A Prospective Evaluation of CT in Acutely Paraparetic Chondrodystrophic Dogs

Introduction

Intervertebral disc disease (IVDD) is a common neurologic disorder in dogs with an estimated prevalence of 19% in dachshunds.¹ Conventionally, spinal myelography was the preferred diagnostic modality for IVDD, with an accuracy of 83–97% in dogs.^2–4 Despite the availability of myelography, the technique is invasive with theoretical risks of infection, seizures, and exacerbation of neurologic signs. ^2,5–10 The added cost for radiographs and exposure of the patient and staff to radiation must be considered. As a result, computed tomography (CT) and MRI have replaced myelography as the favored spinal cord imaging modalities in human medicine.^11–13 In the veterinary field, CT has become much more accessible over the last decade with the growth of specialty practices in most regions. Consequently, the applications for CT in dogs have expanded. Recent studies have described the appearance of intervertebral disk herniations on CT, and other studies demonstrated the suitability of CT for diagnosis of Hansen type I herniations, including multiplanar techniques for improved accuracy.^2,14–18 Comparisons between CT and other modalities have demonstrated improved sensitivity with CT and good agreement in the level and lateralization of disk extrusion with CT, highlighting the applicability of this modality.^2,17 Due to the degree of disc mineralization in chondrodystrophic breeds and the ability of CT to identify calcified lesions, CT offers a rapid, accurate, and thorough imaging assessment for those patients. The purpose of this study was to examine the ability of CT alone to identify spinal cord compression in chondrodystrophic breeds presenting with acute signs of thoracolumbar spinal cord dysfunction.

Materials and Methods

This was a prospective, nonrandomized study performed at a single institution. Informed client consent was obtained for every patient enrolled. Every chondrodystrophic dog matching our inclusion criteria between November 2007 and August 2009 was enrolled regardless of suspected diagnosis.

Chondrodystrophic breeds (Table 1) with acute paraparesis/paraplegia, spinal hyperpathia, and neurolocalization from the 3rd thoracic vertebra (T3) to the 3rd lumbar vertebra (L3) were included. Dogs with neurolocalization other than T3–L3, nonchondrodystrophic breeds, and dogs presenting with a history of spinal pain lasting > 7 days were excluded.

TABLE 1 Patient Information for Sixty-nine Chondrodystrophic Dogs Presenting with Acute Spinal Pathology

F, female; M, male; SD, standard deviation.

View Fullscreen

Data recorded for each patient included age, weight, sex, and neurologic grade. Grading was performed at presentation, 1 day, and 14 days postoperatively via neurologic exam. The grading scheme was modified from Tarlov’s system and is described in Table 2.¹⁹ A complete blood count, serum biochemical profile, and urinalysis were performed on animals with either concurrent illness or those dogs > 7 yr of age. For imaging studies, dogs were premedicated with either intramuscular morphine^a (0.3–0.5 mg/kg) and acepromazine^b (0.03 mg/kg) or diazepam^c (0.25 mg/kg IV) at induction. Anesthesia was induced with propofol^d (4–5 mg/kg IV) and maintained with 0.5–2% isoflurane^e via a rebreathing system with O₂ flow rates between 0.1 L/kg/min and 0.3 L/kg/min. IV isotonic fluid solution was administered at 10 mL/kg/hr for the duration of the anesthetic procedures. All dogs had a single CT scan performed. Lesions were classified according to Hansen’s criteria with Hansen type I lesions representing extruded disks and Hansen type II lesions representing protruded disks.^20–22 Patients that failed to demonstrate an identifiable lesion on CT scan as interpreted by either a surgeon or neurologist subsequently underwent myelography or MRI based on the attending clinician’s preference. Patients in which no compressive lesion was identified via further study were not taken to surgery.

TABLE 2 Neurologic Grading Scale Used for All Patients

View Fullscreen

CT was performed using a single-slice helical CT scanner^f. Patients were positioned in dorsal recumbency with their head toward the gantry and forelimbs extending cranial. A pilot scan was performed for planning purposes from the cranial extent of T3 to the midbody of the 4th lumbar vertebra (L4). If a lesion was identified at the extent of the scan limits, the scan was continued caudally until clear. The gantry was then tilted to generate slices parallel to the plane of the intervertebral discs with slice thickness of 5 mm with 1 mm overlap. Once an “area of interest” was identified, another scan with 2 mm slice thickness and 1 mm overlap was performed with at least two normal images cranial and caudal to the suspected lesion(s). If multiple lesions were suspected, this protocol was performed at each site of interest.

All scans were evaluated by the resident and attending surgeon/neurologist. The site to be explored surgically was determined at that time. Data recorded included the lesion(s) location, lateralization, and character (acute versus chronic). Acute disc extrusions were described as heterogeneous material that was subjectively hyperattenuating compared with the spinal cord (Figure 1).^14,15 Chronic herniations were homogenous with extreme hyperattenuation compared with the spinal cord.¹⁵ All images were saved and sent to a picture archive and communication system^g. Not all lesions identified on CT were surgically decompressed. Dogs with lesions that had a distinct, radio-opaque mass resulting in ≥ 25% spinal cord compression within the vertebral canal were treated surgically. A cut off of 25% was decided at the beginning of the study based on consensus between the study group members.

View Figure

• Download as Powerpoint
• Download Figure

Following imaging, surgery was performed by either a board-certified veterinary surgeon (R.D.) or board-certified veterinary neurologist (T.A.). Cefazolin^h (22 mg/kg IV) was administered perioperatively and a fentanylⁱ constant rate infusion (3–6 μg/kg/hr) was administered intraoperatively. A dorsal thoracolumbar approach was followed by either a right or left hemilaminectomy using a nitrogen powered high-speed drill and rongeurs.²³ The extruded disc material was removed via blunt curette and right angle instruments. The fentanyl infusion was continued for 24–36 hr postoperatively. Any animal previously treated with corticosteroids was continued on a tapering schedule. Tramadol^j (3–4 mg/kg per os [PO] q 8 hr) was administered for 10 days. Animals without motor function had their bladder manually expressed or had an indwelling urinary catheter placed. Phenoxybenzamine^k (0.25 mg/kg PO q 12 hr, typically 5–10 mg total dose PO q 12 hr) and/or diazepam (0.25 mg/kg IV) were used for patients with an upper motor neuron bladder. Bethanechol chloride^l (5–10 mg PO q 8 hr) was prescribed to increase detrusor contractility for select patients.

Following surgery, all CT scans were interpreted by a board-certified radiologist (T.H.) blinded to the presenting clinical signs, surgical findings, and case outcome. The radiologist was instructed to identify lesion location, lateralization, and characterize the degree of spinal cord compression by none, slight (< 25% spinal cord compression), moderate (25–50% spinal cord compression), or severe (> 50% spinal cord compression).

All animals were evaluated 1 day and 14 days postoperatively, and a neurologic exam was performed.

Results

A total of 69 dogs were studied. Represented breeds included dachshund, shih tzu, beagle, basset hound, cocker spaniel, Lhasa apso, cockapoo, pug, puggle, and Pembroke Welsh corgi. Average patient age was 6 yr (range, 2–13 yr; standard deviation [SD], 2.9), average weight was 9.9 kg (range, 3.8–25.9 kg; SD, 6.1) as described in Table 1. Eleven dogs had prior nonspecific histories of either episodic paresis or pain, possibly IVDD related. Average neurologic grade on presentation for was 3.3/5 (SD, 1.09).

CT clearly identified a compressive lesion in 63 of 69 dogs giving a sensitivity of 97% when evaluated by surgeon/neurologist. Of those cases, all 63 lesions were considered Hansen type I by the surgeon/neurologist and surgical decompression followed immediately. Dogs with negative or inconclusive CT scans (occurring in 6 of 69 dogs) were further assessed via additional imaging (Figure 2). Three dogs had surgical decompression based on additional imaging. The remaining 3 dogs had no surgery because no compressive lesion was identified.

View Figure

• Download as Powerpoint
• Download Figure

The most common lesion localization site was the 13th thoracic vertebra to the 1st lumbar vertebra. Fifty-five of the 69 dogs had one lesion, 10 dogs had two lesions, and 1 dog had 4 lesions (Table 3). Eighty-four lesions were noted on the 69 CT scans performed. Seventy-eight lesions were treated with surgery (based on arbritary selection criteria set at > 25% compression of the spinal cord) and 6 lesions were nonsurgical (< 25% compression of the spinal cord). Seventy-six of the surgical lesions had an acute appearance on CT, and 2 lesions appeared chronic. All 6 nonsurgical lesions appeared chronic.

TABLE 3 Distribution of Disc Herniation in the Thoracolumbar Region (78 Lesions)

*

For lesions noted at the extent of the computed tomography (CT) scan, the scan was extended caudally until no additional pathology was noted.

T, thoracic vertebra; L, lumbar vertebra.

View Fullscreen

Surgeon and radiologist agreed on CT lesion level 92% of the time (95% CI, 0.863–0.982). There was no significant difference in CT imaging interpretation between surgeon and radiologist (χ², 3.37; degrees of freedom, 8; P = 0.8487). Lesion level was disagreed upon in six dogs. All of those patients had surgery based on surgeon/neurologist’s CT interpretation, and four of those dogs improved within 2 wk. One patient was euthanized for suspected ascending myelomalacia 1 day postoperatively. The remaining patient had a small amount of disc material at the hemilaminectomy site with significant spinal cord swelling and bruising. The radiologist considered that patient’s CT scan normal. That patient deteriorated neurologically 1 day postoperatively and was euthanized due to complications related to renal failure. A necropsy was not performed.

Surgeon and radiologist agreed on lesion side 91% of the time (95% CI, 0.846–0.973). In 6 of 7 cases with disagreement, the surgeon provided a side, whereas the radiologist documented a ventral (midline) lesion only. For the remaining case, a primary lesion was not documented by either surgeon or radiologist; therefore, a side was not listed.

For cases in which CT was inconclusive (6 of 69 cases), additional imaging was performed using myelography (n = 3), MRI (n = 2), or both (n = 1) as shown in Figure 2. No surgery was recommended based on myelography. One myelogram was performed to assess whether compression was contributing to the patient’s clinical signs. That dog was diagnosed with diskospondylitis. Cord swelling was diagnosed in the remaining three myelography patients (thinning of contrast columns). Surgery was not recommended initially in any of those three dogs. One patient progressively worsened, and MRI was performed 14 days later. MRI revealed multiple hyperintense, intramedullary lesions throughout the thoracolumbar cord with dorsal cord compression over L4. Surgical decompression and biopsy confirmed lymphoma. MRI in the remaining two patients revealed evidence of type II disc herniation. Surgery confirmed nonmineralized disc compression in both patients.

The first follow-up examination was performed 1 day postoperatively. From an average preimaging neurologic grade of 3.3/5, the average grade 1 day postoperatively was 3.4/5 (SD, 1.04). Eighty percent of patients (53 of 66 patients) either improved in their neurologic grade or were unchanged. Eleven dogs worsened by one grade, and two dogs worsened by two grades.

The second follow-up examination was performed 14 days postoperatively. Average neurologic grade was 2.1/5 (SD, 1.05). Compared with the 1 day postoperative status, 98% of patients (60 of 61 patients) either improved or remained stable. Twenty-seven dogs improved by one grade, 17 improved by two grades, 3 improved by three grades, 1 patient improved by four grades, and 12 patients remained stable. One patient worsened neurologically, progressing to plegia (grade 5). Four dogs were euthanized, and 1 dog was lost to follow-up. Three dogs were euthanized due to lack of neurologic improvement, and 1 dog was euthanized due to glomerular nephropathy. The neurologic grade on presentation for those 4 dogs was 4/5 and 5/5 1 day postoperatively.

Discussion

The authors tested the ability of CT to identify spinal cord compression in chondrodystrophic breeds presenting with acute signs of thoracolumbar spinal cord dysfunction. The authors found that CT identified a compressive lesion in the population of dogs tested in 91% cases (97% sensitivity). Furthermore, the surgeon’s interpretation of lesion level and lateralization were not significantly different from that of a board-certified radiologist, with 92% and 91% agreement, respectively, similar to a previous study.⁴

The authors elected to image from T3 to L4 to assess the full extent of the spine within the authors’ neurolocalization parameters. Although 75–85% of intervertebral disc herniations occur between the 11th thoracic vertebra and the 2nd lumbar vertebra, imaging restricted to that region may overlook significant compressions elsewhere.^1,24 Consistent with other reports, the authors documented that 25% of surgical lesions in the included dogs occurred either cranial to the 11th thoracic vertebra or caudal to the 2nd lumbar vertebra (see Table 3).^1,24 That finding reinforces the need to perform a complete screen of the entire thoracolumbar spine when using CT exclusively. Alternatively, based on these study results, scanning from the ninth thoracic vertebra to L4 would have identified the vast majority (95%) of spinal cord lesions. Considering the speed of CT, there appears no distinct reason not to image from T3 to L4 routinely, and the authors continue to do so at their institution. The extent of the spine to be scanned, however, should be based on individual patient and institutional considerations to limit radiation exposure.

For patients requiring additional imaging, the authors found that myelography did not offer any information beyond that provided by CT, but MRI was useful. The patient diagnosed with diskospondylitis had clear vertebral changes identified on CT, and in retrospect, myelography was unnecessary. The patient diagnosed with spinal cord lymphoma had vertebral bone lysis identified on CT; however, a lesion supporting the neurologic localization was not identified on the initial CT scan. MRI performed in this patient 14 days later accurately identified a hyperattenuating mass overlying L4. Two additional dogs were diagnosed with Hansen type II IVDD, confirmed via MRI. Based on those results, the authors recommend MRI, when available, be performed in dogs when CT is inconclusive. The use of CT for identification of either nonmineralized herniations or soft-tissue opacities of the spinal cord needs to be further evaluated.

CT was used without the aid of contrast medium in this study. Calcification within the annular support structure of discs and ability of CT to detect highly radio-opaque material makes IVDD patients ideal candidates for CT imaging alone.^2,14–16 A prospective study evaluating the appearance of herniated disc material on CT found the use of subarachnoid or IV contrast unnecessary as the mineralized disc material was clearly visible.¹⁵ Similarly, a recent study by Schroeder et al. (2011) comparing CT with IV contrast medium (CTIV) and CT myelography (CTM) found little difference between techniques for determining level and laterality of compressive lesions.²⁵ Those authors documented sensitivity values of 97% and 94% for lesion level in the CTIV and CTM groups, respectively, which compares favorably to the results in this current study. Furthermore, the use of CTIV was as diagnostic as CTM for compressive intervertebral discs.²⁵ Although theoretically considered to be a safe procedure, risks with giving IV contrast are reported and should be weighed for each patient.^26,27 Based on this study’s results, CT detected 100% of Hansen type I IVDD lesions without the use of IV contrast or myelography. Therefore, the use of IV contrast in chondrodystrophic dogs is not necessary.

In this report, lesion sidedness had an agreement between surgeon/neurologist and radiologist of 91%. Agreement improved to 99% (77 of 78 cases; 95% CI, 0.962–1.01) if cases without a reported side (radiologist) were excluded. Regardless, the strong agreement between surgeon/neurologist and radiologist emphasizes the ability of CT to assist in preoperative planning. This study’s findings compare with Hecht et al. (2009) who demonstrated very good interobserver agreement of CT, myelography, and surgery for diagnosis of level and lateralization of disc extrusion.¹⁷

Following decompression of surgeon/neurologist-identified lesion(s), the vast majority of dogs improved clinically. That suggests that the lesion(s) were causative; however, the authors cannot directly attribute improvement with 100% certainty to successful surgery. The role of spontaneous recovery as documented by others must be considered.^28,29

Eighty-seven percent of patients (60 of 69 patients) either improved or remained stable 2 wk postoperatively; however, five dogs were not available for follow-up information. When the analysis was limited to patients that had follow-up data available, 94% of patients (60 of 64 patients) demonstrated clinical improvement. That degree of clinical improvement compared to other reports where the majority of dogs were ambulatory within 2 wk and had maintained that improvement months later.^30,31

The authors attempted to fulfill medium and long-term follow-up examinations on the study patients at 3 mo and 6 mo postoperatively; however, only 25% and 35% of patients returned for neurologic examinations and telephone interviews, respectively. Although it did not appear that any patient deteriorated neurologically based on either examinations or telephone interview, those data were not included due to incompleteness and subjectivity.

As with any study, limitations are inherent. Although agreement between surgeon/neurologist and radiologist was good, the site operated was the site chosen by the surgeon/neurologist. A difference in improvement may have been noted if the authors had pursued the site identified by the radiologist. Additionally, having multiple individuals interpret images and testing interobserver agreement would demonstrate the reliability of CT. Establishing a strong agreement between individuals of varying experience levels (i.e., trained surgeon versus resident in training) would better support the authors’ notion of the use of CT as a sole modality. The possibility of overlooking lesions smaller than 2 mm has to be considered. For inconclusive CT scans, additional imaging was performed rather than repeating the CT scan with thinner slices. Despite that limitation, the authors feel that a screening scan of 5 mm followed by a detailed scan of 2 mm slice thickness would be appropriate based on this study’s findings. Currently, there are no recommended reference parameters for CT imaging in the canine thoracolumbar spine for this study population. Another limitation in the study design was the lack of a control population of dogs. Due to ethical concerns, dogs were not randomized into study versus control populations as that would preclude some dogs with true spinal pathology the benefit of diagnostics and/or surgery. In this study, the gold standard for lesion identification was surgical findings. An obvious limitation with this approach is that the intervertebral space(s) identified as abnormal was/were the only space(s) approached for decompression, leaving the possibility of other lesions unidentified. Although the authors attempted to limit bias with medium-term clinical follow-up, the authors’ gold standard has inherent limitations. In an ideal study format, comparing CT, MRI, and surgical findings (with or without postsurgical imaging) would eliminate many shortcomings of the current study; however, patient well-being would not support this design.

Conclusion

CT using the described protocol was a reliable, noninvasive, and clinically useful tool for detecting spinal compression in chondrodystrophic dogs. CT provided a firm diagnosis for lesion location and lateralization in 91% of the included dogs. The authors propose that when available, CT be considered for diagnosis and surgical planning in chondrodystrophic breeds presenting with acute paraparesis/paraplegia and neurolocalization between T3 and L3.