Dynamic Effects on the Lumbar Spinal Canal

[문형철]

재미있게 읽었던 논문

추간판에 압박을 가하고 mri를 촬영하면??

panic bird........

Dynamic Effects on the Lumbar Spinal Canal : Axially Loaded CT-Myelography and MRI in Patients With Sciatica and/or Neurogenic Claudication
  
ⓒ Lippincott-Raven Publishers
--------------------------------------------------------------------------------
Volume 22(24)             15 December 1997             pp 2968-2976
--------------------------------------------------------------------------------

[Clinical Studies - Diagnosis]
Willen, Jan MD, PhD*; Danielson, Barbro MD, PhD†; Gaulitz, Arne MD†; Niklason, Thomas†; Schonstrom, Nils MD, PhD‡; Hansson, Tommy MD, PhD§

From the *Department of Orthopaedics, and the †Department of Radiology, Sahlgrenska University Hospital, Molndal, the ‡Department of Orthopaedics, Ryhov Hospital, Jonkoping, and the §Department of Orthopaedics, Sahlgrenska University Hospital, Goteborg, Sweden.
Supported, in part, by research grants from the Foundation for the Memory of the Consul Thure Carlsson, Lund, and FoU, Bohuslan County Council, Goteborg, Sweden.
Acknowledgment date: July 11, 1997.
Acceptance date: July 11, 1997.
Device status category: 7.
Address reprint requests to: Jan Willen, MD, PhD; Department of Orthopaedics; Sahlgrenska University Hospital/Molndal; 41345 Gotborg; SWEDEN.
--------------------------------------------------------------------------------
Abstract
Study Design. In patients with sciatica or neurogenic claudication, the structures in and adjacent to the lumbar spinal canal were observed by computed tomographic myelography or magnetic resonance imaging in psoas-relaxed position and during axial compression in slight extension of the lumbar spine.

Objectives. To determine the mechanical effects on the lumbar spinal canal in a simulated upright position.

Methods. Portable devices for axial loading of the lumbar spine in computed tomographic and magnetic resonance examinations were developed. 50 patients (94 sites) were studied with computed tomographic myelography, and 34 patients (80 sites) with magnetic resonance in psoas-relaxed position followed by axial compression in slight extension. The dural sac cross-sectional area at L2 to S1, the deformation of the dural sac and the nerve roots, and the changes of the tissues surrounding the canal were observed.

Results.
In 66 of the investigated 84 patients, there was a statistically significant reduction of the dural sac cross-sectional area in at least one site during axial compression in slight extension.
Of the investigated patients, 29 passed the borderlines for relative (100 mm2) or absolute stenosis (75 mm2) in 40 sites. In 30 patients, there was deformation of the dural sac in 46 sites.
In 11 of the patients investigated with magnetic resonance imaging, there was a narrowing of the lateral recess in 13 sites, during axial compression in slight extension.

Conclusions. Axial loading of the lumbar spine in computed tomographic scanning and magnetic resonance imaging is recommended in patients with sciatica or neurogenic claudication when the dural sac cross-sectional area at any disc location is below 130 mm2 in conventional psoas-relaxed position and when there is a suspected narrowing of the dural sac or the nerve roots, especially in the ventrolateral part of the spinal canal in psoas-relaxed position.
The diagnostic specificity of the spinal stenosis will increase considerably when the patient is subjected to an axial load.
--------------------------------------------------------------------------------

Pain and neurologic symptoms in the lower extremities elicited while walking, standing, and maintaining certain postures are pathognomonic for lumbar spinal stenosis. Conversely, forward flexion of the spine, squatting, and lying supine with slight hip flexion typically relieve the symptoms.9,11,14,15

It is well known and accepted that the available space within the lumbar spinal canal decreases with extension, axial loading, and increases in forward flexion.14,15 Traditionally, the myelographic examination in suspected encroachment of the lumbar spinal canal includes exposures in standing position and in flexion and extension of the spine.14,15 The rationale for the upright examination is to provoke the narrowing of the spinal canal known to occur during "normal" loading conditions. These conditions might remain undetected if only the supine examination position is used.

To improve diagnostic precision, in addition to a myelogram, Penning and Wilmink introduced the computed tomographic (CT) study, including flexion and extension exposures of the lumbar spine in the supine position.10 They showed convincingly that extension causes constriction of the dural sac and compression of the nerve roots, whereas flexion of the spine relieves the constriction and the root compression.

In a human cadaveric CT study, the current investigators have shown that the cross-sectional area of the lumbar spinal canal significantly decreases in extension and increases in flexion.13 Axial compression and distraction of the tested spines had the same effect on the size of the spinal canal. The most significant factor affecting the dimensions of the canal was the disc. These results confirm‎ the clinical observation that a change in posture from supine to standing causes considerable narrowing of the spinal canal.

The minimum space within the lumbar spinal canal needed to accommodate the cauda equina without any pressure has been determined experimentally by Schonstrom et al.12 The critical value for the cross-sectional area at which pressure was building up between the nerve roots at L3-L4 was approximately 75 mm2. These results indicated that the constriction of the cauda equina to a size less than 75 mm2 at L3-L4 will affect the normal function of the nerve roots.

Detailed analysis of the content of the spinal canal requires CT scan (with or without myelography) or magnetic resonance imaging (MRI) study. So far, these examinations have been performed with the patient in a supine, psoas-relaxed position (PRP). This procedure results in a flexion of the lumbar spine and consequently in a widening of the spinal canal.9,10 In addition, the supine position reduces the compressive load of the spine approximately 75% when compared with the spinal load in upright standing.6

Because CT and MRI examinations are usually performed with the patient supine, dynamic studies and studies of the spine during "normal" loading conditions are restricted. Instead of creating a symptom-provoking examination position, the standard positioning of the patient during CT and MRI examinations today probably has the opposite effect.

The purpose of the current study was to eval‎uate the effects of axial loading of the lumbar spine during CT and MRI examinations of patients with suspected stenosis of the lumbar spinal canal.17

Methods and Materials
The Computed Tomographic Study.
Each patient in this study was first examined in conventional supine PRP (Figure 1), followed by examination in a supine, slightly extended position during axial compression (ACE, Figure 2) in the gantry of the CT scanner in an attempt to mimic the spine position in a standing subject. The applied load was measured by a balance placed at a foot plate connected to straps between a special patient harness and the foot plate. The device was not fixed to the CT scanner table. Special effort was applied to the construction of the harness concentrating the applied load to the lower part of the chest (Figure 2). The straps were intended to follow the gravity line when loaded.

Figure 1. Patient in a conventional psoas-relaxed position on the examination table.

Figure 2. Patient in supine, slightly extended position during axial compression, resting on a low-friction carpet on the examination table. The patient harness is connected to a foot plate by straps, and the axial load is measured by a balance at the foot plate.

The axial load applied was approximately half the body weight (300-400 N). Eight patients were also examined during extension without compression loading. During loading, all patients were positioned on a low-friction carpet to minimize friction against the CT scanner table.

Before the CT examination, 3-6 mL iohexol 180 mg/mL was injected intrathecally. To mix the contrast with the spinal fluid, the patients were told to turn around and to rest prone for a few minutes. The CT examination was performed on a Siemens Somatom Plus (Siemens, Erlangen, Germany). The exposing parameters were 137 kV and 580 mA. The slice thickness was 3 mm with no overlapping. The transverse slices were chosen parallel to the lower endplate of the vertebra above the investigated disc (Figure 3A).

Figure 3. A, Computed tomographic scan slices chosen were parallel to the lower vertebral endplate. B, The box for transverse magnetic resonance slices was placed parallel to the disc.

The Magnetic Resonance Study.
A corresponding loading device was used for MRI examinations (Figure 2). The MRI device was made entirely of nonmagnetic materials and was later used also in CT examinations.

The MRI examination was executed on a 1.0 Tesla system (Magnetom, Siemens) using a surface coil. All patients were examined with sagittal T1- and T2-weighted turbo spin echo (TSE) images and axial T1- (18 patients) or T2-weighted (16 patients) TSE images with 4 mm slices. The box for transverse slices was placed parallel to the disc and as close to the same position as possible in PRP and during ACE (Figure 3B). After applying the axial load, another transverse study was performed.

Study Parameters.
The cross-sectional area of the dural sac (D-CSA) was measured on the slices through the central part of the disc in all positions by using a standard measurement program in the CT scanner. One of the authors (BD) performed three measurements on each image. The mean CSA and changes in CSA were calculated between the two positions in each patient. To ensure that the slices were made at the same site in PRP and during ACE, the radiologist looked at surrounding tissues-the nerve roots outside the dural sac, the facet joints, and other bone structures.

According to experimental 12 and clinical studies 2,3 D-CSA values below 100 mm2 (equivalent to an AP-diameter of 11.5 mm2 of the dural sac) at L3-L4 indicate a relative central stenosis, whereas values below 75 mm2 indicate an absolute central lumbar spinal stenosis.

Computed Tomographic Study Subjects.
During 1993 and 1994, 50 patients (95 sites) with clinical symptoms and signs of sciatica or neurogenic claudication were investigated with the described CT-myelographic technique. There were 18 women and 32 men with a mean age of 46 years (range, 26-71 years). The mean history of complaints was 3.3 years (range, 0.5-15 years). The patients were numbered consecutively (CT1-CT50).

The clinical data on investigated patients with a statistically significant D-CSA reduction equal to or less than 100 mm2 in one or more sites during ACE are described in Table 1. - 총50명중 18명.

Table 1. Computed Tomography/Myelography

Magnetic Resonance Study Subjects.
During 1994 and 1995, 34 patients (80 sites) were investigated by the MRI technique. There were 18 men and 16 women with a mean age of 50 years (range, 25-74 years). The mean history of complaints was 3.4 years (range, 0.5-10 years). The patients were numbered consecutively (MR1-MR34). The clinical data of the investigated patients with D-CSA equal to or less than 100 mm2 in one or more sites at ACE are described in Table 2. - 총 34명중 11명

Table 2. Magnetic Resonance Imaging

In all 84 patients (175 sites) included in the CT and MRI studies, the following recordings were made in PRP and during ACE:

* D-CSA at each disc and lateral recess;
* deformation of the dural sac from the ventral, dorsal, and lateral side;
* disc herniations;
* foraminal narrowing;
* facet joint arthrosis;
* lateral recess narrowing; and
* elicited leg pain during ACE.
The methods used in the current study were approved by the Ethical Committee at the University of Goteborg.

Reproducibility and Statistical Power.
A study of the measurement error was performed. Six patients of an independent radiologist (AG) were selected. The D-CSA in PRP and ACE was determined in these patients on five occasions with an interval of at least 4 weeks. On every occasion, three measurements were performed. The standard deviation of the measurement error was 6.5 mm2. Statistically, the difference in the D-CSA between the PRP and during ACE had to exceed 15 mm2 to imply a significant difference on a 5% level with the described prerequisites.

Results
The Computed Tomographic Study
In 50 patients, a total of 95 sites from L3 to S1 were analyzed. In 8 patients, at the beginning of the CT-myelographic study, examined in PRP, during ACE, and during extension without compression loading, a statistically significant decrease was found of the D-CSA during ACE at 14 sites. The mean reduction of the D-CSA in these sites from PRP to E was (±SD) 15 ± 15 mm2; range, 1-48 mm2; and from PRP to ACE was 40 ± 18 mm2; range, 17-72 mm2. On the basis of these results, the examination during extension without loading was not performed in the remaining patients.

The D-CSA ranged from 300 to 50 mm2 in PRP.
In 55% of all investigated sites, there was a statistically significant reduction of the D-CSA between PRP and ACE (Figure 4).
Four measurements higher than 130 mm2 in PRP were reduced to or were lower than 100 mm2 during ACE. One PRP value was reduced from 186 to 99 mm2 in ACE at L5-S1 (Figure 5).
Forty of the investigated 50 patients had a D-CSA reduction at one site, at least, between PRP and ACE.

Figure 4. Graphic illustration of dural sac cross-sectional area changes in all sites investigated by computed tomographic scanmyelography. Filled squares represent the values in the psoasrelaxed position, and open circles represent values in axial compression in slight extension.

Figure 5. Note the concentric narrowing of the dural sac with cross-sectional area reduction from 186 mm2 in the psoas-relaxed position (A) to 99 mm2 during axial compression in slight extension (B). Further note the configuration change of the S1 nerve root (sleeve) on the right side.

In 18 patients (24 sites), there was a true decrease of the D-CSA between PRP and ACE, to lower than the suggested critical values for a relative or absolute central stenosis in one or two sites (Table 1). Among these patients, deformation of the dural sac was observed in 30 sites.

The Magnetic Resonance Study.
In 34 patients, a total of 80 sites were analyzed. The D-CSA ranged from 313 to 37 mm2 in PRP. In 50% of all investigated sites there was a statistically significant difference between measurements in PRP and ACE. No PRP value higher than 130 mm2 was reduced below 100 mm2 during ACE (Figure 6). Twenty-six of the 34 patients had a significant reduction of the D-CSA at one or more sites between PRP and ACE.

Figure 6. Graphic illustration of the dural sac cross-sectional area changes in all sites investigated by magnetic resonance imaging. Filled squares represent values in psoas-relaxed position and open circles represent values in axial compression in slight extension.

In 11 patients (16 sites), there was a statistically significant decrease of the D-CSA between PRP and ACE, to lower than the critical values for stenosis (relative or absolute) in one or two sites (Table 2). Among these patients, deformation of the dural sac was observed in 16 sites. In 11 patients, narrowing of the lateral recess was noted in 13 sites.

In the CT and MRI studies, 29 patients showed a significant D-CSA reduction, to lower than 100 mm2 at one or two sites (total, 40 sites; Table 3). The site at L4-L5 was the most frequently involved.

Table 3. D-CSA Reduction at ACE

During ACE, deformation of the dural sac and free nerve roots occurred at the disc and lateral recess, according to the schematic shown in Figure 7, A and B. Examples of changes in the spinal canal during MRI examinations in PRP and ACE are illustrated in Figures 8 and 9. Narrowing of the lateral recess and deformation of nerve roots could be estimated in the MRI study but not measured. The ventral part of the fat pad generally changed configuration from concave to convex at several sites during ACE and tended to move anteriorly, causing an impression in the dural sac. Estimation of the increased narrowing of the intervertebral foramen was not performed, because sagittal images could not be exactly reproduced in PRP and during ACE.

Figure 7.
A, Structure changes in and adjacent to the lumbar spinal canal at L4-L5 in psoasrelaxed position (PRP) and during axial compression in slight extension (ACE). DS = dural sac; L = ligamentum flavum, F = fat pad; UAP = upper articular process L5; LAP = lower articular process L4; N4 = nerve L4; N5 = nerve root sleeve L5.
B, An illustration of the structural changes in and adjacent to the lumbar spinal canal at recess L4-L5. Note disc bulge, thickened ligamentum flavum, and changed configuration of the dorsal fat pad during axial compression in slight extension. N5 = nerve root L5.

Figure 8. Dynamic magnetic resonance investigation in a 55-year-old woman with a 2-year history of neurogenic claudication. The dural sac cross-sectional area in L4-L5 decreased from 88 mm2 in psoas-relaxed position (A) to 55 mm2 during axial compression in slight extension (B) with simultaneous compression of the L5 nerve roots, changed configuration of the fat pad, and thickening of the ligamenta flava. There was significant change of the dural sac cross-sectional area at L5-S1. See patient MR 14 in Table 2.

Figure 9. Dynamic magnetic resonance investigation in a 42-year-old woman with a 6-month history of neurogenic claudication. She had sciatica when standing but improved with forward flexion of the lumbar spine. The spinal canal was narrowed only at L4-L5. The dural sac cross-sectional area was 73 mm2 in psoas-relaxed position (A) and decreased to 55 mm2 in axial compression in slight extension (B). Note the changed fluid signal and the narrowing of the lateral recess in the anterior part of the spinal canal (patient MR 29, Table 2).

Disc herniation was seen at 19 sites in 19 patients. In 9 of these patients, a corresponding rhizopathy developed in the ipsilateral leg during ACE. In another 9 patients who reported leg pain during ACE, a corresponding facet arthrosis was described. Greater protrusion into the dural sac of a disc herniation during ACE could be seen in only 1 site.

Discussion
Dynamic myelographic examinations have long been of value in the eval‎uation of encroachments of the spinal canal.5,14,15 However, since the advent of CT and MRI, few attempts have been made to transfer previous experiences to development of the current technique. In the foreseeable future, it will not be possible to perform a CT or a MRI examination on a patient in a standing position or walking-the activity and position that typically elicit the neurogenic claudication.

Prior experimental and clinical studies 10,13 encouraged the investigators to undertake the current study. Penning and Wilmink 10 demonstrated in 12 patients with neurogenic claudication and facet hypertrophy at disc site L3-L4 and L4-L5 that nerve roots located in the lateral recess could be compressed in extension and relieved in flexion of the lumbar spine. They also showed that the D-CSA at the disc was reduced considerably when the spine was moved from a 45° flexed position to extension in a supine position with the legs straightened. The reduction of the area within the spinal canal between a flexed and extended lumbar spine was also shown experimentally by Schonstrom et al.13

In the beginning of the current study, the investigators found that the axial loading of the slightly extended lumbar spine resulted in a greater reduction of the D-CSA than did extension alone. For that reason, further tests in the extended position without compression loading were excluded.10

The participating patients all had symptoms and signs of sciatica, neurogenic claudication, or both. In more than 50% of all included sites, statistically significant reductions (>15 mm2) of the D-CSA were observed. However, 50% of the measured sites had an area exceeding 130 mm2-within the normal range in a relaxed position, according to results in previous studies.2

A D-CSA value of 100 mm2 has been suggested to be a borderline value for relative stenosis at and distal to L3-L4. In the current study, several measurements that were more than 130 mm2 in PRP approached but did not surpass 100 mm2 during ACE. If the D-CSA in a certain site exceeds 130 mm2 in PRP and if there is no sign of a bulging disc or lateral recess encroachment, ACE is unlikely to evoke further findings of clinical relevance. This statement is reinforced by the findings of Schonstrom and Hansson on pressure changes after constriction of the cauda equina.12

In 29 patients (40 sites) examined by CT and MRI, there was a statistically significant reduction of the D-CSA from PRP to ACE below 100 mm2. Narrowing of the lateral recess was obvious in 11 patients (13 sites) in the MRI study. In a study by Porter and Ward,11 it was concluded that neurogenic claudication often is associated with at least 2 sites of stenosis. This opinion is supported by Hamanishi et al.3 Myeloscopic studies by Ooi et al 8 made in walking patients have shown dilation and engorgement of the cauda equina vessels between stenotic sites. Negative effects on impulse propagation have been shown by Olmarker et al 7 and Takahashi et al 16 in two-level porcine cauda equina compression. Similarly, blood flow in two-level spinal stenosis was shown to be of importance in the pathophysiology of neurogenic claudication by Jespersen et al.4

A severe stenosis at one location should not exclude further attempts to investigate the surrounding sites as carefully as possible to visualize the dural sac and the nerve roots in the lateral recess and in the foramina. According to these results, there is a considerable risk of failing to detect an essential narrowing of the spinal canal if only the PRP is used during the examination. This might be avoided by adding the axially loaded CT-myelographic or MRI study to the standard investigation. The quality of the preoperative eval‎uation will be improved, even if further studies have to be done before all advantages of the method are known.

As generally accepted, the best way to analyze the soft tissues in the spinal canal today is by MRI. Thus, the authors now prefer to perform sagittal MRI images in those patients with suspected spinal stenosis of any kind, followed by axial images in selected sites. However, in patients with degenerative scoliosis associated with stenotic symptoms, a myelogram followed by an axially loaded CT scan might be the examination method of choice.

Conclusion
Axial loading of patients during CT and MRI examinations showed pathologic features not detected in the regular, unloaded PRP. In 29 out of 84 patients with sciatica or neurogenic claudication, the load provocation disclosed relative and absolute central spinal stenosis in 40 sites. The specificity of the spinal stenosis diagnosis increases considerably when the patient is subjected to axial loading.

Our findings suggest that dynamic examination can be recommended when the cross-sectional area of the dural sac is below 130 mm2 in PRP, or when there is a suspected narrow lateral recess, with or without deformation of the anterolateral part of the dural sac or the nerve roots in PRP.

Acknowledgments
The authors thank Anders Oden, PhD, for performing statistical analyses.

References
1. Arnoldi CC, Brodsky AE, Cauchoix J, et al. Lumbar spinal stenosis and nerve root entrapment syndromes. Definition and classification. Clin Orthop 1976;115:4-5. Bibliographic Links

2. Bolender N-F, Schonstrom NSR, Spengler DM. Role of computed tomography and myelography in the diagnosis of central spinal stenosis. J Bone Joint Surg [Am] 1985;67:240-5. Bibliographic Links [Context Link]

3. Hamanishi C, Matukura N, Fujita M, et al. Cross-sectional area of the stenotic lumbar dural tube measured from the transverse views of magnetic resonance imaging. J Spinal Disord 1994;7:388-93. Bibliographic Links [Context Link]

4. Jespersen S, Hansen E, Høy K, Christensen K, Lindblad S, Ahresberg MS, Bunger C. Two level spinal stenosis in minipigs. Hemodynamic effects of exercise. J Spine 1995;24:2765-73. [Context Link]

5. Knutsson F. Volum und Formvariationen des Wirbelkanals bei Lordosierung bzw. Kyphosierung und ihre Bedeutung fur die myelografische Diagnostik. Acta Radiol 1942;23:431-43. [Context Link]

6. Nachemson A, Elfstrom G. Intravital dynamic pressure measurements in lumbar discs. A study of common movements, maneuvers and exercises. Scand J Rehabil Med Suppl 1970;1:1-40. Bibliographic Links [Context Link]

7. Olmarker K, Holm S, Rydevik B. More pronounced effects of double level compression than single level compression on impulse propagation in the porcine cauda equina. Presented to the International Society for the Study of the Lumbar Spine, Boston, Massachusetts, June 1990. [Context Link]

8. Ooi Y, Mita KF, Satoh Y. Myeloscopic study on lumbar spinal canal stenosis with special reference to intermittent claudication. Spine 1990;15:544-9. Bibliographic Links [Context Link]

9. Penning L, Wilmink JT. Biomechanics of lumbosacral dural sac. A study of flexion-extension myelography. Spine 1981;6:398-408. Bibliographic Links [Context Link]

10. Penning L, Wilmink JT. Posture-dependent bilateral compression of L4 or L5 nerve roots in facet hypertrophy. A dynamic CT-myelographic study. Spine 1987;12:488-99. Bibliographic Links [Context Link]

11. Porter RW, Ward D. Cauda equina dysfunction. The significance of two-level pathology. Spine 1992;17:9-15. Bibliographic Links [Context Link]

12. Schonstrom N, Hansson T. Pressure changes following constriction of the cauda equina. An experimental study in situ. Spine 1988;4:385-8. [Context Link]

13. Schonstrom N, Lindahl S, Willen J, Hansson T. Dynamic changes in the dimensions of the lumbar spinal canal: An experimental study in vitro. J Orthop Res 1989;7:115-21. Bibliographic Links [Context Link]

14. Schumacher M. Die Belastungsmyelographie. Fortschr Rontgenstr 1986;145:642-8. [Context Link]

15. Sortland O, Magnaes B, Hauge T. Functional myelography with metrizamide in the diagnosis of lumbar spinal stenosis. Acta Radiol Suppl 1977;355:42-54. Bibliographic Links [Context Link]

16. Takahashi K, Olmarker K, Holm S, Porter RW, Rydevik B. Double-level cauda equina compression: An experimental study with continuous monitoring of intraneural blood flow in the porcine cauda equina. J Orthop Res 1993;11:104-9. Bibliographic Links [Context Link]

17. Willen J, Danielson B, Gaulitz A, Niklason T. Dynamic CT and MRI. Development of method and technique for eval‎uation of lumbar spinal stenosis. Presented to the Scandinavian Society of Orthopaedic Surgeons, Reykjavik, Iceland, June 1994, and to the Society of European Spinal Surgeons, Madrid, Spain, Sept, 1994. [Context Link]

Axial Loading Of The Spine During CT And MR In Patients With Suspected Lumbar Spinal Stenosis.
 
Acta Radiologica

ⓒ 1998 Munksgaard International Publishers Ltd.
--------------------------------------------------------------------------------
Volume 39(6)             November 1998             pp 604-611
--------------------------------------------------------------------------------

[Musculoskeletal Radiology]
Danielson, B. I.1; Willen, J.2; Gaulitz, A.1; Niklason, T.1; Hansson, T. H.3

Departments of 1Radiology and 2Orthopaedics, Sahlgrenska University Hospital, Molndal, and the 3Department of Orthopaedics, Sahlgrenska University Hospital, Goteborg, Sweden.
Correspondence: Barbro Danielson, Department of Radiology, Sahlgrenska University Hospital, Fack, SE-431 80 Molndal, Sweden. FAX +46 31 86 11 98.
Accepted for publication 20 April 1998.
--------------------------------------------------------------------------------

Abstract
Purpose: To eval‎uate the effect of compressive axial loading in imaging of the lumbar spine in patients with clinically suspected spinal stenosis.

Material and Methods: A total of 84 patients were examined, 50 with CT (after intrathecal contrast administration) and 34 with MR. First the dural sac cross-sectional area (CSA) was determined with the patient in the supine psoas relaxed position (PRP). Then the CSA was determined during supine axial compression in slight extension (ACE), obtained with a specially designed loading device. A measurement error study was performed.

Results:
A minimum difference in CSA of 15 mm2 between PRP and ACE was found to be significant.
In 40/50 (80%) of CT-examined patients and in 26/34 (76%) of MR-examined patients a significant difference in CSA was found.
In 25/84 (30%) of the patients there was a significant difference at more than one level.

Conclusion: For an adequate eval‎uation of the CSA, CT or MR studies should be performed with axial loading in patients who have symptoms of lumbar spinal stenosis.
--------------------------------------------------------------------------------

There is normally enough space in the lumbar spinal canal for the uncompromised dwelling of its nerve structures. In spinal stenosis, however, there is by definition a conflict between the available space in the canal and its content. When severe, this conflict will impair the nerve motors as well as the sensory functions, and also induce pain (7, 14, 15).

Clinically it is well known that a change in posture or certain physical activities may have a dramatic effect on the symptoms in spinal stenosis. Typically, standing, walking and sometimes even sitting will increase the patient's symptoms by further reducing the available space within the spinal canal. Also typically, forward lumbar flexion or a relaxed supine position, resulting in a diminished lumbar lordosis and consequently more space in the canal, will reduce symptoms by alleviating the pressure on the nerve roots.

Advanced radiological examinations of the lumbar spine are often undertaken in patients in whom there is suspicion of spinal canal space encroachment. In the myelographic examination, for example, standard imaging in the supine position is combined with exposures in the neutral standing position and in standing with the spine extended and flexed (18). The rationale for the standing position is to provoke the narrowing of the spinal canal known to occur during normal loading conditions: On CT, Schonstrom et al. (14) have found that the dural cross-sectional area (CSA) was the best measure for defining central spinal canal stenosis.

In order to avoid movement of the patient during CT and MR examinations, the patient is positioned as comfortably as possible, e.g. with the hips and knees flexed. In doing so, it is quite likely that the spinal nerve encroachment is lessened in comparison to, for example, upright standing. The possibility that the examinations will detect relevant constrictions are consequently reduced. To obtain clinically adequate information on the CSA, it would seem logical to perform CT and MR during an axial loading of the spine. For this reason, equipment was developed to make it possible to perform these examinations while a compressive axial load was applied to the spine.

The aim of this study was to perform the CT and MR examinations in a manner mimicking the standing position in order to gain a more relevant impression of the space available in the spinal canal in patients with clinically suspected lumbar spinal stenosis.

Material and Methods
The indication for the imaging examinations was a clinically suspected lumbar spinal canal narrowing which resulted in sciatica and/or neurogenic claudication in all the patients. The mean duration of symptoms for the patients studied with CT and MR was 5.5 and 3.5 years respectively.

CT study: Between January 1993 and August 1995, 50 patients were examined who had clinical signs of lumbar spinal stenosis such as pain and neurological symptoms in the lower extremities when walking, standing and/or in certain other postures. Their ages ranged between 26 and 76 years, with a mean age of 45 years. There were 19 women (38%) and 31 men (62%).

Fig. 1. The CT slices were positioned parallel to the caudal end plate of the vertebra above the examined disc.

MR study: Between April 1994 and January 1996, 34 patients were examined for clinical suspicion of spinal stenosis. Their ages were between 25 and 71 years with a mean age of 50 years. There were 16 women (47%) and 18 men (53%).

Fig. 2. The box for transverse MR slices was positioned parallel to the disc being examined.

Compressive loading: To obtain the axial compression, a loading device was constructed with which the lumbar spine was exposed to an axial force similar to that in the spine in the standing position (Fig. 3). A special non-magnetic device with similar properties was constructed for the MR examination. The compressive load applied in each case could be read at the foot-plate and ranged between 300 and 400 Newtons. The patients were asked about their symptoms during the examination, especially during compression.

Fig. 3. Patient in ACE obtained with the specially constructed compression device.

The presence of disc protrusion, disc herniation, and facet joint arthrosis was recorded. We also recorded the amount of deformation (if any) and, concerning the CT-examination, decrease in contrast content in the dural sac during compression, as well as presence or not of recess or foraminal stenosis. Facet joint arthrosis was regarded as present if there was a clearly detectable joint space narrowing, increased bony sclerosis, and osteophyte formation. Recess and foraminal stenosis were defined as a reduction in the space available to the nerve roots (recess less than 3 mm) in combination with loss of epidural fat.

The CSA of the dural sac was determined by using a standard measurement program in the CT unit. The MR images were transferred to the CT unit and all measurements were performed there. In every patient, the image selected was that in which the dural sac seemed to have the smallest area on each disc level in all positions. One of the authors (B.D.) performed three measurements on each image. To ensure that the images chosen for measurements were comparable in every position, the radiologist compared nerve roots, other soft tissues, and bony structures such as facet joints and spine processes. The mean CSA was calculated as well as the reduction of the CSA between the different positions in each patient.

Statistical methods: A measurement error study was performed. Six patients were selected by an unbiased radiologist (A.G.). The CSA in these patients in PRP and ACE was measured on five different occasions with a time interval of at least 4 weeks. Three separate measurements were made on every occasion.

There were two independent measurement errors. One occurred when the structures on the image were identified, and the other one in the measurement on the image. The measurement of the CSA could be repeated on the same image and, by using the mean, the error was reduced. The variance of the total measurement error was Q12+Q22/n, where: Q1 was the standard deviation of the type of error mentioned first; Q2 the standard deviation of the other type; and n the number of repetitions of the measurement procedure. The patients were studied in two positions. The probability that the difference between the measurements in the two positions exceeded a selected value, even if no true difference was present, could thus be calculated.

Results
Measurement error study: The estimated variances of the two types of measurement errors were 0.34 and 0.23 respectively. Thus the standard deviation of the total error was 6.5 mm2. The probability that the difference between the CSA calculated at PRP and ACE would exceed 15 mm2 was less than 5%, provided that no true difference was present.

CT study: Sixteen patients (32%) felt no discomfort during the examination. Twenty-one patients (42%) experienced lumbar back pain (LBP) in ACE and 8 of them felt leg pain as well. Nine patients had LBP in PRP and ACE. Three of them had leg pain as well, of whom 2 experienced aggravated LBP in ACE, and 3 leg pain. Two patients felt pressure in the back in ACE and 2 had neurological symptoms.

The number of examined discs at each vertebral level is shown in Table 1a. A total of 99 discs were eval‎uated. A disc protrusion was found at 77 levels (at L4-L5 in 35 patients and at L5-S1 in 34 patients). A disc herniation was found in 10 patients, most often at L4-L5 (6 patients). Facet joint arthrosis was found at 74 levels.

Table 1 a Results of the CT study

The mean decreases in CSA at different levels between PRP and ACE and between PRP and E are shown in Table 2a. In Fig. 4 the values of the CSA in PRP and ACE are shown for all eval‎uated levels. At 55/99 (56%) levels there was a significant decrease in CSA between PRP and ACE. In 40 (80%) of the patients a decrease in CSA was found at at least one level. In 15 patients there was a decrease at more than one level. At 20 levels (15 patients) the CSA changed from more than to less than 100 mm2 or 75 mm2. At five levels there was a significant increase in CSA (range 18-27 mm2).

Table 2 a CT study, decrease in CSA at different disc levels

Fig. 4. The CSA values obtained in PRP ([black small square]) and in ACE ([white circle]) at each disc level eval‎uated in the CT study.

There was a decrease in CSA at 8/21 (38%) of the eval‎uated levels in E compared to PRP, and the same was found in 4/9 (44%) of the patients.

The dural sac was deformed in 14 patients and in 6 the contrast content in the thecal sac was clearly less prominent in ACE compared to PRP. In 9 patients there was a recess stenosis, and in 5 a foraminal one. Five of those who had leg pain in ACE had a disc herniation, 10 had a facet joint arthrosis, and 3 a recess stenosis.

MR study: Fourteen patients (41%) had no pain when compression was performed. Seven patients (21%) had LBP and 10 (29%) leg pain in ACE. Two patients had LBP in PRP and ACE as well, 1 of whom had leg pain in ACE. One patient had no LBP but experienced sensory disturbances in her toes during ACE.

The number of discs examined is listed in Table 1b. A total of 81 discs were eval‎uated. A disc protrusion was registered at 51 levels, and most often at the L4-L5 level. In 10 patients there was a disc herniation, most often at the L5-L6/S1 level (5 patients). The mean decrease in CSA irrespective of level is shown in Table 2b. The values of the CSA in PRP and ACE at every eval‎uated level is shown in Fig. 5.

Table 1 b Results of the MR study

Table 2 b MR study, decrease in CSA at different disc levels

Fig. 5. The CSA values obtained in PRP ([black small square]) and in ACE ([white circle]) at each disc level eval‎uated in the MR study.

At 37/81 (46%) levels a significant decrease occurred in the CSA during ACE. In 26/34 (76%) of the patients we found a decrease in CSA between PRP and ACE. In 10 patients this occurred at more than one level. At 10 levels (8 patients) the CSA changed from above to below 100 or 75 mm2. There was a significant increase in CSA in ACE at four levels (range 16-35 mm2). The dural sac changed its shape during ACE in 11 patients. A recess stenosis was found in 12 patients and a foraminal stenosis was seen in 1 patient. Five of the patients with leg pain in ACE had a disc herniation and 6 a recess stenosis.

Discussion
For a long time now, advanced radiological examinations have been performed to confirm‎ or disprove clinical suspicion of lumbar disc herniation or spinal stenosis. Myelography has been used as a diagnostic tool but has been replaced to an increasing extent by CT and MR in recent years. It is known that the size of the spinal canal increases in forward flexion and decreases in extension and standing (1, 4, 9, 18). Since patients are usually examined in a relaxed supine position, the results of measuring the CSA in routine CT and MR examinations are unlikely to give a true reflection of the space available when the patient is standing. Therefore it would seem more accurate to examine the patient in a position as similar to the standing position as possible. This is achieved by applying an axial load to the lumbar spine, as described by Schonstrom et al. (17). In an experimental study they used a loading device which allowed static loading in compression and distraction as well as in flexion and extension. They found that the CSA of the spinal canal at disc level was reduced by respectively 40 and 50 mm2 between the examinations in flexion and extension as well as between those in distraction and compression. The difference in axial load between compression and distraction was about 500 Newton. This was of the same magnitude as the change in load between lying and standing that was found in a study of in vivo disc pressure measurements (6).

Hamanishi et al. (3) measured the CSA on MR images in patients examined in a supine position with flexed knees and hips and in an extended hyperlordotic position. They found that the CSA values in the latter position were 93±4% of those obtained in the former position. They concluded that the CSA might be larger than the "true" value when the patient lies in a pain-relieving position with a decreased lumbar lordosis.

Penning & Wilmink (10) performed CT examinations in patients with myelographic evidence of nerve root entrapment. They showed the concentric narrowing of the spinal canal in extension and its widening in flexion with relief of nerve root involvement. Our examinations in extension and compression put an increased load on the spine, giving a closer resemblance to the upright standing position and producing CSA values which ought to be clinically relevant. The reason for our exclusion of the examination in extension after the first 9 patients was that we found a greater reduction in the CSA in ACE compared to E (in 80% and 44% of the patients respectively).

As pointed out earlier by Schonstrom et al. (14) and Verbiest (19), intrathecal injection of contrast medium before the CT examination is an advantage when measurement of the CSA is planned. We found that it was very important to turn the patient round before the CT imaging to prevent sedimentation of the contrast. In some cases we were forced to move the patient around between the examinations in PRP and ACE (Fig. 6).

Fig. 6. CT images at L4-L5 in PRP (a) and ACE (b). The contrast has sedimented in PRP but was more mixed with liquor in ACE after the patient had been turned round.

The transverse MR images eval‎uated were T1- and/or T2-weighted. The reason for the change during the ongoing study was that we changed the software of the MR unit.

Schnebel et al. (12) retrospectively compared contrast CT and MR and found these imaging modalities comparable in demonstrating spinal stenosis. The choice between CT and MR examination therefore mainly depends on the availability of the methods. CT examination following intrathecal contrast administration is an invasive technique with possible adverse effects. It also involves exposing the patient to a radiation dose. The MR examination is more useful in its ability to compensate for deformities of the spine, i.e. scoliosis.

Koehler et al. (5) emphasized the importance of correct window settings when measurements on CT images are performed. Beers et al. (2) found differences between observers in distance measurements in spinal CT. Most of their reported discrepancies were likely to result from systematic differences in the window settings. For identifying a boundary, they recommended a window centre just half way between the densities of the structures on either side of it. In the present study we chose window settings according to this recommendation.

Our CT unit has a maximum tilting capacity of 20°. Therefore it was not always possible to obtain adequate tilted images in some patients, especially in ACE. According to Schonstrom (13) an angulation of the gantry will enlarge the measured CSA. However, this enlargement is smaller than 4% if the difference in angulation between calculations is less than 15°. In our study the difference was always below this figure.

In patients where ACE caused a reduction of the CSA, the most typical pattern was a concentric restriction (Figs 7 and 8). This was in accordance to the findings of Penning & Wilmink (10). In some of our patients there was a more prominent indentation of the fat pad dorsally, an accentuated thickening of the yellow ligaments (Fig. 9), or a more pronounced disc protrusion (Fig. 10). In 1 patient with a large disc herniation, the examination in ACE aggravated the compression of the dural sac (Fig. 11). The effect of ACE on the appearance of disc herniations should be further investigated.

Fig. 7. CT images at L4-L5 in PRP (a) and ACE (b) showing a concentric decrease in CSA from 130 to 100 mm2.

Fig. 8. T1-weighted MR images at L4-L5 in PRP (a) and ACE (b) showing a concentric decrease in CSA from 80 to 50 mm2.

Fig. 9. T2-weighted MR images at L4-L5 in PRP (a) and ACE (b) illustrating a more prominent thickening of ligamentum flavum and indentation of the dorsal fat pad in ACE compared to PRP.

Fig. 10. T2-weighted MR images at L5-L6 in PRP (a) and ACE (b) showing a more impressive disc protrusion in ACE compared to PRP.

Fig. 11. CT images at L5-L6 in PRP (a) and ACE (b). Note the accentuated disc herniation in ACE compared to PRP.

Schonstrom et al. (14) has pointed out that the most accurate method for identifying spinal stenosis was the measurement of the CSA on CT images. They also had the impression that the critical size for the dural sac was below 100 mm2. In experimental studies, Schonstrom et al. (15, 16) found that the critical value of the CSA (i.e. the size at which there is an increase in tissue pressure) was 60-80 mm2 in the lumbar spine. They suggested that below this threshold value, a further constriction of the size of the dural sac would cause symptoms of spinal stenosis. Among our patients ACE induced a CSA change from above to below 100 or 75 mm2 at 20 and 10 levels respectively in the CT and MR studies. Our study revealed that the CSA decreased considerably when the examination was performed in ACE. This could mean that normal posture changes can cause abnormal tissue pressure in the dural sac, implying clinical symptoms and signs.

In experimental studies on porcine cauda equina, Olmarker et al. (8) found that compression of the cauda produced pronounced effects on nerve root impulse propagation, especially when the cauda was compressed at two levels. Porter & Ward (11) examined patients with neurogenic claudication with myelography and CT, and found that 94% had either multilevel central canal stenosis or stenosis of both central and root canals. They concluded that neurogenic claudication was generally associated with at least two levels of stenosis. In their study on patients with intermittent claudication, Hamanishi et al. (3) showed that 90 % had a CSA of less than 100 mm2 at two or more levels. They considered double lesions with CSA below 100 mm2 to be a critical factor in the development of intermittent claudication.

In our series 25/84 (30%) of the patients showed a significant decrease in CSA at more than one level in ACE. It would thus seem to be of great importance that all possible stenotic levels are eval‎uated in every patient.

Since the number of images in the CT examination in ACE was limited (in order to reduce the radiation dose), eval‎uation of the recesses in ACE was not always possible. Future examinations should include this area, and studies on the effect of compression on the recesses are underway. At MR it was possible to eval‎uate the recesses but we did not register facet arthrosis unless very impressive.

In patients with clinical suspicion of lumbar spinal stenosis, we found that to a considerable extent there was a significant decrease in CSA when CT and MR examinations were performed in ACE. Therefore it is a strong indication that CT and MR examinations should be performed with axial loading of the spine.

ACKNOWLEDGEMENTS
This investigation was supported by research grants from: Philips Medical System AB; the Consul Thure Carlsson Memorial Foundation, Lund, Sweden; and FOU Bohuslan County Council, Goteborg, Sweden.

The statistical analysis was performed by A. Oden, Ph.D.

REFERENCES
1. Amundsen T., Weber H., Lilleas F., Nordal H. J., Abdelnoor M. & Magnaes B.: Lumbar spinal stenosis. Clinical and radiological features. Spine 20 (1995), 1178. Bibliographic Links [Context Link]

2. Beers G. J., Carter A. P., Leiter B. E., Tilak S. P. & Shah R. R.: Interobserver discrepancies in distance measurements from lumbar spine CT scans. AJR 144 (1985), 395. Bibliographic Links [Context Link]

3. Hamanishi C., Matukura N., Fujita M., Tomihara M. & Tanaka S.: Cross-sectional area of the stenotic lumbar dural tube measured from the transverse views of magnetic resonance imaging. J. Spinal Dis. 7 (1994), 388. [Context Link]

4. Knutsson F.: Volum- und Formvariationen des Wirbelkanals bei Lordosierung und Kyphosierung und ihre Bedeutung fur die myelographische Diagnostik. Acta Radiol. 23 (1942), 431. [Context Link]

5. Koehler P. R., Anderson R. E. & Baxter B.: The effect of computed tomography viewer controls on anatomical measurements. Radiology 130 (1979), 189. Bibliographic Links [Context Link]

6. Nachemson A. & Elfstrom G.: Intravital dynamic pressure measurements in lumbar discs. Scand. J. Rehab. Med. Suppl. 1 (1970), 1. [Context Link]

7. Olmarker K., Rydevik B., Holm S. & Bagge U.: Effects of experimental graded compression on blood flow in spinal nerve roots. A vital microscopic study on the porcine cauda equina. J. Orthop. Res. 7 (1989), 817. Bibliographic Links [Context Link]

8. Olmarker K., Holm S. & Rydevik B.: More pronounced effects of double level compression than single level compression on impulse propagation in the porcine cauda equina. Int. Soc. Study of the Lumbar Spine, Boston, MA 1990. [Context Link]

9. Penning L. & Wilmink J. T.: Biomechanics of lumbosacral dural sac. A study of flexion-extension myelography. Spine 6 (1981), 398. Bibliographic Links [Context Link]

10. Penning L. & Wilmink J. T.: Posture-dependent bilateral compression of L4 or L5 nerve roots in facet hypertrophy. A dynamic CT-myelographic study. Spine 12 (1987), 488. Bibliographic Links [Context Link]

11. Porter R. W. & Ward D.: Cauda equina dysfunction. The significance of two-level pathology. Spine 17 (1992), 9. Bibliographic Links [Context Link]

12. Schnebel B., Kingston S., Watkins R. & Dillin W.: Comparison of MRI to contrast CT in the diagnosis of spinal stenosis. Spine 14 (1989), 332. Bibliographic Links [Context Link]

13. Schonstrom N. S. R.: The significance of oblique cuts on CT scans of the spinal canal in terms of anatomic measurements. Spine 13 (1988), 435. Bibliographic Links [Context Link]

14. Schonstrom N. S. R., Bolender N.-F. & Spengler D. M.: The pathomorphology of spinal stenosis as seen on CT scans of the lumbar spine. Spine 10 (1985), 806. Bibliographic Links [Context Link]

15. Schonstrom N. S. R., Bolender N.-F., Spengler D. M. & Hansson T. H.: Pressure changes within the cauda equina following constriction of the dural sac. An in vitro experimental study. Spine 9 (1984), 604. Bibliographic Links [Context Link]

16. Schonstrom N. S. R. & Hansson T. H.: Pressure changes following constriction of the cauda equina. An experimental study in situ. Spine 13 (1988), 385. Bibliographic Links [Context Link]

17. Schonstrom N. S. R., Lindahl S., Willen J. & Hansson T. H.: Dynamic changes in the dimensions of the lumbar spinal canal. An experimental study in vitro. J. Orthop. Res. 7 (1989), 115. Bibliographic Links [Context Link]

18. Sortland O., Magnaes B. & Hauge T.: Functional myelography with metrizamide in the diagnosis of lumbar spinal stenosis. Acta Radiol. (1977), Suppl. 355, p. 42. [Context Link]

19. Verbiest H.: The significance and principles of computerized axial tomography in idiopathic developmental stenosis of bony lumbar vertebral canal. Spine 4 (1979), 369. Bibliographic Links [Context Link]