Spinal cord hemorrhage in a patient with neurosarcoidosis on long-term corticosteroid therapy: case report

In the case described here, the diagnosis of neurosarcoidosis was definite according to Judson’s criteria [16] and probable according to Zajicek’s [17]. This first report highlights the possibility of spinal cord hemorrhage in patients with neurosarcoidosis, as suggested by Waubant and colleagues, who described hemosiderin deposits in spinal-cord sarcoidosis [18].

Common causes of atraumatic hematomyelia [19, 20] include vascular malformations (arteriovenous fistula, cavernoma, capillary telangiectasia, venous angioma) [21], neoplasms, Gowers’intrasyringal hemorrhage, spinal radiation, anticoagulation and bleeding disorders. None of these applied to our patient, suggesting that her hematomyelia was caused by bleeding of a spinal cord granuloma, given the proximity of the first inflammatory lesion (T8-L1) to the hemorrhage (T10-T11).

Despite the lack of histological confirmation of the source of bleeding, there is sufficient evidence from cerebral cases [9] to suggest that hemorrhage in neurosarcoidosis may be caused by vascular pathologies (venous, arteriolar, or micro-arterio-venous malformations) induced by the disease itself. Sarcoid granulomas in the central nervous system (CNS) follow perivascular spaces and penetrate brain parenchyma [22]. Post mortem examinations of intracranial hemorrhage in neurosarcoidosis have revealed perivascular inflammatory infiltrate and vessel wall destruction by granulomas [7, 9]. The enlarged blood vessels noticed by the neurosurgeon in our patient seem to support the hypothesis of neurosarcoid vasculitis. The frontoparietal hemorrhage that occurred during the second episode, 3 years before the hematomyelia (Fig. 2a-b), is also suggestive of the fragility of blood vessels in the CNS damaged by the neurosarcoid vasculitis.

Long-term corticosteroids also increase vascular fragility, owing to reduced collagen formation in vessel walls [23], but there was no evidence to suggest that steroids directly caused the bleeding here. Indeed, we decided to increase the corticosteroid dose, as adding an immuno-modulating agent to control the neurosarcoid vasculitis (and spare corticosteroids) was not possible because of an infectious complication. There have not been any randomized controlled trials of pharmacological treatments for neurosarcoidosis, but the algorithm developed by Nozaki and Judson, has proved useful for severe neurosarcoidosis [24].

Kreppel and colleagues’ meta-analysis [25] is the largest review of reports of spinal hematoma published before 1996. These authors reviewed 613 patients belonging to four etiological groups of spinal cord hemorrhage: intramedullary (hematomyelia), subarachnoidal, subdural and epidural. Hematomyelia was found in only 0.82 % of patients, possibly owing to the dearth of MRI-based diagnosis prior to 1996. More recently, Leep and colleagues [19] and Heldner and colleagues [20] reviewed the diagnosis, cause and treatment of hematomyelia. Hematomyelia can present as an acute, subacute, step-wise or chronic myelopathy [19]. As it did in our case, neurological deterioration can occur after the initial hemorrhage, owing to a secondary tissue response [19]. It is essential to monitor vital signs and neurological status in the intensive care unit, to prevent complications [20]. However, there have not been any clinical trials comparing different ways of managing hematomyelia [19, 20], notably the timing of hematoma evacuation. In our case, the decision to operate was made because of a progressive, rapid and inexorable neurological deterioration. There was a definite, but not total, clinical improvement following treatment for this rare and severe complication of neurosarcoidosis.