SPINE Volume 39, Number 15, pp 1183-1189 ©2014, Lippincott Williams & Wilkins
The Assessment of Neuronal Status in Normal and Cervical Spondylotic Myelopathy Using Diffusion Tensor Imaging S. Rajasekaran, PhD,* Janardhan S. Yerramshetty, PhD,† Vishnuprasath S. Chittode, FNB,† Rishi M. Kanna, FNB,* Gopalakrishnan Balamurali, MD,* and Ajoy Prasad Shetty, MS*
Study Design. A prospective observational analysis of diffusion tensor imaging (DTI) datametrics collected from control and patients with cervical spondylotic myelopathy (CSM). Objective. The aims were to study the use of DTI in CSM and to probe whether DTI datametrics and tractography will correlate with magnetic resonance imaging and clinical findings. Summary of Background Data. Magnetic resonance imaging is the current “gold standard” in the assessment of cord status in CSM; however, various parameters such as extent of compression and presence of signal intensity changes do not correlate well with clinical status. DTI is a novel investigation tool with proven applications in brain pathologies but is not routinely used in spinal cord evaluation. Methods. Patients with CSM (n = 35) who required surgical decompression (mean age = 48 yr) and 40 normal individuals (mean age = 38 yr) were included. Diffusion Tensor Imaging of the cervical spine was obtained using a 1.5T magnetic resonance image. Apparent diffusion coefficient, fractional anisotropy, and eigenvalues (E1, E2, and E3) were obtained at each cervical level. The DTI datametrics of CSM patients were compared with normal volunteers and correlated with individual and grouped Nurick grades, which indicate the neurological status of patients. Results. There was significant difference in DTI datametrics between patients with myelopathy and control (P < 0.05), with decrease in fractional anisotropy (0.49 ± 0.081 vs. 0.53 ± 0.07)
From the *Ganga Hospital, Coimbatore, Tamil Nadu, India; and †Ganga Orthopaedic Research and Education Foundation, Tatabad, Coimbatore, Tamil Nadu, India. Acknowledgment date: December 25, 2013. Revision date: February 24, 2014. Acceptance date: March 27, 2014. The manuscript submitted does not contain information about medical device(s)/drug(s). Ganga Orthopaedic Research and Education Foundation, Coimbatore, India; Department of Biotechnology, New Delhi, India; and AOSpine Foundation (Asia Pacific), Kowloon, Hong Kong, funds were received to support this work. Relevant financial activities outside the submitted work: grants/grants pending. Address correspondence and reprint requests to S. Rajasekaran, PhD, Ganga Hospital, 313 Mettupalayam Rd, Coimbatore, Tamil Nadu, India 641043; E-mail: [email protected]
DOI: 10.1097/BRS.0000000000000369 Spine
and increase in apparent diffusion coefficient (1.8 ± 0.315 vs. 1.44 ± 0.145) and eigenvalues (E1: 2.82 ± 0.395 vs. 2.37 ± 0.221, E2: 1.64 ± 0.39 vs. 1.18 ± 0.198, E3: 0.956 ± 0.277 vs. 0.76 ± 0.142). There was also a significant difference between increasing grades of myelopathy when individuals were grouped as—control, selfambulant (Nurick grades 1 and 2), and dependent (Nurick grades 3, 4, and 5). Conclusion. The study shows that DTI is a promising and useful investigational tool in evaluation of CSM. There was a significant difference in all DTI values between control and patients with CSM, and there was a significant trend of change in values between control, self-ambulant, and dependent patients. Our results encourage further investigation of this important modality. Key words: diffusion tensor imaging, magnetic resonance imaging, cervical spondylotic myelopathy, Nurick grading, fractional anisotropy, apparent diffusion coefficient, eigenvalues, tractography. Level of Evidence: 3 Spine 2014;39:1183–1189
ssessment of patients with cervical spondylotic myelopathy (CSM) is done mainly by clinical evaluation and by magnetic resonance imaging (MRI), which give information regarding the extent and levels of cord compression and signal intensity changes. However, the extent of neuronal injury and the functional status of the cord, observed through MRI, do not quantify them accurately in all patients.1–3 Also, it is impossible to accurately predict the prognosis for recovery after the decompression.4 Diffusion tensor imaging (DTI), a new MRI sequence, is being extensively used for imaging brain, largely for qualitative and quantitative assessment of strokes, and mapping of neural tracts for assessing severity of injury.5–8 Recently, its application in imaging spinal cord and assessing spinal diseases and injuries is being evaluated.9–12 This article analyzes the usefulness of this technique in quantification of neuronal injury in patients with CSM. Evaluation of the scalar properties of diffusion of extracellular water molecules within white matter fibers forms the basis of DTI. Several recently published reports about DTI in cervical myelopathy indicate that DTI parameters were www.spinejournal.com
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE140018_LR 1183
10/06/14 8:52 PM
The Assessment of Neuronal Status • Rajasekaran et al
significantly different in compressive myelopathy when compared with control.9–12 However, most of the current literature is based on a small sample size. Also, these studies have not tested the relationship of DTI datametrics with severity of the myelopathy condition. In this study, 35 patients with CSM were evaluated using Nurick grading, and the relationship with DTI datametrics was analyzed.
85 ms; flip angle: −90°; field of view: −220 × 220 mm; acquisition data matrix: −256 × 256; original image resolution: −0.86 mm; reconstructed image resolution: −0.215 mm (with zero-filling interpolation); slice thickness: −4 mm; b value is 0 and 500 s/mm2; motion-probing gradient orientations: −12 directions; and number of axial slices: −40.
MATERIALS AND METHODS
From the acquired DTI images, DTI datametrics were obtained from the middle of C1 body and at the midlevels of C2–C3, C3– C4, C4–C5, C5–C6, C6–C7, and C7–T1. The spinal cord was outlined such that no CSF voxels were included in the region of interest (ROI), which may increase the partial volume effects. Diffusion tensor variables were calculated on a voxel-byvoxel basis (voxel area = 0.86 × 0.86 mm), whose descriptive statistics within the regions of interest were directly reported by the Siemen’s workstation. The following data metrics were obtained from the DTI analysis: apparent diffusion coefficient (ADC), fractional anisotropy (FA) and eigenvalues (E1, E2, and E3). A pilot study was performed to assess the interobserver reliability in 20 cases, and this produced an average intraclass correlation coefficient of 0.90 (ranged between 0.80 and 0.96) between observers. For statistical comparison, average from all levels was considered in the case of controls and only compressed levels (highly compressed in case of multiple levels of compression) in the case of patients. Tractography, which gives a visual representation of fiber tracts, was generated by placing initial seed at C1–C2 level covering the same ROI that was used for obtaining DTI datametrics. The criteria for cessation of tracking were set at FA values greater than 0.2. Four patterns were described as completely intact, waisted, partially interrupted, or completely interrupted (Figure 1).13 Fiber indentation at the compression level without interruption was taken as waisted pattern. Fiber interruption across parts of the axial plane was taken as partially interrupted. Fiber interruption across the entire axial plane was taken as complete interruption.
The study was approved by the institutional review board. Thirty-five patients with CSM who presented between April 2011 and January 2012 were included in the study. The inclusion criteria were patients with CSM with evidence of cord compression in cervical spine magnetic resonance (MR) image, with Nurick grading between 1 and 5. Evidence of cord compression in the MR image was defined as the absence of cerebrospinal fluid (CSF) space around the spinal cord accompanied by indentation of the cord in the sagittal T2 MRI sections. Cervical cord compression secondary to trauma, infection, and other etiologies was excluded from the study. For control, 40 healthy human volunteers (20 males and 20 females) were recruited on the basis of the following inclusion criteria: 1. No history of significant neck pain. 2. No symptoms of cervical spine disease in the form of radiculopathy or myelopathy. 3. No evidence of cervical spinal cord compression in the MR image. 4. No history of cervical spine surgery.
Clinical Assessment Clinical grading of patients with CSM was done on the basis of Nurick grading (1–5 levels). Patients were further categorized into 2 groups on the basis of severity of Nurick grading. Group I included patients falling into grades 1 and 2 (patients who did not need any assistance for ambulation). Group II included patients falling into grades 3, 4, and 5 (patients who needed assistance for ambulation).
A radiologist blinded to the clinical details of the patient analyzed MR images for levels and severity of spinal cord compression and cord signal intensity changes in T2-weighted images. In cases with myelopathy at more than 1 level, the segment having the narrowest anteroposterior cord diameter was considered for further analyses.
Pearson correlation was used to test the relationship between DTI datametrics and Nurick grading and other quantitative variables. Spearman’s correlation analysis was used for correlation between quantitative and ordered categorical variables. Student independent t test was used to compare between cases and control in the study. Correlations and comparisons were considered significant at P value of less than 0.05. All the analyses were done using SPSS v.17 (IBM, New York, NY).
Diffusion tensor images were acquired for both cases and control individuals, with a 1.5 Tesla MR imaging system (SIEMENS Magnetom Symphony; Siemens, Munich, Germany). For DTI acquisition, an axial spin-echo single shot echo planar parallel imaging was applied. Scanning parameters were as follows: repetition time: 6000 ms; echo time:
The Mean age of the 40 neurologically healthy volunteers (20 males and 20 females) recruited as normal individuals was 38 years. The Mean age of the 35 patients with CSM was 48 years that included 33 males and 2 females. The distributions of patients under Nurick grading are presented in Table 1.
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE140018_LR 1184
10/06/14 8:52 PM
The Assessment of Neuronal Status • Rajasekaran et al
Figure 1. Tractography patterns: (A) normal, (B) waisting, (C) partially interrupted, and (D) completely interrupted. Arrows show compressed regions.
Comparison of DTI Datametrics Between Control and Patients With CSM
MR Signal Intensity Changes in Patients With CSM
Patients with CSM when compared with normal individuals had significantly decreased FA (0.49 ± 0.081 vs. 0.53 ± 0.07) and relative anisotropy (0.45 ± 0.092 vs. 0.51 ± 0.089) values and significantly increased ADC (1.8 ± 0.315 vs. 1.44 ± 0.145) and eigenvalues at maximum compressed level (Table 2).
Correlation of DTI Datametrics With Nurick Grading When individuals were compared as 3 groups–-control, group I (patients with CSM who did not need assistance for walking) that included Nurick grades 1 and 2, and group II (patients with CSM who needed assistance for walking) that included Nurick grades 3, 4, and 5—DTI datametrics indicated significant differences and a trend in change, that is, average FA value decreased from control (0.53 ± 0.07) to group I (0.51 ± 0.05) and then to group II (0.49 ± 0.09) (Figure 2A). ADC showed increasing trend from control (1.44 ± 0.145) to group I (1.69 ± 0.21) and then to group II (1.83 ± 0.33) (Figure 2B). Variables with significant differences between the groups are indicated in Table 3. However, DTI datametrics did not correlate significantly with individual Nurick grades.
TABLE 1. Details of Patients With Cervical
Tractography in Patients With CSM We categorized the tractography morphology into 4 groups— control, waisting, partial block, and complete block.13 Although all control had normal tractography pattern, none of the patients exhibited normal pattern. The distribution of different abnormal patterns was highly variable between the Nurick grades with no specific trend. For example, completely blocked pattern was found to be evenly distributed between group I and group II (Table 4). Statistically, χ2 analysis indicated no significant (borderline, P = 0.133) association between interrupted patterns and severity of injury (Nurick grade).
Group I (without ambulatory assistance)
Group II (with ambulatory assistance)
A total of 23 (66%) patients showed signal intensity changes in T2-weighted images. The 12 (34%) patients without any signal intensity change had higher ADC values (1.86 ± 0.35 vs. 1.44 ± 0.15) and eigenvalues (E1: 2.92 ± 0.34 vs. 2.37 ± 0.22, E2: 1.67 ± 0.44 vs. 1.18 ± 0.20, E3: 0.97 ± 0.37 vs. 0.77 ± 0.14) in comparison with normal population. Patients with signal intensity changes too are significantly different from normal individuals, which had the following values (FA: 0.49 ± 0.07, ADC: 1.78 ± 0.30, E1: 2.76 ± 0.42, E2: 1.62 ± 0.38, E3: 0.95 ± 0.22). However, there were no significant differences between patients with and without signal intensity changes.
MRI is the most common tool utilized for preoperative assessment and prognosis of various spine conditions. In cervical spondylotic myelopathic cases, many parameters, such as cord morphology, MR signal intensity, multisegment involvement, areas of spinal canal and dural tube, and ratio of anteroposterior diameter to transverse diameter, were measured using MRI, however, with mixed results. Some studies found significant correlations with pre and postoperative indications and outcomes,14–16 whereas some found no correlation17–19 and some had mixed opinion.20,21 It is imperative to find alternatives, and DTI is one such technique that has huge potential for use in preoperative assessment and for prognostication. www.spinejournal.com
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. SPINE140018_LR 1185
10/06/14 8:52 PM
The Assessment of Neuronal Status • Rajasekaran et al
TABLE 2. Diffusion Tensor Imaging Datametrics
of Patients With Cervical Spondylotic Myelopathy and Control CSM—Indices at Compressed Level
0.49 ± 0.081
0.53 ± 0.070
1.80 ± 0.315
1.44 ± 0.145