Subjects and visits
Patients were from a well characterized cohort of 100 consecutive ACL-injured subjects prospectively recruited at the Lund University Hospital between 1985 and 1989 (35). All 100 subjects had a complete ACL tear and were within 18 days after initial trauma assessed by arthroscopy and x-ray with no significant signs of pre-existing knee OA (Figure 1a and 1b). The participants were treated with early physiotherapeutic knee rehabilitation without primary ACL reconstruction. Synovial fluid was collected early after injury (called acute visit; 0 to 18 days) and prospectively at 1 to 5 visits during the following 7.5 years (Figure 1b). For another study with the purpose to examine the association between knee cartilage quality and knee function, 32 subjects without ACL reconstruction or radiographic signs of OA at the 16-year follow-up (described below) were examined with dGEMRIC 20 years after their ACL injury (36). Since the dGEMRIC method is reliant on the presence of joint cartilage, only subjects having Osteoarthritis Research Society International (OARSI, (37)) atlas grades of ≤ 1 were included in the study. Twenty-five of the 32 subjects examined with dGEMRIC had one or more available synovial fluid sample aspirated following their injury and were included in this study (Figures 1a and 1b, Table 1).
Radiography at the 16 year follow up
Radiographs at the 16 year (range 11-18 years) follow up were obtained in standardized standing anteroposterior knee position with both knees in 20 degrees of flexion and weight bearing on a tilt table; a fluoroscopically positioned x-ray beam was used to optimize medial tibial plateau alignment. The radiographs were independently read by two observers blinded to clinical details. Joint space narrowing (JSN) and osteophytes were graded independently on frontal images on a 4-point scale (range 0-3, 0 = no evidence of JSN or bony change) according to the OARSI atlas (14, 15, 37). The interrater reliability (kappa statistic) was Κ = 0.78 for JSN and Κ = 0.52 for osteophytes (38).
Synovial fluid sampling
Twenty-five subjects were included in this study with any kind of synovial fluid samples, i.e. either from first and/or following visit(s) as follows: 20 subjects had their synovial fluid aspirated at the acute visit within 18 days (median 6 days) after injury, and 22 subjects had their synovial fluids collected at between one and five visits during the subsequent 7.5 years of follow-up (median 4 years); these synovial fluids are called chronic samples (Figure 1b, Table 1). The subjects visited the orthopedic outpatient ward only for study purposes (35, 38). All synovial fluids were collected without joint lavage, and the samples were centrifuged at 3000xg for 10 minutes in room temperature and supernatants were stored at -80ºC.
Molecular marker analyses in synovial fluid
sGAG, in synovial fluid mainly chondroitin and keratan sulfate (CS and KS), was quantified by Alcian Blue precipitation (39). Two different aggrecan epitopes were quantified using immunoassays and the monoclonal antibodies (mAb) 1-F21 and OA-1. According to previous publications, mAb 1-F21 is suggested to recognize a protein sequence within or close to the KS region of aggrecan (18, 40). mAb OA-1 recognizes the ARGS neoepitope generated by aggrecanase cleavage at the TEGE392/393ARGS site in the interglobular domain of aggrecan (41). Cartilage oligomeric matrix protein (COMP) was quantified using a commercial assay from AnaMar AB/IDS (cat. no. AN-14-1006-71); the AnaMar COMP-epitope has not been published. Matrix metalloproteinase-3 (MMP-3) and tissue inhibitor of metalloproteinase-1 (TIMP-1) were quantified using monoclonal and polyclonal antibodies; the MMP-3 immuno-assay recognizes both the pro- and active form of the protease and the complex with TIMP; the TIMP-1 immuno-assay detects only free TIMP-1 (42-44). Data on ARGS-aggrecan was generated for this study, all other biomarker data were available from previous studies on the described ACL cohort (45, 46).
The ratio MMP-3/TIMP-1 was used to investigate differences in these biomarkers alone or as a ratio between the enzyme and its inhibitor. We further investigated the ratios of sGAG/COMP, ARGS-aggrecan/COMP and 1-F21 aggrecan/COMP as biomarkers; ratios like these have been suggested to minimize the influence of varying amounts of obtainable synovial fluid (47).
Assessment with dGEMRIC at the 20 year follow up
Subjects were investigated with dGEMRIC on average 20.6 years (range between 18 and 23 years) after the ACL injury (Figure 1b, Table 1). Briefly, Gd-DTPA2- (MagnevistÒ, Schering AG, Berlin, Germany) was injected intravenously at a dose of 0.3 mmol/kg body weight. To optimize the uptake of Gd-DTPA2- into the cartilage, subjects exercised by walking up and down the stairs for approximately ten minutes, starting five minutes after injection. Two hours after injection, post-contrast imaging of the cartilage was performed using a standard 1.5 T MRI system with a dedicated knee coil (Magnetom Vision; Siemens Medical Solutions, Erlangen, Germany). Central parts of the weight-bearing lateral and medial femoral cartilage were identified, and quantitative relaxation time calculations were performed in a 3 mm thick sagittal slice on each condyle, using sets of six turbo inversion recovery images with different inversion times: TR = 2000 ms, TE = 15 ms, FoV 120 x 120 mm2, matrix = 256 x 256, TI = 50, 100, 200, 400, 800 and 1600 ms. A full-thickness region of interest (ROI) in the cartilage was examined. T1Gd was calculated using the mean signal intensity from each ROI (48), and the dGEMRIC images were analyzed and ROIs were drawn using the MATLAB-based Mokkula software (26). An orthopaedic surgeon performed the ROI measurements. All MRI data was available from a previous study (36).
Western blot of aggrecan
Aggrecan fragments from synovial fluid (pooled from 47 subjects with knee OA or knee injury) were purified by mini-preparations of cesium-chloride density-gradient centrifugation in absence or presence of guanidinium chloride, collecting the associative A1 and dissociative D1 fractions, as described (49). Purified aggrecan (i.e. A1D1 fraction prepared from pooled knee cartilage from ten subjects with OA) was in vitro digested using aggrecanase-1 (ADAMTS-4, a disintegrin and metalloproteinase with thrombospondin motifs-4) or MMP-3 as described (50). The samples were deglycosylated and separated by SDS-PAGE on 3-8% Tris-acetate mini-gels and transferred to PVDF-membranes (39). For the immune-reaction we used antibodies against aggrecan G1-domain (Affinity BioReagents no. PA1-1747, polyclonal IgG diluted 1:400), 1-F21 aggrecan epitope (IgG monoclonal antibody diluted 1:75000), ARGS-aggrecan epitope (IgG monoclonal neoepitope antibody OA-1 diluted to 5.3 µg/ml) and chondroitin sulfate clone 3B3 (Seikagaku no. 270789 IgM monoclonal antibody against chondroitinase treated chondroitin 6-sulfate diluted to 0.33 µg/ml). Secondary antibodies were peroxidase-conjugated horse anti-mouse IgG (CST no. 7076S diluted to 10 ng/ml), goat anti-mouse IgM (Sigma no. 8786 diluted to 10 ng/ml) and goat anti rabbit IgG (KPL no. 074-1516 diluted to 13 ng/ml). The immunobands were visualized using Pierce ECL Plus Western Blotting Substrate (no. 32132) and film (Amersham Hyperfilm ECL) or luminescence image analyser Bio-Rad ChemiDoc MP.
Statistical analysis
Associations between the molecular biomarkers and dGEMRIC T1Gd values were investigated using linear regression models with adjustments for age at injury, sex, body mass index at dGEMRIC examination and time between injury and biomarker sampling. Results from crude (without adjustments) linear regression analyses are presented as a supplement (Table S1). Mann-Whitney tests were used for comparison of biomarker values between acute and chronic subject groups. For correlation analysis Spearman’s rank (rS) was used. For subjects with more than one chronic sample, the average biomarker concentration and the average time after injury were used in the linear regression model. The dGEMRIC values were normally distributed. Biomarker data were log10 transformed to obtain normal distribution. To be able to compare effect sizes between biomarkers, we report standardized effects from the linear regression analyses. The reported effects estimate how many standard deviations the dependent variable (dGEMRIC) will change per standard deviation increase in the predictor variable (biomarker concentration). All tests were 2-tailed and P ≤ 0.05 was considered statistically significant. The statistical analysis was performed with SPSS 24.0 for Windows software package.