Detailing ethnicity and phenotypes is critical for pooling association studiesComment on the article by Huizinga et al.код для вставкиСкачать
ARTHRITIS & RHEUMATISM Vol. 52, No. 2, February 2005, pp 670–678 © 2005, American College of Rheumatology LETTERS is also capable of transporting di- and triglutamate forms of MTX (9). This is important, as Kremer (1) highlights, since the polyglutamated forms of MTX are critical to its intracellular activity and efficacy. Further studies are required to clarify the relative importance of the different cellular efflux proteins in mediating an individual’s response to MTX in rheumatoid arthritis. DOI 10.1002/art.20770 Drug efflux transporters in rheumatoid arthritis: comment on the article by Kremer To the Editor: We read with interest the article by Kremer (1) reviewing the mechanisms of action of methotrexate (MTX) and potential mechanisms of MTX resistance. One facet that Kremer highlighted is the role of drug efflux transporters in mediating resistance to MTX. Although he highlights the role of multidrug resistance–associated proteins (MRPs) in mediating MTX transport, he also emphasizes a study examining permeable glycoprotein (P-gp) expression and response to MTX (2,3). Although it is now established that MRP-1 transports MTX (4), whether P-gp transports MTX is less clear. In order to clarify this issue, we undertook a drug accumulation study using radiolabeled MTX in an in vitro T lymphoblastoid cell line (CEM cells). Parent CEM cells do not express any efflux transporters, but CEM cells pretreated with vinblastine (VBL) constitutively express the efflux transporter P-gp. CEM cells pretreated with epirubicin (E1000) constitutively express the efflux transporter MRP-1. This in vitro system is well characterized and has been used to determine the impact of the multidrug resistant transporters on the cellular accumulation of other compounds (5). We incubated 1 ⫻ 106 cells/ml with radiolabeled MTX for 18 hours at 37°C across a range of drug concentrations. The excess supernatant was discarded and the cells were extracted in methanol. The intracellular drug concentrations were calculated for the 3 cell lines and results compared using the Kruskal-Wallis test for multiple comparisons. The extent of accumulation of MTX in the MRP-1–expressing cells was significantly reduced (P ⬍ 0.001), compared with the CEM and VBL cells (Figure 1). These data support the balance of published evidence suggesting that MTX is a substrate for MRP-1 (4) but is not a substrate for P-gp (6). This is also consistent with the observation that P-gp preferentially transports neutral hydrophobic compounds, whereas MRP-1 preferentially transports anionic compounds such as MTX (6). Instead of P-gp it is likely that the transporters MRPs 1–4 (4,6,7), together with breast cancer resistance protein (8), are more important in mediating MTX resistance. Studies suggest that breast cancer resistance protein Supported in part by the UK Arthritis Research Campaign. Samantha L. Hider, MRCP, MSc, BM, BS University of Manchester Manchester, UK Patrick Hoggard, PhD Saye Khoo, MD, MRCP David Back, PhD University of Liverpool Liverpool, UK Ian N. Bruce, MD, FRCP University of Manchester Manchester, UK 1. Kremer JM. Toward a better understanding of methotrexate [review]. Arthritis Rheum 2004;50:1370–82. 2. Llorente L, Richaud-Patin Y, Diaz-Borjon A, Alvarado dlB, JakezOcampo J, De La FH, et al. Multidrug resistance-1 (MDR-1) in rheumatic autoimmune disorders. Part I: Increased P-glycoprotein activity in lymphocytes from rheumatoid arthritis patients might influence disease outcome. Joint Bone Spine 2000;67:30–9. 3. Ranganathan P, Eisen S, Yokoyama WM, McLeod HL. Will pharmacogenetics allow better prediction of methotrexate toxicity and efficacy in patients with rheumatoid arthritis? Ann Rheum Dis 2003;62:4–9. 4. Hooijberg JH, Broxterman HJ, Kool M, Assaraf YG, Peters GJ, Noordhuis P, et al. Antifolate resistance mediated by the multidrug resistance proteins MRP1 and MRP2. Cancer Res 1999;59:2532–5. 5. Davey RA, Longhurst TJ, Davey MW, Belov L, Harvie RM, Hancox D, et al. Drug resistance mechanisms and MRP expression in response to epirubicin treatment in a human leukaemia cell line. Leuk Res 1995;19:275–82. 6. Jansen G, Scheper RJ, Dijkmans BA. Multidrug resistance proteins in rheumatoid arthritis, role in disease-modifying antirheumatic drug efficacy and inflammatory processes: an overview. Scand J Rheumatol 2003;32:325–36. 7. Zeng H, Chen ZS, Belinsky MG, Rea PA, Kruh GD. Transport of methotrexate (MTX) and folates by multidrug resistance protein (MRP) 3 and MRP1: effect of polyglutamylation on MTX transport. Cancer Research 2001;61:7225–32. 8. Volk EL, Farley KM, Wu Y, Li F, Robey RW, Schneider E. Overexpression of wild-type breast cancer resistance protein mediates methotrexate resistance. Cancer Res 2002;62:5035–40. 9. Chen ZS, Robey RW, Belinsky MG, Shchaveleva I, Ren XQ, Sugimoto Y, et al. Transport of methotrexate, methotrexate polyglutamates, and 17␤-estradiol 17-(␤-D-glucuronide) by ABCG2: effects of acquired mutations at R482 on methotrexate transport. Cancer Res 2003;63:4048–54. DOI 10.1002/art.20744 Methotrexate and long-term treatment of rheumatic disease: comment on the article by Kremer Figure 1. Intracellular accumulation of methotrexate (MTX). Values are the mean and SD (n ⫽ 4). ⴱ ⫽ P ⬍ 0.001 in multidrug resistance–associated protein-1–expressing cells (E1000) versus CEM and vinblastine (VBL) cells. To the Editor: We read with interest the excellent review on methotrexate (MTX) by Kremer (1). The author discusses drug 670 LETTERS Figure 1. Competition of multidrug resistance-associated proteins (MRPs) and folylpolyglutamate synthetase (FPGS) for methotrexate (MTX). MTX is transported into the cell by the reduced folate carrier (RFC). MRPs can transport unmodified MTX out of the cell, but not polyglutamated MTX (MTX-Glu2–6) which inhibits dihyrofolate reductase (DHFR). Reproduced, with permission, from Borst P, Evers R, Kool M, Wijnholds J. A family of drug transporters: the multidrug resistance-associated proteins. J Natl Cancer Inst 2000;92:1295–302. transporters in MTX metabolism, including those which cause drug efflux, and their role in leading to the plateau in clinical response, which is often seen when MTX is used in the long-term treatment of rheumatic diseases. We would like to highlight some of the controversies surrounding the role of these transporters in mediating resistance to MTX. As noted by Kremer, MTX transport into the cell is mediated by members of the solute carrier family of transporters, specifically SLC19A1, which is also called the reduced folate carrier. It is increasingly recognized that genetic variations in this transporter can lead to differential responses to MTX (2,3). After entry into the cell, MTX is converted into the polyglutamated form by the enzyme folylpolyglutamate synthetase (FPGS). The multidrug resistant transporter MDR1/P-glycoprotein (P-gp) and the multidrug resistance– associated proteins (MRPs), both members of the ATPbinding cassette family of transporters, have been implicated in MTX efflux from the cell. Whether MDR1/P-gp includes MTX among its several substrates remains controversial. Norris et al showed that leukemic cell lines resistant to MTX had increased expression and function of the MDR1 gene, and MTX resistance was partially reversed by a P-gp–specific monoclonal antibody (4). Similarly, it has been demonstrated that the peripheral blood lymphocytes (PBLs) of patients with rheumatoid arthritis (RA) who are refractory to treatment with MTX express higher levels of P-gp compared with responders (5). In contrast to this, MDR1-transgenic mice overexpressing P-gp in their bone marrow cells develop bone marrow suppression when treated with MTX (6). Another study showed no difference in the expression of P-gp on PBLs between RA patients who were MTX responders and those who were nonresponders (7). 671 The MRP family has seven members, MRP-1–MRP-7. Of these, MRPs 1, 2, 3, and 4 have been mainly implicated in MTX resistance. Interestingly, it has been found that MRPs 1, 2, and 3 cause MTX resistance in cells during a 4-hour exposure to MTX at high concentrations, but not during a 96-hour continuous exposure to low-dose MTX (8, 9). This may be because, although both FPGS and MRPs compete for MTX, MTX polyglutamates cannot be transported by MRPs (Figure 1). Cells with chronic exposure to low-dose MTX may slowly accumulate these polyglutamates, and remain susceptible to MTX, whereas such a phenomenon does not occur with short-term exposure to high concentrations of MTX. Such resistance to high-dose MTX can be partially overcome by inhibitors of MRP-1, such as sulfinpyrazone and probenecid. We agree with Kremer that drug efflux transporters such as MDR1/(P-gp) and MRPs contribute significantly to the loss of response to MTX. However, although these transporters are crucial in determining resistance to high-dose MTX, as used in cancer treatment, their role in mediating resistance to low-dose MTX, as used in the treatment of rheumatic diseases, remains quite controversial. Nevertheless, concomitant use of inhibitors of these transporters with MTX may be a promising strategy to improve the efficacy of the drug in the future. Prabha Ranganathan, MD Howard L. McLeod, PharmD Washington University School of Medicine St. Louis, MO 1. Kremer JM. Toward a better understanding of methotrexate [review]. Arthritis Rheum 2004;50:1370–82. 2. Rothern L, Aronheim A, Assaraf YG. Alterations in the expression of transcription factors and the reduced folate carrier as a novel mechanism of antifolate resistance in human leukemia cells. J Biol Chem 2003;278:8935–41. 3. Dervieux T, Orentas Lein D, Park G, Barham R, Smith K, Walsh M, et al. Single nucleotide polymorphisms (SNPs) in the folate/ purine synthesis pathway predict methotrexate’s effects in rheumatoid arthritis [abstract]. Arthritis Rheum 2003;48 Suppl 9:S438. 4. Norris MD, de Graaf D, Haber M, Kavallaris M, Madafiglio J, Gilbert J, et al. Involvement of MDR1 P-glycoprotein in multifactorial resistance to methotrexate. Int J Cancer 1996;65:613–9. 5. Yudoh K, Matsuno H, Nakazawa F, Yonezawa T, Kimura T. Increased expression of multidrug resistance of P-glycoprotein on Th1 cells correlates with drug resistance in rheumatoid arthritis. Arthritis Rheum 1999;42:2014–5. 6. Mickisch GH, Merlino GT, Galski H, Gottesman MM, Pastan I. Transgenic mice that express the human multidrug-resistance gene in bone marrow enable a rapid identification of agents that reverse drug resistance. Proc Natl Acad Sci U S A 1991;88:547–51. 7. Hider SL, Morgan C, Bell E, Bruce IN. Methotrexate is not a substrate for P-glycoprotein in patients with rheumatoid arthritis [abstract]. Ann Rheum Dis 2002;61 Suppl 1:199. 8. Hooijberg JH, Broxterman HJ, Kool M, Assaraf YG, Peters GJ, Noordhuis P, et al. Antifolate resistance mediated by the multidrug resistance proteins MRP1 and MRP2. Cancer Res 1999;59:2532–5. 9. Kool M, van der Linden M, de Haas M, Scheffer GL, de Vree JM, Smith AJ, et al. MRP3, an organic anion transporter able to transport anti-cancer drugs. Proc Natl Acad Sci U S A 1999;96: 6914–9. 672 LETTERS DOI 10.1002/art.20893 Reply To the Editor: I would like to thank Dr. Hider and colleagues for describing their interesting study, which demonstrates that there is no effect on MTX efflux from the cellular transport protein P-gp in an in vitro T lymphoblastoid cell line. The scientific and clinical communities will hopefully learn more in the coming years about the mechanisms of resistance to MTX. These potential advances in our understanding of MTX, together with new insights into the very real effects of singlenucleotide polymorphisms on several enzymes in the folate metabolic pathways, could lead to more effective and tailored treatment strategies. I would also like to thank Drs. Ranganathan and McLeod for their thoughtful comments regarding the possible role of MRPs in the phenomenon of MTX resistance. Their summary of the possible differential effects of MRPs with short- and long-term exposure to MTX, as well as the potential role of sulfinpyrazone and probenecid, is again noted as a rich area for future clinical investigation. I also agree with their observation that the role of MRPs in the dose range of MTX commonly used to treat RA is still controversial. Hopefully, we will see more attention and research focused on the role of MRPs, as well as their inhibitors, and how manipulation of their activity could affect the clinical response to MTX, when used to treat patients with RA. Our patients will benefit as more is learned about the mechanisms of this old, but excellent, drug. Joel M. Kremer, MD Albany Medical College Albany, NY DOI 10.1002/art.20812 Study of individual joint pathology in rheumatoid arthritis suggests a single pathology: comment on the editorial by Kirwan To the Editor: We read with interest the recent editorial promoting the concept that there are at least 2 pathologies involved in rheumatoid arthritis (RA), suggesting that there are separate processes for joint inflammation and direct joint destruction (1). This is an interesting concept, but it is predominantly based on studies that used conventional radiographic imaging for determination of joint destruction. The concept of 2 or more pathologies in RA stems from the radiographic observations that erosions progress despite the suppression of synovitis and that these processes are therefore uncoupled. However, we wish to highlight another interpretation of the data based on analysis of individual joint damage, and in particular to cite information from recent magnetic resonance imaging (MRI) studies which enable imaging of both synovitis and erosions. Most studies on RA therapies have evaluated summed groups of joints (tender and swollen joint counts or cumulative Sharp scores) or systemic markers of inflammation, such as erythrocyte sedimentation rate or C-reactive protein levels. Very few studies have examined inflammation at the individual joint level. Boers and colleagues (2) analyzed data from the COBRA (Combinatietherapie Bij Reumatoı̈de Artritis) study using clinical measurements of individual joints and radiographic progression of the same joints. That study demonstrated that local signs of RA at baseline and at 1-year followup strongly predicted radiographic progression of joint damage in the individual joint. It is now accepted that clinical examination may be relatively insensitive and that imaging studies have confirmed a discrepancy between clinical examination and MRI-detected synovitis (3). Even in a careful analysis of clinical examination and radiographic joint damage there is bound to be some degree of mismatch, and it is therefore conceivable that synovitis may be present at a subclinical level and may cause subsequent bone erosion. Indeed, the majority of RA patients with apparent clinical remission who are evaluated using sensitive imaging techniques have been demonstrated to have measurable synovitis (4), explaining the apparent contradiction of erosion still occurring in patients whose disease is clinically in remission (5). The best way to visualize the relationship between synovium and bone is to use the most sensitive imaging technique (MRI) to image individual joints (such as the metacarpophalangeal joints), rather than a complex structure (such as the wrist), and follow the relationship over time. Compounding the insensitivity of clinical examination is the number of time points at which synovitis is measured; the measurement of any potential fluctuation in synovitis is limited by the number of study assessments. It would also seem appropriate to concentrate on short-duration RA, in which secondary osteoarthritis is less likely to confound pathologic interpretation. We have previously reported a study of 40 patients with early RA (⬍12 months duration) who had baseline, 3-month, and 12-month MRI scans performed in the course of a randomized trial (6). In this study, no mismatch between MRI-detected synovitis and subsequent bone damage was observed. Where there was no synovitis, no subsequent MRI erosions developed. The number of new erosions developing at the individual joint level was proportional to the amount of synovitis in a given joint, even given the possible limitations of semiquantitative synovitis measurement. Certainly mechanical factors, such as the location of collateral ligaments, are important at the site of erosions (7), but synovitis is the predictor of erosion development. We do not intend to suggest that MRI should be used as an outcome measure in every RA trial, but we do think that MRI provides the best available tool for studying in vivo the intrajoint pathologic relationships in RA. Simultaneous imaging of the synovium and bone, at enough time points to understand the total synovial “load” on a given joint over time, demonstrates only a single process in RA subjects. In conclusion, we believe rheumatologists would be well served by focusing research and therapy on inflammation suppression as the mechanism for controlling joint damage in RA. LETTERS 673 Philip G. Conaghan, MBBS, FRACP, FRCP Dennis McGonagle, MB, PhD, FRCPI Ai Lyn Tan, MRCP Paul Emery, MA, MD, FRCP University of Leeds Leeds, UK 1. Kirwan JR. The synovium in rheumatoid arthritis: evidence for (at least) two pathologies [editorial]. Arthritis Rheum 2004;50:1–4. 2. Boers M, Kostense PJ, Verhoeven AC, van der Linden S, for the COBRA Trial Group. Inflammation and damage in an individual joint predict further damage in that joint in patients with early rheumatoid arthritis. Arthritis Rheum 2001;44:2242–6. 3. Goupille P, Roulot B, Akoka S, Avimadje AM, Garaud P, Naccache L, et al. Magnetic resonance imaging: a valuable method for the detection of synovial inflammation in rheumatoid arthritis. J Rheumatol 2001;28:35–40. 4. Brown AK, Quinn MA, Karim Z, Wakefield RJ, Conaghan PG, Pollard AS, et al. Neither the ACR remission criteria nor the Disease Activity Score accurately define true remission in rheumatoid arthritis [abstract]. Arthritis Rheum 2002;46 Suppl 9:S243. S243. 5. Molenaar ET, Voskuyl AE, Dinant HJ, Bezemer PD, Boers M, Dijkmans BA. Progression of radiologic damage in patients with rheumatoid arthritis in clinical remission. Arthritis Rheum 2004; 50:36–42. 6. 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. 7. Tan AL, Tanner SF, Conaghan PG, Radjenovic A, O’Connor P, Brown AK, et al. Role of metacarpophalangeal joint anatomic factors in the distribution of synovitis and bone erosion in early rheumatoid arthritis. Arthritis Rheum 2003;48:1214–22. DOI 10.1002/art.20894 Reply To the Editor: I thank Dr. Conaghan and colleagues for pointing out some of the evidence against the thesis that there are at least 2 pathologic processes underway in parallel in the joints of patients with RA. The definitions of “synovitis” and “erosions” are crucial to this debate. Conaghan et al take the view that findings seen on MRI, which they call “synovitis” and “erosions,” are the same as the pathologic activities of the same names. In their excellent and important work on MRI of joints in early arthritis, the Leeds group led by Dr. Emery has consistently found many more MRI “erosions” than radiographic erosions, and has argued that the large majority of patients with early RA have erosions when examined by MRI. In spite of this, many longitudinal studies have shown that a substantial proportion of hospital patients with RA never develop joint damage visible on conventional radiography. Nonerosive RA is well recognized. This calls into question the validity of labeling MRI “erosions” as the same pathology as radiographic erosions. Further, showing that there is a broad correlation between 2 aspects of RA, the presence of synovial inflammation and bony erosions, does not prove one causes the other. For example, we would not accept an assertion that pain in the toes causes erosions in the fingers—and yet the 2 are manifestly correlated in patients with RA. The argument that there are at least 2 parallel pathologic processes in the joints of patients with erosive RA is not watertight, but neither is it based on a single observation about controlling clinical signs while allowing radiographic progression. A second relevant observation is that glucocorticoids have the ability to almost stop the progression of joint erosions, even when given in low doses, which adds little in the long run to current antiinflammatory symptom control. A third observation is that RA synovium exhibits varying histology, and the different histologic abnormalities may relate to the subsequent progression or nonprogression of erosions in different ways. This wider range of evidence is reviewed in the original editorial. John R. Kirwan, BSc, MD, FRCP University of Bristol Bristol, UK DOI 10.1002/art.20866 Careful attention to blood sampling as a preanalytical determinant of circulating matrix metalloproteinase 9 to avoid misinterpretations: comment on the article by Ainiala et al To the Editor: Ainiala et al (1) described increased serum matrix metalloproteinase 9 (MMP-9) concentrations in systemic lupus erythematosus patients with neuropsychiatric complications compared with patients without neuropsychiatric manifestions. The authors concluded that elevated serum MMP-9 concentrations might reflect small-vessel cerebral vasculopathy in these patients. Although I do not have experience in that special field, I suggest that preanalytical problems of MMP determinations due to blood sampling should also be considered to avoid misinterpretation of the study results. Because blood sampling markedly determines the concentration of circulating MMP-9, I would direct the attention of Ainiala et al and that of the interested reader to this issue. The significance of blood collection as a preanalytical determinant appears to have escaped the attention of Ainiala et al, because these facts have been discussed primarily in analytical journals (2,3). A report comprising the results obtained in my laboratory is shown in Figure 1. Briefly, blood samples obtained from 8 healthy subjects were prepared in blood collection devices (Sarstedt, Numbrecht, Germany) by centrifugation at 1,600g for 15 minutes, within 30 minutes after venipuncture. Plastic tubes were used either without additives (S-Monovette system 01.1728; Sarstedt) to obtain pure serum (serum⫺) or with kaolin-coated plastic granulate (S-Monovette system 01.1601; Sarstedt) to obtain serum after enhanced clotting (serum⫹). Plasma samples were collected as citrate, heparin, or EDTA plasma in S-Monovette tubes coated with sodium 674 LETTERS Figure 1. Matrix metalloproteinase 9 (MMP-9) concentrations in serum and plasma. MMP-9 was measured in serum (left) and plasma (right) derived from blood samples obtained from 8 healthy adults. Serum was collected in Monovette tubes without additive (serum⫺) or tubes containing kaolin-coated granulates as clot activator (serum⫹). Plasma was collected in tubes coated with sodium citrate, lithium heparin, or potassium EDTA. Significance levels were calculated by repeated-measures analysis of variance with Tukey’s posterior test. Individual and median values are shown. a ⫽ P ⬍ 0.05 versus serum⫺; b ⫽ P ⬍ 0.05 versus serum⫹; c ⫽ P ⬍ 0.05 versus plasma citrate; d ⫽ P ⬍ 0.05 versus plasma heparin; e ⫽ P ⬍ 0.05 versus plasma EDTA. citrate, lithium heparin, or potassium EDTA. Measurements of MMP-9 were performed with an enzyme-linked immunosorbent assay (MP2211; Medac Diagnostika, Wedel, Germany). MMP-9 concentrations in serum samples collected in tubes with clot activator (serum⫹) were ⬃3-fold higher than those in pure serum samples (serum⫺) and essentially higher than the concentrations observed in plasma samples. Because platelets and leukocytes contain high concentrations of MMP-9, the varying release of MMP-9 from blood cells during the platelet activation or sampling process could cause these differences (3). In addition, changes in white blood cell count and increased apoptotic neutrophils in these patients are well known (4) and could also generate different serum MMP-9 concentrations. Therefore, the direct association of the changed MMP-9 concentration with the cerebral vasculopathy as postulated by Ainilia et al could be only detected with a blood collection method that avoids the preanalytical interferences. Serum samples collected neither with nor without clot activator seem to be appropriate to fulfill that precondition. In summary, in order to ascertain a relationship between MMPs in the peripheral blood and pathologic processes in tissue/organs and to use MMPs as diagnostic markers, the influence of blood sampling must be considered. Recently, the use of blood samples collected with sodium citrate was suggested to avoid the detrimental effect of other anticoagulants or serum and to optimize the diagnostic validity of MMPs in peripheral blood (3). Klaus Jung, MD, EurClinChem University Hospital Charité Berlin, Germany 1. Ainiala H, Hietaharju A, Dastidar P, Loukkola J, Lehtimaki T, Peltola J, et al. Increased serum matrix metalloproteinase 9 levels in systemic lupus erythematosus patients with neuropsychiatric manifestations and brain magnetic resonance imaging abnormalities. Arthritis Rheum 2004;50:858–65. 2. Jung K, Laube C, Lein M, Lichtinghagen R, Tschesche H, Schnorr D, et al. Kind of sample as preanalytical determinant of matrix metalloproteinases 2 and 9 and tissue inhibitor of metalloproteinase 2 in blood. Clin Chem 1998;44:1060–2. 3. Mannello F, Luchetti F, Canonico B, Papa S. Effect of anticoagulants and cell separation media as preanalytical determinants on zymographic analysis of plasma matrix metalloproteinases. Clin Chem 2003;49:1956–7. 4. Ren Y, Tang J, Mok MY, Chan AW, Wu A, Lau CS. Increased apoptotic neutrophils and macrophages and impaired macrophage phagocytic clearance of apoptotic neutrophils in systemic lupus erythematosus. Arthritis Rheum 2003;48:2888–97. DOI 10.1002/art.20895 Reply To the Editor: Dr. Jung points out that preanalytical handling of blood samples used for the determination of MMP-9 should be considered in the evaluation of the results of our study. Dr. Jung himself used plastic blood collection vessels with or without additives and measured serum or plasma MMP-9 by enzyme-linked immunosorbent assay. MMP-9 content was highest in serum from the tubes containing added beads, while serum obtained without beads gave lower values. Heparin or citrate plasma contained less MMP-9 than did serum, and EDTA plasma contained no MMP-9. Dr. Jung suggests that differences between the differently treated samples might be caused by variable release of MMP-9 from blood cells during the sampling process or platelet activation, because platelets and leukocytes contain MMP-9. In our study, blood was drawn into glass tubes (Vacutainer; Becton Dickinson, Oxford, UK) without additive, and the tubes were centrifuged within an hour after venipuncture for the assessment of serum MMP-9. Aliquots of the sera were stored at ⫺70°C until the time of analysis, in a freezer that was not used daily. All samples were processed in the same way, which makes us believe that the results are comparable. The absence of MMP-9 in Dr. Jung’s EDTA plasmas may be attributable to degradation of MMP-9 in the absence of calcium (Makowski GS, Ramsby ML. Use of citrate to minimize neutrophil matrix metalloproteinase-9 in human plasma. Anal Biochem 2003;322:283–6). Citrate will also bind ionized calcium, and heparin might affect the stability of MMP-9 in some way. It is apparent that procedures for the measurement of MMP-9 require standardization. We thank Dr. Jung for his valuable comments. Hanna Ainiala, MD Aki Hietaharju, MD, PhD Tampere University Hospital Tampere, Finland Seppo T. Nikkari, MD, PhD University of Tampere Tampere, Finland LETTERS 675 DOI 10.1002/art.20813 Accuracy of haplotype association studies is enhanced by increasing number of polymorphic loci examined: comment on the article by Meulenbelt et al To the Editor: We read with interest the article by Meulenbelt et al regarding haplotypes of the interleukin-1 (IL-1) gene cluster and osteoarthritis (OA) (1). These authors identified 8 haplotypes from 3 polymorphisms: IL1B ⫹3953, IL1B ⫺511, and the IL1RN variable-number tandem repeat (VNTR). They describe the frequency of 2 haplotypes (1-1-2 and 1-2-1) that confer an increased risk for the development of radiographic OA of the hip and 1 protective haplotype (2-1-1). We commend the study for its investigation of combined polymorphisms across both IL1B and IL1RN, 2 genes that are likely to have equal significance in the development of OA. However, the 3 polymorphisms examined represent only a limited sample of the known extended haplotypes within this region. We previously investigated the association between extended haplotypes within the IL-1 gene cluster and radiographic OA of the knee (2). Our study examined 8 loci across the IL1A, IL1B, and IL1RN region. Although single polymorphic markers revealed no association, we identified a common haplotype associated with significant increased risk of knee OA in 2 separate patient cohorts. In a smaller study, we also examined IL-1 haplotypes from UK patients with severe hip OA who underwent primary or revision hip replacement surgery. We reexamined the haplotypes (generated with the PHASE 2.0 program) (3,4) representing the 3 loci analyzed by Meulenbelt et al, and their predominant risk haplotype (1-1-2) was replicated in our data (odds ratio [OR] 3.87, 95% confidence interval [95% CI] 1.79–8.30, P ⫽ 0.00007). However, when our 8-locus extended haplotype was examined, we identified a common risk haplotype, 2CCTC2CT, (OR 24.81, 95% CI 6.53–101.72, P ⫽ 0.0000003) within this population that was extremely rare in healthy controls. This 2CCTC2CT haplotype includes the 3 alleles of the 1-1-2 risk haplotype (Table 1). We identified several other 8-locus haplotypes that included the 1-1-2 alleles, but these bore no significant risk for OA. The accuracy of haplotype association studies is en- Table 1. Association of IL-1 gene cluster with hip OA using 3-locus haplotype compared with 8-locus haplotype* Haplotype 3-locus† 1-1-2 Total 8-locus‡ 2CCTC2CT Total P Controls Hip OA 42 390 14 44 0.00007 4 390 9 44 0.0000003 * Values are the number. IL-1 ⫽ interleukin-1; OA ⫽ osteoarthritis. † IL1B ⫹3953, IL1B ⫺511, IL1RN variable-number tandem repeat (VNTR); 1 ⫽ most frequent alleles. ‡ Alleles arranged in the following order: IL1A TTA repeat, IL1A ⫺889, IL1B ⫹3953, IL1B ⫺31, IL1B ⫺511, IL1RN VNTR, IL1RN ⫹8006, IL1RN ⫹11100. Underlined alleles are equivalent to the 1-1-2 haplotype. hanced by increasing the number of polymorphic loci examined, particularly in genomic regions where linkage disequilibrium varies widely, as within the IL-1 gene cluster. We believe that the 2CCTC2CT haplotype we describe here represents a significant risk for the development of severe hip OA. To verify these findings, it would be of great interest to see the study by Meulenbelt et al extended by employing 8-locus haplotype analysis. Andrew J. P. Smith Christopher J. Elson, PhD Mark J. Perry, PhD Jeffrey L. Bidwell, PhD, FRCPath University of Bristol Bristol, UK 1. Meulenbelt I, Seymour AB, Nieuwland M, Huizinga TW, van Duijn CM, Slagboom PE. Association of interleukin-1 gene cluster with radiographic signs of osteoarthritis of the hip. Arthritis Rheum 2004;50:1179–86. 2. Smith AJ, Keen LJ, Billingham ME, Perry MJ, Elson CJ, Kirwan JR, et al. Extended haplotypes and linkage disequilibrium in the IL1R1-IL1A-IL1B-IL1RN gene cluster: association with knee osteoarthritis. Genes Immun 2004;5:451–60. 3. Stephens M, Smith NJ, Donnelly P. A new statistical method for haplotype reconstruction from population data. Am J Hum Genet 2001;68:978–89. 4. Stephens M, Donnelly P. A comparison of Bayesian methods for haplotype reconstruction from population genotype data. Am J Hum Genet 2003;73:1162–9. DOI 10.1002/art.20896 Reply To the Editor: We thank Smith and his colleagues for their interest in our observed predisposing association of 2 haplotypes (1-1-2 and 1-2-1), formed by polymorphisms IL1B ⫹3953, IL1B ⫺511, and IL1RN VNTR, with radiographic OA (ROA) of the hip. Interestingly, Smith et al not only confirmed the association of the 1-1-2 haplotype with severe, symptomatic hip OA among patients in the UK but also indicated an extended 8-locus haplotype (2CCTC2CT) across the IL1A, IL1B, and IL1RN genes that was extremely rare in controls and conferred a very high risk for symptomatic hip OA (OR 24.81, 95% CI 6.53–101.72). This 2CCTC2CT haplotype includes the 3 alleles of the 1-1-2 risk haplotype. To verify these findings, we genotyped 4 of the 5 additional polymorphisms suggested by Smith et al in our 55–65–year-old cohort from the population-based Rotterdam study (n ⫽ 791). Using the THESIAS 2.0 program (Tregouet DA, Escolano S. Tiret L, Mallet A, Golmard JL. A new algorithm for haplotype-based association analysis: the Stochastic-EM algorithm. Ann Hum Genet 2004;68:165–77), we tested the effect of the original haplotype (IL1B ⫹3953, IL1B ⫺511, and IL1RN VNTR, 1-1-2) and the extended haplotype (IL1A ⫺889, IL1B ⫹3953, IL1B ⫺31, IL1B ⫺511, and IL1RN VNTR, IL1RN ⫹8006, IL1RN ⫹11100, 1-1-1-1-22-1) in patients with hip ROA (n ⫽ 70) as compared with the 676 LETTERS population we studied, excluding patients with hip ROA (n ⫽ 721). For the original 3-locus haplotype (1-1-2), we observed an OR of 3.2 (95% CI 1.5–6.8); however, with the 7-locus extended haplotype (1-1-1-1-2-2-1, identical to the CCTC2CT haplotype identified in the UK) the risk for hip ROA increased to an OR of 7.3 (95% CI 2.7–19.5, P ⫽ 0.00008). We agree with Smith et al that increasing the number of polymorphic loci at this gene cluster enhanced the accuracy of our haplotype association study. The extended haplotype 1-1-1-1-2-2-1 encompassing the IL1A ⫺889, IL1B ⫹3953, IL1B ⫺31, IL1B ⫺511, and IL1RN VNTR, IL1RN ⫹8006, IL1RN ⫹11100 polymorphisms does, indeed, confer a significant risk for the development of both severe hip OA (in patients in the UK) and the more common hip ROA (in the Dutch population-based sample). The extended haplotype analysis with our data showed increased effects not only of the original 1-1-2 haplotype that was the focus of Smith and colleagues but also of the original 1-2-1 haplotype. This implies that both extended haplotypes harbor functionally relevant variation. Again, this gene cluster associates to hip OA. It would be relevant to know whether other OA phenotypes could also develop as a consequence of genetic variation at this locus. Ingrid Meulenbelt, PhD P. Eline Slagboom, PhD Leiden University Medical Center Leiden, The Netherlands Cornelia M. van Duijn, PhD Erasmus University Medical School Rotterdam, The Netherlands DOI 10.1002/art.20814 Detailing ethnicity and phenotypes is critical for pooling association studies: comment on the article by Huizinga et al To the Editor: In a recent special article, Dr. Huizinga and colleagues presented recommendations for genetic association studies in Arthritis & Rheumatism (1). Proper evaluation of genetic association studies is critical to our understanding and interpretation of such results, when reported. Thus, this article is timely and important, addressing some of the pitfalls encountered in genetic association studies and providing sound, logical guidelines to minimize the plethora of false-positive associations that have hampered this approach (2). A strategy that may provide a more accurate representation of the presence of genetic association is the pooling or meta-analysis of all results (published or not). For such an approach to be feasible, all articles must provide enough information to confidently ascertain the ethnicity, the mode of ascertainment of subjects, the criterian used to diagnose the phenotype, and various disease manifestations. Whenever possible, standard or widely accepted definitions should be used to increase uniformity among various studies. With the growing confidence in genotyping methodology, the greatest variability in pooling results will now likely occur at the population and phenotype levels. Thus, to effectively perform meta-analysis, detailed information regarding the heritage/ ethnicity of the population being studied and assurances regarding the quality of the phenotypic data are critical. Such data are best ascertained by clinicians who have considerable experience with the diagnosis of interest (3). Thus, articles being considered for publication should ensure that the phenotyping is of utmost quality, and that standard criteria have been used to define phenotypes whenever possible. Proton Rahman, MD, MSc, FRCPC Memorial University of Newfoundland St. John’s, Newfoundland, Canada 1. Huizinga TW, Pisetsky DS, Kimberly RP. Associations, populations, and the truth: recommendations for genetic association studies in Arthritis & Rheumatism. Arthritis Rheum 2004;50: 2066–71. 2. Hirschhorn JN, Lohmueller K, Byrne E, Hirschhorn K. A comprehensive review of genetic association studies. Genet Med 2002;4: 45–61. 3. Endicott J. Good diagnoses require good diagnosticians: collecting and integrating the data. Am J Med Genet 2001;105:48–9. DOI 10.1002/art.20897 Reply To the Editor: We thank Dr. Rahman for his positive comments. For meta-analysis, it is necessary that data on putative variation on the population level are given, hence our recommendation to avoid hidden population stratification. One of the major steps forward in the science of rheumatology has been the correct classification of patients by well-studied criteria, so we have not explicitly stressed the relevance of correct phenotypic classification. Nevertheless, we completely agree with Dr. Rahman that the quality of phenotypic information is a key factor in determining the quality of articles on the phenotype–genotype association. Tom W. J. Huizinga, MD, PhD Leiden University Medical Center Leiden, The Netherlands David S. Pisetsky, MD, PhD Durham VA Hospital and Duke University Medical Center Durham, North Carolina Robert P. Kimberly, MD University of Alabama at Birmingham DOI 10.1002/art.20845 Corrected QT interval in anti-SSA–positive adults with connective tissue disease: comment on the article by Lazzerini et al To the Editor: Because previous reports have shown a prolongation of the corrected QT (QTc) interval in anti-SSA–positive children LETTERS without congenital heart block (CHB) (1,2), Lazzerini et al recently investigated the same issue in adults (3). They reported a significant prolongation of the mean QTc interval in adult patients with anti-SSA–positive connective tissue diseases (CTDs) compared with controls (anti-SSA–negative patients with CTDs): the mean ⫾ SD QTc interval was 445 ⫾ 21 in 31 anti-SSA–positive patients versus 419 ⫾ 17 msec in 26 anti-SSA–negative patients (P ⫽ 0.000005 by Student’s t-test). This result suggests that anti-SSA–positive patients may be at increased risk for cardiovascular mortality. We have studied electrocardiograms of 89 adults with CTD. QTc interval duration was compared in patients with and patients without anti-SSA antibodies. The anti-SSA–positive group consisted of 32 patients (29 women and 3 men; mean ⫾ SD age 37 ⫾ 11 years); 15 were positive for both anti–60-kd SSA and anti–52-kd SSA, 15 were positive for anti–60-kd SSA, and 2 were positive for anti–52-kd SSA. Six patients were also positive for anti-SSB antibodies. Twenty-eight of these patients had systemic lupus erythematosus (SLE) and 4 had primary Sjögren’s syndrome (SS). The anti-SSA–negative group (controls) consisted of 57 patients (54 women and 3 men; age 38 ⫾ 12 years) who were negative for both anti-SSA and anti-SSB antibodies. Forty-nine had SLE, 4 had undifferentiated CTD, and 4 had mixed CTD. The QT interval was measured manually with a digitizing pad (SummaSketch II Professional MMII 1812; SummaGraphics, Seymour, CT) connected to a PC computer, by an investigator who was blinded to anti-SSA status. QT interval was corrected for heart rate by the Bazett formula, to yield the QTc value. The anti-SSA–positive and anti-SSA–negative groups were very similar with regard to age and CTD. All of the patients were treated with hydroxychloroquine. None was taking other drugs that could influence the QTc interval (3). The mean ⫾ SD QTc was 409 ⫾ 30 msec in the anti-SSA– positive group and 409 ⫾ 28 msec in the anti-SSA–negative group (P ⫽ 0.78 by Mann-Whitney U test). Five of the anti-SSA–positive patients (16%) and 6 of the anti-SSA– negative patients (11%) had QTc values above the upper limit of normal (440 msec) (P ⫽ 0.35 by Fisher’s exact test). Our data thus do not confirm the findings of Lazzerini et al. They are, however, in accordance with those reported by Gordon et al (4). Those authors studied electrocardiograms in adults with anti-SSA antibodies. No statistically significant difference in QTc was found between a group of 49 anti-SSA– positive patients (mean ⫾ SD 411 ⫾ 19 msec), a group of 62 anti-SSA–negative patients (403 ⫾ 24 msec), and a group of 19 anti-SSA–positive mothers of children with CHB (408 ⫾ 19 msec) (4). It should be noted that our study focused only on surface electrocardiography whereas Lazzerini et al performed other tests more appropriate for determining possible autonomic dysfunction. Nevertheless, our findings were not consistent with theirs but were in accordance with those reported by Gordon et al (4). Another limitation of our study is the marked preponderance of SLE patients in both groups, whereas in the study by Lazzerini et al, 48% of the patients in the anti-SSA– positive group had SS and 65% of those in the anti-SSA– negative group had systemic sclerosis. It remains to be confirmed whether patients with SS and anti-SSA antibodies might indeed have prolonged QTc compared with anti-SSA–positive patients with other CTDs. 677 Nathalie Costedoat-Chalumeau, MD Zahir Amoura, MD Jean-Sébastien Hulot, MD Pascale Ghillani, MD Philippe Lechat, MD Centre Hospitalier Universitaire Pitié-Salpêtrière Christian Funck-Brentano, MD Centre Hospitalier Universitaire Saint Antoine Jean-Charles Piette, MD Centre Hospitalier Universitaire Pitié-Salpêtrière Paris, France 1. Cimaz R, Stramba-Badiale M, Brucato A, Catelli L, Panzeri P, Meroni PL. QT interval prolongation in asymptomatic anti-SSA/ Ro–positive infants without congenital heart block. Arthritis Rheum 2000;43:1049–53. 2. Gordon PA, Khamashta MA, Hughes GR, Rosenthal E. Increase in the heart rate–corrected QT interval in children of anti-Ro–positive mothers, with a further increase in those with siblings with congenital heart block: comment on the article by Cimaz et al [letter]. Arthritis Rheum 2001;44:242. 3. Lazzerini PE, Acampa M, Guideri F, Capecchi PL, Campanella V, Morozzi G, et al. Prolongation of the corrected QT interval in adult patients with anti-Ro/SSA–positive connective tissue diseases. Arthritis Rheum 2004;50:1248–52. 4. Gordon PA, Rosenthal E, Khamashta MA, Hughes GR. Absence of conduction defects in the electrocardiograms of mothers with children with congenital complete heart block. J Rheumatol 2001; 28:366–9. DOI 10.1002/art.20898 Reply To the Editor: Costedoat-Chalumeau et al describe their study of QTc interval duration in 89 adults with CTD (32 anti-SSA– positive and 57 anti-SSA–negative), in which they found no significant differences between the 2 groups. As they note, these data do not seem to confirm the results of our study, in which we showed significant prolongation of the mean QTc interval in 31 anti-SSA–positive patients compared with 26 anti-SSA–negative patients. Possible explanations for the differing results in the 2 studies may arise from the following considerations. Costedoat-Chalumeau et al studied a very selected cohort of CTD patients. Their study population consisted almost exclusively of patients with SLE (close to 90% in each group). Similarly, in the study by Gordon et al that CostedoatChalumeau and colleagues cite (Gordon PA, Rosenthal E, Khamashta MA, Hughes GR. Absence of conduction defects in the electrocardiograms of mothers with children with congenital complete heart block. J Rheumatol 2001;28:366–9), there was a less extreme but still significant preponderance of SLE patients (⬃70% overall). As a consequence, the absence of significant differences in the QTc interval duration between the anti-SSA–positive and anti-SSA–negative subjects may be related to a peculiar “resistance” of SLE patients to the hypothesized electrophysiologic effects of these antibodies. Indeed, it should be noted that in Gordon and colleagues’ 678 LETTERS study, in which the prevalence of SLE was not as high as in the study by Costedoat-Chalumeau et al, the QTc in the antiSSA–positive group was reported to be longer than in the anti-SSA–negative group and this difference, although nonsignificant, approached significance (P ⫽ 0.063). In our study there was also a prevalence, albeit less marked and partially related to epidemiologic factors, of specific forms of CTD, i.e., SS in the anti-SSA–positive group (48%) and systemic sclerosis (SSc) in the anti-SSA–negative group (65%). Costedoat-Chalumeau et al hypothesize that SS patients may have peculiar characteristics compared with patients who have other CTD, more frequently developing QTc interval prolongation in the presence of anti-SSA antibodies. This is a plausible hypothesis, especially considering the results of a recent study by Pirildar et al (Pirildar T, Sekuri C, Utuk O, Kemal Tezcan UK. QT dispersion in rheumatoid arthritis patients with and without Sjögren’s syndrome. Clin Rheumatol 2003;22:225–8), in which the authors showed that QT dispersion and corrected QT dispersion values were significantly longer in rheumatoid arthritis (RA) patients with secondary SS than in RA patients without secondary SS. In that report, the behavior of the QTc interval was not mentioned and the number of anti-SSA–positive patients was very small (7 of 58 [12%]). Nevertheless, these data may reflect an unusual “vulnerability” of cardiac repolarization in patients with SS, possibly leading to a prolongation of the QTc in the presence of anti-SSA positivity. However, when we divided the anti-SSA–positive patients in our study into those with SS (n ⫽ 15) and those with other CTD (n ⫽ 16 [6 with SLE, 5 with undifferentiated CTD, 4 with SSc, and 1 with mixed CTD]), we found that the QTc interval was prolonged in both groups, with no significant difference (mean ⫾ SD 446 ⫾ 24 msec versus 444 ⫾ 17 msec, respectively; P ⫽ 0.77 by Student’s t-test for unpaired data); there also was no significant difference in the proportion of patients with QTc intervals above the upper limit of normal (i.e., 440 msec) in the SS group versus the non-SS group (75% versus 50%). In conclusion, based on the present data, we cannot clearly define the role of anti-SSA antibodies in ventricular repolarization in patients with CTD. However, it is likely that the clinical setting in which these antibodies act may be of relevance, with a possible peculiar “vulnerability” of SS patients or a particular “resistance” of SLE patients, or both. In order to further clarify this, the need for study of a large cohort of patients, including a sufficient number of comparable anti-SSA–positive patients with different forms of CTD, seems crucial. Such a study is now in progress at our institution. Pietro Enea Lazzerini, MD Maurizio Acampa, MD Mauro Galeazzi, MD Franco Laghi-Pasini, MD University of Siena Siena, Italy Errata DOI 10.1002/art.20899 In the article by Palao et al published in the September 2004 issue of Arthritis & Rheumatism (pp 2803–2810), there was an error in the labeling of the lanes in Figure 5b. The lane labeled “AS-FLIP” should have been labeled “NS” and vice versa. DOI 10.1002/art.20929 In the article by del Rincón et al published in the December 2004 issue of Arthritis & Rheumatism (pp 3813–3822), there was an error in the text on page 3818. The last full sentence on that page should have read, “Arterial incompressibility was present in 75 (12%) of these patients, and obstruction or [not ‘and’] incompressibility occurred in 155 (24%) of the patients.” We regret the errors.