Document Type


Publication Date


Publication Title

Osteoarthritis Cartilage


Purpose: There is an outstanding need to identify minimally invasive biomarkers for reliable detection of knee osteoarthritis (OA). Current clinical diagnostic methods are limited since OA symptoms do not always correlate with structural degeneration in the joint. Soluble biochemical markers provide a better readout of disease activity, and a variety of blood, synovial fluid, and urine biomarkers have been explored in OA, including microRNAs. As small, non-coding RNAs, microRNAs are promising biomarker candidates since they are easy to detect in biofluids, are relatively stable (i.e. resistant to enzymatic degradation), and can be reliably quantified such that levels can be linked to disease. Furthermore, microRNAs are known drivers of OA pathology, and their expression may precede joint degeneration, when opportunities for intervention still exist. Based on this, circulating microRNAs have strong potential to serve as biomarkers for knee OA, but a major limitation is lack of reproducibility across studies profiling circulating microRNAs in OA. While sequencing is the gold standard method for unbiased profiling of microRNAs, there are critical experimental design and analysis parameters that can impact the results. The objectives of this study are to identify circulating microRNAs in late-stage radiographic knee OA compared to non-OA controls using existing microRNA-sequencing data, and to validate the findings using our recently established Henry Ford Health System (HFHS) Osteoarthritis cohort.

Methods: We searched the literature for microRNA-sequencing studies profiling circulating microRNAs in OA versus non-OA participants and identified two studies, one conducted in Norway (Aae et al., 2020) and the other in France (Rousseau et al., 2020). We obtained raw sequencing data from the authors and re-analyzed the data by applying our recently reported method for microRNA-sequencing analysis. Among other changes (e.g. normalization method), we re-defined the cohorts to include participants with only Kellgren-Lawrence (KL) grades 3 and 4 in the OA group (compared to KL 0 to 4 and total knee arthroplasty in the original Norway study and KL 2 and 3 in the original France study) and with KL grade 0 in the non-OA group (consistent with the original Norway study and compared to KL 0 and 1 in the original France study). Following differential expression analysis using a multivariate model adjusted for age, sex, and body mass index, we prioritized microRNAs that were common to the OA groups in both cohorts. We next performed validation by real-time PCR in the HFHS Osteoarthritis cohort utilizing plasma samples from participants with unilateral and/or bilateral knee and/or hip OA and non-OA controls.

Results: As reported by the two original microRNA-sequencing studies, there were no significant differences in the Norway cohort and 3 differentially expressed microRNAs in OA (miR-139-5p, miR-1299, miR-200a-3p) in the France cohort, though none achieved validation in real-time PCR experiments. Following our re-analysis, we identified 23 and 82 differentially expressed microRNAs (p<0.1) in the Norway and France cohorts, respectively, with 3 microRNAs in common between the OA groups: miR-126-3p, miR-30c-2-3p, and miR-144-5p. Of these, miR-126-3p had the highest counts-per-million in both cohorts, showed an increased fold change in OA in both cohorts (p<0.05; Figure 1A and 1B), and was found in 100% and 91% of OA samples and 0% and 35% of non-OA samples in the Norway and France cohorts, respectively. Furthermore, a report in 2014 by Borgonio Cuadra et al. identified circulating miR-126 to be elevated in OA (KL 2 and 3) compared to non-OA (KL 0) by both real-time PCR array and real-time PCR validation experiments (Figure 1C). This led us to explore miR-126-3p expression in plasma samples from the HFHS Osteoarthritis cohort where we found a consistent increase in knee OA (symptomatic, KL 3 or 4), irrespective of unilateral or bilateral, compared to non-OA controls (asymptomatic, KL 0), yet no significant increase in hip OA (Figure 1D).

Conclusions: Through application of our microRNA-sequencing analysis method, we identified circulating miR-126-3p to be increased in late-stage radiographic knee OA compared to non-OA controls in two studies originally reporting no validated differences. This finding is supported by previous literature identifying circulating miR-126 to be elevated in knee OA compared to non-OA controls and is extended by our data showing that the increase may be unique to knee OA and not hip OA. Taken together, there are now data from four independent cohorts demonstrating an increase in circulating miR-126-3p in knee OA, suggesting that this microRNA may have utility as a biomarker for OA. [Formula presented]

PubMed ID

Not assigned.



First Page


Last Page