- Email: [email protected]
Higher bone turnover detected by serum levels of tartrate-resistant acid phosphatase 5b (TRACP-5b) is a risk for the radiographic medial knee joint space narrowing in men in early forties without knee pain the three years prospective observational study
Abstracts / Osteoarthritis and Cartilage 24 (2016) S63eS534
growing number of evidence suggested altered levels of adipokines in OA pathogenesis. Fatty acid binding protein 4 (FABP4) is one of the most abundant intracellular lipid transport proteins produced by mature adipocytes. Reports showed that FABP4 could enhance the production of pro-inflammatory cytokines by forming a positive feedback loop. FABP4 has been documented to play an important role in metabolic syndrome such as type II diabetes mellitus and atherosclerosis. However there is very limited knowledge about FABP4 in OA. This study aimed to investigate the role of FABP4 in the pathogenesis of OA. Firstly we wanted to describe the pattern of FABP4 in OA patients by determining FABP4 levels in local and systemic fat and body fluid and compared with non-OA individuals. Secondly we carried out an animal study to see if FABP4 knockout mice are protected from OA. Methods: 1. Human study 75 individuals were recruited including 45 OA cases that undergone total knee replacement (TKR) and 30 non-OA controls that received arthroscopic surgery. Blood was taken one day before the surgery. Subcutaneous adipose tissue (ScAT), synovial fluid (SF) and infrapatellar fat pad (IFP) from OA and control subjects were harvested during surgery (TKR or arthroscopy). Explant culture of ScAT and IFP was performed to examine the basal cytokine production. FABP4 concentrations in SF, plasma and culture media of ScAT & IFP were diluted appropriately and determined by enzyme-linked immunosorbent assay (ELISA). 2. Animal study FABP4-/- mice and their wild-type littermates were fed with standard chow diet and high fat diet for 20 and 26 weeks started from 8 weeks old. At the time of sacrifice, both the right and left knees were harvested and fixed with paraffin section. A semi-quantitative score system recommended by Osteoarthritis Research Society International (OARSI) was used to assess the quality of the cartilage histology. Harris hematoxylin and eosin staining was employ to observe the morphology of the knee joint. Depletion of proteoglycans in cartilage was captured using toluidine blue and fast green staining. Results: 1. Human study The fat-conditioned media of IFP from OA patients contained significantly higher concentrations of FABP4 than ScAT (1192 ± 186.4ng/ ml versus 679.8 ± 138.4ng/ml, p<0.001). However, no significantly difference was found in non-OA controls between IFP and ScAT. In OA cases there was significant correlation between FABP4 in ScAT and FABP4 in IFP (r ¼ 0.47, p ¼ 0.002). No significant difference of FABP4 in ScAT or IFP was found between OA cases and non-OA controls. OA subjects had significantly higher levels of FABP4 in plasma than nonOA controls (18.04 ± 2.31ng/ml versus 5.46 ± 0.62ng/ml, p<0.001); SF from OA subjects had significantly higher levels of FABP4 than non-OA controls (107.8 ± 14.17ng/ml versus 12.2 ± 4.31 ng/ml, p ¼ 0.003), after adjusting for BMI and gender. The significance was lost (p ¼ 0.424) after adjusting for age. In OA subjects, SF FABP4 concentration was significantly higher than paired plasma FABP4 (100.4 ± 16.2ng/ml versus 16.2 ± 2.2 ng/ml, p<0.001). FABP4 concentration in paired synovial fluid and IFP (n ¼ 35) was significantly correlated (Spearman’s r ¼ 0.364, p ¼ 0.032). 2. Animal study Mice with high fat diet had significantly higher OARSI scores than mice with standard chow diet (p ¼ 0.025). FABP-/- mice that provided with high fat diet for same periods did not show more severe OA cartilage than their counterparts that fed with standard chow (p ¼ 0.844). Older mice tended to have more severe cartilage degeneration by having higher OARSI scores (p ¼ 0.020). Conclusions: This is the first study that reported an altered pattern of FABP4 in OA patients. OA patients had higher levels of FABP4 in both plasma and SF compared with non-OA controls. SF from OA patients had significantly higher levels of FABP4 compared to plasma. IFP from OA patients produced more FABP4 than ScAT while no differences in controls. IFP could be a source of FABP4 in the knee joint, and thus lead to high concentration of FABP4 in SF. The significance disappeared after adjusting age. This might be caused by the large age differences between OA and control groups. Further studies are required to investigate the relationship between FABP4 and cartilage and bone to elucidate the biological relevance and potential linkage between FABP4 and osteoarthritis.
S71
106 HIGHER BONE TURNOVER DETECTED BY SERUM LEVELS OF TARTRATE-RESISTANT ACID PHOSPHATASE 5B (TRACP-5B) IS A RISK FOR THE RADIOGRAPHIC MEDIAL KNEE JOINT SPACE NARROWING IN MEN IN EARLY FORTIES WITHOUT KNEE PAIN THE THREE YEARS PROSPECTIVE OBSERVATIONAL STUDY L. Liu y, z, H. Kaneko y, S. Hada y, R. Sadatsuki y, A. Yusup y, M. Kinoshita y, H. Arita y, J. Shiozawa y, Y. Saita y, Y. Takazawa y, H. Ikeda y, K. Kaneko y, z, M. Ishijima y, z. y Dept. of Med. for Orthopaedics and Motor Organ, Juntendo Univ. Graduate Sch. of Med., Tokyo, Japan; z Sportology Ctr., Juntendo Univ. Graduate Sch. of Med., Tokyo, Japan Purpose: Osteoarthritis (OA) is a non-rheumatoid joint disease characterized by progressive degeneration of extra-cellular cartilage matrix (ECM), enhanced subchondral bone remodeling, osteophyte formation and synovial thickening. Healthy subchondral bone protects cartilage from high peak stresses and possible matrix damage. Animal studies showed that during early OA there is a marked reduction in subchondral bone thickness and there are increased numbers of subchondral pores, suggesting that subchondral bone remodeling is involved in disease progression. Tartrate-resistant acid phosphatase 5b (TRACP-5b) is the bone-specific acid phosphatase isozyme localized in osteoclasts. The serum levels of TRACP-5b reflect the number and activity of osteoclasts, indicating conditions of bone resorption. However, it still remained unclear that a high bone mass or its determinants may be directly involved in the pathogenesis of knee OA. Human joints with OA have a broad spectrum of subchondral calcified tissue changes. Some of these changes may precede the initiation of OA, while others are part of the joint’s response to abnormal loading conditions resulting from OA. Thus, the influencing the subchondral calcified tissue response to joint loading in a positive feedback loop may further increase stress on joint tissues. It has long been speculated and revealed by the recent development of imaging technique that the degeneration and destruction of articular cartilage are started silently and earlier than it had long been speculated. The aim of this prospective study was to investigate whether the baseline bone turnover can predict radiographic JSN of the medial knee joint in men in early forties without knee pain. Methods: This prospective study was conducted by analyzing data of the Sportology Core Study in Juntendo University Graduate School of Medicine, Tokyo, which was started at 2012. Eighty seven healthy male volunteers (42.9 y in average) who didn’t have any symptoms for knee pain and experience any traumatic episodes for the knee joints were enrolled in this study. A standing, extended antero-posterior view radiograph of both knee were taken at the time of study entry and at three year follow up. Radiological progression of the knee OA is defined as a 0.3 mm or greater of JSN during 3-years of follow up than in the knee of lower JSW at baseline. The BMD of lumbar spine (L2-L4) (LS-BMD) and femoral neck (H-BMD) were measured using dual X-ray absorptiometry (DXA) using the fan beam Discovery-W scanner. The serum levels of TRACP-5b (sTRACP-5b) were measured by ELISA. The significance of the differences in the data was evaluated using the Mann-Whitney U-test. We examined the relation of tertile groups of sTRACP-5b (using the lower tertile as the referent category) to the risk of the radiographic progression of JSN using the logistic regression model. Results: The subjects were divided into two groups by the presence or absence of the radiological JSN: the JSN during three year of follow-up was observed in twelve of 87 subjects (JSN group), while it was not observed in the remaining seventy-five subjects (non-JSN group). The LS-BMD was positively correlated to the H-BMD at baseline (r: 0.470, p<0.001). The sTRACP-5b were negatively correlated with the LS-BMD at baseline (r: 0.316, p ¼ 0.003), while the sTRACP-5b were not correlated with H-BMD at baseline (r: 0.124, p ¼ 0.261). The sTRACP-5b of JSN group at baseline (326.92 mg/dl) were significantly increased in comparison to those of non-JSN group at baseline (280.83 mg/dl, p<0.01). The LS-BMD of JSN group at baseline (0.931 mg/kg2) was significantly lower than those non-JSN group at baseline (1.005 mg/kg2, p<0.01). However, no significant differences of H-BMD at baseline between the JSN group and non-JSN group were observed. After three years follow-up, sTRACP-5b of JSN group were also significantly increased in comparison to those of non-JSN group. When the subjects were divided into three groups according to the sTRACP-5b at baseline (T1, T2 and T3), the odds ratio (OR) for the JSN after 3-years of follow up
S72
Abstracts / Osteoarthritis and Cartilage 24 (2016) S63eS534
in the upper tertile at baseline (T3) was significantly higher than that in the lower tertile (T1) [Odds ratio: 9.0 (95% CI: 1.0 to 78.9), p<0.05]. This risk for the JSN after 3-years of follow up in the upper tertile at baseline (T3) was still observed after adjustment for age and BMI of the subjects [Odds ratio: 10.5 (95% CI: 1.1 to 97.0), p<0.05]. Conclusions: In a three year prospective study, the higher levels of sTRACP-5b were the risk factor for the radiographic medial JSN in men in early forties without knee pain. 107 DEVELOPMENT OF A SERUM BIOMARKER PANEL PREDICTIVE OF A KNEE OSTEOARTHRITIS DIAGNOSIS
indicated to evaluate this diagnostic approach in the context of preradiographic OA to determine if it could serve as an early predictor of risk for radiographic OA to provide a means of treating the disease at a stage more amenable to structural modification and potential cure. $$MISSING OR BAD IMAGE SPECIFICATION {B0C8B235-1411-495581EA-4750D184C937}$$.
Table 1 Analysis of clinical covariates and 19 MRM markers for OA diagnosis.
HIGHLY
V.B. Kraus y, J. Catterall y, E. Soderblom z, M.A. Moseley z, S. Suchindran z. y Duke Univ. Sch. of Med., Durham, NC, USA; z Duke Univ., Durham, NC, USA Purpose: Osteoarthritis (OA) fits the description provided by Machiavelli over 500 years ago of a malady, in the beginning “is easy to cure but difficult to detect, but in the course of time, not having been either detected or treated in the beginning, becomes easy to detect but difficult to cure”. A precedent for improved outcomes with early identification and early treatment now exists for Rheumatoid Arthritis. We hypothesized that this paradigm should inform the approach to the diagnosis and treatment of OA in order to prevent the disease prior to the onset of irreversible joint tissue pathology. The goal of this study was to develop a serum biomarker panel highly indicative of radiographic OA that could potentially be readily deployed as a diagnostic in a clinic setting. Methods: Using a systematic, iterative, and unbiased mass spectrometry approach, we created a Multiple Reaction Monitoring (MRM) panel for 146 peptides (99 proteins) in serum. The selection of proteins was based on results of discovery proteomic studies in synovial fluid (n ¼ 23), urine (n ¼ 45) and serum (n ¼ 40) from subjects with and without symptomatic radiographic knee OA. To identify the peptides most diagnostic of radiographic knee OA, we tested the MRM panel in the serum of additional subjects (n ¼ 110) from the Prediction of Osteoarthritis Progression (POP) cohort and the Genetics of Generalized OA (GOGO) study. A diagnosis of ANY KNEE OA was defined as the presence of grade >1 (by standardized atlas scoring of knee radiographs) osteophyte, or joint space narrowing (JSN) or >2 Kellgren Lawrence grade. Controls, all from GOGO, were siblings that did not meet the definition of ANY KNEE OA. Quantitative LC/MS/MS was performed on 1 mg of protein digest per sample, using a nanoAcquity UPLC system (Waters Corp) coupled to a Synapt G2 HDMS high resolution accurate mass tandem mass spectrometer (Waters Corp) via a nanoelectrospray ionization source. Samples were assayed in singlicate along with a representative “QC Pool” sample created from equal portions of all individual samples which was run periodically throughout the total acquisition window. Samples were also spiked with a constant amount of yeast alcohol dehydrogenase (ADH). The mean and median technical variability of a measured protein intensity across eight QC Pool injections were 16.3% and 12.4%, respectively, with variations in measured ADH intensities of 10.4%. The optimal multimarker predictors were identified by LASSO regression using the R package “glmnet”. To evaluate the value of added MRM markers for predicting control vs OA status, the data were randomly split 50 times into half training and half testing datasets. Based on the training data, two logistic regressions were fit: (i) using the clinical covariates only; (ii) using the clinical covariates plus 19 selected MRM markers. The fitted scores were used to compute the area under the curve (AUC) of Receiver Operator Characteristic curves. Results: The sample was 82% female, mean age 64 (SD 10) years, mean BMI 28 (SD 6) kg/m2. The prevalence of ANY KNEE OA in the sample was 70%. Age and BMI were significantly associated with ANY KNEE OA (p ¼ 0.0027 and 0.0002, respectively) yielding an AUC of 0.82 (Table 1). A total of 19 MRM markers in combination with age and BMI yielded a near perfect diagnostic predictor with cross-validated AUC ¼ 1.0 in the selected training sample and AUC ¼ 0.974 in the test sample. Evaluation of the selection path for MRM markers and associated AUCs demonstrated that the combination of the top 2 MRM markers with clinical parameters also provided a near perfect diagnostic marker (AUC 0.948). Conclusions: We developed a method of diagnosing radiographic OA based on a serum panel of mass spectrometry based biomarkers. In combination with age and BMI, two biomarkers improved the diagnostic capability for knee OA to near perfect. Future work is now
Parameters (n ¼ 110 subjects)
Training Data (AUC ± SD)
Testing Data (AUC ± SD)
Clinical covariates (age, gender, BMI) Clinical + 19 MRM markers
0.822 ± 0.043 1.000 ± 0.0
0.792 ± 0.047 0.974 ± 0.02
108 DETECTING SUBJECTS AT RISK OF RADIOLOGICAL PROGRESSION: DATA FROM THE OAI J.G. Tamez-Pena y, J. Farber z, S. Totterman z. y ITESM, Monterrey, Nuevo Leon, Mexico; z Qmetrics, Rochester, NY, USA Purpose: Disease modifying osteoarthritis (OA) drugs (DMOAD) trials are hampered by the slow progression of OA. Drug companies address the issue by including patients that are at risk of progression (Early OA, Obesity, malalignment, older age, women, etc.); but current criteria have been proven to be very inefficient. The purposes of this study were: First, identify X-Ray quantitative features that are typical of OA subjects who will progress during the next two years. Second, develop an X-ray based OA progression index that can be used to screen for subjects who are at risk of radiological worsening. Third, evaluate the standard response of the mean (SMR) of the set composed of subjects identified at risk of OA progression. Methods: Quantitative joint space width (qJSW) measurements and semi-quantitative central readings of the knee for Kellgren & Lawrence (KL) grade and radiographic features of the tibio-femoral joint from fixed-flexion knee radiographs from the OAI were used in this study (https://oai.epi-ucsf.org/datarelease/ImageAssessments.asp). The qJSW set is composed of 21 continuous measurements and from those we derived 12 descriptive measurements for a total of 33 quantitative features. We used the qJSW and semi-quantitative assessment of the baseline, 12 month and 24 month x-rays. We also included the 36 and 48 month qJSW data. We complemented the data by including age, gender, weight, height, BMI, site information and the scan date from the clinical and enrollee OAI data sets. Only knees with complete qJSW data and KL scores were included. We used all the longitudinal data to detect and adjust for potential site-specific drift in qJSW. We adjusted qJSW data for height, age and gender correlations using healthy subjects (KL ¼ 0) as a normal reference. After adjustment, all measurements were z-transformed. Then, we split the subjects into a training set (1/3 of the population) and test set (the reaming 2/3). From the training set, we selected subjects that had a baseline KL score of 2 or 3. After that, we labeled subjects as progressors or non progressors based on change between the baseline and 24-month exams of OARSI scores of medial and/or lateral joint space narrowing scores. The labeled train-set was explored by a feature selection algorithm: B:SWiMS (https://cran.rproject.org/web/packages/FRESA.CAD). B:SWiMS returned a set of compact features that were more common in subjects that worsened their semi-quantitative scores. B:SWiMS also returned a logistic model that can be used to gauge the risk of progression of any OA subject. Finally, we tested the performance of the index on the test set. Performance evaluation included ODDS ratios and SRM metrics. Results: 5211 knees (3109 females, age 61.8 ± 9.1) had complete qJSW and semi-quantitate KL scores at baseline. At baseline 988, 803, 2118, 1086 and 216 subjects had KL scores 0, 1, 2, 3 and 4 respectively. The train set consisted on 745 subjects, 207 of them having OARSI JSW progression. B:SWiMS identified that having an abnormal difference between medial and lateral JSW, or a large variance in the medial JSW or abnormal slope between the medial and lateral JSW measurements are baseline characteristics of subjects that will worsen OARSI scores within a 2 year period (Table 1). Those features were incorporated into a logistic model. The model was tested on 1851 knees with either KL ¼ 2
growing number of evidence suggested altered levels of adipokines in OA pathogenesis. Fatty acid binding protein 4 (FABP4) is one of the most abundant intracellular lipid transport proteins produced by mature adipocytes. Reports showed that FABP4 could enhance the production of pro-inflammatory cytokines by forming a positive feedback loop. FABP4 has been documented to play an important role in metabolic syndrome such as type II diabetes mellitus and atherosclerosis. However there is very limited knowledge about FABP4 in OA. This study aimed to investigate the role of FABP4 in the pathogenesis of OA. Firstly we wanted to describe the pattern of FABP4 in OA patients by determining FABP4 levels in local and systemic fat and body fluid and compared with non-OA individuals. Secondly we carried out an animal study to see if FABP4 knockout mice are protected from OA. Methods: 1. Human study 75 individuals were recruited including 45 OA cases that undergone total knee replacement (TKR) and 30 non-OA controls that received arthroscopic surgery. Blood was taken one day before the surgery. Subcutaneous adipose tissue (ScAT), synovial fluid (SF) and infrapatellar fat pad (IFP) from OA and control subjects were harvested during surgery (TKR or arthroscopy). Explant culture of ScAT and IFP was performed to examine the basal cytokine production. FABP4 concentrations in SF, plasma and culture media of ScAT & IFP were diluted appropriately and determined by enzyme-linked immunosorbent assay (ELISA). 2. Animal study FABP4-/- mice and their wild-type littermates were fed with standard chow diet and high fat diet for 20 and 26 weeks started from 8 weeks old. At the time of sacrifice, both the right and left knees were harvested and fixed with paraffin section. A semi-quantitative score system recommended by Osteoarthritis Research Society International (OARSI) was used to assess the quality of the cartilage histology. Harris hematoxylin and eosin staining was employ to observe the morphology of the knee joint. Depletion of proteoglycans in cartilage was captured using toluidine blue and fast green staining. Results: 1. Human study The fat-conditioned media of IFP from OA patients contained significantly higher concentrations of FABP4 than ScAT (1192 ± 186.4ng/ ml versus 679.8 ± 138.4ng/ml, p<0.001). However, no significantly difference was found in non-OA controls between IFP and ScAT. In OA cases there was significant correlation between FABP4 in ScAT and FABP4 in IFP (r ¼ 0.47, p ¼ 0.002). No significant difference of FABP4 in ScAT or IFP was found between OA cases and non-OA controls. OA subjects had significantly higher levels of FABP4 in plasma than nonOA controls (18.04 ± 2.31ng/ml versus 5.46 ± 0.62ng/ml, p<0.001); SF from OA subjects had significantly higher levels of FABP4 than non-OA controls (107.8 ± 14.17ng/ml versus 12.2 ± 4.31 ng/ml, p ¼ 0.003), after adjusting for BMI and gender. The significance was lost (p ¼ 0.424) after adjusting for age. In OA subjects, SF FABP4 concentration was significantly higher than paired plasma FABP4 (100.4 ± 16.2ng/ml versus 16.2 ± 2.2 ng/ml, p<0.001). FABP4 concentration in paired synovial fluid and IFP (n ¼ 35) was significantly correlated (Spearman’s r ¼ 0.364, p ¼ 0.032). 2. Animal study Mice with high fat diet had significantly higher OARSI scores than mice with standard chow diet (p ¼ 0.025). FABP-/- mice that provided with high fat diet for same periods did not show more severe OA cartilage than their counterparts that fed with standard chow (p ¼ 0.844). Older mice tended to have more severe cartilage degeneration by having higher OARSI scores (p ¼ 0.020). Conclusions: This is the first study that reported an altered pattern of FABP4 in OA patients. OA patients had higher levels of FABP4 in both plasma and SF compared with non-OA controls. SF from OA patients had significantly higher levels of FABP4 compared to plasma. IFP from OA patients produced more FABP4 than ScAT while no differences in controls. IFP could be a source of FABP4 in the knee joint, and thus lead to high concentration of FABP4 in SF. The significance disappeared after adjusting age. This might be caused by the large age differences between OA and control groups. Further studies are required to investigate the relationship between FABP4 and cartilage and bone to elucidate the biological relevance and potential linkage between FABP4 and osteoarthritis.
S71
106 HIGHER BONE TURNOVER DETECTED BY SERUM LEVELS OF TARTRATE-RESISTANT ACID PHOSPHATASE 5B (TRACP-5B) IS A RISK FOR THE RADIOGRAPHIC MEDIAL KNEE JOINT SPACE NARROWING IN MEN IN EARLY FORTIES WITHOUT KNEE PAIN THE THREE YEARS PROSPECTIVE OBSERVATIONAL STUDY L. Liu y, z, H. Kaneko y, S. Hada y, R. Sadatsuki y, A. Yusup y, M. Kinoshita y, H. Arita y, J. Shiozawa y, Y. Saita y, Y. Takazawa y, H. Ikeda y, K. Kaneko y, z, M. Ishijima y, z. y Dept. of Med. for Orthopaedics and Motor Organ, Juntendo Univ. Graduate Sch. of Med., Tokyo, Japan; z Sportology Ctr., Juntendo Univ. Graduate Sch. of Med., Tokyo, Japan Purpose: Osteoarthritis (OA) is a non-rheumatoid joint disease characterized by progressive degeneration of extra-cellular cartilage matrix (ECM), enhanced subchondral bone remodeling, osteophyte formation and synovial thickening. Healthy subchondral bone protects cartilage from high peak stresses and possible matrix damage. Animal studies showed that during early OA there is a marked reduction in subchondral bone thickness and there are increased numbers of subchondral pores, suggesting that subchondral bone remodeling is involved in disease progression. Tartrate-resistant acid phosphatase 5b (TRACP-5b) is the bone-specific acid phosphatase isozyme localized in osteoclasts. The serum levels of TRACP-5b reflect the number and activity of osteoclasts, indicating conditions of bone resorption. However, it still remained unclear that a high bone mass or its determinants may be directly involved in the pathogenesis of knee OA. Human joints with OA have a broad spectrum of subchondral calcified tissue changes. Some of these changes may precede the initiation of OA, while others are part of the joint’s response to abnormal loading conditions resulting from OA. Thus, the influencing the subchondral calcified tissue response to joint loading in a positive feedback loop may further increase stress on joint tissues. It has long been speculated and revealed by the recent development of imaging technique that the degeneration and destruction of articular cartilage are started silently and earlier than it had long been speculated. The aim of this prospective study was to investigate whether the baseline bone turnover can predict radiographic JSN of the medial knee joint in men in early forties without knee pain. Methods: This prospective study was conducted by analyzing data of the Sportology Core Study in Juntendo University Graduate School of Medicine, Tokyo, which was started at 2012. Eighty seven healthy male volunteers (42.9 y in average) who didn’t have any symptoms for knee pain and experience any traumatic episodes for the knee joints were enrolled in this study. A standing, extended antero-posterior view radiograph of both knee were taken at the time of study entry and at three year follow up. Radiological progression of the knee OA is defined as a 0.3 mm or greater of JSN during 3-years of follow up than in the knee of lower JSW at baseline. The BMD of lumbar spine (L2-L4) (LS-BMD) and femoral neck (H-BMD) were measured using dual X-ray absorptiometry (DXA) using the fan beam Discovery-W scanner. The serum levels of TRACP-5b (sTRACP-5b) were measured by ELISA. The significance of the differences in the data was evaluated using the Mann-Whitney U-test. We examined the relation of tertile groups of sTRACP-5b (using the lower tertile as the referent category) to the risk of the radiographic progression of JSN using the logistic regression model. Results: The subjects were divided into two groups by the presence or absence of the radiological JSN: the JSN during three year of follow-up was observed in twelve of 87 subjects (JSN group), while it was not observed in the remaining seventy-five subjects (non-JSN group). The LS-BMD was positively correlated to the H-BMD at baseline (r: 0.470, p<0.001). The sTRACP-5b were negatively correlated with the LS-BMD at baseline (r: 0.316, p ¼ 0.003), while the sTRACP-5b were not correlated with H-BMD at baseline (r: 0.124, p ¼ 0.261). The sTRACP-5b of JSN group at baseline (326.92 mg/dl) were significantly increased in comparison to those of non-JSN group at baseline (280.83 mg/dl, p<0.01). The LS-BMD of JSN group at baseline (0.931 mg/kg2) was significantly lower than those non-JSN group at baseline (1.005 mg/kg2, p<0.01). However, no significant differences of H-BMD at baseline between the JSN group and non-JSN group were observed. After three years follow-up, sTRACP-5b of JSN group were also significantly increased in comparison to those of non-JSN group. When the subjects were divided into three groups according to the sTRACP-5b at baseline (T1, T2 and T3), the odds ratio (OR) for the JSN after 3-years of follow up
S72
Abstracts / Osteoarthritis and Cartilage 24 (2016) S63eS534
in the upper tertile at baseline (T3) was significantly higher than that in the lower tertile (T1) [Odds ratio: 9.0 (95% CI: 1.0 to 78.9), p<0.05]. This risk for the JSN after 3-years of follow up in the upper tertile at baseline (T3) was still observed after adjustment for age and BMI of the subjects [Odds ratio: 10.5 (95% CI: 1.1 to 97.0), p<0.05]. Conclusions: In a three year prospective study, the higher levels of sTRACP-5b were the risk factor for the radiographic medial JSN in men in early forties without knee pain. 107 DEVELOPMENT OF A SERUM BIOMARKER PANEL PREDICTIVE OF A KNEE OSTEOARTHRITIS DIAGNOSIS
indicated to evaluate this diagnostic approach in the context of preradiographic OA to determine if it could serve as an early predictor of risk for radiographic OA to provide a means of treating the disease at a stage more amenable to structural modification and potential cure. $$MISSING OR BAD IMAGE SPECIFICATION {B0C8B235-1411-495581EA-4750D184C937}$$.
Table 1 Analysis of clinical covariates and 19 MRM markers for OA diagnosis.
HIGHLY
V.B. Kraus y, J. Catterall y, E. Soderblom z, M.A. Moseley z, S. Suchindran z. y Duke Univ. Sch. of Med., Durham, NC, USA; z Duke Univ., Durham, NC, USA Purpose: Osteoarthritis (OA) fits the description provided by Machiavelli over 500 years ago of a malady, in the beginning “is easy to cure but difficult to detect, but in the course of time, not having been either detected or treated in the beginning, becomes easy to detect but difficult to cure”. A precedent for improved outcomes with early identification and early treatment now exists for Rheumatoid Arthritis. We hypothesized that this paradigm should inform the approach to the diagnosis and treatment of OA in order to prevent the disease prior to the onset of irreversible joint tissue pathology. The goal of this study was to develop a serum biomarker panel highly indicative of radiographic OA that could potentially be readily deployed as a diagnostic in a clinic setting. Methods: Using a systematic, iterative, and unbiased mass spectrometry approach, we created a Multiple Reaction Monitoring (MRM) panel for 146 peptides (99 proteins) in serum. The selection of proteins was based on results of discovery proteomic studies in synovial fluid (n ¼ 23), urine (n ¼ 45) and serum (n ¼ 40) from subjects with and without symptomatic radiographic knee OA. To identify the peptides most diagnostic of radiographic knee OA, we tested the MRM panel in the serum of additional subjects (n ¼ 110) from the Prediction of Osteoarthritis Progression (POP) cohort and the Genetics of Generalized OA (GOGO) study. A diagnosis of ANY KNEE OA was defined as the presence of grade >1 (by standardized atlas scoring of knee radiographs) osteophyte, or joint space narrowing (JSN) or >2 Kellgren Lawrence grade. Controls, all from GOGO, were siblings that did not meet the definition of ANY KNEE OA. Quantitative LC/MS/MS was performed on 1 mg of protein digest per sample, using a nanoAcquity UPLC system (Waters Corp) coupled to a Synapt G2 HDMS high resolution accurate mass tandem mass spectrometer (Waters Corp) via a nanoelectrospray ionization source. Samples were assayed in singlicate along with a representative “QC Pool” sample created from equal portions of all individual samples which was run periodically throughout the total acquisition window. Samples were also spiked with a constant amount of yeast alcohol dehydrogenase (ADH). The mean and median technical variability of a measured protein intensity across eight QC Pool injections were 16.3% and 12.4%, respectively, with variations in measured ADH intensities of 10.4%. The optimal multimarker predictors were identified by LASSO regression using the R package “glmnet”. To evaluate the value of added MRM markers for predicting control vs OA status, the data were randomly split 50 times into half training and half testing datasets. Based on the training data, two logistic regressions were fit: (i) using the clinical covariates only; (ii) using the clinical covariates plus 19 selected MRM markers. The fitted scores were used to compute the area under the curve (AUC) of Receiver Operator Characteristic curves. Results: The sample was 82% female, mean age 64 (SD 10) years, mean BMI 28 (SD 6) kg/m2. The prevalence of ANY KNEE OA in the sample was 70%. Age and BMI were significantly associated with ANY KNEE OA (p ¼ 0.0027 and 0.0002, respectively) yielding an AUC of 0.82 (Table 1). A total of 19 MRM markers in combination with age and BMI yielded a near perfect diagnostic predictor with cross-validated AUC ¼ 1.0 in the selected training sample and AUC ¼ 0.974 in the test sample. Evaluation of the selection path for MRM markers and associated AUCs demonstrated that the combination of the top 2 MRM markers with clinical parameters also provided a near perfect diagnostic marker (AUC 0.948). Conclusions: We developed a method of diagnosing radiographic OA based on a serum panel of mass spectrometry based biomarkers. In combination with age and BMI, two biomarkers improved the diagnostic capability for knee OA to near perfect. Future work is now
Parameters (n ¼ 110 subjects)
Training Data (AUC ± SD)
Testing Data (AUC ± SD)
Clinical covariates (age, gender, BMI) Clinical + 19 MRM markers
0.822 ± 0.043 1.000 ± 0.0
0.792 ± 0.047 0.974 ± 0.02
108 DETECTING SUBJECTS AT RISK OF RADIOLOGICAL PROGRESSION: DATA FROM THE OAI J.G. Tamez-Pena y, J. Farber z, S. Totterman z. y ITESM, Monterrey, Nuevo Leon, Mexico; z Qmetrics, Rochester, NY, USA Purpose: Disease modifying osteoarthritis (OA) drugs (DMOAD) trials are hampered by the slow progression of OA. Drug companies address the issue by including patients that are at risk of progression (Early OA, Obesity, malalignment, older age, women, etc.); but current criteria have been proven to be very inefficient. The purposes of this study were: First, identify X-Ray quantitative features that are typical of OA subjects who will progress during the next two years. Second, develop an X-ray based OA progression index that can be used to screen for subjects who are at risk of radiological worsening. Third, evaluate the standard response of the mean (SMR) of the set composed of subjects identified at risk of OA progression. Methods: Quantitative joint space width (qJSW) measurements and semi-quantitative central readings of the knee for Kellgren & Lawrence (KL) grade and radiographic features of the tibio-femoral joint from fixed-flexion knee radiographs from the OAI were used in this study (https://oai.epi-ucsf.org/datarelease/ImageAssessments.asp). The qJSW set is composed of 21 continuous measurements and from those we derived 12 descriptive measurements for a total of 33 quantitative features. We used the qJSW and semi-quantitative assessment of the baseline, 12 month and 24 month x-rays. We also included the 36 and 48 month qJSW data. We complemented the data by including age, gender, weight, height, BMI, site information and the scan date from the clinical and enrollee OAI data sets. Only knees with complete qJSW data and KL scores were included. We used all the longitudinal data to detect and adjust for potential site-specific drift in qJSW. We adjusted qJSW data for height, age and gender correlations using healthy subjects (KL ¼ 0) as a normal reference. After adjustment, all measurements were z-transformed. Then, we split the subjects into a training set (1/3 of the population) and test set (the reaming 2/3). From the training set, we selected subjects that had a baseline KL score of 2 or 3. After that, we labeled subjects as progressors or non progressors based on change between the baseline and 24-month exams of OARSI scores of medial and/or lateral joint space narrowing scores. The labeled train-set was explored by a feature selection algorithm: B:SWiMS (https://cran.rproject.org/web/packages/FRESA.CAD). B:SWiMS returned a set of compact features that were more common in subjects that worsened their semi-quantitative scores. B:SWiMS also returned a logistic model that can be used to gauge the risk of progression of any OA subject. Finally, we tested the performance of the index on the test set. Performance evaluation included ODDS ratios and SRM metrics. Results: 5211 knees (3109 females, age 61.8 ± 9.1) had complete qJSW and semi-quantitate KL scores at baseline. At baseline 988, 803, 2118, 1086 and 216 subjects had KL scores 0, 1, 2, 3 and 4 respectively. The train set consisted on 745 subjects, 207 of them having OARSI JSW progression. B:SWiMS identified that having an abnormal difference between medial and lateral JSW, or a large variance in the medial JSW or abnormal slope between the medial and lateral JSW measurements are baseline characteristics of subjects that will worsen OARSI scores within a 2 year period (Table 1). Those features were incorporated into a logistic model. The model was tested on 1851 knees with either KL ¼ 2