MR Imaging in Pretruncal Nonaneurysmal Subarachnoid Hemorrhage

Many patients with a spontaneous subarachnoid hemorrhage of unknown origin have CT scan findings that are compatible with a so-called pretruncal (previously called perimesencephalic) hemorrhage. This variant of subarachnoid hemorrhage was first described by van Gijn and coworkers¹ and initially labeled “perimesencephalic nonaneurysmal subarachnoid hemorrhage.” We recently coined the more anatomically correct term “pretruncal nonaneurysmal subarachnoid hemorrhage,” because in most published series of patients cisternal blood was located predominantly in front of the entire brain stem (truncus cerebri).²

After the initial cerebral angiogram is performed, it is common practice to exclude an arteriovenous malformation or venous angioma in the brain stem or cervical spine by MR imaging. No large series of MR imaging in this subset of subarachnoid hemorrhage have been reported since a preliminary study with 4 normal MRIs involving only axial images.³ We report a review of MR imaging in pretruncal subarachnoid hemorrhage. Our study indicates that MR imaging probably is not a cost-effective diagnostic test in the evaluation of this benign clinical entity.

Subjects and Methods

Between 1992 and 1998, 24 patients were admitted to Mayo Medical Center with a pretruncal nonaneurysmal subarachnoid hemorrhage. The diagnosis was suggested on the basis of hemorrhage restricted to the cisterns in front of the brain stem and suprasellar cistern on third generation, high-resolution CT scan (General Electric) and at least 1 normal 4-vessel cerebral angiogram. MR imaging was performed in 14 patients within the same admission period, with an interval of 2 to 7 days (median, 4 days) after the ictus. In the remaining 4 patients, MRI was performed at least 3 months after the initial presentation. Routine sequences included 5-mm T1-weighted sagittal and T2-weighted images. Gadolinium was administered in 14 of 18 patients with 4-mm T1-weighted axial or coronal images. In 7 consecutive patients we additionally performed thin (1.5- to 3-mm) T1-weighted axial sections through the brain stem following a (single or triple) dose of gadolinium contrast. In 1 patient an additional 2-dimensional MRA/MRV was performed, and in 1 patient a phase contrast MRA with velocity encoding of 5 cm/s and 10 cm/s. Three-dimensional time of flight MRA, which included the circle of Willis, was performed in 11 patients. Five patients underwent an MRI of the cervical spine. Gadolinium was administered in 4 of these patients.

Results

No vascular abnormalities were found in 17 of 18 patients with MR imaging of the brain. In 1 patient, a vascular abnormality was found after gadolinium enhancement and interpreted as a capillary telangiectasia. Details have been reported in a separate publication.⁴ Routine MR imaging showed increased T1 signal intensity only in front of the brain stem in all 14 MR scans performed within 1 week of the ictus (Figure 1). The distribution of blood was prepontine and interpeduncular cisterns in 7 patients, additional extension into the premedullar cistern in 3, prepontine cistern only in 3, and interpeduncular cistern only in 1 patient. Additional findings on MRI were hyperintense lesions in the cerebellum and thalamus in 1 patient. Additional fluid-attenuation inverse recovery (FLAIR) images were available in 6 patients and showed cerebrospinal fluid signal abnormality along the sulci in 2 patients (Figure 2). Fifteen patients (63%) underwent an additional 4-vessel cerebral angiograms but were all normal (except for junctional dilatation at the top of the basilar artery and an infundibular widening of the posterior communicating artery in 1 patient each). MRI of the cervical spine was normal in all 5 patients.

Discussion

This subset of subarachnoid hemorrhage is defined by acute or gradual onset of severe headache, normal level of consciousness, absent localizing neurological signs, a limited amount of blood in front of the brain stem, and a normal 4-vessel cerebral angiogram. Serious concern remains about an undiagnosed vascular malformation in patients with an angiogram-negative subarachnoid hemorrhage.⁵ A second cerebral angiogram is usually performed only after review of the first study when there is uncertainty about its quality. We have recently emphasized the possibility of missing a cerebral aneurysm in pretruncal subarachnoid hemorrhage.⁶ This concern is further fostered by a recent report⁷ showing that cervical medullary arterial venous fistulas are a frequently unrecognized cause of subarachnoid hemorrhage. Therefore, MRI of the brain and cervical spine is performed in many academic institutions to exclude any other vascular abnormalities.

Our series is the first large study of MR imaging in pretruncal nonaneurysmal subarachnoid hemorrhage. Axial and sagittal views demonstrated the extension of the hemorrhage in the cisterns exclusively in front of the brain stem but failed to show a vascular lesion in 17 of 18 patients. Our previous finding of a capillary telangiectasia on MRI in 1 case suggested a venous etiology of this disorder.⁴ However, we were unable to confirm this finding, which strongly suggests that this abnormality has been incidental. FLAIR images additionally showed recirculation of subarachnoid blood over the sulci. To our knowledge, this is a new observation in this subset of subarachnoid hemorrhage, but it should not be mistaken for a more diffuse pattern of subarachnoid hemorrhage indicating a possible aneurysmal source. Small lesions were found in the cerebellum and thalamus, likely representing emboli associated with repeated vertebral angiograms.

Our study sample consisted of only 18 MR brain scans. Thus, the 95% confidence interval by the binomial distribution for 1 of 18 (5.6%) is relatively wide (0.1% to 27.3%). Nonetheless, we believe that routine MRI of the brain and spine unnecessarily adds to the cost of evaluation. Perhaps MRI of the spine should be performed only in patients with clinical leads of a ruptured spinal arteriovenous malformation. MRI of the brain may also be useful only to confirm subarachnoid hemorrhage in patients with ambiguous CT scans, such as blood predominantly in the prepontine cisterns, but in all our patients imaged with high-resolution CT scans, the subarachnoid hemorrhage was clearly identified. Nonetheless, over time, newer technologies of MRI will be developed; therefore, MRI as a research tool to investigate the cause of this puzzling but benign variant of subarachnoid hemorrhage may remain useful.

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

Footnotes