tomography.
Introduction
The syndrome, degenerative lumbosacral stenosis (DLSS), was described in chapter five as an acquired narrowing of the vertebral canal, lateral intervertebral foramina, or both at the lumbosacral (LS) junction, which results in compressive radiculopathy of one or more nerve roots of the cauda equina (Chambers 1989, DeRisio et al. 2001). Components of DLSS, which are thought to contribute to compression of the cauda equina include, prolapse of the annulus fibrosus, hypertrophy of the interarcuate ligament, hypertrophy of the capsule of the articular processes, and dynamic narrowing or osteophytosis of the lateral intervertebral neurovascular foramen. The L7 nerve roots, having left the dura at the level of the L6 vertebra, pass through the lateral recess within the vertebral canal of L7 (Axlund and Hudson 2003). They exit the lateral intervertebral neurovascular foramen cranial and lateral to the L7-S1 disc. The lateral intervertebral neurovascular foramen is formed by the pedicle of L7 cranially, the articular processes of L7 and S1 dorsally, the arch of the sacrum caudally, and the body of L7 and the annulus fibrosus of the L7-S1 intervertebral disc ventrally. The lateral intervertebral neurovascular foramen is not an aperture, but resembles a canal or tunnel with an entrance, middle, and exit zones. Dynamic narrowing of the lateral intervertebral neurovascular foramen is thought to occur during extension and potentially may be exacerbated by disc degeneration (Jones et al. 2008).
The LS junction is the most mobile of all the intervertebral articulations in healthy dogs, however, compared to other species the LS junction contributes proportionately less to overall vertebral mobility. In large herbivores the lumbar vertebral column is relatively inflexible and primarily acts as a supporting strut for the heavy abdominal compartment (Gal 1993). Therefore, in herbivores the LS junction is proportionately more mobile in comparison to the remaining vertebral segments than in carnivores. In carnivores the increased flexibility of the lumbar vertebral column allows for an increased stride length, which facilitates a faster gait when chasing prey (Gal 1993). The three- dimensional motion pattern of the caudal lumbar and lumbosacral portions of the vertebral column of dogs has been reported in two studies (Benninger et al. 2004, 2006). The normal LS junction permits bending (flexion/extension) in the median plane, lateral bending in the dorsal plane, and very limited rotation in the transverse plane.
The range of motion of the LS junction in the median plane was 32.8 ± 6.4° in 15 skeletally mature medium to large breed dogs (Early et al. 2013). There is evidence that as DLSS develops, motion of the LS junction is decreased by the changes that occur in the intervertebral disc and supporting structures (Jones et al. 2008). A decrease in the range of motion of the LS junction is in contrast to the concept that instability of the LS junction is a feature of DLSS. However, some dogs with DLSS do have ventral displacement of the sacrum relative to the L7 vertebral body (retrolisthesis
or a “step-lesion”), a feature that is proposed as evidence of instability of the LS junction (Slocum and Devine 1986, Palmer and Chambers 1991a).
The angulation of the LS junction during activity alters the pathway of the nerves of the cauda equina and also affects the dimensions of the lateral intervertebral neurovascular foramina. The cross-sectional area of an intervertebral neurovascular foramen can be estimated in-vivo using computed tomography (CT). Using sagittal reconstructions of CT data, a region of interest (ROI) can be drawn around the bone margins of an intervertebral foramen and the area within that boundary calculated. A significant decrease in the mean radiographic foraminal area with the LS junction in an extended vs. flexed position was detected in dogs in sagittal CT images (Jones et al. 2008). Extension of the LS junction results in the cranial articular processes of the sacrum tipping rostrally into the lateral intervertebral neurovascular foramen. Whilst a significant linear relationship was found between a change in the foraminal area and the angle of the LS junction of dogs with hind limb lameness and lumbosacral pain, this relationship was not significant for dogs without these clinical signs (Jones et al. 2008). Those authors’ hypothesised that one of the normal functions of supportive structures of the LS junction is to maintain the dimensions of the L7-S1 lateral intervertebral neurovascular foramen independent of the changes in LS angulation during flexion and extension. Counter-intuitively, the percentage change in the area of the L7-S1 lateral intervertebral neurovascular foramina from flexion to extension did not differ significantly between dogs with or without clinical signs attributed to degeneration of the LS junction. In the study by Jones et al. (2008) measurements of the foraminal area were derived from sagittal plane images, which were not orientated perpendicular to the path of the L7 nerve roots. Sagittal orientation potentially over-estimated the true cross-sectional area of the foramen at its narrowest point. A double-oblique, parasagittal method of determining the foraminal area perpendicular to the path of the L7 nerve roots has been reported by Higgins et al. (2011). However, this study method was determined to be unreliable due to inter-observer variability and the effects of positioning of the dog for CT slice orientation.
An alternative approach to quantifying the minimal cross-sectional area of a three- dimensional tunnel is to calculate its volume. The three dimensional volume of the L7- S1 lateral intervertebral neurovascular foramen has not been evaluated in dogs with or without DLSS. Measuring the volume of the L7-S1 lateral intervertebral neurovascular foramen might be a valuable clinical and research tool for testing the effects of motion on the LS junction. Establishing values for healthy dogs and those affected with DLSS may be valuable for diagnosis, help determine if surgical intervention is warranted, and help to determine which procedure should be recommended. With the advent of new therapies, many of which differ radically in concept (dorsal stabilisation versus lateral foraminotomy), it is vital that veterinarians have full understanding of the function of the LS junction in both health and disease.
The aims of this study were to develop a method using CT images to determine the volume of the L7-S1 lateral intervertebral foramen and to quantify the effect of positioning of the LS junction on the volume of the foramen. Positioning the dog within the CT gantry with the LS junction in maximal extension, neutral positioning and full flexion was used to mimic the range of motion of the LS junction occurring during activity. The hypotheses were that extreme extension would significantly decrease the volume of the L7-S1 lateral intervertebral neurovascular foramen in both healthy dogs and dogs with DLSS, and that dogs’ with DLSS would have a greater decrease in foraminal volume than unaffected dogs. In addition it was hypothesized that the Greyhound, a breed highly selected for athletic performance and less susceptible to DLSS, would have a lower ratio of L7-S1 lateral intervertebral neurovascular foraminal narrowing than the GSD.
Materials and methods
Three groups of dogs were compared. One group consisted of NZ PDS GSDs with clinical signs referable to the LS junction examined at the MUVTH between May 2009 and April 2013. All were confirmed to have DLSS (Abnormal GSDs = AbNGSD). Physical and neurological examinations localised the clinical signs to the LS region, and screening radiographs and CT scan findings were consistent with DLSS. The second group consisted of NZ PDS GSDs presented to the MUVTH in the same period for signs unrelated to lumbosacral disease (Normal GSDs = NGSDs). There were no historical indications of LS pain or dysfunction and clinical evaluations failed to confirm neurological or orthopaedic abnormalities. While sedated for diagnostic investigation of the primary complaint, each dog underwent LS imaging according to the study protocol. The third group consisted of healthy racing Greyhounds undergoing CT
scanning of the hock joints for an unrelated study on bone density of the central tarsal bone (NGH). An orthopaedic and neurological evaluation was performed and any greyhound with a hind limb gait abnormality or apparent pain on examination of the LS region was excluded. The imaging studies were identical for all groups. Each dog was deeply sedated with 0.005-0.01 mg/kg medetomidine and 0.1-0.2 mg/kg butorphanol administered intravenously (IV) as a single injection. A second dose of medetomidine was given IV to any dog that was insufficiently sedated. Computed tomography was performed using a Phillips Brilliance 16-slice helical scanner (Philips Healthcare, The Netherlands) with the dogs in dorsal recumbency supported by a high-density foam trough. Each dog was alternately strapped to the CT table with the LS junction in neutral, flexed, and extended LS positions. For the neutral LS position the femurs were allowed to rest at 90 degrees to the table, with the hocks supported on a foam block and the stifles taped around a block of foam at the width of the pelvis. For the flexed position the hind limbs were drawn forward and restrained beside the thorax using
heavy sandbags. Flexion of the LS junction was maximised by ensuring the dog’s hind- quarters were lifted off the foam trough (Figure 34A). For the extended views a foam block was placed between the stifles, which were secured with tape at the same width as the pelvis. The caudal edge of the foam trough was positioned directly beneath the LS junction as a fulcrum and the hind limbs were then extended caudally and secured in maximal extension of the coxofemoral joints using velcro straps which were secured to the CT table, (Figure 34B).
Figure 34. Positioning of the dog on the CT gantry for imaging of the lumbosacral junction.
A. Flexed CT positioning B. Extended CT positioning
Careful attention to detail was followed to ensure consistency of positioning at the limits of passive range of motion in flexion and extension of the LS junction. The acquisition window included the mid-body of L5 to the first caudal vertebrae and was positioned so that axial sectioning was parallel to the caudal endplate of L7. Axial images were acquired (120kV, 600MAS) with a 1 mm slice thickness and a 0.5 mm interval in both
bone and soft tissue algorithms (bone window 500 -1500 Hounsfield units, soft tissue window 40 to 400 Hounsfield units), using a matrix filter (768).
The CT images were interpreted by the author and a specialist veterinary radiologist. The extent of dorsal annular prolapse was assessed on sagittal and dorsal plane images. The extent of lateral disc annulus protrusion into the entrance zone was assessed on axial soft tissue sequences. The images were windowed and levelled to allow the L7 nerve roots to be seen exiting the L7-S1 lateral intervertebral neurovascular foramen. A subjectively narrowed bony foramen, loss of peri-neural fat, and an increase in the size of the nerve roots were considered abnormal findings. The soft tissue window data were used to create 3-dimensional images of the LS junction including the lateral intervertebral neurovascular foramina. Clipping tools were used to remove the overlying ilial wings allowing visual assessment of the exit zones of the foramina. An inclusion ROI tool was used to divide the 3-dimensional reconstruction in half along the sagittal plane allowing subjective assessment of the entrance zones to the foramina. The 3-dimensional construct was rotated into a plane tangential to the expected path of the L7 nerve roots. The foraminal size was subjectively compared between flexed, neutral and extended positions.
Dogs were classified as “affected” when the physical examination findings were consistent with the results of advanced diagnostic imaging (loss of perineural fat signal, annular protrusion, subjective foraminal narrowing). Based on these criteria, all dogs in the AbNGSD group, and no dogs in the NGSD or NGH groups, met the criteria for a diagnosis of DLSS to be made.
In order to quantify the volume of the L7-S1 lateral intervertebral neurovascular foramen the CT data were then manipulated on an Extended Brilliance Workstation (EBW, Phillips, the Netherlands). To isolate the foramina from the bulk data, a batch file was created from contiguous slices of the soft tissue windowing data covering the foraminal entrance, middle and exit zones. The batch file was converted into a 3- dimensional volume rendering, then a tissue segmentation protocol was applied with thresholds set to include the soft tissue and fat structures within the foramen but excluding bone and dorsal annulus fibrosus (centre -20, width 275 Hounsfield units), (Figure 35, page 126). The region of interest was defined laterally as the most abaxial extent of the pedicle of L7 just cranial to the lateral intervertebral neurovascular foramen. The cranial aspect of the foramen was identified and a 5 mm reference line was drawn from the most lateral extent of the pedicle across the width of the pedicle to the vertebral canal. A rectangular region of interest was created using an inclusion tool. This procedure defined a 3-dimensional foraminal volume with its cranial, caudal,
articular processes of L7 and S1 and vertebral body of L7 respectively), and the medial and lateral limits determined by the 5 mm reference line. The image of the foraminal volume was then magnified and manipulated to determine the extent of extraneous volume data outside the anatomical limits of the foramen. The volume was trimmed of any excess from beyond the foramen by visually checking sequential axial images and using a freehand exclusion tool.
The volume generated was expressed in mm3 +/- a margin of measurement error. Each foramen was measured five times in each of the three positions creating 30 measurements per dog. If an outlier was noted it was dropped from the dataset and a sixth assessment performed. The five repeated volume measurements were then averaged.
In addition to the volume measurements, the longitudinal angulation of the LS junction was measured (LS junction angle). Using a sagittal reconstruction of the bone windowed sequence, a line was placed across the dorsal vertebral body of L7 and the dorsal sacral body representing the ventral floor of the vertebral canal of L7 and the sacrum respectively (Mattoon and Koblik 1993). Where the two lines intersected an angle was formed which was measured using digital imaging software (E-film). The length of the L6 and L7 vertebral bodies were measured between the end-plates of the respective vertebra at the mid-body height on mid-sagittal sections.
Statistical evaluation
All statistical analyses were performed using R (R Foundation for Statistical Computing, Vienna, Austria, 2013). The degree of intra-observer repeatability of the five repeated measurements for each volume was determined by calculating an interclass correlation coefficient. The age, weight, L6 and L7 length distributions of the groups were compared for significant differences with a linear model. The lateral intervertebral neurovascular foraminal volumes measured in neutral positioning for the NGH and NGSD dogs were pooled. The vertebral lengths of L6 and L7 were plotted
against the respective dogs’ foraminal volume in neutral positioning and a linear
regression model applied to determine the relationship between animal size and foraminal volume. There was no statistical difference between L and R side, therefore 'side' was omitted from the model. A simple linear regression was used to compare the mean volume of the lateral intervertebral foramina in each position (combined left and right data) using the NGH volume data as the reference value. The linear regression was repeated with the NGSD volume data as the reference value. Statistical significance was set at P<0.05.