In Vivo Evidence of Early Instability and Late Stabilization in Motion Segments Immediately Superior to Anterior Cervical Arthrodesis

Abstract

Study Design:

Prospective cohort study.

Objective:

To identify patient factors that affect adjacent segment kinematics after anterior cervical discectomy and fusion (ACDF) as measured by biplane radiography.

Summary of Background Data:

The etiology of adjacent segment disease (ASD) may be multifactorial. Previous studies have investigated associations between patient factors and ASD, although few attempted to link patient factors with mechanical changes in the spine that may explain ASD development. Previous studies manually measured intervertebral motion from static flexion/extension radiographs, however, manual measurements are unreliable, and those studies failed to measure intervertebral motion during rotation.

Methods:

Patients had continuous cervical spine flexion/extension and axial rotation movements captured at 30 images per second in a dynamic biplane radiography system preoperatively and 1 year after ACDF. Digitally reconstructed radiographs generated from subject-specific CT scans were matched to the biplane radiographs using a validated tracking process. Dynamic kinematics and preoperative disc height were calculated from this tracking process. Preoperative MRIs were evaluated for disc bulge. Patient age, sex, BMI, smoking status, diabetes, psychiatric history, presence of an inciting event, and length of symptoms were collected. Multivariate linear regression was performed to identify patient factors associated with 1-year postoperative changes in adjacent segment kinematics.

Results:

Sixty-three patients completed preoperative and postoperative testing. Superior adjacent segment disc height and disc bulge predicted the change in superior adjacent segment ROM after surgery. Inferior adjacent segment disc bulge, smoking history, and the use of psychiatric medications predicted the change in inferior adjacent segment flexion/extension ROM after surgery.

Conclusions:

Preexisting adjacent segment disc degeneration, as indicated by disc height and disc bulge, was associated with reduced adjacent segment motion after ACDF, while lack of preexisting adjacent disc degeneration was associated with increased adjacent segment motion after ACDF. These findings provide in vivo evidence supporting early instability and late stabilization in the pathophysiology of disc degeneration.

Introduction

Anterior cervical discectomy and fusion (ACDF) remains a gold standard treatment for cervical radiculopathy and myelopathy, with more than 150,000 procedures performed in the US per year^1–3. Although ACDF has good short-term outcomes, approximately 25% of patients will develop symptomatic adjacent segment disease (ASD) and require reoperation within 10 years^4,5. The etiology of ASD remains controversial. It has been hypothesized that certain individuals have a predisposition to developing degenerative changes of the spine. Previous research has identified associations between ASD and genetics⁶, age^7,8, BMI⁹, psychiatric history⁸, and preoperative disc degeneration^7,9–12. However, there is also evidence to suggest that ASD may be a consequence of the surgical procedure. In cadaveric biomechanical studies, fusion results in compensatory hypermobility at adjacent motion segments with a subsequent increase in disc stresses that may lead to ASD^13–17. However, similar findings have not been observed in vivo¹⁸. These two existing theories are not exclusive to one another and perhaps both patient and mechanical factors contribute to the development of ASD.

While previous studies have investigated the independent effect of patient factors on the development of ASD, few studies have attempted to link patient factors to mechanical changes in the spine that could explain the development of ASD^8–10,19. Previous research suggests that preoperative disc degeneration, which may be characterized by disc bulge and loss of disc height, has the strongest association with development of ASD^7,9–12,20. Studies that attempted to correlate patient factors with kinematics examined range of motion (ROM) using measurements from static end-range sagittal radiographs^7,21,22. Sagittal radiographs are notoriously unreliable and are also limited to measurements of flexion and extension, while neglecting axial rotation, which comprises 25% of everyday motion^23,24. Biplane radiography can address these limitations by accurately measuring arthrokinematics of the spine dynamically in multiple planes of motion^25,26. Although patient factors may not be easily modifiable, identification of patient factors associated with ASD may provide prognostic information for patients as well as allow surgeons to preemptively address pathology that is likely to become symptomatic and avoid revision surgery.

The purpose of this study was to use multivariate analysis to identify patient factors related to changes in adjacent segment kinematics after ACDF. It was hypothesized that the patient-specific factors of preoperative disc height and disc bulge would be the primary patient factors associated with the change in adjacent segment kinematics after ACDF.

Methods

Subjects

Following institutional review board approval, patients who were between twenty-one and sixty years of age and were scheduled to undergo one-level or two-level ACDF at the C4-C5, C5-C6, C6-C7, C4-C6, or C5-C7 levels due to symptoms associated with myelopathy, radiculopathy or myeloradiculopathy resulting from either degenerative spondylosis, cervical disc herniation or disc degeneration were identified by clinical review. Pregnant women or women planning to be pregnant within 3 years of surgery, patients diagnosed with osteoporosis, and patients with any other injury or disease that interferes with spine function were excluded. Sample size was determined a priori, with 80% power and α = .05, with a total of 55 arthrodesis patients needed to identify patient factors that have a medium effect (or greater) on post-surgical kinematics²⁷.

Data Collection and Processing

Participants performed three trials each of full range-of-motion (ROM) flexion-extension and axial rotation within a biplane radiography system before and one-year after surgery (Figure 1A). Synchronized biplane radiographs were collected at thirty images per second for three seconds during each trial (Figure 1B). Motion was performed to the beat of a metronome set at a rate of forty-four beats per minute to ensure completion of one full movement cycle in approximately three seconds. Total radiation exposure associated with biplane radiography was estimated to be 0.83mSV (estimated using PCXMC, STUK - Radiation and Nuclear Safety Authority, Helsinki, Finland).

Figure 1:

The biplane radiography system and volumetric bone-model-based tracking technique. A) The participant performed dynamic movements within the biplane imaging system. B) Synchronized radiographs were collected using high-speed cameras. C) Bone tissue was segmented from CT scan images. D) Three-dimensional subject-specific bone models of each vertebra were created from segmented bone tissue. E) Subject-specific bone models were placed in a computer-generated reproduction of the biplane system. Simulated X-rays were passed through bone models to generate digitally reconstructed radiographs (DRRs). Bone position and orientation were determined by an optimization process that matches the DRRs to the edge-enhanced radiographs. Intervertebral kinematics were calculated for each motion segment of the cervical spine.

Computed tomography (CT) scans (0.29 × 0.29 × 1.25 mm voxels) of the cervical spine (C1-C7) were acquired preoperatively for each participant. Bone tissue was segmented from the CT images using automated and manual segmentation in Simpleware software (Synopsis Inc, Mountain View, CA) (Figure 1C). A three-dimensional model of each vertebra was generated from the segmented bone tissue (Figure 1D)²⁸. A validated volumetric model-based tracking technique matched digitally reconstructed radiographs constructed from the three-dimensional bone models to the previously obtained biplane radiographs (Figure 1E). This tracking process has a validated in-vivo accuracy of 0.19 mm for tracking individual bone motion, 0.2 mm to 0.4 mm for measuring anterior and posterior disc height, respectively, and 1.1° or better for measuring rotation²⁹. Movement of C3 through C7 bones was included in the present analysis. A low-pass 4^th order Butterworth filter was used to smooth the 6 DOF motion path of each bone (1.7 Hz cutoff for flexion/extension and 1.0 Hz cutoff for rotation, determined based upon residual analysis)³⁰.

Total ROM at each motion segment and disc height were calculated using anatomic coordinate systems centered on each vertebral body with the vertical axis parallel to the posterior vertebral body³¹. All measurements were performed using the same landmark locations on the 3D bone models pre- and postoperative to eliminate uncertainty associated with manual identification of landmarks. The maximum flexion, extension, and right and left rotation at each motion segment were determined from the three trials of each movement. Change in range of motion was calculated as the difference between postoperative and preoperative ROM. Disc height was calculated as the distance from the center of the inferior endplate of the superior vertebral body to the center of the superior endplate of the inferior vertebral body. Patient age, sex, BMI, smoking status, diabetes, psychiatric history, presence of an inciting event, and length of symptoms were recorded from the medical record. Preoperative clinical MRIs (T2, 3.0 mm slice thickness) were evaluated by two observers for presence of disc bulge at C34, C45, C56, and C67.

Statistics

Multivariate linear regression was performed using the backward method, where all independent variables were entered into the equation initially and then the variable with the highest p-value was sequentially removed at each step until only variables that contributed significantly to predicting the dependent variable remained, with p > 0.10 to be removed (SPSS 24.0, IBM Corporation, New York, USA). The pre- to postoperative change in superior and inferior adjacent segment flexion/extension and axial rotation ROM served as the dependent variables. Independent variables included in the analyses were patient age, sex, BMI, smoking status, diabetes, psychiatric history, presence of an inciting event, length of symptoms, as well as surrogate markers for existing degenerative disease including preoperative adjacent segment disc height and the of presence of a disc bulge observed on MRI³².

Results

Seventy-five individuals provided informed consent and agreed to participate in this ongoing study. Sixty-three individuals completed both preoperative and 1-year postoperative testing (Table 1). Average radiation exposure from the cervical spine CT was 2.12 mSV. The mean age was 48.7±7.7 years. There were 31 males and 32 females. The mean change in flexion/extension ROM at the superior adjacent motion segment was 0.7° (range −10.0° to 10.1°) and at the inferior adjacent motion segment was 2.3° (range −11.3° to 11.5°). The mean change in axial rotation ROM at the superior adjacent motion segment was 0.7° (−2.8° to 5.5°) and at the inferior adjacent motion segment was 0.7° (−2.0° to 4.2°). For MRI grading of disc bulge, there was weak agreement at C34 (ICC=0.48) and moderate agreement at C45 (ICC=0.74), C56 (ICC=0.69), and C67 (ICC=0.60) discs.

Table 1:

Demographics, injury characteristics, and operative data of the study participants

Demographics
Age (Years)	48.7 ± 7.7
Sex
Male	31
Female	32
Diabetics (%)	9 (14.3%)
Smokers (%)	14 (22.2%)
Psychiatric History	20 (31.7%)
Injury Characteristics
Inciting Event (%)	21 (33.3%)
Length of Symptoms (Months) (Mean ± SD, Median [Min, Max])	17.7 ± 25.5, 8 [1, 146]
Operative Data
One-Level Fusions	33
C45	1
C56	18
C67	14
Two-Level Fusions	30
C456	10
C567	20
Graft Type
Autograft	10
Allograft	53
MRI Data
Presence of Disc Bulge
Superior Adjacent	41
Inferior Adjacent	18
ROM
Superior Adjacent	Flexion/Extension	Axial Rotation
PRE	13.3 ± 4.2	8.3 ± 2.6
POST	14.1 ± 3.9	9.0 ± 2.9
Inferior Adjacent
PRE	10.9 ± 4.6	4.7 ± 2.4
POST	13.2 ± 5.3	5.4 ± 2.5
Disc Height
Superior Adjacent	3.7 ± 0.7 mm
Inferior Adjacent	3.8 ± 1.1 mm

The multivariate regression analysis identified superior adjacent segment disc height and disc bulge as factors related to the change in superior adjacent segment ROM after surgery. The regression indicated that the superior adjacent segment flexion/extension ROM increased 1.2° for each millimeter of superior adjacent segment disc height prior to surgery (p = 0.032), for example, those with less than 3mm of adjacent segment disc height prior to surgery had an average loss of 0.4° in flexion/extension ROM after surgery, while those with greater than 5mm of disc height had an average increase of 2.2° in flexion/extension ROM after surgery. The regression analysis also indicated that the superior adjacent segment axial rotation ROM increased 0.7° for each millimeter of superior adjacent segment disc height prior to surgery (p = 0.042) and axial rotation ROM decreased 0.6° in the presence of disc bulge prior to surgery (p = 0.044).

The multivariate regression analysis identified inferior adjacent segment disc bulge, smoking history, and the use of psychiatric medications as factors related to the change in inferior adjacent segment flexion/extension ROM after surgery. The regression indicated that the inferior adjacent segment flexion/extension ROM increased 4.0° in the presence of psychiatric medications (p = 0.019), but flexion/extension ROM decreased 3.8° and 2.8° in smokers (p = 0.087) and in the presence of disc bulge (p = 0.013), respectively. There were no significant associations between any independent variables and the change in axial rotation ROM at the inferior adjacent segment.

Discussion

The objective of this study was to perform a multivariate analysis to identify patient factors that affect adjacent segment kinematics one year after ACDF as measured by dynamic biplane radiography. An increase in superior adjacent segment flexion/extension ROM from pre to post-surgery was associated with greater preoperative superior adjacent segment disc height, while an increase in superior adjacent segment axial rotation ROM from pre to post-surgery was associated with greater preoperative superior adjacent segment disc height and the absence of disc bulge. An increase in inferior adjacent segment flexion/extension ROM was associated with psychiatric medications, however smoking and the presence of a disc bulge were associated with a decrease in inferior adjacent segment flexion/extension ROM from pre- to post-surgery. Other patient factors investigated – age, sex, BMI, diabetes, presence of an inciting event, and length of symptoms were not related to changes in superior or inferior adjacent segment ROM from pre-surgery to one-year post-ACDF.

At the superior adjacent segment, the change in flexion/extension and axial rotation ROM one-year post-ACDF were associated with preoperative disc height and the presence of a disc bulge. In the MRI grading system used for cervical degeneration, disc bulge and decreased disc height were indicators of moderate and severe disc degeneration²⁰. These results reflect the progression of disc degeneration proposed by Kirkaldy-Willis and Farfan, where disc degeneration progresses from an early instability to late stabilization³³. In patients without preoperative disc degeneration – larger disc height and absence of disc bulge – there was an increase in adjacent segment motion from pre-surgery to one-year post-ACDF, which may reflect the initiation of instability that starts the degenerative process. For patients with evidence of preoperative disc degeneration – smaller disc height and presence of disc bulge – there was a decrease in adjacent segment motion from pre-surgery to one-year post-ACDF, which may reflect the pathologic stabilization later in the degenerative process. The decrease in adjacent segment motion is likely related to loss of disc height, soft tissue scarring, joint surface irregularities, and bony abnormalities such as osteophyte formation³³. Given that disc height and disc bulge have both been associated with disc degeneration, we performed a secondary collinearity analysis to confirm that disc height and disc bulge were not correlated in our sample.

At the inferior adjacent segment, a similar relationship between ROM and preoperative adjacent segment disc degeneration (only disc bulge) was observed, and additionally, smoking history and the use of psychiatric medications were also identified as factors related to the change in inferior adjacent segment flexion/extension ROM after surgery. A limitation of this study is that our data collection process did not capture the C7-T1 disc space and kinematics, which meant that only 29/63 (46.0%) participants had inferior adjacent kinematics and preoperative inferior adjacent segment disc height or disc bulge recorded, thus, conclusions regarding the inferior adjacent segment level may be subject to underpowering.

Historically, smoking and diabetes have been correlated with an increased rate of pseudoarthrosis, which would result in residual motion at the arthrodesis site and potentially reduce adjacent segment compensatory hypermobility^34–36. Our results partially support this theory. In our analysis, smoking predicted a decrease in inferior adjacent segment ROM during flexion/extension, however, diabetes was not correlated with any change in adjacent segment motion. In a previous analysis of this cohort, solid bony fusion was noted in 48.9% of patients at one-year post-ACDF and there was no significant difference in rate of fusion for patients who were active smokers and/or who were diabetics³⁷. Other patient factors that may increase the rate of pseudoarthrosis and indirectly affect adjacent segment degeneration include advanced age, obesity, and the use of psychiatric medications due to negative changes in bony metabolism^38–40. The use of psychiatric medication was the only one of these factors identified in this study as potentially affecting the change in ROM after ACDF. It is possible that our results could be confounded by the opposing effects on kinematics from pseudoarthrosis and compensatory increased motion after fusion, which could explain the lack of association between some patient factors and adjacent segment kinematics. However, a post-hoc analysis of this cohort found no relationship between motion at the operated site and change in adjacent segment ROM.

A limitation of the current study is the relatively short follow-up time given that ASD develops at a rate of about 3% per year⁵. However, this early timepoint allowed us to obtain in vivo evidence that suggests changes in adjacent segment motion after ACDF may follow the time course proposed by Kirkaldy-Willis of instability (increased motion) in healthy discs and stabilization (decreased motion) in more degenerated discs. These short-term results will serve as a valuable longitudinal datapoint as mid-term and long-term follow-up data continue to become available in this cohort. Additionally, the results of this study are limited to one- and two-level arthrodesis of the C4-C5 to C6-C7 motion segments of the cervical spine. Results may differ in the presence of longer fusion constructs or disc arthroplasty. This study was also limited to morphologic assessment of disc health using standard clinical MRI data. Quantitative MRI, such as T2 mapping, may provide an early indication of changes in disc tissue composition⁴¹ and provide additional insight into the relationship between disc health and changes in adjacent segment motion after ACDF. Finally, preoperative ROM may have been affected by neck pain which may have confounded the results⁴².

One strength of the study was the multivariate analysis, which accounted for relationships between the independent variables that can confound results, and can be used to discern the relative effects of each patient factor on the change in ROM. Furthermore, the sample size of 63 patients was of a sufficient size to identify medium or larger effect size associations if they existed^43,44. Additional strengths include the prospective study design, the use of a validated and accurate system to measure dynamic intervertebral motion, the inclusion of flexion/extension and axial rotation motions in the analysis, and the recording of multiple movement trials to determine the maximum ROM at each motion segment. Although patient factors are not easily modifiable when compared to surgical factors, identifying patient factors that predict ASD may provide prognostic information for patients and allow surgeons to preemptively treat these segments to avoid costly revision surgery.

Conclusion

Following one- and two-level ACDF, preexisting adjacent segment disc degeneration, as indicated by disc height and disc bulge, was associated with reduced motion in the superior adjacent segment at 1-year postoperatively. Lack of preexisting adjacent segment disc degeneration was associated with increased motion in the superior adjacent segment at 1-year postoperatively. These findings provide in vivo evidence to support the theory that disc degeneration may progress from early instability to late stabilization. Follow-up in this cohort is ongoing to determine if mid-term changes in adjacent segment kinematics parallel the predicted course of progression from early instability to late stabilization as well as identifying patient factors that may predict the eventual development of ASD and revision surgery.