3. ERGEBNISSE
3.2. CSF Tau in dogs with IVDH
Cerebrospinal fluid tau protein as a biomarker for severity of spinal cord injury in canine intervertebral disk herniation
A. Roerig, R. Carlson, A. Tipold, V.M. Stein*
Department of Small Animal Medicine and Surgery, University of Veterinary Medicine Hannover, Hannover, Germany
*Corresponding author: Tel: +49-511-953-6311 Email: veronika.stein@tiho-hannover.de (V.M. Stein)
Abstract
Intervertebral disk herniation (IVDH) is a common cause of spinal cord injury (SCI) in dogs. Microtubule-associated protein tau derives predominantly from neurons and axons, making it a potential marker of neuronal injury.
A retrospective study, including fifty-one dogs with thoracolumbar or cervical IVDH and 12 clinically normal dogs, was designed to describe associations between cerebrospinal fluid (CSF) tau concentration, degree of neurological signs and motor functional recovery in dogs with IVDH. Signalement, degree of neurological dysfunction and outcome were recorded. Cisternal CSF tau values were determined by an ELISA. Associations between CSF tau concentration and various clinical parameters were evaluated.
Tau protein was significantly elevated in dogs showing plegia (median: 79.9 pg/mL;
P = 0.016) compared to healthy dogs and dogs with paresis (median: 30.1 pg/mL;
P = 0.025). Plegic dogs that improved by one neurological grade within one week had significantly lower tau protein levels compared to plegic dogs that needed more time for recovery or did not show an improvement (P =0.008). A CSF tau concentration of > 41.3 pg/mL had a sensitivity of 86% and specificity of 83% to predict an unsuccessful outcome in plegic dogs based on receiver-operating characteristics curve analysis (area under the curve = 0.887, P = 0.007, 95%
confidence interval [CI]: 0.717 to 1.057).
In conclusion, CSF protein tau levels are positively associated with the severity of spinal cord damage and therefore may serve as a prognostic indicator in dogs with IVDH.
Keywords: spinal cord injury, dog, CSF analysis, protein tau
Introduction
Tau proteins belong to the microtubule-associated proteins family (Weingarten et al., 1975). These phosphoproteins specifically localize in neurons where they bind to microtubules, promoting their assembly and stability (Weingarten et al., 1975; Binder et al., 1985). Tau is not physiologically secreted; its release into the cerebrospinal fluid (CSF) is most probably due to neuron damage or death (Mori et al., 1995).
Therefore CSF tau protein concentrations can serve as a biological marker for axonal damage in the central nervous system (CNS) (Zemlan et al., 1999; Shiiya et al., 2004).
Elevated levels of CSF tau protein have been described in human neurodegenerative disorders such as Alzheimer’s disease and in multiple sclerosis (Blennow et al., 1995; Andreasen et al., 1998; Terzi et al., 2007). At present, information on tau protein in veterinary medicine is rare. So far tau protein expression in the aging brains of dogs and cats as well as CSF tau levels in dogs with encephalitis have been examined (Head et al., 2005; Pugliese et al., 2006; Tanaka et al., 2011)
Intervertebral disk herniation (IVDH) is a common cause of spinal cord injury (SCI) in dogs, resulting in pain and neurological dysfunction. Severity of neurological signs is determined by neuroanatomical location, velocity and amount of the compressive material as well as duration of compression (Brisson, 2010). IVDH leads to SCI via a combination of primary and secondary events. Primary spinal cord injury refers to the initial mechanical insult whereas secondary injury is a biochemical cascade following the primary event and consisting of vascular dysregulation, neurogenic shock, oxidative stress and excitotoxicity (Dumont et al., 2001).
Several prognostic factors for IVDH were previously studied in CSF samples, such as the percentage of CSF macrophages (Srugo et al., 2011), CSF levels of myelin basic protein and creatinine kinase activity (Levine et al., 2010; Witsberger et al., 2012), concentration of beta-2-microglobulin (Muñana et al., 2007) and CSF glutamate concentration (Olby et al., 1999). These described studies included dogs with thoracolumbar IVDH or dogs with acute signs (Levine et al., 2006; Levine et al., 2010; Srugo et al., 2011). Therefore additional reliable prognostic factors that allow
differentiating between good and poor functional outcome, are needed. Moreover clear cut-off points for unfavourable outcome should be stated (Levine et al., 2010;
Witsberger et al., 2012).
CSF tau levels were highly elevated in human head trauma patients and a correlation was established between clinical improvement and decreased CSF tau levels (Zemlan et al., 1999).We therefore hypothesized that high levels of protein tau reflect the severity of spinal cord damage and that determination of CSF tau concentration has a pivotal value as a prognostic biomarker for motor functional recovery in dogs with IVDH.
Material and Methods
Study design and animals
Medical records of the Department of Small Animal Medicine and Surgery, University of Veterinary Medicine, Hannover, Germany, were retrospectively reviewed (2005 - 2011) for dogs diagnosed with IVDH. The study was performed according to the ethical rules of the University.
On the basis of the severity of the neurological signs 51 dogs were categorized according to Sharp and Wheeler (2005) into five grades: paraspinal hyperesthesia (grade 1), ambulatory paresis, ataxia and proprioceptive deficits (grade 2), nonambulatory paresis (grade 3), plegia with nociception (grade 4), plegia with no deep nociception (grade 5) (Sharp and Wheeler, 2005). Inclusion criteria for this study were dogs suffering from cervical or thoracolumbar IVDH as suggested by magnetic resonance imaging (MRI) findings and in most cases confirmed by decompressive surgery, a severity of grade 2 to 5 at presentation and subsequent analysis of CSF. Dogs were neurologically re-examined seven days after first
The information retrieved from the medical records included signalement (sex, age, breed), time from onset of clinical signs to presentation, previous drug administration with emphasis on glucocorticosteroids, localization of the disk protrusion/extrusion, the occurrence of pathological changes in the myelon detected by MRI and functional neurological outcome (Levine et al., 2009; Boekhoff et al., 2012).All MR images were reviewed by certified neurologists. MR images were evaluated for the presence of an intramedullary hyperintensity over the length of one vertebral body or more (Levine et al., 2009).The outcome was defined to be successful if the dog improved by at least one neurological grade within a week after first presentation at the Department of Small Animal Medicine and Surgery.
Normal research colony dogs (n = 8, Animal Experiment number: 33.42502/05-12.05) as well as four dogs, which were presented at our clinic because of orthopedic, non-neurologic diseases, were used as a control population. All control dogs were required to have normal physical and neurological examinations as well as CSF analysis.
Cerebrospinal fluid (CSF) collection
Cerebrospinal fluid was collected at the time of presentation/admission under general anesthesia from the cerebellomedullary cistern. 200 µL of samples without iatrogenic blood contamination were subsequently used for routine examination (cell count, glucose and protein concentrations) and remaining CSF samples were frozen and stored in polypropylene tubes at -20°C (Lachno et a l., 2011).
Biochemical analysis of Tau protein
Tau protein was determined by a commercially available enzyme-linked immunosorbent assay (ELISA) (Innotest hTAU, Innogenetics), which recognizes non-phosphorylated and non-phosphorylated tau protein (Vandermeeren et al., 1993). It is a solid-phase enzyme immunoassay in which tau protein or tau fragments are captured by a primary monoclonal antibody (AT120) and two biotinylated secondary antibodies (HT7bio and BT2bio). 50 µL of CSF was used for each measurement, each sample was measured in duplicates, and the mean was used for further evaluation. Previous
examinations excluded interferences in the assay format for hemoglobin, bilirubin, albumin or globulin (Nishimura et al., 1998).
Statistical Analysis
Data obtained were analyzed using a commercially available software package (GraphPad Prism Version 5.0, GraphPad Software). Variables were controlled for normal distribution using Kolmogorov-Smirnov normality test. As data were not consistent with a Gaussian distribution, nonparametric tests (Kruskal-Wallis and Mann-Whitney U test) were applied. All statistical tests were two-tailed and a P-value
≤ 0.05 was considered to be statistically significant.
For the purpose of analyzing the relation between tau protein levels and severity of neurological dysfunction, dogs were divided into two groups (A: paresis = grade 2/3, B: plegia with and without deep pain sensation = grade 4/5). Within these groups dogs were further categorized based on occurrence of functional improvement by one grade within 7 days or no functional improvement.
For the purpose of describing duration of clinical signs before admission dogs were divided into three subcategories (acute: < 5 days, subacute: 5 - 14 days, chronic: >
14 days). For analyzing the effect of administration of corticosteroids (yes vs. no), location of the spinal cord compression (cervical vs. thoracolumbar) and spinal cord TW2 hyperintensity (yes vs. no) dogs were divided in two groups, respectively.
Receiver-operating characteristics (ROC) curve analysis was performed to determine the overall effectiveness of CSF tau concentration to predict an unsuccessful outcome in dogs showing plegia due to IVDH. Additionally, ROC curve analysis was used to find out if CSF tau concentrations are able to distinguish between paretic and plegic dogs. The cut-off that maximized the Youden index (sensitivity + specificity -1) was selected as optimal.
Results
Fifty-one dogs with cervical (n = 18) or thoracolumbar (n = 33) IVDH of various breeds and with a different degree of neurological dysfunction met the inclusion criteria. Mean age of the dogs was 7.3 years (range 3 - 12.25 years). Breeds included Dachshund (n = 19), mixed breed (n = 13), Beagle (n = 2), Bernese Mountain Dog (n = 2), Cocker Spaniel (n = 2), Rottweiler (n = 2), French Bulldog (n = 2) and 9 other breeds with one dog each. Twenty-three male, 11 castrated male, 7 female and 10 spayed female dogs were examined. The neurological examination revealed 24 dogs with paresis and/or ataxia with proprioceptive deficits (grade 2) whereas 7 dogs were nonambulatory paraparetic (grade 3). Fourteen dogs displayed plegia with nociception (grade 4) whereas in 6 dogs nociception was lost (grade 5). Dogs with paresis (grade2/3, n = 31) and dogs showing plegia (grade 4/5, n = 20) were combined for further evaluation. Forty-four dogs (grade 2: n = 18, grade 3: n = 7, grade 4: n = 14, grade 5: n = 5) underwent decompressive surgery whereas 6 dogs (all classified as grade 2) were treated conservatively and one dog (grade 5) was euthanized after diagnostic imaging.
Glucocorticosteroids were administered to 23 of the 51 dogs with IVDH (45.1%) before presentation (Table 1).
The control group consisted of 12 dogs with a mean age of 5.0 years (range 1.8 - 8.8 years). One female, two spayed females, two males and seven castrated males were examined. Breeds included Beagle (n = 8) and one of each of the following breeds (mixed breed, Labrador Retriever, Boxer, Chihuahua).
n Location of spinal cord compression
Cervical T2W Hyperintensity of spinal cord
(length > one vertebrae)
Table 1. Clinical signs, treatment and MRI findings of 51 examined dogs with IVDH.
Tau was measurable in seven of 12 dogs of the control group (58.3%) and in 33 of 51 dogs with IVDH (64.7%). Median tau protein level in the control group was 20.6 pg/mL (range: 0 – 51.2 pg/mL) whereas in dogs with IVDH it was 41.8 pg/mL (range: 0 – 778.7 pg/mL). Significant differences (P = 0.049) were detected between CSF tau protein in dogs with IVDH and dogs of the control group. These differences were even more distinct with evaluation of neurological grade (Fig. 1).
control
grade 2/3
grade 4/5 0
100 200 300 400 500 500 600 700 800
*
*
Tau in pg/ml
Figure 1. CSF tau protein concentrations show significant (*) differences between plegic dogs with IVDH (grade 4/5, n = 20) and both, control dogs (n = 12, P = 0.016) and paretic dogs with IVDH (grade 2/3, n = 31, P = 0.025).
The CSF tau protein concentrations were significantly (P = 0.016) higher in dogs with plegia (grade 4/5, median: 79.9 pg/mL, range: 0 – 778.7 pg/mL) compared to control dogs. In dogs with paresis (grade 2/3) CSF tau levels were also higher
(median: 30.1 pg/mL, range: 0 – 193.1 pg/mL) than in control dogs, however, this difference did not reach the level of significance (P = 0.175). Moreover, a difference was noted within the group of dogs with IVDH as plegic dogs showed significantly higher tau concentrations compared to paretic dogs (P = 0.025).
An association could be assessed for CSF tau concentration and functional outcome.
Dogs with IVDH (n = 51) that improved by one neurologic grade within a week (n = 23) had significant lower CSF tau levels (P = 0.015) than dogs which needed more time for neurological improvement or that did not show any improvement (n = 28; Fig. 2). Considering only plegic dogs, this correlation revealed to be even more significant (P = 0.008; Fig. 3).
improvement
no improvement 0
100 200 300 400 400500 600700 800
*
Tau in pg/ml
Figure 2. Association of CSF tau levels in dogs with IVDH (n = 51) and outcome.
Twenty-three dogs showed neurologic improvement whereas 28 dogs did not
improvem ent
no im provem
ent 0
200 400 600
800 *
Tau in pg/ml
Figure 3. Comparison of CSF tau protein concentrations within the group of plegic dogs due to IVDH (grade 4/5, n = 20) with 6 of the dogs showing neurological improvement defined as amelioration by one neurologic grade within one week and 14 with no improvement. Lower CSF tau levels in plegic dogs with IVDH correlate significantly (*) with a better functional outcome (P = 0.008)
CSF tau concentrations were evaluated in relation to MRI T2-weighted (T2W) hyperintensities in the myelon, previous treatment with corticosteroids, duration of clinical signs before admission and localization of IVDH (cervical vs. thoracolumbar).
Protein tau levels in dogs with thoracolumbar IVDH (median: 48.3 pg/mL, range: 0 – 282.6 pg/mL) were higher compared to dogs with cervical IVDH (median: 24.3 pg/mL, range: 0 – 778.7 pg/mL), even though this difference was not significant.
(P = 0.238)
No statistical difference (P = 0.319) in tau levels between dogs with or without the administration of glucocorticosteroids could be detected. Additionally, no correlation of the CSF tau concentrations was found for T2W spinal cord hyperintensity (P = 0.087) and duration of clinical signs at admission (P = 0.759)
The ROC curve analysis suggested an optimal CSF tau cut-off value of > 41.3 pg/mL.
At this point the sensitivity and specificity for predicting an unsuccessful functional outcome in plegic dogs were 86% and 83%, respectively. Area under the ROC curve (AUC) was estimated at 0.887 (95% CI: 0.717 to 1.057) and the overall ability of CSF tau concentration to predict unsuccessful functional outcome was significant (P = 0.07). (Fig. 4)
0.00
0.05
0.10
0.15
0.20 0.0
0.2 0.4 0.6 0.8 1.0
1 - Specificity
Sensitivity
Figure 4. Receiver-operating characteristic curve (ROC) to predict unsuccessful outcome for tau protein concentration in the CSF of 20 dogs showing paraplegia due to IVDH. Sensitivity and specificity of this test at 41.3 pg/ml were 86% and 83%, respectively and indicated by a circle.
A CSF tau cut-off value of > 59.5 pg/mL with a sensitivity of 60% and specificity of 84% to identify plegic dogs out of all dogs with IVDH was recommended by ROC
Discussion
The pathophysiology of intervertebral disk herniation includes primary and secondary injury of the spinal cord and associated vascular structures. The severity of the injury and therefore the extent of neuronal damage is thought to be a function of the amount, speed and duration of cord compression (Kearney et al., 1988; Carlson et al., 2003). Axons are damaged either because of direct physical disruption or nerve transmission is negatively influenced by regional hemorrhages or edema (Anderson and Hall, 1993). Furthermore, secondary injury may lead to neuronal cell death due to necrosis and apoptosis. This contributes to a loss of neurons which most likely has a negative impact on outcome (Li et al., 2000). Necrosis typically occurs shortly after the primary injury, whereas apoptosis can occur weeks following the injury (Crowe et al., 1997).
Several biomarkers have been examined for their potential to indicate axonal injury.
The study of Brisby et al. (1999) demonstrated that patients with disk herniation have increased concentrations of neurofilament protein in the CSF, indicating the damage of axons in the affected nerve root. Additionally, the measurement of glial fibrillary acidic protein (GFAP) in CSF of humans after trauma to the cervical spine revealed the possibility to quantify the degree of nerve cell injury (Guéz et al., 2003).
Neuronal cell death is likely to cause a release of intracellular microtubule binding proteins such as tau into the extracellular space where they are transported by convective bulk flow to CSF (Segal, 1993). Tau protein was hypothesized to be a good marker for axonal injury because this protein is localized primarily in neurons and is enriched in the axonal compartment (Binder et al., 1985).
In humans tau is expressed from a single gene that undergoes alternate splicing resulting in six tau isoforms which range from 352 to 441 amino acids (Goedert et al., 1989). Tau proteins can be separated into two groups, the low molecular weight tau proteins expressed in neurons of the CNS, and the high molecular weight tau proteins most abundant in the peripheral nervous system (Georgieff et al., 1993). Tau proteins play an important role in the neuronal microtubules network involved in axonal transport and neuronal transmission (Drubin, 1986). By regulating microtubule
assembly, tau proteins participate in modulating axonal morphology and growth (Buee et al., 2000).
Tau has been the focus of multiple studies over the last years and different authors have shown that CSF tau levels correlate with the degree of pathological changes in the CNS. In human patients with head trauma a correlation/positive association between clinical improvement and decreased CSF tau levels was observed (Zemlan et al., 1999). Moreover, a severity-dependent elevation of tau in humans with acute SCI has been demonstrated (Kwon et al., 2010). In contrast, as tau is an intraneuronal nonreleased protein, CSF tau levels in dogs free of axonal injury are expected to be low (Mori et al., 1995). This assumption was confirmed in our study by the low median tau concentration in the control group and was proven in a previous study where a mean tau level of 29.3 pg/mL was detected in the CSF of healthy dogs (Tanaka et al., 2011).
A concentration gradient was assessed along the spinal cord for some CSF proteins (Blennow et al., 1993). However, Sjögren et al., (2001) demonstrated that the volume of CSF does not influence the tau concentration measured. This enables the introduction of CSF tau measurements into the clinical routine, where the volume of CSF collected often differs due to different sizes and bodyweights of the various dog breeds.
Results of the present study indicate that in dogs with IVDH, CSF tau concentration was associated with the degree of neurologic dysfunction and the functional outcome. Although CSF tau has not been investigated previously in dogs with IVDH, data from human traumatic brain injury as well as SCI are consistent with these results (Zemlan et al., 1999; Kwon et al., 2010).
Disk herniations with the same dynamic properties and identical volume of extruded material usually cause more severe damage in the thoracolumbar than in the cervical spine, due to the smaller epidural space in the thoracolumbar region (Brisson, 2010).
This observation was confirmed in our study by higher protein tau levels in patients with thoracolumbar IVDH compared to dogs with cervical IVDH, even though this
An area of hyperintensity greater than or equal to the length of the L2 vertebral body in T2W MRI in paraplegic dogs with thoracolumbar IVDH has proven to correlate with poor functional recovery (Schouman-Claeys et al., 1990; Ito et al., 2005). This hyperintensity was thought to be caused by hemorrhage, inflammation, edema, and necrosis (Flanders et al., 1999). As of this multifactorial pathogenesis it is not entirely surprising that there was no clear association between high protein tau levels and spinal cord T2W hyperintensity.
In our study no statistical difference in tau levels between dogs with or without the administration of glucocorticosteroids could be detected. The benefit of this treatment is controversially discussed, partly because of the risk of harmful side effects (Faden et al., 1984; Olby, 1999; Davis and Brown, 2002). In dogs with encephalitis tau protein levels were shown to be significantly lower if they were treated with glucocorticosteroid (Tanaka et al., 2011).
Using a cut-off value of 41.3 pg/mL CSF tau concentration was a significant and sensitive predictor of functional outcome in plegic dogs with IVDH. As the AUC implies a way to measure the accuracy of a diagnostic test, it can be concluded, that measuring protein tau in the CSF of plegic dogs to determine the functional outcome, has a test precision of 89% (Hanley and McNeil, 1982). According to Swets (1988) this is only a moderate accuracy for a diagnostic test. Other studies which investigated possible prognostic biomarkers in the CSF of dogs with IVDH showed comparable test accuracy (Levine et al., 2010; Srugo et al., 2011). A combination of different biomarkers might still result in the most accurate outcome prediction (Witsberger et al., 2012).
With a sensitivity of 60% and a test accuracy of 68% CSF tau concentrations could not identify the neurologic grade. This lack of sensitivity is most likely due to the fact that severity of neurologic signs can not only be explained by the amount of damaged neuronal tissue but is also substantiated in acute edema, demyelination and accompanying inflammation of the myelon. Therefore dogs may show severe neurologic signs even though the loss of neurons and with it the release of protein tau is minimal.
A possible limitation of this study is the collection site of CSF samples from the cerebellomedullary cistern. Lumbar samples are more sensitive for the detection of abnormalities because of the mainly caudal flow of CSF (Thomson et al., 1989, 1990). Thus, present results might underestimate CSF tau levels in dogs with IVDH, particularly thoracolumbar IVDH. Lumbar samples were not chosen because of possible iatrogenic blood contamination.
Previous investigations have shown that the material of test tubes may affect the results obtained for tau. Even though tau levels did not decrease when samples were collected in polystyrene tubes it is recommended to use non-absorbent
Previous investigations have shown that the material of test tubes may affect the results obtained for tau. Even though tau levels did not decrease when samples were collected in polystyrene tubes it is recommended to use non-absorbent