Microstructural Modeling of Collagen Networks in Ligament Using Confocal Microscopy

Ligaments are bands of dense fibrous connective tissue that provide joint stability by connecting bone to bone and are composed of ground substance and fiber reinforcement. Ligament injuries often lead to chronic joint disorders, such as osteoarthritis, due to mechanical deficiencies of the repaired tissue. The underlying structural cause of these mechanical deficiencies is unclear, but is likely related to the fibrous network of load-bearing collagen within ligament, as collagen networks in healthy ligament are aligned but become disorganized after injury. The structure-function relationship of collagen networks can be investigated using microstructural models that mathematically describe the relationship between stress, strain, and structural features of materials. In this research, the variability in collagen networks was accounted for by including a collagen distribution in the standard unidirectional microstructural ligament model, creating a multidirectional model. To validate the multidirectional model, two healthy bovine ligaments with variable collagen networks–the lateral collateral sesamoid ligament (LCSL) and the palmar annular ligament (PAL)–were mechanically tested in the longitudinal and transverse configuration. To image the structure of collagen networks, bovine ligament samples autofluorescence was captured using confocal microscopy. To quantify the collagen structure, the fast Fourier transforms (FFT) of confocal images were determined. An elliptical method determined the preferred collagen fibril orientation θp. A least-squares curve fitting algorithm using the semicircular von Mises distribution predicted the collagen fibril distribution term k. The unidirectional model was fit to experimental data using an established methodology where stress in the transverse configuration is purely due to ground substance. The multidirectional model was fit using two methods. First, the model was fit with strong ground substance stress in the transverse configuration. The model was alternatively fit with stronger collagen network stress in the transverse configuration. Imaging results predicted significantly different k values between the LCSL and PAL where k = 3.97 ± 1.84 for LCSL, k = 1.93 ± 0.52 for longitudinal PAL, and k = 1.01 ± 0.32 for transverse PAL. The study found that the multidirectional model with strong ground substance better correlated to experimental mechanical results for longitudinal and transverse tests of the PAL (R^2 = 0.97 and R^2 = 0.69, respectively) as compared to the unidirectional model (R^2 = 0.92 and R^2 = 0.42, respectively). The multidirectional model with stronger collagen network stress accurately predicted experimental mechanical results for longitudinal and transverse tests of the PAL (R^2 = 0.92 and R^2 = 0.97, respectively). Results show that accurate fitting of the multidirectional model requires stronger collagen networks in the transverse configuration indicating that collagen networks resist loads in the transverse configuration. By fitting the multidirectional model accurately, the model holds potential to predict mechanical function based solely on structural adaptations. This multidirectional model can serve as a useful tool in advancing a mechanistic understanding of ligament healing and developing treatment strategies to strengthen ligament repair