Home

Current Issue

Archives

Guidelines for Authors and Reviewers

Classified Ads

Links

Search PubMed

Subscriptions

Subscriber Registration

Guidelines for Website Users

E-mail Alert Service

Contact Info

Correspondence

To the Editor:

A recent case-control (n = 124, 365) cohort study¹ concluded that, "pre-rheumatoid immunological process as reflected in RF production is not associated with any marked inflammation or tissue injury heightening C-reactive protein (CRP)." That study including 39 male and 85 female pre-RA cases¹ also showed no difference in CRP concentrations between subjects positive and negative for RF at baseline. Authors inferred that the prerheumatoid immunological process reflected in RF production is not associated with any marked inflammation or tissue injury¹.

Our baseline findings in adult males from a case-control study nested within a community cohort (n = 8745 males) offer a different perspective from the reported interpretations¹. The 18 incident male RA cases developed American College of Rheumatology (+) disease 3 to 20 years (median 12) after entry at baseline in 1974². Four controls (CN) from the same entry cohort were matched to each case for age, sex, and race (all Caucasian).

Rheumatoid factors were assayed by 3 methods. The 2 whole sera assays were Behring nephelometer and ELISA (Hemagen Diagnostics Inc, Waltham, MA, USA, according to manufacturers' instructions) techniques against human IgG^3,4. The third assay was an IgM isotype-specific ELISA against rabbit IgG⁵. Elevated concentrations for both whole sera methods are 20 IU/ml. For purposes of this study, elevated levels for the IgM method are 52 IU/ml. Serum CRP was assayed by an ultrasensitive antigen capture ELISA (Hemagen Diagnostics) with a detection threshold of 0.1 µg/ml and elevated levels of 8.0 µg/ml⁶. Acute serum amyloid-A (ASAA) concentrations were assayed by ELISA (Hemagen Diagnostics) with a detection threshold of 1.0 µg/ml and elevated levels for males of 10.0 µg/ml⁷.

Associations among only the elevated concentrations of CRP, ASAA, and RF were determined (Table 1), unlike the previous study¹, which analyzed CRP concentrations across the full ranges of values, stratified by either quintiles or the highest deciles. Fisher's exact test was used to determine p values (2 tailed) for comparisons of the 18 pre-RA versus the 72 controls in dichotomous distributions of their elevated analyte values. This test was also used to assess the probability of associations of elevated CRP with other elevated analyte values in single samples, in either the pre-RA cases (i.e., *, in Table 1) or controls (i.e., in Table 1).

Table 1. Relationships of elevated levels of rheumatoid factors (RF), C-reactive protein (CRP), and acute serum amyloid-A (ASAA) in 18 male incident RA cases before (3-20 yrs) clinical onset (pre-RA) and in 72 matched controls.

Baseline rheumatoid factor elevation (RF+) by one or more methods was found in 6 (33.3%) of 18 pre-RA cases versus 6 (8.3%) of 72 controls (p = 0.012). RF+ by 2 or more methods was observed in 4 (22.2%) pre-RA cases, but in no controls (p = 0.001). Elevated CRP levels (CRP+) were found in 4 (22.2%) pre-RA cases versus 4 (5.6%) controls (p = 0.048). Concordance of CRP+ and any RF+ was found only in the pre-RA cases (p = 0.005), and the overlap was complete in the 4 pre-RA cases having 2 or more RF+ (p < 0.001).

Elevated ASAA levels occurred in 5.6 % of both pre-RA and control subjects. The one pre-RA case with an elevated ASAA level was CRP+ and had all 3 RF+. In controls, CRP and ASAA concentrations were each elevated in 4 subjects, 3 of whom had positive concordance (p < 0.001), and none had elevation of any RF. Thus, elevated CRP concentrations were associated with pre-RA status (p = 0.048), but not independently of RF+. Elevated CRP was associated significantly (p = 0.005) with RF+ only in the pre-RA cases. As expected⁸, CRP+ was associated significantly (p < 0.001) with ASAA+ in controls (Table 1).

A subset of 16 (44.4%) of our 36 pre-RA females had assays of serum CRP, ASAA, and IgM isotype-specific RF. Baseline serum RF+ was significantly associated with pre-RA status to a similar degree as in males. In controls, CRP was significantly (p < 0.001) associated with ASAA (data not shown). In our female subjects, CRP did not associate with either pre-RA status or baseline RF+, as reported¹.

The prerheumatoid immunologic associations in our male sample differ from control findings (Table 1) and support recent observations of significant correlations between hormones and biological mediators or modulators of inflammation long before clinical onset of RA². The prerheumatoid immunological process, as reflected in RF production^1-4, may be associated with longterm, systemic physiological perturbations of the normal counter-regulatory cytokine and modulator networks of inflammation^2,9. Putative asymptomatic synovitis or other intraarticular inflammatory processes may be hypothesized as preceding clinical disease and causing immunological activation¹. However, an alternative physiological perspective was proposed that chronic perturbations of the neuroendocrine, immunological, and microvascular endothelial systems may contribute to imbalances in the normal cytokine counter-regulatory processes that may predispose to chronic synovitis⁹.

By interpreting results from both studies conjointly (Table 1 and¹), and analyzing the sex-specific findings, a broader understanding of RA precursors may be achieved. Further prospective investigations of systemic physiological perturbations before the clinical onset of RA promise to clarify its complex and likely prolonged physiopathogenesis^1,2,9.

ALFONSE T. MASI, MD, DRPH; JEAN C. ALDAG, PhD, University of Illinois College of Medicine, Peoria, Illinois; JEAN SIPES, PhD, National Institutes of Health, Bethesda, Maryland, USA.

The opinions expressed here do not necessarily reflect those of the National Institutes of Health.

REFERENCES

1. Aho K, Palosuo T, Knekt P, Alha P, Aromaa A, Heliövaara M. Serum C-reactive protein does not predict rheumatoid arthritis. J Rheumatol 2000;27:1136-8.

2. Masi AT, Aldag JC, Pirzada FA, et al. Serum hormone and biological mediator correlations were found many years before onset of rheumatoid arthritis in 18 pre-RA males, but not in 72 matched controls [abstract]. Arthritis Rheum 2000;43 Suppl:S158.

3. Masi AT, Chatterton RT, Lu YC, et al. Low serum dehydroepiandrosterone sulfate levels, rheumatoid factor and elevated serum tumor necrosis factor without cytokine receptor elevations predict long-term onset of RA in women: a controlled, prospective study [abstract]. Arthritis Rheum 1995;38 Suppl:S214.

4. Masi AT, Aldag JC, Fecht T, et al. Rheumatoid factor positivity (RF+) and current cigarette smoking (CCS+) of 30+ daily are independent long-term predictors of RA [abstract]. Arthritis Rheum 1997;40 Suppl:S312.

5. Swedler W, Wallman J, Froelich CJ, Teodorescu M. Routine measurement of IgM, IgG, and IgA rheumatoid factors: high sensitivity, specificity, and predictive value for rheumatoid arthritis. J Rheumatol 1997;24:1037-44.

6. Loose LD, de Oliveira R, Sipe JD, Franzblau C, Shanahan WR. A possible systemic component of osteoarthritis (OA): elevated concentrations (by ELISA) of C-reactive protein in serum of OA patients and modulation by tenidap [abstract]. Arthritis Rheum 1993;36 Suppl:S166.

7. de Oliveira RM, Sipe JD, Loose LD, et al. A rapid sensitive enzyme-linked immunosorbent assay for serum amyloid in plasma and tissue culture fluids. Amyloid Int J Exp Clin Invest 1994; 1:23-9.

8. Gabay C, Kushner I. Acute-phase proteins and other systemic responses to inflammation. N Engl J Med 1999;340:448-54.

9. Masi AT. Hormonal and immunologic risk factors for the development of rheumatoid arthritis: an integrative physiopathogenetic perspective. Rheum Dis Clin N Am 2000;26:775-803.

Dr. Aho, et al reply

To the Editor:

Clinical rheumatoid arthritis (RA) is preceded by an immunological process characterized by the occurrence of serum rheumatoid factor (RF) and certain other marker antibodies of RA¹. Any satisfactory explanation for the etiopathogenesis of RA should account for this process. Pre-illness sera from healthy subjects who are known to have later developed RA are available from few population based series. Thus, it is important to try to extract all relevant information from such sera. Although there is consensus concerning the existence of the pre-rheumatoid immunological process, some variation has been noted in certain details. Thus, we did not find any increase in C-reactive protein (CRP) concentrations in pre-illness specimens compared to sera from matched controls², whereas Masi, et al in their letter report a statistically significant difference in men (but not in women). This may have been due to chance but it is possible that there are underlying differences in the immunological background of the patient series. Indeed, some interesting differences between populations have been observed, e.g., anti-RA33, one marker antibody of RA, is present in some 30-40% of Central European patients with RA. Yet this antibody occurs very seldom in Finnish RA patients and it proved to be virtually absent in a large patient series collected from the Bethesda area in the United States³.

In view of the findings by Masi, et al, we now report on determinants of CRP concentrations among the control subjects who did not develop RA². We analyzed such available measures for their associations with CRP that have been connected with or may be involved in the prerheumatoid process. Male sex, poor perceived health, heavy smoking, and high body mass index, independent of each other, were significantly associated with an elevated CRP (Table 1). No association was noted with serum RF, antifilaggrin antibody, dehydroepiandrosterone sulfate, testosterone, alpha-tocopherol, or selenium concentrations (data not shown).

Table 1. Adjusted* odds ratio (95% CI) of elevated serum CRP concentration (³ 1.9 µg/l, the highest tertile) for different factors.

Our findings are in agreement with earlier observations indicating that slightly heightened CRP, yet within the "normal" range, is associated with a multitude of factors. It is likely that the association between smoking and elevated CRP is causal, whereas the sequence of events in most instances is not known. Smoking has to be taken into account as a confounder if some factor seems to be associated with an increased risk of RA.

KIMMO AHO, MD; TIMO PALOSUO, MD; PAUL KNEKT, PhD; PIRKKO ALHA, MSc; ARPO AROMAA, MD; MARKKU HELIÖVAARA, MD, National Public Health Institute, Helsinki, Finland.

REFERENCES

1. Aho K, Palosuo T, Kurki P. Marker antibodies of rheumatoid arthritis: diagnostic and pathogenetic implications. Semin Arthritis Rheum 1994;23:379-87.

2. Aho K, Palosuo T, Knekt P, Alha P, Aromaa A, Heliövaara M. Serum C-reactive protein does not predict rheumatoid arthritis. J Rheumatol 2000;27:1136-8.

3. Goldbach-Mansky R, Lee J, McCoy A, et al. Rheumatoid arthritis associated autoantibodies in patients with synovitis of recent onset. Arthritis Res 2000;2:236-43.

Are Care Pathways the Answer?

To the Editor:

In rheumatology we are used to the idea that there are therapies that in theory ought to work and when tested in small open studies seem to work, yet in the hard world of controlled trials are found wanting. Anti-CD4 therapy for rheumatoid arthritis (RA) is one example. I was disappointed to find neither of the recent otherwise excellent editorials^1,2, commenting on the failure by March, et al³ to find benefit from an evidence based clinical pathway (or guideline), seemed prepared to grasp this option. That is, that in a similar manner guidelines that should work commonly do not in practice achieve their intended results, perhaps because the "best practice" on which they are based is itself unproven, and often even untested. A perhaps provocative corollary is that those multitudes, myself included, participating in the churning out of evidence based "best practice guidelines" should implicitly accept the imemdiate responsibility for testing them in a controlled trial setting rather than puzzling over why their acceptance is so poor.

This is perhaps only a restatement of the last paragraph of Naglie and Alibha¹, but I believe it requires emphasis. This allocation of responsibility, if accepted, might slow down or even interrupt the current torrent of often conflicting guidelines and allow the poor clinician time to catch their breath, and actually get on with treating the patients.

ANTHONY S. RUSSELL, FRCPC, Department of Medicine, University of Alberta, Edmonton, Alberta T6G 2Z2, Canada.

REFERENCES

1. Naglie IG, Alibhal SMH. Improving outcomes in hip fracture patients: are care pathways the answer? [editorial]. J Rheumatol 2000;27:2068-70.

2. Syed KA, Bogoch ER. Integrated care pathways in hip fracture management. Demonstrated benefits are few [editorial]. J Rheumatol 2000;27:2071-3.

3. March LM, Cameron ID, Cumming RG, et al. Mortality and morbidity after hip fracture: Can evidence based clinical pathways make a difference? J Rheumatol 2000;27:2227-31.

Drs. Naglie and Alibhai reply

To the Editor:

We share Dr. Russell's concerns about the lack of evidence supporting many common medical practices. We agree that many interventions that have been supported only by observational data have subsequently been shown to be ineffective when subjected to well designed randomized controlled trials (RCT). As we highlighted in our editorial, critical care pathways should be subjected to the same rigorous standards of proof of benefit as other medical therapies (i.e., methodologically sound RCT) before they are widely implemented.

Care pathways are complicated by the fact that they include several interventions, some of which are supported by RCT, but others which are not. When the term "evidence based" is added to care pathways, the health care community may be misled to believe that all elements of the pathway are supported by RCT. In addition, care pathways influence care beyond the constellation of interventions that they incorporate. Care pathways can draw on resources to introduce and support their implementation and to meet their requirement for documentation. This, in turn, may have unforeseen consequences on patient care. Therefore, one cannot assume that implementation of "evidence based" care pathways will result in improved clinical outcomes. Rather, as Dr. Russell emphasizes, it is essential to evaluate them with RCT to establish that they achieve their intended result of improving patient outcomes.

I. GARY NAGLIE, MD, FRCPC; SHABBIR M.H. ALIBHAI, MD, FRCPC, Departments of Medicine and Health Administration, University of Toronto, Toronto, Ontario, Canada.

Drs. Syed and Bogoch reply

To the Editor:

Dr. Russell states that the cause of failure by March, et al¹ to show a benefit of the use of an integrated pathway for hip fracture care was that the best practice guidelines on which the pathway was based were flawed. He also implies that care pathways in general have this problem, which leads to their not being useful in real-world situations.

As we pointed out in our editorial², there are care pathways that have been successful in a variety of circumstances. These were, for the most part, in the setting of elective surgery. It cannot be said that all care pathways are flawed and should be abandoned. A number of factors determine the success of care pathways, including the best practice guidelines on which they are based. We feel that the most important determinant of effectiveness of a care pathway is the appropriateness of the circumstance in which it is implemented. Hence, in the more "controlled" elective setting, care pathways function relatively well. In hip fractures, due to their urgent nature, it is difficult to match needs and resources.

In general, we feel that care pathways can be useful where care processes are homogeneous, elective, and scheduled (e.g., total knee replacement), but in situations where comorbidity is high and complications common, and matching resources to the flow of patients is unpredictable (e.g., hip fracture), it is unlikely that they would work.

KHALID A. SYED, MD; EARL R. BOGOCH, MD, FRCSC, Department of Surgery, St. Michael's Hospital, University of Toronto, Toronto, Ontario M4Y 1J3, Canada.

REFERENCES

1. March LM, Cameron ID, Cumming RG, et al. Mortality and morbidity after hip fracture: can evidence based clinical pathways make a difference? J Rheumatol 2000;27:2227-31.

2. Syed KA, Bogoch ER. Integrated care pathways in hip fracture management. Demonstrated benefits are few [editorial]. J Rheumatol 2000;27: 2071-3.

Drs. March, et al reply

To the Editor:

Dr. Russell appears to make the point that just because something doesn't show a benefit in a controlled trial doesn't mean it is not a useful intervention. We would make 2 points in response to this. First, we would disagree on the one hand and defend the need for adequately powered controlled trials when deciding on efficacy, particularly in the field of rheumatology, where disease presentation is heterogeneous, natural history fluctuates markedly, and outcome measures are subjective. Second, we would agree to a certain extent in the setting where a study lacked power or when all relevant outcomes were not evaluated. Our paper on clinical pathways for hip fracture¹ did discuss possible explanations for the lack of effect on mortality. Sample size is always an issue in any study, but another important issue is choosing the outcome measures most likely to be influenced by your intervention. In our case, quality of life of hip fracture subjects was not assessed in sufficient numbers to allow analysis, but quality of life adjusted years may have been the more appropriate measure from both an individual and a societal perspective.

Dr. Russell also makes the point that guidelines may fail if not based on proven therapy and laments the existence of numerous and at times conflicting guidelines for the one condition. We agree and strongly advocate the development and implementation of evidence based guidelines that recommend interventions proven in controlled trials wherever possible, and when not available this is made explicit in the guideline. This is a solution to both Dr. Russell's problems and is the basis for our hip fracture pathway.

I believe we also agree on his final point that clinicians should take the responsibility for evaluating the guidelines they develop.

Guidelines aren't going to go away. Clinicians should not ignore them, but should get involved in their development and appraisal so that only the "good" ones are widely implemented.

LYN MARCH, PhD; IAN CAMERON, PhD; ROBERT CUMMING, PHD; PHILIP SAMBROOK, MD, University of Sydney, Department of Rheumatology, Royal North Shore Hospital, St. Leonards, NSW, Australia 2065.

REFERENCE

1. March LM, Cameron ID, Cumming RG, et al. Mortality and morbidity after hip fracture: can evidence based clinical pathways make a difference? J Rheumatol 2000;27:2227-31.

To the Editor:

Dr. Croft's editorial joins two others^2,3 in the same issue of The Journal in a formidable assault on the tender points of fibromyalgia (FM). We will try to deal only with the first of these papers.

We are flattered by Dr. Croft's kind remarks about our work, but we have, sadly, failed to convince him. He does not take issue with our findings that certain tender points (TP) are significantly more prevalent in population subjects with FM than in those with chronic generalized pain (CGP)⁴. He refers to our publications^5,6 and summarizes them well when he states that subjects with FM "tend to be more distressed, more disabled, and more likely to seek health care" than those with CGP¹. He does not like to use the term fibromyalgia and prefers to talk about patients with CGP with higher or lower TP counts. He asks what the classification of patients "on the basis of a high tender point count" (that is, FM) adds to the management of chronic pain. We agree that it adds very little. That is not the issue, however. What, for instance, did the discovery of rheumatoid factor in the 1940s add to the management of rheumatoid arthritis at that time? We have to start somewhere, and defining a condition is not a bad way of doing it. Is it useful to forget about FM, and deal with CGP only? We don't think so, for several reasons: (1) there is no widely accepted definition of CGP; (2) CGP includes, of necessity, well defined entities such as polyarticular osteoarthritis, rheumatoid arthritis, polymyalgia rheumatica, etc; (3) while FM may be simply the extreme end of an epidemiological spectrum of CGP, it is more likely that we will be able to gain a better understanding of its pathogenesis, and potential treatments, by studying the extreme rather than the whole spectrum; this is the argument that Russell has made⁷; it is an argument Dr. Croft rejects because he does not think that a high TP count represents a measure of risk; in fact, it would seem that it does, not in terms of deaths perhaps, but certainly in terms of disability and distress^5,6.

We don't think there is anything mystical about TP counts. Certainly, high TP counts can occasionally be found in people who report no pain⁸. That is interesting, but it does not help our understanding of CGP or FM. TP counts give us an indirect and not very precise assessment of severity of pain and/or distress. If that were their only purpose we would agree that one should measure the "components of the distress itself"¹. However, we think that the main reason for doing TP counts is to classify patients as having FM or not; the classification criteria for FM are meant to allow for a common language in which students of this condition can communicate with each other. A second objective, for the clinician, is that of giving the patient a label for a distressing problem. Some have argued that assigning the FM label creates "medicalization and dependency"⁹. Recent evidence shows that this is not the case¹⁰. Moreover, we doubt that many patients will be satisfied with a diagnosis of chronic pain. Dr. Croft assures us that "we no longer need special pleading to defend the reality of chronic pain and the suffering associated with it"¹. We, alas, live in a less enlightened environment and deal daily with skeptical colleagues, insurers, and employers who would be less than sympathetic to a diagnosis of chronic widespread pain.

KEVIN P. WHITE, MD, PhD. FRCPC; MANFRED HARTH, MD, FRCPC, St. Joseph's Health Care London, London, Ontario, Canada.

REFERENCES

1. Croft P. Testing for tenderness: what's the point? [editorial]. J Rheumatol 2000;27:2531-3.

2. Wolfe F. Sayin' "Stand and deliver for you are a bold deceiver": Faking fibromyalgia [editorial]. J Rheumatol 2000;27:2534-5.

3. Smythe H. Fibromyalgia: can one distinguish it from malingering? More work needed; more tool supplied [editorial]. J Rheumatol 2000;27:2536-40.

4. White KP, Harth M, Speechley M, Ostbye T. A general population study of fibromyalgia tender points in noninstitutionalized adults with chronic widespread pain. J Rheumatol 2000;27:2677-82.

5. White KP, Speechley M, Harth M, Ostbye T. The London Fibromyalgia Epidemiology Study: comparing the demographic and clinical characteristics in 100 random community cases of fibromyalgia versus controls. J Rheumatol 1999;26:1577-85.

6. White EP, Speechley M, Harth M, Ostbye T. Comparing self-reported function and work disability in 100 community cases of fibromyalgia versus controls in London, Ontario: the London Fibromyalgia Epidemiology Study. Arthritis Rheum 1999;42:70-83.

7. Russell IJ,. Is fibromyalgia a distinct clinical entity? The clinical investigator's evidence. Baillieres Best Pract Res Clin Rheumatol 1999;13:445-54.

8. Croft P, Schollum J, Silman A. Population study of tender point counts and pain as evidence of fibromyalgia. BMJ 1994;309:696-9.

9. Hadler NM. Fibromyalgia: la maladie est morte. Vive le malade! J Rheumatol 1997;24:1250-1.

10. White KP, Nielson W, Harth M, Speechley M, Ostbye T. Does the label "fibromyalgia" alter health status and function: a prospective within-group comparison [abstract]. Arthritis Rheum 2000;43 Suppl:S212.

Dr. Croft replies

To the Editor:

Drs. White and Harth defend the use of tender point counts and the use of the FM label. In response, I would like to return to the distinction between classification for research purposes and labelling in clinical practice.

The clinical syndrome, as I understand it, is "chronic widespread pain without obvious underlying musculoskeletal pathology." To reemphasize an observation from the editorial, there is strong published expert opinion that the number of tender points is not in practice used as a basis for defining this syndrome in the clinic. Further, tender point counts are not used clinically to exclude from the chronic widespread pain (CWP) syndrome those "well-defined entities such as polyarticular osteoarthritis, rheumatoid arthritis, polymyalgia rheumatica," referred to by White and Harth. These conditions have their own features that place them in those other groups.

I do not feel qualified to enter the debate as to whether this syndrome should or should not be labelled "fibromyalgia syndrome" in clinical practice, but I do accept White and Harth's point that a label can be helpful to both doctor and patient as signalling the reality and severity of symptoms. As a scientist, however, I have to ask if clinicians can demonstrate the use of "pathological" labels to be truly of benefit to the patient? Thirty years ago in British general practice, the commonest labels for neck and back pain were "cervical spondylosis" and "spinal osteoarthritis." The shift in textbooks and in practice to "neck pain" and "low back pain" as acceptable labels -- although they may be difficult to define and need red flag diagnoses to be excluded -- parallels a shift to more active, less radiography obsessed, more symptom and function oriented approaches to these problems. There may be beneficial effects of a label on outcome, but these need to be demonstrated in good scientific studies.

The problematic question is whether a high tender point count should be used in addition in clinical practice to define a separate entity to "chronic widespread pain without obvious underlying musculoskeletal pathology." The analogy used is rheumatoid factor, the implication being that a high tender point count is the equivalent of a definite pathological diagnosis, even though we do not yet know what that pathology might be. White and Harth hint that this then provides a more real or convincing disease entity than "chronic widespread pain without obvious underlying musculoskeletal pathology" for insurers and employers. This seems to me scientifically unjustified at present. It leaves unanswered the question of what the reality of chronic widespread pain without high tender point counts might be and overlooks the absence of evidence on the utility of the tender point count as a diagnostic procedure in clinical practice. For every instance like rheumatoid factor, there are other examples of presumed pathologies and their accompanying diagnostic features from the history of musculoskeletal pain that have disappeared from our textbooks.

However, I certainly agree with White and Harth that tender point counts are worth doing for research classification and communication, since I believe it is important to continue to investigate whether they represent distinctive neurophysiological pathology or provide clinically effective ways to subclassify for prognosis or treatment. I would also accept that carrying out a tender point count may have a clinical function as a means of communication between doctor and patient, and might play a practical part in the management of chronic widespread pain. However, my view remains that such a science of tender points has not reached the stage that gives the count a special value in clinical practice beyond what an assessment of pain and its associated distress, disability, and handicap can more practically provide.

PETER CROFT, MD, Keele University, Keele, UK.

Epstein-Barr Virus and Rheumatoid Arthritis

To the Editor:

We read with interest the article by Blaschke, et al¹, addressing EBER 1/2 transcripts found within synovial membrane of 2 (8%) of 25 patients with rheumatoid arthritis (RA). Hybridization signals were confined to lymphocyte infiltrates of the rheumatoid synovium. Epstein-Barr virus (EBV)-carrying cells were identified as isolated cells scattered throughout the subsynovial layer. The nature of these cells, however, could not be identified by histological means. We are confused by these results that the positive cells are lymphocytes but could not be identified. Further, there is discussion that recent experimental data indicate that rheumatoid synoviocytes (besides B lymphocytes) could be a target for EBV infection: an EBV infected fibroblast cell line was established from RA synovial tissue for the first time by Koide, et al². However, Koide, et al do not say that this fibroblast cell line is a synoviocyte cell line in that paper.

In 1997, we reported the detection of EBV encoded small RNA 1 (EBER-1) and latent membrane protein 1 (LMP-1) in RA synovial lining cells³. We reported that EBER was observed in synovial lining cells from 8 (23.5%) of 34 RA patients, but in none of 20 patients with osteoarthritis (p < 0.05), nor in the one patient with psoriatic arthritis. Interestingly, EBER was localized in synovial lining cells that were located at the top of villous lesions. This was confirmed also by the 2 monoclonal antibodies to LMP-1.

We believe our article is the first report that rheumatoid lining cells (synoviocytes) -- besides B lymphocytes -- are a target for EBV infection.

SHIGEMASA SAWADA, MD, First Department of Medicine, Nihon University School of Medicine, Nerima Hikarigaoka Hospital; MASAMI TAKEI, MD, First Department of Medicine, Nihon University School of Medicine, Tokyo, Japan.

REFERENCES

1. Blaschke S, Schwarz G, Moneke D, et al. Epstein-Barr virus infection in peripheral blood mononuclear cells, synovial fluid cells and synovial membranes of patients with rheumatoid arthritis. J Rheumatol 2000;27:866-73.

2. Koide J, Takada K, Sugiura M, et al. Spontaneous establishment of an Epstein-Barr virus infected fibroblast cell line from the synovial tissue of a rheumatoid arthritis patient. J Virol 1997;71:2478-81.

3. Takei M, Mitamura K, Fujiwara S, et al. Detection of Epstein-Barr virus encoded small RNA 1 and latent membrane protein 1 in synovial lining cells from rheumatoid arthritis patients. Int Immunol 1997;9:739-43.

Dr. Blaschke replies

To the Editor:

Epstein-Barr virus (EBV) infection in the synovial tissue of patients with rheumatoid arthritis (RA) remains controversial -- some early studies failed to detect the viral genome (EBV-DNA) within the rheumatoid synovium^1,2. Using highly sensitive techniques several investigators were able to localize EBV infection in synovial tissue samples of RA patients in comparison to controls with osteoarthritis^2-7. However, conflicting results exist with respect to the frequency of infected cells and the cell type in which EBV could be identified, as follows.

Lymphocytes. We analyzed EBV infection by in situ hybridization for the small EBV encoded RNA (EBER)⁸. Distribution and morphology of EBER-RNA positive cells within the rheumatoid synovium led to the hypothesis that these cells could represent lymphocytes known to be the primary target site of latent EBV infection in the peripheral blood. Until publication of the data, we failed to further characterize these cells by double staining experiments required to prove this hypothesis. By combination of in situ hybridization for EBER-RNA and immunohistochemistry for CD antigens 2 recent studies report the detection of EBV infection within lymphocytes of the rheumatoid synovium^6,7. In addition, Niedobitek, et al⁶, using double staining for the CD20 and CD79a B cell antigens, showed that these EBV infected lymphocytes belong to the B cell subset.

Synovial fibroblasts. Takei, et al³ first hypothesized that synovial fibroblasts (previously defined as type B synoviocytes) also might be infected with EBV in the rheumatoid synovium: expression of EBER-RNA and the latent membrane protein 1 (LMP-1) of EBV was described in the synovial tissue from 8 of 34 patients with RA. In that study, infected cells were characterized as synovial lining cells only by histological means. Results of negative staining of EBV infected cells for the anti-pan B cell antibody (anti-CD19) were not shown. In addition, the staining pattern (an accumulation of EBER-1 positive cells at the apex of villous lesions of the synovial lining cells) was not observed in other studies^6,7.

The most important evidence for the hypothesis that rheumatoid synoviocytes could be a target for EBV infection results from the isolation of an EBV-carrying fibroblast cell line from RA synovial tissue⁹. This fibroblast cell line, designated DSEK, was studied for the expression of certain cell surface markers and cytokines known to be characteristic for synovial fibroblasts: DSEK cells were shown to be negative for lymphocyte and macrophage markers, but expressed CD44, CD58, and HLA-DR antigens and spontaneously produced interleukin 10, basic fibroblast growth factor, and transforming growth factor ß1.

In a recent study⁷, the same research group reported on the frequent detection of lytic EBV infection in both lymphocytes and synovial lining cells of RA synovial tissues. These results differ from the findings of Niedobitek, et al⁶, who could not localize EBV infection within the synovial lining cells using the same techniques. These discrepancies may perhaps be dependent on the EBV isolate or on the disease stage of the patient population studied. Thus, larger studies are still required to address this issue and to further elucidate the role of synovial Epstein-Barr virus infection for the pathogenesis of RA.

SABINE BLASCHKE, MD, Department of Nephrology and Rheumatology, University of Göttingen, 37075 Göttingen, Germany.

REFERENCES

1. Alspaugh MA, Shoji H, Nonoyama M. A search for rheumatoid arthritis-associated nuclear antigen and Epstein-Barr virus specific antigens or genomes in tissues and cells from patients with rheumatoid arthritis. Arthritis Rheum 1983;26:712-20.

2. Fox RI, Chilton T, Rhodes G, Vaughan JH. Lack of reactivity of rheumatoid arthritis synovial membrane DNA with cloned Epstein-Barr virus DNA probes. J Immunol 1986;137:498-501.

3. Takei M, Mitamura K, Fujiwara S, et al. Detection of Epstein-Barr virus encoded small RNA 1 and latent membrane protein 1 in synovial lining cells from rheumatoid arthritis patients. Int Immunol 1997;9:739-43.

4. Mousavi-Jazi M, Bostrom L, Lovmark C, Linde A, Brytting M, Sundqvist VA. Infrequent detection of cytomegalovirus and Epstein-Barr virus DNA in synovial membranes of patients with rheumatoid arthritis. J Rheumatol 1998;25:623-8.

5. Saal JG, Krimmel M, Steidle M, et al. Synovial Epstein-Barr virus infection increases with the risk of rheumatoid arthritis in individuals with the shared HLA-DR4 epitope. Arthritis Rheum 1999;42:1485-96.

6. Niedobitek G, Lisner R, Swoboda B, et al. Lack of evidence for an involvement of Epstein-Barr virus infection of synovial membranes in the pathogenesis of rheumatoid arthritis. Arthritis Rheum 2000;43:151-4.

7. Takeda T, Mizugaki Y, Matsubara L, Imai S, Koike T, Takada K. Lytic Epstein-Barr virus infection in the synovial tissue of patients with rheumatoid arthritis. Arthritis Rheum 2000;43:1218-25.

8. Blaschke S, Schwarz G, Moneke D, Lutz B, Müller GA, Reuss-Borst M. Analysis of Epstein-Barr virus infection in peripheral blood mononuclear cells, synovial fluid cells and synovial membranes of patients with rheumatoid arthritis. J Rheumatol 2000;27:866-73.

9. Koide J, Takada K, Sugiura M, et al. Spontaneous establishment of an Epstein-Barr virus infected fibroblast line from the synovial tissue of a rheumatoid arthritis patient. J Virol 1997;71:2478-81.

Risk Factors for Development of Lower Limb Pain in Adolescents

To the Editor:

I read with interest the article by Shrier, et al¹. This very impressive study indicates that lower limb pain in adolescents is not associated with decreased flexibility. However, it is my clinical impression that musculoskeletal pains in adolescents are much more commonly associated with hypermobility rather than hypomobility. The literature also generally supports this association^2-4, although one recent study failed to find a relationship between musculoskeletal pain and hypermobility in preadolescents⁵. I am unclear from my reading of their study if the authors were able to evaluate whether those individuals with the most flexibility by the 3 tests used had more lower limb pain than the rest of the group. If the most flexible individuals did not have the highest frequency of lower limb pain, this would be interesting information. It would either suggest that the tests of flexibility used are not good at measuring hypermobility, or that our clinical impression and the literature are mistaken.

PETER MALLESON, MBBS, MRCP(UK), FRCPC, Division of Rheumatology, Department of Pediatrics, University of British Columbia, Vancouver, British Columbia, Canada.

REFERENCES

1. Shrier I, Ehrmann-Feldman D, Rossignol M, Abenhaim L. Risk factors for development of lower limb pain in adolescents. J Rheumatol 2001;28:604-9.

2. Gedalia A, Press J. Articular symptoms in hypermobile schoolchildren: a prospective study. J Pediatr 1991;119:944-6.

3. Gedalia A, Person DA, Brewer EJ, Giannini EH. Hypermobility of the joint in juvenile episodic arthritis/arthralgia. J Pediatr 1985;107:873-6.

4. Kirk JA, Ansell BA, Bywaters EGL. The hypermobility syndrome: musculoskeletal complaints associated with generalized joint hypermobility. Ann Rheum Dis 1967;26:419-25.

5. Mikkelsson M, Salminen JJ, Kautiainen H. Joint hypermobility is not a contributing factor to musculoskeletal pain in pre-adolescents. J Rheumatol 1996;23:1963-7.

Dr. Shrier replies

To the Editor:

We thank Dr. Malleson for bringing up an important point not mentioned in our article¹. We used flexibility to refer to limitations of range of motion due to muscle, whereas the papers cited by Dr. Malleson use the term "hypermobility" to refer to limitations of range of motion due to ligaments and/or capsule. For example, we tested the hamstring range of motion by extending the knee while the hip was flexed at 90 degrees to ensure that limitations were not due to hip or knee capsule/ligament. In contrast, hypermobility in the articles cited was defined by Beighton's criteria: (1) passive dorsiflexion of the little fingers beyond 90 degrees, (2) passive apposition of thumbs to the flexor aspect of the forearm, (3) hyperextension of the elbows beyond 10 degrees, (4) hyperextension of the knees beyond 10 degrees, and (5) forward flexion of the trunk so that the palms of hands rest easily on the floor. The first 4 of these tests clearly test ligamentous laxity and not muscle flexibility. Although the last test is an extreme measure of the Sit and Reach Test, no previous papers cited reported data associating the result of this test alone with the occurrence of musculoskeletal pain.

Although previous papers discussed ligament laxity^2-4, we have rerun our analysis dichotomizing our flexibility measures as hyperflexible (top 25%) and non-hyperflexible (bottom 75%). None of the measures predicted a risk of injury, with p values ranging from 0.4 to 0.8.

We thank Dr. Malleson for the opportunity to distinguish between these subtle but important differences in determinants of range of motion.

IAN SHRIER, MD, PhD, Dip Sport Med (FACSM); DEBBIE FELDMAN, PhD, SMBD, Jewish General Hospital, Montreal, Quebec, Canada.

REFERENCES

1. Shrier I, Ehrmann-Feldman D, Rossignol M, Abenhaim L. Risk factors for development of lower limb pain in adolescents. J Rheumatol 2001;28:604-9.

2. Gedalia A, Press J. Articular symptoms in hypermobile schoolchildren: a prospective study. J Pediatr 1991;119:944-6.

3. Gedalia A, Person DA, Brewer EJ, Giannini EH. Hypermobility of the joint in juvenile episodic arthritis/arthralgia. J Pediatr 1985;107:873-6.

4. Kirk JA, Ansell BA, Bywaters EGL. The hypermobility syndrome: musculoskeletal complaints associated with generalized joint hypermobility. Ann Rheum Dis 1967;26:419-25.

Insulin-like Growth Factor and RA

To the Editor:

The recent article¹ showing an association between low levels of insulin-like growth factor (IGF-1) proteins in rheumatoid arthritis (RA) patients who are sedentary is certainly an interesting observation and not unexpected. It would be interesting to know how may patients in the study had fibromyalgia (FM), since it is known that many FM patients have low levels of IGF-1^2,3. This is not simply of academic interest because RA and FM often coexist⁴. Clearly, increasing the level of activity in sedentary RA patients would be desirable. However, if these RA patients also have FM, but the diagnosis of FM is not made, then appropriate therapies to decrease the pain due to the FM might not be instituted, thus making an attempt to increase physical activity in such patients much more difficult and, in fact, in some patients unrealistic. Many of my patients with RA are quite active. However, many RA patients who have concomitant FM are not as active as their counterparts without FM, all other things being equal.

THOMAS J. ROMANO, MD, PhD, FACP, Wheeling, West Virginia 26003, USA.

REFERENCES

1. Lemmey A, Maddison P, Breslin A, et al. Association between insulin-like growth factor status and physical activity levels in rheumatoid arthritis. J Rheumatol 2001;28:29-34.

2. Bennett R, Clark S, Campbell S, Burckhardt C. Low levels of somatomedin C in patients with the fibromyalgia syndrome. Arthritis Rheum 1992;35:1113-6.

3. Dinser R, Halama T, Hoffman A. Stringent endocrinological testing reveals subnormal growth hormone secretion in some patients with fibromyalgia syndrome but rarely severe growth hormone deficiency. J Rheumatol 2000;27:2482-8.

4. Romano T. Fibromyalgia syndrome in other rheumatic conditions. Lyon Mediterranee Medical, Medecine Du Sud-Est 1996; 32:2143-6.

Dr. Maddison replies

To the Editor:

We thank Dr. Romano for his interest in our work. We agree that he makes an important point that the presence of fibromyalgia (FM), itself associated with low levels of insulin-like growth factor (IGF-1), might have contributed to the low serum levels of IGF-1 seen in our patients with rheumatoid arthritis. However, in this study, the presence or absence of FM features was not recorded in the rheumatoid population we studied and this question cannot be addressed.

On the other hand, it is unlikely that FM made a significant contribution to the results in the nonrheumatoid group with other rheumatic complaints, the majority of whom did not have a generalized pain syndrome. Further, the 5 patients with primary FM showed a range of serum IGF-1 results from low normal levels to low (one patient).

In our study, level of exercise was the variable most closely associated with low circulating levels of IGF-1 and IGFBP-3. Currently we are examining the effect of an exercise intervention on growth hormone and IGF status.

PETER J. MADDISON, MD, Department of Rheumatology, Ysbyty Gwynedd and School of Sport, Health and Exercise Sciences, University of Wales, Bangor, Wales.

Clinical Correlates of Avascular Necrosis in Systemic Lupus Erythematosus

To the Editor:

I read with interest the article by Gladman, et al and their review of the literature analyzing predictive factors of symptomatic osteonecrosis in patients with lupus¹. Although they give due credit for Lew Cozen's landmark work with Edmund Dubois² describing the syndrome in 1960, Gladman, et al should know of a more recent effort spearheaded by Dr. Cozen.

Now 90 years of age, Dr. Cozen is still a vibrant, active member of our medical staff. In 1998, he published an analysis similar to Gladman, et al, but in the orthopedic literature². He spent 2 years following osteonecrosis patients and compared them with SLE patients without osteonecrosis. Analyzing numerous comparisons of 488 patients, he found statistically increased prevalences for hypertension, pleural effusions, cerebritis, nephritis, anemia, and hemolytic anemia among patients with osteonecrosis. No serologic feature, including antiphospholipid antibodies, was predictive. This information should be added to Gladman's informative review.

DANIEL J. WALLACE, MD, Clinical Professor of Medicine, Cedars-Sinai/UCLA School of Medicine, Los Angeles, California 90048.

REFERENCES

1. Gladman DD, Urowitz MB, Chaudhry-Ahluwalia V, Hallet DC, Cook RJ. Predictive factors for symptomatic ostenecrosis in patients with systemic lupus erythematosus. J Rheumatol 2001;28:761-5.

2. Dubois EL, Cozen L. Avascular (aseptic) bone necrosis associated with systemic lupus erythematosus. JAMA 1960;174:966-84.

3. Cozen L, Wallace DJ. Avascular necrosis in systemic lupus erythematosus: Clinical associations and a 47-year perspective. Am J Orthopedics 1998;27:352-4.

Drs. Gladman and Urowitz reply

To the Editor:

We thank Dr. Wallace for his letter and the confirmation that anticardiolipin antibodies were not associated with avascular necrosis (AVN, osteonecrosis), by Dr. Cozen in The American Journal of Orthopedics, which we neglected to include in our reference list. However, it should be noted that while Dr. Cozen's paper describes items associated with the presence of AVN, patients were not matched for disease duration and followup. Indeed, patients without AVN had a shorter followup duration than patients with AVN, therefore it is possible that some items present in the patients with AVN had not had a chance to develop in patients without AVN. Moreover, the analysis provided is only univariate and the p values for the items identified are very small. Nonetheless, it is important for readers to appreciate the work with these reservations in mind.

DAFNA D. GLADMAN, MD, FRCPC; MURRAY B. UROWITZ, MD, FRCPC, Centre for Prognosis Studies in the Rheumatic Diseases, Toronto Western Hospital, Toronto, Ontario M5T 2S8.