Cauda Equina Atrophy in Amyotrophic Lateral Sclerosis on Routine Lumbar Magnetic Resonance Imaging : Journal of Computer Assisted Tomography

Secondary Logo

Journal Logo

Neuroimaging: Spine

Cauda Equina Atrophy in Amyotrophic Lateral Sclerosis on Routine Lumbar Magnetic Resonance Imaging

Matsushima, Satoshi MD, PhD; Omoto, Shusaku MD, PhD; Shimizu, Tetsuya MD, PhD; Baba, Akira MD, PhD; Ojiri, Hiroya MD, PhD

Author Information
Journal of Computer Assisted Tomography 46(6):p 991-996, 11/12 2022. | DOI: 10.1097/RCT.0000000000001349
  • Free

Abstract

Amyotrophic lateral sclerosis (ALS) is a progressive neurodegenerative disorder affecting the upper and lower motor neurons. It requires an accurate and early diagnosis for appropriate therapeutic intervention to be provided in the early stages of the disease. Currently, the diagnosis is generally made using the Awaji criteria and the revised El Escorial criteria; however, the sensitivity is inadequate,1,2 and the establishment of additional biomarkers to make an accurate diagnosis is required. Recently, various advanced magnetic resonance imaging (MRI) techniques, such as voxel-based morphometry, diffusion tensor imaging, functional MRI, and magnetic resonance spectroscopy, have been used to establish biomarkers.3,4 However, these biomarkers have not yet been widely used in daily clinical practice. In addition, although ALS has been characterized by increased signals in the subcortical white matter of the precentral gyrus and along the contiguous corticospinal tracts on T2-weighted images (T2WIs), proton density-weighted images, and fluid-attenuated inversion recovery and decreased signals in the motor cortex on T2WI, fluid-attenuated inversion recovery, and susceptibility-weighted imaging,5–7 the sensitivity of these findings is not sufficient. Moreover, the motor cortical signal loss was correlated with ALS and with age in normal controls; thus, decreased signals in the motor cortex can be seen even in older individuals without ALS.7 To improve the diagnosis of ALS in daily clinical practice, the discovery of additional ALS-specific findings that are available in routine MRI is highly anticipated.

Recently, a case report described the accumulation of 43 kDa phosphorylated transactivation response DNA-binding protein, which is known as the histological hallmark of the central nervous system in ALS, in the cytoplasm of Schwann cells of the peripheral nerves, such as facial nerve, auditory nerve, spinal cord anterior roots, cauda equina, and peripheral nerves in the dorsal root ganglia.8

However, there are few reports on the imaging of peripheral nerves in ALS using MRI. As for cauda equina in particular, to the best of our knowledge, there have been only 2 case reports of cauda equina with gadolinium enhancement on the lumbar roots9,10; there is no report of a coherent assessment of the cauda equina in ALS using MRI. We hypothesized that the volume of the cauda equina could be altered in ALS. In addition, because routine lumbar MRI provides a relatively stable visualization of the cauda equina, we speculated that it could be a novel biomarker of ALS in daily clinical practice. Therefore, we aimed to retrospectively examine the total cross-sectional area of the cauda equina of ALS patients using routine clinical lumbar MRI.

MATERIALS AND METHODS

Subjects

Our institutional review board approved this retrospective study, and informed consent was waived due to the study's retrospective nature. The subjects were 15 patients (11 men, 4 women; mean ± standard deviation [SD] age, 68.5 ± 13.9 years [range, 37–89 years]) who were diagnosed with ALS and had undergone lumbar spine MRI at our institution between October 2013 and September 2020. The diagnosis of ALS was made by neurologists with neuromuscular disorder training following the Awaji criteria and the revised El Escorial criteria (probable ALS in 11 cases and possible ALS in four cases). Cases with other possible neurological diseases or complications of other neurological diseases based on the history, clinical symptoms, clinical findings, and course were excluded. Among them, 14 patients had clinical impairment of the lower motor neuron in the lumbosacral spinal cord region, and one patient did not. The exact onset was unknown in one case; however, the mean ± SD from onset to MRI acquisition in the 14 patients for whom the time of onset was known was 70.93 ± 69.63 weeks (range, 8–240 weeks).

From January 2019 to September 2020, we included 15 cases as non-ALS controls from the patients who underwent lumbar spine MRI for various purposes at our institution who were age- and sex-matched to the above ALS cases. If there were multiple cases of the same age and sex, only one was randomly selected. The final diagnosis among these 15 cases was lumbar spondylosis with spinal canal stenosis in 10 cases, lumbar spondylosis without canal stenosis in 2 cases, disc herniation in 2 cases, and Tarlov cyst in 1 case. Among these cases, none had a diagnosis or history of background disease-causing nerve root atrophy (spinal muscular atrophy, etc) or thickened cauda equina (meningitis, chronic inflammatory demyelinating polyneuropathy, Guillain-Barré syndrome, Charcot-Marie-Tooth disease, sarcoidosis, Sjogren syndrome, Krabbe disease, metachromatic leukodystrophy, amyloidosis, or neoplastic lesions such as malignant lymphoma, meningeal dissemination, and leptomeningeal gliomatosis/melanomatosis, etc). In addition, cases with artifacts in visual evaluation or severe lumbar spinal canal stenosis with no visible cerebrospinal fluid (CSF) space (which could be difficult to measure), tethered cord, or postoperative condition of the lumbar spine were not included in the ALS and non-ALS groups.

Magnetic Resonance Imaging Acquisition

Magnetic resonance images were acquired using 1.5T or 3T clinical MRI scanners (MAGNETOM Avanto, MAGNETOM Vida, or MAGNETOM Skyra; Siemens, Erlangen, Germany). The imaging protocols for the axial and sagittal images of the turbo spin-echo T2WI used for the evaluation were as follows (transverse images were cross-sections parallel to the intervertebral disc):

  1. Axial T2WI on 1.5T (repetition time/echo time [TR/TE] = 3100/107 ms, flip angle = 170 degrees, slice thickness = 4 mm, section gap = 1.2 mm, matrix = 320 × 224, band width = 150 Hz/pixel, scan time = 2 minutes 24 seconds, field of view [FOV] = 180 × 180 mm).
  2. Sagittal T2WI on 1.5T (TR/TE = 4020/114 ms, flip angle = 170 degrees, slice thickness = 4 mm, section gap = 1.2 mm, matrix = 448 × 358, band width = 189 Hz/pixel, scan time = 2 minutes 6 seconds, FOV = 330–350 × 330–350 mm).
  3. Axial T2WI on 3T (TR/TE = 5000/95 ms, flip angle = 150 degrees, slice thickness = 4 mm, section gap = 1.2 mm, matrix = 384 × 257, band width = 303 Hz/pixel, scan time = 2 minutes 25 seconds, FOV = 200 × 200 mm).
  4. Sagittal T2WI on 3T (TR/TE = 4200/92 ms, flip angle 131 degrees, slice thickness = 4 mm, section gap = 1.2 mm, matrix = 448 × 314, band width 294 Hz/pixel, scan time = 1 minute 37 seconds, FOV = 330–350 × 330–350 mm).

Image Analysis

The anonymity of the cases was ensured before data evaluation to reduce bias. This allowed the evaluator to take measurements without knowing whether the patient had ALS at the time of measurement. The total cross-sectional area of the cauda equina at the L3 and L4 vertebral levels was retrospectively measured by 2 independent neuroradiologists with 15 or 20 years of experience using a software (volume analyzer SYNAPSE VINCENT; FUJIFILM, Tokyo, Japan) on the picture archival and communication system. The reason for choosing the L3 and L4 levels instead of the L1, L2, L5, and S1 levels was the need to ensure that the spinal cord and conus were not included within the measurement range and the desire to evaluate at a level where sufficient volumes of the nerve roots were present in the dural sac.

Although it would be better to measure the area of each single nerve root of the cauda equina, it is difficult to manually enclose the region of interest (ROI) of the single nerve root of the cauda equina because of its small size, and it is likely to be inaccurate and error prone. To avoid this, the total cross-sectional area of one section was evaluated. In the transverse T2WI at the L3 and L4 vertebrae levels, which were determined by checking the vertebral levels on the sagittal section of the T2WI, the ROIs were first manually enclosed to include the entire dural sac. At the same time, we were cautious to ensure that the dura and epidural structures were not included within the ROI. This ROI includes the total cross-sectional area of the cauda equina and the CSF, but the low signal of the nerve and the high signal of water on T2WI differ greatly, suggesting that the 2 areas can be separated.

On the histogram of the grayscale range of the dural sac content, where the horizontal axis is the signal value and the vertical axis is the area, we manually set the signal threshold to ensure that all the cauda equina were labeled yellow; all the CSF areas were not labeled, visually confirming the anatomy of the cauda equina (Fig. 1A). By integrating the area below the signal threshold (Fig. 1B), we measured the total cross-sectional area of the cauda equina in each section. The average of these values, measured by 2 independent evaluators, was taken as the final measurement.

F1
FIGURE 1:
Method for measuring the total cross-sectional area of cauda equina in axial T2-weighted images at the level of the L3 vertebral body of a 74-year-old woman with ALS. While visually confirming the anatomy of the cauda equina, we manually set thresholds in signals such that the cauda equina is labeled in yellow and the entire CSF region is not labeled, as seen in the upper row in (colored A). On the histogram of the grayscale range of the dural sac content, the horizontal axis is the signal value, and the vertical axis is the total transverse area. The total transverse area of the cauda equina in the transverse plane is measured by integrating the area below the above signal threshold (765, colored B). The actual measurement, in this case, was 64.35 mm2. The control axial T2-weighted image at the level of the L3 vertebral body of a 72-year-old man with lumbar spondylosis with spinal canal stenosis is shown for reference (C). The actual measurement was 99.05 mm2. Print version in grayscale. Figure 1 can be viewed online in color at www.jcat.org.

Statistical Analyses

All statistical analyses were performed using R version 3.6.1 (R Foundation for Statistical Computing, Vienna, Austria). The Mann-Whitney U test was used to compare the total cross-sectional area values between the ALS patients and non-ALS controls. The cutoff values, sensitivity, specificity, and area under the curve (AUC) of the total area of the L3 and L4 vertebral levels between the ALS and non-ALS patients were measured using the Youden index from the receiver operating characteristic curve. The AUC determinations were as follows: low accuracy (0.5–0.7), moderate accuracy (0.7–0.9), and high accuracy (0.9–1.0).

Two-way mixed class intraclass correlation coefficients (ICC) were calculated and evaluated for the interobserver reproducibility of the results of the 2 independently obtained measurements. The index values of ICC represent the following interpretations: poor (<0.2), fair (0.21–0.4), moderate (0.41–0.6), good (0.61–0.8), and excellent (0.81–1).

For the duration from onset to image acquisition and the total area of each L3 and L4 level, Spearman rank correlation coefficients (ρ) were calculated, and correlations were assessed. P values less than 0.05 were considered statistically significant.

RESULTS

Comparison of the ALS Patients and Non-ALS Controls and Receiver Operating Characteristic Curve Analysis

The total transverse area of the cauda equina in the ALS group at L3 (median, 66.73 mm2; range, 42.12–87.55 mm2; interquartile range [IQR], 62.83–70.57 mm2) was significantly smaller than that of the non-ALS group (median, 90.19 mm2; range, 77.79–99.86 mm2; IQR, 82.13–92.83 mm2; P < 0.001; Fig. 2), and the cutoff value was 76.95 mm2 (sensitivity, 1; specificity, 0.87; AUC, 0.96 [95% confidence interval {CI}, 0.893–1]; Fig. 3). The total transverse area of the cauda equina in the ALS group at L4 was significantly smaller than that of the non-ALS group (ALS: median, 52.9 mm2; range, 42.74–67.45 mm2; IQR, 46.3–57.72 mm2 vs non-ALS: median, 67.63 mm2; range, 56.35–82.26 mm2; IQR, 63.86–74.51 mm2; P < 0.001; Fig. 4), and the cutoff value was 61.04 mm2 (sensitivity, 0.8; specificity, 0.87; AUC, 0.94 [95% CI, 0.859–1]; Fig. 5). In summary, both measurements at the L3 and L4 levels were significantly smaller in the ALS group than in the non-ALS group, with high accuracy. In particular, the L3 level had higher accuracy in the ALS group compared with the L4 level.

F2
FIGURE 2:
Box-and-whisker plots of the total transverse area of cauda equina for ALS patients and non-ALS controls at the L3 vertebral level. ALS: median, 66.73 mm2; range, 42.12–87.55 mm2; IQR, 62.83–70.57 mm2. Non-ALS: median, 90.19 mm2; range, 77.79–99.86 mm2; IQR, 82.13–92.83 mm2.
F3
FIGURE 3:
Receiver operating characteristics curve of the total transverse area cutoff of the cauda equina for differentiating ALS patients and non-ALS controls at the L3 vertebral level. The cutoff value was 76.95 mm2 (sensitivity, 1; specificity, 0.87; AUC, 0.96 [95% CI, 0.893–1]).
F4
FIGURE 4:
Box-and-whisker plots of the total transverse area of cauda equina for ALS patients and non-ALS controls at the L4 vertebral level. ALS: median, 52.9 mm2; range, 42.74–67.45 mm2; IQR, 46.3–57.72 mm2. Non-ALS: median, 67.63 mm2; range, 56.35–82.26 mm2; IQR, 63.86–74.51 mm2.
F5
FIGURE 5:
Receiver operating characteristics curve of the total transverse area cutoff of the cauda equina for differentiating ALS patients and non-ALS controls at the L4 vertebral level. The cutoff value was 61.04 mm2 (sensitivity, 0.8; specificity, 0.87; AUC, 0.94 [95% CI, 0.859–1]).

There were no significant correlations between the total area of L3 and L4 levels and the duration from onset to image acquisition (L3: ρ = −0.03, P = 0.917; L4: ρ = −0.15, P = 0.615). In addition, in the case with no clinical evidence of lower motor neuron dysfunction in the lumbosacral spinal cord region, the total areas were 69.38 mm2 at the L3 level and 52.9 mm2 at the L4 level.

Interobserver Reproducibility

The ICCs among the evaluators of the measured values were excellent, with ICC = 0.88 at the L3 level and ICC = 0.89 at the L4 level.

DISCUSSION

The ALS is a progressive neurodegenerative disorder affecting the upper and lower motor neurons. This study showed that there was atrophy of the cauda equina in the ALS patients compared with non-ALS controls on MRI. There are few reports on the imaging evaluation of the peripheral nerves in ALS. However, there are reports of facial nerve atrophy,11 evaluations of the brachial plexus,12,13 lumbosacral plexus, and peripheral nerves in the upper and lower extremities14 on MRI, and cervical nerve root thinning, and median and ulnar nerves using ultrasound (US).15,16 However, clinically stable peripheral nerve assessment on US may be difficult to perform during the work-up for ALS patients because US is susceptible to the influence of the assessor's technical skills. Although routine lumbar MRI is expected to provide a relatively stable visualization of the cauda equina, to the best of our knowledge, there have been only 2 case reports of cauda equina with gadolinium enhancement on the lumbar roots.9,10 There is no report with a coherent assessment of the cross-sectional area of the cauda equina in ALS using MRI, and this study is the first investigation of this assessment.

The exact etiology of cauda equina atrophy is unclear because histological analysis was not performed in this study; however, several hypotheses may be considered, including the following. It is difficult to accurately assess the signal and volume of the ventral nerve roots, because of their microstructure, on current clinical routine lumbar MRI. However, a report on high-resolution magnetic resonance microimaging using the superoxide dismutase 1 transgenic mouse model for ALS showed a significant increase in signal changes and a decrease in the volume of ventral nerve roots of the lumbar spinal cord. The change was consistent with the axonal degeneration (axon damage and myelin loss) observed in the histological analysis.17 Axonal degeneration is thought to occur in the peripheral nerves in ALS; the possibility of secondary Wallerian degeneration associated with an impaired anterior horn of the spinal cord can be considered as the mechanism of axonal degeneration. In the 2 case reports of lumbar root with gadolinium enhancement,9,10 the primary immune response is less likely to cause a weak response to immunotherapy and glucocorticoid therapy, and the etiology of the gadolinium enhancement is proposed to be endothelial disruption and inflammation due to Wallerian degeneration. The possibility that similar pathogenesis may cause cauda equina atrophy should be considered. The second hypothesis is that the changes are due to primary neuroinflammation because primary neuroinflammation due to immune response has been suggested as an important factor in ALS pathogenesis.18 In addition, the first case report reported the accumulation of 43 kDa phosphorylated transactivation response DNA-binding protein, which is known as the histological hallmark of the central nervous system in ALS, in the cytoplasm of Schwann cells of the peripheral nerves, such as facial nerves, auditory nerves, spinal cord anterior roots, cauda equina, and peripheral nerves in the dorsal root ganglia.8 Thus, the possibility of primary peripheral neuropathy of the cauda equina and its consequent atrophy is also considered.

In contrast to multifocal motor neuropathy, there are also reports of brachial and lumbar plexus swelling in ALS12,13; however, there are conflicting reports on the presence of lumbosacral plexus or upper and lower extremity peripheral nerve swelling.14 One explanation for this contradiction might be that former studies restricted their analysis to the brachial and lumbar plexus, where resolution of single nerve fascicles of peripheral nerves is difficult.14 Unlike the previously mentioned studies targeting the brachial and lumbar plexus, this study used the cauda equina, which is easier to separate visually. It evaluated the total cross-sectional area of the cauda equina rather than a single nerve root. Therefore, we believe that we were able to evaluate the cross-sectional area of the nerve more accurately and establish that atrophy of the cauda equina is an important finding of ALS on routine lumbar MRI. In addition, several reports that concluded the atrophy of peripheral nerves other than the cauda equina, such as reports of facial atrophy on MRI11 and thinning of cervical nerve roots and median and ulnar nerves on US,15,16 also support our results.

The degree of atrophy may be related to the stage of the disease; however, there was no correlation between the disease duration and degree of atrophy in this study. Therefore, various factors beyond the time axis (eg, the presence or absence of damage to anterior horn cells at the level of the lumbosacral spinal cord or the cauda equina itself in individual cases and the degree of damage) may be involved. On the contrary, this may be potentially useful for early diagnosis because it does not rule out atrophy in the early stages of the disease.

We believe that cauda equina atrophy in ALS, which can be assessed using the widely available routine lumbar MRI, is a novel finding with high clinical utility that supports the diagnosis of ALS. We speculate that this finding may improve the accurate and early diagnosis of ALS to avoid inappropriate drug treatment or surgery due to misdiagnosis and can potentially be considered an additional biomarker.

In this study, the case with no clinical evidence of lower motor neuron dysfunction in the lumbosacral spinal cord region showed that both L3 and L4 levels were below the cutoff value. Thus, atrophy may be detected before the appearance of clinical findings. A large sample size is needed in future studies to investigate the association between this clinical finding and cauda equina atrophy.

This study has several limitations. First, it was a retrospective study, and the patient population was small because of the rare nature of the condition. Second, although the total transverse area of the cauda equina at the level of the L3 and L4 vertebrae was mainly considered, there are additional sensory fibers in this area, which may lead to a lower sensitivity. We also cannot exclude the possibility that pathology in the control group could potentially cause cauda equina atrophy. However, even under these limitations, we believe that the results of this study can provide useful information for clinical diagnosis, as significant atrophy and high accuracy were obtained in this study. There are also several limitations to the measurement methods. This study used 2-dimensional (2-D) images taken in daily clinical practice, because we do not take 3-D images regularly. Therefore, we cannot rule out the possibility that we measured the area of the oblique nerve. Hence, more accurate data may be obtained in the future by measuring arbitrary vertical cross-sectional areas relative to the nerve using 3-D images. In addition, concerning the measurement of the total cross-sectional area of the cauda equina in one section, the limitation was that the setting of the ROI and the measurement of the area were manual and visual. However, because the index values of ICC of the measurements are excellent, we believe that our results are reliable. Last, the images obtained at 1.5T and 3T were mixed. However, because the cauda equina and CSF have very different signal intensities on T2WI, even if the magnetic field strengths are different, it is considered to have minimal influence on the separation of these signals.

Although this study was limited to the evaluation of the cross-sectional area of the cauda equina in ALS, in the future, we would like to compare and correlate the cauda equina with various levels of spinal cord atrophy, as well as with various other levels of peripheral nerves that can be measured on MRI.

In conclusion, in this study, the cauda equina was significantly more atrophic in ALS than in non-ALS patients on routine lumbar MRI, with high accuracy. This could be a novel finding of great importance in diagnosing and evaluating ALS using routine MRI in clinical practice.

REFERENCES

1. Geevasinga N, Menon P, Scherman DB, et al. Diagnostic criteria in amyotrophic lateral sclerosis: a multicenter prospective study. Neurology. 2016;87:684–690.
2. Johnsen B, Pugdahl K, Fuglsang-Frederiksen A, et al. Diagnostic criteria for amyotrophic lateral sclerosis: a multicentre study of inter-rater variation and sensitivity. Clin Neurophysiol. 2019;130:307–314.
3. Turner MR, Grosskreutz J, Kassubek J, et al. Towards a neuroimaging biomarker for amyotrophic lateral sclerosis. Lancet Neurol. 2011;10:400–403.
4. Filippi M, Agosta F, Grosskreutz J, et al. Progress towards a neuroimaging biomarker for amyotrophic lateral sclerosis. Lancet Neurol. 2015;14:786–788.
5. Cheung G, Gawel MJ, Cooper PW, et al. Amyotrophic lateral sclerosis: correlation of clinical and MR imaging findings. Radiology. 1995;194:263–270.
6. Zhang L, Ulug AM, Zimmerman RD, et al. The diagnostic utility of FLAIR imaging in clinically verified amyotrophic lateral sclerosis. J Magn Reson Imaging. 2003;17:521–527.
7. Adachi Y, Sato N, Saito Y, et al. Usefulness of SWI for the detection of iron in the motor cortex in amyotrophic lateral sclerosis. J Neuroimaging. 2015;25:443–451.
8. Nakamura-Shindo K, Sakai K, Shimizu A, et al. Accumulation of phosphorylated TDP-43 in the cytoplasm of Schwann cells in a case of sporadic amyotrophic lateral sclerosis. Neuropathology. 2020;40:606–610.
9. Young NP, Laughlin RS, Sorenson EJ. Gadolinium enhancement of the lumbar roots in a case of ALS. Amyotroph Lateral Scler. 2010;11:207–209.
10. Luigetti M, Cianfoni A, Conte A, et al. Gadolinium enhancement of the lumbar leptomeninges and roots in a case of ALS. Amyotroph Lateral Scler. 2010;11:412–413.
11. Miyata M, Kakeda S, Hashimoto T, et al. Facial nerve atrophy in patients with amyotrophic lateral sclerosis: evaluation with fast imaging employing steady-state acquisition (FIESTA). J Magn Reson Imaging. 2020;51:757–766.
12. Gerevini S, Agosta F, Riva N, et al. MR imaging of brachial plexus and limb-girdle muscles in patients with amyotrophic lateral sclerosis. Radiology. 2016;279:553–561.
13. Staff NP, Amrami KK, Howe BM. Magnetic resonance imaging abnormalities of peripheral nerve and muscle are common in amyotrophic lateral sclerosis and share features with multifocal motor neuropathy. Muscle Nerve. 2015;52:137–139.
14. Kronlage M, Knop KC, Schwarz D, et al. Amyotrophic lateral sclerosis versus multifocal motor neuropathy: utility of MR neurography. Radiology. 2019;292:149–156.
15. Nodera H, Takamatsu N, Shimatani Y, et al. Thinning of cervical nerve roots and peripheral nerves in ALS as measured by sonography. Clin Neurophysiol. 2014;125:1906–1911.
16. Schreiber S, Abdulla S, Debska-Vielhaber G, et al. Peripheral nerve ultrasound in amyotrophic lateral sclerosis phenotypes. Muscle Nerve. 2015;51:669–675.
17. Cowin GJ, Butler TJ, Kurniawan ND, et al. Magnetic resonance microimaging of the spinal cord in the SOD1 mouse model of amyotrophic lateral sclerosis detects motor nerve root degeneration. Neuroimage. 2011;58:69–74.
18. Evans MC, Couch Y, Sibson N, et al. Inflammation and neurovascular changes in amyotrophic lateral sclerosis. Mol Cell Neurosci. 2013;53:34–41.
Keywords:

amyotrophic lateral sclerosis; cauda equina; magnetic resonance imaging; peripheral nerve

Copyright © 2022 Wolters Kluwer Health, Inc. All rights reserved.