Clinical Presentation, Diagnosis, and Radiological Findings of Neoplastic Meningitis

Georgios Rigakos, MD; Chrysoula I. Liakou, MD; Naillid Felipe, Dennis Orkoulas-Razis, MHS; Evangelia Razis, MD, PhD


Cancer Control. 2017;24(1):9-21. 

In This Article


Typically, the diagnosis of neoplastic meningitis is based on findings from symptomatology, CSF cytology, and neuroimaging studies (eg, gadolinium-enhanced MRI of the brain and spine). Gadolinium-enhanced MRI is the first study obtained in a patient with a systemic malignancy and neurological symptoms. Frequently, neoplastic meningitis is discovered at the same time as parenchymal CNS disease (38%–83%).[52] In many such cases, radiological findings in the brain parenchyma are not explained by the neurological symptoms, thus leading to further investigation with CSF cytology.

Historically, imaging to investigate neoplastic meningitis involved computed tomography (CT) with a myelogram; however, since the advent of MRI — which is more sensitive than CT — CT is only used in patients who have a contraindication to MRI. MRI performed to investigate the CNS for neoplastic meningitis should be multiplanar, obtained before and after the administration of gadolinium 0.1 mmoL/kg in at least a 1.5 Tesla scanner (gadolinium-enhanced MRI), preferably with thin cuts in the brainstem, and should include the entire neuraxis. Revealing sequences include contrast-enhanced fluid-attenuation inversion recovery and contrast-enhanced T1-weighted sequences.[2]


Not all patients with neoplastic meningitis demonstrate positive findings. In 1 study, 66% of patients with neoplastic meningitis demonstrated findings on MRI.[14] Such findings include ependymal, leptomeningeal, and dural enhancement. Observations seen the cranium may include enhancement or thickening of the cranial nerves, small superficial metastases in the sulci, linear enhancement of the leptomeninges of the cerebellum or the basal cisterns, or ventricular dilation consistent with communicating (nonobstructive) hydrocephalus. In the vertebral column, intradural, nodular enhancement — particularly but not exclusively in the cauda equina — is the hallmark of neoplastic meningitis (Figure 1).[53] Lumbar or sacral nerve roots may appear thickened and associated intramedullary disease may be present. Irritation of the meninges can lead to meningeal enhancement, so lumbar puncture should be performed following MRI.[8] Other false-positive results can occur because of ischemia, infection, inflammation, hemorrhage, irritants, radiotherapy, chemotherapy, granulomas, trauma, or hypoxia (Figure 2). Previous therapy with bevacizumab can affect enhancement.[3] Atypical presentation in the form of symmetrical, curvilinear, band-like edema along the surface of the brain stem has also been described.[53]

Figure 1.

Positive findings of neoplastic meningitis seen on MRI of the spine of a woman aged 45 years with metastatic melanoma. She had been receiving ipilimumab therapy and had been in remission for 5 years when she presented with headache, diplopia, twitching, and numbness of the face. CSF was positive for neoplastic meningitis. The patient underwent surgery for Ommaya shunt placement. During the operation, the brain was covered with a thick yellow layer of metastatic melanoma cells. Linear, meningeal-enhanced thickening can be seen on T1-weighted, gadolinium-enhanced spinal MRI representing leptomeningeal disease (white arrow). The spinal cord appears to be infiltrated by the disease below that level. CSF cerebrospinal fluid, MRI = magnetic resonance imaging.

Figure 2.

False-positive findings on MRI of the spine in a woman aged 52 years with breast cancer and metastatic bone disease. MRI was obtained to evaluate her bone lesions. She had uptake in the meninges in the thoracic and lumbar spine. CSF was negative for neoplastic meningitis on serial lumbar punctures. She died from the disease 2 years after MRI was obtained. A diagnosis of neoplastic meningitis was never confirmed, nor was she treated for neoplastic meningitis. Linear, meningeal-enhanced thickening can be observed on T1-weighted, gadolinium-enhanced spinal MRI (white arrow). This finding is sometimes mistaken for neoplastic meningitis. An artifact (black dot) can also be seen. CSF = cerebrospinal fluid, MRI = magnetic resonance imaging.

Results from a study showed that findings on MRI differ according to clinical presentation.[12] Thus, study patients with no neurological symptoms or signs were unlikely to have radiological findings (14%), whereas study patients with cranial nerve (33%), spinal cord (53%), cerebral (66%), and multilevel abnormalities (83%) were more likely to have findings on MRI.[12]

MRI tends to be less reliable than cytology because the latter has a high rate of specificity but a low rate of sensitivity. Thus, gadolinium-enhanced MRI is commonly used to support the diagnosis, rather than be used as the sole indicator.[14] In 1 study, 31% of patients were diagnosed based on symptomatology and radiological findings when cytology findings were negative.[14] The study also suggests that MRI-proven leptomeningeal seeding is an indicator of poor prognosis and could be used to identify response to intrathecal chemotherapy.

A small study of 68 patients with suspected neoplastic meningitis compared the diagnostic ability of gadolinium-enhanced MRI and CSF cytology and found that the overall sensitivity rate of gadolinium-enhanced MRI in neoplastic meningeal disease was significantly lower than that of CSF cytology (45.5% vs 93.2%); however, in neoplastic meningitis from solid tumors, both methods had a sensitivity rate of 84.6%.[2] The authors suggest that MRI alone can be used to diagnose neoplastic meningitis in this setting. In leukemia and lymphoma, the rates significantly decreased to 20% and 37.6%, respectively.[2] Furthermore, the positive predictive value of MRI in differentiating between infectious and malignant meningitis was highest in lymphoma (83.3%), high in solid tumors (72.7%), but low in leukemia (33.3%).[2]

Other MRI techniques, such as perfusion, may add to the potential for radiographic diagnosis.[54] One group tested the value of magnetic resonance spectroscopy for diagnosing neoplastic meningitis, and they found that magnetic resonance spectroscopy fared better than CSF cytology from the first lumbar puncture in identifying neoplastic meningitis.[55,56]

Other Imaging Tests

CT myelography preceded MRI as the diagnostic imaging technique for neoplastic meningitis. Although CT myelography is not as sensitive as gadolinium-enhanced MRI, it can reveal nerve root thickening, cord enlargement, and CSF flow abnormalities better than MRI.[52] In general, it is performed with a water-based, nonionic dye injected via lumbar puncture; however, if a block is suspected, then the injection is administered using a lateral cervical route. These 2 techniques are comparable, and CT can be used as an alternative to MRI when the latter is contraindicated.[6]

If possible, CSF pathway blockage should be evaluated (present in 31% to 61% of neoplastic meningitis cases) by performing a radionuclide cisternogram.[6] Indium-diethylenetriamine penta-acetic acid or macroaggregated albumin may be used to assess the flow of CSF, which is abnormal in 30% to 40% of patients with neoplastic meningitis.[6] This nuclear medicine study is performed with a lumbar injection of the radionuclide followed by serial imaging of its flow to rule out a block to the flow of CSF. The results of such a test are important as they may have therapeutic implications because blockage remediation with radiotherapy will allow for a uniform distribution of intra-CSF chemotherapy.[57] Although its use is included in clinical practice guidelines, radionuclide cisternogram is not routinely used and other MRI techniques may obviate its need.[6]

A few isolated case reports have been published in which positron emission tomography (PET)/CT has been useful in the diagnosis of neoplastic meningitis.[58–61] Fluorodeoxyglucose, thymidine, or methionine has been used when MRI and cytology failed, but the results of these reports are not enough to establish a role for PET/CT in this setting.[58–61]

Response Assessment in Neuro-oncology

No method of quantifying neoplastic meningitis has been described, so no systematic method exists to evaluate a radiographic response to therapy. During the course of our research, we have found that many studies do not include repeat MRI after diagnosis. Chamberlain et al[52] are developing criteria for the assessment of therapeutic response in the setting of carcinomatous meningitis.


Neoplastic meningitis remains underdiagnosed.[62] To make the diagnosis, clinicians should consider the clinical presentation, findings on laboratory studies, and observations on imaging.

Lumbar Puncture

Cytology. The standard diagnostic test for neoplastic meningitis remains the cytological identification of malignant cells in CSF. MRI appears to be sensitive for detecting metastatic deposits along the neuraxis. However, metastases at a microscopic level are below the resolution available for MRI, which may explain why MRI is less sensitive in detecting neoplastic meningitis than CSF cytology.[62] Malignant cells go undetected in up to one-third of patients who have clinical or radiographic evidence of neoplastic meningitis.[62] Novel assays are being tested that may enhance the early identification of malignant cells in CSF.[63–65] Until then, the diagnosis is generally made after the onset of neurological manifestations, which have a rapidly fatal course for many patients.[5] When symptoms appear, most tumors have widely disseminated throughout the meninges.

Positive findings on CSF cytology require optimal sampling and processing. Malignant cells can be found in the initial lumbar puncture sample in 50% to 70% of patients and in nearly all cases after 3 attempts.[32] Performing lumbar puncture to obtain CSF can — and likely should — be repeated until findings are positive if the patient has no evidence of obstructive hydrocephalus. In the setting of communicating hydrocephalus, lumbar puncture can be performed and CSF diversion through lumbar puncture or ventriculostomy would be indicated. Contradictions to lumbar puncture include bleeding diathesis, skin infections at the puncture site, and vertebral or other skeletal deformities (scoliosis or kyphosis).[66] The procedure should only be performed by experienced clinicians.[66]

Gadolinium-enhanced MRI of the area of maximal symptomatology should precede lumbar puncture, because the latter may lead to false-positive results on MRIs. Cytology of the CSF obtained by lumbar puncture is more likely to be positive than CSF obtained using a ventricular catheter if spinal cord–related symptoms are present and vice versa if cranial-related symptoms are present.[67] Periodic lumbar puncture is recommended in the follow-up of patients with neoplastic meningitis, even in those with ventricular catheters, because cytology results have high rates of false-negative results when using samples taken from ventricular catheters.[32,67]

Essential elements of the CSF laboratory evaluation include cell count and differential, cytology, protein, and glucose concentrations (Table 2). In patients who have primary solid tumors, a finding of malignant cells in the CSF is evidence of leptomeningeal metastases. In approximately 50% of patients with leptomeningeal metastases, the CSF opening pressure will be elevated.[68] Similarly, most patients with leptomeningeal metastases have elevated protein levels and increased CSF cell counts.[68] A low CSF glucose level (hypoglycorrhachia) is seen in 30% of patients with leptomeningeal metastases.[68] However, these abnormal CSF findings alone are nonspecific and may be present in various disorders.

Carcinoma cells in the CSF are diagnostic, with the exception of certain false-positive results in patients who have reactive lymphocytes (which are difficult to distinguish from malignant lymphomatous cells) because of an infectious or inflammatory process in the CSF. However, negative findings on cytology do not rule out the diagnosis, because 30% of patients with leptomeningeal disease have a negative cytological result on the first sample obtained via lumbar puncture.[32] This percentage decreases to 15% after 2 high-volume lumbar punctures and then 10% after 3 lumbar punctures.[32]

Cytological findings are more likely to be positive in patients with extensive leptomeningeal involvement than in patients with focal involvement because CSF obtained from a site distant to the lesion is less likely to yield a positive finding on cytology.[32] Other causes of false-negative results can include an inadequate sample (< 10.5 mL CSF) and delayed processing of samples. CSF pleocytosis and modest protein elevations are consistent with but not indicative of the diagnosis, whereas reduced glucose levels usually are seen with neoplastic meningitis (ie, abnormal glucose transport) or infection (ie, increased glucose utilization).

The lymphocyte count is elevated in more than 50% of patients with neoplastic meningitis, and the presence of eosinophils in a patient with clinical evidence of neoplastic meningitis is suspicious for lymphomatous infiltration, although eosinophilia can also herald a number of other conditions that should be entertained in the differential diagnosis.[69] Xanthochromia can occur from leptomeningeal bleeding and is seen more frequently in neoplastic meningitis from melanoma.[45] Several series have demonstrated that, in some cases, serial CSF sampling via lumbar puncture or sampling from alternate sites (eg, cisternal, ventricular) is required to detect malignant cells.[11,67,70,71] LDH concentrations are elevated in cases of stroke, bacterial meningitis, CSF pleocytosis, head injury, primary CNS tumors, and some metastases.[44,72] Levels of LDH are also elevated in 80% of neoplastic meningitis; therefore, they can be useful in confirming the diagnosis.[72] LDH isoenzyme 5 levels are elevated in neoplastic meningitis from breast or lung primary tumors and melanoma, as well as bacterial meningitis, although they can be normal even when cytological findings are positive.[73] Ferritin levels are sensitive to inflammatory changes in the CSF, but they are nonspecific for early neoplastic meningitis, whereas CSF alkaline phosphatase levels may be elevated in neoplastic meningitis due to lung primary tumors.[70,74,75]

Tumor Markers and Metabolomics

Most tumor markers in CSF have poor rates of sensitivity and specificity;[76] however, if they are present, then their levels should decline with successful therapy.

Their re-elevation can cause disease relapse before any other findings become apparent. Useful markers include carcinoembryonic antigen in adenocarcinomas, α-fetoprotein and β-human chorionic gonadotropin in germ cell tumors, 5-hydroxyindoleacetic acid in carcinoid tumors, and immunoglobulins in multiple myeloma; their presence in CSF is diagnostic.[77,78] Levels of CSF β-2-microglobulin may be useful in detecting neoplastic meningitis caused by hematological spread but not in neoplastic meningitis from solid tumors.[79] Levels may also be elevated after treatment with intrathecal methotrexate.[80] Prostate-specific antigen may be elevated in neoplastic meningitis from a prostate primary tumor.[81,82]

Nonspecific markers can be strong, indirect indicators of neoplastic meningitis, but none are sensitive enough to improve on the cytological diagnosis.[5] Epithelial-associated glycoprotein is present in up to 90% of neoplastic meningitis cases.[83] Cytokeratins measured by tissue polypeptide antigen and tissue polypeptide-specific antigen have a sensitivity rate of 80% to neoplastic meningitis from breast cancer.[84] Neither carcinoembryonic antigen nor β-glucuronidase is helpful in detecting solid tumors or metastases, nor are these values useful in detecting leptomeningeal lymphomatosis. However, if their levels are elevated when neoplastic meningitis is diagnosed, then a return to normal levels of both markers signifies successful treatment.[85,86]

CSF β-glucuronidase values are frequently elevated, but wide fluctuations make it unreliable as a marker, and elevations can also occur with bacterial, viral, fungal, or tubercular meningitis. However, in association with an elevated LDH level, high CSF β-glucuronidase levels can indicate neoplastic meningitis from a breast primary tumor with high sensitivity and specificity rates. CSF fibronectin and myelin basic protein values can be elevated in neoplastic meningitis, bacterial meningitis, tick-borne encephalitis, multiple sclerosis, trauma, and a number of other conditions.[87,88]

Epidermal growth factor, vascular endothelial growth factor, and antithrombin 3 have been suggested as useful biomarkers, although antithrombin 3 has been evaluated in primary CNS lymphoma but not neoplastic meningitis.[89–91] Other markers such as creatinine kinase BB, tissue polypeptide antigen, and β-2 microglobulin are all indirect indicators of neoplastic meningitis and are still not sensitive enough to improve on findings seen on cytology.[91–93]

The metabolome of CSF could be of use in detecting neoplastic meningitis. Use of nuclear magnetic resonance spectroscopy based on variation seen in CSF metabolites is encouraging for the early detection of neoplastic meningitis in an animal model and now in humans.[55,56] The proteomics in CSF samples of children with acute lymphoblastic leukemia were used to predict CNS clot formation, and mass spectrometry of CSF in the setting of glioma was associated with glioma grade and prognosis.[94,95] However, the diagnostic path of proteomics in CSF must be further explored.

Flow Cytometry

Other useful adjuncts to CSF cytology include flow cytometry, measuring of immunophenotype, fluorescence in situ hybridization, chromosomal analysis, and immunohistochemical studies of tumor cells. The underlying diagnostic utility of such studies depends on the underlying systemic malignancy. For example, lymphocytes in the CSF may not be identifiable as malignant by the cytopathologist, but a demonstration of monoclonality (λ- or κ-light chain–directed monoclonal antibody analysis), B-cell lineage, or a specific chromosomal abnormality may differentiate leukemic or lymphomatous meningitis from a normal or reactive T-cell population;[96] however, additional research is necessary. Glial fibrillary acidic protein assessed by immunohistochemistry in CSF may facilitate the identification of malignant glial cells.[1,63–65,97]

Several studies have demonstrated that the sensitivity of flow cytometry is several-fold higher than that of cytology for detecting CSF leukemia or lymphoma.[98,99] Flow cytometry allows for the early detection of neoplastic meningitis before the onset of clinical symptoms and CSF pleocytosis; therefore, its use may enable more effective treatment. Patients with negative findings on cytology but positive findings on flow cytometry are often asymptomatic and have lower CSF cell counts and fewer neoplastic B cells when compared with patients whose cytology findings are positive.[100] Future consensus regarding standardized antibody panels for flow cytometry that uniformly define positivity is likely to advance the early detection of neoplastic meningitis and will help permit its broader clinical applicability.[1,63–65]

Another group of researchers studied patients with epithelial cell cancers to explore how flow cytometry immunophenotyping contributed to the diagnosis and prognosis of neoplastic meningitis.[101]

CSF samples from patients diagnosed with neoplastic meningitis were studied using flow cytometry immunophenotyping. Expression of EpCAM was used to identify the epithelial cells. The prognostic value of flow cytometry immunophenotyping was evaluated in 72 patients diagnosed with neoplastic meningitis and eligible for therapy.[101] Compared with cytology, flow cytometry immunophenotyping had greater sensitivity and negative predictive value (80% vs 50% and 69% vs 52%, respectively), but lower specificity and positive predictive value (84% vs 100% and 90% vs 100%, respectively).[101] The multivariate analysis revealed that the percentage of CSF EpCAM-positive cells predicted an increased risk of death.[101] A cut-off value of 8% EpCAM-positive cells in the CSF distinguished 2 groups of patients with statistically significant differences in overall survival (P = .018).[101] This cut-off value kept its statistical significance regardless of the absolute CSF cell count.

In another study, EpCAM-based flow cytometry showed 100% sensitivity and 100% specificity rates in detecting neoplastic meningitis compared with a sensitivity rate of 61.5% for cytology.[102] Although this study was limited by a small number of participants (n = 29), its results suggest that flow cytometry warrants further study for diagnosing neoplastic meningitis.[102]

However, a caveat to the use of flow cytometry on CSF samples is its low cell number and suboptimal cell environment, meaning that cancer cells degenerate following their in situ removal and even more so when repeatedly centrifuged. Other potentially significant limitations of flow cytometry include a high rate of false-positive results at low cell counts (< 25 cells/μL) and an inability to provide differential data (poor differentiation between monocytes and eosinophils and an inability to detect mitoses and neoplastic cells).[103,104] Thus, the clinical use of flow cytometry for the detection of neoplastic cells in the CSF is limited by variations in the equipment and methods.

The implementation of standardized protocols across clinical laboratories will be necessary before flow cytometry can be routinely implemented in clinical practice over conventional cytology.[101]

Fluorescence In Situ Hybridization

Fluorescence in situ hybridization can potentially aid in the diagnosis of leptomeningeal disease in patients with cancer.[105,106] It has been used to identify genetic changes in cancer cells from the CSF. For example, cells from 13 of 15 neoplastic meningitis CSF samples in 1 study showed numerical chromosomal abnormalities compared with no chromosomal abnormalities observed in the 10 control samples.[107] The study was limited by the use of patients who had already been diagnosed with cytology, thus suggesting that fluorescence in situ hybridization was less sensitive than cytology.[107]

Circulating Tumor Cells and Tumor DNA in Cerebrospinal Fluid

An analysis of CSF circulating melanoma cells was performed using immunomagnetic enrichment of cells expressing CD146 to diagnose neoplastic meningitis.[108] Enumeration of circulating melanoma cells in the CSF was correlated with CSF cytology obtained during the same lumbar puncture and with results on MRI. Among the negative CSF circulating melanoma cells, no study patient had neoplastic meningitis and 3 had brain metastasis.[108] It is worth mentioning that new technologies to collect CSF circulating cancer cells are being evaluated because they have been shown to perform better than cytology in detecting neoplastic meningitis in patients with lung cancer.[109]

Lin et al[110] presented the validation of CSF circulating tumor cells (CTCs) to diagnose neoplastic meningitis from epithelial tumors in patients suspected of having the disease. MRI and CSF analyses were performed using conventional cytology and enumeration of CSF CTCs for all study participants. Samples were considered positive for CSF CTCs when at least 1 CSF CTC was detected in a 3-mL sample (≥ 0.33 CSF CTCs/mL). The diagnostic performance of CSF CTCs was evaluated, and the gold standard was either a positive finding on CSF cytology or unequivocal findings on MRI. The rates of sensitivity were 95% for CSF CTCs compared with 81% for CSF cytology alone and 62% for MRI alone.[110] The rate of specificity was 83%.[110] Thus, this method had superior diagnostic performance when compared with CSF cytology or MRI.

A study of circulating tumor DNA in CSF was also published in patients with leptomeningeal disease.[111] Detecting tumor-derived, cell-free DNA in the blood of patients with brain tumors is challenging, presumably owing to the blood–brain barrier. The CSF may serve as an alternative "liquid biopsy" of brain tumors by allowing circulating DNA within the CSF to be measured so that tumor-specific mutations can be characterized. Many aspects about the characteristics and detectability of tumor mutations in CSF remain undetermined. Because CSF circulates through the CNS and interfaces with the brain as well as malignant tissues, CSF can potentially carry cell-free DNA and CTCs. Although cytology requires morphologically intact tumor cells for positive findings, cell-free DNA can presumably originate from dying but not CTCs anatomically distant from the site of CSF collection. Some studies have examined the nucleic acids in the CSF of individuals with brain tumors by using methods based on polymerase chain reaction, but the characteristics of CSF tumor cell-free DNA have not been comprehensively investigated using high-throughput sequencing.[111,112]

Methylation of promoter 2 of SHP1 in CSF was used as a detector of neoplastic meningitis related to epithelial-derived malignancy with a much higher rate of sensitivity than cytology.[113]