How does a cadaver model work for testing ultrasound diagnostic capability for rheumatic-like tendon damage?

To establish whether a cadaver model can serve as an effective surrogate for the detection of tendon damage characteristic of rheumatoid arthritis (RA). In addition, we evaluated intraobserver and interobserver agreement in the grading of RA-like tendon tears shown by US, as well as the concordance between the US findings and the surgically induced lesions in the cadaver model. RA-like tendon damage was surgically induced in the tibialis anterior tendon (TAT) and tibialis posterior tendon (TPT) of ten ankle/foot fresh-frozen cadaveric specimens. Of the 20 tendons examined, six were randomly assigned a surgically induced partial tear; six a complete tear; and eight left undamaged. Three rheumatologists, experts in musculoskeletal US, assessed from 1 to 5 the quality of US imaging of the cadaveric models on a Likert scale. Tendons were then categorized as having either no damage, (0); partial tear, (1); or complete tear (2). All 20 tendons were blindly and independently evaluated twice, over two rounds, by each of the three observers. Overall, technical performance was satisfactory for all items in the two rounds (all values over 2.9 in a Likert scale 1-5). Intraobserver and interobserver agreement for US grading of tendon damage was good (mean κ values 0.62 and 0.71, respectively), with greater reliability found in the TAT than the TPT. Concordance between US findings and experimental tendon lesions was acceptable (70-100 %), again greater for the TAT than for the TPT. A cadaver model with surgically created tendon damage can be useful in evaluating US metric properties of RA tendon lesions.