Radiographic assessment of the femorotibial joint of the CCLT rabbit experimental model of osteoarthritis

BACKGROUND The purposes of the study were to determine the relevance and validity of in vivo non-invasive radiographic assessment of the CCLT (Cranial Cruciate Ligament Transection) rabbit model of osteoarthritis (OA) and to estimate the pertinence, reliability and reproducibility of a radiographic OA (ROA) grading scale and associated radiographic atlas. METHODS In vivo non-invasive extended non weight-bearing radiography of the rabbit femorotibial joint was standardized. Two hundred and fifty radiographs from control and CCLT rabbits up to five months after surgery were reviewed by three readers. They subsequently constructed an original semi-quantitative grading scale as well as an illustrative atlas of individual ROA feature for the medial compartment. To measure agreements, five readers independently scored the same radiographic sample using this atlas and three of them performed a second reading. To evaluate the pertinence of the ROA grading scale, ROA results were compared with gross examination in forty operated and ten control rabbits. RESULTS Radiographic osteophytes of medial femoral condyles and medial tibial condyles were scored on a four point scale and dichotomously for osteophytes of medial fabella. Medial joint space width was scored as normal, reduced or absent. Each ROA features was well correlated with gross examination (p < 0.001). ICCs of each ROA features demonstrated excellent agreement between readers and within reading. Global ROA score gave the highest ICCs value for between (ICC 0.93; CI 0.90-0.96) and within (ICC ranged from 0.94 to 0.96) observer agreements. Among all individual ROA features, medial joint space width scoring gave the highest overall reliability and reproducibility and was correlated with both meniscal and cartilage macroscopic lesions (rs = 0.68 and rs = 0.58, p < 0.001 respectively). Radiographic osteophytes of the medial femoral condyle gave the lowest agreements while being well correlated with the macroscopic osteophytes (rs = 0.64, p < 0.001). CONCLUSION Non-invasive in vivo radiography of the rabbit femorotibial joint is feasible, relevant and allows a reproducible grading of experimentally induced OA lesion. The radiographic grading scale and atlas presented could be used as a template for in vivo non invasive grading of ROA in preclinical studies and could allow future comparisons between studies.