Purpose: Accurate biomarkers for diagnosis and prediction of osteoarthritis (OA) are needed. In addition, biomarkers have the potential to serve as a measure of the different pathological processes underlying OA. Here, we report on a proteomics screen targeted at two important pathways thought to underlie the etiology of OA: the inflammation and metabolic pathways. The aim of this study was to identify a robust biomarker for OA severity and progression. Methods: We used data from the Rotterdam Study (RS), a population based prospective study with participants aged 45 and older. Participants of the RS underwent blood and radiographic measurements at baseline and after a mean follow-up time of 5 years. We measured 184 proteins (inflammation and cardiometabolic panel) in plasma from 3,517 participants in the RS using the Olink platform. We estimated the association for all available proteomic biomarkers with OA in knee, hip and hand. Specifically, phenotype-protein associations were estimated in three ways: 1) Cross-sectionally of overall OA burden, where we computed a weighted overall OA-score by adding up standardized score for knees, hands and hips for each individual; 2) Cross-sectionally in all joints separately, where we analyzed severity of OA by adding up bilateral KL-scores for knee or hip and all joints of the left and right hand (total amount of joints=30); 3) Longitudinally in knee and hip separately, where we defined cases of OA progression as individuals with at least 1 unit increase in KL-grade, excluding progressors from KL0 to KL1. For each of the three scenarios, we analyzed the relationships with multivariate regression models using linear models for continuous outcomes and generalized linear models with binomial link function for dichotomous outcomes. In model 1, we adjusted for age, sex and cell counts, whereas in model 2 we additionally adjusted for BMI. We report effect estimate (β) per standard deviation (SD) change in protein levels with 95% Confidence Interval (CI) and nominal p-value (significance level<0.05) for each protein. We used FDR-correction for multiple testing. All statistical analyses were performed in R version 3.5.2. Results: We found in total 56 proteins that were significantly associated with one or more OA-outcomes at nominal level. In total, we found 12 significant proteins to be associated with overall OA burden (Figure 1. illustrates these associations in decreasing order of the significance level), out of which 7 proteins stayed significant after BMI adjustment. The most robust association was found between the level of circulating Cartilage acidic protein 1 (CRTAC1) and overall OA burden (β=0.18, p=3.16 x10-6; FDR-corrected p=2.74 x10-4). When we examined the joints separately, we observed associations with severity of OA in the knee (β=0.16, p=1.0 x10-5) and hand (β=0.086, p=9.0 x10-4), while a similar trend was seen in hip (β=0.08, p=0.12), although power in hip was limited in our study. The association with baseline CRTAC1 was also found for progression in knee (β=0.21, p=0.016) and hip (β=0.18, p=0.062). These associations were independent of BMI. Among the findings were several other promising biomarkers, e.g. MMP-10, which we observed to be associated with overall OA burden (β=-0.104, p=0.01; FDR-corrected p=0.28), as well as for severity of hand OA (β=-0.09, p=6.7x10-4) and progression of knee OA (β=-0.24, p=7.3 x10-3). Additionally, COMP, a well-known biomarker for OA, was also found significant for overall OA burden (β=0.15, p=1.17 x10-4; FDR-corrected p=5 x10-3), OA severity in hand (β=0.10, p=5 x10-5) and knee (β=0.09, p=0.01). In a multivariable adjusted model with both COMP and CRTAC1 included, the estimate and significance of the latter remained while the estimate of COMP lowered and lost its significance. Conclusions: We identified a number of promising biomarkers reflecting overall OA burden and increased risk for OA progression. The CRTAC1 protein is a robust, promising biomarker for osteoarthritis severity and progression. Moreover, we showed that the association of CRTAC1 is independent from COMP. This protein is a glycosylated extracellular matrix protein that is found in the interterritorial matrix of articular deep zone cartilage. This protein can also be used to distinguish chondrocytes from osteoblasts and mesenchymal stem cells in culture. This suggests that CRTAC1-levels reflects a cartilage-specific process in the joint. Such a biomarker might be useful for targeting the right patients for clinical trials and designing novel therapies for OA.
Read full abstract