Meningeal biopsy in intracranial hypotension

Intracranial hypotension is a cause of diffuse pachymeningeal enhancement on gadolinium-enhanced MRI ^[1-6] that frequently is associated with subdural collections of fluid. Downward displacement of the cerebrum and cerebellar tonsils, flattening of the pons, and effacement of the optic chiasm also may occur. ^[3-5,7] Examination of the CSF often shows variable increases in protein concentration and lymphocyte count as well as a low pressure. The pathogenetic mechanism of gadolinium enhancement of the meninges, formation of subdural collections of fluid, and changes in the CSF is not known. Hydrostatic pressure changes are presumed to cause the subdural collections of fluid and, through compensatory vasodilatation, to be the mechanism for meningeal enhancement. ^[3-5] Meningeal vasodilatation and leaky vessels may cause the increase in CSF protein and cell count. ^[5]

We report the results of meningeal biopsy in six patients with intracranial hypotension and meningeal gadolinium enhancement on MRI.

Case reports.

Patient 2.

A 38-year-old man with insulin-dependent diabetes presented in August 1992 with a 1-month history of occipital headaches and neck pain. Nonprescription analgesic and nonsteroidal agents had not been of help. The patient was afebrile and had no history of trauma. The results of general medical and neurologic examinations were normal, as were CT of the head and chest radiographs. MRI showed diffuse, thick, pachymeningeal enhancement with gadolinium and bilateral, thin, subdural collections of fluid. CSF examination showed a low pressure, and the fluid had to be removed by syringe. It contained 11 leukocytes/mm³ (85% lymphocytes), 126 mg/dl glucose (with simultaneous blood glucose of 230 mg/dl), and 70 mg/dl protein. There were no malignant cells, and cultures were negative. Subsequent lumbar punctures continued to show low CSF pressure, pleocytosis as high as 88/mm³, and increased protein as high as 109 mg/dl. Empirical treatment with antituberculosis medications did not produce improvement. In early September 1992, meningeal biopsy was performed through a right frontal craniotomy. The histologic slides were reviewed at the Mayo Clinic in consultation. After biopsy, the patient received treatment with doxycycline for 3 weeks because of a low-positive serum Lyme antibody titer; this treatment was without benefit. Extensive testing failed to show evidence of systemic, neoplastic, or infectious disease. A tapering course of prednisone was attempted in September 1992. The patient's condition gradually and spontaneously improved, and by late November 1992, he was asymptomatic.

Patient 3.

In late June 1991, a 35-year-old previously healthy man developed chest and epigastric pain that radiated to his dorsal area and shoulders and settled in the posterior neck. Within several hours, a pounding headache developed when he sat up or stood; it was relieved when he lay down. On examination, he was afebrile. The results of neurologic examination were normal. Head CT was normal. The initial lumbar puncture was unsuccessful, because the pressure was so low that fluid could not be collected. Neck pain resolved within a few days, but pounding orthostatic headaches persisted and were sometimes associated with diplopia on right lateral gaze. MRI of the head and spine 1 week after the onset of symptoms showed a very thin, hypodense subdural collection of fluid along both frontal convexities and a uniform and thick enhancement of the pachymeninx over the convexities, falx cerebri, and tentorium cerebelli, and, to a lesser extent, the cervical spine. The suprasellar cistern was effaced and the optic chiasm was displaced downward. Four CSF examinations were performed over the next 2 weeks, and all showed barely measurable or very low pressures, normal glucose levels and culture results, increased protein levels (90, 690, 283, and 196 mg/dl), and an essentially mononuclear pleocytosis (5, 30, 33, and 7 cells/mm³). Results of iopamidol myelography-CT were unremarkable. No clues about the site of CSF leak were discovered. An open cranial meningeal biopsy was performed. Later, the patient was seen at the Mayo Clinic, and symptomatic management was recommended. His condition gradually improved. Six weeks after biopsy, MRI showed significant improvement in the previously noted abnormalities. In 3 months' time, the patient was entirely asymptomatic.

Patient 4.

In September 1990, a 45-year-old woman had generalized throbbing headaches associated with nausea and vomiting when she was upright; her condition improved when she was recumbent. There was no history of trauma. The results of general physical and neurologic examinations were normal. MRI showed small bilateral subdural collections of fluid, with a signal intensity slightly higher than that of CSF, and thick and diffuse enhancement of the pachymeninx over the convexities, falx cerebri, and tentorium cerebelli. The iter (the opening of the aqueduct of Sylvius) was displaced downward on midsagittal MRI. CSF examination showed low pressure, less than 40 mm CSF. It was reported as "blood tinged." CSF glucose was 54 mg/dl and protein was 99 mg/dl. On subsequent examination, the CSF was clear, with a total cell count of 2 cells/mm (³) and 1 erythrocyte/mm³. There were no blasts or malignant cells. Cultures were negative and the VDRL was nonreactive. Lyme disease serology was normal, as were results of tests for human immunodeficiency virus and angiotensin-converting enzyme. In November 1990, a meningeal biopsy was performed through a right temporal craniotomy. Treatment was symptomatic. The headaches and associated nausea and vomiting gradually resolved within 6 months.

Patient 5.

In May 1989, the diagnosis of normal-pressure hydrocephalus was made in a 66-year-old man who presented with gait unsteadiness, urinary frequency and incontinence, and memory disturbance of several months' duration. Neurologic examination showed gait apraxia, and CT showed enlargement of the ventricular system. After a series of lumbar punctures had been performed, his gait reportedly improved. In July 1989, a medium-pressure ventriculoperitoneal (VP) shunt was installed. There was symptomatic improvement but for only 3 weeks. In October 1989, the VP shunt was revised with a low-pressure valve. The patient then began to complain of headache, nausea, and vomiting when in the upright position; these were relieved by his lying down. A few weeks later, the shunt was revised again to a medium-pressure valve. In November 1991, because of poor balance, left hemifacial spasm of a few months' duration, and a "heavy head" sensation when he was in an upright position, MRI was performed. It showed, in addition to enlargement of the ventricular system, diffuse, thick pachymeningeal enhancement with gadolinium. CSF glucose, cell count, and cytology were normal; cultures were negative. However, CSF protein was 317 mg/dl, and the opening pressure was recorded as low.

In June 1992, the patient was evaluated at the Mayo Clinic. He had wide-based gait and left hemifacial spasm. MRI showed diffuse pachymeningeal enhancement Figure 1 A. CSF was grossly bloody, with a barely measurable pressure, and 56 mg/dl glucose, 46/mm³ total leukocyte count, and 516 mg/dl protein. Cultures were negative. No malignant cells or blast cells were present. The results of extensive laboratory testing to identify any systemic, infectious, or neoplastic disease were normal. The medium-pressure VP shunt was revised to a performance-2 Delta shunt. The orthostatic symptoms improved and subsequently resolved. In September 1992, the patient underwent microvascular decompression of the left facial nerve for treatment of his hemifacial spasm. An intraoperative meningeal biopsy was performed. The results of the preoperative CSF examination were normal except for a protein concentration of 89 mg/dl. MRI of the head showed resolution of the abnormal meningeal enhancement Figure 1 B.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. (A) Enhanced coronal T_1-weighted MRI showing symmetric thickening and enhancement of the pachymeninx (TR, 600 msec; TE, 20 msec). (B) This follow-up enhanced image was obtained 2.5 months later when the patient's condition was clinically improved. The abnormal pachymeningeal enhancement seen in A has resolved (TR, 450 msec; TE, 11 msec).

Results.

Lumbar puncture and CSF analysis.

Initially, CSF pressure was very low in all patients: less than 40 mm CSF in one patient and not measurable or less than 10 mm CSF in the other five. One patient had to sit up for CSF to be collected; in another patient, the fluid had to be removed with a syringe, and in yet another patient, who had "dry" lumbar punctures, cisternal puncture had to be performed. The fluid was xanthochromic in one patient and contained 96 erythrocytes/mm³. It was bloody in one patient and clear in the other five patients. In one patient, it contained 1 erythrocyte/mm³ and, in another, 36 erythrocytes/mm³. In two patients with clear CSF, the initial erythrocyte count was not recorded. Pleocytosis, mostly mononuclear, was noted in four patients. All patients had more than one CSF examination, and most had several. Maximal pleocytosis in these patients (patients 2, 3, 5, and 6) was 88, 33, 46, and 6 cells/mm³. All patients had increased CSF protein, with the maximal recorded concentration of 1,000, 109, 690, 99, and 516 mg/dl in patients 1 through 5. In patient 6, CSF obtained through cisternal puncture was colorless and had a total protein concentration of 71 mg/dl. CSF and intra-arachnoid fluid obtained from the frontal region at the time of biopsy was yellowish and had a protein concentration of 4,600 mg/dl. It was presumed that the fluid obtained at the time of biopsy was likely loculated. None of the patients had decreased CSF glucose, abnormal cells, blast cells, or malignant cells. The results of bacteriologic testing of CSF were normal in all patients.

MRI findings.

The meningeal enhancement with gadolinium in all patients was diffuse, thick, and continuous. The pachymeninx was involved in all cases, but there was no evidence in any case of leptomeningeal enhancement, such as extension of enhancement to cortical sulci or enhancement of leptomeninges around the brain stem. Subdural collections of fluid--hygromas--were noted in all patients. No subdural hematoma was seen. The subdural collections of fluid were bilateral and primarily, but not exclusively, involved the frontoparietal areas. In one case, these collections of fluid were also noted at the level of the cervical spine. The collections of fluid were thin (4 to 6 mm at their maximal thickness), and there was no evidence that they had a mass effect on the brain. Evidence for downward displacement of the brain was noted in two patients.

Pathologic findings.

Information about the gross appearance of the dura mater at the time of biopsy was available for all patients. Grossly, the pachymeninx and leptomeninges appeared unremarkable in all the patients except for the one (no. 6) who had prolonged symptoms. In this patient, the arachnoid was thickened and appeared opaque. On histologic examination, the dura mater was not thickened, and its epidural surface was completely unremarkable in all patients. On the subdural surface, there was a variably thin zone consisting of fibroblasts and small thin-walled blood vessels in an amorphous matrix, but no evidence of inflammation or hemorrhage was noted Figure 2. The absence of hemorrhage, hemosiderin, blood pigments, and inflammatory cells distinguishes this process from the organizing neomembrane of chronic subdural hematoma. This area corresponds to the layer of the dural border cells described by Nabeshima et al ^[8] and Haines et al ^[9] and the zone for potential development of subdural collections of fluid. In all cases of relatively short duration (nos. 1 to 5) the leptomeninges demonstrated only minimally increased collagen, without evidence of inflammation or increased vascularity. In the patient (no. 6) with prolonged symptoms, the leptomeninges showed an increase in connective tissue elements and multiple nodules of arachnoidal cap cells unassociated with evidence of inflammation Figure 3. These foci of arachnoidal cap cell hyperplasia likely represent a nonspecific reactive change to a longstanding local "irritative" process.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. (patient 2). Microscopic appearance of meningeal biopsy specimen in a case of intracranial hypotension. (A) Dura mater (D) has a normal structure. The inner dural surface has an organizing zone of fibroblasts and thin-walled vascular spaces without inflammation (*). (B and C) Higher magnification of zone of fibroblasts and thin-walled vessels in an amorphous matrix in the inner surface of the dura (*). (Hematoxylin-eosin; A, times 100 before 27% reduction, and B and C, times 330 before 27% reduction.)

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. (patient 6). Meningeal biopsy samples. (A) Normal dura mater (D). Part of the outer dura has peeled away (curved arrow), exposing the deep layer (straight arrow). (B) Enlargement of boxed area in A showing leptomeningeal or subdural reaction, with hyperplasia of arachnoid cap cells (arrows) and increased connective tissue. There is no evidence of inflammation or of any abnormal or malignant cells. (Hematoxylin-eosin; A, times 64 before 39% reduction, and B, times 200 before 39% reduction.)

Discussion.

Normal meninges, perhaps with the exception of the falx cerebri, are usually invisible on noncontrast MRI studies. On T_1-weighted images, even without contrast medium, small segments of meninges are sometimes visible as thin lines, with CSF on one side and the signal-void of the inner Table ofcranial bone on the other side. On T_2-weighted images, however, it is difficult to visualize even small segments of meninges because of the high signal from the CSF. ^[10] With gadolinium-enhanced T_1-weighted images, the meninges may be visualized as small linear areas of enhancement along the falx cerebri and tentorium cerebelli and over the convexity of the brain. In normal cases, they are never extensive or thick. With gadolinium, the sensitivity of MRI for detecting meningeal abnormalities has increased markedly. Various meningeal diseases show increased meningeal enhancement. Bacterial or viral meningitis, when severe or advanced, may show gadolinium enhancement of the meninges, not only of the basal cisterns but of the entire meninges of the base of the skull and depth of the cortical sulci. ^[10-13] With treatment of meningitis, the meningeal enhancement may resolve, but it may persist if there is residual meningeal thickening and fibrosis. Gadolinium enhancement of the meninges may occur with subdural collections of fluid and after shunting procedures for hydrocephalus, ^[14] after subarachnoid hemorrhage, after intrathecal chemotherapy, ^[10,12,15] after craniotomy or craniectomy, ^[16,17] and in meningeal carcinomatosis or neurosarcoidosis. ^[18] We have seen persistent diffuse meningeal enhancement in biopsy-proved arachnoid thickening and fibrosis of undetermined cause (Mokri B, unpublished data). Although gadolinium-enhanced MRI is a sensitive technique for demonstrating meningeal disease or irritation, the MRI finding per se is nonspecific. Meningeal gadolinium enhancement in intracranial hypotension was initially described in 1991 ^[19]; subsequent reports followed. ^[1-6]

The syndrome of intracranial hypotension is characterized by headaches that occur with upright posture and that are sometimes associated with nausea and vomiting and less frequently with diplopia, dizziness, tinnitus, photophobia, or changes in hearing. Symptoms are relieved by the patient's lying down. ^[20-23] This syndrome may occur spontaneously, ^[2,20,23,24] after lumbar puncture, ^[20] in connection with overdraining ventricular or spinal shunts, ^[4,25] after head trauma (with or without obvious CSF leak), and from a tear of a spinal nerve root. ^[26] Although some authors have speculated that CSF hyperabsorption causes spontaneous low CSF pressure, ^[27,28] no data support such a mechanism. ^[29]

Our MRI findings are similar to those of Fishman and Dillon ^[3] and Pannullo et al. ^[5] Pachymeningeal gadolinium enhancement, as noted in all our patients, was diffuse and continuous. None of the patients had evidence of leptomeningeal enhancement, extension of enhancement to the cortical sulci, or enhancement of the leptomeninges around the brain stem.

Subdural fluid collections were also present in all the patients. None of these fluid collections was a hematoma; they were hygromas, with some variation in their signal intensity that likely was related to the concentration of protein in the fluid. The collections of fluid were thin, without a noticeable mass effect on the brain, and generally located underneath the enhanced convexity of the pachymeninges; sometimes small segments of these collections were noted between enhancing membranes.

Evidence of downward displacement of the brain was present in two patients. Because MRI is performed with patients in the supine position, such displacements may have been more evident if the head had been imaged with patients in an upright position.

With CT studies, in the pre-MRI era, meningeal enhancement was not present. However, several other imaging features that may be associated with intracranial hypotension were described, including subdural effusions ^[30,31] and downward displacement of the brain (manifested by narrowing of the sylvian fissure and effacement of infratentorial cisterns). ^[32]

Ultrastructural studies of the meninges have shown that the dura mater consists of elongated fibroblasts oriented along its flat axis and extensive extracellular collagen, which accounts for the toughness of this membrane. ^[8,33-36] As reviewed by Haines et al, ^[9] the outer, or periosteal, dura mater is attached to the inner surface of the skull. The inner, or meningeal part has proportionately less collagen but a larger number of fibroblasts. ^[33,34] Still deeper in the dura mater and overlying the arachnoid, the fibroblasts become more elongated and form a layer of "dural border cells." This layer lacks extracellular collagen, but irregularly interdigitating processes of the fibroblasts create extracellular spaces of various configurations and sizes. ^[8,34,36-39] Some cell junctions, usually desmosomes or intermediate junctions ^[8,39] but not tight junctions, occur between the cells. Therefore, with few cell junctions, the absence of tight junctions, the absence of extracellular collagen, and with enlarged extracellular spaces, the layer of dural border cells is a relatively weak zone structurally at the dura-arachnoid interface. Subdural collections of fluid form in this layer. ^[9]

Immediately beneath the layer of dural border cells is a layer of arachnoid barrier cells, characterized by a lack of collagen but the presence of numerous tight junctions. ^[34,36-41] This layer serves as a barrier against the movement of fluids, of substances with large molecular weight, and even of ions. ^[8,42] Deep to this layer is the arachnoid, with fibroblasts that have long, irregular processes attached by cell junctions and reinforced by collagen. The pia mater and all arachnoid layers possess cell junctions and are part of the blood-brain barrier, whereas the dura mater lacks these cell junctions and is not part of the blood-brain barrier. As stated by Fishman and Dillon, ^[3] dural vasodilatation due to decreased intracranial pressure may be responsible for pachymeningeal gadolinium enhancement through greater concentration of gadolinium in the dural microvasculature and in the interstitial fluid of the dura. This hypothesis is supported by the histologic observations on the biopsy material that demonstrated a normal appearance of outer dura mater, arachnoid, and pia mater. The location of the loose zone of fibroblasts and small blood vessels noted in the biopsy samples corresponds to the layer of dural border cells. This is the zone in which subdural collections of fluid occur. ^[9] These collections should represent a phenomenon secondary to hydrostatic pressure changes. Inflammatory reaction was conspicuously absent in all biopsy samples. The CSF pleocytosis (although at times fairly high) and the increased CSF protein are likely secondary phenomena, perhaps related to diapedesis of cells and leakage of proteins due to pachymeningeal vasodilatation and increased permeability rather than to a meningeal inflammatory process. According to the Monro-Kellie hypothesis, ^[43,44] with an intact skull and a constant intracranial volume, a decrease in CSF volume causes a compensatory vasodilatation of the brain and meninges and subdural collections of fluid.

Decreased CSF flow in the lumbar subarachnoid space due to a CSF leak at a higher level may also have contributed to increased CSF protein in some of the cases.

Arachnoid cell proliferation was present in patient 6, who, in contrast to the other five patients, had very longstanding symptoms. We suspect that this proliferative reaction was triggered by chronic venous engorgement and prolonged persistence of subdural zone changes related to intracranial hypotension. Fishman and Dillon ^[3] proposed the same mechanism to explain the meningeal fibrosis in patients with meningeal fibrosis related to overdraining ventricular shunts.

Good and Ghobrial ^[6] reported extensive fibrocollagenous proliferation in the leptomeninges, without evidence of inflammation in the meningeal biopsy sample, of a 40-year-old man with a 6-week history of postural headache due to intracranial hypotension. MRI showed diffuse meningeal enhancement and small subdural collections of fluid. Fishman ^[45] questioned the concept that fibrocollagenous changes in the meninges explained meningeal enhancement in intracranial hypotension. Although we think that meningeal fibrosis can occur as a late consequence of intracranial hypotension, such a change is unlikely to occur in 6 weeks.

Our histopathologic observations support the hypothesis that dural-meningeal MRI abnormalities in intracranial hypotension represent a reactive secondary phenomenon rather than a primary meningeal process and, as Fishman and Dillon ^[3] suggested, are likely related to hydrostatic CSF changes.