An objective index for spinal cord compression on computed tomography in Thoroughbred horses

Abstract Background Computed tomographic myelography can be a useful tool for evaluating vertebral canal stenosis. However, an index of spinal cord compression is yet to be established. Objectives This observational descriptive study aimed to establish an index for spinal cord compression using computed tomography (CT). Methods Twenty‐three Thoroughbred horses (age, 155–717 days; weight, 205–523 kg) with suspected cervical vertebral malformation were subjected to computed tomographic myelography in dorsal recumbency using large‐bore gantry CT to define the entire cervical vertebrae from C1 to C7. Subsequently, the height of the spinal cord was measured in the sagittal plane reformatted using curved multi‐planar reformation (MPR), thereby comparing it with stenotic ratio (i.e. dividing the area of spinal cord by that of the subarachnoid space) measured in the transverse plane. The measurement was performed at the level of each of six intervertebral spaces, for a total of 138 sites. Accordingly, the appropriate cut‐off value for spinal cord height was determined using the receiver‐operating characteristic curve, from which the area under the curve with 95% confidence interval was estimated. Results The spinal cord compression cut‐off value was 7.06 mm, with an area under curve of 0.84. A weak relationship was observed between spinal cord height and stenotic ratio (R 2 = 0.08, p < 0.05). Conclusions Following curved MPR, a cut‐off value of 7.06 mm may serve as an index for spinal cord compression.

losses (Hoffman & Clark 2013). Early diagnosis and prognostic determination are some countermeasures for reducing such economic loss.
X-ray examination, the first and relatively easy-to-use diagnostic tool for orthopaedic disorders, is widely available at many veterinary clinics. Findings on cervical vertebral plain radiographic projections, including subluxation and degenerative joint disease, are suggestive of abnormalities (van Biervliet et al., 2006;Levive et al., 2007). The use of the intervertebral ratio was a limitation of radiographic assessment as it was reported to be unreliable, with variations of 5%-10% (Scrivani et al., 2011). In addition, radiography is unable to identify spinal cord compression given the similar X-ray attenuation of the spinal cord and cerebrospinal fluid. Therefore, myelography, wherein a contrast agent is injected into the subarachnoid space, has been utilised to determine spinal cord compression (Aleman et al., 2014;Rose et al., 2007;van Biervliet et al., 2006). Although lateral cervical myelogram can reveal the site and degree of dorsal and ventral compressions, determining lateral canal stenosis remains difficult (van Biervliet et al., 2006;Szklarz et al., 2019). Moreover, the interpretation of a myelogram is somewhat subjective (van Biervliet et al., 2004).
Magnetic resonance imaging (MRI) is a useful tool for spinal cord imaging that does not require injection of contrast agent into subarachnoid space. Thus, MRI plays an important role in the diagnosis of spinal cord disorders among small animals (Wisner & Zwingenberger 2015). However, MRI of the equine neck has been described only in post-mortem specimens (Janes et al., 2014(Janes et al., , 2015Mitchell et al., 2012;Sleutjens et al., 2014;). This is because it requires a gantry size that is as wide as the shoulder of horses. Nevertheless, the gantry size for MRI was originally designed to accommodate a human-sized trunk, and even foals have a shoulder size much larger than a human trunk, suggesting that the entire neck cannot fit into the gantry. For this reason, the use of MRI for the entire C1 to C7 spinal cord in living horses has not been reported.
Equine computed tomography (CT) was first introduced in 1987 ( Barbee et al., 1987). However, it was not until 2010s that the use of CT examination of the spinal cord began in living horses (Veraa et al., 2016). Contrast agent injection into the subarachnoid space was necessary to delineate the dura mater and pia mater. Thus far, CT myelography has provided useful information on live horses affected with CVM (Lindgren et al., 2020). One study had quantitatively evaluated stenosis using CT myelography by calculating the stenotic ratio (i.e. dividing the area of spinal cord by that of subarachnoid space). The stenotic ratio is used to evaluate the area of the spinal cord relative to the area of the subarachnoid space, with larger values indicating lesser subarachnoid space relative to the spinal cord size (Yamada et al., 2016). However, in this previous study, calculation of the stenotic ratio utilised images with a symmetric appearance of the transverse plane using a multiplanar reformation (MPR) procedure by adjusting the sagittal and dorsal planes simultaneously. Therefore, delineating the area of spinal cord and of subarachnoid space is laborious.
CT images can be reformatted from the transverse plane to the dorsal and sagittal planes in one acquisition. Simultaneous comparison in each vertebra is difficult given that the transverse CT plane is obtained differently for each cervical vertebral level. However, the true sagittal F I G U R E 1 Conventional method was impossible to use as the angled neck configuration fit C1 to C7 into one single plane. Curved multi-planar reformation reformatted the image of the angled neck along the structure in one single plane plane can simultaneously compare each intervertebral site, but during CT image acquisition in horses, the angled neck configuration could not fit C1 to C7 into one plane. Curved MPR reformats the image of the angled neck along the structure in one single plane ( Figure 1).
According to the literature, equine CT is performed in lateral recumbency (Gough et al., 2020;Lindgren et al., 2020). However, lateral recumbency often results in an angled neck due to the difference between the heights of the skull and shoulder on the CT table (Gough et al., 2020;Lindgren et al., 2020). This study used an equine CT scanner in which the gantry had an opening gantry size of 90 cm. This custom-made equine CT patient table was made of radiolucent carbon fibre, which can withstand a load of 500 kg ( Figure 2). Thus, we positioned the horse on the CT patient table, with the patient in dorsal recumbency with the neck straighter than that in lateral recumbency.
In addition, curved MPR was used for image processing. This observational descriptive study aimed to establish an objective index for spinal cord compression.

Study design, sample and setting
This study was an observational descriptive study. Twenty-three

Statistics
Data were statistically analysed using the JMP v.15.2.0 software (SAS Institute Japan, Tokyo, Japan). The appropriate cut-off value for spinal F I G U R E 4 Impingement was defined as a deficit of contrast agent between the border of the dura mater and the border of the pia mater.

RESULTS
CT images of the entire cervical vertebral column from C1 to C7 were obtained in all 23 live horses. Contrast agent leakage from subarachnoid space to subdural space was observed in seven out of 23 cases. A weak negative correlation was found between spinal cord height in the sagittal plane and stenotic ratio in the transverse plane (R 2 = 0.08, p < 0.05) ( Figure 6). Moreover, no significant difference between spinal cord height and age in days was observed (R 2 = 0.0003, p = 0.85) (Figure 7).

DISCUSSION
In dorsal recumbency, using large-bore gantry CT was able to define the entire cervical vertebrae from C1 to C7. The conventional sagittal CT plane cannot accommodate C1 to C7 in one plane due to angular deviations present. This was addressed using curved MPR given F I G U R E 7 Relationship between spinal cord height and age in days. Squares indicate no spinal cord compression. Considering that the central nervous system of a 155-day-old horse has stopped, this index could be applied to horses between 155 and 717 days old F I G U R E 8 Subdural contrast agent leakage may have led to the underestimation of the stenotic ratio that it reformats the image of the angled neck along the structure in one single plane. Additionally, we were able to simultaneously compare each level of apparent spinal cord impingement. Producing a symmetric appearance of the transverse plane by adjusting sagittal and dorsal planes simultaneously was unnecessary. Furthermore, subdural contrast agent leakage may have led to the underestimation of the stenotic ratio (Figure 8), given that it led to considerably increased areas of subarachnoid space being measured, thereby resulting in a low stenotic ratio (Yamada et al., 2016 Curved MPR allows the visualisation of entire cervical vertebral structures with minimal modification of the original data (Kanitsar et al., 2002). Curved MPR sagittal planes have thus far been used only for observing the location of spinal cord compression in veterinary medicine (Yamada et al., 2016). This image-processing method has been well established and is used for measuring various vessels in humans (Nam et al., 2017). Hence, it was considered that the measurement obtained using a curved MPR image was accurate.
Although dorsolateral compression of the spinal cord can be barely observed in the sagittal plane, we observed a weak but significant rela-

CONCLUSIONS
Following curved MPR, a cut-off value of 7.06 mm may serve as an index for spinal cord compression.

CONFLICT OF INTEREST
The authors declare no conflict of interest.

ETHICS STATEMENT
This study protocol was approved by the Animal Experiment and Wel-

DATA AVAILABILITY STATEMENT
The data used to support the findings of this study are available from the corresponding author upon request.