A statistical model for intervertebral disc degeneration: determination of the optimal T2 cut-off values

The aim of this study was to investigate the possibility of quantitative classification in intervertebral disc degeneration using spin-spin relaxation time (T2) cut-off values with regard to morphological classifications.

Lumbar magnetic resonance (MR) imaging was performed on 21 subjects (a total of 104 lumbar disks). The T2 relaxation time was measured in the nucleus pulposus using a sagittal multi-echo spin-echo sequence. The morphological classification of disc degeneration was assessed independently by three experienced neuroradiologists according to the Pfirrmann and Schneiderman classifications. Receiver operating characteristic analysis was performed among grades to determine T2 cut-off values in each classification. Intra- and interobserver differences were calculated using kappa statistics.

Moderate overall interobserver agreement was found between observers in both the Pfirrmann and Schneiderman classification schemes (kappa 0.46 and 0.51), while intraobserver reliability was substantial to almost perfect. The interobserver reliability was only fair in Pfirrmann grades III and IV (kappa 0.33 and 0.36), but the T2 cut-off values still indicated a significant difference between grades (p<0.05).

Interobserver agreement of MR evaluation in patients with intervertebral disc degeneration was only fair to moderate on the classification of more severe disc degeneration in the Pfirrmann and Schneiderman schemes. Based on our results, quantitative T2 cut-off values seem to be a more reliable method to define the degree of disc degeneration, which may help staging intervertebral disc degeneration (IVDD) even if the interobserver reliability is low.