Diffusion-Weighted Magnetic Resonance Imaging of the Spine and Spinal Cord
Water diffusion is the random displacement of proton molecules in a fluid system, which causes signal attenuation on a diffusion-weighted (DW) magnetic resonance (MR) image. A set of diffusion-weighted images can be collected, and from these images, the apparent diffusion coefficient (ADC) can be calculated in each voxel. In biological tissue, the diffusion of water molecules is hindered, relative to diffusion in free water.1 and 2 Several factors, such as cell type, cell number, cell membrane density and permeability, size of macromolecules, tissue viscosity, and temperature, may influence diffusion of the water in living tissue.
DW MR imaging has been widely adopted for the evaluation of a variety of diseases affecting the brain, especially for acute stroke. Diffusion-weighted imaging (DWI) is a well-established modality for detection of cerebral ischemic changes at a time when therapeutic intervention may still be successful.3 and 4 Multiple sclerosis, epilepsy, traumatic brain injury, brain abscesses, brain development disorders, alcoholism, and Alzheimer’s disease are other pathological conditions where DWI has already shown promising results, although on a smaller scale.5, 6, 7 and 8
While there are numerous reports about the high sensitivity and usefulness of DWI in the brain, the experience with DWI in the vertebral column and the spinal cord is much more limited.9, 10, 11, 12, 13, 14 and 15 This is certainly due to the much greater magnetic inhomogeneities in and around the spine, including the small size of the spinal cord itself, involuntary motion, and the low signal-to-noise ratio. Pulse sequences specifically designed for the spine and spinal cord are not widely available and require much more optimization. With the recent MR systems, new radio frequency (RF) coil technology, and software innovations, these problems can be partially overcome.
This review describes the possibilities and usefulness of DWI in different diseases of the spine and spinal cord.
As in the brain, DWI of the spinal cord could contribute to both the diagnostic process, such as the delineation of acute spinal cord ischemia, and the pathophysiologic understanding of many disorders, such as spinal cord ischemia, demyelinating diseases, and neoplasms. As already mentioned, the main technical problems in DWI of the human spinal cord in vivo are the small size of the spinal cord and physiological motion. Specifically, the small size of the spinal cord and adjacent structures requires smaller voxel sizes and, thus, reduces the signal-to-noise ratio (SNR). The spinal cord is approximately 45 cm in length in the adult male, with the largest cross-sectional area in the lower cervical (38 mm circumference) and lumbar regions (35 mm). This means that sagittal acquisition is usually necessary to obtain sufficient coverage in a reasonable scan time (Fig. 1). The greatest problem, however, is the sensitivity of all diffusion measurement techniques to motion, which leads to considerable ghosting artifacts in DWI. The spinal cord can move independently of the surrounding tissues, and movements of the spinal cord, cerebrospinal fluid (CSF) flow, swallowing, and breathing are problematic for DWI of the spinal cord, as well as resulting disturbing artifacts. Susceptibility gradients around the vertebrae and pronounced global field variations adjacent to the cervicothoracic junction and the lungs may cause severe distortions that heavily impair overall image quality. The highly ordered arrangement of axons in the spinal cord, the CSF spaces, and nearby bone structures make ADC and diffusion anisotropy measurements of the cord challenging. In an attempt to diminish imaging artifacts resulting from motions—widely known as ghosting artifacts—several DWI methods have been introduced over the past decade.15 R. Bammer and F. Fazekas, Diffusion imaging of the human spinal cord and the vertebral column, Top Magn Reson Imaging 14 (2003) (6), pp. 461–476. Full Text via CrossRef | View Record in Scopus | Cited By in Scopus (23)15 Each of these techniques has certain advantages as well as drawbacks, and there is still no consensus about what is the optimal diffusion-weighted sequence.
Full-size image (199K)
Figure 1. DWI of the cervical spine in a healthy volunteer. Sagittal diffusion-weighted MR image of the cervical spinal cord in a healthy volunteer using navigated, interleaved, multishot echo planar imaging (IEPI) (5-mm slice thickness, FOV, matrix, b max = 700 s/mm2). (A) Calculated isotropic diffusion-weighted image (DWI). (B) Corresponding apparent diffusion coefficient map (ADC).
Quantitative diffusion measurements in healthy volunteers confirmed the assumption that diffusion coefficients in the spinal cord are comparable to those of the brain and demonstrate diffusion anisotropy.12 Fifteen healthy volunteers were imaged using a multishot, navigator-corrected, spin-echo, echo-planar pulse sequence in a recent study.13 The results of that study confirmed the presence of diffusion anisotropy in the white matter of the human spinal cord. With the diffusion gradients perpendicular to the spinal cord, mean ADC values ranged from 0.40 to 0.57 × 10−3 mm2/s for white and gray matter in their study. With the diffusion gradients parallel to the white matter tracts, calculated ADC values were significantly higher. Bammer and coworkers found in their study that the ratio between diffusion along the fibers and across the fibers differed by at least a factor of 2. Similar findings were observed by other researchers.12 Nagayoshi and coworkers have obtained diffusion-weighted echo planar images of the human cervical cord in vivo in 17 healthy volunteers using a 1.5-T clinical MR unit. White matter of the spinal cord was hyperintense on DWI due to the anisotropic diffusion.16
Diffusion-weighted imaging was also recently successfully obtained for the assessment of the spinal cord in children.17 In one study, line-scan diffusion imaging (LSDI) was used for measurements of ADC and relative diffusion anisotropy in the spinal cord in 12 children. The diffusion anisotropy within the spinal cord in children was similar with in vitro and in vivo animal experiments. The results of the study suggest that LSDI is a feasible and robust technique for DWI of the spinal cord in a pediatric cohort of patients.
Primarily because of the small size of the spinal cord, diffusion quantification, as well as differentiation of gray matter and white matter, remains difficult with currently available techniques. An extension to DWI is diffusion tensor imaging (DTI). DTI promises to allow the characterization of the structural integrity of the spinal cord. Using diffusion anisotropy, it is possible to visualize different fiber tracts, such as the corticospinal tract or transverse bundles, in the region of the pons. In a recent study using navigated diffusion-weighted interleaved echo planar imaging (IEPI), Ries and coworkers were the first to demonstrate DTI in the human spinal cord.18 The water diffusivity in the direction parallel to the fibers was found to be 2.5 times higher than the average diffusivity in directions perpendicular to the fibers and showed cylindrically symmetric anisotropy.18 In another study, axial DTI acquisitions from healthy volunteers, using zonally magnified, oblique, multislice echo planar imaging (ZOOM-EPI), were obtained.19 Although it was not possible to distinguish between gray and white matter, the diffusivity maps showed lower diffusivity in the gray matter, and higher diffusivity in the lateral and posterior regions of the cord (where there are white matter tracts). Single-shot echo-planar axial diffusion tensor sequences with parallel imaging were recently used for tractography of the cervical cord.20 The cervical cord appeared as a tract bundle, cephalocaudally color-encoded in their study. A study published recently showed promising results using single-shot fast spin-echo DTI (DTI-SSFSE) in MR tractography of the brain and cervical spinal cord. DTI-SSFSE showed a much smaller degree of image distortions when compared with the DTI-EPI technique.21 DTI is technically much more challenging than conventional DWI. Larger studies with better data quality and sufficient diagnostic quality are necessary for serious conclusions about the utility of DTI in the clinical routine for patients with spinal cord diseases.
Many diffusion MR imaging techniques are not useful for imaging the spine because of the inhomogeneous magnetic environment and the high lipid content of vertebral bodies, which can lead to strong geometric distortions and chemical shift artifacts. Line-scan diffusion imaging has been proposed as a useful technique, because of the reduced sensitivity to motion. In one study, line diffusion measurements of the spinal column provided images with adequate quality and within a reasonable imaging time.15 Nagakawa and coworkers reported on a study of 119 vertebral compression fractures with a single-shot EPI sequence and observed acceptable image quality with a high specificity in diagnosis.22
Spinal Cord Infarction
Spinal cord infarction is an uncommon condition, usually clinically presented with sudden onset of paralysis, sensory loss, and urinary and bowel dysfunction. Acute spinal cord ischemia has a very poor prognosis, with neuronal death, functional neurological loss, and paraplegia in up to 33% of the cases.23 Atherosclerosis, aortic dissection (3 to 5% risk), and aortic surgery (1 to 10% risk) are the most common causes of acute spinal cord ischemia.24 Other rare causes for spinal cord infarction include the following: cardiac embolism; decompression sickness; coagulopathy; spinal arteriovenous malformations; systemic hypotension; epidural anesthesia; and vasculitis. MR imaging findings in acute spinal ischemia are rather nonspecific and usually present as intramedullary hyperintensity on T2-weighted MR images with cord enlargement. In one clinical study, only 45% of the patients with acute spinal cord ischemia had signal intensity changes on T2-WI MR images.24
In 2000, Gass and coworkers were the first to report DWI hyperintensities in spinal cord infarction.25 Twelve hours after symptom onset, T2-weighted MR images demonstrated high signal intensity lesions in the lumbar cord and conus medullaris. DWI was performed 30 hours after symptom onset and showed hyperintensity in the spinal cord and a reduced ADC value, confirming a centrospinal infarct. On follow-up examination, 11 days later, ADC was increased, suggesting pseudonormalization. It is interesting to note that pseudonormalization of the ADC is normally seen in the late subacute phase of an ischemic stroke. At that point, the reduced ADC value increases, and lesions cannot be distinguished from normal tissue based on ADC. On T2-weighted scans, these lesions appear hyperintense. Depending on the TE of the DWI sequence and the associated T2–shine-through effect, pseudonormal lesions can appear more or less hyperintense. A similar case was examined in our institution with increased signal in the conus medullaris on DWI 24 hours after the onset of symptoms (Fig. 2). Stepper and Lovblad reported a case of spinal cord infarction, seen on echo-planar DWI, in a patient after grafting of the descending thoracoabdominal aorta.26 An area of hyperintensity corresponding to a decrease in diffusion was observed on DWI. Likewise, Sagiuchi and coworkers demonstrated signal abnormalities on DWI in a case of anterior spinal artery stroke of the cervical spinal cord using single-shot echo-planar imaging.27 High signal on DWI and decreased ADC were seen 26 hours after onset of symptoms. Twenty-eight days later, an increase in the ADC was noted, consistent with a chronic stage of an infarction. Weidauer and coworkers described two cases of anterior spinal artery syndrome where a clear diagnostic benefit was observed for DWI over conventional sequences.28 In one of the patients, 4 hours after the onset of symptoms, conventional MR sequences showed no abnormality, whereas with DWI, high signal intensities were noted in the cord that were consistent with spinal infarction. In a recent study, three cases of spinal cord infarction, detected with SSFSE DWI, were described.29 and 30 The findings with DWI, combined with clinical information, enabled the authors to confidently make the diagnosis of spinal cord infarction. High signal intensities on DWI and low ADC values compared with the normal cord were present in these cases as well. In one of their cases, an increase in ADC was observed on follow-up scans, again indicating pseudonormalization. In two other cases, the ADC values remained low, possibly because of venous contributions in spinal cord infarctions, with delayed clearance of necrotic cells by macrophages due to the individual variations of anastomoses.31 and 32 In three patients with spinal cord infarction reported by Küker and coworkers in 2004, a strong diffusion abnormality and T2-WI abnormalities were demonstrated in all cases.33 However, diffusion abnormalities showed a rapid decline, suggesting early signal normalization. The largest series of patients with spinal cord infarction and DWI findings includes six patients. In this study, ADC values ranged from 0.23 × 10−3 to 0.47 × 103 mm2/s.31 The ADC measured in reported cases of spinal cord infarction ranged between 0.23 and 0.9 × 10−3 mm2/s. The shortest time reported in the literature between the onset of clinical symptoms and abnormalities shown on DWI was 3 hours after ictus. Conventional MR sequences failed to show an abnormality in four reported cases at 3, 4, 10, and 46 hours after the onset of spinal ischemia.28, 30 and 34
Full-size image (169K)
Figure 2. Spinal cord infarction in a 30-year-old female. (A) Sagittal T2-weighted MR image (TR/TE 3000/140) of the lumbar spine performed 24 hours after the onset of clinical symptoms demonstrates high signal in the region of the conus medullaris with cord swelling. (B, D) On sagittal (B) and axial images (D) IEPI DWI (b = 700 s/mm2), intramedullar high signal abnormality is present in the corresponding region, located predominantly on the left side of the cord. (C, E) On ADC maps, low ADC values were measured (0.86 × 10−3 mm2/s), consistent with restricted diffusion and spinal cord ischemia.
Although the largest number of patients with described DWI findings is small (six patients), and the exact sensitivity and specificity of DWI for spinal cord infarction must be further investigated, these preliminary data suggest that DWI has the potential to be a useful and feasible technique in the early detection of spinal infarction. According to the published reports, diffusion abnormality can be found after a few hours, but does not last for longer than 1 week, which is shorter than in brain ischemia. However, further study is required, in a much larger patient cohort, to establish the time course in the spinal cord ischemia. Early diagnosis of acute spinal cord ischemia will contribute to improved patient management and allow an earlier application of pharmacological agents that may reduce neuronal apoptosis, and thus, subsequent neurological deficit.
Traumatic Lesions of the Spine and Spinal Cord
The published incidence rate for traumatic spinal cord injury in the USA ranges between 28 and 55 per million people, with about 11,000 new cases every year. Most spinal cord trauma occurs in young, healthy individuals. Males between the ages of 15 and 35 are most commonly affected. The initial cord injury itself is only partially responsible for the functional deficits that ultimately occur. Increased functional loss is related to “secondary injury,” an immune response, which can persist for several days and results in increased lesion size, swelling, and ultimately, the additional degeneration of axonal fiber tracts.35 The exact stage of traumatic injury is often difficult to characterize by conventional MRI, as the ultimate goal is to test for the functional integrity of the axons within the white matter tracts of the spinal cord, which cannot be accomplished with conventional MRI. Similarly, conventional MRI cannot detect possible therapeutic responses to neuroprotective drugs. One of the main problems in evaluating new therapies designed to reduce irreversible spinal cord damage is the difficult assessment of the pathological and pathophysiological state of the cord. DWI could be potentially useful to show the extent of the injury.
The use of DWI was evaluated first in rats with injured spinal cords.36, 37 and 38 Specifically, DWI was performed in injured rats treated with neuroprotective agents and compared with controls. The results showed a significant difference in anisotropy, with higher diffusion anisotropy in the spinal cords of the treated animals than in the untreated animals. In another animal study, diffusion MRI was used to demonstrate spontaneous regeneration in hemi-crushed rat spinal cords.39 In this context, Sagiuchi and coworkers reported recently a case of acute spinal cord injury with a type II odontoid fracture in a human and described DWI findings.10 DWI, performed 2 hours after injury, showed intramedullary hyperintensity and decreased ADC values at the C1-C2 vertebral levels.
The animal research clearly demonstrates that DWI is more sensitive for the evaluation of spinal cord injury and the outcome and regeneration after neuroprotection than conventional MR sequences. Although there is just one published report on DWI in spinal cord injury in humans, the results encourage further development of DWI as a useful method for detecting and visualizing spinal cord injury. One of the potential drawbacks for DWI in traumatic lesions of the spinal cord is the presence of hemorrhagic components in traumatic tissue damage.
Multiple sclerosis (MS) is a chronic inflammatory demyelinating disease of the CNS in which the spinal cord is often involved. Recent data suggest that MS is a T-cell-mediated disease with secondary macrophage activation. The pathologic hallmark of multiple sclerosis is inflammatory demyelination, which can lead to irreversible tissue loss or partial demyelination in cases where reparative processes occur with subsequent remyelination. MRI is, by far, the most sensitive technique for detecting MS lesions. MS plaques are seen best with T2-weighted MR sequences and are hyperintense on T2-WI and iso-hypointense on T1-weighted MR images. MS lesions in the cord are usually wedge-shaped and primarily located in the lateral and posterior columns.40
Clark and coworkers have used a conventional cardiac-gated, navigation diffusion-sensitized SE sequence for in vivo DWI of the spinal cord to measure ADC in normal controls and in patients with MS.41 These authors found that MS lesions usually present with increased rates of diffusivity with a significantly higher isotropic diffusion coefficient, compared with healthy controls (1180 ± 120 × 10−6 mm2/s (n = 4) versus 910 ± 50 × 10−6 mm2/s (n = 3)). Differences in diffusion anisotropy, however, although present, did not reach statistical significance. The decrease in anisotropy is probably due to the loss of myelin from white matter fiber tracts, with a net expansion of the extracellular space fraction. Perilesional inflammatory edema surrounding acute lesions may also contribute to the decreased anisotropy. Reduced anisotropy is also seen in MS brain lesions.42 One of the interesting findings in the study by Clark and coworkers was the large deviation in lesion values.41 The results of the correlative study with postmortem high-resolution MRI of the spinal cord in MS showed two types of abnormalities40: (a) focal high signal lesions representing demyelinated plaques; and (b) less well-defined areas of mildly increased signal representing partial demyelination. This morphological lesion heterogeneity would explain the large standard deviation in the lesion values measured in the study by Clark and coworkers. Postmortem MR findings in MS lesions of the spinal cord were recently correlated with quantitative measurements of axonal density and myelin content on a 7.0-T MR machine.43 In this study, a strong correlation was found between myelin content and axonal density in the cord, and magnetization transfer ratio (MTR), T1, proton density, and diffusion anisotropy. Moderate correlation was observed with ADC. One of the possible explanations is the variability of the extracellular space in MS lesions, independent of the number of axons. Loss of myelin should have more impact on the enlargement of the extracellular space with an increase in ADC.
The superiority of DWI in the detection of MS-related abnormalities in the spinal cord was proven in animal research. In one experimental study performed on excised swine spinal cords with experimental allergic encephalomyelitis (EAE), T1- and T2-weighted MRI revealed white matter abnormalities in only 5 of the 10 EAE diseased spinal cords, while high b-value, q-space, diffusion-weighted MRI (q-space DWI) detected white matter abnormalities in all diseased spinal cords.44 Furthermore, high b-value, q-space DWI was able to detect abnormalities in the normal-appearing white matter in spinal cords where no plaques were identified by conventional MR images.
To assess whether diffusion tensor-derived measures of cord tissue damage are related to clinical disability, both mean diffusivity (MD) and fractional anisotropy (FA) histograms were acquired from the cervical cords obtained from a large cohort of MS patients.45 Cord and brain MD and FA histograms were created from 44 patients with MS and 17 healthy controls. The study showed that average cervical cord FA was significantly lower in MS patients compared with controls. Good correlation was found between the average FA and average MD and the degree of disability. The same research group demonstrated in another study that patients with primary progressive MS have reduced average cervical cord fractional anisotropy, and increased cord mean diffusivity.46 Based on their results, the authors concluded that the quantification of the extent of diffuse cervical cord pathology can be performed with diffusion tensor MR imaging.
In our series (unpublished data) the majority of acute MS lesions had high signal on DWI and increased diffusivity when measurements were performed (Figure 3 and Figure 4). However, some of the lesions showed high signal on DWI but low ADC values on ADC maps, suggesting severe inflammation and restricted diffusivity (unpublished data).
Full-size image (173K)
Figure 3. Multiple sclerosis (MS) plaque in the cervical spinal cord. (A) High signal intensity lesion in the cervical spinal cord is shown on sagittal STIR image. (B, C, D) The lesion has low signal on T1-weighted TSE MR image (TR/TE 400/10) and shows peripheral enhancement on postcontrast images (C, D). (E) On axial IEPI DWI (b = 700 s/mm2), high signal is present in the left side of the cord at the corresponding level. High signal is present on the ADC map (not shown), and region of interest (ROI)-based analysis revealed an ADC for the normal-appearing spinal cord (NASC) of 0.88 × 10−3 mm2/s, and for the MS lesion, 1.058 × 10−3 mm2/s, indicating elevated diffusivity in the acute MS lesion.
Full-size image (177K)
Figure 4. MS plaque in the cervical spinal cord. (A, B) Sagittal and axial T2-weighted MR image of the cervical spine with a hyperintense lesion located dorsally in the spinal cord at the C4 to 5 level. (C) On postcontrast T1-WI MR image, homogenous enhancement of the lesion is observed. (D, E) High signal is noted on axial DWI (D), with increased ADC (1.13 × 10−3 mm2/s) compared with normal-appearing white matter (1.07 × 10−3 mm2/s) measured on ADC map (E), indicating elevated diffusivity in an active MS lesion.
The exact value of DWI in MS of the spinal cord has not been completely evaluated yet. DWI may be of benefit in demonstrating remyelination in MS plaques in the spinal cord, and changes in ADC values would indicate or differentiate lesions with and without remyelination.
Degenerative spondylosis may cause significant narrowing of the spinal canal and can ultimately lead to compression of the spinal cord. Compromise of the spinal cord can also result from protruded or herniated discs, especially in the case of a narrow spinal canal. Clinically, such patients usually present with chronic progressive signs of myelopathy. Bammer and coworkers have found reduced ADC values in spondylotic myelopathy, whereas the surrounding cord demonstrated elevated diffusivity.47 The former is presumably due either to cord compression or to vascular compromise, while the latter is due to surrounding edema. Ries and coworkers recently showed a patient suffering from severe spinal stenosis with cord compression.18 The extent of diffusion abnormality was much larger than the area seen on conventional MRI. In one case in that study, DWI scans appeared hypointense at the level of the stenosis, and quantitative measurements indicated significantly increased ADC values. However, it is not clear whether or to what extent susceptibility differences (caused by protruded discs or osteophytes that moved toward the anterior aspect of the cord) may have also led to strong signal dephasing in this particular patient. Tsuchiya and coworkers demonstrated a significant increase in ADC within areas of myelomalacia due to chronic compression.20 Cystic necrosis of the central gray matter and the syrinx, as well as atrophy in myelomalacia, explain the increased proton mobility. Recently, a multishot echo-planar sequence was used to calculate ADC and apparent diffusion tensor (ADT) in 36 patients with symptomatic cervical spondylosis.48 In this study, 17 of 21 patients with myelopathy had an abnormal ADC and 15 of 19 had an increased ADT. The sensitivity of DWI for the detection of myelopathy was 80% compared with T2-weighted MRI, with a sensitivity of 61%. The results of the study confirmed previously published findings that DWI is more sensitive than T2-weighted MRI in detecting cervical myelopathy.49 The changes in cervical myelopathy may be located in the interstitial space, with increased internal pressure. An increase in internal pressure may be due to compression of the cord from bone formations and disks, as well as narrowing of the spinal canal at that level. Recently, 15 patients with clinical symptoms of acute or slowly progressive spinal cord compression and 11 healthy volunteers were prospectively evaluated using conventional MR sequences and diffusion tensor MR imaging.50 In their series, FA had a much higher sensitivity and specificity (73.3 and 100%) in the detection of spinal cord abnormalities compared with T2-weighted FSE imaging (46.7 and 100%) and ADC (13.4 and 80%). Early detection of myelopathy would be of benefit for patients who are potential candidates for decompressive surgery.
Infections of the Spine and Spinal Cord
MR is the method of choice for detecting and demonstrating spinal infections. Typical MR imaging findings in pyogenic spondylodiscitis are low signal on T1-WI, and high signal on T2-WI MR images with enhancement on postcontrast T1-WI images of the affected vertebral bodies; high signal on T2-weighted MR images with loss of intranuclear cleft and peripheral enhancement of the involved disk; and irregularity of the vertebral endplates.51 Large paraspinal collections, involvement of more than one segment, skip lesions, contiguous subligamentous spread, and odd locations (laminae, pedicle, spinous process) all indicate a tuberculous origin. However, MR imaging findings in pyogenic or tuberculous spondylitis are relatively unreliable and both conditions can result in images that mimic images of malignancy. A conclusive diagnosis requires biopsy and culture, which are invasive and not always definitive. Differentiation between tuberculous and pyogenic spondylitis is difficult both clinically and radiologically.
A recent study evaluated the efficacy of diffusion-weighted MR imaging in differentiating spinal infections from malignancy.52 The authors did not find any statistically significant differences in ADC values between 69 tuberculous bone lesions and 9 pyogenic infections. However, the mean ADC of malignancy was significantly lower than the mean ADC of tuberculous and pyogenic infections. The mean ADC of infectious paraspinal masses was significantly higher than that of malignant soft-tissue masses. A higher ADC in the bone marrow, compared with normal vertebral bodies, was found in both infections and malignancy in this study. The greater diffusion in infectious processes compared with malignancy was explained by higher water content and replacement of bone marrow fat by inflammatory cells and proteins in infections (Figure 5 and Figure 6).
Full-size image (164K)
Figure 5. Patient with pyogenic spondylodiscitis at the L4 to 5 level. (A) Marked enhancement of the affected vertebral bodies and patchy enhancement of the disk L4 to 5 were demonstrated on sagittal postcontrast T1-weighted TSE MR image (TR/TE 400/10). (B) High signal was noted in involved vertebral bodies on isotropic DWI image. (C) The vertebral bodies are hyperintense on the corresponding ADC map, with ADC values higher than in normal vertebral bodies, most probably due to the increased water content and presence of inflammatory cells in infectious spondylitis.
Full-size image (139K)
Figure 6. Patient with tuberculous spondylitis in the thoracic spine. (A) High signal intensity was observed in two vertebral bodies of the thoracic spine on T2-weighted TSE MR image (TR/TE 3000/140). (B) On postcontrast T1WI TSE MR image (TR/TE 400/10), intense enhancement of vertebral bodies is observed. No enhancement was shown in the disk space, indicating preservation of the disk. (C, D) The affected vertebral bodies are hyperintense on DWI image (C) and ADC map (D), with high ADC values consistent with the elevated diffusion in an infectious process.
Spinal epidural abscesses are a serious complication of spinal infections. Their avascular center is not accessible to intravenous antibiotics, and thus, usually require surgical drainage. Epidural abscesses can be classified as (a) acute abscess formations with frank pus in the epidural space; and (b) chronic epidural abscesses with granulation tissue. On MR imaging, epidural abscesses have high signal on T2-WI and low signal on T1-WI with two types of enhancement.53 A homogeneous enhancement corresponds to an abscess with inflammatory tissue without purulent collection; a peripheral enhancement indicates a true abscess formation with purulent fluid. There are numerous reports about the usefulness of DWI in distinguishing brain abscesses from necrotic neoplasms.53, 54, 55, 56 and 57 Typically, brain abscesses have restricted diffusion and show a high signal on DWI and low ADC values.53, 54, 55, 56 and 57 In one case report on DWI in epidural abscess formation,58 a high signal was observed on DWI with a corresponding low signal on ADC maps, consistent with restricted diffusion. In our series of patients with spinal infections, all epidural purulent collections were hyperintense on DWI with low ADC values (unpublished data) (Fig. 7).
Full-size image (227K)
Figure 7. Epidural abscess formation in the lumbar spine. (A) On postcontrast T1-WI TSE MR image with spectral fat saturation (SPIR), fluid collection with peripheral enhancement is shown in the epidural space at the level L4-S2. (B, C) High signal was observed on sagittal (B) and axial (C) isotropic DWI image. (D) Low signal was observed on ADC map, indicating restricted diffusion in purulent collection.
Using conventional MR imaging, the differentiation of benign from malignant acute vertebral compression is difficult. Morphologic criteria that favor a malignant cause are as follows: paravertebral soft tissue masses; involvement of the posterior elements; and irregularity of the posterior vertebral margin. A wedge-shaped deformity, preservation of the posterior vertebral margin, absence of soft-tissue masses, and lack of pedicle abnormalities favor an osteoporotic origin. All fractures are usually hypointense on T1-weighted MR images, have a high signal on T2-weighted MR images, and show enhancement on postcontrast studies.59 Signal intensity abnormalities detected on conventional MR sequences cannot be reliably used for differentiation of benign and malignant vertebral fractures.
Results from recent studies have raised the hope that DWI might be an important adjunct in the diagnostic work-up of acute vertebral fractures (Figure 8 and Figure 9).59, 60, 61, 62, 63, 64, 65, 66, 67 and 68 In the first published study in 1998 about the usefulness of DWI in vertebral fractures, DWI showed pathologic compression fractures of the spine to be hyperintense to adjacent normal vertebral bodies, contrary to the benign compression fractures that were hypo-isointense to normal vertebral bodies.60 The enthusiastic conclusion of the study was that restricted diffusion and high signal on DWI in pathologic fractures are due to reduction of the interstitial space and replacement with tumor tissue. The major problem with this explanation is the fact that the SSFP sequence used in that study contained information from both T1 and T2 effects, and, at least partially, the high signal on DWI could be the result of T2 effects. A SSFP sequence with a diffusion gradient in only one direction was used in the above-mentioned study; therefore, quantification was also not possible. In contrast to that study, Castillo and coworkers did not find any advantage for DWI of the spine in the detection and characterization of vertebral metastases (in the absence of compression fractures) compared with unenhanced T1-weighted MR imaging.59 In this particular study, 53% of vertebral metastases showed low signal on DWI, which was attributed to the normal extracellular space and relatively free water movement. This could be the case in sclerotic and treated metastases. Reduced extracellular space and tumor cell accumulation are the cause of restricted water mobility, seen in 34% of metastases with high signal on DWI.
Full-size image (163K)
Full-size image (115K)
Figure 8. Patient with osseous metastases from gastric cancer and cord compression. (A) On sagittal T2-weighted TSE MR image, slight hyperintensity is observed in three thoracic vertebrae, with a marked reduction in size and dorsal dislocation as well as compression of the spinal cord. (B, C) Affected vertebral bodies have low signal on T1-WI TSE MR image (B) (TR/TE 400/10) with enhancement on postcontrast image (C). (D) Involved vertebrae have high signal on isotropic DWI image. High signal is also present in the spinal cord below the level of compression. (E) On the ADC map, the following ADC values were measured: 1.29 × 10−3 mm2/s for the normal spinal cord, and 0.74 × 10−3 mm2/s for the intramedullary lesion, suggesting restricted diffusion due to the compression myelopathy and venous congestion.
Full-size image (192K)
Figure 9. Osseous metastases in a patient with a history of prostatic cancer. (A) High signal is demonstrated on this T2-WI TSE MR image in multiple thoracic vertebral bodies, with compression of the spinal cord at the Th7 level due to the dorsal dislocation of the vertebral body (Th7) into the spinal canal. (B, C) The cord is expanded and shows intramedullary high signal on DWI with low ADC, indicating compression myelopathy.
Zhou and coworkers were the first to report on quantitative analysis of DWI abnormalities in vertebral fractures. ADC was measured in their study using a fast spin-echo technique (FSE).64 The results of the study have addressed several important issues. Four metastatic lesions were hypointense on DWI, which was explained by low tumor cellularity and increased water content. The authors found a statistically significant difference in ADC values between malignant and benign vertebral fractures, although considerable overlap in values for both groups was an obvious problem for threshold settings. However, that study has attracted some attention and raised some concerns.67 One of the major concerns was the influence of vertebral marrow lipid content, which may range from 20 to 70% in healthy individuals and may contribute to the quantitative analysis of ADC.69 To avoid that problem, use of fat saturation or selective water excitation was recommended.67 On the other hand, the lipid content of metastatic vertebras is significantly lower due to the replacement of normal marrow with tumor cells or edema; thus, the impact of lipid signal intensity is low.64 This is, of course, only true for severe metastatic replacement of the vertebra, but not for incomplete replacement of normal marrow. Herneth and coworkers reported on the ADC values for metastases and normal vertebral bodies with a navigated interleaved diffusion-weighted echo-planar method (IEPI-DWI) on a 1.0-T clinical scanner.66 The mean ADC for metastases (0.69 × 10−3 mm2/s) was significantly lower than that for adjacent normal bone marrow (1.66 × 10−3 mm2/s), and pathologic fractures had a lower ADC (0.65 × 10−3 mm2/s) compared with benign fractures (1.62 × 10−3 mm2/s). The study has clearly shown the feasibility of EPI-DWI for the quantitative assessment of ADC in distinguishing pathologic from benign vertebral fractures. In another report published in 2004, with 46 patients with osteoporotic or traumatic fractures and 31 patients with vertebral metastases, SSFSE was rated useful in differentiating the two groups.68 Low signal was detected on DWI in 95% of benign vertebral fractures. Metastases showed a low signal in 80% of the patients, and a high signal in only 10% of the patients.
Despite encouraging reports in the literature on the utility of diffusion techniques in distinguishing pathologic from benign vertebral fractures, there is still no consensus about the optimal sequences to use, and no threshold ADC values that are useful for everyday practice.
Epidural spinal cord compression is one of the most important emergencies and requires prompt and adequate treatment. The most common neoplastic lesions associated with epidural spinal cord compression are lung and breast cancers, followed by lymphoma, myeloma, prostate cancer, and sarcoma. About 10% of patients with non-Hodgkin’s lymphoma will present with spinal cord compression.70 Epidural lymphomas account for 9% of epidural spinal tumors.71 The signal intensity on T1-weighted images is usually lower than or equal to muscle, and also low on T2-weighted MR images. It has been postulated that, because of high cellularity, lymphoma will have restricted diffusion. Sixteen of 19 cerebral lymphomas appeared hyperintense on DWI and hypointense on ADC maps in one study.72 Quantitative analysis of their data has shown significantly lower diffusivity in lymphomas compared with high-grade astrocytomas. The results suggested that diffusivity of brain tumors may correlate with their cellularity; an increase in cellularity (such as in lymphomas) would decrease the extracellular space and lead to restricted water diffusion. If this is true, one would expect the same DWI features in epidural lymphomas in the spine. No published reports exist on DWI findings in lymphomas located in the spinal canal. In two cases of epidural lymphomas in our series, both showed marked high signal on DWI and low ADC values consistent with restricted diffusion (Fig. 10). It remains to be verified if quantitative analysis and ADC measurements of different epidural masses benefit diagnostic process and may help in distinguishing epidural lymphomas from other neoplastic lesions, as well as epidural abscesses and hematomas (Fig. 11).