Misdiagnosis of Spinal Dural Arteriovenous Fistula: An Analysis of 12 Cases

Background: Spinal dural arteriovenous stula (SDAVF) is a rare spinal vascular disease. The clinical misdiagnosis rate is very high. The highest misdiagnosis rate is reported in orthopedics. The specic reason for misdiagnosis remains unclear. Objective: To investigate the clinical and imaging manifestations of SDAVF, analyze the causes of misdiagnosis, propose countermeasures, and improve the orthopedists and other specialists’ understanding of this disease. Methods: The clinical data, diagnosis and treatment of 12 patients who had SDAVF that was misdiagnosed as a different disease were retrospectively analyzed, and the modied Aminoff-Logue Disability Scale (ALS) scores before and during follow-up were compared. Results: From 2014 to February 2019, 15 patients were diagnosed with SDAVF at our institution. Twelve (80%) were misdiagnosed; of these, 6 (50%) were misdiagnosed more than twice, and 6 patients (50%) were misdiagnosed at least once. The initial diagnoses included lumbar spinal stenosis and lumbar disc herniation (LDH) in 6 patients, cervical spinal stenosis in one patient, benign prostatic hyperplasia (BPH) in one patient, myelitis in 2 patients, and syringomyelia in one patient. After the initial diagnosis, one patient was misdiagnosed with LDH, one was misdiagnosed with subacute combined degeneration of the spinal cord, and 4 were misdiagnosed with myelitis. The clinical manifestations mainly included weakness and numbness of the lower limbs, urinary symptoms, and numbness of the perineal area. In the 12 misdiagnosed patients, magnetic resonance imaging(MRI) showed signs of spinal cord edema and typical or atypical ow-void patterns. One patient had undergone extended cervical decompression and lumbar decompression. All patients eventually underwent microsurgical treatment. The average follow-up duration was 0.9 years. The modied ALS scores showed signicant improvement in gait, bladder function and bowel movement, and the differences before and during follow-up were statistically signicant (P<0.05).


Background
In the middle-aged and elderly population, patients presenting with lower limb weakness or/and numbness are more likely to seek medical advice rst from an orthopedic service, and most orthopedists diagnose diseases based on their own expertise. Since most orthopedists lack experience with spinal dural arteriovenous stula (SDAVF), they may misdiagnose the disease. According to literature, the misdiagnosis rate for this disease is as high as 80%, and the highest rate is reported in orthopedic services [1]. Many patients have severely damaged spinal cord function when they are diagnosed, which leads to permanent spinal cord injury [2]. Although digital subtraction angiography (DSA) has become the gold standard for diagnosing SDAVF, spinal-cord magnetic resonance imaging (MRI) is still a preferred imaging method in clinical practice, and its diagnostic accuracy is also closely related to the orthopedist's clinical experience [3]. The purpose of this paper is to retrospectively analyze the clinical data of 12 patients who had SDAVF that was misdiagnosed as another disease, summarize the clinical presentation, and analyze the reasons for the misdiagnosis. This work can enhance orthopedists and other specialists' understanding of this disease to improve the accuracy of the diagnosis.

General data
From 2014 to February 2019, 15 patients were diagnosed with SDAVF at our institution; 12 (80%) patients were previously misdiagnosed, and the misdiagnosis was corrected later through DSA. The patients included 10 men and 2 women with an average age of 54 years (range: 41-74 years).

Imaging examinations
Among the 12 patients, 2 patients underwent MRI of the lumbar spine at another hospital, and one patient did not undergo MRI before admission (this patient underwent computed tomography (CT) of the lumbar spine). The remaining patients had at least one MRI examination for the entire spine. The diagnosis of all the patients was con rmed by DSA, and single stulas was presented in all 12 patients. The stula was located in the thoracic segment in 10 patients, in the thoracolumbar segment in one patient, and in sacrococcygeal region in one patient.

Surgical procedure
The diseased segments were identi ed before surgery. The patient was placed in the prone position, and uoroscopy was performed to accurately locate the stula and the spinous process of the corresponding vertebral segment, which were marked by the injection of a small amount of methylene blue on the skin.
A longitudinal posterior midline incision (the marked site was considered the center) was made for laminectomy or semi-laminectomy. Under microscopy, the dural matter was incised and the arachnoid was separated to identify the drainage vein, which was then electrocoagulated and divided.

Follow-up and assessment
The modi ed Aminoff-Logue Disability Scale (ALS) [4] (Table 1) was used to score the functional status of spinal cord during physical examination before surgery and 6 months after discharge. SPSS 13.0 was used to perform paired t-tests to compare the modi ed ALS scores before and after treatment. P<0.05 was considered statistically signi cant. Spinal cord MRI was performed at follow-up visits from 3 months to 1 year after the surgery.

Results
Clinical manifestations of SDAVF All 12 patients had subacute onset of SDAVF, which mainly manifested as transverse spinal cord injury (SCI) caused by progressive venous hypertensive myelopathy (VHM). According to the initial manifestations reported in their medical history, 3 patients (25%) had weakness of both lower extremities, 2 patients (17%) had paresthesia in the perineal region or lower limbs, one patient (8.3%) had bladder dysfunction, and 6 patients (50%) had 2 or more of the above symptoms (Table 2). At admission, the course of disease of 12 patients was from 5 to 23 months. Of those 12 patients, 5 had grade III lower limb muscle strength and increased muscle tone; 2 had grade III left lower limb muscle strength and grade I right lower limb strength with increased muscle tone; 2 had grade III lower right lower limb muscle strength and grade II right lower limb muscle strength with increased muscle tone; one patient had grade I lower limb muscle strength with decreased muscle tone; and 2 had grade IV lower limb muscle strength with increased muscle tone. Twelve patients had numbness of the body and lower limbs below the level of sensory disturbance, and the level of sensory disturbance did not match the location of the SDAVF stula. Furthermore, 12 patients had di culty urinating, and 4 patients required an indwelling catheter.  (Figure 3) (for visualization purposes, the typical signs of spinal cord edema (hyperintensity) and dorsal and ventral bead-like ow voids are described as a "white radish plus black sesame seeds" sign). Tortuous abnormal vascular shadow was observed in one patient (Figure 2), and weedlike atypical ow voids signs were observed in the dorsal spinal cord in 4 patients (for visualization purposes, hyperintensity associated with spinal edema and dorsal and/or ventral atypical ow voids are described as a "white radish plus weed" sign) (Figure 1, Figure 4) (see Table 2 for complete explanations).

Types of misdiagnosis
Of the 15 patients, 12 (80%) patients were misdiagnosed, and 6 (50%) were misdiagnosed more than twice. The initial misdiagnosis included lumbar spinal stenosis (LSS) and lumbar disc herniation (LDH) in 6 patients, cervical spinal stenosis in one patient, benign prostatic hyperplasia (BPH) in 2 patients, myelitis in 2 patients, and syringomyelia in one patient. After the initial diagnosis, one patient was misdiagnosed with LDH, one was misdiagnosed with subacute combined degeneration of the spinal cord, and 4 were misdiagnosed with myelitis (Table 3) Treatments administered between the rst and second misdiagnoses and the con rmed diagnosis One patient was misdiagnosed with cervical spinal stenosis and underwent expansive laminoplasty, then was misdiagnosed with cauda equina syndrome caused by LDH and underwent removal of the nucleus pulposus and interbody fusion with bone graft and internal xation (Figure 1). Two patients who were misdiagnosed with BPH received conservative medical treatment (Figure 3). A total of 7 patients were misdiagnosed with myelitis and subacute combined degeneration of the spinal cord at the initial and second diagnoses, respectively, and 4 of them were treated with hormone therapy.

Follow-up and assessment
All patients completed the follow-up. The mean duration of the follow-up was 0.9 years (0.5 to 1 year). The gait score was 3.33 ± 0.89 before surgery and decreased to 1.42 ± 0.91 at the follow-up, and the difference was statistically signi cant (t = 5.20, p < 0.0001). The bladder function score was improved from 2.17 ± 0.72 minutes before surgery to 1.17 ± 0.94 at the nal follow-up, and the difference was statistically signi cant (t = 2.93, p = 0.003). Bowel function was also improved from 2.08 ± 0.67 before surgery to 1.0 ± 0.85 at follow-up, and the improvement was statistically signi cant (Table 4). Overall improvement was seen in the patients, but the degree of improvement was not identical among the patients. In particularly, the patients with severe symptoms and bladder/bowel dysfunction experienced nonideal improvement. Spinal-cord MRI was performed at follow-ups from 3 months to 1 year after the surgery and showed that spinal cord edema was signi cantly improved or had disappeared, and abnormal ow voids were not noted (Figures 2 and 3).

Discussion
SDAVF is a type of intraspinal vascular malformation that has been gradually recognized in clinics in the past 20 years. It occurs when the arteries supplying the nerve roots or dura mater communicate with the spinal drainage vein when passing through the dura mater at the intervertebral foramen. Because the incidence of SDAVF is extremely low and the clinical manifestations are not typical [5], early diagnosis is not easy, and the misdiagnosis rate is high [6,7]. If diagnosis and treatment are delayed, spinal venous pressure continues to increase and can cause spinal cord ischemia and edema and even irreversible injuries, such as necrosis and demyelination. Therefore, it is necessary to improve the clinical and imaging understanding of this disease to reduce misdiagnosis and missed diagnoses.

Analysis of clinical manifestations
The disease is more common in elderly men, with a male-to-female ratio close to 4:1 [8]. Ten of the 12 patients in this study were men, accounting for 83.3% of all patients. SDAVF usually begins with progressive spinal dysfunction. The most common symptoms include gait abnormality, decreased myodynamia, paresthesia, sphincter dysfunction and sexual dysfunction. Almost half of all patients initially present more than one of these symptoms. In particular, in cases of patients with bladder dysfunction, physicians should be especially vigilant about considering this disease as a possible diagnosis. Within half a year of onset, symptoms of decreased myodynamia, abnormal sensation and bladder/bowel dysfunction can all appear.

Imaging analysis
The typical imaging ndings usually include hypointense T1 or isointense signals, hyperintense T2 signals of spinal cord edema, and bead-like ow void signs in the dorsal and ventral spinal cord. However, some atypical images include local "weed" signs locally in the dorsal or ventral spinal cord. When these weed signs are present, the possibility of this disease should be considered to avoid a missed diagnosis.

Analysis of causes of misdiagnosis
1.This is a rare disease with an annual incidence of 5 to 10 per 1 million [9]; therefore, it is easily ignored by some specialists. 2. Additionally, the onset of the disease is not typical, and there are no obvious causes. The clinical manifestations may indicate the involvement of multiple health care specialties, including orthopedics, urology, neurology, pain management, and neurosurgery. Most patients seek medical advice from multiple specialists. However, specialists usually make diagnoses based on their own expertise and thus fail to pay adequate attention to this disease. This is an important reason for the misdiagnosis of this disease [10]. 3. Most patients are middle-aged and elderly men with an onset of sensorimotor dysfunction in the lower limbs; therefore, a diagnosis of degenerative changes of the lumbar spine is often the rst choice. BPH is considered when patients present with bladder dysfunction or a decline in sexual function. 4. Physical examinations are not detailed enough to provide a conclusive diagnosis. Although the patient may already have increased muscle tone and positive pathological signs of this disease, the diagnosis of lumbar degenerative diseases, such as LDH, is made only based on MRI ndings. 5. MRI may not be available in primary hospitals, so only lumbar CT is used for diagnosis. It is also possible that if MRI is used as a screening method and the initial symptoms are relatively mild, spinal cord MRI is not considered immediately because of its high cost. 5. Physicians do not fully understand the manifestations of this disease on spinal cord MRI images and only consider in ammatory changes or secondary edema of the spinal cord caused by LDH compression when mild changes in long T2 signals are observed, especially when atypical ow voids are observed near the spinal cord. 6. The level of the spinal cord scanned by MRI is too high or too low, causing a missed diagnosis.

Differential diagnosis
The median time from the onset of the patient's clinical symptoms to the diagnosis of the disease is 12 to 44 months, which is an important reason for the high disability rate associated with the disease. Therefore, early diagnosis is key to reducing the disability rate [11]. In this study, 1. Patients misdiagnosed with LDH accounted for approximately 50% (16/33) of all misdiagnosed patients. Most patients with LDH present with root pain that is obviously related to the LDH. However, the pathological signs are negative, which is a very important indicator of SDAVF. In addition, some patients with SDAVF may also have LDH, and this is also an important reason for missed diagnosis. 2. Myelitis progresses rapidly, and spinal cord swelling can be seen on spinal cord MRI images, but there is no ow void sign. In patients with atypical ow void signs, spinal angiography can be an option for con rming the diagnosis. 3. Patients with BPH usually do not have speci c clinical manifestations or signs of lower extremity abnormality, and the diagnosis is generally not di cult to make. For patients who cannot be diagnosed, further spinal cord MRI or spinal angiography should be performed.
How to prevent misdiagnosis and missed diagnosis Primary physicians should obtain detailed a medical history from patients and perform comprehensive physical and neurological examinations. The relevant specialists should improve their understanding of the clinical features of this disease. MRI can be used as a preliminary screening method to analyze myeleterosis on longitudinal and axial views. Furthermore, it is necessary to strengthen collaboration among various departments to reduce the occurrence of misdiagnosis and missed diagnosis.

Limitations Of This Study
This study had limitations. Since this was single-center retrospective study, the sample size was small, and the follow-up period was short. Although there were many limitations, the results can deepen different specialists' understanding of this disease and can improve diagnostic accuracy.

Conclusion
When patients, especially those with bladder and bowel dysfunction, have symptoms such as weakness of the lower limbs and/or numbness, the diagnosis of SDAVF should be considered. Furthermore, if MRI shows patterns resembling a white radish plus black sesame seeds or a white radish plus weeds, the diagnosis of SDAVF should be strongly considered, and further angiography should be performed to con rm the diagnosis.

Consent for publication
A written format of informed consent, which includes information on the use, disclosure and publication of patient information on the condition of anonymity, was signed and obtained from all individual participants in the study.
Baohui Yang collected the cases and drafted the manuscript. Shuai Cao analyzed the data. Xijing He designed the study. Haopeng Li performed the operation. All authors read and approved the nal manuscript. 11. Safaee, M.M,Clark AJ, Burkhardt JK, et al., Timing, severity of de cits, and clinical improvement after surgery for spinal dural arteriovenous stulas. J Neurosurg Spine, 2018. 29(1): p. 85-91. Figure 1 A 59-year-old man rst visited another hospital due to "progressive weakness of the lower limbs and di culty urinating for one month" and underwent cervical, thoracic, and lumbar spine MRIs (Figures 1a,   1b, and 1c). Figure 1b, a thoracic spine MRI image, shows a hyperintense signal consistent with spinal cord edema and weed-like ow voids of the dorsal spinal cord at T9-10 (white radish plus weed pattern). The patient was misdiagnosed with cervical spinal stenosis and was treated with open-door expansive cervical laminoplasty (Figures 1d and 1e). His symptoms were not resolved after surgery. He was misdiagnosed with LDH at his second admission and underwent lumbar decompression (Figures 1f and   1g). His symptoms worsened progressively after surgery. He was admitted to our clinic more than 9 months after the surgery and was diagnosed with thoracic SDAVF, which was con rmed by angiography (Figure 1h).

Figure 2
A 49-year-old women was misdiagnosed with primary myelitis and was admitted to our hospital due to "numbness and weakness of both lower limbs and bladder dysfunction for more than 7 months". A: Preoperative thoracic spine MRI shows bead-like ow voids and spinal cord edema (white radish plus black sesame seed pattern). B. Lumbar MRI shows abnormal tortuous vascular shadows around the cauda equina. C. Spinal DSA was performed and con rmed the diagnosis of SDAVF with the stula in the sacrococcygeal region. D. F. Follow-up MRI at 3 months after surgery shows that ow void signals and spinal cord edema have disappeared.

Figure 3
A 56-year-old man was initially misdiagnosed with BPH and then misdiagnosed with myelitis. He was admitted to our hospital due to "lower limb numbness and weakness and bladder dysfunction for more than 9 months". A: Preoperative MRI of the thoracic spine shows signs of ow voids and spinal cord edema (white radish plus black sesame seed pattern). B. Spinal DSA was performed and con rmed the diagnosis of SDAVF with the stula in the T9 level. C. Follow-up MRI at 3 months after surgery shows the disappearance of ow voids and improvement of spinal cord edema.