| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Both authors: Department of Radiology, Royal National Orthopaedic Hospital, Brockley Hill, Stanmore, Middlesex, London, England HA7 4LP.
Received March 8, 2005;
accepted after revision May 25, 2005.
Address correspondence to A. Saifuddin
(asaifuddin{at}aol.com).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionAPPENDIX 1: Castellvi et...References |
|---|
MATERIALS AND METHODS. Five hundred consecutive lumbar spine MRI studies were reviewed. A standard protocol of sagittal and axial T1-weighted and T2-weighted spin-echo sequences was used. The sagittal images were assessed for the presence of an LSTV, and axial images were assessed for the level of origin of the iliolumbar ligaments.
RESULTS. Of the 500 patients, 433 (86.6%) had normal lumbosacral segmentation and 67 (13.4%) had a transitional lumbosacral junction. The iliolumbar ligament was identified at L5 in all patients with normal lumbosacral segmentation (n = 433), bilaterally in 432 and unilaterally in one. Using the identification of the iliolumbar ligaments as a marker of the L5 vertebral level, we numbered 46 of the 67 LSTV as L5 transitions and 21 as S1 transitions.
CONCLUSION. The iliolumbar ligament is readily identifiable on axial lumbar spine MRI and always arises from L5. We suggest that its position can be used to confidently assign lumbar levels in patients with LSTV.
Keywords: iliolumbar ligaments • lumbar spine • MRI • transitional vertebra
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAPPENDIX 1: Castellvi et...References |
|---|
Although LSTV can be sensitively identified on sagittal lumbar spine MRI, based on abnormal morphology of the lumbosacral junction [5], no standard method is established for their numbering. Techniques that have been used include the addition of cervicothoracic localizer scans [3, 6] and determining lumbar levels by identification of the right renal artery [7, 8]. However, both of these techniques are prone to substantial error.
Both anatomic and MRI studies have indicated that the iliolumbar ligament (ILL) arises almost exclusively from the L5 transverse process [9-15]. However, these studies are limited by relatively small numbers (Table 1). We hypothesize that if ILLs can be reliably identified on routine axial MR images through the lumbosacral junction and if they arise solely from L5, then their level of origin can serve as a marker of lumbar levels, which may enable the confident numbering of LSTV.
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAPPENDIX 1: Castellvi et...References |
|---|
All patients were imaged at 1.0 T using a dedicated lumbar spine coil and a standard protocol for lumbar spine MRI. Both T1-weighted spin-echo images (TR/TE, 637/12) and T2-weighted fast spin-echo images (4,399/120) were obtained through the lumbosacral spine in the sagittal plane and through the lumbosacral junction in the axial plane. Additional axial images through any areas of apparent disease were also performed as necessary. Axial images were performed from pedicle to pedicle through the lumbosacral junction. The field of view was 200 x 200 mm for axial scans and 300 x 300 mm for sagittal scans, with a matrix of 256 x 256, slice thickness of 4.5 mm, and a 0.4-mm gap for both axial and sagittal imaging.
The images were assessed for the presence or absence of a transitional lumbosacral junction based on the method of O'Driscoll et al. [5]. In addition, the position of the right renal artery (RRA) in relation to the L1-2 disk space was noted on T2-weighted fast spin-echo sagittal images. The T1-weighted spin-echo axial images through the lumbosacral junction were examined to identify the ILLs. These appeared as either a single or a double hypointense band arising from the transverse process and extending to the posteromedial aspect of the iliac crest (Figs. 1A and 1B). The slice number at which the ILL was identified was noted and correlated with the sagittal localizer scan to allow identification of the ILL's lumbar vertebral level of origin.
|
|
|
|
|
|
|
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAPPENDIX 1: Castellvi et...References |
|---|
Of the 500 lumbar spine MRI studies assessed, 433 (86.6%) patients had normal lumbosacral segmentation and 67 (13.4%) had transitional lumbosacral junctions. With the system used by O'Driscoll et al. [5], 53 had type 1 morphology of the lumbosacral junction, 273 had type 2 morphology, 124 had type 3 morphology, and 50 had type 4 (transitional) morphology (Figs. 2A and 3A). In addition to the 50 type 4 lumbosacral junctions on sagittal imaging, 17 other patients (2 with type 3, 12 with type 2, and 3 with type 1 morphology) clearly had transitional lumbosacral junctions based on the axial images, which showed either unilateral (Fig. 4) or bilateral enlargement of the transverse processes with extension down to the sacral ala.
The ILL was identified at L5 in all patients with normal lumbosacral junctions, bilaterally in 432 and unilaterally in one. In one patient, the ILL was seen bilaterally at L5 with a smaller ligament bilaterally at L4. In no other case was an ILL seen to arise at the L4 level. The ILL was identified in 21 patients with an LSTV. In these cases, the LSTV was numbered S1 because the level above where the ILL arose was classified as L5. The ILL was not identified in 46 patients with LSTV. In these cases, the LSTV was labeled as L5. Smaller ligaments were identified between these transitions and the ilium, which were identical in appearance to the sacroiliac ligaments or the sacroiliac part of the iliolumbar ligament (see Discussion section).
The RRA was identified on the sagittal T2-weighted fast spin-echo images in 456 of the 500 patients (Fig. 3A). In the other 44 cases it was not seen. In only 367 of the 456 cases was the RRA closest to the L1-2 disk space. Therefore, the RRA was identified adjacent to the L1-2 disk in only 73.5% of cases. It was not seen in 8.5% of the cases and was at the mid vertebral body level or closer to other disk spaces in 17.5% of cases.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAPPENDIX 1: Castellvi et...References |
|---|
The presence of an LSTV is best identified on a true 30° angled anteroposterior radiograph of the lumbosacral junction (Ferguson view) together with an anteroposterior view including the thoracolumbar junction to enable assessment of the vertebral level. A classification system for LSTV was proposed by Castellvi et al. [1] in 1984 based on the conventional radiographic appearances (Appendix 1). Conventional radiographic studies have shown that the lumbosacral transitional disk is of significantly reduced height compared with normal lumbosacral intervertebral disks, and that on lateral radiographs the transitional vertebra has a squared appearance, with the ratio of the anteroposterior diameter of the superior vertebral end-plate to the inferior endplate being less than 1.37 [19]. These morphologic changes allow the identification of LSTV on sagittal MRI studies of the lumbar spine. O'Driscoll et al. [5] looked at 100 consecutive sagittal T1-weighted and T2-weighted lumbar spine MRI studies and classified them into four types according to the appearance of the disk between the uppermost sacral segment and the remainder of the sacrum: type 1 to indicate no disk material present, type 2 to indicate a small residual disk not extending for the whole anteroposterior diameter of the sacrum, type 3 to indicate a well-formed disk extending for the whole anteroposterior diameter of the sacrum, and type 4 to indicate the same as type 3 but with the addition of an abnormal upper sacral outline analogous to the radiographic "squaring" appearance. This study showed an excellent correlation between the presence of a fused LSTV (Castellvi type III or IV) and a type 4 MRI appearance. Patients with a pseudarthrosis (Castellvi type IIa or IIb) were not reliably identified with this method. However, assessment of axial images through the lumbosacral junction identifies pseudarthroses because of the unilateral or bilateral enlargement of the transverse processes of the LSTV, which can be followed on consecutive axial images down to the sacral ala. Therefore, the identification of LSTV is reliably made using MRI. However, these techniques do not address the problem of numbering the LSTV. This is of clinical relevance when assigning lumbar levels when reporting MRI scans. Most surgical errors occur in patients with LSTV [20], but this should be avoided with standard surgical practice that correlates the pathologic levels with intraoperative radiography.
If an LSTV is suspected on MRI, the decision of whether it represents a sacralized L5 or lumbarized S1 must be made. This may be complicated by the presence of a concomitant thoracolumbar transition [20]. The true nature of the lower vertebral segmentation can be established only on conventional radiographs that include the thoracolumbar junction so that hypoplastic true ribs may be differentiated from large transverse processes, thus allowing correct identification of the L1 vertebral body [17]. Once this has been correctly established, the LSTV level can be confirmed.
However, with current clinical practice, MRI studies of the lumbar spine are commonly reported without available radiographs. Attempts have therefore been made to number LSTV based on MRI. Peh et al. [6] looked at the ability of radiologists to correctly identify lumbar levels on sagittal and coronal T1 localizer MR scans. They compared assessments of sagittal and coronal lumbar localizer views, with these assessments combined with an additional cervicothoracic localizer. In terms of assignment of lumbar vertebral levels, concordance occurred in 80% of patients.
Hahn et al. [3] also recommend a whole-spine localizer to enable counting caudad from C2 to identify and number LSTV. In their series, they found 24 of 200 LSTV cases and numbered them based on this method with the assumption that there were seven cervical and 12 thoracic segments. In practice this will not distinguish between patients with thoracolumbar transitions [5]—for example, between T11-L5, T12-L6, and T12-L5.
It is proposed that the position of the RRA may be used to identify lumbar levels on sagittal MRI, because it usually lies closest to the L1-2 disk space [7]. However, this method also appears insensitive. Although the RRA is commonly seen to lie adjacent to the L1-2 disk, this criterion cannot be used in approximately 25% of cases, either because the artery cannot be identified or because it is situated midway between disk spaces. These findings were confirmed in the present study. It would therefore appear that this technique is not sufficiently sensitive for numbering lumbar levels [8].
Lee et al. [21], in a recent article, looked at the position of the aortic bifurcation, RRA, and conus medullaris when locating lumbar segments using a whole-spine localizer for verification. They found the conus medullaris to be variable and unhelpful. However, the RRA was between the lower half of L1 and the upper half of L2 in 92% of the 210 patients studied, and the aortic bifurcation was at L4 in 83%. In cases of LSTV in this series (n = 20), they found caudal position of these structures in lumbarization of S1 and cephalad position of these paraspinal structures in sacralization of L5. The same limitations with regard to sensitivity apply, as described previously, and the positions of these structures cannot be used to confidently assign levels when reporting a lumbar spine MRI series alone.
The ILL is one of three vertebropelvic ligaments, the others being the sacrotuberous and sacrospinous ligaments. Functionally, the ILL restrains flexion, extension, axial rotation, and lateral bending of L5 on S1 [22]. Fetal dissection suggests that it forms at 11 weeks of gestation and is well developed by birth [9]. The literature suggests that an L5 origin is invariable and that occasionally a weaker band may coexist from L4. Yamamoto et al. [10] reported an occasional weak band arising from L4. Pool-Goudzwaard et al. [11], in a dissection study, describe a complex ligament of at least seven components, ascribing an L4 component to 10 of 17 of their specimens. However, some larger cadaver and MRI studies (133 dissections, 14 MRI cases) have not found an L4 component [12-14]. Recent cadaver and MRI studies of the ILL indicate that it arises from the transverse process of L5 and is divided into two main parts: first, a broad, flat anterior part inserting onto the iliac tuberosity; and, second, a thinner fusiform, posterior part that inserts onto the iliac crest [13, 14, 23]. The anatomic literature is summarized in Table 1.
|
|
The proportion of LSTV (13.4%) seen in our study of 500 patients is consistent with previously reported studies [1-5]. We were able to identify the ILL in all patients using standard axial T1-weighted spin-echo sequences through the lumbosacral junction in the presence of normal lumbosacral segmentation. This indicates that the ILL is a reliable marker of the L5 vertebral level on axial MRI. We were then able to use this criterion for numbering LSTV. If a normal ILL was identified at the level above the transitional vertebra, this identified the L5 level so the transitional vertebra was then labeled as a transitional S1. If a normal ILL was not identified, the transitional vertebra was labeled as a transitional L5. In L5 transitions, smaller ligaments analogous to the sacroiliac part of the ILL [11] were commonly identified and no ILL could be identified at the level above.
In conclusion, identification of the ILL on contiguous axial MR images through the lumbosacral junction indicates the L5 vertebral level. Consequently, the numbering of LSTV can be reliably made by reference to the level of origin of the ILL.
APPENDIX 1: Castellvi et al. [1] Classification of Lumbosacral Transitional Vertebrae
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAPPENDIX 1: Castellvi et...References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAPPENDIX 1: Castellvi et...References |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  E. J. CHURCH and T. G. ODLE Diagnosis and Treatment Of back Pain Radiol. Technol., November 1, 2007; 79(2): 126 - 151. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. L. Bressler Numbering of Lumbosacral Transitional Vertebrae on MRI Am. J. Roentgenol., February 1, 2007; 188(2): W210 - W210. [Full Text] [PDF]
|
|
|
|
|
|
A. Saifuddin and R. J. Hughes Reply Am. J. Roentgenol., February 1, 2007; 188(2): W211 - W211. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
