Systemic lupus erythematosus (SLE) is a systemic autoimmune disease that is characterized by autoantibody production and inflammatory disease involving multiple organs. Premature atherosclerosis is a common complication of SLE and results in substantial morbidity and mortality from cardiovascular disease (CVD). The reasons for the premature atherosclerosis in SLE are incompletely understood, although chronic inflammation is thought to play an important role. There is currently no known preventative treatment of premature atherosclerosis in SLE. Mycophenolate mofetil (MMF) is an immunosuppressive agent that is commonly used for treatment of patients with SLE. In order to study the impact of this drug on murine lupus disease including premature atherosclerosis development, we treated gld.apoE−/− mice, a model of SLE and accelerated atherosclerosis, with MMF. We maintained seven-week old gld.apoE−/− mice on a high cholesterol Western diet with or without MMF. After 12 weeks on diet, mice receiving MMF showed decreased atherosclerotic lesion area compared to the control group. MMF treatment also improved the lupus phenotype, indicated by a significant decrease circulating autoantibody levels and ameliorating lupus nephritis associated with this model. This data suggests that the effects of MMF on the immune system may not only be beneficial for lupus, but also for inflammation driving lupus-associated atherosclerosis.
Citation: Richez C, Richards RJ, Duffau P, Weitzner Z, Andry CD, Rifkin IR, et al. (2013) The Effect of Mycophenolate Mofetil on Disease Development in the gld.apoE−/− Mouse Model of Accelerated Atherosclerosis and Systemic Lupus Erythematosus. PLoS ONE 8(4): e61042. https://doi.org/10.1371/journal.pone.0061042
Editor: Christine Beeton, Baylor College of Medicine, United States of America
Received: October 23, 2012; Accepted: March 5, 2013; Published: April 8, 2013
Copyright: © 2013 Richez et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported in part by a research grant from Aspreva Pharmaceuticals to I.R.; and by grant: K01 AR055965-02 from NIAMS/NIH to T.A. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIAMS or NIH. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: This study was funded in part by Aspreva Pharmaceuticals. However, this does not alter the authors' adherence to all the PLOS ONE policies on sharing data and materials.
Systemic lupus erythematosus (SLE) is a complex systemic autoimmune disease involving multiple organs that is characterized by autoantibody production and chronic inflammation . Over time, management of SLE patients has improved and life expectancy of these patients has increased to reach a 10-year survival rate about 70% . However, several studies have revealed that atherosclerosis-attributed vascular events are significantly more frequent in these surviving lupus patients, compared to age-related individuals without SLE , .
Atherosclerosis is characterized by a chronic inflammatory state where immune cell activity is linked to plaque formation and remodeling . A plaque is formed within the lumen of medium- and large-sized arteries due to physiological imbalances caused by chronic inflammation; the plaque is described as a progressive accumulation of lipid, inflammatory cells, smooth muscle cells, and connective tissue within the intima of arteries . It has become widely accepted that atherosclerosis is an inflammatory disease, and that the immune system plays a pivotal role in disease development. Therefore, it is reasonable to suggest that the chronic inflammatory condition encountered in SLE and the activation of immune cells may predispose patients to an increased risk of premature atherosclerosis leading to cardiovascular disease (CVD). For these reasons, immunomodulatory therapy might be of benefit in ameliorating atherosclerosis in patients with SLE. However, with the exception of hydroxychloroquine  and some statins , the availability of beneficial treatments to decrease CVD risk in SLE is limited.
Mycophenolate mofetil (MMF) is an immunosuppressive drug used in the treatment of patients with SLE, particularly those with nephritis . It is also approved to prevent transplant rejection, especially in heart and kidney transplantation. MMF is an ester pro-drug which is metabolized in the body to the active compound mycophenolic acid (MPA). MPA is a noncompetitive inhibitor of a rate-limiting purine biosynthetic enzyme, inosine-5′-monophosphate dehydrogenase (IMPDH). IMPDH is involved in de novo synthesis of purines, and lymphocytes rely exclusively on this de novo pathway for nucleotide synthesis , . Therefore, MMF selectively targets lymphocyte proliferation. Importantly, MMF has been shown to reduce immune-mediated vascular injury in transplantation-associated atherosclerosis (known as coronary allograft vasculopathy)  and to attenuate plaque inflammation in patients with symptomatic carotid artery stenosis . These findings further suggest a potential role for MMF in the treatment of atherosclerosis.
In the study presented here, we utilized a mouse model that displays synergy between lupus and atherosclerosis . The gld.apoE−/− mouse model incorporates the gld inactivating mutation in Fas ligand (FasL) which develops lupus-like autoimmunity together with splenomegaly and lymphadenopathy; and the apoE−/− strain which spontaneously displays increased plasma levels of cholesterol and triglycerides and the development of atherosclerosis, particularly when mice are given a high cholesterol “Western diet”. In the study presented here, we used the gld.apoE−/− mouse model to reflect the accelerated atherosclerosis that occurs in patients with immune disorders to determine if MMF is effective in the treatment of lupus-associated atherosclerosis.
Animals and Study Protocol
The gld.apoE−/− mice used in this study were obtained by crossing gld and apoE−/− mice as previously described . Starting at 7 weeks of age, the mice were maintained on Adjusted Calories Western diet (21% fat) (#88137, Harlan-Teklad) for 12 weeks, supplemented with 200 mg/kg/day mycophenolate mofetil (MMF) (n = 8) or not supplemented (control) (n = 8). The concentration was chosen based on an equivalent mouse dosage value converted from a human dosage of 2000 mg/day, using equations presented by Ng . MMF was incorporated into our experimental Western diet by Harlan-Teklad Special Diets using CellCept® Oral Suspension (Roche Laboratories) and the conversion ratios we provided. This study was carried out in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by the Institutional Animal Care and Use Committee of Boston University School of Medicine (Protocol number: AN-14843). All efforts were made to minimize suffering.
Calculation of Drug Dosage in Experimental Groups
The amount of drug incorporated into the diet of each experimental group was based on established dosages of MMF used in human lupus patients. An average dose is 2000–3000 mg MMF/day, therefore for the studies presented here, we chose 2000 mg. The human dosage was converted into an equivalent dosage appropriate for mice, as metabolism of drugs differs greatly between humans and mice. Equations presented by Ng  were used to convert drug dosages among species of the same Class (in this case, placental mammals). First, the minimum energy cost (MEC) was established for the murine model according to the equation: ; where K is a constant of 70 (for placental mammals), and BW is the body weight of the model animal (average mouse body weight is 0.025 kg). Therefore, the MECmouse value was calculated to be 4.4. Next, the MECmouse value was converted into a corresponding drug dosage. The next conversion step is to calculate a Universal MEC (UMEC) based on the MEC of a human (1694, a value established in many pharmocokinetic studies ) and the established human dose of 2000 mg/day. The equation to calculate this is: ; where MEChuman is 1694 and the drug dose is 2000 mg/day, yielding a UMEC value of 1.18. An appropriate murine dosage equivalent to the 2000 mg/day human dosage may be calculated, by using the same equation to calculate UMEC, instead using the MEC value of a mouse: ; where MECmouse is 4.4 and the UMEC is 1.18, giving a murine drug dosage of 5.2 mg/day. By incorporating mouse body weight (0.025 kg) into this calculation, a value of 208 mg/kg/day is obtained. This dosage is equivalent to the 2000 mg/day human dosage (28.57 mg/kg/day, assuming a 70 kg human). According to the manufacturer’s product insert, there are 35 g of pure drug in the 110 g CellCept® oral suspension powder. Assuming that mice eat 5 g of chow per day and that the average body weight of a mouse is 0.025 kg, a ratio of CellCept® oral suspension powder that should be incorporated into each kilogram of Harlan Teklad Western diet was calculated using a pharmacokinetic equation provided by Sedgwick  and Gibson .
Analysis of Atherosclerosis, Splenomegaly, Lymphadenopathy, and Body Weight
After 12 weeks of Western diet, the final body weights were recorded, the mice were euthanized, and the spleen and submandibular lymph nodes were removed and weighed. The heart containing the aortic root, was frozen in OCT. Frozen sections of aortic root were stained with Oil Red O solution and microphotographs were taken and values of total atherosclerotic lesion area were analyzed using Adobe Photoshop 7.0.
Paraffin embedded tissue was sectioned (5 µm) and slides were stained with hematoxylin and eosin. Cross-sectional areas of at least 25 glomeruli were measured in each animal using computer-assisted pixel counting (Photoshop CS3; Adobe).
Circulating anti-nuclear antibodies (ANA) were measured by immunofluorescence using HEp-2 coated slides (The Binding Site Inc., San Diego, California). Slides were incubated for 1 h with serial log-scale dilutions (1∶100 to 1∶90000; as previously described by Komori, et al.)  of mouse serum, washed in PBS, and then incubated with FITC-labeled goat anti-mouse IgG (whole molecule; Sigma-Aldrich, St. Louis, Missouri). Slides were viewed using fluorescent microscopy. Serum cholesterol levels were determined using a total cholesterol microtiter procedure according to the manufacturers instructions (Wako Diagnostics, Richmond VA).
Body Weight and Food Intake
MMF has an excellent oral bioavailability. However, this treatment frequently induces gastrointestinal symptoms in humans , which could decrease drug intake. In order to ensure that the mice in each experimental group were ingesting similar amounts of food, the amount of food ingested was recorded weekly over the course of the experiment, and total ingestion of each group is presented in Figure 1A. Total body weights of the mice were also measured at intervals throughout the experiment (Figure 1B). There was no difference between the mice treated with MMF and control mice with regard to total food intake or body weight.
The Development of Splenomegaly, Lymphadenopathy and Autoantibodies are Inhibited by MMF Treatment
After 12 weeks treatment, the spleen and submandibular lymph nodes of the gld.apoE−/− mice were removed and weighed. The mice receiving MMF had a significantly smaller spleen weight compared to mice not receiving MMF (control): 0.41±0.08 g versus 0.63±0.05 g in the MMF group (p = 0.02, using Mann Whitney test) (Figure 2A), as well as a lower, but not significant, lymph node weight compared to control (mean ± SEM): 0.23±0.02 g versus 0.44±0.09 g, respectively (p = 0.09) (Figure 2B). To determine if MMF effects were due to a modification of lymphocyte population, we performed immunohistochemical staining for T-cells and B-cells in lymph node sections from treated and untreated mice. Similar to a previous report , no significant effect was observed in the appearance or distribution of T- or B-cells within the lymph nodes regardless of treatment (data not shown). We then examined the effect of MMF dose on autoantibody production. In humans, a decrease in autoantibody titers has been shown during the course of MMF therapy . We found a similar result in our lupus-associated atherosclerosis mouse model with a significant decrease of anti-nuclear antibody (ANA) titer in the MMF group compared to the controls (median [IQR]): 1.5 [1.5–2] versus 4 [4–5.5], respectively (p = 0,001) (Figure 2C).
A, Spleen from mice, treated with 200 mg/kg/day MMF or untreated, was harvested and weighed (*, p<0.05). B, The lymph nodes were also harvested and weighed. C, ANA titer was determined from serum samples using HEp-2-coated slides and scored using the value of the last positive dilution (**, p = 0.001). Data are expressed as means ± SEM.
Improvement of Nephritis in gld.apoE−/− Mice Receiving MMF
MMF is currently used as induction therapy, maintenance therapy, or both for patients with lupus nephritis , , , , . Although B6 mice deficient in Fas (lpr) or FasL (gld) display mild or no kidney disease , the lack of apoE associated with the gld mutation induces significant kidney disease . We assessed the extent of renal damage associated with the lupus phenotype, in control versus the MMF treated experimental gld.apoE−/− group. Glomerular tuft size and cell count were significantly decreased in gld.apoE−/− after treatment with MMF (respectively, p = 0.002 and p = 0.004) (Figure 3A–C). Of note, infiltration of inflammatory cells as well as modest crescent formation was observed in control mice, but not in MMF treated mice. Our findings demonstrate the efficacy of MMF on lupus parameters in our mouse model, as previously described in human SLE and other mouse models.
The mice were treated with MMF or control for 12 wk. A, Representative H&E-stained sections of kidney in both groups. Glomerular tuft size (B) (*, p = 0.002) and cell count (C) (**, p = 0.004) were measured by computer-assisted pixel counting. Values shown are the mean ± SEM.
MMF Decreases the Severity of Atherosclerosis in gld.apoE−/− Mice
To determine the effect of MMF on atherosclerotic parameters, the aortic root of each mouse was stained with Oil red O to reveal areas of atherosclerotic lesion after 12 weeks on Western diet (WD), (Figure 4A). A significant decrease in lesion area was observed in the MMF-treated mice compared to control (Figure 4B) (p = 0.02). Serum levels of total cholesterol were also examined to determine whether the protective effects of MMF on atherosclerosis severity could be explained by differences in serum cholesterol levels. However, there was no detectable difference between mice treated with MMF compared to control: 716±63 mg/dl versus 669±32 mg/dl, respectively (p = 0.38) (Figure 4C).
Aortic root atherosclerotic lesion area in MMF-treated or control mice. A, Representative photographs of aortic root stained with Oil Red O from mice maintained on Western diet for 12 wk and treated with 200 mg/kg/day MMF or untreated. B, Atherosclerotic lesion area of Oil Red O-stained aortae was quantified in both groups (*, p = 0.02). C, Total serum cholesterol was quantified in control mice and mice treated with 200 mg/kg/day MMF.
The purpose of this experiment was to investigate the effects of mycophenolate mofetil (MMF), on development of premature atherosclerosis in a murine mouse model of accelerated atherosclerosis and systemic lupus erythematosus (SLE). The gld.apoE−/− model is ideal to use for this experiment based on previous findings of synergistic disease presentation of both SLE and atherosclerosis in mice . Our study investigated the effect of a physiologically relevant dose of MMF on disease development. The 200 mg/kg/day concentration of MMF in mouse diet is roughly equivalent to the 2000 mg/day dose of MMF approved for use in human renal transplant prophylaxis and commonly used for the treatment of human SLE. Severity of cardiovascular disease was assessed by quantifying the atherosclerotic lesion area in the aortic root. To assess the hallmarks of SLE associated with the gld.apoE−/− model, spleen and submandibular lymph node were weighed. As such, the data show that the dosage of 200 mg/kg/day yields a significant ameliorating effect on atherosclerosis, splenomegaly and lymphadenopathy presentation. These data suggest that MMF treatment of patients with SLE could not only be beneficial to lupus, but also decrease the risk of cardiovascular disease.
With advances in medical care, the quality of life has improved and survival rate has increased for patients with SLE . However, with this increased survival rate, there is also a correlated increase in CVD. Chronic inflammation has been implicated as a contributing factor for the development of premature atherosclerosis in SLE patients , , , . Therefore, there is an interest in using anti-inflammatory or immunomodulatory therapies for this condition. MMF, an immunosuppressive agent, is currently used for the treatment of SLE patients, particularly those with kidney involvement, as well as for the prevention of rejection in transplant patients.
MMF was first demonstrated to inhibit T-cell function, however, MMF also exerts inhibitory effects on other immune cells and effectors, including downregulation of cell adhesion molecules and attenuation of monocyte and macrophage responses . MMF has also been shown to suppress a number of the inflammatory events that are involved in the development of atherosclerosis. T-lymphocyte infiltration to atherosclerotic plaque and circulation to sites of inflammation is abrogated by MMF treatment . MMF has been shown to reduce the expression of vascular adhesion molecules in atherosclerosis by inhibiting the nuclear factor NFκB which is required for their transcriptional upregulation . Raisanen et al  showed that MMF treatment reduces the appearance and proliferation of smooth muscle cells in the intima, which normally contribute to atherosclerotic plaque formation by recruitment of extracellular matrix and self-proliferation. Thus, MMF has properties that could be considered anti-atherogenic: inhibiting T-cells, blocking leukocyte adhesion and inhibiting proliferation of smooth muscle cells; therefore making it a potentially valuable drug to prevent the development of atherosclerosis in patients with SLE.
A recent publication examined the effect of MMF on atherosclerosis development by using bone marrow transplantation to generate a mouse model of lupus with associated atherosclerosis . van Leuven et al. report that the LDLr−/−. Sle mouse shows decreased atherosclerosis after MMF treatment, although lesion area does not decrease with atorvastatin treatment alone. However, despite the beneficial effects to atherosclerosis, there are no discernable differences in circulating autoantibody levels or kidney pathology. While it is known that B6 mice deficient in Fas (lpr) or FasL (gld) display mild or no kidney disease, the gld.apoE−/− model develops glomerular tuft enlargement and significant inflammatory cell infiltrate to the kidney . The results presented in the current study show a decrease in both lupus-disease and cardiovascular disease. It is reasonable to suggest that differences in the disease models utilized (Ldlr−/− versus apoE−/−, respectively; and Sle1.2.3 versus gld, respectively) or the dosage (40 mg/kg/day versus 200 mg/kg/day) could have contributed to the contrasting results.
Substantial evidence exists to support a critical role of inflammation in the pathology of both SLE and CVD caused by atherosclerosis. The chronic inflammation associated with SLE correlates to the increased risk in CVD seen in patients , , . The current study suggests that MMF, in addition to this well-known efficacy on lupus nephritis, might be a promising agent for the prevention of atherosclerosis in SLE. A more extensive analysis of atherosclerosis and SLE in these MMF-treated gld.apoE−/− mice will need to be performed to fully determine how MMF impacts the accelerated atherosclerosis of our lupus mouse model and potentially bring a better understanding of the link between atherogenesis and autoimmune disease.
Conceived and designed the experiments: IRR CDA TA. Performed the experiments: CR RR TA ZW. Analyzed the data: CR PD. Wrote the paper: IRR CR TA.
- 1. Tsokos GC (2011) Systemic lupus erythematosus. N Engl J Med 365: 2110–2121.
- 2. Pons-Estel GJ, Alarcon GS, Scofield L, Reinlib L, Cooper GS (2010) Understanding the epidemiology and progression of systemic lupus erythematosus. Semin Arthritis Rheum 39: 257–268.
- 3. Westerweel PE, Luyten RK, Koomans HA, Derksen RH, Verhaar MC (2007) Premature atherosclerotic cardiovascular disease in systemic lupus erythematosus. Arthritis Rheum 56: 1384–1396.
- 4. Urowitz MB, Gladman D, Ibanez D, Bae SC, Sanchez-Guerrero J, et al. (2010) Atherosclerotic vascular events in a multinational inception cohort of systemic lupus erythematosus. Arthritis Care Res (Hoboken) 62: 881–887.
- 5. Galkina E, Ley K (2009) Immune and inflammatory mechanisms of atherosclerosis. Annu Rev Immunol 27: 165–197.
- 6. Ross R (1999) Atherosclerosis–an inflammatory disease. N Engl J Med 340: 115–126.
- 7. Jung H, Bobba R, Su J, Shariati-Sarabi Z, Gladman DD, et al. (2010) The protective effect of antimalarial drugs on thrombovascular events in systemic lupus erythematosus. Arthritis Rheum 62: 863–868.
- 8. Ferreira GA, Navarro TP, Telles RW, Andrade LE, Sato EI (2007) Atorvastatin therapy improves endothelial-dependent vasodilation in patients with systemic lupus erythematosus: an 8 weeks controlled trial. Rheumatology (Oxford) 46: 1560–1565.
- 9. Dooley MA, Jayne D, Ginzler EM, Isenberg D, Olsen NJ, et al. (2011) Mycophenolate versus azathioprine as maintenance therapy for lupus nephritis. N Engl J Med 365: 1886–1895.
- 10. Allison AC, Eugui EM (1993) Immunosuppressive and other effects of mycophenolic acid and an ester prodrug, mycophenolate mofetil. Immunol Rev 136: 5–28.
- 11. Franklin TJ, Cook JM (1969) The inhibition of nucleic acid synthesis by mycophenolic acid. Biochem J 113: 515–524.
- 12. Gibson WT, Hayden MR (2007) Mycophenolate mofetil and atherosclerosis: results of animal and human studies. Ann N Y Acad Sci 1110: 209–221.
- 13. van Leuven SI, van Wijk DF, Volger OL, de Vries JP, van der Loos CM, et al. (2010) Mycophenolate mofetil attenuates plaque inflammation in patients with symptomatic carotid artery stenosis. Atherosclerosis 211: 231–236.
- 14. Aprahamian T, Rifkin I, Bonegio R, Hugel B, Freyssinet JM, et al. (2004) Impaired clearance of apoptotic cells promotes synergy between atherogenesis and autoimmune disease. J Exp Med 199: 1121–1131.
- 15. Ng R (2004) Drugs development and preclinical studies. In: Sons JW, editor. Drugs: From Discovery to Approval. 107–138.
- 16. Sedgwick C, Pokras MA (1988) Extrapolating rational drug doses and treatment periods by allometric scaling. Proceedings of the 55th Annual Meeting of the American Animal Hospital Association. Washington, D. C: 156–157.
- 17. Komori H, Furukawa H, Mori S, Ito MR, Terada M, et al. (2006) A signal adaptor SLAM-associated protein regulates spontaneous autoimmunity and Fas-dependent lymphoproliferation in MRL-Faslpr lupus mice. J Immunol 176: 395–400.
- 18. Epinette WW, Parker CM, Jones EL, Greist MC (1987) Mycophenolic acid for psoriasis. A review of pharmacology, long-term efficacy, and safety. J Am Acad Dermatol 17: 962–971.
- 19. Van Bruggen MC, Walgreen B, Rijke TP, Berden JH (1998) Attenuation of murine lupus nephritis by mycophenolate mofetil. J Am Soc Nephrol 9: 1407–1415.
- 20. Appel GB, Contreras G, Dooley MA, Ginzler EM, Isenberg D, et al. (2009) Mycophenolate mofetil versus cyclophosphamide for induction treatment of lupus nephritis. J Am Soc Nephrol 20: 1103–1112.
- 21. Chan TM, Li FK, Tang CS, Wong RW, Fang GX, et al. (2000) Efficacy of mycophenolate mofetil in patients with diffuse proliferative lupus nephritis. Hong Kong-Guangzhou Nephrology Study Group. N Engl J Med 343: 1156–1162.
- 22. Ginzler EM, Dooley MA, Aranow C, Kim MY, Buyon J, et al. (2005) Mycophenolate mofetil or intravenous cyclophosphamide for lupus nephritis. N Engl J Med 353: 2219–2228.
- 23. Ong LM, Hooi LS, Lim TO, Goh BL, Ahmad G, et al. (2005) Randomized controlled trial of pulse intravenous cyclophosphamide versus mycophenolate mofetil in the induction therapy of proliferative lupus nephritis. Nephrology (Carlton) 10: 504–510.
- 24. Kelley VE, Roths JB (1985) Interaction of mutant lpr gene with background strain influences renal disease. Clin Immunol Immunopathol 37: 220–229.
- 25. Lockshin MD, Salmon JE, Roman MJ (2001) Atherosclerosis and lupus: a work in progress. Arthritis Rheum 44: 2215–2217.
- 26. Manzi S, Meilahn EN, Rairie JE, Conte CG, Medsger TA Jr, et al. (1997) Age-specific incidence rates of myocardial infarction and angina in women with systemic lupus erythematosus: comparison with the Framingham Study. Am J Epidemiol 145: 408–415.
- 27. Riboldi P, Gerosa M, Luzzana C, Catelli L (2002) Cardiac involvement in systemic autoimmune diseases. Clin Rev Allergy Immunol 23: 247–261.
- 28. van Leuven SI, Kastelein JJ, Allison AC, Hayden MR, Stroes ES (2006) Mycophenolate mofetil (MMF): firing at the atherosclerotic plaque from different angles? Cardiovasc Res 69: 341–347.
- 29. Huang HD, Liu ZH, Zhu XJ, Chen ZH, Li LS (2002) Inhibition of mycophenolic acid on NF-kappaB activity in human endothelial cells. Acta Pharmacol Sin 23: 649–653.
- 30. Raisanen-Sokolowski A, Vuoristo P, Myllarniemi M, Yilmaz S, Kallio E, et al. (1995) Mycophenolate mofetil (MMF, RS-61443) inhibits inflammation and smooth muscle cell proliferation in rat aortic allografts. Transpl Immunol 3: 342–351.
- 31. van Leuven SI, Mendez-Fernandez YV, Wilhelm AJ, Wade NS, Gabriel CL, et al. (2012) Mycophenolate mofetil but not atorvastatin attenuates atherosclerosis in lupus-prone LDLr(−/−) mice. Ann Rheum Dis 71: 408–414.