Introduction

Metastases to the brain are among the most clinically significant, because even a single one is likely to cause serious disability. Because most may be at least partially protected from otherwise effective systemic therapies by the blood–brain barrier, they present special management problems. But whether their pathogenesis differs from that of metastases at other sites is not known.

Origin of Primary Tumors

Since the first successful systemic treatment of a disseminated neoplasm – acute lymphoblastic leukemia (ALL) in children – the CNS has been recognized as a refuge for neoplastic cells. Permeation of the meninges at the base of the skull and spinal canal is mainly seen in certain types of lymphomas, though these occasionally produce nodular brain lesions. Such lesions are usually derived from epithelial neoplasms. Most are the result of hematogenous dissemination. The exceptions are those from prostatic carcinoma, and some breast cancers, where tumor reaches the brain from skull lesions, or the dura attached to involved bone. This occurs only rarely in myeloma and lymphomas.

Of all epithelial tumors, both small and non-small cell lung cancer are associated with the highest incidence of brain metastases (BM), which occur in 30–60% of cases, and become manifest early in the course of the disease, often even before the primary tumor is diagnosed ( ). Lung metastases may be especially likely to elicit edema, with ring enhancement and disruption of the blood–brain barrier, so that they produce focal deficits and other symptoms early ( ). Melanomas and renal cell carcinomas are also associated with a high incidence of BM ( ), but they are less common than breast or lung cancer.

Breast cancer is the second most common source of brain metastases, being a very common tumor in America; they ultimately develop in 35% of patients with metastatic disease ( ). They mainly occur in HER2+and triple negative breast cancer, arising less frequently from hormone receptor positive tumors ( ). Their natural history differs from those due to lung cancer. They are not usually symptomatic until late in the course of the disease, after metastases to bone and lung become detectable. The number of metastases per brain in breast cancer patients appears, in some, but not all studies ( ), significantly greater than those in the brains of lung cancer patients. But this may be, in part, due to gender; female lung cancer patients have been reported to have more metastases per brain than males ( ). The incidence of cerebellar metastases is highest in breast cancer patients ( ), but this does not necessarily indicate a predilection for cerebellar involvement. The incidence of breast metastases in the cerebellum correlates strongly with the total number of brain metastases present ( ). Thus, the occurrence of metastases in a given brain region may be a function of the total number of BM which have occurred.

Most lung cancer patients present with incurable disease, and systemic therapies are only effective in a minority. But the majority of breast cancer patients are potentially curable at presentation, especially with adjuvant therapies. Breast cancer metastases at any site are far more sensitive to systemic chemotherapy than those from lung cancer, and also respond to endocrine therapy in tumors expressing hormone receptors. Such treatments may delay overt progression of micrometastases for years, and 50% of patients with bone and lung metastases may be controlled by systemic therapy for many months. But the drugs used in systemic therapies may not reach the brain, so that metastases there may accumulate while other metastases are suppressed by therapy (Hengel et al., 2012). Thus, in breast cancer, the brain may be a refuge for metastatic disease.

Neoplastic Cell Meningitis

The leptomeninges are most often involved in ALL and certain lymphomas. This is assumed to be present in all ALL cases. It is rare in acute myeloblastic leukemia, though it occasionally occurs in the monoblastic subsets (M4 and M5). The lymphomas include Burkitt’s, T-cell, and a subset of especially aggressive and high stage diffuse large B-cell lymphomas. In the latter group, the presence of the typical 8;14 (myc) chromosome translocation should be ruled out, regardless of whether typical Burkitt’s histology is observed, because a posi- tive result implies an increased risk of meningeal involvement. Carcinomatous meningitis involvement is far less common. It is most frequently observed in small cell and breast carcinomas.

In fact, few such patients have a typical meningeal syndrome. They are usually alert, and without neck stiffness. The meninges at the bases of skull and cord are the most frequent sites of involvement. The usual clinical syndrome is that of mononeuritis multiplex, due to entrapment of nerve roots by meningeal growth. Ophthalmoplegias due to cranial nerve involvement are common, as are symptoms of lumbar root involvement.

Diagnosis

Early BM are often clinically silent but, ultimately, cause focal neurologic deficits of gradual onset; they are sometimes associated with more diffuse problems, such as amnesia or dementia. Only occasionally do they present with seizures, and most never cause them. Nevertheless, anti-epileptic therapies are frequently prescribed. This may be based on experience with primary intracerebral brain tumors, surgical resection of which can result in epileptogenic foci. Although brain imaging is usually performed for brain-related symptoms, it is also indicated in all lung cancer patients regardless of whether symptoms are present ( ). Asymptomatic metastases are often discovered in this way ( ).

BM which have caused neurologic deficits are almost always detectable by imaging, which can usually readily distinguish them from primary brain tumors. The former are typically sharply circumscribed, though they often display ring enhancement ( ). Primary intracerebral brain tumors are diffusely permeative at their margins. In fact, when adequate imaging fails to demonstrate BM, a different cause for focal deficits is likely.

Pre-contrast computed tomography (CT) scanning cannot rule out BM, but occasionally rules them in. Post-contrast CT scanning usually detects BM which have reached the point of causing neurologic deficits. Magnetic resonance imaging (MRI) is the most sensitive and specific method of demonstrating brain metastases. It can often detect small asymptomatic lesions which are not apparent on post-contrast CT scanning, and is more effective in detecting posterior fossa metastases ( ). If possible, it should be the primary diagnostic procedure of choice.

Biopsy of brain metastases is not usually necessary. In patients without a known neoplastic diagnosis, careful physical examination and imaging – especially of the chest – will disclose a source in many cases, and the diagnosis can be made by biopsy of a probable primary tumor. It is especially important to rule out apparently localized melanomas, which may metastasize early. Localized primary small cell tumors, whether bronchogenic or arising at other sites, are often found to have metastasized to the brain.

In patients at risk for primary brain lymphomas, such as those immunosuppressed post-transplant, or with AIDS, evaluation of cerebrospinal fluid Epstein–Barr DNA can help rule them in or out ( ). Clinical findings are of paramount importance in the diagnosis of neoplastic cell meningitis. It is not usually detected by CT scanning; MRI is positive in carcinomatous meningitis and about 50% of lymphomatous meningitis. Lumbar puncture to reveal neoplastic cells in the cerebrospinal fluid may be specific, but is quite insensitive. Examination of spinal fluid cells is diagnostic in less than 50% of cases, though flow cytometry may increase sensitivity. The presence of, for example, an abducens palsy is sufficient to establish the diagnosis.

Pathogenesis

BM are initially small, bland lesions which have not permeated adjacent parenchyma. Many are asymptomatic when imaging for staging detects them. But their growth is often associated with a local inflammatory, hypervascular reaction around them, denoted as ring enhancing by their appearance on imaging. The resulting edema is usually associated with neurologic deficits, and may lead to midline shift, herniation and sudden death. There is some evidence that BM from lung cancer acquire these changes more rapidly than those from other tumors ( ).

There is a possible anatomic explanation for the high incidence of BM in lung cancer. Cells from most tumors must pass through the pulmonary capillary circulation to reach organs other than the lung itself, which is the commonest site of metastases. Cells from primary lung cancers presumably enter the systemic circulation directly from draining veins. They thus reach the cerebral circulation without the interposition of other small vessels. Yet, specific cellular mechanisms which may facilitate the adhesion and growth of lung cancer cells in the brain still seem probable.

Studies in animal melanoma models revealed specific cell lineages from primary melanomas with a strong predilection for spread to the brain. Such cells were shown in vitro to bind to the surface of brain cells and to proliferate in this environment. A small population of primary tumor cells was shown to have a predilection for forming BM, and this population was enhanced by multiple in vitro passages ( ). The assumption was that such tumor cells had gained mutations enabling brain metastases. But the discovery of the epithelial–mesenchymal transition ( ) suggests that metastatic cells may not have gained mutations, but have acquired stem cell attributes. There is increasing evidence that metastatic cells do not necessarily acquire mutations additional to those in the primary tumors ( ).

That human BM arise from tumor stem cells expressing specific brain markers is reasonable to assume but difficult to demonstrate ( ). Tissue from such lesions is usually obtained at post-mortem examinations, in which the metastases are devitalized, so that markers cease to be detectable. Perhaps the best evidence for the existence of such markers is the predilection of specific breast cancer subsets for the brain. Were it possible to characterize brain specific markers in subsets of the cells from any tumors, a pharmacologic approach to inhibiting brain metastases might be possible. Animal models of BM have provided information concerning marker expression ( ), and have also provided data on the relative roles of cooptation of pre-existing vessels and de novo angiogenesis in the development of BM vasculature ( ). The lack of de novo angiogenesis in BM contrasts to its presence in primary brain tumors.