This blog is being written on a flight back from San Diego where I have just attended the OARSI (Osteoarthritis Research Society International) meeting and the ASBMR (American Society of Bone and Mineral Research) conference. While it was great that they were co-located in the same city, it was a very tough time trying to go between the two meetings, which both had some great science and debates on imaging in osteoarthritis and bone.
The meetings were very good from BioClinica’s view point as we had 2 posters, including a plenary poster where I presented the reproducibility part of the osteoarthritis imaging study we conducted on behalf of Novartis to create a 95 point model of vertebral shape from plain film radiographs. This has been several years in development in close collaboration with Optasia, a software company out in the UK. The poster we presented demonstrated the underlying reproducibility of the two radiologists who reviewed the 95 point mark up mask. In other words, the reproducibility of the underlying osteoarthritis imaging technique was excellent and as good as any other published data using 6 point mark ups (this is obviously a very shortened review of the whole poster which can be found here.
The other poster was based off a clinical study we performed in conjunction with GSK and Rosiglitazone (Avandia). In this study in which we evaluated the bone effects of the drug, we also performed QCT assessments which last year we showed in a poster with a novel femoral neck quadrant analysis originally described by Ken Poole et al. The reproducibility of the osteoarthritis imaging technique was poor and while we showed a trend, we did not show anything statistically significant with this analysis. However, we felt it was worthwhile evaluating the thresholds we used with the QCT to define the cortical/cancellous bone boundary. This is where there is a significant amount of "partial volume" effects for those who wish to have the technical explanation. In other words, while we often think of there being a sharp boundary between the cortical shell and trabecular bone visually, this is not the case since the trabecular bone structure merges with the cortical bone, creating a challenge at the voxel level with QCT, in a simple explanation. The work we did was to reanalyze all the QCT scans with new thresholds and found that with the higher values we were able to demonstrate statistically significant differences in bone loss between Rosiglitazone and Metformin on only 30 – 34 subjects. The folks from GSK were pleased that this study was able to produce something of scientific value and we may have helped validate a new imaging biomarker in the process.
Learn more about osteoarthritis imaging in clinical trials. Sign up for our free webinar "MRI in Osteoarthritis – Current Knowledge and Future Perspectives in Clinical Trials!"