Disc Degeneration Assessed by Quantitative T2* (T2 star) Correlated with Functional Lumbar Mechanics

Abstract

Study Design

Experimental correlation study design to quantify features of disc health, including signal intensity and distinction between the annulus fibrosus (AF) and nucleus pulposus (NP), with T2* magnetic resonance imaging (MRI) and correlate with the functional mechanics in corresponding motion segments.

Objective

Establish the relationship between disc health assessed by quantitative T2* MRI and functional lumbar mechanics.

Summary of Background Data

Degeneration leads to altered biochemistry in the disc, affecting the mechanical competence. Clinical routine MRI sequences are not adequate in detecting early changes in degeneration and fails to correlate with pain or improve patient stratification. Quantitative T2* relaxation time mapping probes biochemical features and may offer more sensitivity in assessing disc degeneration.

Methods

Cadaveric lumbar spines were imaged using quantitative T2* mapping, as well as conventional T2-weighted MRI sequences. Discs were graded by the Pfirrmann scale and features of disc health, including signal intensity (T2* Intensity Area) and distinction between the AF and NP (Transition Zone Slope), were quantified by T2*. Each motion segment was subjected to pure moment bending to determine range of motion (ROM), neutral zone (NZ), and bending stiffness.

Results

T2* Intensity Area and Transition Zone Slope were significantly correlated with flexion ROM (p=0.015; p=0.002), ratio of NZ/ROM (p=0.010; p=0.028), and stiffness (p=0.044; p=0.026), as well as lateral bending NZ/ROM (p=0.005; p=0.010) and stiffness (p=0.022; p=0.029). T2* Intensity Area was also correlated with LB ROM (p=0.023). Pfirrmann grade was only correlated with lateral bending NZ/ROM (p=0.001) and stiffness (p=0.007).

Conclusions

T2* mapping is a sensitive quantitative method capable of detecting changes associated with disc degeneration. Features of disc health quantified with T2* predicted altered functional mechanics of the lumbar spine better than traditional Pfirrmann grading. This new methodology and analysis technique may enhance the assessment of degeneration and enable greater patient stratification for therapeutic strategies.

Introduction

Intervertebral disc (IVD) degeneration is a primary cause for low back pain, one of the most prevalent musculoskeletal impairments in the U.S. affecting an estimated 70–85% of the population at some point in their life¹. Currently, assessment of patients with low back pain often includes diagnostic imaging of the intervertebral discs to identify their morphological health. This data is complimented with a clinical evaluation, aimed at assessing the functional biomechanics of lumbar spine motion and the presentation of pain. Coupling these data provides a picture of an individual’s overall spinal health. Unfortunately, this often results in contradictory or inconclusive outcomes in spite of the phenomenological link between the intervertebral disc’s constituent makeup and mechanics. The ability to discern differences in biochemistry, structure, and mechanics, the likely nexus for this challenge, has become the focus for both imaging and biomechanics studies.

Conventional T2 weighted sagittal magnetic resonance imaging (MRI) sequences have been utilized to create a subjective grading scale for disc health based on morphological features^2,3. The grading system is applied in the clinical decision making process and yet lacks specificity^4,5. The Pfirrmann grading system with a score from 1–5 is based upon the MRI signal intensity, clarity of the transition zone between the annulus fibrosus (AF) and nucleus pulposus (NP), and disc height². This subjective scoring system largely fails to correlate well with pain or provide clinically useful patient stratification, especially in detecting early signs of degeneration^6–9. However, the underlying processes of the Pfirrmann grading system appear to be robust and quantification of these features may advance disc health assessment.

Recently, new quantitative MR imaging techniques have been developed including T1ρ, T2 mapping, chemical exchange saturation transfer (CEST), Magnetic Transfer Ratio, and T2* mapping which may offer promise in more accurately discriminating between patient discs of different health^10–16. Many of these techniques utilize the physics of water molecule relaxation within a magnetic field to produce images, which highlight the presence of proteoglycans (T1rho), hydration level (T2), and structure of the macromolecule matrix along with hydration level (T2*). Quantitative T2* (T2 star) mapping is an emerging technique with the added benefit of a short acquisition time, high signal-to-noise ratio, and three-dimensionality over traditional T2 mapping^11,17,18. T2* has been shown to probe biochemical properties of the tissues and has been beneficial in cartilage^{10,11,17–24}. Specifically, T2* relaxation times provide information about spatial macromolecule architecture in conjunction with water molecule mobility^11,18. The relationship between T2* value and histological grades of degeneration has been established in the hip joint, where a decrease in T2* relaxation time was significantly correlated with a higher degree of cartilage degeneration¹⁰.

To the authors knowledge, only two studies have been published which apply the T2* technique to the intervertebral disc^17,25. Hoppe et al. proposed a technique for assessment of disc degeneration using axial T2* maps based on Watanabe et al. classification and found a significant correlation between T2* parameters and Pfirrmann grades^25,26. Welsch et al. correlated quantitative T2 and T2* relaxation times with Pfirrmann Score in patients suffering from low back pain¹⁷. They reported both techniques correlated strongly with Pfirrmann Score regarding the nucleus pulposus, but T2* alone had the ability to identify changes in the posterior annulus fibrosus. They also performed a parametric analysis between T2* relaxation times at different regions of interest (ROI) and reported the highest spatial variation from AF to NP in Pfirrmann grade 1, and a decline at higher grades.

These T2* seminal works provide the basis for our investigation which aims to utilize the quantitative nature of the imaging technique and distinguish critical features of the Pfirrmann grading system. The features of the Pfirrmann grading system include disc signal intensity and clarity of the transition zone between the AF and NP. The interrogation of these features using the T2* images will enable the quantitative grading of disc degeneration without subjectivity or bias but with clinically recognized features of distinction. These efforts utilized ROI data alone to quantify disc health despite knowing that specific features of classification systems enable those systems to predict morphological changes better than others.

The ability of T2* mapping to detect changes related to the intrinsic biochemical properties has a potentially profound impact on diagnostic imaging of the intervertebral disc. Throughout the degeneration process there is a loss of proteoglycans and a change in the anabolic/catabolic balance of collagen homeostasis and its architecture. The alterations of the biochemical content and its structural integrity change the mechanical properties of the degenerating disc impacting the global mechanics of the spine^{27–29,30–36}. Biomechanically, these changes lead to altered spinal stability and potentially cord occlusion, neural compressive lesions or nerve root pinching, causing pain.

Therefore, the purpose of this study was to create T2* maps of IVDs throughout the degenerative spectrum, quantify parameters of the Pfirrmann grading scale, and correlate these imaging parameters with mechanics outcomes of functional spinal units in an attempt to probe the relationship between structural constituent properties and functional mechanical parameters.

Methods

Eighteen osteoligamentous lumbar spines (L3-Sacrum), acquired from our University Bequest Program, (age: 53.2±15.5 years; range: 21–71 years), were examined using conventional and quantitative magnetic resonance imaging protocols and biomechanically exercised in flexion/extension, lateral bending, and axial rotation. A correlation study design was utilized to examine the relationship between the intervertebral disc morphology, based on MR imaging, and functional biomechanics of the lumbar spine.

Classic Disc Grading

MR imaging was performed on a Siemens 3T scanner (Magnetom Trio; Siemens Healthcare) on each IVD (L3-L4, L4-L5, L5-S1), 54 in total. Conventional T2-weighted sagittal images were acquired for Pfirrmann disc grading (Figure 1A). Each disc was graded via the Pfirrmann scale independently by seven spine surgeons (DWP) and three experienced spine researchers (AME, DJN, HM). These scores were then averaged and rounded to the nearest integer.

Figure 1. Magnetic Resonance Images of Lumbar Spine and Intervertebral Disc Using T2 and Quantitative T2* Techniques.

Left: T2* Map of Lumbar Spine. Top Right: Healthy Lumbar Intervertebral disc imaged with classical T2 MRI. Bottom Right: Same disc using T2* map showcasing the five regions of interest from anterior to posterior (1 to 5).

T2* Disc Health Quantification

T2* relaxation maps were obtained [TR(ms): 500; TE(ms): 4.18, 11.32, 18.46, 25.60, 32.74, 39.88; Voxel Size(mm): 0.5×0.5×3.0, Slices: 33]. Quantitative T2* maps were calculated (MapIt, Siemens Healthcare) and mean T2* values were recorded using the open-source Osirix Imaging Software across five mid-sagittal plane regions of interest (ROI) from anterior (ROI 1) to posterior (ROI 5) (Figure 1C). Further analysis utilizing MATLAB (Mathworks Inc. Natick, MA), evaluated the central ROI (ROI 3) mean relaxation value in all sagittal slices across the coronal plane as the intensity changed from lateral to medial to lateral. Using a least squares method, the slope of the transition zone (ms/mm) between the annulus fibrosus and nucleus pulposus on each side was linearly regressed and then averaged. This metric quantified the distinction between the NP and AF. To quantify the signal intensity of the disc, the area underneath the curve was calculated by the trapezoidal integration method and normalized to the width of the IVD, measured by MRI (Figure 2).

Figure 2. Quantification of Imaging Parameters, Transition Zone Slope and T2* Intensity Area, from T2* Relaxation Profile in the Coronal Plane of ROI 3.

Top Left: Coronal Profile of Healthy Disc with Plot of Average T2* Relaxation Time (ms) overlaid. Top Right: Corresponding quantification of Transition Zone Slope and T2* Intensity Area of the healthy disc. Bottom Left: Coronal Profile of Degenerated Disc with Plot of Average T2* Relaxation Time (ms) overlaid. Bottom Right: Corresponding quantification of Transition Zone Slope and T2* Intensity Area of degenerated disc.

Biomechanical Testing

After imaging, the specimens were embedded in polymethylmethacrylate and tested in a six-axis Spine Kinetic Simulator (8821 Biopuls, Instron, Norwood, MA)³⁷. Pure moments of 7 Nm with no axial preload (0 N) were applied with the spine unconstrained to displace inferiorly; this methodology is utilized in numerous studies and was also the initial basis for the ASTM F2423 standard, although it now includes an axial preload^37–40. Input moments and forces were controlled by a 6 DOF load cell (AMTI M4380, Watertown, MA) in three cardinal planes. Inferiorly, movements were unconstrained using a passive X-Y table (resistance <0.1N) to minimize shear forces.

Three cycles of was performed sinusoidally (0.015 Hz) and the final cycle was used for analysis. Load and moment data were collected on two six-axis load cells (AMTI M4380, Watertown, MA) at 100 Hz on a data acquisition board connected to personal computer. Intersegmental angular displacements were recorded at 100 Hz using a 3D visual motion analysis 5-camera system (Vicon MX-F40NIR, Vicon Motion Systems, Centennial, CO) capturing a four ball infrared reflecting marker set attached to each vertebral body. Range of motion (ROM), neutral zone (NZ), ratio of NZ/ROM, and bending stiffness were quantified for each functional spinal units in all bending directions³⁹. Figure 3 displays these metrics over top a representative motion profile of a healthy functional unit.

Figure 3. Representative Motion Profile.

A healthy functional unit motion profile with biomechanical metrics graphically defined. These include range of motion (ROM), neutral zone (NZ), and bending stiffness.

Statistical Treatment

The mechanical outcomes were compared with both T2* measurements and Pfirrmann grade using Pearson’s Correlation Tests (α=0.05). Furthermore, each T2* measurement was also correlated with respect to Pfirrmann grade (α=0.05). One-way single measure intra-class correlation (ICC) was calculated to determine the intra-observer variability in assessing Pfirrmann grades to the discs.

Results

All specimens were imaged and received the full battery of biomechanical testing to reveal moment-angular displacement curves for flexion/extension (FE), lateral bending (LB), and axial rotation (AR). One Pfirrmann grade 5 disc was unable to be quantified using the T2* technique due to the cadaveric nature of the tissues and this disc’s complete collapse.

The graders of disc health had an intra-class correlation coefficient value of 0.686. Pfirrmann grade was significantly correlated with ROI’s 2–5 from T2* maps, as well as the transition zone slope and T2* intensity area (Table 1).

Table 1.

Correlations between Pfirrmann Grade and T2* imaging parameters

	Pfirrmann Grade
	r	p
Transition Zone Slope	−0.604	<0.001*
T2* Intensity Area	−0.774	<0.001*
ROI 1	0.001	0.996
ROI 2	−0.698	<0.001*
ROI 3	−0.821	<0.001*
ROI 4	−0.737	<0.001*
ROI 5	−0.381	<0.001*

(r: Pearson’s correlation coefficient; p: p-value * represents significance at p=0.05.).

Range of motion measurements and stiffness correlations with imaging parameters are displayed in Tables 2 and 3, respectively. Pfirrmann grade was significantly correlated with only LB NZ/ROM (p=0.001) and stiffness (p=0.007). T2* Intensity Area and the Transition Zone Slope were also significantly correlated with LB NZ/ROM (p=0.005; p=0.010) and stiffness (p=0.022; p=0.029). T2* Intensity Area was also correlated with LB ROM (p=0.023). Select correlational plots for lateral bending are displayed in Figure 4. In addition, T2* Intensity Area and the Transition Zone Slope was significantly correlated with Flexion ROM (p=0.015; p=0.002), NZ/ROM (p=0.010; p=0.028), and stiffness (p=0.044; p=0.026). Several regions of interest also correlated with these variables as well as AR ROM (p=0.034) and AR stiffness (p=0.006).

Table 2.

Correlations between T2* parameters and range of motion and ratio neutral zone to range of motion

			Transition Zone Slope	T2* Intensity Area	ROI 1	ROI 2	ROI 3	ROI 4	ROI 5	Pfirrmann Grade
Flexion	ROM	r	0.414	0.334	0.118	0.322	0.311	0.353	0.090	−0.220
		p	0.002*	0.015*	0.398	0.019*	0.024*	0.01*	0.521	0.109
	NZ/ROM	r	−0.352	−0.305	.084	−0.187	−0.323	−0.293	−0.114	0.136
		p	0.010*	0.028*	.554	0.185	0.019*	0.035*	0.422	0.332
Lateral Bending	ROM	r	0.179	0.312	0.164	0.206	0.262	0.188	0.335	−0.223
		p	0.199	0.023*	0.241	0.139	0.058	0.177	0.014*	0.105
	NZ/ROM	r	−0.402	−0.378	−0.346	−0.363	−0.406	−0.399	−0.258	0.468
		p	0.003*	0.005*	0.011*	0.007*	0.003*	0.003*	0.062	0.001*
Axial Rotation	ROM	r	−0.143	−0.148	−0.061	−0.292	−0.130	−0.191	0.027	0.193
		p	0.307	0.290	0.667	0.034*	0.355	0.171	0.848	0.162
	NZ/ROM	r	0.136	0.036	0.128	0.151	0.055	0.099	−0.063	0.071
		p	0.332	0.799	0.361	0.280	0.696	0.481	0.653	0.608

(r: Pearson’s correlation coefficient; p: p-value * represents significance at p=0.05.)

Table 3.

Correlations between T2* parameters and stiffness measurements

			Transition Zone Slope	T2* Intensity Area	ROI 1	ROI 2	ROI 3	ROI 4	ROI 5	Pfirrmann Grade
Stiffness (deg/Nm)	Flexion	r	−0.306	−0.278	0.105	−0.079	−0.308	−0.163	−0.228	0.230
		p	0.026*	0.044*	0.454	0.574	0.025*	0.244	0.101	0.094
	Extension	r	0.002	0.191	0.106	0.249	0.126	0.172	−0.021	−0.031
		p	0.988	0.170	0.451	0.072	0.368	0.217	0.879	0.822
	Lateral Bending	r	−0.300	−0.314	−0.269	−0.292	−0.341	−0.197	−0.186	0.362
		p	0.029*	0.022*	0.052	0.034*	0.013*	0.157	0.182	0.007*
	Axial Rotation	r	0.276	0.241	0.186	0.270	0.233	0.371	0.229	−0.208
		p	0.043*	0.082	0.182	0.051	0.093	0.006*	0.099	0.132

(r: Pearson’s correlation coefficient; p: p-value * represents significance at p=0.05.)

Figure 4. Representative Correlational Plots.

Top: Lateral Bending Stiffness and Bottom: Lateral Bending NZ/ROM correlated with ROI 3 Average T2* Value (ms), Transition Zone Slope (ms/mm), and T2* Intensity Area (ms). r: Pearson’s correlational coefficient and p: p-value are displayed on corresponding plot.

Discussion

Quantitative T2* MRI has been shown to be beneficial in cartilage imaging by probing the biochemical content of the tissue. Specifically, T2* relaxation times provide information about spatial macromolecule architecture and its interaction with water molecule mobility^{10, 11}. T2* relaxation is a combination of inherent “true” T2 relaxation and additional relaxation due to magnetic inhomogeneities, on both a microscopic and macroscopic scale (1/T2* = 1/T2 + 1/T2’, where T2’ is the relaxation due to magnetic field inhomogeneities). While T2* is affected by bulk inhomogeneities in the magnetic field, which are typically not of interest, it is also affected by the differences in tissue composition at microscopic level, such as the change from cartilage to bone, annulus to nucleus pulposus or susceptibility-induced changes related to para- or diamagnetic depositions within the disc⁴¹. These principals extrapolate to other tissues, such as the intervertebral disc. This study highlights the quantitative nature of the T2* imaging to distinguish features utilized in the Pfirrmann grading system, the current gold standard for assessing disc health, and correlate them with intersegmental spinal functional mechanics. The features of the Pfirrmann grading system quantified by T2*, including disc signal intensity and clarity of the transition zone between the AF and NP, were found to be more sensitive in detecting change in the global mechanics of the lumbar spine.

Disc health was correlated with the bending stiffness of the lumbar spine in flexion, but not in extension. Flexion bending stiffness increased with degeneration severity, resulting in a decrease of flexion ROM. Furthermore, the NZ/ROM ratio, a relative measure of joint laxity, was shown to increase with worsening degeneration, indicating a decrease in the stability of the segment. There is more laxity in the joint until the end range of motion, where the tissue is much stiffer, allowing for little motion outside of the neutral zone. These findings were significantly correlated with the Transition Zone Slope, T2* Intensity Area, and specific ROI’s, however no correlation was determined with Pfirrmann grade. Similarly to flexion, lateral bending stiffness was found to increase with degeneration, while ROM decreased. Pfirrmann grade was only significantly correlated with LB NZ/ROM and stiffness, while parameters of T2* were correlated with flexion, lateral bending, and axial rotation range of motion and stiffness measurements.

Previous studies have also reported a decrease in range of motion for both flexion and lateral bending, while axial rotation increased with degeneration^31,34–36. Mimura et al. reported an increase in lateral bending NZ/ROM ratio with disc degeneration similar to the results herein³⁴. The literature also suggests an increase in the bending stiffness of both lateral bending and flexion with degeneration³². Haughton et al. observed a decrease in axial rotation bending stiffness with degeneration³³. Thus, the mechanical results of the present study align with the prevailing literature and exhibit identifiable changes across the disc degeneration spectrum.

The observed changes in global mechanics are supported by the changes locally in the AF and NP through degeneration. The mechanical properties of the AF in tension are deteriorated, as there is a shift of collagen production from Type I to Type II^42–48. This results in an increase in AR ROM while stiffness decreases⁴⁹. Conversely, flexion and lateral bending places the ipsilateral AF in compression and previous literature has shown that the compressive modulus of the AF increases with degeneration^28,29,50, causing the bending stiffness of the functional spinal unit to increase. This phenomenon is exacerbated when the NP loses hydration due to the loss of proteoglycans, associated with degeneration. The NP no longer acts like a pressure vessel and the disc height decreases while a larger compressive load is placed on the outer AF.

It is important to note that the correlation coefficients obtained in this present study are weak to moderate, but was determined to be significant because of the large sample size. In order to expand the clinical utility of these imaging metrics, a post-hoc multiple linear regression analysis was performed to predict the biomechanical outcome measures by using combinations of all the T2* MR imaging parameters (ROI’s 1–5, Transition Zone Slope, and T2* Intensity Area). The correlation coefficients are displayed in Table 4. Also included in the table is the highest correlation coefficient achieved from the univariate linear regression analysis and the percent improvement in prediction. The improvement in agreement between the two analysis techniques illustrates the ability of various T2* parameters to identify independent characteristics of disc health that relate to the functional mechanics of the lumbar spine.

Table 4.

Correlation Coefficient Comparison between Multiple Linear Regression Analysis and Univariate

	Flexion			Lateral Bending			Axial Rotation
	ROM	NZ / ROM	Stiffness	ROM	NZ / ROM	Stiffness	ROM	NZ / ROM	Stiffness
Univariate (r)	0.414	0.352	0.308	0.335	0.406	0.341	0.292	0.151	0.371
Multivariate (r)	0.536	0.434	0.496	0.484	0.575	0.611	0.486	0.344	0.486
Improvement (%)	29.5	23.3	61.0	44.5	41.6	79.2	66.4	127.8	31.0

(r: Correlation Coefficient). Multivariate coefficient includes all T2* Imaging Parameters (ROI’s 1–5, Transition Zone Slope, and T2* Intensity Area). The Univariate value represents the highest correlation coefficient achieved between the all imaging metrics and the corresponding biomechanical output.

As with any cadaveric study, the limitations to this work should be considered to contextualize the results. Biomechanical tests were conducted on cadaveric tissue without musculature at a low loading rate commensurate with flexing fully forward over a 12 second duration. Although pure moments were applied to the spine and coupled motions could have occurred, only the primary direction of bending was analyzed for each rotation sequence. Although three levels were all compared against each other, there was no statistical difference as a function of disc level within a single Pfirrmann grade (p>0.10). Also, imaging was free of any motion artifacts, and therefore idealized, due to stationary specimens, which were at room temperature.

Previous studies have investigated the relationship between T2* relaxation times and Pfirrmann grade. Welsch et al. investigated the relationship between T2* relaxation times, grouped into 5 ROI’s from anterior to posterior, and Pfirrmann grades of subjects with low back pain¹⁷. Only including grades 2–5 they found a significant correlation between Pfirrmann grade and all but the anterior most ROI. Hoppe et al. performed a similar study using 7 ROIs13. They reported a significant correlation with Pfirrmann grade for all ROIs except for the anterior and posterior most regions. Their study did not include severely degenerated discs. The present study incorporated 54 discs from the entire degeneration spectrum, which followed a normal distribution (7 grade 1, 12 grade 2, 17 grade 3, 11 grade 4, 7 grade 5). Similarly to Welsch et al. there was a significant correlation between Pfirrmann grade and ROI’s 2–5 as well as the newly defined variables, T2* Intensity Area and Transition Zone Slope.

Quantitative T2* assessment of disc health has the potential to be more sensitive in detecting the local changes associated with disc degeneration, particularly due to its intrinsic nature to detect water mobility, or the interaction of water within the macromolecular network^11,18. T2* also predicted altered kinematics better than traditional Pfirrmann grading. Therefore it is postulated that the T2* relaxation times may be correlated to the concentration of proteoglycan or collagen, which are related to its biomechanical competence. This technique allows for the quantification of key characteristics of disc health, without subjectivity or bias. These include the T2* Intensity Area and more uniquely the Transition Zone Slope, which defines the distinction between the NP and AF. T2* also benefits from a short acquisition time, high signal-to-noise ratio, three-dimensionality and is currently available on clinical scanners.

T2* MRIs produce a sensitive measure capable of detecting alterations in the lumbar intervertebral disc morphology which are correlated with their mechanical performance. These morphologic changes are likely due to altered biochemical content during degeneration. Affecting the local mechanical properties of the tissue, and ultimately leads to affected kinematics and spinal function. Further investigation is needed to establish the relationship between T2* relaxation times and precise biochemical content, such as water, proteoglycans and collagen. Nonetheless, this work highlights a clinically available MR imaging sequence (T2*) and disc health quantification methods that are correlated with the functional mechanics of the spinal segment. This new methodology and analysis technique may enhance quantitative data on the degenerative cascade, and enable greater patient stratification for therapeutic strategies with a currently available technology.

Key Points.

1. Key features of the Pfirrmann grading scale were quantified using T2* magnetic resonance imaging.

2. These metrics, T2* relaxation times, T2* Intensity Area, and Transition Zone Slope, predicted altered kinematics better than traditional Pfirrmann grading.

3. This new methodology and analysis technique is a potential diagnostic tool for the assessment of disc degeneration.