Magnetic resonance imaging computerized assessment in rheumatoid arthritisComment on the article by Goldbach-Mansky et al.код для вставкиСкачать
ARTHRITIS & RHEUMATISM Vol. 50, No. 3, March 2004, pp 1008–1018 © 2004, American College of Rheumatology LETTERS extension along the bone diaphysis.” However, in many cases (especially in PIP joints), synovitis begins in the region of the diaphysis and then bulges over to the region of the periarticular bones, which would be the other way around. In most cases, we see a combination of synovitis and effusion. Because these different phenomena appear at the same time, we would prefer to use a scoring system that includes both parameters. The results show that it is possible to clearly differentiate between healthy and pathologic inflammatory findings in small finger and toe joints. However, for evaluating a scoring system, more pathologic joints as well as a healthy control group to distinguish between healthy and pathologic findings should be included in longitudinal studies. DOI 10.1002/art.20202 Ultrasonographic assessment of finger and toe joint inflammation in rheumatoid arthritis: comment on the article by Szkudlarek et al To the Editor: In their interesting study, Szkudlarek et al (1) demonstrated that musculoskeletal ultrasonography of selected small finger and toe joints achieved high interobserver agreement rates for the identification of synovitis and bone erosions. This is an important issue, and we thank the authors for performing this study, especially because one of the main disadvantages of musculoskeletal ultrasound was considered to be operator dependence, which was disproven in this study. However, we think that the study raises some points with regard to preselection of the joints and the scoring systems introduced. The authors decided to evaluate 5 preselected joints (the second and third metacarpophalangeal [MCP] joints, the second proximal interphalangeal [PIP] joint, and the first and second metatarsophalangeal [MTP] joints), because they thought that the accessibility of these joints for assessment with ultrasound was representative of all small joints of the hands and feet. However, the distribution of joint involvement does change in patients with rheumatoid arthritis: the second and third MCP joints are clinically involved more often than are the fourth and fifth MCP joints (2). Also, more erosions have been described in the second and fifth MCP joints compared with the third and fourth MCP joints (3). Furthermore, the accessibility and therefore the interpretation of pathologic findings of the second MCP joint is easier than in other finger joints. Because the patient cohort was rather small and an ultrasonographic examination of all PIP and MCP joints takes about 10–15 minutes (4), we would suggest an examination of all of these joints, which is supplemented by the authors’ statement that the “agreements varied from joint to joint probably reflecting the anatomic localization and availability for examination” (1). Semiquantitative scoring systems are frequently used in radiology (e.g., magnetic resonance imaging evaluation of synovitis and bone lesions) (5). However, we believe the ultrasonographic scoring systems introduced for calculating interobserver agreement may have some limitations. We perform ultrasonographic examinations of finger and toe joints from the dorsal and volar sides and in 2 planes (see ref. 6), but in the report by Szkudlarek et al (1), we did not see a clear description of whether the authors performed the examination according to a standardized protocol. Our experience shows that synovitis and effusion can be detected predominantly from the volar side. It would have been of great interest to see the typical anatomic landmarks (e.g., metacarpal head, tendon, synovial membrane) as well as a statement about the orientation (volar or dorsal view, proximal or distal) and specification (MCP, PIP, or MTP) of the joints in the figures. It therefore remains unclear whether synovitis and effusion appear similar or equal in all 3 joint regions. In the study by Szkudlarek et al, synovitis was classified on a 4-grade scale (0–3), with a score of 1 defined as “filling the angle between the periarticular bones, without bulging over the line linking tops of the bones,” and a score of 2 was defined as “bulging over the . . . periarticular bones but without Alexander K. Scheel, MD Georg-August-University Göttingen, Germany Marina Backhaus, MD Charité University Hospital Berlin, Germany 1. Szkudlarek M, Court-Payen M, Jacobsen S, Klarlund M, Thomsen HS, Østergaard M. Interobserver agreement in ultrasonography of the finger and toe joints in rheumatoid arthritis. Arthritis Rheum 2003;48:955–62. 2. Weidekamm C, Koller M, Weber M, Kainberger F. Diagnostic value of high-resolution B-mode and Doppler sonography for imaging of hand and finger joints in rheumatoid arthritis. Arthritis Rheum 2003;48:325–33. 3. Wakefield RJ, Gibbon WW, Conaghan PG, O’Connor P, McGonagle G, Pease C, et al. The value of sonography in the detection of bone erosions in patients with rheumatoid arthritis: a comparison with conventional radiography. Arthritis Rheum 2000;43:2762–70. 4. Backhaus M, Burmester GR, Sandrock D, Loreck D, Hess D, Scholz A, et al. Prospective two year follow up study comparing novel and conventional imaging procedures in patients with arthritic finger joints. Ann Rheum Dis 2002;61:895–904. 5. Østergaard M, Klarlund M, Lassere M, Conaghan P, Peterfy C, McQueen F, et al. Interreader agreement in the assessment of magnetic resonance images of rheumatoid arthritis wrist and finger joints: an international multicenter study. J Rheumatol 2001;28: 1143–50. 6. Backhaus M, Burmester GR, Gerber T, Grassi W, Machold KP, Swen WA, et al, for the Working Group for Musculoskeletal Ultrasound in the EULAR Standing Committee on International Clinical Studies including Therapeutic Trials. Guidelines for musculoskeletal ultrasound in rheumatology. Ann Rheum Dis 2001;60: 641–9. DOI 10.1002/art.20205 Reply To the Editor: We thank our colleagues, Drs. Scheel and Backhaus, for their interest in and comments on our recent study. We are pleased to be given the opportunity to elaborate on the description of our methodology. Except for our intention to assess finger as well as toe joints, the only criterion for our preselection of joints for examination was their accessibility for ultrasound. In this 1008 LETTERS 1009 respect, the PIP joints are most accessible and can be assessed in 4 aspects (dorsal, palmar, medial, and lateral). The accessibility of the MCP joints varies, with the second and fifth MCP joints being accessible in 3 aspects (dorsal and palmar and, respectively, medial and lateral), while the third and fourth MCP joints are accessible in only 2 aspects (dorsal and palmar). This variability of accessibility is also seen in the MTP joints, where the first MTP joint is accessible in 2 aspects (medial and dorsal, due to the presence of the sesamoid bones on the plantar side of the joint), the fifth MTP joint is accessible in 3 aspects (dorsal, plantar, and lateral), while the second, third, and fourth MTP joints are accessible in 2 aspects (dorsal and plantar). We considered the 5 preselected joints to be representative of the finger and toe joints involved in rheumatoid arthritis, with respect to location, size, and accessibility. Thus, the distribution of joint changes in rheumatoid arthritis had no influence on our selection. However, we think that the differences described by Drs. Scheel and Backhaus suggest that our choice of joints for examination included different frequencies of involvement. The examinations in our study were performed according to a protocol on which we reached consensus before its beginning. The preselected joints were examined from all accessible aspects (see above), in both the longitudinal and transverse aspects, as shown in the figures. The longitudinal aspects were used for scoring synovitis, joint effusion, and power Doppler flow signal, while both longitudinal and transverse aspects were used for scoring the destructive changes. Our experience is that synovitis and joint effusion are most often visualized on the dorsal side of the MCP and MTP joints, but in the PIP joints the findings predominate on the palmar side. Likewise, in our experience, it is most often joint effusion in the PIP joints that can be visualized along the diaphysis. On the basis of a cross-sectional study, we cannot conclude which of the findings is primary. We can only state that in our patient population, the combination of synovitis and joint effusion often occurred in the MTP and PIP joints and seldom occurred in the MCP joints. In order to elucidate patterns of localization of these parameters in future longitudinal studies, we prefer to score them separately. We find our results encouraging for the further use of ultrasound imaging in rheumatoid arthritis and agree with Drs. Scheel and Backhaus that a final evaluation of a scoring system should occur in longitudinal studies, with a higher number of participants, including both patients and healthy controls. Marcin Szkudlarek, MD, PhD Michel Court-Payen, MD, PhD Søren Jacobsen, MD, DMSc Mette Klarlund, MD, PhD University of Copenhagen Hvidovre Hospital Hvidovre, Denmark Henrik S. Thomsen, MD, DMSc University of Copenhagen Herlev Hospital Herlev, Denmark Mikkel Østergaard, MD, PhD, DMSc University of Copenhagen Hvidovre Hospital Hvidovre, Denmark, and University of Copenhagen Herlev Hospital Herlev, Denmark DOI 10.1002/art.20196 The validity and predictive value of magnetic resonance imaging erosions in rheumatoid arthritis: comment on the article by Goldbach-Mansky et al To the Editor: We were pleased to see the editorial by GoldbachMansky et al on the use of magnetic resonance imaging (MRI) in the evaluation of bone damage in rheumatoid arthritis (RA) (1). In particular, the authors highlighted the increased sensitivity of MRI over plain radiographs in detecting bone damage. In addition, they raised some important issues regarding the validity of MRI lesions and their pathophysiologic significance. It is important to realize that the appearance of a particular MRI bone abnormality will depend on the sequences acquired. Most important, the presence or absence of bone edema can complicate interpretation by the reader, so use of a set of sequences that visualizes both trabecular bone loss and edema is important. The Outcome Measures in Rheumatology Clinical Trials (OMERACT) consensus view (2) defined an MRI erosion as a sharply marginated bone lesion with correct juxtaarticular localization and typical signal characteristics visible in 2 planes, with a cortical break seen in at least 1 plane, and also defined signal characteristics. Evidence is accumulating that MRI erosions represent true bone damage and are indeed the predecessors of radiographic erosions. Such evidence includes the following: 1. Based on the above definitions, MRI erosions correlated 100% with sonographically determined cortical breaks in the radial half of the second metacarpophalangeal (MCP) joints, where ultrasound has access equivalent to that of MRI (3). 2. Biopsy of a small number of these lesions has demonstrated necrotic tissue consistent with erosion pathology (4). 3. Miniarthroscopy has confirmed the presence of bone pathology in all examined patients with early or established RA who had MRI bone erosions in their MCP joints (5). 4. Baseline MRI erosions in early RA markedly increase the risk (relative risk ⬃5) of radiographic erosions 1–2 years later (6,7). 5. Recent reports have documented that MRI bone changes have prognostic value in predicting long-term (5–6-year) radiographic outcome, in established (8) as well as early RA (9). In the study of early RA, the baseline MRI status of the wrist was highly predictive of the total modified Sharp score (hands and feet) 6 years later (9). In the study of established RA, 78% of new radiographic erosions were detectable by MRI 1–5 years earlier (8). 6. Even though 1-year followup studies have demonstrated only low numbers of baseline MRI erosions being visualized as radiographic erosions (10–13), longer-duration studies have demonstrated an increased rate (6,8). McQueen et al (6) reported that at 2 years, 1 of 4 MRI abnormalities had progressed to radiographic erosions, while Østergaard et al (8) found that 51% of MRI erosions of the wrist were detected by radiographs 5 years later, and that the presence of 1010 LETTERS a baseline MRI erosion in a radiographically noneroded bone increased the risk of a radiographic erosion at 5 years more than 4-fold. Goldbach-Mansky and colleagues are to be congratulated on their comparison of MRI lesions with computed tomography (CT)–based lesions, because CT, like ultrasound, can determine true cortical breaks. It is to be expected that the MRI erosions will have larger volumes than those of corresponding lesions on CT images if associated bone marrow edema is included in the volumes measured. This differentiation of bone erosion and bone edema is not always easy, particularly if optimal MRI sequences and definitions are not applied. The CT and MRI acquisitions, the definitions of CT and MRI erosions, and the criteria for the exact delineation of the erosions used in the study by Goldbach-Mansky et al were not provided in the editorial or in the short abstracts published so far. Consequently, discussion of these topics must await publication of the original data. It is likely that MRI lesions will never correspond exactly with radiographic lesions, due to the following: a) the tomographic nature of MRI compared with the monoplanar radiograph; b) the sensitivity of MRI: very small defects may be visible on MRI, while radiographs are known to be insensitive to small erosions (3); c) use of modern therapeutic strategies means many lesions will not progress in size, and, consequently, that small baseline MRI erosions will never become detectable on radiographs; and d) strict MRI definitions are not being used (see above). Goldbach-Mansky and colleagues raised the issue of interrater reliability in scoring MRI erosions. Results from the most recent OMERACT 6 interrater reliability exercises, both cross-sectional and longitudinal, were recently published (14,15). These results reflect that with increased training, there is improved agreement across readers from multiple centers, but it is correct that further improvements are desirable. However, the moderate agreement in readings from multiple centers is not reflected in reports of single intrareader or dual interreader agreements between trained readers, which have shown very good agreement rates (intraclass correlation coefficients generally ⬎0.8) (16,17). The field of MRI is moving forward rapidly, and the usefulness of MRI in determining synovial volumes and activity cannot be overstated, given the known relationship of MRI with erosion progression (11,13). With the increasing use of effective disease-suppressant therapy, it is important to have a tool that will sensitively detect bone erosions before radiographic progression is evident. We agree with the authors that at present we would not recommend MRI for use in large clinical trials involving multiple centers and readers, but we do believe that its use in proof-of-concept studies is highly important in providing relevant and timely information about response to therapy. Philip G. Conaghan, MD, FRACP Mikkel Østergaard, MD, PhD Copenhagen University Hospital at Hvidovre Hvidovre, Denmark Dennis McGonagle, MB, FRCPI, PhD Philip O’Connor, MD, FRCR Paul Emery, MA, MD, FRCP Leeds General Infirmary Leeds, UK 1. Goldbach-Mansky R, Woodburn J, Yao L, Lipsky PE. Magnetic resonance imaging in the evaluation of bone damage in rheumatoid arthritis: a more precise image or just a more expensive one? Arthritis Rheum 2003;48:585–9. 2. Conaghan P, Edmonds J, Emery P, Genant H, Gibbon W, Klarlund M, et al. Magnetic resonance imaging in rheumatoid arthritis: summary of OMERACT activities, current status, and plans. J Rheumatol 2001;28:1158–62. 3. Wakefield RJ, Gibbon WW, Conaghan PG, O’Connor P, McGonagle D, Pease C, et al. The value of sonography in the detection of bone erosions in patients with rheumatoid arthritis: a comparison with conventional radiography. Arthritis Rheum 2000; 43:2762–70. 4. McGonagle D, Gibbon W, O’Connor P, Blythe D, Wakefield R, Green M, et al. A preliminary study of ultrasound aspiration of bone erosion in early rheumatoid arthritis. Rheumatology (Oxford) 1999;38:329–31. 5. Ostendorf B, Peters R, Dann P, Becker A, Scherer A, Wedekind F, et al. Magnetic resonance imaging and miniarthroscopy of metacarpophalangeal joints: sensitive detection of morphologic changes in rheumatoid arthritis. Arthritis Rheum 2001;44: 2492–502. 6. McQueen FM, Benton N, Crabbe J, Robinson E, Yeoman S, McLean L, et al. What is the fate of erosions in early rheumatoid arthritis? Tracking individual lesions using x rays and magnetic resonance imaging over the first two years of disease. Ann Rheum Dis 2001;60:859–68. 7. Lindegaard H, Hørslev-Petersen K, Vallø J, Junker P, Østergaard M. Baseline MRI erosions in early rheumatoid arthritis MCP and wrist joint bones markedly increase the risk of radiographic erosions at 1 year follow-up [abstract]. Arthritis Rheum 2002;46 Suppl 9:S521. 8. Østergaard M, Hansen M, Stoltenberg M, Jensen K, Szkudlarek M, Pedersen-Zbinden B, et al. New radiographic bone erosions in the wrists of patients with rheumatoid arthritis are detectable with magnetic resonance imaging a median of two years earlier. Arthritis Rheum 2003;48:2128–31. 9. Benton NM, Perry D, Crabbe J, Stewart N, Robinson E, Yeoman S, et al. Magnetic resonance scanning at first presentation predicts modified Sharp score at six years in patients with rheumatoid arthritis [abstract]. Arthritis Rheum 2002;46 Suppl 9:S370. 10. McQueen FM, Stewart N, Crabbe J, Robinson E, Yeoman S, Tan PL, et al. Magnetic resonance imaging of the wrist in early rheumatoid arthritis reveals progression of erosions despite clinical improvement. Ann Rheum Dis 1999;58:156–63. 11. Østergaard M, Hansen M, Stoltenberg M, Gideon P, Klarlund M, Jensen KE, et al. Magnetic resonance imaging–determined synovial membrane volume as a marker of disease activity and a predictor of progressive joint destruction in the wrists of patients with rheumatoid arthritis. Arthritis Rheum 1999;42:918–29. 12. Klarlund M, Østergaard M, Jensen KE, Madsen JL, Skjødt H, Lorenzen I. Magnetic resonance imaging, radiography, and scintigraphy of the finger joints: one year follow up of patients with early arthritis. The TIRA group. Ann Rheum Dis 2000;59:521–8. 13. Conaghan PG, O’Connor P, McGonagle D, Astin P, Wakefield RJ, Gibbon WW, et al. Elucidation of the relationship between synovitis and bone damage: a randomized magnetic resonance imaging study of individual joints in patients with early rheumatoid arthritis. Arthritis Rheum 2003;48:64–71. 14. Lassere M, McQueen F, Østergaard M, Conaghan P, Shnier R, Peterfy C, et al. OMERACT Rheumatoid Arthritis Magnetic Resonance Imaging Studies. Exercise 3: an international multicenter reliability study using the RA-MRI Score. J Rheumatol 2003;30:1366–75. 15. Conaghan P, Lassere M, Østergaard M, Peterfy C, McQueen F, O’Connor P, et al. OMERACT Rheumatoid Arthritis Magnetic Resonance Imaging Studies. Exercise 4: an international multicenter longitudinal study using the RA-MRI Score. J Rheumatol 2003;30:1376–9. LETTERS 16. Bird P, Lassere M, Shnier R, Edmonds J. Computerized measurement of magnetic resonance imaging erosion volumes in patients with rheumatoid arthritis: a comparison with existing magnetic resonance imaging scoring systems and standard clinical outcome measures. Arthritis Rheum 2003;48:614–24. 17. Østergaard M, O’Connor P, Conaghan P, Ejbjerg B, Szkudlarek M, Peterfy C, et al. Is intravenous contrast injection necessary for MRI assessment of inflammatory and destructive changes in rheumatoid arthritis wrist and MCP joints? European Surgical Research 2002;34 Suppl 1:141. DOI 10.1002/art.20197 Magnetic resonance imaging computerized assessment in rheumatoid arthritis: comment on the article by Goldbach-Mansky et al To the Editor: We believe it is important to clarify 3 issues arising from the recent editorial by Goldbach-Mansky et al (1). These relate to computer segmentation processes, validity, and interrater reliability. Goldbach-Mansky et al stated that the live wire paradigm developed by the National Institute of Health has been shown to be “more repeatable and more efficient than manual tracking methods , including the method used by Bird et al.” The reference provided to support this assertion (3) describes the measurement of magnetic resonance imaging (MRI) bone volumes of the talus and the calcaneus using 30 2-dimensional sequences from an existing data set. The live wire technique was compared with a manual tracing method using 3 observers, and the authors reported that the live wire method was faster and more repeatable. It appears, however, that the magnitude of difference between the 2 segmentation techniques, although statistically significant, is not substantial in real terms. Furthermore, the segmentation processes have been examined in large structures that do not approximate the degree of difficulty encountered when outlining small bony erosions in the wrist. Therefore, it is simply not the case that the live wire method is more repeatable and faster than a manual method, and, more important, it does not appear that the method has been directly compared with the erosion outlining method used in our study. Second, we agree that the validation of MRI is an important step in the development of MRI as an outcome measure, but this should be explored cautiously. The authors present their own experiment, which compares MRI with computed tomography (CT) in the assessment of erosion volumes in 4 patients with rheumatoid arthrits (RA) (4). Because the data are currently available only in abstract form, it is difficult to examine the details. However, the results of the study demonstrate that there were substantial differences between the size of erosion volumes when comparing MRI and CT. The authors concluded that MRI overestimates the size of erosions in patients with RA, because CT is “considered to be the best way to assess bony changes.” Such a conclusion does not acknowledge the fact that this type of construct validity assumes a surrogate gold standard only, and therefore the only conclusion that can be reached is that there is or is not a correlation between the 2 methods. Additionally, the interrater reliability of both MRI and CT was poor, with the total number 1011 of erosions scored on CT varying between the observers by nearly 2-fold. Therefore, the study effectively illustrates that although validation should remain a firm goal for MRI researchers, it is mandatory that we have confidence in the reliability of the measurement instrument and the measurement technique before embarking upon validation studies. Interrater reliability is an essential part of any measurement process, and the Outcome Measures in Rheumatology Clinical Trials (OMERACT) MRI-RA group has examined this as a central issue over the past 6 years. GoldbachMansky et al reported that the recent OMERACT exercise (5) used different scoring systems, but this is erroneous. The OMERACT exercise referred to used a standarized scoring system for all readers, estimating erosion size in the metacarpophalangeal joints and the carpus. The development of the OMERACT MRI-RA scoring system is an iterative process, and, as such, it represents an international grading system that will continue to be modified and improved over time. In the same way, the erosion volume method has been the subject of an international trial of interrater reliability (6) and also is the subject of an ongoing study to assess the effect of training and calibration in improving interrater agreement. Our study (2) sought to examine the feasibility and reliability of the erosion volume method, using a software program that is freely available on the internet and is accessible to all research groups. By doing this, we attempted to keep the process of development transparent so that the development of MRI computerized assessment in RA remains a tangible goal rather than a means to its own end. MRI measurement in RA is still in its infancy, and it is only continuing peer review and assessment that will develop such measurement to its full potential. Paul Bird, BMed (Hons), FRACP Marissa Lassere, MBBS (Hons), PhD, FRACP, FAPHM Ron Shnier, MBBS, FACR John Edmonds, MBBS, MA, FRACP St. George Hospital and University of New South Wales Sydney, New South Wales, Australia 1. Goldbach-Mansky R, Woodburn J, Yao L, Lipsky PE. Magnetic resonance imaging in the evaluation of bone damage in rheumatoid arthritis: a more precise image or just a more expensive one? Arthritis Rheum 2003;48:585–9. 2. Bird P, Lassere M, Shnier R, Edmonds J. Computerized measurement of magnetic resonance imaging erosion volumes in patients with rheumatoid arthritis: a comparison with existing magnetic resonance imaging scoring systems and standard clinical outcome measures. Arthritis Rheum 2003;48:614–24. 3. Falcao AX, Udupa JK, Samarasekera S, Sharma S. User-steered image segmentation paradigms: live wire and live lane. Graph Model Im Proc 1998;60:233–60. 4. Bedair H, Murphy M, Fleming D, Hill S, Schaub C, Thornton B, et al. A comparison of MRI and CT in detecting carpal bone erosions in early rheumatoid arthritis [abstract]. Arthritis Rheum 2001;41 Suppl 9:S222. 5. Østergaard M, Klarlund M, Lassere M, Conaghan P, Peterfy C, McQueen F, et al. Interreader agreement in the assessment of magnetic resonance images of rheumatoid arthritis wrist and finger joints: an international multicenter study. J Rheumatol 2001;28: 1143–50. 6. Bird P, Ejbjerg B, McQueen F, Østergaard M, Lassere M, Edmonds J. OMERACT Rheumatoid Arthritis Magnetic Resonance Imaging 1012 Studies. Exercise 5: an international multicenter reliability study using computerized MRI erosion volume measurements. J Rheumatol 2003;30:1380–4. DOI 10.1002/art.20122 What are the real effects of arthritis self-management education programs on pain and disability? Comment on the article by Warsi et al To the Editor: We read with interest the report by Warsi et al (1) regarding the meta-analysis of the effects of arthritis selfmanagement education programs on pain and disability. Metaanalysis is a powerful tool for bringing together the results of different studies, but meta-analysis itself must be undertaken rigorously if criticism is to be avoided (2). Most important, such an analysis should be performed within the framework of a systematic review of the literature, to avoid bias and ensure appropriate combinability of studies. We have undertaken such a systematic review for the Cochrane Collaboration (3,4), examining the effects of patient education for adults with rheumatoid arthritis (RA) on health outcomes (pain, disability, psychological well-being, disease activity). Our review was restricted to randomized controlled trials (RCTs) of patient education interventions in which patients with a confirmed diagnosis of RA participated. We included all types of patient education programs, not only programs involving selfmanagement education as did Warsi and colleagues. There are several reasons why the meta-analysis by Warsi et al will be criticized and why its value is limited. First, the search strategy used was not comprehensive, because it omitted studies published in the last 5 years, considered only English-language publications, and searched only in Medline, HealthSTAR, and the reference list of retrieved articles. The Cochrane Collaboration advises investigators to search at least in the electronic databases Medline and Embase, and in the Cochrane Controlled Trials Register (5). (The overlap in journals listed in Medline and Embase is only ⬃34%.) Warsi et al did not search PsycINFO, the most comprehensive database of citations to psychosocial studies, and did not try to locate unpublished studies. By relying mainly on Medline, it can be expected that only 30–80% of all relevant studies will be identified (6). In our review, we searched Medline, Embase, PsycINFO, and the Cochrane Controlled Trials Register, from 1966 to September 2002 and in all languages, and we searched for unpublished studies. When publications provided incomplete data, we contacted the authors for more information. As a result, 50 studies were identified (4), including at least 18 randomized clinical trials that were not included by Warsi et al. Those trials dealt with patient education programs for patients with RA that contained a self-management education component. Six of the articles had been published before October 15, 1998 (7–12). Warsi et al did not assess the methodologic quality of studies included in their review, yet quality assessment of individual studies is necessary to limit bias in conducting the systematic review, gain insight into potential comparisons, and LETTERS guide the interpretation of findings (5). The Cochrane review used 2 independent assessors of methodologic quality to evaluate 4 criteria: selection bias, attrition bias, detection bias, and performance bias (4). There is great variation among the studies included in the review by Warsi et al with regard to research design (nonrandomized studies were included), types of interventions, types of disease (RA, osteoarthritis, polyarthritis, fibromyalgia), assessment periods, and assessment instruments. The authors therefore used a random-effects model in their statistical analyses and performed a subgroup analysis on interventions that closely resembled the Arthritis Self-Help Course (taught through chapters of the Arthritis Foundation). However, they did not analyze other differences between studies in relation to outcomes. The Cochrane review presents separate analyses for 3 types of interventions: information only, counseling, and behavioral treatment (mainly self-management programs). Furthermore, extensive sensitivity analyses were performed using only studies with high scores for methodologic quality, only larger studies, studies using the same instrument to assess each outcome, and studies that assessed outcomes at a fixed time point (after 2–4 months). Warsi et al concluded that arthritis self-management education programs lead to small but significant reductions in pain and disability. However, the 95% confidence intervals (95% CIs) for the effect sizes for both pain and disability included 0, meaning the effects were not significant. The results of our more comprehensive review of RCTs of patient education programs for people with RA showed similar overall results. However, in the subanalysis of educational interventions that included techniques aimed at behavioral change (mostly self-management programs) we found, at first followup, a significant beneficial effect of such interventions on disability (standardized mean difference ⫽ ⫺0.23, 95% CI ⫺0.36, ⫺0.10). This effect was quite robust, as shown by sensitivity analyses, but the benefit was not maintained after longer followup. We support Warsi et al in their call for independent high-quality trials of patient education (some of which have been included in the Cochrane review), and we believe that future research should seek to identify which patient characteristics (including the diagnostic category) are relevant to beneficial outcomes, and which components of patient education programs are effective. Erik Taal, PhD Robert P. Riemsma, PhD John R. Kirwan, MD Johannes J. Rasker, MD University of Twente Enschede, The Netherlands 1. Warsi A, LaValley P, Wang PS, Avorn J, Solomon DH. Arthritis self-management education programs: a meta-analysis of the effect on pain and disability. Arthritis Rheum 2003;48:2207–13. 2. Egger M, Smith GD, Sterne JA. Uses and abuses of meta-analysis. Clin Med 2001;1:478–84. 3. Riemsma RP, Taal E, Kirwan JR, Rasker JJ. Patient education programmes for adults with rheumatoid arthritis. BMJ 2002;325: 558–9. 4. Riemsma RP, Kirwan JR, Taal E, Rasker JJ. Patient education for LETTERS 5. 6. 7. 8. 9. 10. 11. 12. adults with rheumatoid arthritis. Cochrane Database Syst Rev 2003;2:CD003688. Clarke M, Oxman AD, editors. Cochrane reviewers’ handbook 4.2.0 [updated March 2003]. In: The Cochrane Library: Oxford: Update Software. Updated quarterly; 2003. Available at: http:// www.cochrane.org/resources/handbook. Dickersin K, Scherer R, Lefebvre C. Identifying relevant studies for systematic reviews. BMJ 1994;309:1286–91. Brus HL, van de Laar MA, Taal E, Rasker JJ, Wiegman O. Effects of patient education on compliance with basic treatment regimens and health in recent onset active rheumatoid arthritis. Ann Rheum Dis 1998;57:146–51. Goeppinger J, Arthur MW, Baglioni AJ Jr, Brunk SE, Brunner CM. A reexamination of the effectiveness of self-care education for persons with arthritis. Arthritis Rheum 1989;32:706–16. Kraaimaat FW, Brons MR, Geenen R, Bijlsma JW. The effect of cognitive behavior therapy in patients with rheumatoid arthritis. Behav Res Ther 1995;33:487–95. Lindroth Y, Brattstrom M, Bellman I, Ekestaf G, Olofsson Y, Strombeck B, et al. A problem-based education program for patients with rheumatoid arthritis: evaluation after three and twelve months. Arthritis Care Res 1997;10:325–32. Parker JC, Smarr KL, Buckelew SP, Stucky-Ropp RC, Hewett JE, Johnson JC, et al. Effects of stress management on clinical outcomes in rheumatoid arthritis. Arthritis Rheum 1995;38: 1807–18. Taal E, Riemsma RP, Brus HL, Seydel ER, Rasker JJ, Wiegman O. Group education for patients with rheumatoid arthritis. Patient Educ Couns 1993;20:177–87. 1013 quality scores in the meta-analysis. We assessed the methods of each study and outlined these in table format. However, we elected to not explicitly include a score in the analysis because of the lack of any standard method for scoring the quality of such literature. Other authors have shown that the results of meta-analyses including such quality scores are very sensitive to the scoring system chosen (Juni P, Witschi A, Bloch R, Egger M. The hazards of scoring the quality of clinical trials for meta-analysis. JAMA 1999;282:1054–60). We found it encouraging that the results of the review by Taal et al on programs for RA and the results of our review were similar—that the long-term effects of self-management education are not significant. Taal and colleagues also examined the results of these programs at first followup and found significant improvement. However, these short-term improvements did not persist. Because RA and osteoarthritis are chronic diseases, optimal self-management education programs would require the demonstration of a long-term benefit. Unfortunately, as a group, the current programs have not demonstrated such an effect. Further research efforts need to focus on increasing the persistence of benefit for arthritis self-management education programs, determining subgroups of patients that are most likely to benefit, and improving the methods for conducting and reporting the results of such trials. Daniel H. Solomon, MD, MPH Michael P. LaValley, PhD Philip S. Wang, MD, ScD Jerry Avorn, MD Brigham and Women’s Hospital Harvard Medical School Boston, MA DOI 10.1002/art.20207 Reply To the Editor: We appreciate the comments by Dr. Taal and colleagues regarding our review article on arthritis selfmanagement education programs. They recently published a similar review on education for patients with RA (Riemsma RP, Kirwan JR, Taal E, Rasker JJ. Patient education for adults with rheumatoid arthritis. Cochrane Database Syst Rev 2003; 2:CD003688). Taal and associates express concern regarding the results of our search strategy. Although it is true that the search strategy they used yielded several articles that we had not found, the majority of nonoverlapping references either were published after 1998, the end of our search period, or include interventions that we did not consider primarily educational. We required that education be a major focus of the intervention programs, to limit the heterogeneity of the included studies. Thus, several articles were excluded that described interventions primarily involving innovations in physical therapy, occupational therapy, or psychological counseling programs. Taal et al also suggest “that only 30–80% of all relevant studies will be identified” by relying on Medline; however, more recent research suggests that the proportion has improved substantially, and that the risk of bias from using only Medline in a meta-analysis is small (Sampson M, Barrowman NJ, Moher D, Klassen TP, Pham B, Platt R, et al. Should meta-analysts search Embase in addition to Medline? J Clin Epidemiol 2003;56:943–55). Another criticism concerns our decision not to include DOI 10.1002/art.20208 Association of the proinflammatory haplotype (MICA5.1/TNF2/TNFa2/DRB1*03) with polymyositis and dermatomyositis To the Editor: Polymyositis (PM) and dermatomyositis (DM) are systemic inflammatory connective tissue disorders that likely result from interactions between genetic and environmental risk factors. Earlier studies indicated that certain HLA class II alleles, including HLA–DRB1*03 (HLA–DR3) in Caucasian populations (1) and HLA–DRB1*14 in Korean patients (2), confer risk for development of PM and DM. However, it is not clear whether this is a primary association or an association due to other genes in the HLA region. Because of linkage disequilibrium, the markers in the HLA region may be important not alone, but in the context of common haplotypes. The 8.1 ancestral haplotype (HLA–A1;B8;DRB1*03) includes the TNF2 allele of the TNFA gene (the conventional name for the G-308 TNFA allele is TNF1 and that for the A-308 TNFA allele is TNF2). This haplotype has been associated with high in vitro production of tumor necrosis factor (TNF) by peripheral blood mononuclear cells and also with high circulating serum levels of TNF, and was thus considered a proinflamma- 1014 LETTERS Table 1. TNFA ⫺308 alleles, genotypes, and haplotype frequency in PM/DM patients compared with controls* TNF1 TNF2 TNF1/TNF1 TNF1/TNF2 TNF2/TNF2 MICA5.1/TNF2/TNFa2/DRB1*03 PM/DM (n ⫽ 65) Controls (n ⫽ 65) P Pcorr OR 95% CI 60 (92) 38 (58) 27 (42) 33 (51) 5 (8) 34 (26) 61 (94) 22 (34) 43 (66) 18 (28) 4 (6) 12 (9) NS 0.0081 0.0081 0.0116 NS 0.0005 – 0.016 0.024 0.0348 – 0.030 – 2.75 0.36 2.69 – 3.48 – 1.35–5.61 0.18–0.74 1.30–5.58 – 1.71–7.09 * Values are the number (%) of patients (or, for MICA5.1/TNF2/TNFa2/DRB1*03, the number [%] of 130 chromosomes). PM/DM ⫽ polymyositis/dermatomyositis; Pcorr ⫽ corrected P (corrected for 2 alleles [TNF2], 3 genotypes [TNF1/TNF1 and TNF1/TNF2], or 61 haplotypes [MICA5.1/TNF2/TNFa2/DRB1*03]); OR ⫽ odds ratio; 95% CI ⫽ 95% confidence interval; TNF1 ⫽ G-308 TNFA; NS ⫽ not significant; TNF2 ⫽ A-308 TNFA. tory haplotype (3,4). In patients with juvenile DM the TNF2 allele of the TNFA gene has been found to be associated with increased in vitro production of TNF by peripheral blood mononuclear cells and with a more severe disease course (5). The present study was conducted to analyze associations between individual genotypes from the HLA region (single-nucleotide polymorphisms [SNPs] in HLA–DRB1, and TNFA genes, and microsatellite markers in TNF and MICA genes), as well as various haplotypes including these genetic markers, with PM and DM in an adult population. Sixty-five patients who fulfilled classification criteria for PM (n ⫽ 36) or DM (n ⫽ 29) (6) were studied. Data from healthy unrelated controls, who were matched to the patients by age, sex, and race, were included for comparison. All patients and controls were genotyped for HLA–DRB1 and TNFA ⫺308 SNPs and for TNF and MICA microsatellites. All subjects gave their informed consent to participate in the study. The protocol was approved by the ethics committee at Karolinska Hospital. We analyzed 4 markers within this HLA region, which spread over 1.2 Mbp on the short arm of chromosome 6. The frequencies of the TNF2 allele of the TNFA ⫺308 gene (Table 1), the TNFa2 and DRB1*03 alleles, as well as MICA5.1-5.1 genotype were all significantly higher in patients with PM/DM compared with controls. Our analysis confirmed earlier findings that the HLA–DR3 genotype is associated with PM/DM in Caucasians (2). We also demonstrated that the TNF2 allele of the TNFA gene was associated with PM and DM in adult patients, consistent with previous findings in juvenile DM (5). Our investigation also revealed that the frequency of the MICA5.1 allele in a homozygous state (MICA5.1-5.1 genotype) was significantly increased in adult PM and DM patients compared with controls, although correction for multiple comparisons weakened this finding. There was no difference between PM and DM patients regarding the frequencies of any investigated markers, and there was no effect of sex on the frequencies of the investigated markers. Using the expectation-maximization algorithm and correcting the P value for the maximum number of observed haplotypes, we determined that among 61 reconstructed haplotypes, 1 particular haplotype with 4 markers (MICA5.1/ TNF2/TNFa2/DRB1*03) was significantly more frequent in the PM and DM patients compared with the controls (52% and 18%, respectively, P ⬍ 0.05). This haplotype was identified in 34 PM/DM patients and in 12 controls (Table 1). The 8.1 ancestral haplotype (HLA–A1;B8;DRB1*03), together with the TNF2 allele, is common in Caucasian populations and was previously demonstrated to confer susceptibility to autoimmune diseases such as systemic lupus erythematosus and Sjögren’s syndrome (4). Extending this ancestral proinflammatory haplotype with MICA5.1 and TNFa microsatellite markers resulted in a haplotype with 4 markers (MICA5.1/TNF2/TNFa2/DRB1*03), which was found to be significantly more frequent in PM and DM patients compared with controls. Additional genetic factors as well as environmental factors are likely involved in the development of myositis, since some patients lack some of the analyzed alleles of the proinflammatory haplotype. In conclusion, our observations suggest that the ancestral haplotype (A1;B8;DRB1*03) together with the TNF2 allele, rather than the HLA–DR3 gene itself, is an important susceptibility factor for the development of polymyositis and dermatomyositis. Our data also demonstrate that the TNFa2 microsatellite and MICA5.1 alleles should possibly be included in the extended 8.1 ancestral haplotype. Supported by grants from the Swedish Rheumatism Association, the King Gustaf V 80-Year Foundation, the Swedish Research Council (2001-74-X14045), the Vårdal Foundation, the Professor Nanna Svartz Foundation, Magnus Bergvalls stiftelse, Stiftelsen Clas Groschinskys Minnesfond, and the Karolinska Institutet Foundation. Adla B. Hassan, MD, PhD Liene Nikitina-Zake, MD Carani B. Sanjeevi, MD, MSc, PhD Ingrid E. Lundberg, MD, PhD Karolinska Institutet Stockholm, Sweden Leonid Padyukov, MD, PhD Karolinska Institutet Stockholm, Sweden and Mechnikov Research Institute for Vaccines and Sera Moscow, Russia 1. Hausmanowa-Petrusewicz I, Kowalska-Oledzka E, Miller FW, Jarzabek-Chorzelska M, Targoff IN, Blaszczyk-Kostanecka M, et al. Clinical, serologic, and immunogenetic features in Polish patients with idiopathic inflammatory myopathies. Arthritis Rheum 1997;40: 1257–66. 2. Rider LG, Shamim E, Okada S, Pandey JP, Targoff IN, O’Hanlon TP, et al. Genetic risk and protective factors for idiopathic inflammatory myopathy in Koreans and American whites: a tale of two loci. Arthritis Rheum 1999;42:1285–90. LETTERS 3. Lio D, Candore G, Colombo A, Colonna Romano G, Gervasi F, Marina V, et al. A genetically determined high setting of TNF-alpha influences immunologic parameters of HLA-B8,DR3 positive subjects: implications for autoimmunity. Hum Immunol 2001;62: 705–13. 4. Degli-Esposti MA, Leaver AL, Christiansen FT, Witt CS, Abraham LJ, Dawkins RL. Ancestral haplotypes reveal the role of the central MHC in the immunogenetics of IDDM. Immunogenetics 1992;36: 345–56. 5. Pachman LM, Liotta-Davis MR, Hong DK, Kinsella TR, Mendez EP, Kinder JM, et al. TNF␣-308A allele in juvenile dermatomyositis: association with increased production of tumor necrosis factor ␣, disease duration, and pathologic calcifications. Arthritis Rheum 2000;43:2368–77. 6. Bohan A, Peter JB. Polymyositis and dermatomyositis (first of two parts). N Engl J Med 1975;292:344–7. DOI 10.1002/art.20097 Effect of infliximab treatment on T cell cytokine responses in spondylarthropathy: comment on the article by Zou et al To the Editor: We read with interest the article by Zou et al (1) describing the down-regulation of T cell cytokines by infliximab treatment in patients with ankylosing spondylitis (AS). Intracellular flow cytometric analysis of tumor necrosis factor ␣ (TNF␣) and interferon-␥ (IFN␥) in CD4⫹ and CD8⫹ peripheral blood lymphocytes after nonspecific (phorbol myristate acetate [PMA]/ionomycin) and antigen-specific (peptides derived from aggrecan) stimulation indicated that infliximab down-regulates the production of both Th1 cytokines for at least 6 weeks. These results and their interpretation are intriguing in a number of respects. First, there was a large difference in the baseline data (before infliximab treatment) between the treated group in Zou and colleagues’ recent study (16.5%, 35.7%, 18.4%, and 23.8% IFN␥⫹ CD4⫹, IFN␥⫹ CD8⫹, TNF␣⫹ CD4⫹, and TNF␣⫹ CD8⫹ cells, respectively), the placebo group in that study (13.0%, 23.7%, 9.0%, and 8.6%, respectively) and AS patients in a previous similar study by the same authors (8.7%, 24.9%, 5.1%, and 2.7%, respectively) (2). This variability raises questions concerning the validity of the technical approach and hampers correct interpretation. Second, the findings after infliximab treatment contrast with those described in 2 previous reports. In the first, we analyzed, by the same technology, IFN␥ and interleukin-2 (IL-2) levels in 20 infliximab-treated spondylarthropathy (SpA) patients and demonstrated restoration of a normal Th1 profile (3). Differences between that study and the one by Zou et al (1) were that we used freshly isolated lymphocytes whereas Zou et al used lymphocytes after freezing/thawing, and that we analyzed different SpA subtypes whereas Zou and colleagues analyzed only AS. The second report (4), by the same group of authors as in the current report by Zou et al (1), describes a study in which the investigators used the same approach as in their recent study (freezing/thawing; AS patients only). In that study the results were similar to ours, i.e., TNF␣ and IFN␥ responses increased after 2 weeks of inflix- 1015 imab treatment. This is intriguing since 2 similar studies performed by the same group led to opposite conclusions. In order to address the discrepancies with our study (3), we performed additional experiments. First, we assessed the variability of intracellular cytokine detection on frozen peripheral blood mononuclear cells (PBMCs) since Zou et al indicate that they “cannot exclude the possibility that the freezing/thawing process altered the cytokine expression by T cells” (1). Intraassay variability after freezing/thawing was assessed by analyzing IFN␥ in 24 samples in duplicate: the mean variance was 0.52% (SD 0.43%) for CD4⫹ cells and 0.65% (SD 0.52%) for CD8⫹ cells. Interassay variability was assessed by analyzing 5 samples for IFN␥, IL-2, and TNF␣ after thawing, at 2 different time points: the mean variance was 2.63% (SD 1.89%) and 3.35 (SD 3.73%) for CD4⫹ and CD8⫹ T cells, respectively. The variability due to freezing/thawing was assessed by analyzing 5 samples for IFN␥, IL-2, and TNF␣ production by paired freshly isolated and frozen PBMCs: freezing/thawing induced a clear increase of IFN␥ (from 15.7% to 23.4% in CD4⫹ cells, from 25.1% to 34.3% in CD8⫹ cells; P ⫽ 0.007), a decrease of IL-2 (from 48.0% to 36.6% in CD4⫹ cells, from 9.1% to 4.8% in CD8⫹ cells; P ⫽ 0.022) (Figure 1A), and a nonsignificant increase of TNF␣ (from 53.1% to 57.3% in CD4⫹ cells, from 19.4% to 23.2% in CD8⫹ cells; P ⫽ 0.203). Thus, although intra- and interassay variability is small, freezing/thawing of the samples induces a systematic bias. Second, considering the above-mentioned differences between studies, we performed a new study of IFN␥, IL-2, and TNF␣ production by frozen CD4⫹ and CD8⫹ T cells stimulated with PMA/ionomycin at baseline and at week 6 of infliximab treatment in 7 HLA–B27⫹ AS patients, i.e., we used exactly the same patient population, treatment, sample collection and preparation, and analysis as did Zou et al (1). In CD4⫹ cells there was a nonsignificant decrease in IFN␥ production (15.2% at baseline versus 10.8% at week 6; P ⫽ 0.063) and a decrease in IL-2 production (36.2% and 31.9% at baseline and week 6, respectively; P ⫽ 0.028) (Figure 1B), but no difference in TNF␣ production (50.9% and 49.0%, respectively; P ⫽ 0.499). In CD8⫹ lymphocytes, infliximab had no effect on IFN␥ (34.3% and 34.1%, respectively; P ⫽ 0.398), IL-2 (9.1% and 11.3%, respectively; P ⫽ 0.866) (Figure 1B), or TNF␣ (18.78% and 24.5%, respectively; P ⫽ 0.866). Third, in order to exclude a bias caused by possible differences between SpA subgroups, we reanalyzed our data focusing on the 9 AS patients included in our original study. Confirming our previous results in SpA patients, the analysis in AS patients at baseline and at week 6 of infliximab treatment showed an increase in IFN␥ levels in CD4⫹ T cells (13.1% and 18.9%, respectively; P ⫽ 0.066) and CD8⫹ T cells (33.6% and 48.1%, respectively; P ⫽ 0.038) and an increase in IL-2 in CD4⫹ cells (22.0% and 39.1%, respectively; P ⫽ 0.008) and CD8⫹ cells (9.0% and 15.5%, respectively; P ⫽ 0.008) (Figure 1C). Although TNF␣ was not measured, this study shows no systematic difference in the Th1/Th2 profile between AS and other SpA subtypes. In summary, our results indicate that freezing/thawing as done by Zou and colleagues (1) induces a systematic bias in T cell cytokine profiles after PMA/ionomycin stimulation. Using frozen samples in a protocol completely identical to that used by those authors, we were unable to confirm the downregulation of IFN␥ and TNF␣ production by T lymphocytes of 1016 LETTERS Figure 1. Intracellular T lymphocyte cytokine levels analyzed by 4-color flow cytometry after phorbol myristate acetate/ionomycin stimulation. Results are expressed as the percentage positive cells. A, Effect of freezing/thawing (fresh versus frozen lymphocytes from the same samples) on interferon-␥ (IFN␥) and interleukin-2 (IL-2) production by CD4⫹ and CD8⫹ T lymphocytes from 5 healthy controls. B, Effect of infliximab treatment (week 0 versus week 6) on IFN␥ and IL-2 production by CD4⫹ and CD8⫹ T cells from 7 ankylosing spondylitis (AS) patients, after freezing/thawing. C, Effect of infliximab treatment (week 0 versus week 6) on IFN␥ and IL-2 production by freshly isolated CD4⫹ and CD8⫹ T cells from 9 AS patients. AS patients during infliximab treatment. In contrast, data from studies of freshly isolated lymphocytes confirm that infliximab restores the production of Th1 cytokines in HLA–B27⫹ AS patients, as was previously described for different subtypes of SpA (3). Dr. Baeten is a Senior Clinical Investigator of the Fund for Scientific Research–Vlaanderen. Dominique Baeten, MD, PhD Bernard Vandooren, MD Leen De Rycke, MD Eric M. Veys, MD, PhD Filip De Keyser, MD, PhD Ghent University Hospital Ghent, Belgium 1. Zou J, Rudwaleit M, Brandt J, Thiel A, Braun J, Sieper J. Down-regulation of the nonspecific and antigen-specific T cell cytokine response in ankylosing spondylitis during treatment with infliximab. Arthritis Rheum 2003;48:780–9. 2. Rudwaleit M, Siegert S, Yin Z, Eick J, Thiel A, Radbruch A, et al. Low T cell production of TNF␣ and IFN␥ in ankylosing spondylitis: its relation to HLA-B27 and influence of the TNF-308 gene polymorphism. Ann Rheum Dis 2001;60:36–42. 3. Baeten D, van Damme N, van den Bosch N, Kruithof E, De Vos M, Mielants H, et al. Impaired Th1 cytokine production in spondyloarthropathy is restored by anti-TNF␣. Ann Rheum Dis 2001;60: 750–5. 4. Braun J, Xiang J, Brandt J, Maetzel H, Hainel H, Wu P, et al. Treatment of spondyloarthropathies with antibodies against tumour necrosis factor ␣: first clinical and laboratory experiences. Ann Rheum Dis 2000;59 Suppl 1:i85–9. LETTERS DOI 10.1002/art.20209 Reply To the Editor: In their letter discussing our report, Baeten et al raise some important questions about the investigation of intracellular T cell cytokine staining in general and about the specific results presented by us. We would like to address these questions here. Unlike measuring soluble cytokines by enzyme-linked immunosorbent assay, standardization of the flow cytometry method is difficult. There are many variables such as the medium used for short-term culture of the T cells, the anticytokine antibodies and the type of fluorescence dye coupled to these antibodies, and the experience and skill of the person who performs the experiments. These make it impossible, in our opinion, to compare baseline levels of cytokine-positive cells in experiments performed by different investigators and at different time points. Because there is always the risk that the experimental conditions vary from one time point to another, we decided in our study to use frozen cells obtained from patients before and during infliximab treatment. These cells were thawed and investigated by the same person on the same day. We believe this affords the best conditions to assess any change in the potential of T cells to secrete cytokines during a treatment trial. We agree that there is a difference between T cell responses using fresh cells versus frozen/thawed cells, but this difference is stable. However, in contrast to the examples presented by Baeten et al, we have always found downregulation of T cell function when comparing frozen T cells with fresh ones. We indeed reported in a previous publication that we observed an increase in the number of IFN␥- and TNF␣positive T cells during infliximab therapy in AS patients (1). However, these were very early results, reported in preliminary form, from an investigation of individual patients in an overview of the first clinical and laboratory experiences in treating AS patients with infliximab. The results reported from that study were obtained after 1 or 2 weeks of treatment. In our more recent study we did not perform assessments at those early time points, but only after 6 weeks and 12 weeks of treatment. Thus, as stated also in our recent publication, we cannot comment on cytokine production by T cells in the first days after inflixmab infusion. The preliminary results prompted us to perform 2 systematic studies in AS patients treated with either infliximab or etanercept. In those studies, T cells from AS patients before and during treatment with one of the TNF blockers or with placebo were investigated at the same time and by the same person. In the group initially treated with placebo we did not find any difference in the percentage of cytokine-positive T cells during placebo treatment, whereas we found the same changes as in the TNF blockade–treated patients once placebo treatment was switched to infliximab (recent study) or etanercept (2). Interestingly, in patients treated with infliximab, we observed a down-regulation of the percentage of TNF␣- and IFN␥-positive cells, while these cells were up-regulated in patients treated with etanercept. Baeten et al reported in 2001 (3) that impaired Th1 cytokine production in SpA patients was restored by antiTNF␣ treatment. However, in their data from 4 weeks after the 1017 second or third infliximab infusion, a time point that is close to our 6-week postinfusion measurements, there were no significant changes in the number of IFN␥⫹ T cells (20.9% before infusion versus 22.0% after infusion in the study by Baeten and colleagues). Furthermore, the results presented in the letter by Baeten et al show, similar to our results, a down-regulation of the number of IFN␥⫹ CD4⫹ T cells at week 6 after the initiation of treatment with infliximab. The small number of patients tested by those authors and the fact that we found an even clearer down-regulation of the number of Th1 cytokine– positive T cells after 12 weeks might explain why this difference was not significant. For the percentage of TNF␣-positive T cells as well, we found a clearer down-regulation after 12 weeks of continuous treatment, compared with 6 weeks. Finally, restoration of an impaired Th1 response by infliximab would be in clear contrast to the observed exacerbation of latent tuberculosis in infliximab-treated patients. Intact function of Th1 cells is crucial to fight Mycobacterium tuberculosis effectively. Thus, our findings could better explain this clinical observation. In summary, we cannot fully explain the differences found in our study and the experiments reported by Baeten et al. We have suggested some possible explanations, and we are confident that we used a proper experimental design in our study. Future treatment studies should also address this very important question of whether TNF␣-blocking agents suppress or restore Th1 function and whether there is a difference between the available TNF␣-blocking agents with a similar study design and also including other diseases such as rheumatoid arthritis and Crohn’s disease. Joachim Sieper, MD Jiangxiang Zou, MD Martin Rudwaleit, MD Jurgen Braun, MD University Hospital Benjamin Franklin Berlin, Germany 1. Braun J, Xiang J, Brandt J, Maetzel H, Haibel H, Wu P, et al. Treatment of spondyloarthropathies with antibodies against tumour necrosis factor ␣: first clinical and laboratory experiences. Ann Rheum Dis 2000;59 Suppl 1:i85–9. 2. Zou J, Rudwaleit M, Brandt J, Thiel A, Braun J, Sieper J. Up regulation of the production of tumour necrosis factor ␣ and interferon ␥ by T cells in ankylosing spondylitis during treatment with etanercept. Ann Rheum Dis 2003;62:561–4. 3. Baeten D, van Damme N, van den Bosch F, Kruithof E, De Vos M, Mielants H, et al. Impaired Th1 cytokine production in spondyloarthropathy is restored by anti-TNF␣. Ann Rheum Dis 2001;60: 750–5. DOI 10.1002/art.20210 Recurrent pregnancy loss in the context of antiphospholipid antibodies: comment on the article by Triolo et al To the Editor: We read with interest the article by Triolo et al (Triolo G, Ferrante G, Ciccia F, Accardo-Palumbo A, Perino A, 1018 LETTERS Castelli A, et al. Randomized study of subcutaneous low molecular weight heparin plus aspirin versus intravenous immunoglobulin in the treatment of recurrent fetal loss associated with antiphospholipid antibodies. Arthritis Rheum 2003; 48:728–31). There is indeed renewed controversy regarding the benefit of low molecular weight (LMW) heparin with or without aspirin for the treatment of antiphospholipid antibody (aPL)–positive recurrent pregnancy loss, despite the fact that a few years ago it was considered the standard of care. It would be useful to have clarification on 2 issues raised by the study design. First, given the lack of consensus in the literature regarding the value of intravenous immunoglobulin (IVIG) for aPL-positive recurrent pregnancy loss, it was surprising to see it referred to as one of the most efficacious therapeutic regimens for this problem. Could the authors please describe their rationale for selecting this particular treatment rather than aspirin alone, or even placebo, as a comparator for LMW heparin? Second, although we understand and fully sympathize with the difficulties in accruing sufficient numbers of patients who fulfill these inclusion criteria, the small sample sizes in the 2 treatment arms in this study raise some concern. In addition, there was no information in the Methods section regarding the expected event rates for either treatment. An analysis of the raw data, for both first trimester losses and live birth rates, did not result in the same P values as those provided, and indicated insufficient statistical power to draw any conclusions regarding the comparative efficacy of IVIG and LMW heparin. We and other investigators studying recurrent pregnancy loss in the context of aPL are still struggling to determine the most appropriate treatment. We very much appreciate the authors’ presentation of their experience with these 2 regimens. C. A. Laskin, MD C. A. Clark, BSc K. A. Spitzer, BSc University of Toronto Toronto, Ontario, Canada DOI 10.1002/art.568 Applications Invited for Editor of Arthritis & Rheumatism, 2005–2010 and Editor of Arthritis Care & Research, 2005–2009 During the summer and fall of 2004, the American College of Rheumatology Committee on Journal Publications will review applications for the position of Editor, Arthritis & Rheumatism, 2005–2010 term and the position of Editor, Arthritis Care & Research, 2005–2009 term. The official term of the next Arthritis & Rheumatism editorship is July 1, 2005–June 30, 2010; however, some of the duties of the new Editor will begin during a transition period starting April 1, 2005. The official term of the next Arthritis Care & Research editorship is July 1, 2005–June 30, 2009; however, some of the duties of the new Editor will begin during a transition period starting April 1, 2005. The deadline for completed applications is June 1, 2004, and the final selection will be announced by November 2004. It is requested, but not required, that those who plan to apply for either position submit a nonbinding letter of intent by April 15, 2004. For additional information or to request an application or submit a letter of intent, contact Jane Diamond, Managing Editor, at the ACR office.