Abstract
Objectives
In daily clinical practice, the assessment of the thickness of the cauda equina on lumbar spine magnetic resonance imaging is an important parameter. However, its relevance to the size of the dural sac in non-pathological conditions is unknown. To examine the relationship between the size of the dural sac and the apparent thickness of the cauda equina nerve root using lumbar spine magnetic resonance imaging in non-pathological conditions.
Methods
We retrospectively measured the dural sac diameter and vertebral body diameter, counted the apparent number, and calculated total cross-sectional area of the cauda equina, dural sac ratio and the area of one apparent nerve root of cauda equina in 100 cases. Spearman's rank correlation coefficient (ρ) was used.
Results
Dural sac ratio and diameter were positively correlated with the area of one apparent nerve root (ρ=0.77, P<0.001; ρ=0.74, P<0.001; respectively) and negatively correlated with the apparent number of cauda equina in a single cross-section (ρ=–0.63, P<0.001; ρ=–0.52, P<0.001; respectively).
Conclusions
A larger dural sac ratio and diameter was associated with an apparently thicker cauda equina and lower visible number. In a larger dural sac, the physiologically clumped and apparently thicker cauda equina should not be misdiagnosed as pathological.
Keywords: Cauda equina, dural sac diameter, dural sac ratio, dural ectasia, dural sac
Introduction
Lumbar spine magnetic resonance imaging (MRI) is frequently performed for various purposes in daily clinical practice, and it is common to evaluate images of the vertebral body, disc, spinal canal, spinal cord and cauda equina. When evaluating the cauda equina, the distribution, thickness characteristics, such as thickening and atrophy, space-occupying lesions and contrast effects, etc. are mainly evaluated.
Adhesive arachnoiditis should be considered if there is clumping or displacement, 1 and if it shows thickening, various diseases, such as neoplasm, inflammation, demyelination, deposition and metabolic diseases, need to be considered as a differential diagnosis. In addition, evaluation of the dural sac for spinal canal stenosis and dural ectasia is important to perform at the same time. In particular, dural ectasia may have underlying diseases such as Marfan syndrome,2–5 neurofibromatosis-1, 6 Ehler–Danlos syndrome, 3 Loeys–Dietz syndrome, 7 scoliosis 8 and ankylosing spondylitis. 9
There have been several reports quantitatively examining dural ectasia. 2 , 4 , 5 , 7 , 8 In particular, in recent reports, the dural sac ratio (DSR), based on vertebral and dural sac measurements, is often used to evaluate dural ectasia.
In daily clinical practice, we may experience cases in which the cauda equina appears to be thick when the dural sac tends to be wide, although not to the degree of a pathological dural ectasia. In such a scenario, it is essential to determine whether the thickening of the cauda equina is pathological.
Therefore, we hypothesise that the cauda equina appears thicker in cases of a wide dural sac under non-pathological conditions. To the best of our knowledge, there have been no reports on the relationship between the width of the dural sac and the apparent cauda equina using MRI. Therefore, in this study, we examined the relationship between the width of the dural sac and the apparent thickness of the cauda equina under non-pathological conditions.
Materials and methods
Study subjects
Our institutional review board approved this retrospective study (no. 32-031 (10106)). The need for informed consent was waived due to the retrospective nature of the study. All the procedures followed were in accordance with the ethical standards of the institution and with the Declaration of Helsinki and its subsequent revisions. The anonymity of the cases was ensured prior to evaluation of the data. Out of 2253 patients who underwent lumbar spine MRI including sagittal and transverse images of T2-weighted images (T2WI) at our institution, for various purposes between July 2018 and December 2019, 100 patients (43 men, 57 women, mean ± standard deviation (SD) 42.3 ± 13.26 years (range 9–69 years)) were included in the study who did not meet the following exclusion criteria.
Exclusion criteria
We excluded cases based on clinical information, history and imaging findings of conditions that could cause thickening of the cauda equina, adhesions within the dural sac, or conditions affecting the diameter of the dural sac or vertebral body to be measured.
The exclusion criteria were as follows: cauda equina syndrome; the increased signal intensity of nerve root cauda equina on T2WI; inflammatory, immunological, deposit/metabolic, neoplastic, or connective tissue disorders (meningitis, chronic inflammatory demyelinating polyneuropathy, Guillain–Barré syndrome, Charcot–Marie–Tooth disease, sarcoidosis, Sjogren’s syndrome, neoplastic lesions such as malignant lymphoma, meningeal dissemination and leptomeningeal gliomatosis/melanomatosis, Marfan syndrome, neurofibromatosis-1, Ehler–Danlos syndrome, Loeys–Dietz syndrome, Krabbe disease, metachromatic leukodystrophy, amyloidosis, etc.); scoliosis; spondyloarthropathy; ankylosing spondylitis; intraspinal canal lesion; spinal canal stenosis (due to spondylosis, hypertrophy of the ligamentum flavum, hypertrophic facet joint degeneration, ossification of posterior longitudinal ligament, disc bulging/protrusion/extrusion/sequestration, compression fracture, epidural lipomatosis and/or engorgement of epidural venous plexus, etc.); spondylodiscitis; spondylolisthesis; subarachnoid haemorrhage; cerebrospinal fluid (CSF) leakage; lumbar puncture; radiation therapy; trauma; surgery for the intervertebral disc, vertebrae, or spinal canal; adhesive arachnoiditis; dural arteriovenous fistula; and deformity of vertebra (such as scalloping of the vertebral body, compression fracture and congenital deformity).
The presence of the above clinical findings and medical history were confirmed from the medical records, and cases with the above imaging findings were excluded by two board-certified diagnostic radiologists of the Japan Radiological Society (SM and TS) with more than 15 years of experience. In addition, we excluded cases that were difficult to measure due to image quality degradation caused by artefacts.
MRI acquisition
Magnetic resonance images were acquired using 1.5 T (n=61) or 3 T (n=39) clinical MRI scanners (MAGNETOM Avanto or MAGNETOM Skyra; Siemens, Erlangen, Germany).
The imaging protocol of the axial and sagittal images of the turbo spin-echo T2WI used for the measurements was as follows. The axial views were a cross-sectional view parallel to the vertebral body and the intervertebral disc.
Axial T2WI on 1.5 T (TR/TE 3100/107 ms, flip angle 170°, slice thickness 4 mm, section gap 1.2 mm, matrix 320 × 224, band width 150 Hz/pixel, scan time 2 min 24 s, field of view (FOV) 180 × 180 mm).
Sagittal T2WI on 1.5 T (TR/TE 4020/114 ms, flip angle 170°, slice thickness 4 mm, section gap 1.2 mm, matrix 448 × 358, band width 189 Hz/pixel, scan time 2 min 6 s, FOV 330 ∼ 350 × 330 ∼ 350 mm).
Axial T2WI on 3 T (TR/TE 5000/95 ms, flip angle 150°, slice thickness 4 mm, section gap 1.2 mm, matrix 384 × 257, band width 303 Hz/pixel, scan time 2 min 25 s, FOV 200 × 200 mm).
Sagittal T2WI on 3 T (TR/TE 4200/92 ms, flip angle 131°, slice thickness 4 mm, section gap 1.2 mm, matrix 448 × 314, band width 294 Hz/pixel, scan time 1 min 37 s, FOV 330 ∼ 350 × 330 ∼ 350 mm).
Image analysis
Two independent board-certified diagnostic radiologists (SM and TS) retrospectively measured and calculated the following using software on the Picture archival and communication system (volume analyser Synapse Vincent; Fujifilm, Tokyo, Japan). These two independent radiologists were the same two individuals who confirmed the clinical findings, medical history and exclusive imaging findings.
The following 1–3 measurements, which are often used in the evaluation of dural ectasia as described above, 2 , 4 , 5 , 7 , 8 were used in the present study as well.
Dural sac diameter (DSD) (Figure 1): The antero-posterior diameters of the dural sac perpendicular to the long axis of the spinal canal were measured at the level of the L4 vertebra at the midline-sagittal section of the T2WI. During the measurement, the antero-posterior diameters of sites of high signal due to CSF were measured, being careful not to include epidural fat that is lower in signal than CSF.
Vertebral body diameter (VBD) (Figure 1): In the midline-sagittal section of T2WI, the antero-posterior diameters of the middle of the L4 vertebral body were measured at the same level of the measurements of the DSD as described above.
DSR=DSD/VBD. Based on the above results, the DSR was calculated as DSD/VBD.
Figure 1.
Measurement method of dural sac diameter (DSD) and vertebral body diameter (VBD). The anteroposterior diameter of the central part of the L4 vertebra was measured as VBD on the midline-sagittal section of the T2-weighted images. The anteroposterior diameter of the dural sac perpendicular to the spinal canal at the same level was measured as DSD.
In this study, to evaluate the thickness, the area per one apparent nerve root of cauda equina was considered. However, because one of the nerve roots of the cauda equina is a microstructure, it is difficult and error prone to measure the transverse area by enclosing the region of interest (ROI) manually. Therefore, the thickness of the one apparent nerve root of the cauda equina was measured by the following method.
Total area of cauda equina in a single cross-section (TACE) (Figure 2(a and b))
Figure 2.
Measurement of total area of cauda equina in a single cross-section (TACE). The region of interest was first manually enclosed to include the entire dural sac (a). On a histogram of grayscale range of this dural sac content with signal value on the horizontal axis and area on the vertical axis, as shown in (b), the total cross-sectional area of the cauda equina in one cross-section was measured by setting the signal threshold and integrating the area of the signal below the threshold. With respect to the threshold setting, we determined this value by visually checking the values such that all cauda equina were labelled and all areas of the cerebrospinal fluid were not labelled, as shown in (a).
In the transverse image on T2WI at the L4 vertebral level, the ROI was first manually enclosed to include the entire dural sac (Figure 2(a)). At this time, we enclosed the ROI with great care to prevent dural and epidural structures from entering the ROI. Within this ROI, the total cross-sectional area of the cauda equina and the cross-sectional area of the CSF were included. Because the signal intensities on T2WI differ markedly between the low signal of the nerve root and the high signal of the CSF, the area of the two can be separated.
On the histogram showing the grayscale range of the dural sac content with the signal value on the horizontal axis and the area on the vertical axis, the total cross-sectional area of the cauda equina was measured by setting the signal threshold and integrating the area below the threshold (Figure 2(b)), so that all the cauda equina were labelled and all the CSF areas were not labelled (Figure 2(a)) while visually confirming the anatomy of the cauda equina. The thresholds for segmentation of the cauda equina were determined manually by two independent diagnostic radiologists using the afore-mentioned software (volume analyser Synapse Vincent; Fujifilm, Tokyo, Japan).
Apparent number of cauda equina in a single cross-section (ANCE)
In the same T2WI transverse image as above, we manually counted the total number of cauda equina by defining an apparent single nerve root that could be visually contoured as a round or oval shape. However, when it was difficult to count due to obliquity or clumping of nerve roots, the counting was done by referring to the upper and lower slices.
Area of one apparent nerve root of cauda equina in a single cross-section (AOCE)
Based on the above measurements and counts, the TACE/ANCE was calculated and given as AOCE.
For these values, the average values of two independent evaluators were used as the final values.
Statistical analyses
IBM SPSS statistics software version 25 was used for the statistical analysis. Two-way mixed-class intraclass correlation coefficients (ICCs) were calculated and evaluated for the inter-observer reproducibility of the results of the two 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). In addition, Spearman's rank correlation coefficients (ρ) were calculated for DSR and their respective TACE, ANCE, and AOCE to evaluate the correlation. For DSD and VBD, the correlations of the above three items were also evaluated by calculating the ρ value, respectively. P values <0.05 were considered statistically significant.
Results
Inter-observer reproducibility
The inter-observer reproducibility of the measured values was excellent for DSD, VBD, TACE and ANCE. The respective ICC for DSD was 0.963 (P<0.001), that for VBD was 0.954 (P<0.001), that for TACE was 0.938 (P<0.001) and that for ANCE was 0.92 (P<0.001).
Results and correlations of measurements
The mean ± SD of the respective measurements were 13.56 ± 2.2 mm for DSD, 31.83 ± 2.974 mm for VBD, 65.88 ± 9.046 mm2 for TACE and 18 ± 2.96 for ANCE. Based on these, the calculated DSR and AOCE were 0.43 ± 0.083 and 3.707 ± 0.66 mm2, respectively. There was a positive correlation between DSR and AOCE (ρ=0.77, P<0.001). The correlation between DSR and TACE was weak (ρ=0.24, P=0.018), and there was a negative correlation between DSR and ANCE (ρ=-0.63, P<0.001) (Figure 3(a–c)).
Figure 3.
A representative scatter plot of the results of the correlations associated with dural sac ratio (DSR). (a) There was a positive correlation between DSR and the area of one apparent nerve root of cauda equina in a single cross-section (AOCE) (ρ=0.77, P<0.001). (b) The correlation between DSR and the total area of cauda equina in a single cross-section (TACE) was weak (ρ=0.24, P=0.018). (c) There was a negative correlation between DSR and the apparent number of cauda equina in a single cross-section (ANCE) (ρ=–0.626, P<0.001).
Similarly, in DSD we found a positive correlation with AOCE (ρ=0.74, P<0.001). The correlation with TACE was weak (ρ=0.35, P<0.001). There was a negative correlation with ANCE (ρ=–0.52, P<0.001) (Figure 4(a–c)).
Figure 4.
A representative scatter plot of the results of the correlations associated with dural sac diameter (DSD). (a) We found a correlation between DSD and the area of one apparent nerve root of cauda equina in a single cross-section (AOCE) (ρ=0.74, P<0.001). (b) The correlation between DSD and the total area of cauda equina in a single cross-section (TACE) was weak (ρ=0.35, P<0.001). (c) There was a negative correlation between DSD and apparent number of cauda equina in a single cross-section (ANCE) (ρ=–0.522, P<0.001).
VBD was weakly correlated with AOCE, TACE and ANCE (ρ=–0.27, P=0.006, ρ=0.18, P=0.073 and ρ=0.37, P<0.001, respectively). These results suggest that the larger the DSR and DSD, the thicker the apparent cauda equina nerves appear to be, with a smaller number of them. Images of a representative case illustrating these findings and those of a control case are shown in Figure 5(a–d).
Figure 5.
Images of representative case and control case. (a, b) Control case of a 29-year-old man. Measurements are dural sac diameter (DSD) 12.54 mm, dural sac ratio (DSR) 0.42, apparent number of cauda equina in a single cross-section (ANCE) 24, area of one apparent nerve root of cauda equina in a single cross-section (AOCE) 2.41 mm2. (c, d) A representative case of a 49-year-old woman. In the sagittal image of T2-weighted images (T2WI) (c) compared to that of the control case (a), the dural sac was visually depicted more widely (DSD 15.17 mm, DSR 0.54). In the transverse image of T2WI (d) compared to that of the control case (b), the cauda equina nerve roots appear obviously thicker and are depicted in smaller numbers (ANCE 12.5, AOCE 4.96 mm2).
Discussion
In the present study, we hypothesised that in the absence of pathological dural sac enlargement and true thickening of the cauda equina, the cauda equina may appear thicker in cases visually observed with a wide dural sac. To the best of our knowledge, there have been no reports of investigations on the association between the width of the dural sac and the apparent thickness of the cauda equina, and this study is the first examination of this association.
In this study, the correlation between DSR/DSD and AOCE was positive, while the correlation between DSR/DSD and TACE was weak. There was a negative correlation between DSR/DSD and ANCE. These results suggest that the larger DSR and DSD were associated with the thicker apparent cauda equina and the smaller apparent number of cauda equina.
Although the detailed aetiology of this is unclear, we speculate the following possibilities. As a first hypothesis, dural ectasia occurs in Marfan syndrome due to relatively high CSF pressure in the terminal portion of the spinal canal. 10 Thus, it is presumed that the pressure of CSF is slightly higher physiologically when the dural sac is wide, even in physiological conditions. Therefore, when the dural sac is wider, it is expected that the cauda equina will congregate in non-inflammatory conditions due to the circumferential pressure on the nerve roots caused by increased CSF pressure. Therefore, we speculate that a nerve root may be depicted as a thickened single nerve root in appearance, with an apparent decrease in the number of cauda equina in one transverse image on MRI, although it is a state in which thin cauda equina is actually gathered together without reduction of the true number. In addition, there was a negative correlation between DSR/DSD and ANCE; however, the degree of correlation was weaker than that between DSR/DSD and AOCE. It could be speculated that this may include an element of the original individual differences in the number of cauda equina.
A second hypothesis is the possibility of thickening of the cauda equina associated with venous perfusion disturbances. In the presence of spinal canal stenosis, it has been reported on dynamic MRI that the cauda equina has a slower contrast effect in the arterial phase and a residual contrast effect in the venous phase and that circulatory disturbances in the cauda equina are present. 11 In this report, impaired perfusion in the venous system rather than in the thick-walled high-pressure arteries is considered as the contributing factor. 11 Although such spinal canal stenosis is not present in the present study, a similar phenomenon speculates that minor venous oedema may result from minor venous perfusion impairment, caused by mechanical pressurisation of the cauda equina due to increased CSF pressure. Therefore, it is possible to speculate that the cauda equina may appear thicker. However, cases presenting with an increased signal intensity of the cauda equina suggestive of venous perfusion impairment on the T2WI were initially excluded, and it is difficult to explain the negative correlation between the width of the dural sac and the apparent number of cauda equina in this hypothesis, even if it was assumed that cases presenting with nerve enlargement that did not show increased signal intensity on imaging were included.
From the above, we believe that the first hypothesis is more probable.
The present results may reflect the fact that when the dural sac is physiologically wider, the nerve roots of cauda equina may be more aggregated and appear thicker on MRI. If the dural sac is wide, and thickening of the cauda equina is suspected, the possibility of a physiological change should be considered, rather than misidentifying it as a pathological condition, such as adhesive arachnoiditis or a tumour, inflammation, demyelination, or deposition/metabolic disease that causes thickening of the cauda equina.
If these findings are widely recognised by radiologists, neurologists, neurosurgeons and orthopaedic surgeons who evaluate MRI in daily clinical practice, such misdiagnosis and the implementation of unnecessary additional examination and treatment can be avoided. We believe that these are important novel findings to be aware of when evaluating lumbar spine MRI in daily clinical practice.
Our study has several limitations. Concerning the study subjects, the possibility of having potential lesions, such as adhesion, cannot be completely excluded because the subjects were not perfectly healthy volunteers. However, we have set many exclusion criteria and attempted to exclude as many patients as possible, and this influence is expected to be low.
Next, there are limitations regarding the evaluation methods. First, this was a retrospective study using two-dimensional images that were used in daily clinical practice, and we have not performed any three-dimensional measurements. Therefore, we cannot rule out the possibility that the area of the oblique nerve was measured. Hence, more accurate data may be obtained in the future by measuring an arbitrary vertical cross-sectional area of a nerve using three-dimensional images.
In addition, another limitation was the marking of the ROI and the measurement of the area were manual and visualised for the measurement of the total cross-sectional area of the cauda equina in a single transverse section. The apparent number of cauda equina was counted with reference to the upper and lower slices to be as accurate as possible; however, it was also a visual assessment.
We believe that these values are reliable because the inter-observer reproducibility is high, even though they may have been influenced by subjective elements of the observer.
Finally, another limitation was that we did not distinguish the filum terminale in this study because it was difficult to identify accurately and separate a single thin normal filum terminale in two-dimensional images for evaluation. Therefore, the filum terminale has been included within the total area and apparent number of cauda equina. However, even if a single thin filum terminale was included, we speculate that it would account for only a small percentage and have a very small effect on the assessment.
Conclusion
In cases with a physiologically wide dural sac, the nerve roots of cauda equina may be more aggregated and may appear thicker on MRI. These are important new findings that should be considered when evaluating the cauda equina using routine MRI. This should be considered when evaluating lumbar spine MRI and should not be misinterpreted as pathological cauda equina thickening or adhesive arachnoiditis.
Footnotes
Conflict of interest: The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Ethics approval: This retrospective study was approved by the ethics committee of the Jikei University School of Medicine (no. 32-031 (10106)). All procedures performed were in accordance with the ethical standards of the institution and with the Declaration of Helsinki and its later amendments. The need for informed consent was waived due to the retrospective nature of the study. This article does not contain any studies with animals performed by any of the authors.
Funding: This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
ORCID iDs: Satoshi Matsushima https://orcid.org/0000-0002-9940-5205
Akira Baba https://orcid.org/0000-0001-6913-5307
References
- 1.Anderson TL, Morris JM, Wald JT, et al. Imaging appearance of advanced chronic adhesive arachnoiditis: a retrospective review. AJR Am J Roentgenol 2017; 209: 648–655. [DOI] [PubMed] [Google Scholar]
- 2.Oosterhof T, Groenink M, Hulsmans FJ, et al. Quantitative assessment of dural ectasia as a marker for Marfan syndrome. Radiology 2001; 220: 514–518. [DOI] [PubMed] [Google Scholar]
- 3.Villeirs GM, Van Tongerloo AJ, Verstraete KL, et al. Widening of the spinal canal and dural ectasia in Marfan's syndrome: assessment by CT. Neuroradiology 1999; 41: 850–854. [DOI] [PubMed] [Google Scholar]
- 4.Chuzel Q, Dupuis-Girod S, Rousset M, et al . Assessment of dural ectasia using computed tomodensitometry as a criterion in Marfan syndrome. J Comput Assist Tomogr 2019; 43: 282–287. [DOI] [PubMed] [Google Scholar]
- 5.Lundby R, Rand-Hendriksen S, Hald JK, et al. Dural ectasia in Marfan syndrome: a case control study. AJNR Am J Neuroradiol 2009; 30: 1534–1540. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Razek AAKA. MR imaging of neoplastic and non-neoplastic lesions of the brain and spine in neurofibromatosis type I. Neurol Sci 2018; 39: 821–827. [DOI] [PubMed] [Google Scholar]
- 7.Kono AK, Higashi M, Morisaki H, et al. Prevalence of dural ectasia in Loeys–Dietz syndrome: comparison with Marfan syndrome and normal controls. PLoS One 2013; 8: e75264. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Abul-Kasim K, Overgaard A, Ohlin A. Dural ectasia in adolescent idiopathic scoliosis: quantitative assessment on magnetic resonance imaging. Eur Spine J 2010; 19: 754–759. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Liu CC, Lin YC, Lo CP, et al. Cauda equina syndrome and dural ectasia: rare manifestations in chronic ankylosing spondylitis. Br J Radiol 2011; 84: e123–e125. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Fishman EK, Zinreich SJ, Kumar AJ, et al . Sacral abnormalities in Marfan syndrome. J Comput Assist Tomogr 1983; 7: 851–856. [DOI] [PubMed] [Google Scholar]
- 11.Kobayashi S, Suzuki Y, Meir A, et al. Circulatory dynamics of the cauda equina in lumbar canal stenosis using dynamic contrast-enhanced magnetic resonance imaging. Spine J 2015; 15: 2132–2141. [DOI] [PubMed] [Google Scholar]