Abstract
Background: Superficial siderosis (SS) of the CNS is caused by repeated slow hemorrhage into the subarachnoid space with resultant hemosiderin deposition in the subpial layers of the brain and spinal cord. Despite extensive investigations, the cause of bleeding is frequently undetermined.
Objectives: To review the clinical and imaging features of 30 consecutive patients with SS and provide insights into the underlying causes of subarachnoid bleeding in this disabling disorder.
Methods: The authors reviewed the medical records of 30 consecutive patients with clinical and MRI evidence of SS.
Results: The commonest neurologic manifestations included gait ataxia and hearing impairment. A clinical history of subarachnoid hemorrhage was relatively rare. Possible predisposing conditions were identified on history in 22 patients, the commonest being a prior trauma (15 patients). In addition to the characteristic MRI findings of SS, 18 patients had abnormalities on MRI possibly related to chronic bleeding. The most common of these was the presence of a fluid-filled collection in the spinal canal seen in 14 patients.
Conclusions: A history of prior subarachnoid hemorrhage is often absent in patients with superficial siderosis (SS). A past history of trauma is common. Prior intradural surgery may be an additional risk factor. Xanthochromia or the presence of red blood cells in the CSF is a common finding. Only rarely does angiography demonstrate the bleeding source. The presence of a fluid-filled collection in the spinal canal is a common finding on MRI and is likely related to the SS. With longitudinally extensive cavities, a dynamic CT myelogram may help localize the defect and direct the site of laminectomy. Surgical repair of a dural defect, if present, should be considered. Surgical correction of bleeding should be documented by CSF examination months after surgery. Friable vessels in the dural defect are a possible source of the chronic bleeding.
Superficial siderosis (SS) of the CNS is an uncommon and often unrecognized disorder caused by chronic slow or repeated bleeding into the subarachnoid space.1 Hemosiderin deposition in the subpial layers of the brain and spinal cord causes parenchymal damage. The cardinal features of SS of the CNS are deafness and cerebellar ataxia, and these occur in approximately 90% of cases.1
Such cases were almost never diagnosed pre mortem before the advent of MRI. CT scanning rarely demonstrates specific features of chronic hemosiderin deposition. MRI reveals a rim of hypointensity on T2-weighted images, enveloping the surface of the brainstem, cerebellum, and sometimes cortical fissures. Even with MRI, the diagnosis may still be missed because the imaging abnormalities follow the contours of the brain and can be overlooked. With no symptoms to suggest intracranial bleeding, a bloody CSF may be written off as a traumatic tap. The bleeding site goes unidentified in half the cases. Progression to a bed-bound state may occur over decades in up to one-fourth of these patients.1
Our understanding of this disorder and how best to treat is hindered by the small number of cases in reported series. The largest reported series of SS is from 1969 and included nine patients.2 The majority of the other reports include fewer cases or a single case or are compilations of multiple reports. The current series reports the spectrum of clinical, laboratory, and imaging manifestations in 30 patients with SS of the CNS evaluated at our institution since 1989.
Methods.
The current series of patients were obtained via a computerized search of the Mayo Clinic (Rochester, MN) coded medical records system (which incorporates clinical, pathologic, and autopsy diagnoses) and the records system of the Mayo Clinic Rochester Diagnostic Radiology Department. The search covered the period 1976 to 2004 and identified patients with a diagnosis of SS. All patients in this series were diagnosed after 1989. Patients with imaging evidence of siderosis limited to previous tumor or infarction were excluded. All MRIs were performed using a 1.5 T magnet and were reviewed by a neuroradiologist. Other investigations done include cerebral or spinal angiography and CT–myelography. Details of Cases 7,3 26,3,4 28,5 29,5 and 305 have been reported earlier.
Results.
Clinical features.
SS of the CNS was documented in 30 patients (17 men and 13 women), with an age range of 30 to 82 years (mean age 59 years). The commonest presentation was that of hearing loss and slowly progressive gait ataxia (see table E-1 on the Neurology Web site; go to www.neurology.org). Hearing was spared in only four patients. The most disabling manifestation was gait ataxia, present in all but three patients. Seven patients had a prior history of cord or spine injury as pre-existing possible causes for a myelopathy (table 1; see also table E-1). Corticospinal tract signs were present in an additional 12 patients. Thus, the predominant phenotype resembled spinocerebellar degeneration with progressive ataxia, hearing loss, and corticospinal tract signs.
Table 1 Demographics, possible cause per history, and imaging (MRI, angiogram, and CT–myelogram)
Sensory symptoms were present in 12 patients and were usually related to long-standing stable mono- or oligoradiculopathies. A sensory level was present in only Case 28. Lower motor neuron signs (other than those attributable to pre-existing radiculopathy) were not seen. Cognition was subjectively or objectively impaired in six patients. Other symptoms that were felt to be possibly related to the SS included seizures in four, urinary symptoms in five, visual complaints in five, and impaired smell in five patients. Olfaction was not specifically examined in most patients. Symptoms typically progressed over many years. Episodic headache was seen in four patients, and some headache episodes were suggestive of subarachnoid bleeding.
Neuroimaging.
All patients had a strikingly similar MRI appearance (figure 1; see table E-2). The cerebellum and brainstem were encompassed by a dark, hypointense rim on T2-weighted images in all except Cases 4 and 24. Often this hemosiderin deposition was also present in the sylvian fissures (19 patients), cerebral convexities (11 patients), and interhemispheric fissure (12 patients). In Case 4, the imaging findings of hemosiderin deposition were seen over the cerebral convexities but not in the posterior fossa; this patient had a history of prior supratentorial intraparenchymal hemorrhages secondary to possible amyloid angiopathy. Imaging findings consistent with hemosiderin deposition along the ventricles was noted in only Case 11. Cerebellar atrophy was present in all cases except Cases 4 and 24 (see table E-2). Often the atrophy was most marked in the superior vermis. Spinal cord imaging was available in all patients except Case 4 and was remarkable for diffuse T2-weighted hypointensity of the pial surface of the cord in all except Cases 20 and 24. Another common finding was a similar rim of T2 hypointensity along the nerve roots of the cauda, at times with clumping of the nerve roots suggesting arachnoiditis. In Case 21, old blood with reactive changes led to a hypointense signal on T2-weighted images in the cul-de-sac of the lumbosacral thecal sac. A tumor was entertained in the differential diagnosis. A biopsy showed dural tissue with fibrosis and chronic inflammation.
Figure 1. Axial T2-weighted (A through D) and proton density (E) brain MRI in Cases 6 (A, C), 4 (B), 17 (D), and 18 (E), showing signal hypointensity due to hemosiderin deposition around the medulla and cerebellar vermis (A), along the cerebral convexities (B), around the midbrain (C), around the pons and along the cerebellar folia (D), and along the interhemispheric and sylvian fissures (E). Sagittal T1-weighted (F) and sagittal T2-weighted (G, H) spine MRI in Cases 7 (F, H) and 15 (G), showing hemosiderin deposition along the cervical cord (F), thoracic cord (G), and the roots of the cauda equina (H). Axial T2 (I, J) spinal MRI in Cases 15 (I) and 7 (J), showing hemosiderin deposition around the cord (I) and along the cauda, resulting in nerve root clumping suggestive of arachnoiditis (J). Also presented is a CT showing a rim of hyperdensity due to hemosiderin around the brainstem in Case 7 (K).
CSF.
CSF evaluation was performed on 22 patients (table 2), revealing xanthochromia or red blood cells (RBCs) in 20. The RBC count was elevated in 19 patients and ranged from 18 to 20,394 cells/μL. Xanthochromia was present in 14 patients. CSF protein was elevated in 21 patients, ranging from 52 to 1,628 mg/dL. CSF glucose levels were consistently normal.
Table 2 CSF findings
Possible etiology suggested by history.
Only rarely did patients experience symptoms suggestive of a single episode or repeated episodes of subarachnoid hemorrhage. Four reported headaches, some of which were suggestive of subarachnoid hemorrhage. A possible predisposing factor for the SS was identified on history in 22 patients and included a history of head injury in 7, spine injury in 7, brachial plexus injury in 3, possible cerebral amyloid angiopathy in 1, and a dural tear in 1 (table 1; see also table E-1). Case 22 had a prior history of trauma and was subsequently identified as having a spinal arteriovenous fistula. Cases 28 and 29 had a history of brachial plexus and spine injury. Six patients had a history of a neurosurgical procedure prior to developing symptoms related to SS: shunt placement for a idiopathic intracranial hypertension in Case 3, a diskectomy in Case 5, a shunt placement for a fourth ventricle cyst in Case 10, resection of a cerebellar astrocytoma in Case 11, cerebellar cavernous hemangioma excision in Case 18, and excision of a thoracic meningioma in Case 19.
Possible etiology suggested by imaging.
In addition to the findings of SS, MRI frequently provided possible clues to the etiology of the SS that had not always been suspected on history. This included a fluid-filled collection in 14 patients (table 1; figure 2, D and F, and figure 3; see also table E-2). In four of these (Cases 14, 17, 27, and 30), the collection was fairly localized and suggested the possibility of a meningocele or pseudomeningocele. Case 28 had a fluid-filled collection and pseudomeningocele. Additional findings of possible significance included evidence of chronic hemorrhage related to possible amyloid angiopathy in Case 4 (figure 2A), increased T2 signal involving the anteromedial temporal lobes in Case 9 (figure 2B), and a cerebellar venous angioma in Case 18 (figure 2C). An odontoid fracture in Case 13 (figure 2E) and intramedullary signal change in Cases 7, 14, 28, and 29 were seen as markers of prior injury.
Figure 2. (A) Axial T2-weighted brain MRI in Case 4, showing areas of old lobular hemorrhage possibly related to amyloid angiopathy. (B) Axial fluid-attenuated inversion recovery brain MRI in Case 9, showing increased T2 signal involving the anteromedial temporal lobes and insula (left greater than right). (C) Axial T1-weighted brain MRI with contrast material in Case 18, showing left cerebellar venous angioma and postoperative changes related to resection of a cerebellar cavernous hemangioma. (D) Sagittal T2-weighted spine MRI, showing postoperative arachnoid loculations in Case 5. (E) Sagittal T2-weighted spine MRI, showing odontoid fracture in Case 13. (F) Sagittal T2-weighted and axial T2-weighted (inset) spine MRI, showing right-sided pseudomeningocele at C7 to T1 in Case 17. (G) Sagittal T1-weighted spine MRI with contrast material, showing prominent vessels on the ventral surface of the cervical cord in Case 26.
Figure 3. Sagittal T2-weighted spine MRI (A1, B1, C1, E1, F1, G1) and cervical myelogram (D1) in Cases 2 (A1–2), 6 (B1–2), 7 (C1–2), 12 (D1–2), 15 (E1–2), 19 (F1–2), 26 (G1–2), showing a transdural leak (D1–2) or a ventral fluid-filled collection (A1–2, B1–2, C1–2, E1–2, F1–2, G1–2), which is also seen on the CT myelogram (A2, D2, E2) and axial T2-weighted MRI (B2, C2, F2, G2). Cases 6 (B1–2), 19 (F1–2), and 26 (G1–2) also show atrophic thoracic cord surrounded by adhesions (dorsally at T7, T8 in Case 6, and ventrally at T9 in Case 26).
Myelography (19 patients) and cerebral or spinal angiography (22 patients) was done to look for the possible source of the subarachnoid bleeding (table 1; see also table E-2). Of the 14 patients with an area of fluid collection seen on MRI, a CT–myelogram was available in 10 and was abnormal in all except Cases 11 and 17. In an additional three patients, CT–myelogram showed abnormalities that included a poorly localized transdural leak of contrast (Case 12) or presence of a localized area of fluid collection suggestive of a meningocele or pseudomeningocele or CSF loculation (Cases 13 and 29). Dynamic CT–myelography was done in Cases 7 and 26 and showed a defect at T1 to T2, leading to the fluid-filled collection that filled with contrast material and thus confirmed free communication between the collection and the subarachnoid space.
Information provided by angiography was of uncertain significance. This included fibromuscular carotid changes in Case 16, an unruptured carotid cavernous aneurysm in Case 17, a cerebellar venous angioma in Case 18, and a spinal arteriovenous fistula in Case 22. Prominent veins along the cord or in the posterior fossa were seen in Cases 2, 7, 9, and 26 and were believed to be secondary to chronic venous hypertension secondary to sclerosis of epidural veins due to SS (figure 2G).
Surgical intervention, course, and follow-up.
Because of the progressive course and evidence of active bleeding into the CSF, a bleeding site that could be surgically obliterated was sought in all cases. Such surgical intervention directed at the suspected bleeding source was undertaken in seven patients (table 3). Of these, Cases 17, 26, and 29 had CSF analysis done weeks after surgery. The study showed absence of xanthochromia or RBCs in all three. A repeat MRI done after surgery in Cases 7 and 26 showed decrease of the fluid-filled collection (figure 4). At 1 year post surgery, Cases 7 and 29 remained stable and Case 26 reported improvement. Case 22, who underwent resection of a spinal vascular malformation, had evidence of clinical progression 10 years after surgery; however, it was unknown whether the surgery corrected the CSF bleeding.
Table 3 Surgical intervention and follow-up
Figure 4. Preoperative (A, C) and postoperative (B, D) sagittal and axial (inset in C and D) T2-weighted spine MRI in Cases 7 (A, B) and 26 (C, D), showing resolution (A, B) and decrease (C, D) in the size of the fluid-filled spinal canal collection.
Discussion.
Ferritin is known to be involved in the CNS tissue response to CSF hemoglobin and hemoglobin breakdown products.6 The roles of other iron-related proteins such as transferrin, ferritin repressor protein, and iron regulatory factor are less well characterized. Animal studies suggest that the ferritin repressor protein-immunoreactive Bergmann glia and ferritin-containing microglia in the molecular layer of the cerebellum are involved in conversion of heme to ferritin and finally hemosiderin.7
The clinicians evaluating the patients usually did not initially suspect SS. Rather, the history and examination suggested a neurodegenerative disorder. Imaging subsequently established the diagnosis of SS, with the characteristic MRI T2 hypointensity outlining the brain and spinal cord surfaces. The commonest presenting feature in our patients was slowly progressive gait ataxia and hearing impairment, at times with corticospinal signs. This presentation is consistent with prior reviews of patients with SS of the CNS.1 Blood pooling in the posterior fossa may predispose to this clinical picture.8 The selective vulnerability of the eighth cranial nerve is probably due to its long glial segment that predisposes it to iron deposition.9
SS has been reported to present as a myelopathy,10 and in a prior review of the literature, a “myelopathic syndrome” was commonly recognized.1 In some of our patients, sensorimotor symptoms and signs were consistent with a myelopathy. However, these findings were not specific and could have reflected involvement at other levels of the nervous system. Urinary symptoms suggestive of a neurogenic bladder were documented in four patients and could have been due to hemosiderin deposition on the nerve roots or CNS. Visual complaints and cognitive decline have been previously reported1,11 but were relatively uncommon in our series. A distinct pattern of cognitive and social impairment in SS has been recognized.11 Areas affected include speech production, visual recall memory, executive functions, and the ability to represent the mental state of others. Loss of smell, often with subjective taste deficits, was infrequently reported; however, olfactory dysfunction is often overlooked in the clinic. Seizures have rarely been reported in SS12 and were seen in four cases in our series.
The first clue to SS is typically from MRI, although SS can sometimes be identified by CT scanning, with a hyperdense ring around the brainstem.1 The increased recognition of this entity, however, is in large part due to the advent of MRI. MRI is very sensitive to hemosiderin and is the diagnostic procedure of choice.8,13 In our patients, T2 hypointensity due to hemosiderin deposition was consistently seen outlining the cerebellum, brainstem, and pial surface of the cord. In some patients, it was apparent along the roots of the cauda equina, with evidence of arachnoiditis. These imaging findings may also be seen in the sylvian and interhemispheric fissures and over the hemispheric convexities. Hemosiderin deposition along the ventricles is rare and was present in only Case 11, but it has been reported earlier in relationship to intraventricular tumors.14 Cerebellar atrophy is nearly always seen and preferentially involves the superior vermis. Gait ataxia is more common than limb ataxia due to preferential vermian involvement. Case 9 showed increased T2 signal involving the medial temporal lobes and insula (figure 2B). A left temporal lobe biopsy done at an outside institution showed “nonspecific neurodegenerative changes.” A patient with idiopathic SS was noted at autopsy to have necrosis in the temporal and insular regions with hemosiderin-containing subcortical microglia.15
In nearly all of our patients undergoing CSF examination, there was evidence of active bleeding. Even in the two patients without spinal fluid RBCs or xanthochromia on a single CSF examination, detection of bleeding may have been missed; prior authors have commented on intermittent hemorrhage in some cases and the need to repeat CSF studies to demonstrate bleeding.1 Resolution of CSF xanthochromia and RBCs after surgical obliteration of the bleeding source can confirm that the responsible site was operated. Postsurgical CSF examination is not a common practice in the treatment of this disorder but should be.
It is unclear that elimination of the source of chronic bleeding slows or halts the clinical progression of this disorder. However, the very consistent documentation of active bleeding and the relentless progression noted in this and other series make this an obvious therapeutic strategy. The first step is to identify the bleeding site. In our series, the history and investigations identified a possible clue for the etiology of SS in all except Cases 8, 20, 23, and 25.
In many cases, the bleeding site is either initially unknown or is inferred from circumstantial clues. Among our patients, symptomatic subarachnoid hemorrhage with sudden onset of a severe headache was relatively rare. This was in contrast to a prior review of SS cases, where it was documented in approximately one-third of patients.1 Because of the occult nature of the bleeding, a high index of suspicion needs to be maintained in order not to overlook the diagnosis. Twenty-two of our 30 patients did have a prior history suggestive of a possible predisposing factor to SS. The commonest of these was a history of trauma in 15. History of a neurosurgical procedure was present in six, one of whom (Case 3) also had a history of spine injury. Only 2 of the 22 patients with a possible predisposing cause did not have a history of trauma or prior intradural surgery. These were Case 4 who had a prior history of a cerebral hemorrhage and Case 16 who had a history of a low-pressure headache secondary to a dural tear.
Major trauma possibly causes traction of nerve roots and damages the medullary and radicular veins running along the roots.1 This may make the veins and venules fragile and vulnerable to subsequent minor trauma. Intradural surgery may inadvertently provide access to the CSF for a bleeding source within the brain.16 SS may also occur years after intradural neurosurgical procedures.17,18
An unexpectedly common finding in our retrospective review of these cases was the recognition of an abnormal fluid collection in the spinal canal in over half the patients; this included 14 cases documented by MRI and an additional 2 patients on CT–myelography. In one other, a dural leak on CT–myelogram indicated the presence of a dural defect. Of these 17 cases with a possible dural pathology, a prior history of trauma was present in 9, intradural surgery in 3, and dural tear in 1. In some cases, it was unclear on MRI whether the collection was intradural or extradural. The differential diagnosis entertained was that of an arachnoid cyst, meningocele, pseudomeningocele, or meningeal diverticulum. Of the seven patients who had some surgical intervention for their superficial siderosis, six were from this category. SS due to traumatic pseudomeningoceles after brachial plexus injury5,19,20 or meningeal diverticula following root avulsion21 has been treated surgically in other reports. The longitudinal extent of the fluid-filled collection as seen in some of our cases and its therapeutic implications have not been well recognized.
Friable vessels were seen at the site of the scar tissue in the area of pseudomeningocele formation in Case 28. The absence of CSF RBCs or xanthochromia after repair of the dural defect in Cases 17, 26, and 29 and reduction in the size of the fluid-filled collection in Cases 7 and 26 further suggest that the dural defect was the likely source of chronic bleeding. Abnormal vessels at sites of dural defect are the likely source of slow chronic bleeding, and these may not be detected on angiographic studies. In one reported patient, coagulation of a small spider angioma in an arachnoid scar resulted in CSF clearing.22 In another, cauterization of vessels in a meningeal diverticulum resulted in reduction of bleeding.21 In a report, slow chronic oozing of blood was noted in a hole in the pseudomeningocele during repair20; coagulation of the bleeding site and repair of the pseudomeningocele resulted in resolution of the xanthochromia. Normalization of CSF iron and ferritin 2 months after removal of a cauda equina ependymoma has been reported.23 The syndrome of intracranial hypotension due to spontaneous CSF leaks has been associated with increased RBCs in the CSF.24 It is speculative if a similar mechanism could be operative in the chronic subarachnoid bleeding in superficial siderosis.
In most of the reported patients, there is paucity of long-term clinical follow-up after interventions. Few surgical cases have documented resolution of subarachnoid bleeding by way of postsurgical CSF analysis. Judging clinical outcomes is difficult, given the slow progression of SS and the possible episodic nature of the bleeding. It is unclear if the process continues despite eradication of the bleeding. Progression of disease despite ending exposure to neurotoxins has been described in manganese-related neurotoxicity.25
Footnotes
-
Additional material related to this article can be found on the Neurology Web site. Go to www.neurology.org and scroll down the Table of Contents for the April 25 issue to find the title link for this article.
Disclosure: The authors report no conflicts of interest.
Received October 26, 2005. Accepted in final form January 13, 2006.
References
Disputes & Debates: Rapid online correspondence
- Superficial siderosis
- Reply from the Author
REQUIREMENTS
If you are uploading a letter concerning an article:
You must have updated your disclosures within six months: http://submit.neurology.org
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.