Patients and control individuals
A cohort of 238 patients was recruited from community physicians to an early synovitis study at the US National Institutes of Health (protocol 94-AR-194). The referrals were derived from over 20 different physicians, most of whom were rheumatologists. All patients had undergone a preliminary rheumatologic evaluation before study entry. Inclusion in the study was based on the presence of persistent synovitis in at least one peripheral joint, which had been present for at least 6 months but less than 1 year. Patients with traumatic, septic, and crystal-induced arthritis were specifically excluded. Patients with well defined diffuse connective tissue diseases such as systemic lupus erythematosus and scleroderma were also excluded.
A tender joint count was determined by assessing for the presence of joint line and/or stress tenderness in 68 peripheral joints, and a swollen joint count was determined by evaluating for effusion and/or synovial thickening in 66 peripheral joints (hips were excluded). The total number of affected joints was calculated based on the presence of either tenderness or swelling in each of the joints examined. Sacroiliitis was defined on the basis of having a history of persistent inflammatory low back or buttock pain in conjunction with tenderness over the sacroiliac joint(s). Radiographic confirmation of sacroiliitis was sought but was not considered to be an essential part of the definition. Enthesitis was considered to be present if the insertion of the Achilles' tendon or the plantar fascia was either swollen or tender on examination, and the patient complained of persistent pain in the hind foot area.
RF was measured using nephelometry (in which a level >20 IU/ml was considered positive) and using enzyme-linked immunosorbent assay (ELISA) as previously described [13, 14]. Anteroposterior and lateral radiographs of the hands, wrists, feet, knees, and other affected joints were either available for evaluation or were obtained at the time of assessment. Radiographs were available for analysis in 196 out of the 238 patients in the cohort. The radiographs were evaluated for the presence of erosions by an experienced musculoskeletal radiologist. Patients were deemed to have an erosive arthropathy if one or more definite erosions were demonstrable in any peripheral joint radiograph. No attempt was made to quantify the degree of erosive damage when present. Patients with clinical evidence of sacroiliitis were evaluated using anteroposterior and oblique views of the sacroiliac joints.
The American College of Rheumatology criteria for RA [18] and the European Spondlyarthropathy Study Group criteria [19] were applied to each member of the cohort, based on the clinical, radiographic, and laboratory data obtained. The European Spondlyarthropathy Study Group criteria were slightly modified by considering patients as having an 'asymmetric' arthropathy if they did not meet the American College of Rheumatology symmetry criterion for RA. Because patients were recruited on the basis of having peripheral joint synovitis and not axial disease, the presence of 'inflammatory spinal pain', when present, was insufficient as the only major European Spondlyarthropathy Study Group criterion. Also, alternating buttock pain and sacroiliitis were regarded as one minor criterion rather than two. Patients not fulfilling either set of criteria were classified as having UA for the purposes of the present study, even if a more specific rheumatic diagnosis was suggested clinically. A group of 20 normal individuals (age 32 ± 9.5 years; 14 females and 6 males) served as a control group, and data from that group were used to assign a positive cutoff point only. For the analyses throughout the study, patient groups were compared with each other, and thus the lower age of this normal group was not deemed to be a confounding factor.
Quantification of IgG-AGE in high-molecular-weight complexes
A solid-phase aminophenyl boronic acid (APB)-ELISA [15] was used to measure the IgG-AGE in the high-molecular-weight complexes, which were isolated from serum by a polyethylene glycol precipitation method. Sera were made to a 2.5% final concentration with polyethylene glycol 8000 and incubated for 16 hours at 4°C. After centrifugation at 13 000 g for 15 min, the supernatant was discarded and the precipitate was resuspended back to the original serum volume with phosphate-buffered saline (PBS). The IgG-AGE was measured by ELISA of the AGE proteins that were captured via cis-diols to the solid-phase immobilized APB. APB (Sigma, Oakville, Ontario, Canada) 2 mg/ml in 0.2 mol/l carbonate/bicarbonate buffer (pH 9.4) was reacted with Reacti-Bind maleic anhydride activated polystyrene 96-well plates (Pierce, Rockford, IL, USA) for 16 hours at 37°C. Plates were washed with EPPS buffer (0.15 mol/l NaCl, 0.02 mol/l EPPS [Sigma], and 0.01 mol/l l) MgCl2; pH 8.6) three times. The test samples (100 μl) from the 2.5% polyethylene glycol precipitate (diluted 1/500–1/4000 as necessary to keep the values within the standard curve), positive and negative controls, and an appropriate standard curve using IgG1-AGE (0.625–10 μg/ml), all diluted in EPPS buffer and in duplicate, were incubated for 1 hour at 37°C. After washing the plates three times with PBS/0.1% Tween 20, the plates were blocked with 100 μl 1% goat serum in PBS/Tween 20 for 1 hour at 37°C. The plates were washed three times with PBS/Tween 20.
To detect specifically the bound IgG, 100 μl horse radish peroxidase conjugated F(ab')2 fragments of goat antihuman IgG (heavy chain specific; Jackson, BioCan, Mississauga, Otario, Canada) diluted 1/20000 in PBS/Tween were added and the plates were incubated for 1 hour at 37°C. After washing the plates three times with PBS/Tween, the substrate o-phenylene diamine was added. The reaction was stopped by the addition of 4 mol/l H2SO4 approximately 30 min later. The optical density at 492 nm (reference 690 nm) was measured using an ELISA plate reader (SLT LabInstruments, Fisher, Montreal, Quebec, Canada). The plates could be regenerated once for reuse by a series of washes. These were, in sequence, elution of the AGE-modified proteins with 0.1 mol/l sorbitol (American Chemicals Inc., Montreal, Quebec, Canada; two elutions of 5 min incubation, and then one rinse at room temperature), and then four washes with 0.02 mol/l NaOH, followed by five washes with 0.05 mol/l acetic acid and then 10 washes with distilled water. Plates were stored between uses in distilled water containing a bactericidal agent, namely 0.02% Proclin (Superlco, Sigma). The cutoff for identifying those with elevated levels of IgG-AGE was the mean plus 2 SD of IgG-AGE levels from 20 normal individuals.
Measurement of antibodies to IgG-AGE
IgM and IgA anti-IgG-AGE antibodies were detected in serum or plasma by ELISA, as previously described [13, 14], with the testing laboratory being blinded to the diagnosis. IgG of all four subclasses, which were fully glycated in vitro, were used at a concentration of 2 μg/ml to coat the wells of an enzyme immunoassay plate (ICN, Montreal, Quebec, Canada). After washing the plates, the sera or plasma, diluted 1:1000 in duplicate, were incubated in the AGE-modified immunoglobulin-coated wells for 2 hours at 37°C. After washing the plates in PBS/Tween (0.1%), the bound antibodies were detected using peroxidase conjugated F(ab')2 fragments of antihuman IgM, or IgA (Jackson) diluted 1/10000 in PBS/Tween. To follow the reactivity over time, and keep consistent results, serum from one normal individual (approximating the mean reactivity) and a positive control was tested each time the assay was performed. After washing the plates three times with PBS/Tween, the substrate o-phenylene diamine was added. The reaction was stopped by the addition of 4 M H2SO4 approximately 30 min later. The optical density at 492 nm (reference 690 nm) was measured, using an ELISA plate reader (SLT LabInstruments). Cutoff values were determined from the sera of 20 normal control individuals, and were the mean plus 2 SD. In order to standardize the optical densities over time, the experimental values obtained were corrected to the positive control that was included in every assay. For selected sera, the titers for reactivity against the IgG-AGE as well as against bovine serum albumin with and without AGE damage were determined.
Because IgG1, IgG2, and IgG4 are structurally very similar, and in general when a sera was positive against IgG1-AGE it was also reactive against IgG2-AGE as well as IgG4-AGE, for clarity only the results for the response to is structurally IgG1-AGE are presented. Because IgG3 quite different (60 amino acid long hinge region versus the 12–15 amino acid hinge in IgG1, IgG2, and IgG4, along with several other differences in the heavy chain), the data for the immune response to IgG3-AGE are also presented. In preliminary studies (not shown), investigations into the specificity of the assay for RFs were done. Sera containing RF and anti-IgG-AGE antibodies were preincubated with either an AGE [Nε-(carboxymethly)lysine] or IgG (which lacked AGE damage) to see if the reactivity could be blocked in the subsequent ELISA. We found that IgG at 1 mg/ml inhibited 30% of the binding of polyclonal RF from one patient with RA to IgG, but it did not inhibit the binding of polyclonal anti-IgG-AGE antibodies to IgG-AGE. Higher concentrations of IgG (4 mg/ml) inhibited the binding of polyclonal RF by up to 90%, and anti-IgG-AGE by 30%, but this could have been because the polyclonal IgG might have had AGE on it. Preincubation with Nε-(carboxymethly)lysine (50 mg/ml) inhibited the binding of polyclonal anti-IgG-AGE antibodies to IgG-AGE by 40% but did not inhibit RF binding to IgG. Lysine at the same concentration did not inhibit either. Thus, this assay appears to measure anti-IgG-AGE antibodies and not simply RFs.
Statistical analysis
Patient groups were compared using analysis of variance or the Kruskal–Wallis test for continuous variables, and using the χ2 test for proportions. The significance levels were adjusted for multiple comparisons using the Bonferroni method where applicable. Statistical analyses were performed using Epi Info statistical software (Centers for Disease Control and Prevention, Atlanta, GA, USA; http://www.cdc.gov/epo/epi/epiinfo.htm) and SPSS statistical software (SPSS Inc., Chicago, IL, USA).