One-hundred-and-sixty RA patients were included in the multicenter, randomized, double-blinded, parallel-group, placebo-controlled CIMESTRA trial 2324. Briefly, patients fulfilled the American College of Rheumatology 1987 revised criteria for RA 25. Further, the patients appeared with active disease less than six months, less than or equal to two swollen joints at baseline, and were aged 18 to 75 years 2324. Health Assessment Questionnaire (HAQ score, 0 to 3) 26, Visual Analogue Scale (0 to 10) (VAS pain, global and doctor) and Disease Activity Score in 28 joints (DAS28) 27 were calculated. Fourteen-hundred-and-seventy-six healthy twin-individuals aged 18 to 67 years served as controls 22. The trial was approved by the local ethics committee (j. nr M1959-98) and fulfilled the Declaration of Helsinki and the International Conference on Harmonisation 1996 revised guidelines for Good Clinical Practice (j.nr NCT00209859). Signed informed consent was obtained from all study participants.

The treatment protocol compared methotrexate (MTX) plus cyclosporine vs. MTX plus placebo. During the first eight weeks patients were assessed fortnightly and every four weeks thereafter. Subsequently, whenever synovitis was present MTX dose was escalated by 2.5 mg from 7.5 mg/week to maximum 20 mg/week followed by a stepwise cyclosporine/placebo-cyclosporine increment (0.5 mg/kg) every four weeks from 2.5 mg/kg to maximum 4.0 mg/kg. In addition, intra-articular betamethasone (7 mg/l) was injected into swollen joints at any visit (maximum four joints or 4 ml per visit). During the second year, hydroxychloroquine (200 mg/day) was added and cyclosporine/placebo was tapered to zero, while MTX was continued 2324. During the open extension study from three to four years the treatment strategy continued to aim at tight synovitis control. Oral glucocorticoids were allowed in the open extension study.

Serum was obtained from routinely drawn non-fasting blood samples collected between 08.00 a.m. to 2.00 p.m. Samples were allowed to clot at room temperature followed by centrifugation at 3,000 × g for 10 minutes. Sera were stored at -80°C.

SP-D was measured at baseline, after two weeks, one and six months, and after one, two, three and four years using a five-layered sandwich ELISA as previously described 19. In controls, SP-D was only measured at baseline. All analyses were done in duplicate and serial samples from the same patient were analyzed simultaneously. The inter-assay coefficients of variation were 3.5 and 3.8% for low (367 ng/ml) and high (2,470 ng/ml) quality controls, respectively, and the intra-assay coefficients of variation were 1.7% for both quality controls. C-reactive protein (CRP) (mg/l) and erythrocyte sedimentation rate (ESR) (mm/hour) were assayed by standard methods. IgM-rheumatoid factor (IgM-RF) (cut-off level < 16 IU/ml) and anti-CCP (cut-off level < 24 U/ml) (Euro Diagnostica AB, Malmö, Sweden) were measured by ELISA as previously described 282930.

Radiographs of hands, wrists, and forefeet were obtained at baseline (n = 155), and annually thereafter. After four years 137 radiographs were available, but only 133 patients had radiographs available at baseline and at four years. Radiographs were scored according to Sharp-van der Heijde by an independent senior radiologist who was aware of the sequence of x-ray recordings 31. The annual estimated progression rate in total Sharp-van der Heijde Score (TSS), Joint Space Narrowing score (JSN) and erosion score (ES) was calculated according to disease duration and TSS, JSN and ES at baseline for each patient 32. Radiographic progression was defined as the smallest detectable difference from baseline (= one unit).

Corresponding serum and synovial fluid samples were available from 20 RA patients with joint effusions before treatment. Synovial fluid was collected by aseptic technique before injection of glucocorticoid and stored at -80°C. Before analysis, the samples were centrifuged 30 minutes at 400 × g and subsequently the supernatant was incubated four hour at 37°C with bovine testicular hyaluronidase (Sigma H3884, St Louis, MO, USA) to reduce viscosity (2 μl hyaluronidase (1 mg/ml in 0.2 M TRIS, 0.1 M sodium acetate, pH 7.0) to 300 μl synovial fluid). Subsequently, they were centrifuged at 20.000 × g for 10 minutes at 4°C. The supernatant was assayed for SP-D by ELISA. The possible trapping of SP-D in the synovial fluid pellet was studied by incubating the pellet with ethylenediaminetetraacetic acid (EDTA) 0.52 M in a TRIS-buffered saline (TBS) buffer (pH 7.4) at 37°C in 30 minutes followed by centrifugation in four minutes at 20.000 × g and 4°C. A total of 50 μl of the resulting supernatant was re-calcified with 60 μl of 1 M CaCl₂, and pH was adjusted to 7.9 by adding 28.5 μl 1 M TRIS pH 8.6 prior to analysis.

Gel filtration chromatography was done on available synovial fluid samples (n = 11) and corresponding sera. Hyaluronidase-treated samples (200 μl) were applied to an analytical Superose 6 column connected to a fast-performance liquid chromatography system (former Amersham Biosciences, now GE Healthcare, Uppsale, Sweden) using TBS (pH 7.4) containing 10 mM EDTA and 0.05% emulphogen as eluent at a flow rate of 24 ml/hr. Fractions of 0.2 ml were collected and quantified by the SP-D ELISA. SP-D was eluted as two structurally different forms with high and low molecular weight (SP-D multimers (fraction 10 to 18) and SP-D trimers (fraction 24 to 38)). Size chromatography on healthy serum followed by SDS-PAGE and Western blotting has yielded protein bands at > 250 kDa for multimeric SP-D, and 90 kDa, 43 kDa and 40 kDa for trimeric SP-D 1319.

Genomic DNA was isolated from EDTA stabilized whole blood. Applied Biosystems (Assay-by-design) (Foster City, California, USA) designed primers and probes for the non-synonymous substitutions of DNA-bases of the SP-D gene resulting in the Met/Thr variant. The genotyping procedure has been described previously 13. Human leucocyte antigen (HLA)-DRB1 genotyping for shared epitope (SE) was performed by polymerase chain reaction-based sequence-specific oligonucleotide probing, as described elsewhere 3334. Herein, we define the shared epitope as the presence of HLA-DRB1*04 and/or HLADRB1*01 and/or HLADRB1*10.

All statistical analyses were conducted using STATA version 9.2 (StataCorp, College Station, Texas, USA). Comparisons between groups were done by Mann-Whitney U-test or Fischer's Exact Test, and if analysing more than two groups, Kruskal-Wallis test was used. Spearman Rank Correlation analysis was applied when appropriate. Comparison between patients and controls was performed using linear regression models, where control twins were clustered in pairs. Linear regression was also applied in the prospective analysis of SP-D in RA patients, where repeated measurements in the individual patient were clustered. We used logistic regression to assess whether baseline SP-D could predict radiographic progression after four years with adjustment for gender, age, smoking, anti-CCP and radiographic status at baseline. Robust estimation of standard error was calculated. To approximate a normal distribution, SP-D was logarithmically transformed when used as continuous, dependent variable in linear regression analyses.

One individual from each healthy twin pair was used for genotype and allele frequency estimation. The genotype frequencies were tested for Hardy-Weinberg equilibrium by (χ²-analysis. Comparisons of genotype and allele frequencies in patients and controls were performed by logistic regression with adjustment for gender and age or by Fishers Exact test.

Since SP-D did not differ between treatment arms, data from all RA patients were pooled. Analysis was by intention-to treat (N = 142). Completers' analysis was also performed and gave similar results (data not shown). Results are presented as median (95% confidence interval) if not otherwise stated. P-values ≤ 0.05 and P ≤ 0.01 were considered significant with single and multiple testings, respectively.

Of 160 patients included, 61 (38%) did not complete the four-year protocol. The reasons for drop-out were adverse events (11), treatment failure (10), patients' request (13) and other (27). Fifty-six (35%) left the study during the first two years. Patients who dropped out did not differ from completers with regard to demographic and clinical variables at baseline (data not shown). At baseline one patient had serum SP-D of 8,106 ng/ml. This patient subsequently developed severe pulmonary fibrosis and was excluded from the statistical analyses.

The demographic characteristics of the RA patients at baseline and the control population are shown in Table 1. Among the 142 patients included in the intention-to-treat analyses, all data for composite disease activity measures were available in 134 individuals. Seventy-eight percent, 66% and 69% had achieved ACR50, ACR70 and DAS28 < 2.6 after four years.

Including patients with radiographs available at both baseline and after four years (N = 133), 53%, 23% and 49% progressed radiographically according to TSS, JSN and ES score, respectively. Of note, however, radiographic progression at four-year follow-up was small in terms of Sharp/van der Heijde units (median (iqr): TSS 2 (0 to 7) to 5 (0 to 11), JSN 0 (0 to 2) to 0 (0 to 4) and ES 2 (0 to 5) to 3 (0 to 8)).

Baseline SP-D in RA patients was 693 ng/ml (649; 770) vs. 913 ng/ml (879; 945) in controls (P < 0.001). This difference persisted after adjustment for age, gender and current smoking status (P < 0.001) and was also present at four years after adjustment for confounders (P < 0.001). Compared to baseline, SP-D had increased in RA patients at four years (893 ng/ml [810; 1013] vs. 693 ng/ml [649; 770], P < 0.001) even when adjusting for gender, age and smoking status (p < 0.001). However, at four years, SP-D was still lower in RA patients as compared to controls with adjustment for confounders (P < 0.001). There was no significant correlation between age and SP-D in the RA population (rho = 0.06, P = 0.42). Likewise, there was no significant gender difference among RA patients. In contrast, SP-D increased significantly with age in healthy subjects (rho = 0.21, P < 0.001), and control males had significantly higher levels of SP-D compared to females (Table 2). Both RA and control smokers had significantly higher SP-D than non-smokers (Table 2). Disease activity markers and HAQ score were inversely correlated to SP-D at baseline (CRP: rho = -0.30, P < 0.001, DAS28: rho = -0.23, P = 0.003 and HAQ: rho = -0.21, P = 0.008). No significant difference in SP-D at baseline was observed between patients with respect to anti-CCP, IgM-RF status or any SE present (P = 0.50, P = 0.14, and P = 0.24, respectively). Furthermore, SP-D did not differ between smoking SE positive vs. non-smoking SE positive patients (P = 0.13).

The CRP change from baseline to four years (Δ) correlated inversely to the SP-D change (ΔCRP vs. ΔSP-D, rho = -0.39 and P < 0.001). We found no association between SP-D and radiographic data including estimated annual progression rate (data not shown). Baseline SP-D did not predict radiographic progression (Total Sharp score) at four years (P = 0.46)

Synovial fluid was obtained from 20 patients at baseline. Median SP-D in synovial fluid was 275 ng/ml (221; 299). SP-D in corresponding sera was 678 ng/ml (592; 829). SP-D in synovial fluid and serum levels correlated significantly (rho = 0.69, P < 0.001), Figure 1. Synovial fluid SP-D was not significantly associated with sex, age, CRP, autoantibodies, any SE or radiographic findings (data not shown). There was no detectable SP-D in the debris enriched pellets resulting from centrifugation of the synovial fluid. Results from the gel filtration chromatography are outlined in Figure 2. Multimeric SP-D was barely detectable in synovial fluid as compared to serum, where both multimeric and trimeric molecular variant SP-D (trimeric subunits) were detected.

The Met11Thr polymorphism was in Hardy-Weinberg equilibrium in both RA and controls (data not shown). The distribution of genotypes and allele frequencies is presented in Table 4. When adjusting for gender and age, there was no overrepresentation of Thr11Thr in RA patients as compared with controls (Table 3). Circulating SP-D did not differ between genotypes in RA patients, whereas healthy individuals with the Thr11Thr genotype appeared with the lowest level as previously reported 22. The genotypes were not associated with specific disease features including DAS28, CRP, joint counts, auto-antibodies, HAQ or x-ray findings (data not shown). The Met11Thr allelic variation could neither predict x-ray progression nor disease activity outcome after four years and the size distribution of SP-D in synovial fluid did not differ between genotypes (data not shown).