Article focus

• A finite element model of the complete spine was used to assess biomechanics after surgery

• Coupled motions including multi-level cervical and thoracolumbar fusions were assessed using during realistic whiplash scenarios

Key messages

• Cervical fusion increases ALL strain at the adjacent segment in a realistic whiplash scenario.

• C6-C7 experienced the highest strain under these conditions, but cervical fusion had minimal effect. C3-C4 had a disproportionately high increase in strain following C4-C5 fusion.

• Thoracolumbar fusion had no effect on cervical spine strain.

Strengths and limitations

• This study determined that adjacent segment strain following spinal fusion disproportionately affects certain motion segments.

• A quantitative rationale for adjacent segment disease as a biomechanical consequence of fusion as well as adjacent segment disease as natural progression of underlying disease was established.

• A standardized 75-kg adult male and standardized Euro NCAP whiplash scenario was modelled. Other scenarios were not evaluated by this model.

Introduction

Whiplash injuries, caused by a sudden or unexpected neck flexion or extension, are the most often reported injuries in low-velocity rear-impact vehicle collisions.¹ The economic impact of whiplash is estimated to be $3.9 billion annually in the United States, or more than $29 billion when litigation costs are considered.² Some clinical challenges of whiplash include associated secondary gain that may confound injury,² associated non-organic signs of disability,² and the difficulty of establishing consistent diagnostic MRI findings.³ Nonetheless, large defined population studies have shown that the incidence of patients presenting with whiplash related complaints may increase even when a concomitant decrease in insurance claims is observed.⁴ Additionally, the overall body of literature provides evidence supporting a lesion-based model of whiplash injury.⁵ Injury to the anterior longitudinal ligament (ALL) appears to be a marker to severe whiplash injuries in both cadaveric studies⁶ and post-mortem studies.^7,8

Most studies on whiplash have focused on patients with no pre-existing cervical spine disease. Cervical spine arthrodesis is a common procedure, with over 1.1 million patients undergoing the procedure over an eight-year period in the United States.⁹ The effects of cervical arthrodesis on the adjacent motion segments above and below the level of fusion still remain a point of debate. It is clear that there is increased strain in the soft tissue and vertebrae in adjacent motion segments, due to compensation for the new loss of flexibility.^10,11 However, motion-sparing procedures, such as cervical disc arthroplasty, have failed to show decreased rates of adjacent segment disease (ASD), at least for single level surgery.¹²

We have previously quantified the biomechanical effects of an 8 g whiplash following cervical arthrodesis on adjacent segment strains seen in the ALL using a validated finite element (FE) model.¹¹ In our original study, the impact scenario selected was based on whole cadaver experiments performed by Mertz¹³ in 1967 and whole cervical spine model experiments performed by Ivancic¹⁴ in 2004, reflecting an 8 g peak acceleration. Although those historical impact pulses provided a method for validation and assessment of failure, newer data is available that reflects more realistic real-world, rear-impact collisions.¹ We sought to study cervical spine biomechanics after spinal fusion using the low-speed acceleration pulses described by the European New Car Assessment Programme (Euro NCAP). Additionally, our prior study used a simplified FE model that did not capture the additional effects caused by coupled motions from the lumbar spine, nor the effects created by the interaction between the model and the driver’s seat. Advances in computational resources have allowed us to address these limitations.

We sought to provide quantitative data on the potential effects of adjacent segment strains following cervical and lumbar arthrodesis and test the hypothesis that specific motion segments experienced disproportionate changes in ALL strain and that there was no effect of lumbar fusion on cervical spine biomechanics.

Materials and Methods

Seat Interaction Model

In the second set of simulations, a seat and floor were added using rigid 4-by-4 element shells through LS-PrePost. The seat angle was fit to the curvature of our validated FE model, and in accordance with the angles described in SAE J826 H-point mannequins fitted with a Head Restraint Measuring Device and the Euro NCAP whiplash testing protocol.^17,18

• Initially, the seat and floor were fixed in all six degrees of freedom (x-, y-, z-translational and x-, y-, z-rotational) and positioned 1 mm below the THUMS model. Gravity was then applied for three seconds to relax the THUMS model into the seat in preparation for the ensuing whiplash exposure. Whiplash was again simulated as described above. The FE model with the added seat and floor is shown in Figure 1.

Fig. 1

Image of the validated FE model, with a rigid seat and floor shown using a semi-transparent filter.

Results

Representative snapshot images of the whiplash simulation are in Figure 2. Figure 3 compares the differences in resultant head acceleration between the T1 Acceleration Model and the Seat Interaction model. The T1 Acceleration Model showed a 10 g increase and 8 ms phase delay in peak head acceleration.

Fig.

Snapshots of the a) T1 Acceleration Model and b) Seat Interaction Model. At 0 ms in the T1 Acceleration Model, a 16 km/h pulse with a 10 g peak acceleration is applied leftward to all dummy nodes at or below the T1 vertebrae (grey). By 75 ms, most of the load has been applied. By 265 ms, all of the load has been applied and the dummy is moving forward at a constant velocity. At 0 ms in the Seat Interaction Model, the same pulse is applied leftward to the rigid seat (grey). At 75 ms, the seat is slightly past mid-impact with the dummy. By 265 ms, the impact is complete resulting in the dummy separating from the seat.

Fig. 3

Comparison of resultant head accelerations in two models loaded using the same Euro NCAP pulse. The T1 Acceleration Model experienced a 7.2 g peak head acceleration at 70 ms. The Seat Interaction Model experienced a 17.5 g peak head acceleration at 78 ms. A 5-point averaging filter was applied to the Seat Interaction Model curve to smooth out short-term fluctuations.

Discussion

A validated complete human FE model was updated to investigate the effects of cervical and lumbar fusions on peak ALL strains seen in the neck, under a realistic whiplash impact pulse with the addition of a rigid car seat and floor. Based upon the FE analysis, cervical fusion increases the risk of injury at specific levels that may not have pre-existing risk (i.e. C3-C4), but does not appear to dramatically increases peak ALL strains in portions of the spine that are already high risk (i.e. C6-C7). Additionally, while changes to motion at segments adjacent to a fusion can be measured, these changes are limited to the immediate segments (at most). Fusion at the thoracolumbar spine did not appear to affect the cervical spine in our model.

The inclusion of seat interaction and thoracolumbar coupled motions resulted in a 23% increase in the ALL strains (p < 0.0001), indicating that modelling the seat is important in injury prediction. There was no significant difference in the way the models assessed the changes in adjacent segment ALL strain following fusion, suggesting that the more computationally efficient T1 Acceleration Model may be adequate to quantify these effects.

Increase in ALL strain was seen following all cervical spine procedures. Statistically significant increases in ALL strain between one- and two-level fusions were seen in both the T1 Acceleration Model and Seat Interaction Model (p = 0.019 and p = 0.03, respectively) when analyzing the motion segments between C2-C6. C6-C7 showed high levels of ALL strain under all tested conditions including the baseline, non-operative condition. Our strain of 0.3 in the baseline model was close to the tolerance of the ALL according to Yoganandan,¹⁹ but well below the 0.8 from the dynamic tests of Bass.²⁰ The addition of a seat back increased ALL strain by 23.0% (p < 0.0001). However, the percent change in ALL strain following surgery remained consistent for all conditions (p = 0.76 for single-level; p = 0.48 for two-level fusion). While absolute strain values varied, the consistent percent change suggests that our model is robust in its assessment in change of adjacent segment ALL strain following cervical fusion under a broad range of conditions (i.e. with and without a seat back, and EURO-NCAP acceleration vs Yoganandan’s acceleration pulse). The increased strain in the whole-human test condition was likely a result of the additional shorter impulse delivered to the neck due to the decoupling of the body from the seat and the vertical acceleration component from straightening of the compliant lumbar and thoracic spines. Additionally, the presence of soft-tissue interactions surrounding the muscle elements may have further contributed to the larger strains to the cervical spine. Lumbar fusion had no meaningful effect on cervical strain (-1.0% change, p = 0.61).

Complete ALL failure has been documented to occur at threshold strains between 0.426 and 0.476,¹⁹ with partial injury occurring at strains above 0.222.¹⁴ Our results suggest partial injury during whiplash to the ALL at C6-C7 for a patient with a normal, unfused spine. Patients who have undergone fusion appear to have an increased risk of injury. In most of our simulations, one or more ALL segments would have reached partial failure thresholds following fusion.¹⁴

Lumbar fusion does not appear to have any impact in the cervical spine during whiplash. The claim that neck symptoms have been exacerbated by prior lumbar fusion in the absence of sagittal imbalance is not supported by this study.

Several important clinical conclusions can be made from this data. Modern reports of ASD indicate C4-C5 and C6-C7 as the levels most affected by ASD.²¹ Plate distance from the vertebral endplate has also been shown to affect adjacent segment ossification.²² Our data shows that C6-C7 experiences the highest strain at baseline, and that single-level cervical fusion did not appear to have meaningful difference in biomechanical strain. In contrast, C3-C4, which under normal conditions experiences lower ALL strain and is infrequently a treated motion segment in isolation, experienced the largest increases in cervical ALL strain following arthrodesis in comparison with the other motion segments. Thus, our data is in agreement with contemporary thinking that ASD reflects both a natural history of spondylosis (C6-C7) as well as biomechanical factors (C3-C4). Perhaps motion-sparing surgery at C4-C5 will have advantages over fusion when considering adjacent disease at C3-C4.

Several limitations exist for future consideration but do not affect the conclusions drawn above. Our material properties were isotropic and did not account for strain rate variability. Although our FE model is validated for the acceleration rate simulated, the correlation between cadaveric and FE simulations may be different at different accelerations. Additionally, in our FE model, the ligaments were rigidly attached to vertebral bodies; there may be high strains at these attachments clinically. In real-world vehicle collisions, muscles contract either in anticipation or as an instantaneous response to the impact, which may exacerbate or alleviate these strains depending on the subject-dependent muscle activations.^23,24 Although a realistic whiplash impulse was used, our data cannot be applied directly in a forensic situation, as our seat is rigid and based on the geometry in the SAE J826 standard. Our seat did not include a headrest or incorporate whiplash prevention and mitigation technologies currently found in most vehicles. Additionally, the aetiology of whiplash syndrome may involve the facet capsules.^25,26 Unfortunately, the capsules are not modelled with sufficient fidelity in our model for a confident strain analysis. However, these limitations do not change the points made in this study. Finally, our computational model only represents at 50 percentile American male. Changes in height or weight may result in absolute differences in ALL strain. However, the conclusions drawn – specifically that peak ALL strain increases in the cervical spine adjacent to a fusion, and that the increase in strain disproportionately affects different motion segments – is unlikely to change when differently sized models are used. Importantly, the use of a standardised model (‘AM50’) allows better comparison with available literature and existing mechanical and cadaveric models.

In conclusion, this study quantifies the effects of whiplash neck injury in scenarios wherein the occupant has previously undergone cervical or lumbar arthrodesis. Peak ALL strains were found to be higher in the motion segments adjacent to the level of fusion compared with a healthy spine, and strains directly increased with longer fusions. C3-C4 was the motion segment at greatest biomechanical risk for adjacent segment strain following arthrodesis, while C6-C7 showed high strain levels regardless of arthrodesis status. Lumbar fusion had no statistical or clinically meaningful effect on ALL strains in the cervical spine. This biomechanical study suggests that the debate about whether adjacent segment disease is the effect of natural history or the effect of couple motions may require investigation of specific motion segments.