Spinal subarachnoid diverticula in dogs: A review

Abstract

Spinal subarachnoid diverticula are fluid dilations of the subarachnoid space that can cause a compressive myelopathy in dogs. These diverticula are usually associated with high motion areas in the cervical and caudal thoracic spine. The definitive etiopathogenesis has not been determined but likely involves congenital or acquired causes. Pugs, French bulldogs, and Rottweilers are overrepresented breeds. Clinical signs typically include ataxia, paresis, and upper motor neuron urinary and/or fecal incontinence; pain is a less common feature. Diagnosis is based on advanced imaging, with magnetic resonance imaging now being favored given the additional detail of the spinal cord parenchyma that can be obtained. Outcomes are better with surgical intervention than with medical therapies, though there is a lack of long-term data. No superior surgical technique has been identified, and questions remain on the significance of addressing leptomeningeal adhesions at the time of surgery. Clinical signs can recur, though not always due to recurrence of diverticulum formation, and pugs may be less likely to have a successful long-term outcome.

Introduction

Spinal subarachnoid diverticula (SAD) are fluid dilations in the subarachnoid space that can cause compression of the spinal cord and an associated myelopathy. In humans, these diverticula make up 1% to 3% of all mass-like lesions of the spinal canal (1) and, although considered rare in dogs, the overall incidence has not been reported. The first case report in a dog was published in 1968 and was termed a leptomeningeal cyst (2). Since then, 39 English publications have described a total of 324 cases with various terms for the condition including spinal meningeal cyst (3), meningeal cyst (4), arachnoid cavitation (5), arachnoid pseudocysts (6), spinal intradural arachnoid cysts (7), and most commonly arachnoid or subarachnoid cysts (8–21). More recently, a shift towards the term “subarachnoid diverticulum” has been used to describe the synonymous lesion as the lack of a histopathological epithelial lining renders the term “cyst” inaccurate. A classification scheme has been adapted from human medicine in which Type 1 is extradural without involvement of the nerve roots, Type II is extradural with involvement of the nerve roots, and Type III is intradural (22). This classification scheme aims to differentiate between types of meningeal cysts based on anatomic location and has been used in dogs, with all cases using this description being Type III (3,6,17,18,23–27). Though many case reports have been published on SAD in dogs, the etiology remains unknown, with broad categories of congenital and acquired subarachnoid diverticula being described. Associated leptomeningeal adhesions and arachnoid webs have been reported in conjunction with SAD in dogs and humans. Arachnoid webs are rare intradural pathologies that share many imaging and intraoperative similarities with the canine SAD (28,29). Thoracolumbar and cervical spinal cord segments are most commonly affected with clinical signs consistent with a progressive myelopathy in these regions. Though diagnosis of the condition is becoming more frequent with the advent of advanced imaging techniques, a standardized medical or surgical treatment protocol has not been identified, making it difficult to determine overall prognosis. A review of the literature is provided to discuss the typical etiologies and clinical features, diagnosis, treatments available, and associated outcomes.

Etiopathogenesis

The definitive etiology of subarachnoid diverticula is unknown. Previous case descriptions have discussed congenital and acquired disease as broad etiological categories though the disease is likely multifactorial. It has been speculated that spinal dysraphism (3–5,8) is a congenital cause in dogs. In children, subarachnoid diverticula have been associated with neural tube defects (30), although most often these are accompanied by meningoceles/meningomyeloceles rather than subarachnoid diverticula (2). Because age is variable across all reports, it is difficult to prove a solely congenital or developmental cause. One of the more popular hypotheses discussed by many authors is the theory of a 1-way valve formed by arachnoid proliferation in which cerebrospinal fluid (CSF) is allowed to flow into the region with pressure changes but cannot flow out (2,4,5,8,10,14,17,18,20,25,26). This supports a congenital etiology despite the range of ages over which clinical signs become apparent as it allows for progressive expansion and therefore, progressive compression of the spinal cord.

Certain breeds, most notably pugs, rottweilers, and French bulldogs, are over-represented in the literature. Pugs make up 28% (92/324), Rottweilers 19% (62/324), and French bulldogs 13% (43/324) of the reported cases that identified the breed of dog affected (2,6,7,16–18,21,23,24,27,31–39). This suggests a genetic or hereditary cause for subarachnoid diverticula formation. One case series described a consistent phenotype and common sire in 7 related pugs with signs of progressive cervical myelopathy between 3 and 4 mo of age, suggesting a hereditary component for cervical subarachnoid diverticula in pugs (7). Another report discussed 2 young Shih Tzu littermates with similar onset and cervical lesions consistent with subarachnoid diverticula (11). With the cases involving an over-representation of Rottweilers (6,12,16,17), a genetic predisposition has been proposed due to their conformation and large heavy heads.

Despite the support for congenital lesions or genetic predispositions, concurrent or previous disease at the level of the diverticulum has been described and warrants consideration in the pathogenesis of acquired etiologies. In humans, acquired diverticula have been reported secondary to infection, inflammation, trauma, idiopathic, or iatrogenic (lumbar puncture or surgery) causes (1). In dogs, there have been cases secondary to trauma (18), caudal cervical spondylomyelopathy (18), cryptococcal meningitis (40), a migrating porcupine quill (41), and tumors/lumbosacral sarcoma (42). More commonly, authors have identified intervertebral disc herniation and/or its associated surgical treatment as a potential cause for acquired subarachnoid diverticula. Four previous cases described dogs with previous intervertebral disc herniations treated surgically with no evidence of a diverticulum at the time of diagnosis, but a diverticulum present on later imaging (13,16,32,43). Arachnoiditis has also been proposed as a potential cause (2,17) but the lack of inflammatory cells on histopathology and CSF analysis (2–5,8–11,13,15,17,18) makes this less likely. It is unknown what role concurrent disease at a nearby site plays if present at the time of diagnosis of the subarachnoid diverticulum — whether the SAD is the cause of the clinical signs or whether the other disease process is responsible. However, it is worth noting that in 1 study (32) 21.3% of dogs, mainly pugs (33.3%) and French bulldogs (61.5%), had neurologic disease diagnosed at the same or an adjacent site. In particular, in pugs, caudal articular process dysplasia has been speculated as having a potential association with SAD (44). An additional study of neurologic disease in French bulldogs exclusively, found that 11.3% had a subarachnoid diverticulum as the cause of their neurologic signs. Of these cases, 64% also had adjacent segment disease (35). And, in a recent study in pugs with thoracolumbar SAD, 58% had intervertebral disc disease at the same or adjacent sites (39). These findings suggest that concurrent disease can play a role in subarachnoid diverticulum formation, and the predisposition in some breeds may be due to their higher prevalence of adjacent segment disease compared to other breeds.

The significance of leptomeningeal adhesions in the formation of or contribution to subarachnoid diverticula is unknown, although they have been noted in a small number of case reports and studies (2,5,12,13,17,18,21,33,34,39,42). Pia-arachnoid adhesions can be caused by previous surgery (13), trauma (18), tumors (42), or anything causing vertebral instability (24,45). Chronic vertebral instability or meningeal irritation can lead to the formation of adhesions as the area between the pia and arachnoid becomes scarred. The adhesions can be focal thickenings but can also encircle the spinal cord leading to a so-called constrictive myelopathy (46) as well as further disruption of normal CSF flow. This leads to additional fluid and subarachnoid space enlargement (24,45) either causing subarachnoid diverticulum formation and/or worsening associated spinal cord compression.

Clinical features

Signalment

In the largest retrospective study of subarachnoid diverticula in dogs, pugs, Rottweilers, and French bulldogs were the most common breeds affected. In the same population, 78% of dogs were male (32). This is consistent with 2 other published studies in which 80% were male (6,39), indicating a clear sex predisposition that was not noted earlier given the many single case reports or smaller case series making it difficult to establish significance. The reason for male dogs to be affected more commonly is unknown and it remains unclear what role hormones play in subarachnoid diverticula formation.

Of all published cases, the age at diagnosis ranged from 18 wk to 13 y old (2–27,31–43,47,48). In a study of 122 dogs, the overall median age of onset of signs was 36 mo (32) and did not vary with diverticulum location. However, in a smaller study (18), dogs with SAD in a cervical location were younger with a mean age of 31 mo compared to a mean age of 6.2 y in those with thoracolumbar diverticula.

Location

In the 4 largest studies including various dog breeds, the incidence of cervical location ranged from 41% to 71% compared to a thoracolumbar location at 29% to 58% (16,18,32,34). Certain breeds or different body weights, however, are more likely to be associated with specific localizations. Pugs, apart from the case series of related pugs (7), tend to have a thoracolumbar localization (23,32,34,39) and they are often older with mean age of onset in this breed reported as 4.9 y in 1 study (32) and 7.3 y (39) in another. French bulldogs also have a predisposition for thoracolumbar localization with 1 study reporting that 88% were located in the thoracolumbar region (35). Conversely, Rottweilers tend to have cervically located lesions (6,12). In general, dogs with cervical diverticula were significantly larger (18,32) and dogs with thoracolumbar lesions were more likely to be small breeds (7,18,26,32,33). In a review of all extramedullary spinal cysts in dogs, of those with subarachnoid diverticula, 85% with cervical lesions were large or giant breed dogs and 82% with thoracolumbar lesions were small or medium breed dogs (25). The reason for this remains unclear, however, as previously suggested, the conformation of larger dog breeds with heavy heads likely plays a role in the location of the lesions affecting these breeds of dog. Lesions also tend to occur in more mobile areas of the spine with C2–C3 the most commonly reported site in the cervical spine, followed by C5–C6 (5–7,10–13,17,18,26,31,32,37). The most common region for the thoracolumbar spine is T9–T13 (26), and 77.3% of French bulldogs have lesions within the region of T9–T12 (35). The variation in spinal cord compression with a dynamic lesion here may also explain the variable onset and range of clinical signs. Typically, subarachnoid diverticula are single lesions, but they can occur as multiple lesions (12,16,17,32,34,37). As patients are generally imaged based on examination findings and lesion localization, rather than the entire spinal cord, the incidence of multiple lesions may be underestimated. If found, the significance of multiple lesions is unknown as, if outside the lesion localization determined from examination, these may be incidental findings that do not contribute to clinical signs. Most lesions are located in the dorsal midline subarachnoid space (83% to 90%) (25,26,32) but they can also occur ventrally, laterally, or in a combination of these surrounding the spinal cord. In 1 study (16), dogs with bilobed or multilobed lesions were all Rottweilers, and in another study of 10 Rottweilers (6), all bilobed lesions occurred in a caudal cervical location.

Clinical signs

In general, clinical signs reflect those of any compressive myelopathy including ataxia and paresis, and hypermetria in some cases due to disruption of the spinocerebellar tracts, with signs typically being bilateral given that most lesions are located along the midline. Proprioceptive ataxia was the most common clinical sign in 1 study (32), present in 92.6% of cases, and hypermetria was present in 21.3%. Paresis, however, was much less common, and 50% of the dogs with paresis had concurrent neurologic disease that could have accounted for this finding. These findings are likely due to the more commonly dorsal localization of the diverticula, affecting the sensory pathways more than the motor tracts. Signs are typically slowly progressive in 91% to 94% of dogs (18,32) but can wax and wane, while some dogs are asymptomatic with diverticula found incidentally (5). Reasons for the typically slow progression are an increase in the amount of fluid and therefore compression of the spinal cord over time, loss of plasticity with chronic spinal cord compression, and potentially progression of any concurrent neurologic disease processes.

Upper motor neuron (UMN) fecal incontinence (8,18,20, 23,24,32–34,48), urinary incontinence (18,27,32,34,38), or both (2,12,14,18,37,39,42,48) have been reported, most commonly with thoracolumbar lesions (18,26,32). As lesions are usually located dorsally, fecal incontinence typically occurs secondary to compression of the dorsally located ascending sensory pathways (18,32,48). Disruption of these pathways interrupts relay of sensory information to the sensory cortex, via the thalamic nuclei, for conscious recognition of rectal distension and defecation. An overall incidence of incontinence has been reported to range from 8% to 8.2% (26,32). One study reported an incidence of urinary incontinence, fecal incontinence, or both at 3.3%, 4.1%, and 4.1%, respectively (32). Although this incidence is low overall, UMN fecal or urinary incontinence in combination with other clinical signs should raise the suspicion for a subarachnoid diverticulum as a differential diagnosis.

The presence of spinal pain has been variable in the literature with the most recent reports (32,26,49) quoting an incidence of 18.9% of dogs showing discomfort. However, of all published cases, only 10% were specified to be painful on palpation or as part of their history (2,8,14,16,18,20,32,38,41,42). Given that some dogs have concurrent disease, as well as leptomeningeal adhesions, this may play a role in the detection of spinal discomfort. The incidence of spinal pain in dogs with subarachnoid diverticula is much lower than that reported for neuropathic pain in 44.5% of human patients with intradural spinal arachnoid “cysts” (30). This discrepancy between dogs and humans may reflect the difficulty detecting discomfort in veterinary patients, the owner’s ability to recognize early signs of pain, and/or our understanding of the underlying cause of the disease.

Diagnosis

Diagnosis has classically been achieved via myelography though, with the advent of advanced imaging, computed tomography (CT) myelography, and magnetic resonance imaging (MRI) have become more commonplace. Imaging is the main diagnostic test used, although CSF analysis findings and histopathology of surgically addressed lesions have been described in the literature as adjunctive findings. Survey spinal radiographs can detect vertebral abnormalities or associated spinal malformations, or rarely enlargement of the spinal canal at the affected site (8), but they are typically normal and not helpful in the diagnosis of subarachnoid diverticula (3,10,13,17).

Myelography and CT-myelogram typically reveal a tear-drop-shaped collection of fluid and contrast medium in the subarachnoid space causing focal compression of the spinal cord (10,17,26,32,35,50). It is described as starting as a gradual dilation of the subarachnoid space that ends abruptly in a teardrop-shaped accumulation of contrast medium (17,24,50). There are some benefits of CT-myelography over myelography alone, as it can help determine lateralization more accurately (13,16), can provide some indication of spinal cord atrophy (17), and allows visualization of ventral spinal cord attenuation more readily (10). Despite these advantages, it does not provide information about the spinal cord parenchyma or the presence of adhesions.

As for most myelopathies, MRI is the imaging modality of choice given its greater detail of soft tissue structures. Magnetic resonance imaging is ideal for evaluating the spinal cord parenchyma to find concurrent lesions such as edema or syringomyelia (16,20,26,37,49) and adhesions (17). The overall incidence of syringomyelia with SAD is not reported, though it was present or suspected in some reports (6,13, 16,18,20,21,31,36,38,40) and 1 report (21) showed resolution of syringomyelia on repeat MRI following surgical treatment of the SAD. A recent imaging study using cine balanced fast field echo (cine bFFE) found an incidence of syringomyelia in 50% of dogs, usually occurring cranial to the SAD more commonly than caudal (51). However, many earlier studies relied on myelography or CT myelography for diagnosis which may underestimate the actual presence of syringomyelia. Additionally, it is difficult to determine whether syringomyelia is present secondary to the SAD or concurrent disease, making the true incidence associated with SAD unclear. On MRI, SAD appears as T2W hyperintense, T1W hypointense, FLAIR hypointense dilations of the subarachnoid space, consistent with CSF and can have the characteristic tear-drop shape on sagittal images seen on myelography or CT myelogram (Figures 1A, B; Figures 2C, D). Though MRI is the imaging modality of choice, T2W images alone make it difficult to differentiate between cerebrospinal fluid and surrounding fat (50). T1W images can help with this differentiation as fat remains hyperintense and fluid becomes hypointense; however, it can still be unclear. Because of this, additional imaging sequences are available in high field scanners to aid in diagnosis and help determine the extent of the SAD. One such sequence that is commonly used is the half-Fourier acquisition single-shot turbo spin-echo pulse sequence (HASTE) in sagittal plane which increases the signal intensity of cerebrospinal fluid, while fat has little or no signal (47) so it helps visualize the subarachnoid space and has a relatively short acquisition time (Figure 1C). In a study comparing diagnosis of SAD with T2W imaging alone and HASTE, the sensitivity of T2W alone was 25% compared with the sensitivity of T2W and HASTE at 52.8%. Though not perfect, false negatives were reduced from 75% to 47.2% after the addition of HASTE (47). Figure 2 shows a dorsal subarachnoid diverticulum, with Figures 2A and B highlighting the difference between T2W images and HASTE sequences. Another study favored the additional sequence of 3D constructive interference in steady state (3D-CISS) over HASTE due to its lower susceptibility for artefact, superior differentiation between cerebrospinal fluid and spinal cord, and ability to detect arachnoid webs or adhesions (37). Compared to T2W images, Tauro et al (37) found that the 3D-CISS sequence was 100% sensitive in diagnosing subarachnoid diverticula, although its specificity was reduced to 92.1% (compared with T2W at 97.4%) given the misdiagnosis of SAD in 3 control studies that did not have the disease. Although it can increase diagnosis of SAD, the 3D-CISS sequences do take longer to acquire than HASTE. 3D-CISS also helps visualize the “scalpel sign,” named for its appearance when compressing the spinal cord, which is present with arachnoid webs (37). It is also important to note that meningeal/subarachnoid fibrosis is a separate constrictive pathology that can be difficult to differentiate from SAD given its T2W hyperintense, T1W hypointense appearance on MRI and its often dorsal location (52). Additional sequences are susceptible to over-interpretation and reviewer experience should be taken into consideration when evaluating images.

Figure 1.

A — A T2W transverse image of a ventral SAD at the level of the first thoracic vertebra (T1). The arrow points to a T2W hyperintense fluid dilation in the ventral subarachnoid space. B — Shows the same lesion, indicated by the arrow, is now hypointense on T1W imaging in a transverse plane. C — A HASTE image highlighting the ventral subarachnoid fluid dilation (arrow) that starts gradually and ends abruptly consistent with a subarachnoid diverticulum.

Figure 2.

A T2W sagittal image (A) and a sagittal HASTE image (B) highlight the difference between the 2 sequences with the dorsal subarachnoid diverticulum detected at the level of T11–L1 indicated by the arrow (more easily noted on the HASTE sequence). At the level of the arrow on the sagittal images, the T2W and T1W transverse plane images (C, D) show a T2W hyperintense, T1W hypointense subarachnoid fluid dilation indicated by arrowheads.

Cerebrospinal fluid analysis and histopathologic findings help support the diagnosis and rule out other diseases. Typically, CSF is normal, though about 20% will show non-specific elevations in total protein and 9.2% to 10% have evidence of a mild mononuclear pleocytosis (26,32,49). One study reported that 70.8% had a normal cell count (32). Histopathologic findings of surgically excised tissue lack an epithelial lining and generally show fibrosis or connective tissue proliferation (6,7,16,17,24,25, 32,39), though some are completely normal tissue (17,39). Few cases have shown inflammatory changes (17,33,39).

Treatment and prognosis

Medical management aims to reduce production of CSF, lower its volume, and reduce any surrounding inflammation; surgical management focuses on decompression of the spinal cord for improvement in clinical signs. Although medical and surgical treatments are often discussed, there is little detail in the literature regarding standardization of treatments, as well as widely variable follow-up times, making it difficult to determine optimal management and assess overall outcome. Prior to 2017, medical management was only described in 5 cases (5,7,18,23) with treatments ranging from exercise restriction, to tapering courses of “anti-inflammatory” prednisone [either an unspecified dose (5) or 0.36 mg/kg body weight, PO, q12h, reduced by 50% every 2 wk and discontinued after 6 wk (23)], and 2 with no specified management. The only study assessing medical management and its outcome as it compares to surgical management is a study of 96 dogs in which 52% of dogs were treated medically, with the treatment of choice being variable courses of prednisone in 44/50 dogs (34). Based on recheck visits and owner assessments, over a median follow-up of 24 mo, only 26% showed improvement on prednisone. Anecdotally, corticosteroids are often recommended before surgery, but in a recent study 5 cases in which steroids were used for 2 mo or longer in the perioperative period, showed no association with a better outcome (39). Whether or not it helps in removing the secondary intramedullary changes observed in the cord (edema, pre-syrinx, syrinx/syringomyelia) is also unknown.

Although prednisone is often the medication of choice for reducing CSF production and improving its absorption through the arachnoid villi, there is no evidence of its efficacy for this condition or that proves this mechanism of action. Omeprazole, a proton pump inhibitor of gastric parietal cells, has been purported as an adjunctive treatment to reduce cerebrospinal fluid production. It is routinely used in management of hydrocephalus and syringomyelia secondary to caudal occipital malformation syndrome, although no studies have been conducted on its use in treatment of SAD. The lack of information on reliable medical management, the well-known side effects of corticosteroid administration, and because clinical signs are secondary to spinal cord compression all support surgical intervention as the superior option.

Surgical management has been described with various procedures and approaches including hemilaminectomy, dorsal laminectomy, and ventral slot, based on the location of the diverticulum, combined with some variation of durotomy, durectomy, diverticulum fenestration, dural marsupialization, and vertebral stabilization. To date, no one procedure is superior and follow-up times are widely variable making assessment of success with each separate procedure difficult to determine. Although not statistically significant, 1 study did find a trend towards a better outcome when marsupialization of the dura was performed (18). Conversely, in a larger study evaluating the outcome in 46 dogs treated surgically (34) 82% showed improvement and most of these dogs (28/46) had a durectomy performed while only 3/46 had the dura marsupialized; another study showed no relationship between marsupialization and outcome (39). This suggests that marsupialization may not be superior to other techniques. As it can be technically challenging to perform and risks tension on the dura, its efficacy may be more related to the skill of the surgeon performing it rather than the technique itself.

What is becoming a common theme, regardless of the other surgical procedures performed, is to dissect leptomeningeal adhesions at the time of surgery; in human surgery, the adhesions (arachnoid webs) are dissected and the dura is closed (29). One study found that when ventral adhesions were dissected, compared to just opening the diverticulum itself, the spinal cord was able to return to a more normal position ventrally (24) suggesting that adhesions play a role in SAD pathogenesis and clinical signs. Alhough dissection of these adhesions was successful, recurrence is unknown and when fibrotic tissue is present complete removal may not be possible. One study aimed to restore normal flow of CSF after obstruction secondary to leptomeningeal adhesions by placing a shunt tube as a bypass (45). Of the 7 dogs undergoing this surgery, 4 had improvement with 3/7 having a complete recovery over a 5- to 17-month follow-up period further supporting the benefit of addressing pia-arachnoid adhesions in outcome. Because adhesions are hypothesized to be secondary to instability (24,45) causing small, repetitive injury to the area, the question remains whether stabilization should be a feature of surgical management. A recent paper has investigated the role of vertebral stabilization as part of SAD management in a group of 5 dogs (38) and results did not show a clear advantage of stabilization. More cases are needed to assess the role of stabilization on outcome, but it is still questionable whether or not stabilization is even indicated as the reason for adhesion formation has not been definitively determined. However, in a recent retrospective evaluation of vertebral stabilization in pugs with known caudal articular process dysplasia undergoing surgery for intervertebral disc herniation, subarachnoid diverticula, or pia-arachnoid fibrosis (52), 3/5 pugs with SAD showed improvement 10.2 to 24 mo after surgery. Again, larger studies of vertebral stabilization as part of treatment of SAD are needed, though this highlights its potential in patients with concurrent disease processes in which instability plays a role in pathogenesis.

Most papers have reported an overall successful outcome and improvement in clinical signs, mainly gait, in the short term. However, few have assessed long-term outcome (over 12 mo) with surgical management in a larger cohort of variable breeds of dogs, apart from Mauler et al (34) and 1 other study in 17 dogs (18). In this study, in the 11 dogs that had follow-up over 12 mo (14 mo to 72 mo) that were managed surgically, 8/11 (4 cervical location, 4 thoracolumbar location) had improvement or normalization of gait though 1 of these dogs worsened and was euthanized at 14 mo post-surgery. Those with cervical lesions were larger to giant breed dogs with an age range of 7 mo to 2 y, and those with thoracolumbar lesions were all smaller breeds with an age range of 11 mo to 9 y. From all previously published cases, there are 15 describing UMN incontinence (7 fecal, 2 urinary, 6 both) (2,18,23,27,48). Overall, 9/15 dogs (3/7 fecal, 1/2 urinary, and 4/6 both) showed improvement or resolution of incontinence (2,18,23,27,48) with 2 having complete resolution within 2 wk (48).

Despite a lack of longer-term data, surgical management seems to have a better outcome in terms of gait and incontinence than the reported medical treatment options, although success with surgical procedures is likely multifactorial. No definitive predictors of prognosis have been identified. One study (18) did not find the following statistically significant but did show a trend towards a positive outcome in patients that were less than 3 y old and those that had a duration of clinical signs less than 4 mo. The mean survival times in 1 study for dogs treated medically and surgically, if euthanized as a result of the SAD, were 15.2 mo and 12 mo, respectively (34). Despite reports that patients can show a recurrence of signs at a rate of 20% (18), the reason for this is largely unknown as most owners at that time do not elect further workup. One retrospective study (33) found that the median time to recurrence of clinical signs after surgery was 20.5 mo, but on repeat imaging only 3/8 showed that this was secondary to the SAD. Other causes found as sources of recurrent signs were formation of a laminectomy membrane and herniation of the spinal cord through the laminectomy defect, highlighting the importance of covering the laminectomy site following surgery. This suggests that the overall prognosis for SAD may be better as recurrence of signs may not be due to regrowth of the actual diverticulum. However certain breeds, mainly pugs, may have an overall worse prognosis. A recent study evaluating short- and long-term outcomes following surgery in pugs (39) found that though 82% had a successful short-term post-operative outcome (6 mo), in the long term (12 mo or longer) 86% of pugs showed deterioration with 50% of those undergoing repeat MRI that showed a recurrent SAD or new SAD. Interestingly, leptomeningeal adhesion dissection was not found to influence outcome in this study. Reasons for the apparently worse prognosis in this breed are not definitively known, though as previously stated, they do have a higher rate of concurrent disease that may contribute to their overall outcome. For example, caudal articular process dysplasia, present in 91.2% to 97% of pugs (53,54), can be found in clinically normal or abnormal dogs (44,46,49,56). The secondary effects of this vertebral malformation are vertebral instability and constrictive myelopathy characterized by spinal cord compression and fibrous adhesions (44,46,56). It has been reported that 22% of dogs with caudal articular process dysplasia had a concurrent SAD (44). This disease may provide a potential explanation for the worse prognosis associated with SAD in this breed though, as previously mentioned, vertebral stabilization in these patients may improve post-operative outcome in this breed (55). Of the pugs in the previous report (39), 6/25 had identified vertebral articular dysplasia based on high field MRI; however, it could not be reliably evaluated on low field MRI making it difficult to eliminate this as a contributing factor for post-operative deterioration.

In conclusion, spinal subarachnoid diverticula in dogs are becoming more commonly recognized with advancement in imaging techniques, though a definitive etiology is unknown. The disease likely represents a congenital etiology in younger patients and acquired etiology in older patients. It should be considered a differential diagnosis in dogs with a progressive, non-painful myelopathy where incontinence is noted, particularly in overrepresented breeds. Though no clear surgical technique shows a superior outcome, surgical intervention is warranted in these cases given the lack of evidence to support medical therapies. Surgical treatment, even with variable follow-up times, allows for a higher rate of success but certain breeds, mainly pugs, may carry an overall worse prognosis. In cases of recurrence of clinical signs, re-investigation with advanced imaging is recommended to identify the definitive cause. Further research should include the role of leptomeningeal adhesions and whether there is a need for stabilization, the correlation of duration of clinical signs with outcome, the influence of concurrent disease on overall outcome especially in certain breeds, and the validity of medical therapies. CVJ