Prospective analysis of imaging prediction of pseudarthrosis after anterior cervical discectomy and fusion: computed tomography versus flexion-extension motion analysis with intraoperative correlation.

STUDY
DESIGN: A prospective comparison of quantitative motion analyzed (QMA) flexion/extension radiographs versus computed tomography (CT) as an analytical predictor of cervical pseudarthrosis. Intraoperative confirmation of the fusion was performed.
OBJECTIVE: To prospectively compare motion analyzed flexion/extension radiographs to CT to predict pseudarthroses. Define motion thresholds on flexion/extension radiographs to define pseudarthroses. SUMMARY OF BACKGROUND DATA: Assessment of postoperative fusion success is an important factor in assessing success after anterior cervical spine fusion. Gross intervertebral motion can be used as a measure; however, the current "gold standard" for determining fusion status is a CT to assess bridging bone. Defining the amount of intervertebral motion at the fusion site has been previously addressed and definitions have varied widely.
METHODS: Data were analyzed at 47 fusion segments. Intervertebral motion at the fusion site was measured from flexion/extension radiographs taken at least 1 year after the cervical spine fusion. Motion was quantified from digitized radiographs by an independent researcher using proprietary quantitative motion analysis (QMA) software. CT scans on all patients were analyzed for fusion status by a neuroradiologist. Those patients determined to have a symptomatic pseudarthrosis were revised and intraoperative motion at the facet joints was documented. Correlation between intraoperative findings, CT and QMA was performed.
RESULTS: Using greater than 4° of measured motion on flexion/extension radiographs resulted in a Spearman correlation P-value of 0.096 (95% confidence interval: -0.06 to 0.66). Using greater than 1° of motion, the Spearman correlation P < 0.0001 (95% CI: 0.54-0.90). The positive predictive value (PPV) using 4° of motion as the criterion was 100%, indicating a high specificity. The negative predictive value (NPV) was 52%, indicating a low sensitivity. Using greater than 1° of motion, the PPV was 100% and the NPV was 73%. Findings from CT showed an identical PPV and NPV to assessments made using greater than 1° of rotation. Specificity and positive predictive value were 100% for all criteria. Using a lack of bridging on CT or more than 1° of intervertebral motion during flexion/extension increased the sensitivity to 85% and the negative predictive value to 85%.
CONCLUSION: A threshold level of 4° of motion is commonly used to identify a pseudarthrosis. Our prospective study suggests that this value has a high PPV, but a low specificity and would miss many of the pseudarthroses that have angular motion less than 4° (sensitivity 23%). By lowering the threshold for angular motion to 1°, the sensitivity improves to 77%. CT scan has been touted as the gold standard, and it has a high positive predictive value of 100%. However, its NPV was slightly lower than using 1° of motion on QMA analyzed flexion-extension films (73% vs. 79%). In conclusion, although CT scan has long been regarded as the gold standard for determining a pseudarthrosis in the cervical spine, the interpretation is subjective and vulnerable to both type I and type II errors. Analysis of motion using Quantitative Motion Analysis is seemingly less subjective than CT and in our prospective study was more predictive of an operatively confirmed pseudarthrosis.