Ninety-eight North American volunteers (sixty-three men and thirty-five women) were enrolled in this study. The men had a mean age, height, and weight (and standard deviation) of 24.6 ± 5.7 years, 180.6 ± 7.5 cm, and 83.2 ± 15.1 kg, respectively. The women had a mean age, height, and weight of 25.3 ± 6.1 years, 164.3 ± 7.1 cm, and 59.2 ± 7.1 kg, respectively. Subjects with a history of cervical spine abnormalities; symptoms suggestive of degenerative or inflammatory disease; evidence of infectious, neoplastic, or traumatic conditions; or congenital malformations of the cervical spine were excluded. Informed consent was obtained from all participants, and the study was approved by the institutional review board.
Computerized tomography scans of the cervical spine (Somatom Plus 4; Siemens, Erlangen, Germany) were obtained for all volunteers. A preliminary lateral scout scan was performed, followed by 3.0-mm helical scans of the cervical spine and reconstructions at 1.5-mm intervals. The scans were obtained with the volunteers in the supine position and with the neck in the neutral position. Computerized tomography images were analyzed with use of standardized bone windows with ImageJ, a public-domain image analysis and processing program developed by the National Institutes of Health24. Morphologic characteristics were obtained from the third cervical vertebra (C3) to the seventh cervical vertebra (C7).
Axial images through the site of maximal pedicle width were utilized to measure the metrics shown in Figure 1. Pedicle length was defined as the distance between the pedicle-lateral mass junction and the pedicle-vertebral body junction. Pedicle axis length was defined as the length of the pedicle axis from its projection on the posterior surface of the lateral mass to the anterior vertebral body surface. Pedicle transverse angulation was defined as the angle between the pedicle axis and the vertebral body midsagittal axis. Medial offset of the pedicle axis was defined as the mediolateral distance between the pedicle axis projection on the dorsal lateral mass surface and the lateral border of the cranial vertebral inferior articular process.
Axial images reformatted perpendicular to the pedicle axis through the pedicle isthmus were used to measure the metrics shown in Figure 2. Pedicle width was defined as the mediolateral outer diameter of the pedicle at its isthmus. Pedicle height was defined as the superoinferior outer diameter of the pedicle at its isthmus.
Oblique images reformatted through the horizontal midpoint of the pedicle were used to measure the pedicle sagittal offset and the pedicle sagittal angulation (Figs. 3-A and 3-B). Pedicle sagittal offset was defined as the superoinferior distance between the pedicle axis projection on the dorsal lateral mass surface and the inferior border of the superior articular facet (Fig. 3-A). This value was recorded as negative if the pedicle axis projection was cranial to the inferior border of the superior articular facet. Pedicle sagittal angulation was defined as the angle between the pedicle axis and a line parallel to the inferior end plate (Fig. 3-B). This value was recorded as positive if the pedicle axis angulated cranially and negative if it angulated caudally.
Statistical Analysis
The mean and the standard deviation were computed for all eight parameters. Three-factor analysis of variance was used to determine significant differences (p < 0.05) in pedicle geometry on the basis of sex (male or female), side (right or left), and spinal level (C3 to C7). Bonferroni-Dunn post-hoc analysis was used to determine significant differences between specific factor levels. The critical p value was adjusted to <0.005 for comparisons between spinal levels. Independent analyses were performed for each geometrical parameter.
Intraobserver and interobserver repeatability was assessed by quantifying the coefficient of repeatability25. For intraobserver repeatability, all eight measurements were repeated for ten randomly selected volunteers by the primary observer (S.V.M.). For interobserver repeatability, all eight measurements were repeated for fifteen randomly selected volunteers by an independent secondary observer. Primary and secondary observers were qualified orthopaedic surgeons with experience in spine surgery. Measurement value differences were calculated as the second measurement by the primary observer (intraobserver) or the measurement by the secondary observer (interobserver) subtracted from the first measurement by the primary observer. The mean and the standard deviation of these differences were calculated for each geometrical measurement across all five spinal levels and on right and left sides. The coefficient of repeatability for intraobserver and interobserver assessments was computed as the standard deviation multiplied by a factor of 2.0.
The computerized tomography scans of 490 subaxial cervical vertebrae from the ninety-eight volunteers were analyzed. Six linear and two angular parameters were measured for 980 pedicles. There were no significant differences (p > 0.05) between the right and left sides for any parameter. Significant sex-related differences (p < 0.05) were observed in association with seven parameters, with the pedicle transverse angulation as the sole exception (Tables I and II). Significant level-related differences (p < 0.05) were observed in association with all parameters.
Pedicle width (range, 2.1 to 9.9 mm) was significantly dependent (p < 0.05) on sex and spinal level. The mean width increased from 5.4 mm at C3 to 7.2 mm at C7. Post-hoc analysis revealed that pedicle width at each level was significantly different from that at all other levels (p < 0.005), with the exception of C3 relative to C4. The mean pedicle width was greater in men than in women at all levels. The width was <4.0 mm in seventeen pedicles in nine volunteers (seven women and two men), including seven pedicles at C3, six at C4, and four at C5. A pedicle width of <4.0 mm was bilateral in four vertebrae and unilateral in nine vertebrae.
Pedicle height (range, 3.5 to 9.3 mm) was significantly dependent (p < 0.05) on sex and spinal level. The mean height increased from 6.3 mm at C3 to 6.7 mm at C7. Post-hoc analysis revealed that the C7 pedicle height was significantly greater (p < 0.005) than the pedicle height at all other levels. The mean pedicle height was greater in men than in women at all levels. Pedicle height was <4.0 mm in six pedicles (all in women), including two pedicles at C3, one at C4, one at C5, and two at C6; these six pedicles were from three volunteers, with a single volunteer accounting for four unilateral small pedicle height measurements. The mean pedicle height was greater than the mean pedicle width from C3 to C5 and was approximately equal to the width at C6. The mean pedicle width (7.2 mm) was marginally greater than the mean pedicle height (6.7 mm) at C7. The mean pedicle width and height at all subaxial levels were sufficient to accommodate 3.5-mm screws in 98% of the volunteers. Pedicle width and height dimensions of <4.0 mm were rare and occurred in only 1.7% of all pedicles. Of those pedicles, 82% were in women and 72% were unilateral.
Pedicle length (range, 3.4 to 7.8 mm) was significantly dependent (p < 0.05) on sex and spinal level. The mean pedicle length was shortest at C4 (5.3 mm) and longest at C6 (5.8 mm). Post-hoc analysis revealed that the pedicle length was significantly smaller at C3 and C4 than it was at C5, C6, and C7 (p < 0.005) and that the length at C5 was significantly smaller than it was at C6 (p < 0.005). The mean pedicle length was greater in men than in women at all levels.
Pedicle axis length (range, 23.0 to 45.1 mm) was significantly dependent (p < 0.05) on sex and spinal level. The mean axis length was longest at C3 (33.1 mm) and shortest at C7 (31.3 mm). Post-hoc analysis revealed that the axis length was significantly shorter at C7 than at all other levels (p < 0.005). The mean pedicle axis length was greater in men than in women at all levels.
Pedicle transverse angulation (range, 21.8° to 59.3°) was significantly dependent (p < 0.05) on spinal level. The mean pedicle transverse angulation was smallest at C7 (33.5°) and largest at C4 (47.8°). Post-hoc analysis revealed that the pedicle transverse angulation values at each level were significantly different from those at all other levels (p < 0.005), with the exception of C3 relative to C4 and C5. The pedicle transverse angulation was >45° for 72.5% of the pedicles at C3, 78.1% of the pedicles at C4, 66.3% of the pedicles at C5, 24.0% of the pedicles at C6, and 3.1% of the pedicles at C7.
Pedicle sagittal angulation (range, -14.2° to 26.2°) was significantly dependent (p < 0.05) on sex and spinal level. Pedicles were directed cranially at C3 (13.7°), with gradually increasing caudal angulation inferiorly to C7 (-2.8°). Post-hoc analysis revealed that pedicle sagittal angulation values at each level were significantly different from those at all other levels (p < 0.005), with the exception of C6 relative to C7. The pedicle sagittal angulation was directed more caudally in women than in men at C3, C5, C6, and C7.
Medial offset (range, 0.0 to 10.5 mm) was significantly dependent (p < 0.05) on sex and spinal level. The mean offset was 1.5 mm at C3 and increased to 5.3 mm at C7 (Fig. 4). Post-hoc analysis revealed that the medial offset at each level was significantly different from those at all other levels (p < 0.005), with the exception of C4 relative to C5. The mean medial offset was lower in women than in men at all levels.
Sagittal offset (range, -1.6 to 8.9 mm) was significantly dependent (p < 0.05) on sex and spinal level. The mean sagittal offset decreased from 4.2 mm at C3 to 1.6 mm at C6 (Fig. 4). Post-hoc analysis revealed that the sagittal offset at each level was significantly different from those at all other levels (p < 0.005), with the exceptions of C5 and C6 relative to C7. Sagittal offset was lower in women than in men at C3, C5, and C7. The pedicle axis projected inside the facet joint cranial to the inferior border of the superior facet in 0.5% of the pedicles.
The intraobserver coefficient of repeatability was less than the interobserver coefficient of repeatability for all measurements (Table III), indicating better agreement between the first and second measurements made by the primary observer. A high level of repeatability was demonstrated for the measurements of pedicle height, width, length, axis length, and transverse angle, with coefficient-of-repeatability values between 4.6% and 10.9% of mean dimensional values for intraobserver repeatability and between 8.6% and 16.4% of mean dimensional values for interobserver repeatability.
Cervical pedicle screw use in the subaxial spine is largely confined to the C7 vertebral level, where the lateral mass is less well developed and pedicle dimensions are larger. The reason for the infrequent use of this technique in the United States is multifactorial and likely is related to the neurovascular risk, the medicolegal environment, training, and a more complex three-dimensional anatomy compared with the trapezoidal anatomy of the lateral mass. The present study provides normative data on pedicle geometry from a large sample of young adult North American volunteers. Our data demonstrated that pedicle morphology is, and therefore screw placement and orientation should be, significantly different (p < 0.05) at each subaxial cervical level and between men and women. It is necessary to account for these differences during surgical planning in order to reduce the risks of pedicle perforation and neurovascular injury. The mean width and height at all subaxial levels were sufficiently large to accommodate standard 3.5-mm pedicle screws in 98% of the young, healthy volunteers enrolled in the present study. Pedicle dimensions of <4.0 mm were rare exceptions (noted in association with 1.7% of the pedicles) and generally were observed in women. Width and height measurements increased caudally.
Pedicle morphometry has previously been evaluated in cadaver spines or patients who underwent surgical intervention with use of physical measurement devices17-19,21 or medical imaging modalities2,6,7,22,23. Sample populations included older specimens or patients6,8,13,18-22, limited sample size2,18,19, or unidentified age and sex2,17,19,22. Despite these differences in measurement technique and study population, our results are consistent with previous data. The mean C3 to C7 pedicle height measurements that we determined were 8.2% to 16.3% less than those in previous studies2,7,17-19,21, and the mean pedicle width measurements were 5.1% to 20.0% greater than those in previous studies2,6,7,17-19,21-23. The mean axis length that we determined was similar to previously reported dimensions7,23. Our findings are also consistent with previous reports in the literature with regard to the dependence of pedicle morphology on level19 and sex18,21. In addition, our results demonstrated significant sex differences (p < 0.05) for seven of eight measurements and significant level differences (p < 0.05) for all eight measurements.
Angular measurements of the pedicle axis in the transverse and sagittal planes provide a quantitative description of the direction of pedicle screw insertion. In a previous study, Abumi et al. recommended that the transverse angulation should be medially inclined from 25° to 45°26. However, in a more recent study, Sakamoto et al. recommended screw insertion angles of approximately 50° from C3 to C6 in order to orient the screw coaxial with the pedicle axis and to reduce the risk of vertebral artery injury20. We found that transverse and sagittal plane angulations were significantly dependent on spinal level. Transverse angulation was approximately 45° at C3 through C5 and decreased caudally to approximately 33° at C7 for both sexes. Sagittal angulation demonstrated cranial orientation at C3 and C4, approximately horizontal orientation at C5, and caudal orientation at C6 and C7, findings that are similar to those of previous investigations6,7,19.
We identified larger pedicle sizes in men for all six linear dimensions and different angular measurements between men and women. The mean pedicle width and height were 18.6% and 18.3% greater in men than in women. This finding is consistent with the results of a study involving the Japanese population that demonstrated pedicle width and height to be 5.3% and 19.2% greater in men6. We also identified that pedicles measuring <4 mm in width or height were more common in women. Smaller body size is associated with smaller vertebral dimensions27-30. This finding further emphasizes the need for presurgical planning to account for smaller pedicle dimensions and different pedicle orientations in women.
The literature has not provided consistent recommendations regarding the use of surface morphometry to define the pedicle screw entry point. One study suggested that the entry point should be slightly lateral to the center of the articular mass and close to the inferior margin of the cranially adjacent vertebra31. Another study indicated that the entry point was better localized with use of the lateral vertebral notch, defined as the most medial part of the ridge of the pars interarticularis32. In a third study, the investigators were unable to determine consistent entry points for pedicle screw insertion in cadaver spines21. The results of the present study indicated that although transverse and sagittal offsets were significantly different at each level, there was a consistent trend from cranial to caudal in the subaxial vertebrae. The pedicle entry point location in the axial plane, defined by the medial offset, was approximately 1.5 mm medial to the lateral margin of the C3 lateral mass and moved medially at caudal levels. Vertical screw location, defined by the sagittal offset, was approximately 4.0 mm inferior to the superior C3 articular process and moved superiorly at caudal levels. To our knowledge, the present study is the first to quantify the ideal pedicle screw entry points in terms of lateral mass geometry in young, healthy volunteers. Quantitative data may permit screw placement more coaxial to the pedicle axis and may reduce the risk of wall violations.
The risks associated with pedicle screw insertion in the cervical spine can be mitigated by a three-dimensional appreciation of pedicle anatomy. The limitations of our normative data set are that they may be best applied to a young North American population without degenerative changes at the facet joints. Intraoperative positioning of the head and distraction of the facet joints can alter the relationship between adjacent vertebrae, altering sagittal-plane screw placement. Subaxial cervical pedicle screw insertion will continue to depend on the combination of a thorough understanding of pedicle anatomy, careful assessment of preoperative imaging studies, intraoperative recognition of abnormalities, the so-called feel of cancellous bone, and intraoperative radiographic visualization. Nevertheless, the present study offers a quantitative description of normative and disease-free pedicle anatomy and associated relationships, which we believe will facilitate the surgeon's ability to insert subaxial cervical pedicle screws. Because of the sex and age-dependence of spinal anatomy, an effort should be made to assess pedicle morphology preoperatively at each spinal level on a patient-specific basis. 