Human Disc Nucleus Properties and Vertebral Endplate Permeability

Dissection

Thirteen freshly frozen cadaveric human lumbar spines from levels L1–L5 and L1–L4 (age range 32–85 years; average age 63 ± 16 years; 4 females and 9 males) were obtained from donor banks. Overall, 51 motion segments were harvested. The specimens were kept at −20°C until they were thawed for MRI assessment. The intervertebral disc (IVD) specimens were scanned in a 3T MRI Scanner (GE Healthcare, Milwaukee, WI) and graded by three trained radiologists using the Pfirrmann degeneration scale.²⁰ In order to obtain the average IVD height, three locations were selected and measured from mid-sagittal MRI images using NIH Image software: one from each lateral side and one at the center. The posterior processes were removed using a bone saw (Exakt Model 300, Band Saw, Norderstedt, Germany). Subsequently, the surrounding muscle and other soft tissue were dissected away from the spine. The motion segments were then cut transversely in each vertebra using a bone saw to obtain specimens consisting of half-vertebra/disc/half-vertebra (Figure 1). Next, they were refrozen for the following two steps, to preserve nuclear tissue. The motion segments were cored at the center of the nucleus pulposus using an 8.25 mm (inner diameter) diamond coring tool (Starlite Industries, Rosemont, PA) and a drill press. In the transverse plane, the center of the nucleus pulposus was approximated by marking a point one-third of the anterior-posterior length along the sagittal midline. Next, the adjacent vertebra/endplate core was separated from the nuclear tissue. In order to keep the cartilage endplate intact, the nucleus tissue was cut approximately 1 mm away from the cranial and caudal ends of the adjacent vertebra. Then, the nucleus was carefully divided in three roughly equal parts along the axial length of the nuclear core and labeled in reference to the adjacent vertebrae: superior, center and inferior (Figure 1). The 102 vertebral/endplate cores were stored at −20°C for later use in microCT imaging.

Imaging

The vertebral bone cores were thawed and imaged in a commercial micro-computed tomography system (μCT 40, Scanco Medical, Brüttisellen, Switzerland) with an X-ray tube voltage of 70 kV and 180o acquisition. Each core was placed in a cylindrical sample holder in a bath of protease inhibitors diluted 1:10 with distilled water (P2714 Protease inhibitor cocktail, Sigma-Aldrich, St Louis MO) to keep the cartilage in the specimens from degrading.²¹ A spatial resolution with an isotropic voxel size of 8 μm (matrix 2048 × 2048, 1000 projections/180o, FOV 16.4 mm) was chosen. Specimens were scanned along the length of each core, encompassing the vertebral endplate surface and 3 mm of bone underneath it. Next, image cross sections were reconstructed to create a 3D structure using the manufacturers’ cone beam reconstruction algorithm.²² Then each 3D structure was transformed into 1000–1100 serial-8 μm thick-sliced images in a sagittal orientation. Next, a light Gaussian filter (to remove high frequency noise) followed by a fixed threshold was applied (to segment the images into a bone and pore phase).²³ The irregular vertebral endplate surface was then identified using a custom made algorithm developed using MATLAB software. The semiautomatic technique is fully described elsewhere.²⁴ For each sagittal μCT image, this algorithm identified a single region of interest (ROI) that included an 8 μm (one pixel) thick endplate surface contour. For each contour, the bone volume fraction (BV/TV) was calculated, where BV is bone volume and TV is tissue volume (equivalent to the ROI volume). The porosity was calculated as the average of 1-BV/TV for the series of sagittal images for each specimen.

Permeability testing

After the vertebral bone specimens were scanned, the specimens were trimmed to approximately 5 mm from the vertebral endplate surface using a low concentration diamond wafering blade on a Buehler Isomet Circular Low Speed Saw. The marrow was then removed from the trabecular bone using a water pik (Waterpik, CO) and the cores were stored at −20° C. A custom built permeameter was used to measure the permeability of the endplate cores. The permeameter consists of a series of pipes connected to a pressurized fluid reservoir (Figure 2) and was validated using porous stainless steel discs of known permeability (Mott Corp, Farmington, CT).²⁵

Permeability measurements were performed on 51 vertebral/endplate cores, one from each motion segment. The locations inferior or superior were randomized and the matching core was saved for a future study. To assess the permeability of the cylindrical core samples, they were first thawed and then placed in perforated cryogenic vial caps (Cryogenic Vial, Corning Cat. No.2018). Then, cyanoacrylate glue was carefully added to seal the cartilage endplate periphery to prevent leaks around the specimen. The samples were oriented so that the driving hydraulic pressure was against the cartilage endplate so that fluid passed in the flow-out, disc to bone direction.²⁶ We chose the flow-out direction for two reasons. First, several groups have demonstrated that nucleus pressure compacts cartilage endplate leading to decreased matrix pore size and lower permeability.^26,27 Because of this, we adjusted the apparatus flow to correspond with physiologic disc pressures so cartilage endplate compaction in our experiment would match that present in vivo. In that way, we expect our data to be more representative of in situ conditions. A similar choice was made by other groups^28,29 and consequently this also allowed a representative comparison to historical data. Second, the cartilage endplate is only loosely adherent to the subchondral bone, and our preliminary attempts at measuring permeability in the flow-in direction damaged the specimens by separating tissues. The cartilaginous portion of the sample in the cryogenic vial was equilibrated in protease inhibitors diluted 1:10 in distilled water for half an hour to reduce cartilage degradation. The sample in the vial was then inserted in the sample mount and ready for testing. The permeameter was filled with deionized water and pressurized to 1 MPa – the average physiological pressure of the disc.³⁰ The fluid flow rate was calculated by measuring the weight of the fluid passing through the specimen at two-second intervals using a precision electronic balance (Ohaus, Cat No. AV2102CU-US, 2100g capacity, 0.01g resolution). The upstream fluid pressure (P_u) was generated using a nitrogen tank and was measured by a pressure sensor (Honeywell Sensotec, Columbus, Ohio, Model A-205). The downstream pressure (P_d) was atmospheric. Specimen’s permeability (k) was calculated using Darcy’s Law to relate the Darcy velocity (v) to the pressure drop across the specimen (P_u − P_d):

where Q is the flow rate (m³/s), A is the cross-sectional area for fluid flow, μ is the fluid viscosity (Pa-s), the Darcy permeability constant k, is the intrinsic property of the tissue (m²), and L_s is the specimen length (m). Each specimen’s permeability was measured twice, once intact, and a second time; after careful removal of the cartilage endplate using blunt dissection and Toluidine blue tissue marker to differentiate the cartilage from the bone. The cartilaginous endplate thickness (L_cep) was then measured using a resistance micrometer.³¹ The cartilage endplate permeability was calculated assuming ‘permeabilities in series’: that is, for the intact specimen, the flow is equal through both the cartilage endplate (cep) and subchondral bone (bone) (see also Figure 2):

From the intact specimen flow experiments, total permeability (k_total) in Equation 1 was determined. From the bone-only flow experiments, bone permeability (k_bone) was determined. Using these quantities, and the pressure-drop/flow-rate relationships from the intact specimen test, cartilage endplate permeability (k_cep) was determined from Equation 2 by first solving for P_d1 and then for k_cep. The final results were reported in fluid mobility units (m⁴/N-s) and defined as the tissue permeability normalized by the viscosity of the fluid (k/μ).^32–34

Biochemistry

Cell density was quantified by measuring DNA fluorescence using the Picogreen assay method (Molecular Probes Invitrogen Detection Technologies, Eugene, OR) following the manufacturer’s protocol. First the stored nuclear tissue was thawed out for 3 minutes. Each nuclear section was finely minced, weighed to obtain wet weight, and then lyophilized in order to obtain dry weight. Samples were digested using papain (Sigma-Aldrich) prepared at a dilution of 1:100 in PBS for 24–36 hours at 60° C and then centrifuged at 3000 rpm for 10 minutes. After digestion, a pellet containing cells and supernatant containing GAGs were separated. The pellet was then resuspended in 300 mL of TE buffer (Sigma-Aldrich) and used to estimate nucleus cell density, while the supernatant was used to determine nucleus glycosaminoglycan (GAG) content. In a validation study, it was determined that the amount of DNA in the supernatant was negligible compared to the amount of DNA found in the pellet; for that reason the DNA was only obtained from the pellet. The amount of DNA was quantified by using the results from a standard curve produced using a calf thymus DNA concentration. Finally, the cell density was calculated by dividing the amount of DNA by 6 picograms of DNA which are contained per cell.³⁵

Nucleus GAG content was quantified using the dimethylmethylene blue assay (DMMB) method with the supernatant from the nucleus digestion.³⁶ The supernant was diluted to 1:200 and then 40 μL of the solution was added to 250 μL of the DMMB dye solution. A wavelength reading was measured using a spectrophotometer (SpectraMax M3, Sunnyvale, CA). The GAG content was calculated based on a standard curve using an aqueous solution of chondroitin sulfate C from shark cartilage (chondroitin 6-sulfate, Sigma Aldrich, St. Louis, MO). The results were normalized by volume (mm³) from the water content previously obtained.

Data Analysis

All statistical analyses were performed using the JMP (Version 7.0, SAS Institute Inc.). Interobserver variability was by determined by a kappa score. Means and standard deviations for the variables were obtained. Standard analysis of variance (ANOVA) and analysis of covariance (ANCOVA) procedures were performed to compare specimen group means and to estimate the effect of the specimen variables (degeneration grade entered as an ordinal predictor, and porosity, cartilage endplate thickness, and disc height entered as continuous predictors) on the measured parameters of interest (mobility, cell density, GAG content, GAGs/cell). When appropriate, Tukey post-hoc tests were used to identify group differences. Standard least squares multiple regression models were used to investigate relationships between variables, with the strength of association quantified by the correlation coefficient R².

Endplate Mobility, Porosity & Cartilage Thickness

The mean total mobility was 3.26 × 10⁻¹⁰ ± 4.43 × 10⁻¹⁰ m⁴/N-s. For the cartilage endplate alone, the mobility was 1.19 × 10⁻¹⁰ ± 1.64 × 10⁻¹⁰ m⁴/N-s while that for the subchondral bone alone was 2.21 × 10⁻⁹ ± 1.37 × 10⁻⁹ m⁴/N-s. Total Mobility (TM) had a positive correlation with age and bone porosity (R²=0.16, p<0.01; TM (x 10⁻¹⁰) = 0.11*age − 3.80 and R²=0.09, p=0.05; TM (x 10⁻¹⁰) = 16.13*bone porosity − 5.33, respectively). Bone Mobility (BM) also had positive correlation with age and bone porosity (R²=0.10, p=0.03; BM (x 10⁻¹⁰) = 0.27*age + 4.89 and R²=0.23, p<0.0001; BM (x 10⁻¹⁰) = 75.15*bone porosity − 18.19, respectively). In addition, bone mobility had an inverse relationship with GAG content (R²=0.16, p=0.004). The mean endplate porosity was 54.6 ± 7.5%, and was positively correlated with age (R²=0.13, p<0.01, Figure 3) and degeneration (R²=0.22, p=0.004). It was observed that cartilage thickness increased with age (R²=0.08, p=0.03). A one-way repeated measures ANOVA indicated there was no statistically significant degeneration trend with spinal level with donor as a covariate (p=0.94).

Cell density

The mean cell density was 1651 ± 1015 cells/mm³ (Table 1). The results of the one way repeated measures ANOVA demonstrated that the three cell density regions were not statistically different. Therefore the average cell density was used. There was a significant decrement between grade 1 and 2 discs, then a progressive increase with degeneration grades 2 through 5 (Figure 4). When specimen factors were analyzed individually, cell density decreased as disc height increased (R²=0.13 p<0.02), but was not directly related to subchondral bone porosity (p>0.5), total mobility (p>0.4), or age (p>0.2). Assuming an exponential relationship between cell density and disc height⁷ gives

where, CD is cell density (cells per mm³), and h is disc height (mm); (R²=0.15, p=0.005, Figure 5). When controlling for disc height, cell density was inversely correlated to age (R²=0.26, p<0.003).

GAGs

The results of the one way repeated measures ANOVA demonstrated that the three GAG content regions were not statistically different. Therefore, the average cell density was used. Average GAG content non-linearly decreased with age (R²=0.83, p<0.0001; average GAG=230118*age^−2.49, Figure 6) and Pfirrmann grade (R²=0.52, p<0.0001), with grade 1 significantly higher than grades 2–5, and grade 4 being significantly lower than grades 1 and 2. When controlling for age, GAG content did not depend on endplate permeability (p>0.05). However, endplate porosity decreased as GAG content increased (R²=0.20, p<0.002). Disc height was non-linearly correlated with average GAG content, where height decreased precipitously when GAG content was below 10 μg/mm³ (R²=0.20; p<0.001).

GAGs/cell

GAG content was normalized by cell density as an indirect measure of cell function. Average GAGs/cell decreased with degeneration, with statistically-significant differences between grades 1–3 and grade 4–5 (R²=0.24; p<0.0001, Figure 7). It was also weakly correlated with disc height (R²=0.17, p<0.01) and inversely correlated with age (R²=0.14, p<0.01). There was no relationship with mobility (p>0.3), or porosity (p>0.07). When modeled together, disc height and age explained 25% of the variation of GAGs/cell (p=0.001). When GAGs/cell was calculated for the endplate-adjacent regions only, this was weakly inversely correlated with adjacent bone porosity (R²=0.05, p=0.04).

Degeneration

Pfirrmann grade increased with age (R²=0.31, p=0.002), with grades 1, 3 & 4 being statistically different. With increasing degeneration, subchondral bone porosity increased (R²=0.22; p<0.0004) and disc height decreased: height of grade 3 discs was less than grade 2, and grade 4 and 5 discs were less than grades 2 and 3 (R²=0.57, p<0.0001).