Surgical Procedure
This study was approved by the Institutional Animal Care and Use Committee. Eight two-month-old neutered male domestic pigs underwent general anesthesia with 4 mg/kg of Telazol (tiletamine and zolazepam) administered intramuscularly, which was maintained with inhalational isoflurane (2% to 3%). All animals received antibiotics preoperatively (35 mg/kg of cefazolin intravenously) and postoperatively (3 mg/kg of ceftiofur intramuscularly twice a day for three days).
The animals were placed in the prone position and were prepared and draped in a sterile fashion. The T7 to T14 vertebrae were identified with use of fluoroscopy and marked on the skin, and the paraspinal muscles were dissected off the lamina superficial to the periosteum. However, the facet joints were left intact. The operation then proceeded according to whether the pig had been randomly assigned to the control, single-screw, or double-screw group.
Control group (two animals): Following soft-tissue dissection of the paraspinal muscles, the incision was left open for sixty minutes—the estimated time necessary to place pedicle screws in the screw-treated groups.
Single-screw group (three animals): The animals underwent single-pedicle-screw fixation of the right pedicles from T7 to T14. The starting point for the pedicle screws was identified with use of anatomic landmarks on the posterior elements of the spine and was confirmed with fluoroscopy. The anatomic landmark for the superior-inferior point was the midportion of the transverse process, and the medial-lateral point was just lateral to the superior facet joint. A pedicle finder, directed 30° to 35° in the axial or frontal plane, was then used to advance down the pedicle. Intraosseous pedicle penetration was confirmed with use of a ball-tipped probe to palpate five pedicle walls and to determine pedicle depth. The pedicle was then tapped with a 2.5-mm-diameter tap, after which a 3.5-mm cortical screw (Synthes, Paoli, Pennsylvania) of appropriate length (usually 22 to 28 mm) was placed.
Double-screw group (three animals): The treatment in this group was the same as that in the single-screw group except that two screws were placed in each pedicle and oriented in the sagittal plane. In the double-screw group, the starting point for the first screw was on the superior portion of the pedicle and the starting point for the second screw was just inferior to the first. The orientation of the two screws was parallel.
Following screw placement, the soft tissues were closed in layers with use of absorbable suture. Anteroposterior and lateral radiographs of the entire spine were made immediately following the surgery in all animals.
Postoperative Period and Specimen Preparation
Analgesics (1.5 mg/kg of flunixin meglumine) were administered intramuscularly twice a day for three days postoperatively. No postoperative immobilization was utilized. The animals were radiographed to assess the curve in the coronal and sagittal planes at two-month intervals. On the anteroposterior radiographs, the vertebrae with the greatest amount of tilt were selected as the cephalad and caudad end vertebrae. Lines were drawn perpendicular to the end plate of the vertebrae. The angle subtended by these lines was the coronal Cobb angle. On the lateral radiographs, T7 and T14 were selected as the cephalad and caudad end vertebrae, respectively, and lines were drawn perpendicular to the end plate of T7 and T14. The angle formed at the intersection of these lines was the sagittal Cobb angle. All animals were killed at six months, and the entire spine (T1 to L5), including the ribs, was harvested.
Axial computed tomography scans were performed with use of a high-speed computed tomography scanner (Philips Medical Systems, Bothell, Washington) with a 14-cm field of view, a scanning power of 200 mA, 120 kV, and a 1.4-mm slice thickness. The inclination of the vertebrae was adjusted to ensure that all transverse images were parallel to the superior margin of the vertebra. The axial rotation of each vertebra was measured with the Synapse analysis system (Fujifilm Medical Systems USA, Stamford, Connecticut) (see Appendix).
Assessment of fusion over the operatively treated levels (T7 through T14) was performed by visual inspection, manual palpation, and evaluation of the radiographs and computed tomography images by two independent observers. The spines were then separated into individual vertebrae, and the T7 to T14 vertebrae were used for subsequent analysis.
The height of the vertebra, on the left and right, in the midcoronal plane and the anterior and posterior heights of each vertebral body in the midsagittal plane were measured with use of manually operated digital calipers (Absolute Digimatic System; Mitutoyo, Tokyo, Japan). The measurements were used to determine whether the vertebra had structural wedge changes during the six-month period.
Each vertebra was cut in the transverse plane into four 4-mm-thick segments. These consisted of superior, intermediate-1, intermediate-2, and inferior sections, in which the intermediate-1 section was through both pedicles. All transverse sections were parallel to the superior margin of the vertebra to ensure true axial sectioning. The specimens were fixed in 70% ethanol, undecalcified, and embedded in methylmethacrylate. Each specimen was cut into slices, which were ground and polished with use of the EXAKT MicroGrinding System (EXAKT, Norderstedt, Germany) to a final thickness of 30 to 40 µm. The slices were stained with a rapid bone stain (Surgipath Medical Industries, Richmond, Illinois). Histomorphometric analysis was performed with use of a digitizing image analysis system coupled to a light microscope (KS300; Carl Zeiss Vision, Hallbergmoos, Germany) at 4× and 20× magnification to evaluate the following parameters.
Screw orientation: Each pedicle screw position was evaluated to determine whether (1) the screw crossed the neurocentral synchondrosis and (2) the screw penetrated the medial or lateral wall of the pedicle.
Spinal canal area: This was measured in the mid-transverse (intermediate-1) section of each vertebra.
Pedicle length and width: These were measured in the mid-transverse (intermediate-1) section of each vertebra (see Appendix).
Vertebral area and trabecular bone volume: These were measured in the mid-transverse (intermediate-1) section of each vertebra. A line was drawn through the mid-sagittal plane of the vertebra, dividing the vertebra into left and right portions, and the vertebral area and the trabecular bone volume of each portion were measured.
Histological grading of the neurocentral synchondrosis: This was performed at each section level of each vertebra with use of a custom 6-point (0 to 5-point) histological scale, with 0 indicating no closure of the neurocentral synchondrosis, 1 indicating <25% closure, 2 indicating 25% to 49% closure, 3 indicating 50% to 74% closure, 4 indicating 75% to 99% closure, and 5 indicating 100% closure (see Appendix).
Statistical Analysis
Paired t tests were used to determine differences between the left and right sides with respect to the pedicle morphology, vertebral area, and trabecular bone volume in each group. The parameters included (1) coronal and sagittal Cobb angles, (2) thoracic spine length, (3) vertebral body rotation, (4) vertebral height, (5) spinal canal area, and (6) histological grade of the neurocentral synchondrosis. These parameters were compared among the three groups with use of one-way analysis of variance and Tukey multiple comparison methods. Pearson correlation and simple linear regression analyses were used to characterize the relationship between closure of the neurocentral synchondrosis scale and the magnitude of the coronal curve. Significance was defined as p < 0.05.
All animals survived without neurological complications and remained normal and healthy for six months. At the time that the animals were killed, a scoliotic curve was not seen in either animal in the control group. One of the three animals in the single-screw group had scoliosis, measuring 47°, and all three animals in the double-screw group had scoliosis, measuring 30°, 42°, and 42° (Table I, Fig. 1). All of the curves were located at the operatively treated levels with the convexity toward the side of the pedicle screw fixation. In the animals with a scoliotic curve, the mean increase in the Cobb angle from zero to two months was 31.3° (95% confidence interval, 20.3° to 42.3°), which was significantly greater than the increase seen from two to four months (mean, 4.8°; 95% confidence interval, -0.9° to 10.5°) or from four to six months (mean, 4.3°; 95% confidence interval, 0.2° to 8.4°) (p < 0.0001). The average increase in the Cobb angle did not differ significantly between the two to four-month period and the four to six-month period. The lateral radiographs demonstrated an average increase in kyphosis of 12° (95% confidence interval, -5.6° to 29.6°) in the control group, 18.7° (95% confidence interval, -6.2° to 43.6°) in the single-screw group, and 26.7° (95% confidence interval, 21.6° to 31.8°) in the double-screw group (Table II). With these small numbers, these differences were not found to be significant.
At the time that the animals were killed, at six months, the mean increase in the length of the thoracic spine (from the superior margin of T1 to the inferior margin of T15) was 12.2 cm (95% confidence interval, 11.3 to 13.1 cm) in the control group, 16.0 cm (95% confidence interval, -9.2 to 41.2 cm) in the single-screw group, and 11.5 cm (95% confidence interval, 10.5 to 12.5 cm) in the double-screw group.
Vertebral rotation in the axial plane occurred toward the screw side and was significantly greater in the double-screw group (mean, 15.2°; 95% confidence interval, 11.7° to 18.7°) than in the single-screw group (mean, 6.1°; 95% confidence interval, 1.6° to 10.6°) or the control group (mean, 0°; 95% confidence interval, 0° to 0°) (p < 0.001).
No evidence of spine fusion over the operatively treated levels (T7 to T14) was detected by direct visualization. The manual palpation test demonstrated that spine mobility at the operatively treated levels was the same among the three groups. Furthermore, there did not appear to be fusions on radiographs or on computed tomography images (Fig. 2).
The average height of the thoracic vertebrae from T1 to T15 was 25.6 mm (95% confidence interval, 25.3 to 25.9 mm) in the control group, 28.9 mm (95% confidence interval, 28.7 to 29.1 mm) in the single-screw group, and 26.0 mm (95% confidence interval, 25.8 to 26.2 mm) in the double-screw group. Comparison of the heights of the left and right parts of the vertebra and comparison of the anterior and posterior heights of the vertebra revealed no significant difference at any level except for T9 in the double-screw group. At T9 in that group, the height of the right (convex, screw-side) part of the vertebra (mean, 27.3 mm; 95% confidence interval, 26.3 to 28.3 mm) was 18% greater than the height of the left part (mean, 22.5 mm; 95% confidence interval, 21.9 to 23.1 mm) (p = 0.01).
With regard to pedicle screw orientation, all screws crossed the neurocentral synchondrosis in the treated animals. In the single-screw group, 25% (six) of the twenty-four screws penetrated the medial wall of the pedicle; this percentage was significantly greater than the percentage of screws (2%, one of forty-eight) that penetrated the medial wall of the pedicle in the double-screw group (p = 0.005).
The mean spinal canal area, as measured in the intermediate-1 section of each vertebra, was 145.4 mm2 (95% confidence interval, 141.6 to 149.2 mm2) in the control group, 146.9 mm2 (95% confidence interval, 141.2 to 152.7 mm2) in the single-screw group, and 144.5 mm2 (95% confidence interval, 141.7 to 147.4 mm2) in the double-screw group. There was no significant difference among the three groups.
The values regarding the pedicle morphology in the three groups are shown in Table III. There were no significant differences in the pedicle morphology between the left (untreated) and right (sham-operation) sides in the control group. In the single-screw group (Fig. 3), the length of the left pedicle was 9% greater (p = 0.008) and the anteroposterior length of the left pedicle was 10% greater than the values for the right (screw-side) pedicle (p < 0.001); no significant difference was found between the widths of the left and right pedicles. In the double-screw group (Fig. 4), the length of the left pedicle was 22% greater and the anteroposterior length of the left pedicle was 20% greater than the values for the right pedicle (p < 0.001), and the width of the right pedicle was 27% greater than the width of the left pedicle (p < 0.001).
The left and right vertebral areas and trabecular bone volumes are shown in Table IV. The left (untreated) vertebral area was 6% greater than the right vertebral area in the double-screw group (p = 0.01), 5% greater in the single-screw group (p = 0.01), and not significantly different in the control group. No significant difference was found between the trabecular bone volumes on the left and right sides in any group.
The histological evaluation demonstrated that each neurocentral synchondrosis was completely open and undisturbed (grade 0) in the control group. In the single-screw group, the neurocentral synchondrosis was intact and completely open (grade 0) on the left, while there was, on the average, <50% closure of the neurocentral synchondrosis on the screw side (mean grade, 2.9; 95% confidence interval, 1.5 to 3.9). In the double-screw group, on the average, >75% of the neurocentral synchondrosis was closed on the screw side (mean grade, 4.4; 95% confidence interval, 3.8 to 4.7). The neurocentral synchondrosis closure in the double-screw group was significantly greater than the closure in the single-screw group (p < 0.001), and the closure in both groups was greater than that in the control group (p < 0.001).
Table V shows the histological grades for the right (screw-side) neurocentral synchondrosis at each section level (from superior to inferior) of the vertebra in the three groups. In the single-screw group, there was, on the average, >75% closure of the neurocentral synchondrosis at the two intermediate levels, <50% closure at the superior level, and <25% closure at the inferior level. The percentage of neurocentral synchondrosis closure at the two intermediate levels was significantly greater than that at the superior and inferior levels (p < 0.0001). In the double-screw group, there was, on the average, approximately 100% closure of the neurocentral synchondrosis at the two intermediate levels, >75% closure at the superior level, and >50% closure at the inferior level. There was no significant difference in the percentage of neurocentral synchondrosis closure between the superior level and the two intermediate levels, whereas the percentages of closure at those levels were significantly greater than that at the inferior level (p < 0.0001). The double-screw group had significantly greater closure of the neurocentral synchondrosis than the single-screw group at all levels except for the intermediate-1 level, and both groups had significantly greater closure than the control group (p < 0.001).
We identified a significant correlation between the amount of neurocentral synchondrosis closure and the magnitude of the coronal curve (Pearson r = 0.92; 95% confidence interval, 0.62 to 1.00; p = 0.001) (Fig. 5)—that is, as the closure of the neurocentral synchondrosis increased, the curve magnitude toward the side in which the screw had been placed became larger.
The neurocentral synchondrosis is a three-dimensional cartilaginous junction on both sides of the vertebra located between the vertebral body and the posterior arch of the vertebra. Unlike the epiphyseal plate of a long bone, this growth plate is bipolar and therefore contributes to the growth of both the vertebral body and the posterior arch1-5. It can be assumed that symmetric growth of the right and left parts of the neurocentral synchondrosis results in spinal growth without deformity and that asymmetric growth can lead to deformation of the vertebral bodies and ultimately to spinal deformity1,6-8. However, authors of previous animal studies have been unsuccessful in creating spinal deformities with asymmetric closure of the neurocentral synchondrosis7-9; as a consequence, there were no further attempts to create scoliosis in this fashion. It is important to remember that the authors of those studies probably did not confirm that the disturbance or closure of the neurocentral synchondrosis was adequate on the intended side. The present study demonstrates that idiopathic-like scoliosis develops secondary to asymmetric closure of the neurocentral synchondrosis, with lateral curvature and axial plane rotation toward the side of the closure. Closure of the neurocentral synchondrosis was confirmed with use of a histological grading system, and the amount of closure correlated with the degree of lateral curvature. In the previous studies, the investigators may have created some disturbance of the neurocentral synchondrosis but not enough asymmetric closure to produce the scoliotic deformity; the situation may be similar to that in the single-screw group in our study, in which scoliosis developed in only one of three animals.
We demonstrated a direct correlation between greater closure of the neurocentral synchondrosis and greater scoliotic deformity. In the double-screw group, the two pedicle screws oriented in the sagittal plane induced nearly 100% neurocentral synchondrosis closure in the central portion of the vertebra as well as >75% neurocentral synchondrosis closure in the superior portion of the vertebra and >50% closure in the inferior portion of the vertebra. In the single-screw group, although the single pedicle screw provided >75% neurocentral synchondrosis closure in the central portion of the vertebra, there was significantly less closure in the superior (<50%) and inferior (<25%) portions of the vertebra. We believe that, because the unilateral double pedicle screws achieved complete closure of the neurocentral synchondrosis, they consistently produced a structural curve, whereas when the inferior and superior parts of the neurocentral synchondrosis were left open, as they were in the single-screw group, not enough asymmetric growth was provided to produce scoliosis.
Our histomorphometric analyses showed that the pedicles and vertebral body were symmetric without spinal deformity in the control group. In the double-screw group, the length of the pedicle, the anteroposterior length of the pedicle, and the vertebral area on the untreated side were 22%, 20%, and 6% greater than the respective values on the screw side. The pedicle width on the screw side was 27% greater than the width on the untreated side. In contrast, in the single-screw group, the length of the pedicle, the anteroposterior length of the pedicle, and the vertebral area on the untreated side were 9%, 10%, and 5% greater than the respective values on the screw side. There was no difference in the pedicle widths of the two sides. These results further demonstrate that unilateral pedicle screws crossing the neurocentral synchondrosis arrest growth activity on the screw side while the relatively excess growth of the untreated side induces vertebral rotation toward the screw side and creates a scoliotic curve.
The spinal deformity seen in this study included axial plane rotation averaging 15° toward the side of the screws, which is an essential characteristic of scoliotic deformity. Wedging of the vertebral body could be observed at the apical vertebra in a scoliotic pig (Case 3) in the single-screw group. In the double-screw group, wedging of the vertebra was found only at the T9 level (the apex), at which the height of the right (convex) side of the vertebra was 18% greater than the height of the left side. There was no significant difference in the average vertebral height among the three groups. These results suggest that the neurocentral synchondrosis is probably involved in the vertebral deformities in the axial plane rather than in the longitudinal plane. Fusion of the neurocentral synchondrosis may result in the asymmetric vertebral growth in the transverse plane that induces the vertebral axial rotation. The wedging of the vertebral body is most likely secondary to the spinal scoliotic deformity, which is similar to the clinical situation.
A previous animal study9 demonstrated that pedicle screws passing through the neurocentral synchondrosis in an immature pig disturb the growth of the spinal canal and create spinal stenosis. Our histomorphometric analysis does not confirm these findings as it demonstrated no difference in the area of the spinal canal between the control and treated groups. No treated animal in this study showed spinal stenosis. Other investigators have confirmed this lack of spinal stenosis in very young children in whom congenital scoliosis was treated with hemivertebra resection and bilateral pedicle screw stabilization10.
The contribution of the neurocentral synchondrosis to human scoliosis was reported in 1909 by Nicoladoni, who considered the vertebral torsion to be due to premature cessation of growth in the neurocentral synchondrosis on the convex side of the curve1. Since that time, authors of other studies1-9,11-17 have concentrated primarily on determining whether asymmetric growth of the neurocentral synchondrosis could be a cause of adolescent idiopathic scoliosis. Assessment of the role of the neurocentral synchondrosis in adolescent idiopathic scoliosis has included some debate11-15 on the age at which it closes because it must be shown that the neurocentral synchondrosis is open and active during the adolescent growth period when the curve is developing. To date, the timing of neurocentral synchondrosis closure is still controversial, with most authors believing that it occurs prior to the age of ten years and refuting the role of the neurocentral synchondrosis in adolescent idiopathic scoliosis. Although we did not specifically perform this study as a model for adolescent idiopathic scoliosis, it can be assumed that the strategy of modulating the activity of the neurocentral synchondrosis would not be useful for adolescent patients since the neurocentral synchondrosis appears to be closed after the age of ten years. Instead, modulation of the growth of the neurocentral synchondrosis may be utilized in the treatment of very young patients if it can be proven that the neurocentral synchondrosis is actively growing. Previous investigations have demonstrated that the neurocentral synchondrosis is visualized on magnetic resonance images and the average width and thickness were 8.0 and 1.2 mm on the left and 7.9 and 1.3 mm on the right, which were not significantly different18.
Our results revealed no difference in the average increase in the length of the thoracic spine among the three groups at six months, and no fusion was found at the operatively treated levels. These findings suggest that this surgical procedure does not arrest longitudinal growth of the spine and therefore could be a fusionless technique for treatment of scoliosis. At present, surgical treatment of scoliosis relies on fusion of the spine—that is, creating a solid arthrodesis of previously mobile spinal motion segments, resulting in loss of motion. Although this is effective in decreasing spinal deformity and preventing curve progression, it "steals" spinal motion and has detrimental effects on the long-term health of the spine. Methods to treat scoliosis without fusion would improve the overall functional outcome of these patients. If one can create scoliosis in a straight normal spine, it may be possible to treat a scoliotic spine similarly to prevent further curve progression and perhaps decrease the deformity with continued growth. Asymmetric growth inhibition of the neurocentral synchondrosis with the use of pedicle screw fixation is a method for creating scoliosis that does not result in fusion of motion segments, and it appears to have potential for treatment of scoliosis in young patients.