Oral Presentation 51st International Society for the Study of the Lumbar Spine Annual Meeting 2025

Effective disc age: a statistical model for age-dependent and level-specific lumbar disc degeneration using magnetic resonance imaging (MRI) (115448)

Harrah R Newman 1 , John M Peloquin 1 , Kyle D Meadows 1 , Barry A Bodt 1 , Edward J Vresilovic 1 , Dawn M Elliott 1
  1. University of Delaware, Newark, DE, United States

INTRODUCTION: Disc degeneration is the structural and compositional deterioration of disc, with associated changes in mechanical properties and function. [1] From magnetic resonance imaging (MRI), disc degeneration is often assessed by grading schemes based on qualitative assessment of disc height and regional signal intensity. [2], [3] The current disc grading schemes are subjective, qualitative, ordinal and do not consider subject age or disc level, despite disc degeneration dependence on both (Fig A). These limitations inhibit differentiation between discs with normal aging (non-pathological) and discs with accelerated degeneration (which may be pathological). We sought to develop a statistical model of age- and level-dependent disc degeneration.

METHODS: With IRB approval, 84 human subjects (n=7 per sex per age decade (18-29, 30-39, 40-49, 50-59, 60-69, 70-83 years old)), underwent supine MRI. [4] Disc geometry was quantified, including: mid-sagittal height, anterior-posterior width, mid-sagittal area, volume, wedge angle, anterior and posterior bulges. T2 relaxation times were measured in the nucleus and across anatomic regions. [5] Subject traits of height, weight, and sex were also considered. Stepwise linear regression was run on these 16 potential predictors with the subject’s true age as the response variable. Standard least squares model fit was used to estimate effective disc age with 1-, 2-, 4-, 6-, 8-, and 16- predictor models. The same predictor(s) were used each disc-level model. Models were evaluated by comparing the subjects’ true age to effective disc age by goodness of fit (R2) and root mean square error (RMSE).

RESULTS: The best predictors were determined by stepwise regression (Fig B). Representative fits for the 1-predictor model (Fig C) shows increasing data scatter at inferior levels. Overall, L1-L4 had the best fits, followed by the L4-L5 (Fig D-F). As expected, fits generally improved with more model predictors, however minimal improvements occurred from 4 to 6 to 8 predictors (Fig D-F). The RMSE accuracy was ~10 years across levels, with little improvement after 4 predictors (Fig F). Level L5-S1 had poorest fit and accuracy. The 4-predictor model was further investigated. In this model, the nucleus T2 time was the most significant predictor for the upper disc levels, L1-L3. The nucleus T2 time and disc volume both contributed significantly for L3-L5. The disc size was the most significant predictor for L5-S1, and this was the only level where geometry was more important than T2 time.

DISCUSSION: We recommend proceeding with the 4-predictor model to quantify effective disc age. The model requires 1) the nucleus T2 relaxation time, 2) subject height, 3) disc mid-sagittal area, and 4) disc volume. The model provides an excellent effective age calculation (R2 > 0.6) for disc levels L1-L5 and a moderate effective age calculation (R2 = 0.43) for the L5-S1. We offer a data-driven, quantitative tool to distinguish normal aging from all other forms of accelerated disc degeneration. This effective age model allows for future research to focus on discs with accelerated degeneration, and is recommended for use in clinical and research applications.

6734ba1320797-ISSLS2025_HRN_Fig.png

  1. [1] M. A. Adams and P. J. Roughley, “What is intervertebral disc degeneration, and what causes it?,” Spine (Phila. Pa. 1976)., vol. 31, no. 18, pp. 2151–2161, 2006, doi: 10.1097/01.brs.0000231761.73859.2c.
  2. [2] C. Pfirrmann, A. Metzdorf, M. Zanetti, J. Hodler, and N. Boos, “Magnetic Resonance Classification of Lumbar Intervertebral Disc Degeneration,” Spine (Phila. Pa. 1976)., vol. 26, no. 17, pp. 1873–1878, 2001.
  3. [3] J. F. Griffith et al., “Modified Pfirrmann grading system for lumbar intervertebral disc degeneration,” Spine (Phila. Pa. 1976)., vol. 32, no. 24, pp. 708–712, 2007, doi: 10.1097/BRS.0b013e31815a59a0.
  4. [4] K. D. Meadows, J. M. Peloquin, H. R. Newman, P. J. K. Cauchy, E. J. Vresilovic, and D. M. Elliott, “MRI-based measurement of in vivo disc mechanics in a young population due to flexion, extension, and diurnal loading,” JOR Spine, no. September 2022, pp. 1–14, 2023, doi: 10.1002/jsp2.1243.
  5. [5] K. D. Meadows, C. L. Johnson, J. M. Peloquin, R. G. Spencer, E. J. Vresilovic, and D. M. Elliott, “Impact of pulse sequence, analysis method, and signal to noise ratio on the accuracy of intervertebral disc T2 measurement,” JOR Spine, vol. 3, no. 3, pp. 1–12, 2020, doi: 10.1002/jsp2.1102.