Clinical manifestations

Presentation and course

Most commonly, leptomeningeal metastasis becomes evident late in the course of malignancy, often when metastases are present at other sites. The onset of symptoms in most cases ranges from several days to weeks.

The clinical manifestations often can involve multiple segments of the CNS axis. The clinical features may be broadly classified into various syndromes: the cerebral syndrome, the cranial nerve syndrome, and the spinal syndrome. Patients frequently have symptoms originating from more than one syndrome at a time. In turn, this symptomatic classification may have only limited clinical- and research-related utility. The distribution and frequency of symptoms have been mainly reported and established by the early studies by Dr. Jerome Posner. The cerebral syndrome includes symptoms of headache, mental status alterations, seizures, nausea, and vomiting (Table 1). Headaches are usually positional and holocephalic and can be associated with nausea and vomiting. Mental status changes consist of confusion and decreased alertness. Seizures can occur as well but, along with focal deficits, are considered less common than headaches. With increasing sensitivity of diagnostic techniques, we may be detecting leptomeningeal metastasis more frequently prior to symptom development. In turn, the clinical syndromes described decades ago may have less direct relevance to contemporary patient care.

Table 1. Cerebral Signs and Symptoms Present Initially in 140 Patients with Leptomeningeal Metastasis from Solid Tumors

The cranial nerve syndrome involves dysfunction of one or multiple cranial nerves that can lead to visual loss with involvement of cranial nerve II; diplopia with involvement of cranial nerves III, IV, and VI; hearing loss and vertigo with involvement of cranial nerve VII and VIII; and dysphagia and dysarthria with involvement of cranial nerves IX, X, and XII.

Table 2. Cranial Nerve Signs and Symptoms Present Initially in 140 Patients with Leptomeningeal Metastasis from Solid Tumors

The spinal syndrome includes manifestations related to the spinal roots, spinal cord, and spinal meninges (Table 3). The clinical features may include weakness, paresthesias, sensory loss, decreased reflexes, neck or back pain, radicular pain, and dysfunction of bladder and bowel. Both weakness and sensory loss may be in a patchy distribution due to polyradiculopathy or in a distribution consistent with myelopathy. The roots most commonly involved are those of the cauda equina, perhaps because gravity causes tumor cells to accumulate at the lower levels of the spinal canal. Bladder dysfunction with urinary urgency, incontinence, or retention (with thoracic spine involvement) is common.

Alternatively, symptoms can be classified into the localizable (due to tumor affecting a specific neuroanatomic location) or nonlocalizable (typically due to increased intracranial pressure) (52).

Table 3. Spinal Signs and Symptoms Present Initially in 140 Patients with Leptomeningeal Metastasis from Solid Tumors

Clinical vignette

A 73-year-old man had history of stage 1A (T1N0M0) adenocarcinoma of the lung (with EGFR exon 19 deletion) after right lower lobe lobectomy in 2009, with CNS progression in the right occipital lobe in 2016 after resection and stereotactic radiosurgery. He was then started on gefitinib without treatment interruptions. In late 2017, he started experiencing progressive gait instability and headaches over a 3-week period. Brain MRI showed a heterogeneously enhancing lesion in the right occipital lobe associated with ependymal spread and interval progression of pial spread of the tumor. Multiple cranial nerves were involved, and there was interval ventricular size enlargement.

MRI of the spine showed enhancement along the dorsal aspect of the thoracic spinal cord.

CT of chest, abdomen, and pelvis also showed progression of disease in his chest and new boney disease in the thoracic spine. Whole-brain radiation therapy and ventriculoperitoneal shunt placement were offered to the patient and his family; however, ultimately, the patient decided to pursue hospice care.

Comment. This case highlights the significant morbidity that leptomeningeal disease can present with as well as the common late manifestation and association with systemic progression. The case preceded US Food and Drug Administration approval of osimertinib. Osimertinib would have been a reasonable medication to offer the patient and to add to the goals of care equation given the positive EGFR-mutation and good CNS penetration of the drug.

Biological basis

Etiology and pathogenesis

Potential mechanisms for the development of leptomeningeal metastasis include direct entry via the choroid plexus, hematogenous metastasis directly to the leptomeninges, direct invasion of the meninges from bony vertebral metastases (likely an infrequent mechanism), spread via nerve sheaths (especially with abdominal malignancies), and shedding of cells from solid brain metastases into CSF (especially after surgical breach of the ventricle at the time of metastasis surgery, particularly within the posterior fossa). Other postulated routes include spread via the perineural, endoneural, and perivascular lymphatics in nerve roots or cranial nerves; and spread via Batson’s venous plexus. These are likely less common routes of spread.

Interesting work suggested the role of complement component 3 (C3) and the complement receptor C3a that is expressed on the choroid plexus epithelium. On activation of the receptor, the blood-brain barrier is opened, allowing for plasma components to enter the CSF space and establishing the environment for leptomeningeal metastasis development (09). Also, it was shown that cancer cells appear to survive in the CSF by outcompeting macrophages for iron by expressing iron-binding protein lipocalin-2 (LCN2). In mouse models, iron chelation therapy inhibited the growth of leptomeningeal metastasis cells (14).

Epidemiology

Early studies, including autopsy reports, shed light on the prevalence of leptomeningeal metastasis in patients with cancer presenting with neurologic symptoms (approaching 20%) (49; 54; 06). Leptomeningeal metastasis can occur as a complication from solid tumors, hematologic malignancies, or primary brain tumors. Leptomeningeal metastasis is diagnosed clinically in 5% to 8% of patients with solid tumors and in 5% to 15% of patients with leukemia and lymphoma (06). Among the published cohorts of patients with leptomeningeal metastasis with solid tumors, the most common primary tumors were carcinoma of the breast (35% to 50%) and lung (25%) and melanoma (5% to 12%) (54; 29). Incidence of leptomeningeal metastasis within each tumor subtype ranges from 5% to 8% of metastatic breast cancers, 9% to 25% of lung cancers (more so in small cell lung cancer), and from 6% to 18% of melanomas (53). Overall, the incidence of leptomeningeal metastasis may be increasing due to improved systemic treatments that poorly penetrate the blood-brain barrier (eg, trastuzumab in HER2-positive breast cancer). This creates a safe haven for cancer cells in the CNS. Leptomeningeal metastasis was reported in 2% of patients with non-small cell lung cancer (NSCLC) at diagnosis of intracranial involvement (46), whereas it is less common at time of intracranial involvement of melanoma (04). In a large retrospective review of 2411 patients with brain metastases, among patients with newly diagnosed intracranial disease, the incidence proportion of concurrent leptomeningeal metastases was 11.4% versus 2.9% among patients with breast versus non-breast primaries (P < .001). The development of leptomeningeal metastasis among initially leptomeningeal metastasis-naïve patients was also more common among patients with breast versus non-breast primaries (HR = 1.49 [1.05-2.11], P = .03) (32). Our perspectives may be skewed, however, by the means in which leptomeningeal metastasis is diagnosed in various studies. Many studies include radiographic diagnosis of leptomeningeal metastasis without a gold standard of detection of malignant cells in the CSF (which often has less than optimal yield). In turn, relying on radiographic detection may be over- or under-calling the leptomeningeal metastasis incidence.

Improved diagnostic technologies have likely contributed to the increased perceived incidence of leptomeningeal metastasis as well. Lymphoma and leukemia frequently invade the CSF space. Intrathecal prophylaxis chemotherapy is commonly used to prevent this. Among primary brain tumors, the incidence of drop metastases is higher in medulloblastoma, ependymoma (15%), and pilocytic astrocytoma than in astrocytoma and oligodendroglioma (approximately 4.7%) (02).

Prevention

The most important prevention measures are those for prevention of malignancies, eg, avoidance of cigarette smoking or early detection of malignancies at a curable stage, such as breast examination and mammography. With acute lymphocytic leukemia and certain subtypes of lymphoma, the chemotherapy drugs commonly used for treatment penetrate the CNS poorly, resulting in a high risk of meningeal recurrence. CNS "prophylaxis" with radiation or intrathecal chemotherapy is commonly employed in the treatment of leukemia, lymphoma, and small cell lung cancer; this can be, rather, considered treatment of clinically undetected disease and not true prophylaxis. It has been suggested that en bloc surgical removal of brain metastasis may confer reduced risk of leptomeningeal seeding compared with piecemeal removal, particularly when the tumor is near CSF spaces (01). Although controversial, treatment with whole-brain radiotherapy in the setting of parenchymal metastasis may lower the risk of future leptomeningeal metastasis compared with stereotactic radiosurgery alone (42; 27). This is not routine practice, however, as other studies refute this idea (40). Interestingly, it has been proposed that preoperative stereotactic radiosurgery of brain metastases reduces the incidence of leptomeningeal metastasis (41). This is actively being investigated within the context of a National Cancer Institute (NCI) cooperative trial.

Differential diagnosis

Viral infections (especially in the immunocompromised), subarachnoid hemorrhage, and lumbar puncture procedures can lead to nerve root enhancement and arachnoiditis that can mimic leptomeningeal metastases on imaging (43). Spontaneous intracranial hypotension, especially in the setting of CSF leaks after surgeries, can cause diffuse pachymeningeal enhancement and should be considered in the correct clinical setting (37). Late effects of prior radiation treatments, especially when combined with intrathecal chemotherapy can also mimic leptomeningeal metastases, causing enhancement of the spinal nerve roots as well (26). In addition, in the era of immunotherapy and checkpoint blockade, neurologic complication such as Guillain-Barre syndrome can cause nerve root enhancement on MRI (50). In patients with advanced cancer, however, the differential diagnosis is usually narrow, and linear enhancement within the sulci of the brain and on the nerve roots is less likely to represent any process other than leptomeningeal metastases.

Diagnostic workup

Imaging studies. The anatomic pattern of leptomeningeal metastasis can be divided into three broad groups (47). These may have relevance to the diagnostic evaluation.

The first pattern consists of meningeal deposits forming plaques and small nodules on the meningeal surface. This is the pattern most characteristic of leptomeningeal metastasis from solid tumors. Tumor cells often invade the brain, spinal cord, or nerves, sometimes via Virchow-Robin spaces, often leading to neurologic symptoms. There is moderate shedding of cells into CSF.

The second pattern is one of thin layers of cells on the meninges, sometimes only one cell thick. It is typical of leukemia and lymphoma. Symptoms may be less severe, and the patients are sometimes asymptomatic. There is high shedding of cells into the CSF, but the probability of a positive cytology may be limited as much by the ability of the pathologist to identify these isolated cells (typically lymphocytes) as malignant, as by the presence of abnormal cells in CSF.

The third pattern consists of isolated nodules, sometimes large, on the cranial or spinal meninges. This is the pattern most typical of CNS primary tumors, especially medulloblastomas and ependymomas. Symptoms are more likely to be caused by compression (especially of spinal cord or nerve roots) than by CNS invasion. There is low shedding of cells into CSF, and cytology is more often negative than positive.

Gadolinium MRI is the preferred imaging technique for both brain and spine imaging because it is more sensitive than contrast CT (for head imaging) and CT myelography (for spine imaging). It is desirable to do MRI before lumbar puncture because lumbar puncture can give rise to temporary meningeal enhancement due to arachnoiditis that can be misleading. However, the pattern of enhancement post-lumbar puncture is usually smooth pachymeningeal (not leptomeningeal) enhancement. In patients with suspected leptomeningeal metastases, the entire neuraxis should be imaged. The leptomeningeal enhancement can be nodular or scattered over the surface of the brain, spine, or cranial nerves and spinal nerve roots in a “sugar coated” manner as above.

T1 hyperintensity is often associated with FLAIR hyperintensity. Either CT or MRI can demonstrate hydrocephalus caused by leptomeningeal disease.

CSF. Examination of CSF should be considered unless contraindicated, eg, by an intracranial mass that carries risk for herniation, noncommunicating hydrocephalus, spinal cord compression, coagulopathy, anticoagulant therapy, or thrombocytopenia. Even when the diagnosis is strongly supported by the clinical picture and imaging studies, CSF examination is helpful to confirm the diagnosis, rule out other etiologies, and, in some cases, establish a baseline for treatment monitoring response.

A positive cytology is normally taken as proof of leptomeningeal metastasis. With leptomeningeal metastases from solid tumors (non-CNS), the cytology is thought to be positive in about 60% on the first attempt and an additional 22% on the second but only an additional 2% on the third attempt (54). These studies, however, were conducted in an earlier era and may not reflect contemporary CSF analysis sensitivity. They are also plagued by a lack of a true gold standard against which cytology should be compared. Regardless, it is desirable to send a large volume (10 mL or more) for cytology to increase the yield (20).

Common abnormalities on routine CSF tests include elevated white blood cells (pleocytosis) (65% on the initial lumbar puncture among 470 patients), elevated protein (82%), and hypoglycorrhachia (45%) (54). At least one of these three routine tests will be abnormal in 97% of patients with leptomeningeal metastasis. These are all nonspecific findings.

Biochemical markers in CSF have been evaluated in leptomeningeal metastasis (55). Beta-glucuronidase, carcinoembryonic antigen (CEA), and lactic dehydrogenase isoenzyme 5 (LDH-5) are commonly increased with meningeal involvement from breast carcinoma, lung carcinoma, and melanoma. These are very rarely used in routine clinical practice. Other markers useful in a more restricted range of tumors include alpha-fetoprotein (AFP) and the beta subunit of human chorionic gonadotrophin (Beta-HCG) for germ cell tumors and certain ovarian and testicular carcinomas and beta2-microglobulin for lymphoma. False-positive values for most of these markers are common in infectious meningitis and other disorders causing CNS inflammation, and an elevation in one of these markers is generally not sufficient to diagnose leptomeningeal metastasis. Because of their poor sensitivity and lack of specificity, these biochemical markers have fallen out of favor except for their use with germ cell tumors. It is possible to obtain a tumor cell block from centrifuged CSF that can be used for immunohistochemistry and other molecular analyses.

Rare cell capture technology is a technique that allows quantification of the number of circulating tumor cells (CTCs) in the CSF per unit volume; it shows higher sensitivity than cytology in diagnosing leptomeningeal metastases. Moreover, this technology may become important when monitoring and quantifying treatment response. Similarly, quantification of circulating tumor cells and cell-free circulating tumor DNA (ctDNA) have shown superior sensitivity to CSF cytology (36). Mutations or fusions of oncogenes, including EGFR, ALK, ROS1, BRAF, etc. can be tested in the CSF. Assays for evaluating CSF ctDNA are commercially available.

In patients without known malignancy whose diagnosis of leptomeningeal metastasis by imaging and CSF was not confirmed, meningeal biopsy of an area that appears abnormal on MRI may be indicated. In patients with known malignancy who have an appropriate clinical syndrome, abnormal routine CSF tests, and no evidence for alternative diagnoses, a clinical diagnosis of leptomeningeal metastasis followed by treatment is appropriate.

Management

General management. When symptomatic hydrocephalus is present, ventriculoperitoneal shunting is usually recommended unless a decision is made to offer only simple supportive measures. There is theoretical risk of tumor spreading to the peritoneal cavity, but it is believed that the benefits still outweigh the risks. Shunts performed for leptomeningeal metastasis often fail due to blockage by tumor cells or protein. At the time of shunting, consideration should be given to including a scalp reservoir and an on-off valve in the shunt tubing if intrathecal chemotherapy is planned.

The benefit provided by dexamethasone or other corticosteroids is uncertain. The symptomatic improvement provided by corticosteroids is less impressive with leptomeningeal metastasis than with solid metastases in the brain. Dexamethasone should probably be omitted or used only in low to moderate doses, except in patients with large meningeal deposits compressing the spinal cord and in those who experience substantial symptomatic benefit.

Physicians must decide together with the patient and family whether antineoplastic treatment with radiation, intrathecal chemotherapy, or systemic chemotherapy is warranted in view of the poor prognosis even with aggressive treatment. For the patients with very poor functional status, a decision to offer only supportive care may be appropriate. In addition, one should consider whether the tumor is of a type that often responds to radiation or chemotherapy (eg, small cell lung carcinoma, breast carcinoma, germ cell neoplasms, lymphoma, leukemia) or of a type that infrequently responds (eg, non-small-cell lung carcinoma, melanoma, gastrointestinal tumors).

For patients regarded as candidates for aggressive treatment, several strategies may be considered. When no promising systemic treatment option is available, intrathecal chemotherapy directly into a CSF reservoir (ie, Ommaya) is commonly used, especially for leptomeningeal metastasis from breast cancer. Cancers such as non-small-cell lung carcinoma and melanoma are less likely to respond to intrathecal chemotherapy.

Radiation. Radiation might be used only for bulky tumors, tumors blocking CSF pathways, or progression after intrathecal chemotherapy. If leptomeningeal metastasis causes hydrocephalus or is associated with multiple brain metastases, whole-brain radiation therapy is often needed to alleviate the obstructive hydrocephalus along with ventriculoperitoneal shunting. However, oncologists should be cautious when combining whole-brain radiation therapy with intrathecal methotrexate given risk of leukoencephalopathy.

With the development of proton radiation therapy, this method may be preferred over photon radiation when irradiating the spine in order to lessen the side effects on the bone marrow. In a phase I trial, 24 patients with leptomeningeal metastasis received hypofractionated craniospinal radiation (56). Ten percent of patients developed dose-limiting toxicities with leukopenias and fatigue. Median overall survival was 8 months after a median follow-up of 11 months. This may provide new treatment strategy for patients with leptomeningeal metastasis, especially if combined with systemic chemotherapy.

Systemic chemotherapy. Systemic chemotherapy has traditionally not been successful in the management of leptomeningeal metastasis as many chemotherapy drugs penetrate the normal CNS poorly. Most studies have been small single or oligo-institution studies utilizing single therapeutic regimens; therefore, generalizability and comparison with other regimens is difficult. Some examples of systemic treatments that have been tried are described below. In a phase 2 study of 19 patients with leptomeningeal metastases, temozolomide at 100 mg/m2 (week on/ week off schedule), led to clinical benefit in only three patients, with a median time of progression of 28 days (48). Capecitabine has been reported in a few cases and in a retrospective study of patients with leptomeningeal metastases from breast cancer. Median overall and progression-free survival were reported at 13 and 8 months, respectively (17). In a prospective cohort study of patients with leptomeningeal metastasis from breast cancer, bevacizumab, etoposide, and cisplatin (the BEEM regimen) showed survival advantage over intrathecal methotrexate (13). These results may be confounded by selection bias.

Systemic tyrosine kinase inhibitors with good CNS penetration have, however, shown promising results in patients with leptomeningeal metastasis. The BLOOM study reported 62% (95% CI, 45% to 78%) overall response rate with osimertinib 160 mg daily (a third generation EGFR inhibitor) in patients with EGFR-positive non-small-cell lung cancer and leptomeningeal metastasis. Duration of response was 15.2 months (95% CI, 7.5 to 17.5 months) (56). Median overall survival was 11.0 months (95% CI, 8.0 to 18.0 months) with 68% maturity. Overall survival rates were numerically less impressive in clinical trials with the older generation EGFR tyrosine kinase inhibitors, afatinib and erlotinib (3.8 and 3.4 months, respectively) (51; 38). Similarly, ceritinib showed 62.5% (24.5, 91.5) intracranial disease control rate in patients with anaplastic lymphoma kinase-positive non-small-cell lung cancer and leptomeningeal metastasis (05).

Immunotherapy has been evaluated in leptomeningeal metastasis as well. In a single-arm, phase 2 study of pembrolizumab in patients with solid-tumor malignancies (17 with breast cancer, two with lung cancer, and one with ovarian cancer) and leptomeningeal metastases, patients received 200 mg of pembrolizumab intravenously every 3 weeks (10). The study met the primary endpoint of a Simon 2-stage designed trial, as 12 of 20 (OS3, 0.60; 90% confidence interval, 0.39 to 0.78) patients were alive 3 months after enrollment. Median overall survival, however, remained short at 3.6 months (90% CI, 2.2 to 5.2).

In another systemic therapy, ANG-1005, a peptide-drug conjugate, facilitating paclitaxel entry into the CSF, has demonstrated promising results with some patients experiencing responses (30).

Intrathecal chemotherapy. Intrathecal administration of chemotherapy bypasses the blood-brain barrier and achieves high concentration in the CSF. The results from Ettinger and colleagues show that the CSF methotrexate concentration with intraventricular administration is in the range of 1600 to 8500 times that of a similar dose given intravenously (18). Intrathecally administered chemotherapy drugs, however, cannot be expected to penetrate more than a few millimeters into the CNS or into bulky tumors (28).

Intraventricular administration is generally preferred to lumbar injection. Shapiro showed that methotrexate injected at lumbar puncture reaches the brain only in low concentrations, whereas methotrexate injected into the lateral ventricles is distributed more uniformly (49). Glantz and colleagues demonstrated the superiority of intraventricular chemotherapy over intralumbar chemotherapy, particularly for short-acting agents (23). In addition, multiple punctures of a scalp reservoir are better tolerated than multiple lumbar punctures. Lumbar treatment may be advocated for "prophylaxis" (eg, in leukemia), where the number of treatments is small. After placement of a ventricular catheter and reservoir, some neuro-oncology practices perform a CSF flow study with a radionuclide such as ¹¹¹Indium to assure adequate flow out of the ventricles into the cranial and spinal subarachnoid space (11; 21). However, it is unclear if this approach provides added value. If flow is blocked by tumor, radiation should be given, with intra-CSF treatment to follow only if the block is relieved to avoid complications of clustering of chemotherapy in a confined space.

Sustained-release liposomal cytarabine was approved in the United States for intrathecal treatment of leptomeningeal metastasis. The cytarabine is released slowly from multivesicular lipid particles, prolonging the cytotoxic drug concentration in CSF for more than 14 days and exposing tumor cells throughout all phases of the cell cycle. Two trials supported its efficacy in lymphomatous meningitis (better than free cytarabine) and leptomeningeal metastasis (comparable to intrathecal methotrexate). Unfortunately, the pharmaceutical company has now discontinued the production of this drug.

Chemotherapy drugs administered intrathecally include methotrexate, cytarabine, and thiotepa. In adults, methotrexate is usually given in a dose of 10 to 15 mg. To reduce the chance of systemic side effects, some practices follow intrathecal methotrexate by folinic acid. Cytarabine is usually given in a dose of 50 to 75 mg. In 52 patients randomly assigned to methotrexate 10 mg or thiotepa 10 mg twice weekly, no difference in response rate or disease outcome was observed (25). Either drug may be mixed in artificial CSF (Elliott's B solution) or preservative-free normal saline, but it is important that solutions with preservatives be avoided. Combining intrathecal chemotherapy with hydrocortisone or systemic dexamethasone may reduce the incidence of symptoms of arachnoiditis. With either drug, intrathecal chemotherapy is generally performed twice weekly for approximately 4 weeks. If the patient is stable or improves, treatment is continued weekly for several weeks and less frequently thereafter. If the patient worsens clinically or if the CSF worsens, treatment should be discontinued.

CSF administration of monoclonal antibodies has been investigated. These include trastuzumab for HER2+ tumors (34; 35; 36). In a phase I/II clinical trial, the recommended phase II dose for intrathecal trastuzumab was 80 mg twice weekly. Twenty-six patients were included, and partial response was seen in five (19.2%) patients, stable disease was observed in 13 (50.0%), and eight (30.8%) of the patients had progressive disease. Median overall survival for phase II dose-treated patients was 8.3 months (95% CI 5.2-19.6) (31).

In a phase 2 study of 30 patients with leptomeningeal metastases from lung cancer who failed EGFR tyrosine kinase inhibitors, intrathecal pemetrexed 50 mg with dexamethasone was associated with a median overall survival of 9 months (19). These more favorable results may reflect a better natural history for this specific patient population.

Outcomes

Prognosis of leptomeningeal metastasis from solid tumors remains dismal. Without treatment, survival is generally measured in weeks rather than months. Balm and colleagues found that among patients who were not treated, approximately one fourth of their patients, survival averaged only 3 weeks (03). Several retrospective series of aggressively treated patients have reported average survival of 5 to 7 months (54; 21), but they may have achieved better results than others by carefully selecting patients for treatment or by reporting only patients who were stable after radiotherapy. The clinical trial by Glantz and colleagues documented median survivals of 3.5 months with liposomal cytarabine and 2.6 months with methotrexate (22). In the methotrexate versus thiotepa trial cited above, median survival for patients receiving methotrexate was 15.9 weeks and 14.1 weeks for patients treated with thiotepa (25). Survival seems to have considerably improved, in the appropriate subset of patients, with the promising new tyrosine kinase inhibitors; median survival was 11 months in the BLOOM study with osimertinib (56). In a large retrospective study of 2411 patients with intracranial metastases, 273 had leptomeningeal metastases. Patients with leptomeningeal metastasis secondary to breast versus non-breast primaries displayed lower all-cause mortality (HR 0.70 [0.52-0.93], P = .01; median survival: 5.2 vs. 2.4 months, respectively), with a greater numerical difference observed in patients with leptomeningeal metastasis at intracranial involvement (7.4 vs. 2.6 months, respectively) (32).

Many patients with leptomeningeal metastasis from solid tumors do not improve clinically despite aggressive treatment. Among 252 patients from multiple series, 28% improved clinically with treatment, 32% stabilized for a period of time, and 40% progressively worsened (54; 03).

Prediction of the prognosis of individual patients may help in treatment decisions. The prognosis is much worse when the systemic cancer is progressing than when it is stable (25). Patients with more favorable tumors, such as breast carcinoma, lymphoma, and leukemia, can be expected to survive longer than those with unfavorable tumors, such as melanoma and non-small-cell lung carcinoma (24). Long duration of symptoms at presentation is predictive of better prognosis (03), as is a good level of function, including ability to care for one's self. Different series have indicated that supratentorial involvement or cranial neuropathies predict shorter survival (25; 03). Bulky disease evident by MRI and blocked CSF pathways are associated with a worse prognosis, especially when the block cannot be corrected with radiotherapy (21). High CSF protein has been found to have negative predictive value, whereas intrathecal chemotherapy is associated with better survival (03).

Complications of the disorder include electrolyte disturbances (notably hyponatremia), infections such as pneumonia and urinary tract infections, thromboembolic complications, such as deep vein thrombosis and pulmonary embolism, and decubitus ulcers.

Side effects of radiotherapy are usually mild; serious complications are rare. Alopecia, mild fatigue, anorexia, and weight loss are common side effects of cranial radiotherapy. Pancytopenia is common with radiotherapy to the entire spine but uncommon with partial spine or cranial irradiation.

Proton therapy is likely safer than photon when craniospinal radiation is performed. The more serious late complications of radiotherapy such as radiation myelopathy are uncommon, partly because the radiation doses used for treatment of this condition are usually modest and partly because few patients survive long enough to be at risk for such complications. When combined with methotrexate, radiation undoubtedly contributes to leukoencephalopathy as above.

Complications of ventricular catheter and reservoir placement are common (39; 12). Among 1076 patients, complications included infection (early or late) in 7.5%, technical complications such as catheter disconnection or nonfunctioning catheter in 6.8%, intracranial hemorrhage in 1.4%, and poor wound healing in 0.5%.

Intrathecal chemotherapy can cause a variety of complications. Lumbar or ventricular injection of chemotherapy may cause symptoms and signs of meningeal irritation including headache, nuchal rigidity, fever, and vomiting as well as symptoms of encephalopathy. Meningeal irritation and acute encephalopathy are more prominent with methotrexate than with cytarabine, and they usually resolve within 1 to 3 days. Lumbar injection of methotrexate or cytarabine rarely causes myelopathy; the etiology is unknown.

An important and serious complication is the late occurrence of leukoencephalopathy (08; 44). Although methotrexate (especially intraventricular) and cranial radiotherapy can each independently cause this problem, the risk is greatest when the treatments are combined. Obbens described leukoencephalopathy in only 2% of 295 patients treated with intraventricular methotrexate, but the incidence may be above 50% in patients who survive 1 year or more after intensive treatment with radiation and intraventricular methotrexate (39; 33). The risk is thought to be much lower with cytarabine and liposomal cytarabine than with methotrexate. Patients with this syndrome typically have dementia and a gait disturbance. Seizures, quadriparesis, and disturbed control of the bowel and bladder may ensue. Leukoencephalopathy often progresses to cause the patient's death. A more limited form of this syndrome may consist only of mild cognitive deficits, a problem most commonly seen in long-term survivors of leukemia who have undergone treatment or prophylaxis with intrathecal methotrexate.