We showed the potential of applying machine learning to generate predictive models for the knee OA incidence. Imaging-based information were found particularly important in the proposed models. Furthermore, our analysis confirmed the relevance of known BM for knee OA. Overall, we propose five highly predictive small models that can be possibly adopted for an early prediction of knee OA.

Biomarkers of joint tissue turnover, cytokines, chemokines and peptide arrays were measured in different cohorts and studies. Amongst those were previously tested biomarkers such as osteocalcin, Carboxy-terminal cross-linked fragment of type II collagen (CTX-II) and cartilage oligomeric matrix protein (COMP). A majority of the biomarker were classified as I, B or B biomarkers according to the BIPED criteria. Work is continuing on testing biomarkers in OA. There is still a huge, unmet medical need to identify, test, validate and qualify novel and well-known biomarkers. A pre-requisite for this is better characterization and classification of biomarkers to their needs, which may not be reached before higher understanding of OA phenotypes has been gained. In addition, we provide some references to some recent guidelines from Food and Drug Administration (FDA) and European Medicines Agency (EMA) on qualification and usage of biomarkers for drug development and personalized medicine, which may provide value to the field.

Little is known about local and systemic biomarkers in relation to synovitis and pain in end-stage osteoarthritis (OA) patients. We investigated the associations between the novel extracellular matrix biomarker, C1M, and local and systemic interleukin 6 (IL-6) with synovitis and pain. Serum C1M, plasma, and synovial fluid IL-6 (p-IL-6, sf-IL-6) were measured in 104 end-stage knee OA patients. Contrast-enhanced magnetic resonance imaging was used to semiquantitatively assess an 11-point synovitis score; pain was assessed by the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) and the Neuropathic Pain Questionnaire (NPQ). Linear regression was used to investigate associations between biomarkers and synovitis, and biomarkers and pain while controlling for age, sex, and body mass index. We also tested whether associations between biomarkers and pain were confounded by synovitis. We found sf-IL-6 was associated with synovitis in the parapatellar subregion (B = 0.006; 95% confidence interval [CI] 0.003-0.010), and no association between p-IL-6 and synovitis. We also observed an association between C1M and synovitis in the periligamentous subregion (B = 0.013; 95% CI 0.003-0.023). Furthermore, sf-IL-6, but not p-IL-6, was significantly associated with pain, WOMAC (B = 0.022; 95% CI 0.004-0.040), and NPQ (B = 0.043; 95% CI 0.005-0.082). There was no association between C1M and WOMAC pain, but we did find an association between C1M and NPQ (B = 0.229; 95% CI 0.036-0.422). Lastly, synovitis explained both biomarker-NPQ associations, but not the biomarker-WOMAC association. These results suggest that C1M and IL-6 are associated with synovitis and pain, and synovitis is an important confounding variable when studying biomarkers and neuropathic features in OA patients.

BackgroundOsteoclasts have been strongly implicated in osteoarthritic cartilage degradation, at least indirectly via bone resorption, and have been shown to degrade cartilage in vitro. The osteoclast resorption processes required to degrade subchondral bone and cartilage—the remodeling of which is important in the osteoarthritic disease process—have not been previously described, although cathepsin K has been indicated to participate. In this study we profile osteoclast-mediated degradation of bovine knee joint compartments in a novel in vitro model using biomarkers of extracellular matrix (ECM) degradation to assess the potential of osteoclast-derived resorption processes to degrade different knee joint compartments.MethodsMature human osteoclasts were cultured on ECMs isolated from bovine knees—articular cartilage, cortical bone, and osteochondral junction ECM (a subchondral bone-calcified cartilage mixture)—in the presence of inhibitors: the cystein protease inhibitor E-64, the matrix metalloproteinase (MMP) inhibitor GM6001, or the vacuolar-type H+-ATPase (V-ATPase) inhibitor diphyllin. Biomarkers of bone (calcium and C-terminal type I collagen (CTX-I)) and cartilage (C2M) degradation were measured in the culture supernatants. Cultures without osteoclasts were used as background samples. Background-subtracted biomarker levels were normalized to the vehicle condition and were analyzed using analysis of variance with Tukey or Dunnett’s T3 post hoc test, as applicable.ResultsOsteochondral CTX-I release was inhibited by E-64 (19% of vehicle, p = 0.0008), GM6001 (51% of vehicle, p = 0.013), and E-64/GM6001 combined (4% of vehicle, p = 0.0007)—similarly to bone CTX-I release. Diphyllin also inhibited osteochondral CTX-I release (48% of vehicle, p = 0.014), albeit less than on bone (4% of vehicle, p < 0.0001). Osteochondral C2M release was only inhibited by E-64 (49% of vehicle, p = 0.07) and GM6001 (14% of vehicle, p = 0.006), with complete abrogation when combined (0% of vehicle, p = 0.004). Cartilage C2M release was non-significantly inhibited by E-64 (69% of vehicle, p = 0.98) and was completely abrogated by GM6001 (0% of vehicle, p = 0.16).ConclusionsOur study supports that osteoclasts can resorb non-calcified and calcified cartilage independently of acidification. We demonstrated both MMP-mediated and cysteine protease-mediated resorption of calcified cartilage. Osteoclast functionality was highly dependent on the resorbed substrate, as different ECMs required different osteoclast processes for degradation. Our novel culture system has potential to facilitate drug and biomarker development aimed at rheumatic diseases, e.g. osteoarthritis, where pathological osteoclast processes in specific joint compartments may contribute to the disease process.Electronic supplementary materialThe online version of this article (10.1186/s13075-018-1564-5) contains supplementary material, which is available to authorized users.

Osteoarthritis is the most common form of joint disease. This presents the osteoarthritis research community and pharmaceutical companies developing disease-modifying osteoarthritis drugs (DMOADs) with great opportunities. There are different osteoarthritis subtypes, which complicates the traditional approaches for developing new treatments. If we can identify new markers that can distinguish different subtypes, this can greatly facilitate drug development from early discovery to late clinical development.