In this thesis, we investigated the capability of Raman spectra to predict the composition, structure, and functional properties of healthy and degraded articular cartilage. The enzymatic treatment induced biochemical changes in the cartilage matrix. On the other hand, the mechanical damage resulted in structural changes to the tissue. These structural changes present themselves in the Raman spectrum as changes in peak width and/or shift in position. In this thesis, we posit that using ML algorithms, within the full fingerprint region of the spectrum, would provide the needed data for the algorithms to detect the structural changes along with biochemical ones.
All groups, except COL24, showed higher PG content relative to control (Table 7), this can be attributed to the osteochondral specimens being taken from different parts of the patella as was shown in Figure 10, where different positions result in different load-bearing requirements.
Studies have shown that non-load-bearing locations exhibit low PG content113. Control group plugs were taken from the top of the patella, while the rest of the plugs were taken from the center, where higher loading scenarios occur. This does not contradict the fact that control samples showed the best biomechanical properties as they were not degraded. Nevertheless, Raman spectra of the degraded groups showed a corresponding significant increase in the SO3−
1060 cm-1 peak of chondroitin sulfate, a side chain in aggrecan the most abundant proteoglycan, relative to the control group.
Apart from the previously mentioned chondroitin sulfate peak, related to PG, the IMP damage spectrum did not show any significant changes from the control spectrum. This suggests only a higher PG content relative to the control group, due to different anatomical locations, a change confirmed by the digital densitometry measurements. The drastic drop in biomechanical properties of the impact samples can be attributed to the rupture of the collagen fiber network, resulting in a decrease in proteoglycan content and a corresponding increase in water content114,115. Although no changes in superficial collagen orientation were observed, the effect of the chondral rupture can be observed in the average orientation of the middle zone.
45
The ABR group spectra (Figure 17) showed an increase in the glycosaminoglycans (GAGs) related peaks, which is consistent with the higher PG content relative to the control group, which is also confirmed by the digital densitometry measurements. There was a slight decrease in the average biomechanical properties (Table 7) of the samples in this group. These changes can be credited to the induced fibrillation (fissures) from the abrasion process. Although this fibrillation damages the integrity of the superficial layer which protects and maintains the deeper layers3, eventually leading to OA116, this progression often takes years. This is backed up by the reference measurements as the samples display a healthy distribution of PG content across all zones, accompanied by a healthy collagen orientation. The overall average orientation increase, relative to control, can be attributed to the loss of cartilage thickness, which likely caused the averaging of a portion of the middle zone with the superficial and the same with the middle and deep.
The mean Raman spectra of the COL24 group had an overall low signal-to-noise ratio, which may be attributed to the high surface roughness of the degraded samples, resulting in the sample surface being out of focus during the scan. For this group, the whole spectrum showed a significant decrease relative to the control mean. This can be attributed to severe disruption of the collagen fibrillar network and loss of PG due to the compromised network. For COL90, a slight decrease was observed in the amide III doublet, hydroxyproline, tyrosine, proline, and tryptophan bands, all compounds related to collagen, without much change to the GAGs bands ranging from 1425 to 1317 cm-1. This is supported by the reference values, as COL90 did not show any change in PG content, relative to the ABR samples taken from the spot next to it. On the other hand, COL24 resulted in a significant loss of PG content, relative to control and all other groups. As expected, both collagenase groups suffered a decrease in their biomechanical properties, especially the instantaneous and dynamic moduli. This is reasonable, as collagenase D cleaves all three alpha chains of fibrillar collagens, breaking the triple helical structure of collagen, which is responsible for the stiffness of collagen99. This likely ledto a decrease in the collagen’s function of withstanding mechanical load. Also, as collagen contributes to the form of cartilage and its thickness100, the loss in cartilage thickness in the collagenase groups is reasonable regarding their respective degradation time.
Finally, the GAGs bands ranging from 1425 to 1317 cm-1 showed a slight decrease in the TRYP30 group along with a slight decrease to hydroxyproline, tyrosine, and proline bands, suggesting a decrease in PG content, an expected outcome as trypsin cleaves peptides on the
46
C-terminal side of lysine and arginine residues (Sigma-Aldrich Inc., St. Louis, MO, USA), effectively lowering the PG content. As a result, the TRYP30 group showed a lower equilibrium modulus than control, ABR, and COL90. This is reasonable, given that cartilage proteoglycan content is responsible for equilibrium loading scenarios, as it provides the cartilage with its osmotic properties, which are critical to its ability to resist compressive loads.
47