ABSTRACT
OBJECTIVE: Successful repair of articular cartilage (AC) defects would be a major advantage due to the low ability of AC to heal spontaneously. Sensitive methods to determine changes in AC composition and structure are required to monitor the success of repair. This study evaluates the ability of unsupervised cluster analysis applied to Fourier transform infrared (FTIR) microspectroscopy to discriminate between healthy and repaired AC. METHODS: Osteochondral lesions (3 mm in depth) were surgically created in patellar grooves of rabbit femurs and were either left to heal spontaneously (n = 6) or surgically repaired with autologous chondrocytes in type II collagen gel (n = 6). After 6 months, tissues were harvested, FTIR microspectroscopy was conducted and Fuzzy c-means (FCM) cluster analysis applied to spectra of pairs of intact and repaired AC samples from each rabbit. Two spectral regions [amide I and carbohydrate (CHO)] were analyzed and the results from the two types of repair were compared. RESULTS: Two separate regions of repair were detected with FCM. The estimated proteoglycan content (from CHO region) in the repaired AC was significantly lower than that in intact AC. The spontaneously repaired AC was better distinguished from the intact AC than the collagen II gel repaired AC. The most distinct clustering was observed for spontaneously repaired samples using CHO region. CONCLUSIONS: This study revealed that unsupervised cluster analysis applied to FTIR microspectroscopy can detect subtle differences in infrared spectra between normal and repaired AC. The method may help in evaluation and optimization of future AC repair strategies.
