Agreement Analysis for a Novel Protocol in Measuring Change in Bone Volume Using Serial CT and a Sheep Model of Osteoarthritis
Presented by: Candace Flynn
Authors: Candace Flynn, Mark Hurtig
Affiliations: University of Guelph
Introduction: Osteoarthritis (OA) research is challenged by the current difficulty in staging the progression of the disease, creating a demand for lengthy, expensive studies. We have previously demonstrated that non-invasive analysis of CT images can capture differences in the width and length (cm) of the medical tibial plateau (MTP) by analyzing the trajectory of OA over time (Flynn & Hurtig 2018).
Hypothesis / Objectives: To assess the inter-rater and within-rater agreement of a novel approach to measuring the change in cortical bone volume in the MTP. We hypothesized that this 3D approach would reduce error in human measurement and therefore possess a stronger correlation with histological parameters of osteoarthritis compared to the former two-dimensional measures of width and length.
Materials and Methods: Four groups of 11 adult ewes underwent arthroscopic meniscal release to induce OA within the right stifle. After twelve weeks, groups received intra-articular injections of saline, hyaluronate, low dose human umbilical SMUP cells, or high dose SMUP cells. 3D reconstructed images (pre-operative and week 48) of the proximal tibia were superimposed and the difference in volume was computed. A subset of 9 animals was analyzed by two trained observers to assess agreement of the protocol using an intra-class correlation test. Pearson and Spearman correlation tests were conducted to evaluate the association between bone volume, histology and MTP width and length.
Results: Intra-class correlation coefficients for between-rater agreement was very strong and significant (ICC = 0.947, p< 0.0001). Reproducibility of results was also good, averaging 83.6% for within-rater agreement (p< 0.001). Volume was observed to correlate significantly with MTP length (r =0.31, p<0.05). Interestingly, the association between bone volume and histological score was found to be stronger (r=0.32, p<0.05) than the associations observed between histology and MTP width and length.
Conclusions: Our previous work demonstrated that induction of OA lesions via meniscal release could be followed non-invasively using serial CT imaging. We have now demonstrated a novel approach to analyzing joint shape change that permits a visual approximation of where cortical drift is occurring on the proximal shell of the MTP.
Acknowledgements, Funding, and Conflicts of Interest: None to disclose