Abstract
Introduction: It has long been recognised that static plain x-rays are a sub-optimal method for the assessment of lumbar fusion. Blumenthal and Gil showed that radiographic assessment of fusion corresponded with operative findings only 69% of the time. Santos et al suggest that both plain x-rays and flexion/extension x-rays overestimate the fusion rate when compared to helical computed tomography (CT). To date there has been no correlation of CT assessment of fusion with surgical exploration. In this study we present an animal model of lumbar spine pseudarthrosis and compare three imaging modalities with micro-cut CT scanning and cadaveric assessment.
Methods: Approval was gained from the QUT animal ethics committee. Eleven mixed bred ewes were assigned to either a fusion group or an intentional pseudarthrosis (IP) group. A dorsal approach to the facet joints of L2/3 was made. The facet joints were destabilised by resecting the articulating surfaces with a rongeur. In the fusion group, the spinous processes of the destabilised segment were wired tightly together and a bone graft harvested from the iliac crest was placed into the joint space. In the IP group the bone graft bed was prepared similarly except that a small proportion of the articulating surface was left intact and a 1.0 cm2 roll of oxidised cellulose was placed into the facet joint space bilaterally. In the IP group the spinous processes were wired around an interspinous spacer which was later removed to create a similar degree of laxity in the fixation of each of the IP specimens. The animals were sacrificed at 6 months and static and dynamic lateral radiographs obtained. The spine was removed en bloc, and high speed fine cut (2mm) CT Scanning performed. The specimens were individually assessed for fusion by micro-cut CT scanning. Eight independent, blinded orthopaedic surgeons, were asked whether they considered the spine to be fused based on
-
plain x-ray
-
plain xray and flexion/extension x-rays and
-
plain x-ray and flexion/extension x-rays and CT scan.
These results were correlated with a fusion rate based on the micro CT. The specificity and sensitivity of these radiological measures in diagnosing pseudarthrosis and inter-rater reliability using Fleiss’ Kappa scores for each method were calculated.
Results: For assessing pseudarthrosis identified by microCT the plain film sensitivity was 0.41 and the specificity was 0.47. For assessing pseudarthrosis with plain and flexion extension xrays the sensitivity was 0.55 and the specificity was 0.33. For assessing pseudarthrosis with plain flexion extension xrays and CT the sensitivity was 0.81 and the specificity was 0.88. The Kappa score for plain films was 0.15, for flexion extension was 0.07 and CT was 0.54.
Discussion: This study suggests that plain radiographs and flexion extension radiographs are an unreliable measure of posterior lumbar fusion. The current clinical gold standard for assessment of fusion (CT) was able to correctly identify non-union in 80% of cases. Whilst no alternatives to structural assessment of the fusion mass with CT currently exist it is important to recognise the limitations of this technique.
Correspondence should be addressed to Dr Owen Williamson, Editorial Secretary, Spine Society of Australia, 25 Erin Street, Richmond, Victoria 3121, Australia.
