Odontoid Fracture as Proximal Junctional Failure in Patients With Multilevel Spine Fusions

Abstract

Study Design:

Retrospective study.

Objective:

Proximal junctional failure (PJF) commonly occurs as a recognized potential outcome of fusion surgery. Here we describe a unique series of patients with multilevel spine fusion including the cervical spine, who developed PJF as an odontoid fracture.

Methods:

We performed a single site retrospective review of patients with prior fusion that included a cervical component, who presented with an odontoid fracture between 2012 and 2019. Radiographic measurements included C2-C7 SVA, C2-C7 lordosis, T1 slope, Occiput-C2 angle, proximal junctional kyphosis, and cervical mismatch. Associated fractures, medical comorbidities, and treatments were determined via chart review after IRB approval.

Results:

Nine patients met inclusion criteria. 5 reported trauma with subsequent onset of pain. All patients sustained a Type II odontoid fracture. 5 with associated C1/Jefferson fractures. In all patients, pre-injury Occiput-C2 angle was outside normative range; C2-C7 SVA was greater than 4 cm in 6 patients; T1-slope minus cervical lordosis was greater than 18.5 degrees in 6 patients. 7 patients were treated operatively with extension of fusion to C1 and 2 patients declined operative treatment.

Conclusion:

In this series of 9 patients with multilevel fusion with type II odontoid fractures, all patients demonstrated abnormal pre-fracture sagittal alignment parameters and a greater than normal association of C1 fractures was noted. Further study is needed to establish the role of poor sagittal alignment with compensatory occiput-C2 angulation as a predisposing factor for odontoid fracture as a proximal junctional failure mechanism.

Introduction

The significance of sagittal alignment in thoracolumbar spinal deformity correction is well established, with the use of pelvic incidence (PI) as a framework to guide corrections.^1,2 Development of an analogous measurement in the cervical spine has long been a goal. The use of T1 slope (T1 S) has been considered the key to understanding patient appropriate cervical lordosis (CL).^3,4 The calculation of cervical mismatch (T1S-CL) has been suggested to describe this association. ⁵

The upper cervical region, including the occiput and the C1 and C2 vertebrae are anatomically unique. Measurements of CL and cervical mismatch do not account for this region, despite this area accounting for significant range of motion. Occiput-C2 (O-C2) angulation has been proposed to describe the sagittal alignment here.^6,7

Functionally, the components of the spine coordinate to maintain the skull horizontal to maximize vision utility. ⁸ In the setting of cervical fusion, the remaining mobile spine attempts to compensate and maintain horizontal vision. ⁹ When lower and mid cervical vertebrae are fused, this compensation must occur at the upper cervical region.

The expressed intention of optimizing cervical mismatch is to produce a cervical construct that is as least obtrusive to cervical function and load bearing as possible. Fusion imparts rigid segments where previously mobile segments resided, resulting in compensation by and increased stress to adjacent levels, which leads to an increased risk of proximal junctional failure (PJF). It is presumed that the greater the discrepancy from an ideal cervical alignment, the more stress, and as such, the greater the risk of PJF on adjacent segments.

We present here a series of patients with prior multilevel fusion, that included the cervical spine, and high cervical compensation because of cervical mismatch that predisposed to an odontoid fracture. We hypothesize the odontoid fracture is a unique manifestation of PJF.

Methods

Patient Population

This is a retrospective analysis of patients who presented for evaluation between 2012 and 2019 to either the Neurosurgical or Orthopedic services at the University of Minnesota with an odontoid fracture and prior history of a cervical fusion. Inclusion criteria consisted of an age older than 18 years, odontoid fracture of any type, and prior cervical fusion of any number of levels and approaches. This review was approved by the University of Minnesota Institutional Review Board (IRB).

Radiographic Analysis and Data Collection

The basic demographic data of age and sex was collected. The Electronic Medical Record (EMR) of each patient was reviewed for a possible mechanism of injury, as well as for a detailed account of prior cervical surgical interventions. Prior bone density testing and known comorbid disease processes were included. Each patient received a cervical CT scan at the time of initial evaluation, which was reviewed to document the type of odontoid fracture and concomitant fractures, when applicable. Radiographic measurements included C2-C7 SVA, C2-C7 lordosis (CL), T1 slope (T1 S), Occiput-C2 angle (O-C2), cervical mismatch (TIS-CL), and proximal junctional kyphosis (PJK). When available, an upright x-ray film, from prior to the patient’s fracture, was used preferentially to complete the aforementioned radiographic measurements. If a suitable pre-fracture image was not available, imaging from the initial evaluation was used. O-C2 angulation was measured using McRae’s line, as the hard palate was unfortunately frequently not visible to draw McGregor’s line. Pre-fracture imaging of the entire spine was rare, as such, measurements of the thoracolumbar spine were not included.

Statistical Analysis

The mean and standard deviation was calculated. Each data set was tested for normality. Regression analysis was performed between data sets. All statistical analysis was conducted using Minitab (Minitab 17, State College, PA).

Results

9 patients met inclusion criteria: 5 males and 4 females. Mean age, at time of presentation, 68 (49-85) years. 2 patients were known to have underlying osteoporosis (T-scores -2.1 and -2.6). The bone mineral density of the remaining patients was unknown. 1 was secondary to chronic steroid use in the setting of a chronic inflammatory condition. 3 patients were either current smokers or had a prior history of smoking. The patients included in the series had a wide range of prior fused levels: C2-C7, C2-T2, C2-S1, 2 C2-Pelvis, C3-C7, C4-C7, C4-T1, and C5-C7. Only 1 patient had a single anterior approach for decompression and instrumentation (C3-C7). All other patients had undergone multiple spinal surgeries that consisted of anterior and posterior approaches, involving multiple institutions, over the course of many years.

5 patients initially presented to the Emergency Department following a fall with immediate onset of neck pain. Of the 5 patients with a fall: 3 fell forward, striking their head; 1 fell down stairs; 1 had an undescribed “fall.” The remaining 4 patients could not recall a specific traumatic event, however, imaging was obtained due to concerns for ongoing neck pain. 4 patients presented neurologically intact (ASIA E). 4 patients presented with only mild weakness (ASIA D). 1 patient had originally presented to another institution some years prior and an exam was not available. CT demonstrated all patients sustained a Type II odontoid fracture—5 with associated C1/Jefferson fractures. Pre-fracture imaging was available for 7 patients. This constituted 6 upright cervical x-rays and 1 cervical CT. Pre-fracture imaging was not available for 2 patients, so the cervical x-rays obtained at the time of presentation were used. For 1 patient, this was an upright image, the orientation is unknown for the other patient.

The results of the radiographic measurements can be found in Table 1. Figure 1 depicts the individual value plot for each measurement, with respective nominal ranges included.

Table 1.

Results of Cervical Alignment Radiographic Measurements.

Measurement	Mean	Standard deviation	Normal
C2-C7 SVA	+53.6mm	24.5	< 40mm 10
T1S	39.4°	22.4	22° ± 7 11
CL	+7.8°	14.5	17° ± 7 12
O-C2 angle	+51.2°	14.1	14° ± 7 6
Cervical mismatch	+31.6°	24.5	16.5° ± 2 12

Figure 1.

Individual value plot of: A, C2-C7 SVA. B, Cervical lordosis. C, Occiput-C2 angulation. D, Cervical mismatch; UL, upper limit; LL, lower limit. ● Pre-injury measurements. ○ Post-injury measurements.

As depicted above, cervical alignment was outside what is considered normal ranges for every patient for more than 1 parameter. C2-C7 SVA was greater than 40 mm in 6 patients. The cervical lordosis of 8 patients was outside the normal range. All patients’ occiput-C2 angle and cervical mismatch were outside normal ranges. 4 patients had an upper instrumented vertebrae (UIV) below C2 allowing for a PJK measurement. A simple linear regression of occiput-C2 on PJK demonstrated a trend that as PJK increased, the occiput-C2 angle increased (R² = 90.3%, R²(adj) = 85.4%, P-value = .05). The small sample size makes it difficult to draw more definitive conclusions.

8 patients were offered surgical fixation at the time of initial evaluation. 1 patient was placed in a halo, failed to fuse, and was subsequently offered surgical fixation. 2 patients declined surgical intervention and were placed in a collar. No evidence of fusion was present on follow-up imaging. Of the patients who underwent surgery, 5 underwent fusion at the C1-C2 level, with extension to prior construct or fusion mass. 1 patient, with a prior lower cervical fusion, underwent C1-C2 fusion without extension to prior fusion for motion preservation. 1 patient underwent Occiput-C2 fusion due to significant pannus formation. Figures 2 and 3 depict select cases that demonstrate the pre-operative cervical alignment, fracture, and fixation construct.

Figure 2.

Depiction of patient with prior lower cervical fusion: A, Failed prior subaxial wiring. B, Displaced and angulated odontoid fracture. C, C1-C2 posterior instrumentation and fusion without extension to inferior fusion mass for motion preservation.

Figure 3.

Depiction of patient with overcompensated occipital-C2 angulation: A, Associated C1 burst fracture. B, Displaced odontoid fracture. C, C1-C2 posterior instrumentation and fusion with extension to inferior fusion mass.

Discussion

The concept of cervical sagittal alignment is a developing one, primarily derived from asymptomatic population studies and postoperative disability metrics.^13-16 To the knowledge of the authors, this is the first published series of cervical sagittal alignment measurements in the setting of PJF manifesting as an odontoid fracture. ¹⁷ Analogous to our growing understanding of the distal consequences of fusion, namely sacroiliac joint and hip dysfunction,^18-20 this study highlights the potential for proximal consequences.

In this patient series, trauma was the primary cause of fracture, which is the most common cause of odontoid fracture in patients without prior cervical fusion. ²¹ Patients in the series experienced Type II fractures, some associated with cervical extension trauma, which also coincides with previously established literature. ²² However, in this series, the rate of associated C1 fractures was greater than what would be expected in a population without prior cervical fusion (56% vs 9%, ²¹ respectively).

The occiput-C2 angulation results depict a patient population with pre-existing hyperextension in the upper cervical region (Figure 4). This likely represents a compensatory mechanism for poor sagittal alignment. ²³ Whether this represents compensation for local cervical malalignment, a more global malalignment, or some combination thereof is unknown given the paucity of full spine imaging in this patient population. The available C2 SVA and CL results suggest at least some contribution from cervical malalignment, given the high SVA and low CL measurements of many patients.

Figure 4.

Occipital-C2 pre-fracture hyperextension of a patient included in the study.

The mechanism of fracture for the patients who sustained a cervical hyperextension trauma coincides with the classic teaching of odontoid fractures. The mechanism of fracture for patients without a history of trauma is not clear, but one could surmise repetitive microtrauma, as a contributing factor, from upper cervical hyperextension as the patient worked to maintain a horizontal gaze. In both traumatic and non-traumatic presentations, the patients were without the ability to further extend at the cervical spine. Concurrently, the upper cervical spine was bracketed by long moment arms cephalad and caudal. Due to the compensatory hyperextension between the occiput and C2, it is our theory that a majority of the stress was concentrated at the odontoid, as it would resist further extension, resulting in the presenting odontoid fractures. Given the complicated dynamic biomechanics involved, this might also explain the higher than expected rate of associated C1 fractures. While the upper cervical spine is hyperextended, the anterior ligaments would experience tension and the posterior osseous structures compression, contributing to the high rate of associated C1 fractures. Further study is warranted to confirm the observed association in this series and proposed theories. The addition of computational modeling and cadaver studies could also shed light on the suspected failure mechanism.

At a minimum, the presented data suggests the importance of considering cervical sagittal alignment when generating an operative plan for patients. Further, the data suggests careful consideration of cervical alignment may also lower the risk of adjacent fractures, necessitating repeat operative interventions.

The small sample size and heterogeneous patient population are the primary limitations of our study, which could benefit from a large multicenter approach or well-designed meta-analysis. Despite anecdotal reports of this failure mode at other institutions, very little information can be found in the literature. For this reason, we present the data to raise awareness and spur discussion of this failure mode. This study is also limited by the availability of full spine radiographs to assess global sagittal balance and possible contributing drivers of cervical hyperextension. As low-radiation imaging technologies become more widely integrated into practices, this information will likely become more readily available.

Conclusions

This is a retrospective, single site, series of patients with previous multilevel cervical fusion, who presented with PJF manifesting as an odontoid fracture. These patients presented with a greater than normal association of C1 fractures. Pre-fracture cervical alignment parameters demonstrated predominately positive C2 SVA and inadequate cervical lordosis with resultant high O-C2 compensatory angulation. The compensatory cervical extension, in response to poor cervical sagittal alignment, may represent a predisposing factor for odontoid fracture as a PJF mechanism.