis and individual courses of disease. Jürgen Braun, MD Rheumazentrum Ruhrgebiet Herne, Germany and University Hospital Benjamin Franklin Joachim Sieper, MD University Hospital Benjamin Franklin and Rheumatology Research Center Berlin, Germany 1. Bañares A, Jover JA, Fernández-Gutiérrez B, Benı́tez del Castillo JM, Garcı́a J, Vargas E, et al. Patterns of uveitis as a guide in making rheumatologic and immunologic diagnoses. Arthritis Rheum 1997;40:358–70. 2. Van Oosten BW, Barkhof F, Truyen L, Boringa JB, Bertelsmann FW, von Blomberg BM, et al. Increased MRI activity and immune activation in two multiple sclerosis patients treated with the monoclonal anti-tumor necrosis factor antibody cA2. Neurology 1996;47:1531–4. 3. Kasner L, Chan CC, Whitcup SM, Gery I. The paradoxical effect of tumor necrosis factor alpha (TNF-alpha) in endotoxin-induced uveitis. Invest Ophthalmol Vis Sci 1993;34:2911–7. 4. Sartani G, Silver PB, Rizzo LV, Chan CC, Wiggert B, Mastorakos G, et al. Anti-tumor necrosis factor alpha therapy suppresses the induction of experimental autoimmune uveoretinitis in mice by inhibiting antigen priming. Invest Ophthalmol Vis Sci 1996;37: 2211–8. 5. Mo JS, Matsukawa A, Ohkawara S, Yoshinaga M. Involvement of TNF alpha, IL-1 beta and IL-1 receptor antagonist in LPS-induced rabbit uveitis. Exp Eye Res 1998;66:547–57. 6. Smith JR, Levinson RD, Holland GN, Jabs DA, Robinson MR, Whitcup SM, et al. Differential efficacy of tumor necrosis factor inhibition in the management of inflammatory eye disease and associated rheumatic disease. Arthritis Rheum (Arthritis Care Res) 2001;45:252–7. 7. Fries W, Giofre MR, Catanoso M, Lo GR. Treatment of acute uveitis associated with Crohn’s disease and sacroileitis with infliximab. Am J Gastroenterol 2002;97:499–500. 8. Reiff A, Takei S, Sadeghi S, Stout A, Shaham B, Bernstein B, et al. Etanercept therapy in children with treatment-resistant uveitis. Arthritis Rheum 2001;44:1411–5. 9. Sfikakis PP, Theodossiadis PG, Katsiari CG, Kaklamanis P, Markomichelakis NN. Effect of infliximab on sight-threatening panuveitis in Behcet’s disease. Lancet 2001;358:295–6. 10. Munoz-Fernandez S, Hidalgo V, Fernandez-Melon J, Schlincker A, Martin-Mola E. Effect of infliximab on threatening panuveitis in Behcet’s disease. Lancet 2001;358:1644. 11. El-Shabrawi Y, Hermann J. Anti-TNFA therapy with infliximab in the treatment of HLA B27 associated acute anterior uveitis: a one year follow up [abstract]. Arthritis Rheum 2001;44 Suppl 9:S123. 12. Braun J, Brandt J, Listing J, Zink A, Alten R, Krause A, et al. Treatment of active ankylosing spondylitis with infliximab: a randomised controlled multicenter trial. Lancet 2002;359:1187–93. 13. Benitez-Del-Castillo JM, Garcia-Sanchez J, Iradier T, Banares A. Sulfasalazine in the prevention of anterior uveitis associated with ankylosing spondylitis. Eye 2000;14:340–3. 14. Clegg DO, Reda DJ, Abdellatif M. Comparison of sulfasalazine and placebo for the treatment of axial and peripheral articular manifestations of the seronegative spondylarthropathies: a Department of Veterans Affairs cooperative study. Arthritis Rheum 1999;42:2325–9. 15. Brandt J, Haibel H, Reddig J, Sieper J, Braun J. Successful treatment of severe undifferentiated spondyloarthropathy with the anti-tumor necrosis factor ␣ monoclonal antibody infliximab. J Rheumatol 2002;29:118–22. 16. Keane J, Gershon S, Wise RP, Mirabile-Levens E, Kasznica J, Schwieterman WD, et al. Tuberculosis associated with infliximab, a tumor necrosis factor alpha-neutralizing agent. N Engl J Med 2001;345:1098–104. 17. Sandborn WJ, Hanauer SB, Katz S, Safdi M, Wolf DG, Baerg RD, et al. Etanercept for active Crohn’s disease: a randomized, doubleblind, placebo-controlled trial. Gastroenterology 2001;121: 1088–94. DOI 10.1002/art.10473 Is expression of intracellular citrullinated proteins in synovial tissue specific for rheumatoid arthritis? Comment on the article by Baeten et al To the Editor: The presence in serum of autoantibodies directed to citrullinated antigens is highly specific for rheumatoid arthritis (RA) (1). This might theoretically be explained by an immune response to specific overexpression of citrullinated antigens in rheumatoid synovium (2) or, alternatively, by an abnormal humoral response to the presence of citrullinated proteins, which could be present in any inflamed synovial tissue. A recent report by Baeten and colleagues suggested that the expression of citrullinated proteins in synovial tissue is specific for RA (3). In that study, citrulline-positive cells were detected in the synovial membrane of 50% of the RA patients but in none of the controls. In a separate experiment, Baeten et al examined biopsy samples obtained from 33 patients from the Academic Medical Center/University of Amsterdam: 19 with RA, 3 with osteoarthritis, 5 with psoriatic arthritis, 1 with reactive arthritis (ReA), and 5 with other arthritides. The biopsy sections were coded and randomly analyzed. Citrulline-positive cells were detected in the synovial membrane of 33% of the patients with RA but in none of the controls, confirming the results of MassonBessiere et al (2). These data could suggest a novel diagnostic tool for discriminating RA from other inflammatory joint diseases by examination of synovial biopsy samples. The expression of intracellular citrullinated proteins in synovial tissue was also investigated at the Academic Medical Center/University of Amsterdam. We included 14 patients with RA, 10 with osteoarthritis (OA), 7 with ReA, and 23 with other arthritides. The rabbit anti–L-citrulline polyclonal antibody (6 g/ml) and staining protocol were identical to those used by Baeten and colleagues (3). Citrulline-containing antigens were observed in synovial cryostat sections of anti– citrulline-containing peptide (anti-CCP)–positive and antiCCP–negative patients. Of importance, replacing the rabbit anti–L-citrulline antibody with control rabbit anti–fluorescein isothiocyanate (anti-FITC) antibody (Dako, Glostrup, Den- LETTERS mark) revealed positive staining in synovial tissue of several RA patients, suggesting nonspecific binding in the negative controls. The same staining was seen when lower concentrations of antibody were used (3 g/ml and 1.5 g/ml). Excluding cells that stained with the negative control anti-FITC antibody in consecutive sections, we found synovial tissue expression of citrullinated proteins in 57% of the patients with RA, 30% of the patients with OA, 29% of the patients with ReA, and 26% of the patients with other arthritides. Subsequently, a recombinant single-chain variable fragment (scFv) monoclonal antibody was selected against a cyclic citrullinated peptide from a patient antibody-fragment phage-display library (Raats J: unpublished observations). This scFv and patient antibodies affinity-purified with a citrullinated peptide were both used for immunohistochemical staining. Using a control scFv antibody, staining was absent in the negative controls. Again, we detected citrullinated antigens in 2825 both RA and non-RA tissue (4). Staining with scFv monoclonal antibody was noted in synovial lining cells and in (peri)vascular areas in 43% of the RA patients, 54% of the OA patients, 56% of the ReA patients, and 43% of the patients with other arthritides. Similar results were obtained when horseradish peroxidase–labeled affinity-purified human IgG against citrullinated peptides was used. Expression of citrullinated antigens in synovial tissue was not associated with the presence of anti-CCP autoantibodies in serum or synovial fluid. Thus, we could not confirm the specificity of intracellular citrullinated protein expression in rheumatoid synovial tissue, and we observed nonspecific binding by rabbit immunoglobulins to RA synovium. To understand the discrepancy between the results obtained in the 2 laboratories, we evaluated in Amsterdam the sections that were stained in Ghent. Similar to the findings in our earlier experiments, 2 populations of citrulline-containing Figure 1. Double immunohistochemical staining in rheumatoid synovial tissue. A, Field of CD38⫹ plasma cells (red) and citrulline-positive cells (rabbit anti–L-citrulline polyclonal antibody) (blue). C, Field of CD38⫹ plasma cells (red) and control antibody (blue) (rabbit anti–fluorescein isothiocyanate [anti-FITC]). An alkaline phosphatase technique was used to detect citrulline-positive cells and control antibody; peroxidase was used to detect CD38⫹ plasma cells. Arrows indicate doublestained cells. B, Citrulline-positive cells (rabbit anti–L-citrulline polyclonal antibody). D, Control antibody (rabbit anti-FITC). (Single-stain peroxidase technique, Mayer’s hemalum counterstained; original magnification ⫻ 400.) 2826 LETTERS cells were identified: strongly positive cells (ⱕ3 cells/biopsy section) and cells that exhibited weaker staining. The cells with intense staining were found mainly in RA synovium, whereas weaker staining was detected in both RA and control tissue. Only sections with strongly positive cells were judged as positive by the investigators in Ghent, whereas we included both cell populations in our evaluation. Using the same microscopic evaluation, we could confirm the results from Ghent. However, cells appeared strongly positive, in most cases the same cells were positive in control experiments using the irrelevant anti-FITC rabbit antibody, suggesting nonspecific binding of rabbit immunoglobulins to cells that are specifically increased in RA synovium. Because the number of plasma cells is specifically increased in rheumatoid synovial tissue (5), and, according to previous experience, rabbit immunoglobulins may bind nonspecifically to these cells, we hypothesized that rabbit anti–Lcitrulline polyclonal antibody may bind to (presumably rheumatoid factor–producing) plasma cells. This hypothesis was confirmed by double-staining experiments using the rabbit anti–L-citrulline antibody with phenotypic markers to detect T cells (CD3), B cells (CD22), plasma cells (CD38 and CD138), and macrophages (CD68) (Figure 1). These experiments revealed clear coexpression of the strongly positive cell population with plasma cell markers, but not with other cells. Similar double-staining was found using the control anti-FITC antibody, confirming nonspecific binding of rabbit immunoglobulins to synovial plasma cells, especially in RA. Based on these results, we want to stress that rabbit immunoglobulins may nonspecifically bind to plasma cells in RA synovium. Therefore, the use of proper negative controls is essential. Conceivably, some commercially available negative controls are selected on the basis of negative staining and, therefore, different controls should be included. It appears likely that certain citrullinated proteins may be specifically up-regulated in rheumatoid synovial tissue, whereas others could be present in both RA and other arthritides. However, rabbit anti–L-citrulline polyclonal antibody may not be the most reliable antibody to detect citrullinated proteins in synovial tissue by immunohistology. T. J. M. Smeets, BSc Academic Medical Center/University of Amsterdam Amsterdam, The Netherlands E. R. Vossenaar W. J. van Venrooij, PhD University of Nijmegen Nijmegen, The Netherlands P. P. Tak, MD, PhD Academic Medical Center/University of Amsterdam Amsterdam, The Netherlands 1. Schellekens GA, de Jong BA, van den Hoogen FH, van de Putte LB, van Venrooij WJ. Citrulline is an essential constituent of antigenic determinants recognized by rheumatoid arthritis-specific autoantibodies. J Clin Invest 1998;101:273–81. 2. Masson-Bessiere C, Sebbag M, Girbal-Neuhauser E, Nogueira L, Vincent C, Senshu T, et al. The major synovial targets of the rheumatoid arthritis-specific antifilaggrin autoantibodies are deiminated forms of the alpha- and beta-chains of fibrin. J Immunol 2001;166:4177–84. 3. Baeten D, Peene I, Union A, Meheus L, Sebbag M, Serre G, et al. Specific presence of intracellular citrullinated proteins in rheumatoid arthritis synovium: relevance to antifilaggrin autoantibodies. Arthritis Rheum 2001;44:2255–62. 4. Smeets TJ, Vossenaar ER, Kraan MC, van Mansum WAM, Raats JM, van Venrooij WJ, et al. Expression of citrullin-containing antigens in RA synovium. Arthritis Res 2001;3:A2. 5. Kraan MC, Haringman JJ, Post WJ, Versendaal J, Breedveld FC, Tak PP. Immunohistological analysis of synovial tissue for differential diagnosis in early arthritis. Rheumatology 1999;38:1074–80. DOI 10.1002/art.10474 Reply To the Editor: We thank Smeets et al for their comments on our recent study, in which we investigated by immunohistochemistry the presence of citrullinated proteins in the synovial membrane of 36 patients with RA and in 52 controls with inflammatory or noninflammatory disease. Using a polyclonal rabbit antibody, we observed the presence of positively staining cells in 50% of the patients with RA but in none of the controls. As clearly mentioned in the report, no positive staining was seen with an isotype- and concentration-matched control antibody (rabbit immunoglobulin fraction; Dako, Glostrup, Denmark), the results were confirmed with another antibody monospecific for proteins containing modified citrulline (1), double immunofluorescence showed no colocalization with B cell and plasma cell markers, and positive staining with the anticitrulline antibody was seen in both rheumatoid factor (RF)–positive and RF-negative patients. Moreover, the anticitrulline reactivity colocalized with affinity-purified antifilaggrin autoantibodies. In an independent study, Smeets et al tried to confirm these results in 54 synovial biopsy samples, using one of the antibodies used in the Ghent study. They concluded that positive staining with the anticitrulline antibody was seen in both RA (8 of 14) and control (11 of 40) synovia, and that nonspecific staining with a control antibody was observed in “several” RA patients. These results are clearly discrepant with our observations and raise three questions. First, is positive staining with the anticitrulline antibody specific for RA synovium? Second, what exactly does this antibody recognize? Third, independent of the polyclonal rabbit antibody, is the presence of citrullinated proteins specific for RA synovium? In order to clarify the first question regarding the RA specificity of the staining with the polyclonal rabbit anticitrulline antibody, we agreed to perform a collaborative study with Smeets et al. Synovial sections obtained from 33 different patients from Amsterdam were sent to Ghent for staining with the anticitrulline antibody. We were blinded as to the diagnoses of the patients from whom the samples were obtained. We observed anticitrulline-positive cells in 8 patients. After staining and scoring, the sections were sent back to Amsterdam and the diagnoses revealed: 19 sections were from patients with RA, and 14 were from controls (psoriatic arthritis, ReA, OA). All sections scoring positively for anticitrulline in Ghent were from RA patients. Thus, anticitrulline staining was positive in 8 of the 19 RA patients and in none of the 14 controls. LETTERS These figures are in complete accordance with those in our previous report, confirming that this staining is specific for RA synovium and could thus be a novel diagnostic tool. Smeets et al, whose colleagues previously explored the potential of synovial histopathology in the diagnosis of undifferentiated arthritis (2), reanalyzed the same sections, and, after exclusion of “weakly positive cells” (judged as negative in Ghent), confirmed our findings. With regard to the second question, based on nonspecific staining in negative controls and on the results of doublestaining experiments in their own study (but not in our collaborative study!), Smeets et al argue that the staining may be attributable to nonspecific binding of polyclonal rat antibodies to synovial plasma cells (which implies that these plasma cells are specifically present in RA synovium; for example, RF-producing plasma cells). Although it is speculative to comment on only partially presented data (because of questions such as, in how many RA patients was nonspecific staining seen, was the nonspecific binding seen in all RA patients that were scored anticitrulline positive and in none of the anticitrulline-negative patients, was there any nonspecific staining in the non-RA patients, was the anticitrulline staining correlated with RF status and/or the presence of synovial plasma cells?), we completely agree that staining with polyclonal rabbit antibodies may induce nonspecific binding to plasma cells. Therefore, we included in our original study a concentration-matched control polyclonal rabbit antibody: in none of the control sections was there positive intracellular staining as seen with the anticitrulline antibody. Moreover, citrulline-positive cells were seen not only in the sublining but also in the lining layer and in synovial cell cultures, which makes the plasma cell hypothesis unlikely. As already mentioned, double stainings with plasma cell markers were negative, and positive anticitrulline staining was seen in both RF-positive and RF-negative RA patients. Of major importance, it must be clarified that in our collaborative study with Smeets et al, we simultaneously stained parallel sections with the same polyclonal rabbit control antibody that was used in our original study, and no positive staining was seen. In contrast with the statement of Smeets and colleagues, we did not perform stainings with the irrelevant anti-FITC antibody nor any double staining in this collaborative study. The fact that they are probably referring to separate experiments performed in Amsterdam makes direct comparison and interpretation most speculative. Therefore, the hypothesis of nonspecific binding to RF-producing plasma cells certainly deserves further attention but is, at present, supported by only little hard evidence. Finally, Smeets et al mention that it is likely that certain citrullinated proteins may be specifically up-regulated in rheumatoid synovial tissue, whereas others could be present in both RA and other arthritides. Although we tend to agree with this statement, there are at present only few published data on this point (3). Masson-Bessiere et al indicate that indeed several citrullinated proteins are present in RA synovium, of which the alpha and beta chains of fibrin are selectively recognized by antifilaggrin autoantibodies (3). However, it is not yet known which of these citrullinated proteins are also present in non-RA inflamed synovium. Therefore, the development and validation of new immunologic tools, such as the recombinant scFv monoclonal antibody 2827 against a cyclic citrullinated peptide mentioned by Smeets et al, and further studies investigating the biochemical identity of citrullinated proteins in synovium are mandatory. In conclusion, based on results of a collaborative study with Smeets et al, we want to confirm that positive staining with the polyclonal rabbit anticitrulline antibody is specific for RA synovium and should thus be explored as a new tool in the early diagnosis of undifferentiated arthritis. Conversely, it is clear that further biochemical characterization of the citrullinated proteins present in synovium of RA patients and inflammatory disease controls is needed to investigate their role in the pathogenesis of the disease. We hope this debate will stimulate other researchers to explore these two questions. Dominique Baeten, MD, PhD Eric M. Veys, MD, PhD Filip De Keyser, MD, PhD Ghent University Hospital Ghent, Belgium 1. Senshu T, Akiyama K, Kan S, Asaga H, Ishigami H, Manabe M. Detection of deiminated proteins in rat skin: probing with a monospecific antibody after modification of citrulline residues. J Invest Dermatol 1995;105:163–9. 2. Kraan MC, Haringman JJ, Post WJ, Versendaal J, Breedveld FC, Tak PP. Immunohistochemical analysis of synovial tissue for differential diagnosis in early arthritis. Rheumatology 1999;38:1074–80. 3. Masson-Bessiere C, Sebbag M, Girbal-Neuhauser E, Nogueira L, Vincent C, Senshu T, et al. The major synovial targets of the rheumatoid arthritis-specific antifilaggrin autoantibodies are deiminated forms of the alpha- and beta-chain of fibrin. J Immunol 2001;166:4177–84. DOI 10.1002/art.10525 Is acquired C1 inhibitor deficiency associated with lupus a distinct disease entity? Comment on the article by Cacoub et al To the Editor: We read with interest the recent report of acquired C1 inhibitor deficiency associated with systemic lupus erythematosus (SLE) (Cacoub P, Fremeaux-Bacchi V, De Lacroix I, Guillien F, Kahn M, Kazatchkine M, et al. A new type of acquired C1 inhibitor deficiency associated with systemic lupus erythematosus. Arthritis Rheum 2001;44:1836–40). Cacoub et al make an interesting, albeit unconvincing, case for a new type of acquired C1 inhibitor deficiency. The case for a distinct disease entity hinges on the temporal occurrence of angioedema in 3 patients with borderline low antigenic levels of C1 inhibitor (values ranging from 6 mg/dl to 8 mg/dl [normal 9–15]) on a background of hypocomplementemia. Interestingly, C1 inhibitor functional activity is not presented. As the authors state, hypocomplementemia due to classical pathway activation is a frequent finding in SLE, and the slight reduction in C1 inhibitor levels may reflect merely complement activation rather than a primary abnormality. In order to assess the true significance of the marginally low C1 inhibitor levels in these 3 patients, it is essential that C1 2828 LETTERS inhibitor levels in SLE patients without angioedema be presented. Moreover, it is noteworthy that C1 inhibitor levels have returned to normal in all 3 patients. The transient nature of this apparent C1 inhibitor deficiency calls into question whether this represents a distinct entity. Patrice Cacoub, MD Hôpital La Pitié-Salpêtrière Véronique Frémeaux-Bacchi, MD Hôpital Européen Georges Pompidou Paris, France Aamir Aslam, BSc, MRCP Siraj A. Misbah, MSc, FRCP, FRCPath John Radcliffe Hospital Oxford, UK 1. Ochonisky S, Intrator L, Wechsler J, Revuz J, Bagot M. Acquired C1 inhibitor deficiency revealing systemic lupus erythematosus. Dermatology 1993;186:261–3. 2. Fawaz-Estrup F. Human parvovirus infection: rheumatic manifestations, angioedema, C1 esterase inhibitor deficiency, ANA positivity, and possible onset of systemic lupus erythematosus. J Rheumatol 1996;23:1180–5. DOI 10.1002/art.10522 Reply DOI 10.1002/art.10378 To the Editor: We thank Aslam and Misbah for their comments on the 3 cases illustrating a new type of acquired C1 inhibitor deficiency. C1 esterase inhibitor deficiency is an uncommon disorder characterized by recurrent episodes of angioedema without urticaria. From a clinical point of view, the symptoms resemble those seen in inherited or acquired C1 inhibitor deficiency. The diagnosis of C1 inhibitor deficiency was based on low plasma levels of C1 inhibitor antigen. At the time of angioedema, the antigenic and functional levels of C1 inhibitor were below normal in all patients (functional C1 inhibitor activity was assessed using a chromogenic assay). Only 2 cases of quantitative deficiency of C1 inhibitor antigen in patients with SLE in the absence of angioedema have been reported (1,2). Part of our report described 3 SLE patients with angioedema and low C1 inhibitor antigenic levels. A genetic C1 inhibitor deficiency can be ruled out because there was no quantitative or functional C1 inhibitor deficiency after steroid treatment. These tests were repeated at least 4 times (2 months apart) for all patients, and the results were consistently confirmed. Two forms of acquired angioedema have been previously described. Type I occurs in patients with lymphoproliferative disorders, with no detectable circulating antibodies to C1 inhibitor. Type II acquired angioedema is characterized by the presence of anti–C1 inhibitor antibodies and a circulating low molecular weight, functionally inactive form of C1 inhibitor. The search for a lymphoproliferative disorder or biologic evidence of type II acquired C1 inhibitor deficiency yielded negative results in all 3 patients. In a patient with typical angioedema and no evidence of type I or type II acquired angioedema, the presence of major classical pathway–mediated complement consumption, including low levels of C3, suggests the third type of acquired angioedema, and patients should be tested for systemic lupus. Although acquired angioedema has been assumed to be related to antigenic or functional deficiency of C1 inhibitor, a third type exists that is associated with transient C1 inhibitor deficiency and occurs in the context of autoimmune disease. To the Editor: We read with interest the report of Keen et al (1), who demonstrated the relationship between a common polymorphic Sp1 binding site in the first intron of the COL1A1 gene, bone mineral density (BMD), and bone remodeling. However, discordant results have been reported (2,3). Our intention was to determine whether Sp1 genotypes are related to BMD, markers of collagen turnover, and ultrasound stiffness. We studied 108 postmenopausal Caucasian women (mean ⫾ SD age 62.5 ⫾ 8.9 years), who underwent measurement of BMD (gm/cm2) at the lumbar spine (L1–4) and at the hip using dual x-ray absorptiometry. Stiffness of the heel was assessed by ultrasound transmission measurement (CUBA Clinical Contact Ultrasound Bone Analyzer; Cooper Surgical, Lake Forest, CA). To investigate levels of collagen turnover, carboxyterminal propeptide of type I procollagen (PICP) and carboxyterminal telopeptide of type I collagen (ICTP) were determined using a radioimmunoassay kit (Orion Diagnostica, Helsinki, Finland). The Sp1 COL1A1 genotypes were assessed as described by Grant et al (4). Polymerase chain reaction was followed by endonuclease digestion with Mls I (Fermentas, Vilnius, Lithuania). Fragments were separated on 3% agarose gel, and Sp1 genotypes were identified. Alleles were coded as “S” (absence of restriction site) and “s” (presence of restriction site). There were no significant differences in anthropometric and environmental variables (age, body mass index [BMI], years since menopause [YSM], calcium intake, or physical activity) in the COL1A1 genotype subgroups. None of the probands had been using hormone replacement therapy. Due to the small number of subjects with the ss genotype, use of analysis of covariance (ANCOVA) was inappropriate. Therefore, the ss genotype was excluded from the analysis. Differences in BMD, stiffness, and biochemical markers between the SS and Ss genotype groups were evaluated by ANCOVA after adjustment for the potential confounding effect of BMI and YSM. The COL1A1 genotype distributions were similar to Is the Sp1 polymorphism in the COL1A1 gene a risk factor for postmenopausal osteoporosis? Comment on the article by Keen et al LETTERS 2829 Table 1. Bone density, stiffness index, and biochemical markers of collagen turnover in relation to COL1A1 genotypes* Genotype Variable SS Ss ss P, SS vs. Ss† BMD L1–4, gm/cm2 BMD total hip, gm/cm2 Stiffness index score S-PICP, g/liter S-ICTP, g/liter 0.85 (0.74–0.973) 0.76 (0.72–0.86) 66.6 (56.4–85.3) 122.3 (90.5–150.0) 2.81 (2.37–3.69) 0.85 (0.77–0.96) 0.76 (0.69–0.91) 74.8 (59.4–84.9) 122.8 (102.6–141.4) 3.19 (2.24–3.80) 0.9 (0.75–0.97) 0.75 (0.63–0.89) 60.4 (59.8–66.9) 146.65 (136.4–156.9) 4.07 (3.33–4.81) 0.87 0.67 0.86 0.62 0.53 *Values are the median (lower–upper quartile). S ⫽ absence of restriction site; s ⫽ presence of restriction site; BMD ⫽ bone mineral density; PICP ⫽ carboxyterminal propeptide of type I procollagen; ICTP ⫽ carboxyterminal telopeptide of type I collagen. † Values were determined by analysis of covariance of variables in SS and Ss genotypes. those previously reported (1,4) and were in Hardy-Weinberg equilibrium (P ⫽ 0.981): SS 65%, Ss 32%, and ss 3%. No statistically significant differences were observed between the SS and Ss genotype groups with regard to BMD at the lumbar spine or hip. The stiffness index scores did not differ between SS and Ss genotypes. No differences in either the marker of collagen formation (PICP) or collagen degradation (ICTP) were documented (Table 1). Our results are consistent with those of some other investigators (2,3) but contradict those of studies (5) that have assumed the Sp1 polymorphism to be an independent marker of increased bone fragility, regardless of BMD. The stiffness index score, calculated from quantitative ultrasound, might reflect bone quality. However, in the present study a significant association between COL1A1 genotypes and stiffness was not demonstrated. Some investigators (1,6) have reported that the impact of COL1A1 genotypes on bone quality might be mediated through impaired collagen metabolism. We failed to confirm a relationship between indices of collagen turnover (PICP, ICTP) and Sp1 genotypes. This study has important limitations. It was conducted with a limited sample of postmenopausal women. Neither osteoporotic fractures nor rate of bone loss was investigated. ANCOVA included only the SS and Ss genotypes. If, as has been postulated, the s allele is associated with an increased risk for reduced BMD and fracture (consistent with allelic dose– response), Ss should represent a partially unfavorable genotype. In the present study, women with the ss genotype had even higher BMD at the lumbar spine than did women with the SS and Ss genotypes, which is not a unique finding (3). In conclusion, the present data do not support the hypothesis that the Sp1 polymorphism might contribute to BMD, bone remodeling, or bone quality in postmenopausal women. Supported by research intention no. 000000023761 from the Ministry of Health, Czech Republic. K. Zajı́c̆ková, MD I. Žofková, MD, PhD, DSc M. Hill, PhD, DSc Institute of Endocrinology Prague, Czech Republic 1. Keen RW, Woodford-Richens KL, Grant SFA, Ralston SH, Lanchbury JS, Spector TD. Association of polymorphism at the type I 2. 3. 4. 5. 6. collagen (COL1A1) locus with reduced bone mineral density, increased fracture risk, and increased collagen turnover. Arthritis Rheum 1999;42:285–90. Lidén M, Wilén B, Ljunghall S, Melhus H. Polymorphism at the Sp 1 binding site in the collagen type I ␣ 1 gene does not predict bone mineral density in postmenopausal women in Sweden. Calcif Tissue Int 1998;63:293–5. Heegaard A, Jorgensen HL, Vestergaard AW, Hassager C, Ralston SH. Lack of influence of collagen type I ␣ 1 Sp1 binding site polymorphism on the rate of bone loss in a cohort of postmenopausal Danish women followed for 18 years. Calcif Tissue Int 2000;66:409–13. Grant SF, Reid DM, Blake G, Herd R, Fogelman I, Ralston SH. Reduced bone density and osteoporosis associated with a polymorphic Sp1 binding site in the collagen type 1 ␣ 1 gene. Nat Genet 1996;14:203–5. Weichetová M, Štěpán JJ, Michalská D, Haas T, Pols HA, Uitterlinden AG. COLIA1 polymorphism contributes to bone mineral density to assess prevalent wrist fractures. Bone 2000;26:287–90. Garnero P, Borel D, Grant SFA, Ralston SH. Sp1 binding site polymorphism in the collagen type I alpha 1 gene, peak bone mass, postmenopausal bone loss and bone turnover: the OFELY study [abstract]. J Bone Miner Res 1997;12:S490. DOI 10.1002/art.10487 Mutations of bone morphogenetic protein receptor type II are not found in patients with pulmonary hypertension and underlying connective tissue diseases To the Editor: It has recently been reported that pleiomorphic heterozygous mutations in the gene encoding bone morphogenetic protein receptor type II (BMPR2) are a cause of familial primary pulmonary hypertension (PPH) (1–4). BMPR2 is located on chromosome 2q33, and the gene product is type II transforming growth factor ␤ (TGF␤) receptor expressed in many tissues, including the lung. The mechanism by which the described mutations lead to abnormalities in the function of this receptor and thus contribute to the development of PPH is unclear, but mutations in BMPR2 are thought to lead to proliferation of pulmonary arterial smooth muscle cells be- 2830 LETTERS cause of ineffective transduction of Smad phosphorylation by the mutant protein (5,6). Pulmonary hypertension is a well-recognized complication of connective tissue diseases such as systemic sclerosis (SSc) and systemic lupus erythematosus (SLE), and the histopathology of the pulmonary vascular lesions is very similar to that seen in PPH (7,8). Therefore, we sought to determine if mutations in BMPR2 were present in patients with pulmonary hypertension and underlying connective tissue disease. Genomic DNA was obtained from 12 patients with connective tissue disease and pulmonary hypertension in the absence of thromboembolic disease or pulmonary fibrosis. Nine of the patients (6 Caucasian women, 1 African American woman, 1 Hispanic woman, and 1 Caucasian man) met the preliminary American College of Rheumatology (ACR; formerly, the American Rheumatism Association) criteria for SSc (9). Two patients (both Hispanic women) met the ACR criteria for SLE (10), and 1 Hispanic woman had been diagnosed with mixed connective tissue disease based on the presence of high-titer anti-RNP antibodies, myositis, Raynaud’s phenomenon, pulmonary hypertension, and the absence of criteria for SSc or SLE. Each patient had documentation of pulmonary hypertension by echocardiographic and/or invasive hemodynamic measurements with normal left ventricle function. Thromboembolism and pulmonary fibrosis as causes of pulmonary hypertension were ruled out by history and physical and radiographic examination. Genomic DNA from 10 healthy volunteers having the same ethnicity as the patients was used as controls. The 13 exons of BMPR2 were amplified and directly sequenced using previously published primers (2). A single-nucleotide polymorphism (SNP) was discovered in exon 12 at position ⫹2811 (G 3 A) coding for a silent mutation, and its presence was confirmed by cloning and resequencing. The Online Mendelian Inheritance in Man (OMIM) database (http://www3.ncbi.nlm.nih.gov/Omim/) lists 17 allelic variants of BMPR2 associated with PPH (OMIM 600799). None of these allelic variants or mutations was found in our pulmonary hypertension patients. Allele frequencies of the SNP in exon 12 were identical among patients and controls. The small size of this cohort is a potential limitation of this study, but based on the previously reported frequencies of BMPR2 mutations in sporadic cases of pulmonary hypertension of 25% (2,6), we would expect 3 members of our cohort to carry BMPR2 mutations. A similar study of a larger group of patients with concomitant connective tissue disease and pulmonary hypertension would be very useful, especially if it were to evaluate the presence of mutations in other TGF␤-related genes. The fact that individuals affected by the same mutation within a family manifest the disease at varying ages indicates that environmental exposures or the interaction of other genes may be important in pathogenesis of the disease (1–4). One possibility is that other proteins belonging to the TGF␤ superfamily pathway may also be important in the pathogenesis of pulmonary hypertension in patients with connective tissue diseases. These genes include endoglin and activin-receptor–like kinase 1, which have been implicated in PPH associated with hereditary hemorrhagic telangiectasias (11). Supported by a grant from the LASR 2000 fund. Dr. Tan’s work was supported by a grant from the NIH (NCRR-3M01-RR02558-1251). Monty B. Tew, MD Frank C. Arnett, MD John D. Reveille, MD Filemon K. Tan, MD, PhD University of Texas Health Science Center Houston, TX 1. Deng Z, Morse JH, Slager SL, Cuervo N, Moore KJ, Venetos G, et al. Familial primary pulmonary hypertension (gene PPH1) is caused by mutations in the bone morphogenetic protein receptor-II gene. Am J Hum Genet 2000;67:737–44. 2. Lane KB, Machado RD, Pauciulo MW, Thomson JR, Phillips JA III, Loyd JE, et al. Heterozygous germline mutations in BMPR2, encoding a TGF-␤ receptor, cause familial primary pulmonary hypertension. Nat Genet 2000;26:81–4. 3. Machado RD, Pauciulo MW, Thomson JR, Lane KB, Morgan NV, Wheeler L, et al. BMPR2 haploinsufficiency as the inherited molecular mechanism for primary pulmonary hypertension. Am J Hum Genet 2001;68:92–102. 4. Newman JH, Wheeler L, Lane KB, Loyd E, Gaddipati R, Phillips JA III, et al. Mutation in the gene for bone morphogenetic protein receptor II as a cause of primary pulmonary hypertension in a large kindred. N Engl J Med 2001;345:319–24. 5. Loscalzo J. Genetic clues to the cause of primary pulmonary hypertension. N Engl J Med 2001;345:367–71. 6. Scott J. Pulling apart pulmonary hypertension. Nat Genet 2000; 26:3–4. 7. Black CM, Dubois RM. Organ involvement: pulmonary. In: Clements PL, Furst DE, editors. Systemic sclerosis. Baltimore: Williams and Wilkins; 1996. p. 299–331. 8. Quismorio FP. Pulmonary manifestations of systemic lupus erythematosus. In: Wallace DJ, Hahn BH, editors. Dubois’ lupus erythematosus. Baltimore: Williams and Wilkins; 1997. p. 673–92. 9. Subcommittee for scleroderma criteria of the American Rheumatism Association diagnostic and therapeutic criteria committee. Preliminary criteria for the classification of systemic sclerosis (scleroderma). Arthritis Rheum 1980;23:581–90. 10. Tan EM, Cohen AS, Fries JF, Masi AT, McShane DJ, Rothfield NF, et al. The 1982 revised criteria for the classification of systemic lupus erythematosus. Arthritis Rheum 1982;25:1271–7. 11. Trembath RC, Thomson JR, Machado RD, Morgan NV, Atkinson C, Winship I, et al. Clinical and molecular genetic features of pulmonary hypertension in patients with hereditary hemorrhagic telangiectasia. N Engl J Med 2001;345:325–34. DOI 10.1002/art.533 Erratum In the article by Leask et al published in the July 2002 issue of Arthritis & Rheumatism (pp 1857–1865), the name of the twelfth author, C. Bernabeu, PhD (Centro de Investigaciones Biologicas, Madrid, Spain) was inadvertently omitted. We regret the error.