Degeneration of intervertebral discs(IVDs) has been implicated as a major etiologic component of lower back pain¹. Morphological magnetic resonance(MR) imaging allows a grading of degeneration as assessed by the widely accepted Pfirrmann scale². Diffusion-weighted imaging(DWI) has been applied for the quantification of IVD degeneration as the apparent diffusion coefficient(ADC)³, and is expected to reflect microstructural changes such as matrix composition(water, proteoglycan, and collagen) and matrix integrity. Conventional DWI analysis is based on an assumption that the water molecules follow a Gaussian distribution. However, human tissue including the IVD is a complex and restricted environment that hinders the distribution of water molecules, resulting in distributions that are far from Gaussian⁴. Q-space imaging(QSI) analysis does not assume a Gaussian shape for the underlying probability density function of water molecule diffusion⁵. QSI has shown promise for evaluating the microstructure of tissues in vivo because it can provide additional diffusion metrics, namely the root mean square displacement(RMSD) and apparent kurtosis coefficient(AKC)⁶, which give in vivo microstructural information that complements the ADC values. We therefore hypothesized that QSI analysis would be able to provide more information about degenerative changes in the microstructural complexity in IVDs.The purpose of the current study is to evaluate the correlation between the quantitative measurements obtained from QSI analyses and the IVD degeneration grades in patients suffering from lower back pain.

Thirty patients(mean age, 57.4 years) suffering from lower back pain were prospectively recruited in our institution. Prior to the study, the research protocol was approved by the institutional review board, and informed consent was obtained from each patient. Patients with history of spine surgery, major systemic disease(e.g. autoimmune disease), serious illness(e.g. tumor, infection), or spinal fractures, were excluded. Images were acquired using 3-Tesla MR(Signa HDx; General Electric). After routine fast spin-echo(FSE) T2-weighted sagittal and axial imaging, QSI data were acquired in the axial plane of the IVD between the fourth and fifth lumbar vertebrae. FSE T2-weighted images were not only used for anatomical reference but for the visual Pfirrmann grading of IVDs². QSI was performed in accordance with previously published literature⁷ by using a spin-echo diffusion weighted echo-planar imaging sequence（Figure 1）. QSI analyses were performed using the free software Volume-One 1.72(http://www.volume-one.org/) and dTV.II.FZR(http://www.ut-radiology.umin.jp/people/masutani/dTV.htm). Five equally sized circular regions-of-interest(ROIs) on the central slice of the axial plane(Figure 2) were manually drawn to measure RMSD and AKC values. The most anterior and most posterior ROIs(ROI 1 and ROI 5) were interpreted as anterior and posterior annulus fibrosus(AF), respectively. The ROIs in between(ROIs 2-4) were interpreted to represent the nucleus pulposus(NP). Spearman's rank correlation coefficient was used to measure the association between the qualitative clinical grading of disc degeneration(Pfirrmann scale) and the quantitative diffusion values(RMSD and AKC). A P value less than 0.05 was considered to be significant.

There were 2 discs(10.0%) classified as Pfirrmann grade 1, 3(10.0%) as grade 2, 10(33.3%) as grade 3, 10(30.0%) as grade 4, and 5(16.7%) as grade 5. Histograms of the RMSD and AKC values within each ROI are shown in Figure 3. RMSD values in the NP regions of IVDs were found to decrease with the increasing Pfirrmann grades, while AKC values were found to increase. The results of correlation analysis are shown in Figures 4 and 5. For the ROIs 2-4(representing NP), there were significant negative correlations between the Pfirrmann grades and the RMSD values(Spearman's ρ= -0.58, -0.75, and -0.74 for ROI 2, ROI 3, and ROI4, respectively; all P<0.01), while there were significant positive correlations between the Pfirrmann grades and the AKC values(ρ= 0.61, 0.72, and 0.70; all P<0.01). No significant correlations could be assessed for ROIs 1 and 5(representing AF) between the Pfirrmann grades and the RMSD and AKC values.

Water molecule diffusion in the IVDs is restricted in a complex manner by several factors such as extracellular matrix(e.g. collagen fibers and proteoglycan). In general, RMSD is not influenced by the viscosity of water, but by the space for free water movement⁵. AKC describes the deviation of the water diffusion pattern within a voxel from a Gaussian distribution, which is thought to reflect the changes in microstructural complexity⁸. Our results suggest that the degenerative process of IVDs involves narrowing of the space for free water movement and a generally higher degree of microstructural complexity.The RMSD and AKC values obtained from QSI analyses may be sensitive biomarkers for IVD degenerative microstructural changes in which we are unable to assess with conventional diffusion-weighted imaging metrics based on an assumption of a Gaussian shape and model of water molecules.