Follow‐up MRI appearance of the surgical site in dogs treated for thoracolumbar intervertebral disc herniation and showing ongoing or recurrent neurological symptoms

Abstract Reherniation and reoperation rates of 4.5%–36% are reported in canine patients treated for intervertebral disc herniation (IVDH). Decision‐making for surgical reintervention can prove challenging, especially since common postoperative changes are poorly described on MRI. The purpose of this single‐center, retrospective, descriptive study was to describe the MRI characteristics of the surgical site in dogs treated for thoracolumbar IVDH and presenting for ongoing or recurrent neurological signs. Twenty‐one patients were included for a total of 42 MRI studies. Chondrodystrophic breeds, specifically Dachshunds, were overrepresented. Mean number of days between surgery and second MRI was 335 (range 2–1367). Metallic susceptibility artifacts were seen in seven of 21 cases (33%), but these were limited in extent, spanning on average 1.3 vertebral bodies. In 11 cases, spinal cord compression suspected to be clinically significant was found at the surgical site; the extradural compressive material consisted of intervertebral disc material only, or a combination of intervertebral disc material and hematoma or inflammatory changes in 10 cases, and a displaced articular process and fibrous tissue in one case. The latter is a newly described complication of mini‐hemilaminectomies. Paravertebral soft tissue changes and vertebral new bone formation varied according to the postoperative stage at which the patients were imaged. The results of this study supported the use of MRI as a diagnostic modality for spinal imaging following IVDH surgery, and showed that the presence of extradural disc material at a spinal surgical site is common along with various vertebral and paravertebral changes.


INTRODUCTION
Recurrence of clinical signs affects an estimated 4.5-12.3% 1,2 or 6.4-36% 3-5 of canine patients operated for intervertebral disc herniation (IVDH), with or without concurrent disc fenestration respectively. In such patients, decision-making for surgical reintervention is made difficult by the lack of characterization of common postoperative changes on magnetic resonance imaging (MRI). The choice of imaging modality itself may be subject to query, due to concerns regarding MRI-specific artifacts following surgery.
Residual spinal cord compression is commonly reported following surgery, with variable rates depending on the technique performed. [6][7][8][9][10] When neurological signs have persisted or recurred postoperatively, the most commonly identified causes include herniation of additional intervertebral disc material (IVDM) from the surgical site, incomplete removal of the herniated IVDM, and formation of hematoma, seroma, or fibrous scar tissue. 10,11 However, deciding on surgical reintervention is complicated by the fact that the location and amount of extradural material do not always correlate with the severity of the reported clinical signs, 12,13 and additionally spontaneous regression of extradural IVDM has been reported in one dog. 14 Therefore, reoperation may not always be indicated in patients showing ongoing pathological changes of the spinal cord and extradural space, and decision-making would be more accurate if clinically significant or insignificant postsurgical findings were better characterized. This however poses the challenge of performing a longer general anesthetic and postoperative MRI study in patients that may not clinically benefit from it, and it may therefore be difficult to justify this in a clinical setting. Finally, although MRI is commonly accepted as the modality of choice for imaging the central nervous system, there is a risk of susceptibility artifacts rendering an MRI study non-diagnostic following spinal surgery due to the presence of microscopic paramagnetic particles left from drilling bone, with one study speculating this risk to be high on low-field MRI in postoperative patients. 15 Susceptibility artifacts in MRI studies are visualized as a focal distortion of the image in proximity to adjacent materials of different magnetic susceptibility. 16 They are greatest on gradient echo sequences, and will be larger with increasing magnetic field strength. 16 In postoperative patients, these artifacts have been reported to be primarily caused by ferromagnetic objects, and this includes not only implants but also microscopic particles that can result from the use of a drill or rongeurs to obtain bony windows. 15,[17][18][19] The objectives of this study were therefore to search for a population of dogs having undergone surgery for thoracolumbar IVDH as well as a postoperative MRI, and (1) assess the frequency and diagnostic impact of susceptibility artifacts on postoperative MRI studies; and (2) describe the changes seen at the surgical site on MRIs performed to investigate ongoing or recurrent neurological signs.

Data recording and analysis
Patients' records were reviewed by a veterinary radiology resident (A.-L.P. intensity of the nucleus pulposus, respectively, and this is illustrated in Figure 1. Susceptibility artifacts were classified as absent, moderate, or marked, these last two categories depending on whether they allowed partial or no assessment of the adjacent vertebral canal, respectively.
As susceptibility artifacts hindered the assessment of certain imaging criteria, this was taken into account when calculating prevalence rates of other imaging findings. The length of the cord affected by the susceptibility artifact was estimated in a number of vertebral bodies included in the area of signal distortion, using the T1W and T2W sagittal and transverse sequences. When present, extradural material was classified as markedly, moderately, mildly, or not compressive if it was occupying over 50% of the diameter of the vertebral canal, between 50 and 25%, less than 25%, or associated with an unchanged shape of the spinal cord and intact CSF column, respectively. The bony defects were assessed subjectively as "clearly visible" or "poorly visible", and new bone formation was classified as "absent" or "present". Tethering of the

Case demographics
Clinical and imaging details for each case are provided in Support-

Technical parameters for the included MRI studies
In total, 42 MRI studies were available for review. All MRI studies but one were performed using a 0.27 Tesla permanent MR magnet (Esaote VetMR Grande, Genova, Italy). One postoperative MRI was performed using a 1.5 Tesla MR magnet (Phillips Achieva, Phillips Healthcare, Best, Netherlands). Besides the required T2W and T1W sagittal and transverse sequences focused on the surgical site, additional sequences were also reviewed when available including T1W

Second MRI examination
The mean number of days between surgery and second MRI study was 335 (range 2-1367 days) days. Eight patients were included in the early postoperative period (range 2-14 days, mean 6 days), and 13 patients in the late postoperative period (range 52-1367 days, mean 537 days).

Susceptibility artifacts
Susceptibility artifacts were present in seven cases (7/ When assessing specifically the central T2W hyperintensity of the intervertebral discs that were not herniated on the first MRI study but herniated on the second study, six were classified as completely degenerate, two as partially degenerate, and one as nondegenerate on the initial MRI.

Changes to the vertebrae
The changes to the vertebrae could not be accurately described in the presence of adjacent susceptibility artifact, and therefore the  ipsilateral to the tethering was visible, along with widening of the contralateral CSF column. This is illustrated in Figure 5. Two of these patients had a good neurological outcome, and 2 were lost to follow-up.

Soft tissues
Unilateral changes in the paravertebral soft tissues were also appreciated in the 16 postoperative MRI studies that were not affected by susceptibility artifacts, both in the early (7 studies) and in the late (9 studies) postoperative groups. -the subcutaneous positioning of these signal voids allowed their visualization despite susceptibility artifacts). These changes were never seen in patients imaged later than 14 days after surgery.
From 14 days onward, five of nine (56%) patients showed a focal reduction in the volume of the epaxial muscles ipsilateral to the surgery, and these muscles had a similar increased signal intensity on both T2W and T1W sequences when compared to the preoperative studies. A tract of T2W and T1W hyperintense tissue could be identified in two (2/9, 22%) patients of this group along the spinous process, in the area of the surgical approach. This is also illustrated in Figure 5.

DISCUSSION
Findings from this study supported using follow-up MRI of dogs treated for IVDH for characterizing the changes at the surgical site, with only a few occurrences of MRI studies being rendered poorly diagnostic due to susceptibility artifacts. Several of these postoperative changes could also be related to the stage of the postoperative period, with some being more or less common depending on the time elapsed since surgery. Due to the presence of pain or neurological deficits in the patients included in this study, the clinical impact of these changes diagnosed on MRI could not be ascertained. The population included in the present study showed characteristics that were similar to those from other publications on canine IVDH, with chondrodystrophic breeds and more particularly Dachshunds being overrepresented, and most patients being middle-aged (median age of 6 years 2 months old). 4,[22][23][24][25] In 16 of 21 patients (76%) included in this study, metallic susceptibility artifacts were either not present or did not hinder the evaluation of the vertebral canal. This does not support the findings of an earlier study, that speculated that susceptibility artifacts would be common postoperatively and therefore MRI may not be an adequate imaging modality for patients that had undergone spinal surgery. 15 It is also interesting to note that when present, the susceptibility artifact would affect the visualization of the adjacent vertebral canal, but this was limited in length and therefore most studies impacted by susceptibility artifacts would remain diagnostic for most of the spinal cord in the field of view. Certain setting adjustments may also be performed to reduce the impact of susceptibility artifacts on the diagnostic quality of MRI sequences, but this was beyond the scope of the present study. Finally, it was also noted that the marked susceptibility artifacts were associated with mini-hemilaminectomy procedures, whereas the moderate ones were all associated with hemilaminectomy procedures. Although the equipment used and burring intensity and process were similar for both surgical techniques, a possible reason for this might be better access and therefore more thorough intraoperative saline flushing of the surgical site in hemilaminectomies as compared to mini-hemilaminectomies. However, the number of cases is low and this may therefore not be a significant reproducible finding.
Extradural IVDM at the surgical site was described in 11 patients In these 11 patients, the extradural causes for spinal cord compression included ones commonly reported, such as IVDM, hemorrhage, and inflammatory changes, but interestingly in one case, the compression was caused by a displaced articular process. The latter was one of two cases that suffered from the displacement of a cranial articular process of L1 following a mini-hemilaminectomy T13-L1. Although there has been a report of spinal instability resulting from bilateral minihemilaminectomy and pediculectomy in a dog, 26 displacements of an articular process has not yet been described as a potential complication of mini-hemilaminectomies. As evidenced by the present study, this may warrant surgical reintervention if the displaced articular process extends into the vertebral canal.
Nine patients showed a disc hernia distinct from the initial surgical site on their second MRI. When reviewing the central T2W hyperintensity of these intervertebral discs on the first MRI study of each patient, most (8/9, 89%) showed some degree of loss of this hyperintensity, indicating degeneration. This is consistent with the pathophysiology of IVD disease and previous publications. 5,25,27 Bony defects were more clearly discerned  After the initial postoperative two weeks, just over half of patients (56%) showed a reduced volume of the operated epaxial muscles compared to the contralateral musculature, occasionally associated with an increase in signal intensity compatible with fatty infiltration.
Although no muscle biopsies were performed, possible explanations for this include loss of innervation and/or loss of blood supply, causing muscular atrophy.
The T2W and T1W hyperintense tract visible in two patients of the late postoperative group along the spinous process of the operated vertebra were hypothesized to be fatty infiltration in the area of dissection of the epaxial muscles from their bony attachments.
Finally, approximately three quarters of the included population (76%) showed a linear tract in the subcutaneous fatty tissues in the area of the surgical approach, of a signal intensity suggestive of fibrosis or scar tissue.
Limitations of this study include first of all its retrospective nature, which introduces variables such as non-standardized MRI sequences, and record-keeping. The authors however implemented strict inclusion criteria to mitigate this. Several different combinations of residents and specialists were involved in the surgeries, and although surgical approaches for both hemilaminectomies and mini-hemilaminectomies are standardized to some degree, certain aspects of these techniques will inherently vary with the surgeon involved. Surgical reports did not offer details such as the length of the incision, the number of muscle detachments, and the size of the bony window. All the data were also collected from a single institution, and may therefore be biased by a number of factors such as the clinicians involved, the surgical material used, and the imaging acquisition methods. Case numbers were also small, although still comparable to other studies on the subject.
Because several patients were treated conservatively following their second MRI, surgical confirmation was often not available for the findings of the follow-up MRIs. All the patients included in the present study underwent a second MRI due to ongoing or relapsed neurological signs; it is therefore uncertain whether the changes described would be found in patients that recover uneventfully, and to what degree they contributed to the recurrence of neurological signs in our patients.
Assessment of bony changes would also potentially have been more accurate on CT than MRI, and therefore further studies using CT to evaluate bony healing and remodeling following spinal surgery may be of interest. Finally, it is also important to consider that the low diagnostic impact of the susceptibility artifacts in this study may only be valid for low-field MRI, as it is less sensitive to local magnetic field inhomogeneities compared to high-field MRI.
In conclusion, findings from this sample of dogs indicated that, contrary to what was suggested in a previous study, metallic susceptibility artifacts are uncommon in postoperative low-field MRI studies, and that the disruption caused to the diagnostic quality of the MR images is focally limited to the surgical site. MRI can therefore be consid-