Mylène Jansen

SF biomarker changes upon KJD 337 16 promotes skin wound healing. 39,40 Its direction of change (opposite to that of FGF-2/TGF β - 1) is perhaps surprising: activin A is a TGF β superfamily member 24 and is strongly FGF-2- dependent in the joint in our preclinical studies. 38,39 It may be that these apparent paradoxes are because different joint tissues are involved: FGF-2 and TGF β-1 may derive from the capsule, say, whereas joint-offloading may reduce the cartilage injury response, reducing activin A. Whilst activin A appears to be a highly sensitive read-out of the intervention, it does not show association with benefit, perhaps because its change is so consistent. There are some limitations of this study. It is important to avoid over-interpretation of what was a small experimental study, and we have reported confidence intervals of association of change in molecules with change in KOOS 4 , to reflect the level of certainty. No correction was made for multiple testing: some apparent associations could have been found by chance. Identification of truly ‘normal’ individuals or those with OA who are of exactly the same demographic and stage of disease and who are willing to undergo sampling of SF is challenging. Our ‘Normal comparator’ group included those individuals undergoing surgery for musculoskeletal tumors, which could potentially have influenced circulating levels of analytes. The OA comparator group was selected from a bank of samples collected at the time of arthroplasty to be as similar as possible in terms of sex and age. However, it is possible by the nature of the intervention that some would have had more advanced stage of disease or slight differences in other factors to those in the Joint Distraction group, meaning they were not directly comparable. However, the validity of our findings do not rely solely on such Comparator data (with most comparisons made within the cohort of those undergoing distraction). These results need to be validated in a larger, independent study, where potentially relevant additional covariates (such as age, BMI) and structural imaging outcomes are incorporated 13 (correlation of marker change with imaging-based measures of cartilage thickness or volume in this current small study was not included as there would have been lack of power to detect an effect). As this intervention was part of usual care and not a clinical trial, this may have led to increased missing data and somewhat less striking improvement in clinical outcomes. This contrasts with previously published data in KJD suggesting a clinically significant long-term response. 14,16–18 A comparison between patients treated in clinical practice and those in (randomized) clinical trials revealed no clear differences in clinical outcomes between these 2 settings at 12 months (data to be published elsewhere). This study was not designed to detect a difference in clinical outcome and likely lacked power to detect differences at 12 months. In summary, we have shown a measurable molecular response in SF to joint distraction, which appears to be associated with patient-reported outcomes. This observation supports the accurate