Superficial Siderosis (Superficial Hemosiderosis) of the CNS
Superficial siderosis (SS) of the central nervous system is an uncommon disease characterized by accumulation of hemosiderin in the meninges, brain surface, spinal cord and cranial nerves. The deposition of hemosiderin, which may be cytotoxic to underlying tissue, results from chronic bleeding into the subarachnoid space. In many cases, the precise source of bleeding is not identified. The most common clinical presentation is one of progressive sensorineural hearing loss and ataxia, as was seen in this patient. Superficial siderosis should be considered in the differential diagnosis of progressive sensorineural hearing loss and/or ataxia because (1) it is easily diagnosed by MRI, which has high sensitivity for detecting heme products, (2) it is a potentially treatable condition, if a cause of bleeding can be identified, and (3) diagnosis of SS may avoid unnecessary searches for other causes of hearing loss and ataxia.
The first clinico-pathological description of SS of the CNS was a case presented to the Chicago Neurological Society in 1908. An extensive review by Fearnley, et al. (1995), noted only 87 cases reported in the world literature, of which only 63 cases had sufficient details to be reviewed. Most of these cases were diagnosed at autopsy, as the first antemortem diagnosis (made by surgical biopsy) was not made until 1965 (1). The male:female ratio is 3:1 and age of onset of clinical symptoms ranges from 14 to 77 (1), though some degree of siderosis pathology can be seen even in neonates who survive hemorrhagic lesions for some time (2). Duration of disease can be up to 38 years (1). With the advent of MRI scanning, early detection of this condition is becoming more common; since the Fearnley (1995) review, reports of SS have been increasing.
The cardinal symptom of SS is hearing loss, found in 95% of cases. Nearly all cases progress bilaterally over a course of 1-15 years, ultimately leaving the patient totally deaf or only with a small island of low tone preservation (1). Cerebellar involvement is nearly as common (88% of cases), resulting in progressive cerebellar ataxia. Of these cases, 56% involved both the limbs and gait, while 44% involved predominately gait. A myelopathic syndrome with bilateral pyramidal signs occurred in 50% and an additional 26% had minor symmetrical pyramidal signs such as hyperreflexia (1). Bladder disturbance occurred in 24% of cases, while sensory signs were found in 13% (1).
Although dementia has been frequently cited in other, smaller reviews and case reports, Fearnley's extensive review found dementia in only 24% of cases (though when present, it was often severe). Anosmia has been reported in 17% of cases, though this is likely a conservative estimate, as this sign may not always have been sought (1). It is not clear whether SS predisposes to seizures, as bleeding into other areas of the CNS often does.
Macroscopically, there is a brownish discoloration of the leptomeninges and of the adjacent CNS parenchyma up to a depth of 3 mm (1). Particularly pigmented are the cerebellum (especially the vermis), basal frontal lobe and olfactory bulbs, temporal cortex, brainstem and cranial nerves (especially VIII), spinal cord and nerve roots. Microscopically, there is hemosiderin deposition, neuronal loss, reactive gliosis, and demyelination (1).
The conversion of blood pigment within the CSF to hemosiderin requires contact with living tissue. The deposition of hemosiderin within CNS tissue is the final result of the following sequence of events (3):
Koeppen et al. induced experimental siderosis in rabbits by weekly intracisternal injections of red blood cells (RBCs). After six weeks and six months, tissues of the cerebellum and cortex were examined histologically and by immunohistochemistry. Hemoglobin cleared rapidly from the CSF in this model, as none of the samples showed xanthochromia after an RBC injection during the previous week. It is generally held that most RBCs in the subarachnoid space are destroyed by cellular attack. Heme is then metabolized to iron-free derivatives mainly by the microsomal enzyme heme oxygenase (4). In tissue, however, significant changes were found. The earliest change observed is an acceleration of ferritin biosynthesis. Immunohistochemistry for the presence of ferritin displays reactive microglia in the cerebellar molecular layer. Both heme and iron in the CSF stimulate ferritin synthesis. The processes of Bergmann glia of the cerebellum come into direct contact with the fluids of the subarachnoid space. These processes take up iron, which stimulates ferritin biosynthesis and enhances additional iron uptake by the Bergmann glia. Iron storage is thought to be first in the form of iron-ferritin and later hemosiderin (2). Bound or stored iron may later be released under conditions of oxidative stress, catalyzing further free radical production and lipid membrane injury.
In experimental models, total iron and ferritin levels in the cerebellar cortex do not rise significantly during observations from six weeks to six months (3). What is noted, however, is a dramatic shift in the relative contributions by heavy (H) and light (L) ferritin subunits. The normal H/L subunit ratio of near 1.0 rises to over 4.0 during the initial period but, at six months, falls to less than 0.5. Meanwhile, hemosiderin deposition also does not occur until after six months of repeated RBC injections, coinciding with the shift from H-ferritin to L-ferritin (3). It is felt that L-ferritin promotes cerebral iron storage, as it does in the liver, resulting in hemosiderin formation. In fact, hemosiderin may perhaps only occur with the formation of larger amounts of L-ferritin and is thus not exclusively related to the severity of the iron excess (3).
The selective vulnerability of the cerebellum may be influenced by the abundance of microglia and presence of Bergmann glia in the cerebellar molecular layer. Microglia synthesize ferritin (and most hemosiderin formation occurs in microglia), while Bergmann glia are a source of ferritin repressor protein (whose dissociation by heme or iron leads to increased ferritin production) (4). Anatomically, the restriction of damage to vermis and paravermis likely reflects their close anatomical relationship to the roof of the 4th ventricle and compartmentalization of CSF flow within the meninges, consistently increasing exposure of these cerebellar surfaces to materials circulating in the CSF (5).
Similarly, the vulnerability of cranial nerve VIII likely results from its course through the pontine cistern, which not only contains a large pool of CSF but also has a greater flow of CSF (1), potentially delivering a larger amount of iron and heme. More importantly, the VIIIth nerve has a long glial segment (the portion of the nerve surrounded by CNS glia) exposed to CSF, resulting in a greater length of hemosiderin deposition and, thus, a greater chance of axonal damage (1). The transition from CNS to PNS in this nerve is located near the internal acoustic meatus rather than near the brainstem as for most other cranial nerves. The only other cranial nerves with longer glial segments are I and II, which are entirely glial. Not surprisingly, the olfactory tract and bulb are frequently involved on pathological examination, though their clinical involvement is likely underestimated. Lack of involvement of the optic nerve, however, is not clearly explained by the above hypotheses, and therefore additional factors must clearly be involved in determining vulnerability.
Chronic repeated bleeding into the CSF is essential for the development of superficial siderosis, although a source of bleeding is only found in roughly half of cases (1). As noted above, hemosiderin begins to accumulate only after extended injections of RBCs in experimental models. Due to the tightness of the blood brain barrier for iron, systemic hemochromatosis does not regularly lead to iron excess in the CNS. In the review of Fearnley et al., a potential source of bleeding was identified in 54% of cases. Of these, the cause was dural pathology in 47%, either a CSF cavity lesion (such as hemispherectomy or meningocele) or cervical root pathology (such as root avulsion). Another 35% had a tumor, often ependymoma, allowing bleeding in the subarachnoid space, while 18% had a vascular abnormality.
Routine laboratory studies are typically unremarkable. CSF examination reveals hemorrhage and xanthochromia in roughly half of cases (1). The lack of CSF xanthochromia in many cases is not surprising, considering the amount of time that may have elapsed since the initial bleeding episodes, and experimental evidence for rapid resolution of xanthochromia after blood is introduced into the subarachnoid space (see above).
CT head scanning can occasionally identify atrophy of the cerebellar vermis or a hyperdense rim over the brain surface, corresponding to hemosiderin deposition. The best method of diagnosis is MR imaging. Gradient-echo T2-weighted studies are extremely sensitive to the presence of hemosiderin, identified as a rim of hypointensity around the brainstem and cerebellum (greatest at the vermis) in all cases. Similar changes are also often seen at the VIIth nerve and spinal cord periphery. These findings are considered pathognomonic for superficial siderosis. T1-weighted images, both pre- and post-contrast, are normal, aside from occasional vermal atrophy and the underlying bleeding source, if any. Of note, routine spin-echo T2-weighting is less sensitive to hemosiderin than gradient-echo imaging, which often is not used unless hemorrhage is suspected (6). Furthermore, fast spin-echo, which is rapidly replacing conventional spin-echo sequences, is even less sensitive to hemosiderin (7). Thus, gradient-echo sequences should be requested for every case of suspected superficial siderosis, to obtain maximal sensitivity.
In searching for a bleeding source, MRI of the spine should be included to exclude tumors, arteriovenous malformations or root avulsions. MR angiography of high resolution should be a sufficient screen for intracranial aneurysms instead of dye angiography, unless clinical suspicion for aneurysm is elevated.
In all cases, a source of subarachnoid bleeding should be sought, and surgical removal or ablation may be considered if a potentially treatable cause of bleeding can be identified. However, surgery has only been described in a few patients and its long-term clinical effects are not known. Unfortunately, it remains unclear whether the pathological process continues despite cessation of bleeding. Pharmacological therapy with iron chelators in a few cases has not shown convincing efficacy (1). This may be related to the short duration of therapy studied and that most chelators do not cross the blood brain barrier. Trientine, a copper chelator used in Wilson's disease, has been found to lower serum iron and does cross the blood brain barrier, but results in a few cases have been mixed. Antioxidants may be theoretically beneficial but, again, most do not easily cross the blood brain barrier or accumulate in the tissues of interest, and have not been studied clinically in cases of SS of the CNS.
Several clinical and diagnostic features of our case are typical of superficial siderosis of the CNS. The cavity left by the patient's posterior fossa surgery, thirty years prior to the onset of her present illness, is a potential source of protracted or recurrent low-grade subarachnoid hemorrhage. In addition, her previous neck trauma may also have contributed to her condition. A search for occult tumor was unrevealing, and no other vascular anomalies could be identified.
The long latency between surgery and initial symptoms is not unusual. A retrospective study of 21 patients with resected cerebellar tumors found that four had developed SS, with symptoms appearing 8-22 years after surgery (8). In a reported case similar to this, a woman had progressive visual impairment and headache at age 18. Surgical exploration of the posterior fossa revealed no tumor, but the patient received radiotherapy. Her symptoms resolved following these interventions, but 36 years later, she developed gait ataxia and later hearing loss. MRI showed the prior craniotomy and a pseudo-meningocele, along with SS and increased red blood cell count and protein levels in the CSF (9). The reason for this prolonged latency from a presumed initial insult to symptom onset is unknown. Whether this is due to a prolonged delay from surgery to the time of protracted bleeding, from bleeding to pathological damage, from pathological damage to the time of actual clinical manifestations, or from delay during each of these events, is uncertain.