1466. The Association of Lumbar Spondylolisthesis and MRI Endplate Changes
Authors * Denotes Presenting Author
  1. Azar Dagher *; University of Maryland
Thompson et al[1] introduced a work showing the significance of endplate signal changes in predicting provocative discography. More recent analysis of this data base showed how the associations vary per level [2,3] and level-specific multivariate analysis showed that edema and fatty changes (Modic 1 and 2 respectively) are contributors to provocative discography only at L4-5. L4-5 has been consistently shown in many works to be the level most prone to spondylolisthesis[4,5]. Smaller population studies have linked Modic changes with translational instability [6-9] while others have not [10]. The purpose of this study is to test the statistical associations of spondylolisthesis with Modic changes and whether spondylolisthesis is also significant for other MRI predictors of discogenic pain using the large data base from the above studies [1-3].

Materials and Methods:
The large Health Insurance Privacy And Portability Act-compliant retrospective database of lumbar MRI with discography (736 patients, 2457 disc levels) has been previously described[1-3]. Institutional Review Board exemption was granted for this new analysis. Relevant variables for this study were tears in the posterior annulus (High Intensity Zone-HIZ), disc herniation of any type, Modic endplate changes of any type (1-edema, 2 fatty change, 3-sclerosis) and spondylolisthesis of any grade. Statistical analysis used a 2 x 2 contingency table of the presence and absence of spondylolisthesis against the presence or absence of each of the MRI variables at the same level. Included was the pairing of spondylolisthesis with Modic change at any lumbar level to potentially reinforce the association of endplate signal changes with generalized lumbar instability, not necessarily at the same level. The p-value, sensitivity, specificity, positive predictive value (PV+), and negative predictive value (PV-) were calculated for each pairing. 95% confidence intervals were calculated using a binomial expansion. A p value < 0.05 was used for significance.

Only Modic changes have a significant association with spondylolisthesis: p = 6.97E-4, sensitivity = 0.10, specificity = 0.95, PV+ = 0.21, PV- = 0.88. With spondylolisthesis at any lumbar level, there is even higher level of significance: p = 2.37E-21, sensitivity = 0.18, specificity = 0.96, PV+ = 0.38, PV- = 0.89. The associations of disc tears and herniations with spondylolisthesis were not significant.

Spondylolisthesis is highly specific and mildly sensitive for Modic changes in the lumbar spine with a high degree of significance, unlike the associations with advanced degenerative disc disease. In other terms, the significance of normal anterior-posterior alignment (such as on an x-ray) is that Modic change will likely be absent (high PV-). If one were to take spondylolisthesis as a marker of instability in the anterior-posterior axis, then one can conclude that shear forces in this direction are a factor in producing endplate signal changes.