Micro-computed tomography evaluation of vertebral end-plate trabecular bone changes in a porcine asymmetric vertebral tether

Authors

  • Jean-Michel Laffosse,

    Corresponding author
    1. Service de Chirurgie Orthopédique et Traumatologique—Institut Locomoteur, Centre Hospitalier Universitaire de Toulouse-Rangueil, 1 Avenue Jean Poulhès, TSA 50032, 31059 Toulouse Cedex 9, France
    2. Laboratoire de Biomécanique, Centre Hospitalier Universitaire, Purpan, France
    • Service de Chirurgie Orthopédique et Traumatologique—Institut Locomoteur, Centre Hospitalier Universitaire de Toulouse-Rangueil, 1 Avenue Jean Poulhès, TSA 50032, 31059 Toulouse Cedex 9, France. T: +33 5 61 32 27 12; F:+33 5 61 32 22 32.
    Search for more papers by this author
  • Thierry Odent,

    1. Service de Chirurgie Pédiatrique, Hôpital Necker Enfants Malades, Paris, France
    2. Département de Chirurgie, Ecole Nationale Vétérinaire de Lyon, Lyon, France
    Search for more papers by this author
  • Franck Accadbled,

    1. Laboratoire de Biomécanique, Centre Hospitalier Universitaire, Purpan, France
    2. Service de Chirurgie Orthopédique Pédiatrique, Hôpital des Enfants, Toulouse, France
    Search for more papers by this author
  • Thibault Cachon,

    1. Département de Chirurgie, Ecole Nationale Vétérinaire de Lyon, Lyon, France
    Search for more papers by this author
  • Charles Kinkpe,

    1. Service de Chirurgie Orthopédique Pédiatrique, Hôpital des Enfants, Toulouse, France
    Search for more papers by this author
  • Eric Viguier,

    1. Département de Chirurgie, Ecole Nationale Vétérinaire de Lyon, Lyon, France
    Search for more papers by this author
  • Jérôme Sales de Gauzy,

    1. Laboratoire de Biomécanique, Centre Hospitalier Universitaire, Purpan, France
    2. Service de Chirurgie Orthopédique Pédiatrique, Hôpital des Enfants, Toulouse, France
    Search for more papers by this author
  • Pascal Swider

    1. Laboratoire de Biomécanique, Centre Hospitalier Universitaire, Purpan, France
    Search for more papers by this author

Abstract

We conducted a micro-CT analysis of subchondral bone of the vertebral end-plates after application of compressive stress. Thoracic and lumbar vertebral units were instrumented by carrying out left asymmetric tether in eleven 4-week-old pigs. After 3 months of growth, instrumented units and control units were harvested. Micro-CT study of subchondral bone was performed on one central and two lateral specimens (fixated side and non-fixated side). In control units, bone volume fraction (BV/TV), number of trabeculae (Tb.N), trabecular thickness (Tb.Th), and degree of anisotropy (DA) were significantly higher, whereas intertrabecular space (Tb.Sp) was significantly lower in center than in periphery. No significant difference between the fixated and non-fixated sides was found. In instrumented units, BV/TV, Tb.N, Tb.Th, and DA were significantly higher in center than in periphery. BV/TV, Tb.N, and Conn.D were significantly higher in fixated than in non-fixated side, while Tb.Sp was significantly lower. We noted BV/TV, Tb.N, and Tb.Th significantly lower, and Tb.Sp significantly higher, in the instrumented levels. This study showed, in instrumented units, two opposing processes generating a reorganization of the trabecular network. First, an osteolytic process (decrease in BV/TV, Tb.N, Tb.Th) by stress-shielding, greater in center and on non-fixated side. Second, an osteogenic process (higher BV/TV, Tb.N, Conn.D, and lower Tb.Sp) due to the compressive loading induced by growth on the fixated side. This study demonstrates the densification of the trabecular bone tissue of the vertebral end-plates after compressive loading, and illustrates the potential risks of excessively rigid spinal instrumentation which may induce premature osteopenia. © 2009 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 28:232–240, 2010

The pathophysiology of intervertebral disc degeneration has been widely studied.1–3 It has been demonstrated that chronic compression can modify mechanical, histologic, biochemical, and cellular properties of the intervertebral disc.4–7 Moreover, relationships have been established on one hand between disc degeneration and the architecture of trabecular vertebral bone,8, 9 and on the other hand between cell death in annulus and bone mineral density of the vertebral end-plate.10 Moore et al.9 showed, in an adult ovine model, that an experimental outer annular tear was associated with an increase in the number of trabeculae, while their thickness and the intertrabecular space decreased. The authors linked these changes to alterations in the biomechanical behavior of the local structures. Simpson et al.8 showed a close relationship between disc degeneration and trabecular architecture of the adjacent vertebral bodies, where bone volume fraction increased with the degree of degeneration. Gruber et al.10 found a significant correlation between the percentage of cell death in the annulus and the bone mineral density of the vertebral end-plate as estimated by using a densitometer in an aging sand rat model. However, few studies have examined vertebral bone structure after application of compressive stress.11, 12 Stilwell11 observed, in a monkey model of scoliosis, apposition of cortical bone on the concave side and bone loss on the convex side of the curve. Pazzaglia et al.,12 in a growing rat model in which the tail vertebrae were subjected to asymmetric stress (compression/distraction), showed increased density and size of bone trabeculae in areas of compression. To our knowledge, no study has examined vertebral trabecular bone by micro-computed tomodensitometry (micro-CT) after application of compression in a growing animal. This technique allows precise analysis of the various morphological and three-dimensional parameters of bone tissue.13 It is the reference modality for examination of bone microarchitecture and it presents close correlations with morphological or mechanical findings.14

Our aim was to carry out a micro-CT study of changes of the bone trabecular network in response to chronic application of asymmetric compressive stress to a vertebral unit (vertebra-disc-vertebra) in a growing pig model.

MATERIALS AND METHODS

Surgical Procedure and Specimen Preparation

All experiments were carried out in accordance with the current regulations of the national veterinary schools of Lyon and Toulouse, France, on the care and use of experimental animals. Eleven skeletally immature 4-week-old domestic pigs were operated under general anesthesia in surgically aseptic conditions and were given pre- and postoperative prophylactic antibiotic treatment. Using a short posterior approach, two pedicle screws (4-mm diameter, 28-mm titanium screws; Medtronic®, Minneapolis, MN) were inserted and connected in situ by a rigid pin on the left side of the spine between T5–T6 and L1–L2. No local or general complication was observed. After 3 months of growth, the animals were euthanized with an overdose of penthotal. The thoracolumbar spines were removed en bloc and immediately immersed in 4°C isotonic saline solution. CT scan was carried out to confirm the correct positioning of the pedicle screws (Fig. 1). In one animal, one of the thoracic pedicle screws was not correctly positioned and this vertebral unit was excluded. We harvested the instrumented thoracic (T5–T6) and lumbar (L1–L2) vertebral units, as well as the thoracic and lumbar vertebral units situated two levels below the corresponding instrumented units, which were used as controls because they were considered sufficiently distant from the corresponding instrumented levels and from the initial surgical approach. The vertebral end-plates of each vertebral unit were separated carefully using a scalpel to avoid any damage to the vertebral end-plates. From each vertebral end-plate, we removed three cylindrical plugs (5-mm diameter and 6-mm high) by means of a cylindrical punch without damaging the trabeculae. These three samples consisted of one central sample (central group) and two lateral samples: one left lateral sample on the instrumented side (fixated side) and one right lateral sample on the opposite side to the instrumentation (non-fixated side) (Fig. 2). Cortical vertebral bone was not sampled. Each specimen was named according to the vertebral unit (thoracic instrumented or control, lumbar instrumented or control), to the vertebral end-plate (cranial and caudal vertebral end-plate with reference to the intervertebral disc) and to the location (central group, fixated side, non-fixated side). Specimens were frozen at 20°C in Ringer solution with 10% dimethyl sulfoxide (DMSO) cryoprotectant, within 6 h of sampling.15 They were later gradually thawed at room temperature for the microstructure study.

Figure 1.

Control CT-scan showing correct positioning of the pedicle screw on the axial section (A), and on sagittal (B) and coronal (C) 2D reconstructions. After 3-months growth, instrumentation resulted in asymmetric shape of the disc on the left side (gray arrow) as compared with the right side (white arrow) on 2D frontal reconstruction (C) and on the anterior view of an instrumented vertebral level after sampling (D).

Figure 2.

Three samples are harvested from each vertebral end-plate of thoracic and lumbar vertebral units. Black circles show pedicle screws' positioning related to vertebral end-plate. VEP, vertebral end-plate; NCS, non-compressed side; C, central; CS, compressed side.

Microstructural Analysis

A micro-CT system [µCT40 (spatial resolution 12 µm); Scanco Medical, Bassersdorf, Switzerland; http://www.scanco.ch/] was used to examine the microarchitecture of the trabecular subchondral bone of the vertebral end-plate. Each specimen was placed in a cylindrical sample holder in Ringer solution in order to prevent any movement and any alteration of the micro-CT parameters from drying during the scanning process. Each plug was scanned continuously with increments of 12-µm thickness for each slice. The voxel size was 12 × 12 × 12 µm. To avoid artefacts, boundaries and cartilage zones were carefully excluded, selecting regions of interest (ROI) strictly located in the center of the trabecular bone structure between the vertebral end-plate cartilage and the growth plate cartilage (Fig. 3). The ROI were at a distance from the vertebral body areas where the screws were placed. The same procedure was used to select all the ROIs in all plugs. The three-dimensional parameters used to describe trabecular bone structure were: bone volume fraction (BV/TV); specific bone surface (BS/BV); trabecular number (Tb.N); trabecular thickness (Tb.Th); trabecular separation (Tb.Sp); connectivity density (Conn.D); degree of architectural anisotropy (DA), which measures the orientation of the trabeculae within the ROI (the higher the DA, the more marked the alignment of the bone in a given direction relative to other directions); and the structure model index (SMI), which indicates the more plate-like or more rod-like structure of the trabeculae (SMI from 0 to 3, respectively).13

Figure 3.

The subchondral bone layer is the only region of interest for micro-CT examination. CEP, cartilage end-plate; GPC, growth plate cartilage.

Microscopic Study

Histologic study was carried out after micro-CT examination. Specimens were fixed in 10% formalin, embedded in paraffin, and stained with hematoxylin and eosin. The specimen was first cut longitudinally to obtain two symmetrical hemicylinders. A histologic section was taken from each hemicylinder and, on each section, two measurements of the thickness of the various layers were obtained, by microscope and digital camera. For each layer, the mean of the four values obtained was used for analysis (Fig. 4).

Figure 4.

Stages in sample preparation for histologic examination. (A) longitudinal sectioning, (B) flat histologic sections, (C) histologic view (original magnification, × 30). Dotted lines indicate lines along which the thickness of the different layers was measured in two locations. For a given layer, the mean of the four values was used for analysis.

Statistical Analysis

Statistical analysis was done using Statview® software (SAS Institute Inc., Cary, NC). A preliminary statistical study (data not displayed) showed no significant difference between thoracic and lumbar vertebral units or between the cranial or caudal end-plates regarding the intervertebral disc studied. The microstructural parameters of the instrumented vertebral units on the one hand, and the control units on the other, were pooled in order to make up the groups of our study. The normality of the distribution of the values has been verified using the Kolmogorov–Smirnov test, and Fisher's exact test was carried out to compare the variance of the groups. Student's t-tests were used to compare variables when distribution was normal, and non-parametric statistical methods were used when normality was not confirmed. Statistical significance was defined as p < 0.05.

RESULTS

The microstructural parameters of the instrumented and control groups are shown in Tables 1 to 3.

Table 1. Microstructural Parameters in the Control and Instrumented Groupsa
GroupControl GroupInstrumented GroupRelative Difference (%)p
  • BV/TV, bone volume density; BS/BV, specific bone surface; Conn.D, connectivity index; Tb.N, trabecular number; Tb.Th, trabecular thickness; Tb.Sp, trabecular separation; SMI, structure model index; DA, degree of anisotropy.

  • a

    Numerical data are given as the mean ± SD. The statistically significant differences are in italics.

  • b

    Unpaired t-test

  • c

    Non-parametric Mann–Whitney test.

BV/TV (%)36.79 ± 7.8833.98 ± 6.94−7.620.009b
BS/BV (1/mm)24.47 ± 4.1824.74 ± 3.755.410.03b
Conn.D (1/mm3)108.18 ± 25.15111.76 ± 32 3.310.78c
Tb.N (1/mm)3.63 ± 0.413.48 ± 0.44−3.940.019b
Tb.Th (µm)94.04 ± 14.68101.86 ± 17.58−7.670.001b
Tb.Sp (µm)267.18 ± 37.49282.27 ± 38.645.650.006b
SMI0.59 ± 0.230.75 ± 0.31−20.710.015c
DA1.30 ± 0.251.25 ± 0.23−4.040.14b
Table 2. Microstructural Parameters in the Two Groups and the Different Locations Studieda
 LocationControl GroupInstrumented GroupRelative Difference (%)p
  • BV/TV, bone volume density; BS/BV, specific bone surface; Conn.D, connectivity index; Tb.N, trabecular number; Tb.Th, trabecular thickness; Tb.Sp, trabecular separation; SMI, structure model index; DA, degree of anisotropy.

  • a

    Numerical data are given as the mean ± SD. The statistically significant differences are in italics.

  • b

    Non-parametric Mann–Whitney test.

BV/TV (%)Left lateral34.97 ± 5.8334.10 ± 5.52−2.470.54
 Central40.78 ± 8.3436.58 ± 8.33−10.290.04
 Right lateral34.93 ± 7.9731.49 ± 7.05−9.830.06
BS/BV (1/mm)Left lateral24.74 ± 4.0825.43 ± 3.952.760.14
 Central21.08 ± 3.6023.16 ± 3.249.900.10
 Right lateral24.32 ± 3.9725.56 ± 3.655.090.003
Conn.D (1/mm3)Left lateral107.94 ± 27.44125.59 ± 36.4416.340.01
 Central105.74 ± 22.66103.61 ± 29.14−2.010.93
 Right lateral110.55 ± 28.31106.18 ± 25.97−3.950.16
Tb.N (1/mm)Left lateral3.48 ± 0.323.49 ± 0.390.410.70
 Central3.87 ± 0.453.65 ± 0.53−5.690.04
 Right lateral3.55 ± 0.353.33 ± 0.32−6.31<0.0001
Tb.Th (µm)Left lateral97.57 ± 14.4190.38 ± 12.41−7.360.001
 Central111.64 ± 19.90101.48 ± 16.8−9.100.04
 Right lateral97.14 ± 14.6990.7 ± 12.15−6.650.003
Tb.Sp (µm)Left lateral280.77 ± 34.61279.23 ± 35.80−0.550.79
 Central248.36 ± 37.38273.55 ± 47.3410.140.02
 Right lateral271.39 ± 34.24293.13 ± 29.878.01<0.0001
SMILeft lateral0.59 ± 0.280.64 ± 0.21−6.460.47
 Central0.85 ± 0.350.62 ± 0.25−27.070.14b
 Right lateral0.73 ± 0.300.53 ± 0.23−26.710.02
DALeft lateral1.17 ± 0.101.15 ± 0.07−1.710.38b
 Central1.58 ± 0.281.45 ± 0.29−7.970.16
 Right lateral1.18 ± 0.091.16 ± 0.09−1.790.46
Table 3. Relative Difference and p-Value of Microstructural Parameters According to Location for the Control and the Instrumented Groupsa
  Central vs. Right LateralCentral vs. Left LateralLeft Lateral vs. Right Lateral
Δrel (%)pΔrel (%)pΔrel (%)p
  • Δrel, relative difference; NS, nonsignificant; BV/TV, bone volume density; BS/BV, specific bone surface; Conn.D, connectivity index; Tb.N, trabecular number; Tb.Th, trabecular thickness; Tb.Sp, trabecular separation; SMI, structure model index; DA, degree of anisotropy.

  • a

    The statistically significant differences are in italics.

BV/TV (%)Control16.70.00216.6<0.00010.1NS
 Instrumented16.20.0027.30.0098.30.03
BS/BV (1/m)Control−13.30.0002−14.8<0.00011.7NS
 Instrumented−9.40.0005−8.90.0002−0.5NS
Conn.D (1/mm3)Control−4.3NS−2.0NS−2.4NS
 Instrumented−2.4NS−17.50.0218.30.002
Tb.N (1/mm)Control8.90.00211.3<0.0001−2.2NS
 Instrumented9.70.0024.60.044.90.007
Tb.Th (µm)Control14.90.000214.4<0.00010.4NS
 Instrumented11.90.00212.3<0.0001−0.3NS
Tb.Sp (µm)Control−8.50.003−11.5<0.00013.5NS
 Instrumented−6.70.02−2.0NS−4.70.003
SMIControl16.5NS23.5NS−5.7NS
 Instrumented15.9NS−3.7NS20.3NS
DAControl34.3<0.000134.9<0.0001−0.5NS
 Instrumented25.8<0.000125.4<0.0001−0.4NS

Microstructural Parameters of the Control Group

Bone volume fraction (BV/TV), number of trabeculae (Tb.N), and trabecular thickness (Tb.Th) were significantly higher in the center than in the periphery (fixated side or non-fixated side) of the vertebral end-plates (p = 0.002, p = 0.002, and p = 0.0002, respectively). Mean intertrabecular space (Tb.Sp) was significantly lower in the center (p = 0.003), as was the specific bone surface (BS/BV) (p = 0.0002). The degree of anisotropy (DA) was significantly higher in the center than in the periphery (p < 0.0001). There was no significant difference between central and lateral locations for connectivity density (Conn.D) or the structure model index (SMI). No significant difference between the fixated and non-fixated sides was found for any of the parameters.

Microstructural Parameters of the Instrumented Group

Conn.D was significantly higher on the fixated side than in the center (p = 0.02), but was similar in the center and the non-fixated side. BV/TV, Tb.N, and Tb.Th were significantly higher in the center than in the periphery (fixated side or non-fixated side) (p = 0.009, p = 0.04, and p = 0.002, respectively). Tb.Sp was significantly lower in the center than in the fixated side (p = 0.02), but was not lower than in the non-fixated side. DA was significantly higher in the center than in the periphery (fixated side or non-fixated side) (p < 0.0001). There was no significant difference in SMI. BV/TV, Tb.N, and Conn.D were significantly higher in the fixated than in the non-fixated side (p = 0.03, p = 0.007, and p = 0.002, respectively), while Tb.Sp was significantly lower (p = 0.003). There was no significant difference between the fixated or non-fixated sides for BS/BV, Tb.Th, SMI, and DA.

Comparison of Microstructural Parameters between Control and Instrumented Groups

BV/TV, Tb.N, and Tb.Th were significantly lower in the instrumented levels than in controls (p = 0.009, p = 0.02, and p = 0.001, respectively). Inversely, Tb.Sp was significantly higher in the instrumented levels (p = 0.006). The differences between these parameters were greater in the central locations and in the non-fixated sides of the end-plates (Table 2). SMI was significantly lower in the instrumented levels (p = 0.015). There was no significant difference in Conn.D and DA, but in the instrumented levels, a significant increase of Conn.D was observed in the fixated side (Δrel = 16% and p = 0.01).

Microscopic Study

Whatever the biological level considered (cartilage end-plate, subchondral bone, or growth plate), and in both the instrumented group and the control group, we consistently found decreased thickness in the center compared to the periphery (fixated and/or non-fixated side). For all layers and locations examined, there was no significant difference of thickness between the samples from the fixated or non-fixated sides, or between the instrumented or control groups (Tables 4 and 5).

Table 4. Histomorphometric Dataa
 LocationControl GroupInstrumented GroupRelative Difference (%)p
  • a

    Numerical data are given as the mean ± SD. The statistically significant differences are in italics.

  • b

    Non-parametric Mann–Whitney test.

Cartilage end-plate (µm)Left lateral16.3 ± 7.823.3 ± 18.442.50.07b
 Central12.2 ± 813.2 ± 5.48.30.62b
 Right lateral17.1 ± 9.316.7 ± 10.4−2.40.88
Subchondral bone (µm)Left lateral147.8 ± 60.5144.3 ± 49.4−2.40.83
 Central77.8 ± 32.386.5 ± 3711.10.38
 Right lateral128.6 ± 57.7150.6 ± 48.317.10.14
Growth plate cartilage (µm)Left lateral34.5 ± 16.437.4 ± 11.78.40.48
 Central23.7 ± 9.126.9 ± 8.913.90.21
 Right lateral30.3 ± 11.538.1 ± 15.325.50.04b
Table 5. Relative Difference and p-Value of Histomorphometric Parameters According to Location for the Control and the Instrumented Groupsa
  Central vs. Right LateralCentral vs. Left LateralLeft Lateral vs. Right Lateral
Δrel (%)pΔrel (%)pΔrel (%)p
  • Δrel, relative difference; NS, nonsignificant.

  • a

    The statistically significant differences are in italics.

Articular cartilage (µm)Control−29.00.007−25.60.02−4.6NS
 Instrumented−21.20.03−43.40.00439.3NS
Subchondral bone (µm)Control−39.50.0002−47.4<0.000114.9NS
 Instrumented−42.6<0.0001−40.1<0.0001−4.2NS
Growth plate cartilage (µm)Control−22.00.002−31.40.00313.6NS
 Instrumented−29.20.0006−27.9<0.0001−1.9NS

DISCUSSION

To our knowledge, this is the first micro-CT study of the microarchitecture of the subchondral bone of the vertebral end-plate after application of asymmetric loading in a growing animal. Our animal model was the pig—a quadruped—which makes it difficult to transpose our results to humans16; but biped animal models such as monkeys are hardly available, and chickens, used as a scoliosis model, are not a suitable for surgical experimentation. Spinal growth speed in domestic pigs is high,17 from 120 mm/month between the 1st and 2nd months to 80 mm/month between the 3rd and 4th months). The spine (from T1 to L5) grew up from 300 mm (at 4 weeks of age) to 450–500 mm (at 4 months of age), which is close to the human adult spine length, as well as the overall size of vertebra and intervertebral disc.17 Moreover, the biological, histologic, and mechanical properties of the pig are acknowledged as being very close to those of human,18, 19 so we consider that this model represents a reasonable approximation of an adolescent.16, 20 We did not generate compression with the instrumentation, but the existence of local compressive forces exerted by the growth plates have been demonstrated on adolescents by Bylski-Austrow et al.21 As the distance between the pedicle screws was constant because of the rigidity of the instrumentation, we assumed that this fixation—carried out in an animal with great growth potential—acted as a tether, and the actual growth generated compressive force within the instrumented segment. As fixation was asymmetric, compression predominated on the fixated side. Gross observation showed decreased height of the two hemivertebrae between the pedicle screws (Fig. 1). However, other authors have shown that insertion of pedicle screws can lead to a decrease in vertebral height, pedicular length, and vertebral canal diameter due to involvement of the neurocentral synchondrosis.22, 23 In our study, microscopic examination showed that the growth plates of the vertebral end-plate, which account for the greatest part of the gain in height of the vertebral body, were not damaged by screw insertion, and that the thickness of the various biological layers did not significantly differ. We also observed narrowing of the disc located within the instrumentation, predominating on the fixated side (Fig. 1). We considered that this confirmed that asymmetric tether with compression forces had been achieved. The major limitation of our study is the absence of pressure monitoring over growth. Meir et al.24 have described a protocol to measure intradiscal pressure. Even if some reproducible results were obtained in human intervertebral disc, hydrostatic intradiscal pressure measurement is very sensitive to the position of the needle transducer within the disc. To carry out a reliable positioning of the transducer into the porcine disc, which presents important changes in size related to its growth, would have been dramatically difficult. On the other hand, Dennison et al.,25 who have developed an intervertebral disc-measuring protocol in a porcine model, acknowledged that the sensors' insertion could provoke disc tears, due to their size, that can potentially initiate or accelerate degenerative changes. Issever et al.26 showed that the mere presence of metallic material could induce significant changes of trabecular microarchitecture, similar to the reaction around prosthetic implants. In our study, careful avoidance of the growth plate of the vertebral end-plate constituted a biological barrier to confounding factors related to the presence of the material itself. In addition, in our study, the screws were situated away from the bone area examined. In this way, the vertebral end-plate (and all its three layers), which are the weak link of the spine in growing subjects,27 were entirely preserved. Because of the high inter-individual variations of the macroscopic and micro-CT parameters (Table 6), we decided to harvest both the control and the instrumented levels on the same animal. At 4 months of age, the macroscopic aspect of two thoracic or lumbar adjacent vertebrae were very similar in a same animal; and, to date, no publications, to our knowledge, have shown any significant difference between two adjacent vertebrae in terms of micro-CT parameters. We considered that a vertebral unit situated two levels below the corresponding instrumented units were an acceptable control. This hypothesis has been confirmed by the absence of difference between thoracic and lumbar vertebrae. Moreover, modifications of the intervertebral discs at levels adjacent to vertebral fusion, whatever the surgical technique used,4, 28–30 and modifications of their associated mechanical properties can affect the structure of the vertebral end-plates.8, 9 In order to reduce this effect, we chose our control levels well away from the instrumented levels. The numerical values obtained in this study are similar to those from Teo et al.31 in a porcine model, and from other authors in human models,14, 32–34 except for bone fraction volume (BV/TV) and the number of trabeculae (Tb.N) (Table 6). We examined trabecular bone immediately below the vertebral end-plate, where the bone network appears more dense, whereas the other authors examined the cancellous bone of the vertebral body, at a distance from the end-plates.

Table 6. Numerical Data of Micro-CT Studies of Vertebraea
Micro-CT ParametersHuman Lumbar Vertebrae (Simpson, 20018)Human Lumbar Vertebrae (Müller, 200234)Human Lumbar Vertebrae (Gong et al., 200632, 33)Human Thoracic Vertebrae (Sran et al., 200714)Porcine Lumbar Vertebrae (Teo et al., 200631)Mice Tail Vertebrae (Issever et al., 200326)Our Studyb
  • a

    Data are given as mean and standard deviation (SD) when this was available.

  • b

    Shows only the values found in our control group.

Resolution (µm) 142015141812
BV/TV (%)11.4 (2.0)8 (3)8.03 (1.50)13 (4.6)20 (6)22–3436.79 (7.88)
BS/BV (mm−1)24.5 (1.0)32.8 (7.0)35.4 (10.4)24.5 (4.2)
Conn.D (1/mm3)3.93 (1.44)108.18 (25.15)
Tb.N (mm−1)1.3 (0.3)1.30 (0.23)1.19 (0.19)0.770 (0.159)2.08 (0.44)3–5.83.63 (0.41)
Tb.Th (µm)84.4 (3.8)60 (20)110 (10)169 (32)100 (10)60–7594 (15)
Tb.Sp (µm)710 (177)650 (160)1,010 (135)350 (60)180–320267 (37.5)
DA1.21 (0.06)1.43 (0.14)1.37 (0.07)1.5–21.30 (0.25)
SMI1.72 (0.41)1.53 (0.27)0.67 (0.40)0.4–1.40.59 (0.23)

In the control vertebral units, we found a denser trabecular network (increased BV/TV, Tb.N, and Tb.Th, with decreased Tb.Sp) with greater spatial disorganization (higher DA) in the center of the vertebral end-plates compared with the periphery. Gong et al. studied the influence of regional variations in microstructural parameters in human models,32, 33 and also found higher DA in the center. One hypothesis is that most of the loading transfer occurs in periphery through the cortical bone. This could characterize the mechanical vulnerability of the center of the vertebral end-plate, as the orientation of the trabeculae is not as well adapted to axial loading. In the instrumented units, in the center and on the opposite side to compression, there was a decrease in BV/TV, Tb.N, Tb.Th, and an increase in Tb.Sp, reflecting rarefaction of bone tissue. On the fixated side, Tb.Th was decreased and Conn.D was increased, while the other parameters were not significantly modified, reflecting overall preservation of bone microarchitecture. This local trabecular reorganization is due to two opposing processes according to Wolff's law. The first process is osteolytic. It is due to the rigidity of the spinal instrumentation. Fixation results in stress shielding which bears a major part of the mechanical stresses. Bone is no longer subjected to physiological demands, resulting in bone loss as can be seen in osteoporosis.35–37 The second process is osteogenic. On the fixated side, growth of the vertebral segment generates compressive forces. The bone is mechanically stimulated, compensating for bone loss (no decrease in BV/TV, Tb.N, Tb.Th, and no increase in Tb.Sp). The significant increase of Conn.D on the fixated side is due to these two phenomena. Lysis produces perforations of the trabeculae, artificially increasing the number of trabecular connections.35 Local compression stimulates trabecular connectivity26 to increase the strength of bone tissue. The structure model index and the degree of anisotropy (DA) reflect the three-dimensional organization of bone material. We found no change in these parameters, probably due to the intrication of the two processes which act in opposite directions and counterbalance each other, as well as the high deviation standard of the parameters.

Despite its limitations and the absence of intradiscal pressure monitoring, our study revealed that the trabecular bone tissue of the vertebral end-plates becomes more dense after application of compressive loading. Robert et al.38 demonstrated, in scoliosis, intervertebral disc's alterations, similar to disc senescence and degeneration,39 associated with calcification of the cartilage end-plate. They related this calcification to the altered loading in scoliosis. The densification of the vertebral end-plate could induce occlusion of marrow bone contacts, which has been associated with disc degeneration.40 Gruber et al.10 have found a significant correlation between cell death in the annulus and bone mineral density of the vertebral end-plate. Our study was designed to analyze only the vertebral end-plate changes under compression, and we did not demonstrate any disc degeneration, but we hypothesized that our animal model could be useful to link the micro-architectural changes due to mechanical stress with possible disc degeneration. Our model also illustrates the potential risks of spinal instrumentation, which, if too rigid, can induce premature osteopenia,26 with possible instrumentation loosening due to bone anchorage. Such instrumentations, which are difficult procedures in osteoporotic individuals,41 may therefore require augmentation techniques42 to prevent this risk. Our study protocol should enable better identification of microstructural changes in the vertebral end-plates, adjacent to the fusion where disc degeneration30 occurs, related to altered mechanical stresses.8, 9, 28, 43 It will also be useful in evaluating devices for dynamic instrumentation without fusion in an attempt to preserve bone microarchitecture, and to better characterize the “protective” role of this material.

Acknowledgements

The authors thank Dr. A. Gomez-Brouchet and Professor M. B. Delisle of the Laboratory of Anatomy and Cytopathology, Toulouse-Rangueil University Hospital, for help in expert histologic analysis and interpretation.

Ancillary