Changes in intervertebral sagittal alignment of the cervical spine from supine to upright

Cervical sagittal alignment is a critical component of successful surgical outcomes. Unrecognized differences in intervertebral alignment between supine and upright positions may affect clinical outcomes; however, these differences have not been quantified. Sixty‐four patients scheduled to undergo one or two‐level cervical arthrodesis for symptomatic pathology from C4–C5 to C6–C7, and forty‐seven controls were recruited. Upright sagittal alignment was obtained through biplane radiographic imaging and measured using a validated process with accuracy better than 1° in rotation. Supine alignment was obtained from computed tomography scans. Coordinate systems used to measure supine and upright alignment were identical. Distances between adjacent bony endplates were measured to calculate disc height in each position. For both patients and controls, the C1–C2, C2–C3, and C3–C4 motion segments were in more lordosis when upright as compared with supine (all p < 0.001). However, the C4–C5, C5–C6, and C6–C7 motion segments were in less lordosis when upright as compared with supine (all p ≤ 0.004). There was an interaction between group and position at the C1–C2 (p = 0.002) and C2–C3 (p = 0.001) motion segments, with the controls demonstrating a greater increase in lordosis at both motion segments when moving from supine to upright. The results indicate that cervical motion segment alignment changes between supine and upright positioning, those changes differ among motion segments, and cervical pathology affects the magnitude of these changes. Clinical Significance: Surgeons should be mindful of the differences in alignment between supine and upright imaging and the implications they may have on clinical outcomes.


| INTRODUCTION
Restoring native cervical sagittal alignment is believed to improve clinical outcomes after spine surgery. 1,2 It has been noted that cervical sagittal alignment is a predictor of spine degeneration [3][4][5][6] and in some cases may predict clinical outcomes after surgery. 1,2,7 Further, upright cervical MRI has emerged to assess the complex effects of weight-bearing forces upon cervical spine pathology. [8][9][10][11] Intraoperative fluoroscopic imaging is used in spinal fusion and disc arthroplasty cases to ensure optimal sagittal alignment, typically with the patient recumbent. 12,13 This contrasts with physiologic upright weight-bearing cervical films obtained in the preoperative and postoperative setting for evaluation of sagittal alignment with emerging clinical significance. 8,[14][15][16][17] Despite these positional imaging differences and the importance of optimal sagittal alignment, little is known about the quantitative differences in cervical sagittal alignment between supine and upright positions, particularly at the intervertebral motion segment level.
One previous study compared overall cervical spine alignment (C2-C7) between supine and upright positions. 18 They found more (5.93 ± 9.0°) cervical spine lordosis was evident on upright sagittal radiographs than on supine computed tomography (CT) and attributed this difference to forces of gravity. Their assessment was limited to overall spine alignment and failed to assess differences at each cervical motion segment. Additionally, only asymptomatic individuals were evaluated, making the generalizability of results to patients unclear.
Finally, manual measurements were made on the radiographs, which are subjective and dependent upon user expertise. The effect of supine versus upright position on cervical motion segment sagittal alignment and disc deformation remains unknown.
This study aimed to quantify differences in cervical spine intervertebral sagittal alignment and disc height between the upright and supine positions and to determine if differences between supine and upright alignment are affected by symptomatic pathology. We hypothesized that intervertebral lordosis would increase only in the upper cervical spine (C1-C2) when moving from supine to upright. We further hypothesized that the change in alignment would be smaller in patients with pathology than in healthy controls due to reductions in disc compliance with pathology. Finally, we hypothesized that anterior and posterior disc height in the subaxial cervical spine would not change when moving from supine to upright in either controls or patients due to the compensation occurring exclusively in the C1-C2 motion segment.

| Subjects
This study was performed under University of Pittsburgh Institutional Review Board approval (Institutional Review Board protocol PRO19070070). A Level II evidence prospective cohort of patients between 21 and 60 years of age and scheduled to undergo one-level or two-level anterior cervical discectomy and fusion (ACDF) at the C5-C6, C6-C7, C4-C6, or C5-C7 levels for symptomatic cervical pathology was recruited. Inclusion criteria were: Symptomatic cervical pathology with operative indications for progressive symptoms refractory to conservative treatment for myelopathy, radiculopathy, or myeloradiculopathy resulting from degenerative spondylosis, symptomatic cervical disc herniation, or symptomatic disc degeneration. Controls had no neck pain, injury, or disease that would interfere with spine function, such as previous spine surgery or trauma, and had no prior psychiatric condition or neurological disorder. Controls had no radiographic evidence of disease. Pregnant women, breastfeeding women, persons diagnosed with low bone density (which would adversely affect automated alignment measurements), and persons exposed to radiation as part of their current or previous job or as part of the treatment were excluded. This is a secondary analysis of data from a previous study of cervical spine kinematics in healthy young adults 19 and from an ongoing study designed to assess the effects of patient factors, surgical factors, and the number of levels fused on adjacent segment kinematics after ACDF. 20-22

| CT scan collection
CT scans (0.29 mm × 0.29 mm × 1.25 mm voxels) of the cervical spine (C1-C7) were acquired for all controls and preoperatively for each patient. CT scans were collected immediately before or after laboratory testing for all participants (GE-Discovery-CT750 HD). The same thin pillow was placed under each participant's head during the CT scan.

| Supine spine orientation
Bone tissue was segmented from the CT scan images using automated and manual segmentation in Simpleware software (Synopsis Inc) ( Figure 1A). A three-dimensional model of each vertebra was generated from the segmented bone tissue. 23 An anatomic coordinate system was created for each vertebra by placing four markers on the superior endplate and four markers on the inferior endplate corresponding to the center anterior, center posterior, left, and right edges of each endplate ( Figure 1B). The coordinate system was centered on the vertebral body with the vertical axis parallel to the posterior vertebral body. 24,25 The position and orientation of each vertebra relative to the adjacent inferior vertebra were calculated following established standards for reporting spine kinematics, 26 with lordosis defined as negative and kyphosis as positive. 27 An additional point in the center of each endplate was identified, and disc height (anterior, central, and posterior) was measured as the distance between corresponding endplate points in the midsagittal plane of the vertebral body 28 (Figure 2).

| Upright spine orientation
Participants were seated upright and directed to look straight ahead while being imaged within a biplane radiography system ( Figure 1C). Synchronized biplane radiographs of the cervical spine were collected at 30 images per second for 0.1 s. The biplane radiographs were distortion OYEKAN ET AL. | 1539 corrected, and the imaging system was calibrated before making any measurements. 29,30 A validated volumetric model-based tracking technique matched digitally reconstructed radiographs, constructed from the bone tissue segmented from the CT scans to the biplane radiographs ( Figure 1D). This automated tracking process has an in-vivo validated accuracy of 0.3 mm or better and 0.5°or better for measuring position and rotation, respectively. 25 After the registration process was completed for each bone, the orientation of each vertebra relative to the adjacent inferior vertebra and the disc heights were calculated following the same methods used for the supine orientation ( Figure 1E). Importantly, the same marker locations were used for the supine orientation and the upright orientation to define the anatomic coordinate system in each vertebra.

| Statistics
Although CT and biplane radiography both provide 6°of freedom kinematics (three translations and three rotations), only F I G U R E 1 Data collection and processing workflow. (A) CT scans were collected and used to create subject-specific three-dimensional models of each cervical vertebra. (B) Anatomic coordinate systems were established in each vertebra by manually placing four markers on each endplate. (C) Participants were seated upright and directed to look straight ahead while synchronized biplane radiographs were collected. (D) A validated volumetric model-based tracking technique matched digitally reconstructed radiographs, constructed from the bone tissue segmented from the CT scans, to the biplane radiographs. (E) After the registration process was completed for each bone, the orientation of each vertebra relative to the adjacent inferior vertebra was calculated by the angle between the respective green arrows in the bone-based coordinate systems, and the disc heights were calculated by the distance between the red anatomic landmarks placed on the anterior, central, and posterior endplates. CT, computed tomography.
F I G U R E 2 Disc height measurement. Disc height was measured in the supine and upright positions using 3 points along the midsagittal plane of each vertebral body. The measurement points were identical in the supine and upright positions. orientation in the sagittal plane was analyzed for this study.
Average differences in orientation were calculated as supine orientation minus upright orientation so that positive differences indicated more lordosis when upright and negative differences indicated less lordosis when upright. Similarly, positive differences in disc height indicated less disc height when upright and negative differences indicated more disc height when upright. A two-way mixed analysis of variance was conducted to investigate the effects of group (controls vs. patients) and position (supine vs. upright) on overall cervical sagittal orientation, intervertebral sagittal orientation, and intervertebral disc height while accounting for age as a covariate. Significance was set to p < 0.05 for all tests. Based upon this analysis, setting power = 0.80, the study was powered to detect medium effects (f = 0.27). 19

| RESULTS
Data from 111 individuals who provided informed consent were included in the analysis. Forty-seven controls (23 men and 24 women; mean age 35 ± 12 years) and sixtyfour patients (32 men and 32 women; mean age 48 ± 8 years) were enrolled before one or twolevel cervical arthrodesis at the C5-C6, C6-C7, C4-C6, or C5-C7 levels. Patients were older (p < 0.001) with no difference in sex (p = 0.912) compared with controls.

| Cervical sagittal alignment
There was a significant effect of group on overall C1-C7 sagittal alignment after controlling for the effects of age (p < 0.001) with controls in 9.3 ± 5.7°more lordosis, on average ( Figure 3A). There was a significant position by group interaction (p = 0.005), with cervical lordosis decreasing 2.4°in patients, but increasing 4.0°in controls, when moving from supine to upright ( Figure 3B).
There was a significant effect of position on intervertebral alignment at all cervical motion segments after controlling for the effects of age (Figure 4). The C1-C2, C2-C3, and C3-C4 motion segments were in more lordosis when upright as compared with supine; however, the C4-C5, C5-C6, and C6-C7 motion segments were in less lordosis when upright as compared with supine (all p ≤ 0.004) ( Figure 5).
There was a significant effect of group on intervertebral alignment at the C6-C7 cervical motion segment after controlling for the effects of age, with controls (−5.8 ± 7.8°) in more lordosis than patients (−2.0 ± 7.6°) (p = 0.019) ( Figure 6).
There was an interaction between group and position at the C1-C2 (p = 0.002) and C2-C3 (p = 0.001) motion segments after controlling for the effects of age, with the control subjects demonstrating a greater increase in lordosis at both motion segments when moving from supine to upright ( Figure 7).

| Intervertebral disc height
There was a significant effect of position on anterior disc height at all cervical motion segments after controlling for the effects of age (all p ≤ 0.015). The C2-C3 and C3-C4 segments exhibited more anterior disc height when upright as compared to supine; however, the C4-C5, C5-C6, and C6-C7 segments exhibited less anterior disc height when upright as compared with supine (Supporting Information: Figure S1).
There was a significant effect of position on center disc height at the C4-C5, C5-C6, and C6-C7 cervical motion segments after controlling for the effects of age (all p ≤ 0.030).
There was a significant effect of position on posterior disc height at the C2-C3 and C3-C4 cervical motion segments after controlling for the effects of age (all p ≤ 0.012). The C2-C3 and C3-C4 segments exhibited less posterior disc height when upright as compared to supine (Supporting Information: Figure S3). There was a significant effect of group on posterior disc height at the C4-C5 and C6-C7 cervical motion segments after controlling for the effects of age (all p ≤ 0.041), with arthrodesis patients exhibiting less posterior disc height than controls (Supporting Information: Figure S4).

F I G U R E 4
Intervertebral orientation for each participant in supine and upright positions. Datapoints above (green shaded region) and below (yellow shaded region) the diagonal line indicate less or more lordosis, respectively, during upright imaging compared with supine.
F I G U R E 5 Group average intervertebral orientation in supine and upright positions. The C1-C2 through C3-C4 motion segments were in more lordosis when upright than when supine; however, the C4-C5 through C6-C7 motion segments were in more lordosis when supine than when upright. Note that all comparisons were made within a subject; however, the error bars indicate intersubject variability (±1 standard deviation).

F I G U R E 6
Group average lordosis at the C6-C7 motion segment. Overall, controls were in more lordosis than arthrodesis patients at the C6-C7 motion segment. Error bars indicate ±1 standard deviation. This study demonstrated that the C1-C2, C2-C3, and C3-C4 motion segments are in a more lordotic alignment when upright, while the C4-C5, C5-C6, and C6-C7 segments were generally less lordotic when upright in comparison to supine. This finding applied to both patients and controls. However, concerning overall sagittal alignment, the control cervical spines were in more lordosis while upright than supine, while the opposite was true for patients. This finding is consistent with previous research, which concluded asymptomatic cervical spines are globally in greater lordosis on cervical radiographs than on CT scans. 18 That previous study concluded that the increased lordosis when upright may be attributable to the differing effect of gravity on the spine between the upright and supine positions.

| DISCUSSION
Our study expands upon those results by also evaluating symptomatic patients and assessing intervertebral changes in alignment and disc height when moving from supine to upright.
Our study is the first to demonstrate that the C4-C5, C5-C6, and C6-C7 motion segments have relatively less lordosis when moving from supine to upright. This may potentially be attributed to compensatory lower cervical spine movements to maintain the horizontal gaze. Our hypothesis that intervertebral lordosis would increase only in the upper cervical spine (C1-C2) was not supported, as we found intervertebral lordosis also increased in C2-C3 and C3-C4 motion segments when moving from supine to upright. Our finding that lordosis decreased in C4-C5, C5-C6, and C6-C7 when moving from supine to upright was unexpected.
These results demonstrate that gravity and potential effects of maintaining horizontal gaze, may affect the sagittal orientation of the entire cervical spine.
The patients in our study were scheduled to undergo onelevel or two-level ACDF for symptomatic degenerative spondylosis, disc herniation, or disc degeneration. They were globally in less lordosis while upright compared with controls. The significance of the loss of physiologic cervical lordosis is a controversial subject with mechanisms that are not entirely understood. 31 Our findings contribute to the growing body of literature evaluating the relationship of loss of lordosis with conditions including neck pain or symptomatic cervical pathology. The loss of lordosis may be potentially attributable to loss of disc height associated with intervertebral disc degeneration. 32 This assertion was substantiated by data from our patient group, where less intervertebral disc height (measured in the center of the disc) was correlated to less lordosis at the C4-C5 (p = 0.045), C5-C6 (p = 0.010), and C6-C7 (p = 0.008) motion segments. Our hypothesis that changes in alignment would be smaller in patients with pathology than in healthy controls was confirmed with a significant position-bygroup effect on overall sagittal orientation. As indicated earlier, we theorize this occurs due to the degenerative process over the entire cervical spine, not only at the symptomatic motion segments, leading to stiffer intervertebral discs throughout the entire cervical spine. This finding agrees with our previous results that demonstrated bony changes occur throughout the entire cervical spine in the presence of one-level or two-level pathology. 33 Our findings further agree with previous results characterizing overall and level-dependent changes in orientation the lumbar spine when moving from supine to upright, including reduced lordosis, and reduced segmental motion associated with intervertebral disc degeneration. [34][35][36] Controls and patients had increased C2-C3 and C3-C4 anterior disc height with a corresponding decrease in posterior disc height when upright compared with supine. This finding implies that if disc height in the upper cervical spine is assessed in the patient while they are supine, it is likely that posterior disc height will decrease when the patient assumes a physiologic weight-bearing upright position. Our hypothesis that disc height would increase while supine, as compared with upright, was confirmed at the anterior and center disc regions for the C4-C5, C5-C6, and C6-C7 segments.
Strengths of our study include the high degree of accuracy for the measurement system, including the ability to measure intervertebral changes using identical coordinate systems for supine and upright imaging, eliminating the possibility that any observed differences were due to inconsistency in manually identifying anatomic landmarks in supine and upright radiographs.
Similarly, our radiographic findings are strengthened by additional regional disc deformation measurements which provide quantitative results that corroborate our sagittal rotation findings and highlight the importance of accounting for differences in intervertebral disc geometry when moving from supine to upright when constructing computational models (e.g., finite element models). Limitations to our study include the lack of assessment of horizontal gaze to confirm speculated mechanisms of changes in alignment and demographic differences in age between the control and pathologic patient groups, which were addressed by accounting for age in the analysis. We further conducted a secondary analysis to compare our young (age ≤ 35 years) controls to older (age > 35 years) controls and found more lordosis evident in young controls at C1-C2 (young: −11.4 ± 4.8°vs. old: −7.4 ± 6.5°; p = 0.024) and C2-C3 (young: −9.2 ± 5.2°v s. old: −3.9 ± 6.0°; p = 0.024) on CT scan alone. No differences were identified in static upright radiograph alignment (all p > 0. 1) or between CT and upright alignment (all p > 0.07) when comparing the young and older controls. If we had included only older controls in our analysis, the differences between supine and upright would have been even larger in the control cohort, but none of the differences observed between controls and patients would have changed. Finally, the amount of change in intervertebral alignment needed to achieve a clinically significant change in symptoms remains unknown.
The clinical implications of our study primarily pertain to clinicians performing anterior cervical decompression and fusion or disc arthroplasty surgery. These procedures are performed in supine positioning where intraoperative fluoroscopic radiographs are obtained to critically assess sagittal alignment, instrument localization, implant position, and fusion construct angles. However, preoperative radiographs are obtained in a physiologic upright position while supine CT scans are used to estimate anticipated intraoperative position creating a previously poorly understood discrepancy in alignment. Clinicians must be aware of the effects positioning has upon sagittal alignment to maximize clinical outcomes.
In conclusion, our study demonstrates that cervical motion segment alignment changes between supine and upright positioning and those changes differ among motion segments. The C4-C5, C5-C6, and C6-C7 motion segments exhibit less lordosis when upright as compared to supine. Cervical pathology affects the magnitude of these changes, possibly due to differences in disc compliance. Surgeons should be mindful of the differences in alignment between supine and upright imaging due to the impact they may have on clinical outcomes.

ACKNOWLEDGMENT
This study was supported by NIH Grant #1R01AR069543.