Multimodal Characterization of Articular Cartilage Degeneration in the Humeral Head Using Raman Spectroscopy, Biomechanics, and Imaging

Tissue Samples

The humeral heads from eleven fresh frozen disarticulated cadaveric shoulders (6 Male/5 Female) with a mean age of 61 ± 4 years old were used in this study (Medcure, Providence, RI, USA). The specimens were then divided into two groups based on visual macroanalysis of the humeral head cartilage. All specimens were obtained in this age range to ensure both groups were of similar age, and we were not comparing young, healthy samples to older degenerated samples. Specimens with no visual signs of cartilage degeneration were designated to the intact control group (N = 4; mean age 62 ± 3) and specimens with visual focal defects to the cartilage were designated to the defect group (N = 7; mean age 60 ± 4). Sites with full cartilage erosion (exposed subchondral bone) were not subjected to analysis. Of the humeral heads that had evidence of degeneration, 3 came from specimens where there was a concurrent rotator cuff tear identified upon dissection by inspection of the supraspinatus tendon. The humeral heads were then sectioned using a rectangular grid pattern to create square osteochondral blocks using a fine-tooth handsaw (N = 7–10 blocks/humeral head)¹⁹. Overall, this yielded 90 osteochondral samples across all donors for subsequent testing. Specimens that had no signs of macro-degeneration were assigned to the intact group (n = 66 osteochondral specimens) and those with visual focal defects were assigned to the defect group (n = 24 osteochondral specimens). Tissue samples were subjected to Raman spectroscopy, mechanical indentation, biochemical analysis (GAG, COL, H₂O content), μCT imaging, histology, and Outerbridge visual grading. All specimens were kept hydrated in 1× PBS and stored at −20 °C.

Raman Spectroscopy

The Raman instrumentation is comprised of a custom Raman probe, with a threaded, hollow needle tip possessing a distal mounted ∅2mm diameter sapphire ball lens (InPhotonics, Norwood, MA, USA), coupled to an NIR diode laser (ex = 785 nm; 125 mW output; IPS) and a spectrometer (Eagle Raman-S, Ibsen Photonics, Farum, Denmark) equipped with a near infrared (NIR) optimized charge coupled device (Ivac, Andor, Belfast, Northern Ireland) Andor)¹⁹. The distal end of this needle probe contains a ∅2mm sapphire ball lens, with the focal distance being close to zero requiring direct contact with the cartilage surface for sufficient spectral acquisition. For Raman spectral acquisitions, the probe tip was positioned perpendicular to the articular surface of each osteochondral explant (N = 90; N = 66 Intact, N = 24 Defect), and spectra were acquired using a 30s integration time. Following subtraction of the probe background, a 3^rd order polynomial baseline correction and subsequent area-under-curve normalization were applied to the 800–1800 cm⁻¹ spectral “fingerprint” region. The resulting “fingerprint” spectrum was decomposed using the following multivariate regression model²⁰:

The derived regression coefficients (${\mathrm{GAG}}_{\mathrm{score}}$, ${\mathrm{COL}}_{\mathrm{score}}$, ${\mathrm{H}}_2{\mathrm{O}}_{\mathrm{score}}$, ${\mathrm{Bone}}_{\mathrm{Score}}$) reflect the relative contribution of each individual constituent to the overall composite spectrum, thereby providing a non-destructive assessment of individual osteochondral explant composition (Figure 1). GAG_REF, Collagen_REF, H₂O_REF, and Bone_REF are reference spectra measured from purified chemicals of each constituent (Figure 1C): GAG_REF= chondroitin sulfate [bovine cartilage; C6737 Sigma]; Collagen_REF= type-II collagen [chicken sternal cartilage; C9301 Sigma]; H₂O_REF= PBS; Bone_REF= Mature bovine subchondral bone.

Mechanical Testing

Each osteochondral explant (N = 90) was subject to indentation testing at the center of the cartilage layer using a custom mechanical testing machine (CellScale Biomaterials, Waterloo, ON, Canada). A ∅2mm diameter spherical indenter was used to apply a 100 μm tissue displacement at a loading rate of 5 μm/s to obtain the cartilage force-displacement response. The compressive modulus was then obtained assuming Hertzian contact theory for the loading curve²¹.

Outerbridge Scoring

A visual assessment of the cartilage breakdown was performed for all specimens with a visual defect (N = 24) and a random subset of visually intact specimens (N = 10) to assess the relationship between clinical assessments of cartilage and mechanical function. A single experienced orthopaedic shoulder surgeon graded the specimens according to the Outerbridge classification system²². The Outerbridge classification was chosen because it represents the current convention for assessing cartilage health during arthroscopic procedures and has previously been used to assess shoulder joint articular cartilage²³.

Biochemical Analysis

A central ∅4mm chondral sub-punch was collected from the center of each osteochondral explant (N = 76; N = 56 Intact, N = 19 Defect), co-localized with the region subjected to mechanical and Raman assessments. Upon separation of the cartilage from the bone, the wet weight of the cartilage was recorded. Following lyophilization, the tissue dry weight was measured, providing the gravimetric water content. Subsequently, the tissue was digested with proteinase K and analyzed for GAG and collagen content using 1,9-dimethylmethylene blue (DMMB) and orthohydroxyproline (OHP) assays, respectively.

Histological Evaluation

A random subset of degenerated (N = 3) and intact (N = 3) osteochondral specimens was fixed in formalin, embedded in paraffin, sectioned, and stained with safranin-O/fast green to assess the presence of GAG. Slides were imaged with a 20× objective on a standard bright-field microscope to provide representative images of intact and degenerated tissue (Figure 3A and 3 B).

Micro-computed Tomography

Osteochondral specimens were scanned (N = 23 degenerated, N = 19 intact controls from defect-free humeral heads) (μCT40, SCANCO Medical, Brütisellen, Switzerland) under the following conditions: The x-ray tube was operated at 70 kVp, an intensity of 114 mA and an integration time of 300 ms. This resulted in an isotropic voxel size of 15μm. The bone density and micro-architectural indices (Bone Mineral Density (BMD, mgHA/cm³ ), Bone Volume Fraction (BV/TV, mm³/mm³), Trabecular Thickness (Tb.Th, mm), and Connectivity Density (Conn.D, 1/mm³) were calculated along the central 20 slices of each specimen, where a 2mm volume of interest was manually drawn directly adjacent to the cartilage layer, allowing for the evaluation of changes in trabecular bone.

Statistical Analysis

All data is presented as median (interquartile range). We used the proc mixed from the SAS software, version 9.4 for the Windows System to implement a mixed model for testing the differences in the selected “compositional/mechanical/ μCT variables” (i.e., Modulus, GAG, COL, H2O, BMD, BV/TV, Conn.D, Tb.Th) by the presence of “damage” (i.e., Intact or Defect; RCT or RCT + OA) in the “cartilage sample” (i.e., each specimen). We used a linear mixed-effects model with random intercepts to address multiple measurements from the same subject. Spearman’s rank test was used to determine the correlation between Outerbridge grade and compressive modulus. A nonlinear (2^nd order polynomial) regression was performed to determine the extent to which variation in the Raman GAG_score accounted for the variation in compressive modulus as assessed by the coefficient of determination (adjusted R²). Two-tailed p-values < 0.05 were considered significant. The specimens obtained were based on what was available, and no a priori sample size or power calculations was performed.

Spatial Maps of Humeral Head Degeneration from Raman Spectroscopy

Spatial maps of the cartilage mechanical function and Raman-derived biomarkers for specimens with increasing levels of degeneration are shown in Figure 2 to illustrate the compositional and mechanical changes that occur as cartilage degenerates. Shown are two intact controls, one specimen with a small defect (< 0.5 cm²) and one specimen with a larger defect (> 1.5 cm²), to highlight the changes in the degenerated regions as the focal defect increases in size. Here we see the progression from an intact control with seemingly uniform mechanical and compositional properties, to the complete loss of mechanics and composition when the defect comprises most of the humeral head.

Cartilage Mechanical Health

Cartilage degeneration resulted in a significant decrease in mechanical function, as evidenced by a reduction in the compressive modulus. The osteochondral specimens with focal defects had a lower elastic modulus than the intact group (Table 1 & Figure 3A). Clinically, cartilage health can be assessed during arthroscopic procedures, where the degree of degeneration can be graded on the Outerbridge scale (0-IV). Therefore, we compared the relationship between Outerbridge grade and the elastic modulus to assess if the gold standard for cartilage degeneration can explain mechanical degradation. In the osteochondral specimens with defects and 10 randomly chosen visually intact specimens (Outerbridge Grade 0). No correlation (p = 0.13) between the elastic modulus and Outerbridge classification (Figure 3B) was found in this subset of specimens.

Histological Assessment of Cartilage

The three intact (Figure 3C) and defective (Figure 3D) specimens were qualitatively evaluated to verify the presence/absence of GAG. In intact samples, we observe uniform cartilage thickness and GAG in the deep zone. Although one sample (Figure 3C–3) appears to have a loss of GAG, which could be attributed to early degenerative changes, where cartilage thinning has not yet occurred. Furthermore, there is a lack of GAG in the superficial layer, which may be attributed to the advanced age of these specimens, as they are likely to have undergone degeneration due to aging²⁶. Similarly, the advanced age of these specimens may explain the absence of a GAG signal in the cartilage's superficial region. In the defect samples, there is heterogeneity in the cartilage thickness, where there is a loss of GAG signal directly under the site of cartilage thinning (Figure 3D).

Comparison of Cartilage Degeneration to Intact Controls

Regions containing focal defects exhibited significant compositional changes relative to intact controls, as measured by both Raman spectroscopy and biochemical analysis (Table 1 and Figure 4). Raman spectroscopy and biochemical analysis revealed a significant reduction in GAG levels between intact and focal-defect samples. Biochemical analysis also revealed a significant increase in tissue hydration. While the Raman-derived H₂O and COL did not reach significance between the defect and controls, there was a slight decrease in COL and an increase in H₂O. Raman-derived biomarkers were also found to have positive correlations to their corresponding biochemical assays (GAG_score vs GAG content, R² = 0.54, p < 0.001; COL_score vs. COL content, R² = 0.19, p = 0.016; H₂O_score vs H₂O content, R² =0.19, p < 0.001).

Correlation between Raman-derived biomarkers and mechanical function

Overall, intact cartilage had a higher compressive modulus than regions with focal lesions. As a result of degeneration, there was a loss of GAG within the cartilage, which was seen in both the GAG_score and GAG content. A 2^nd-order polynomial fit using the Raman-derived GAG_score captured 68% of the variability in the compressive modulus of the osteochondral specimens (Figure 5), indicating that stiffer tissue was associated with higher GAG_score (adjusted R² = 0.68, p < 0.001).

Bone microarchitecture

μCT imaging revealed changes to the underlying trabecular structure of the specimens with a defect (Table 1 and Figure 6). There was a significant reduction in BV/TV (p = 0.011). Yet, there were no statistically significant differences in BMD, Conn.D, or Tb.Th.

Comparison of Degeneration with RC Injury

To assess if RC injury results in differences in the compositional and mechanical changes of cartilage, we compared the Raman-derived biomarkers and compressive modulus of the osteochondral samples with macroscopic damage (focal defects) from humeral heads with RC tears versus those without RC tears (Figure 7). This resulted in n = 10 osteochondral specimens in the RC tear + OA groups and n = 14 osteochondral specimens in the OA only group. Overall, no statistically significant differences in the compositional or mechanical properties between specimens with RCT + OA versus OA (Table 2 and Figures 7A–D) were seen. No statistically significant differences were found between the RCT + OA group compared to the OA only group for BV/TV, BMD, Conn.D, or Tb.Th.

Cartilage degeneration was generally seen in the central portion of the humeral head (Figure 7), consistent with previous studies evaluating end-stage OA^{10, 11}. However, different phenotypes of OA may exist due to concurrent pathology, such as RC arthropathy. Three of the humeral heads with cartilage degeneration also had RC tears; therefore, we plotted the location of degeneration to qualitatively identify changes in the wear pattern. Cartilage degeneration was located centrally when there was no RC tear (Figure 7G); however, the specimens with a RC tear (n = 3) exhibited an inferior shift in the degeneration sites (Figure 7H).