Clinical manifestations

Presentation and course

Many patients with brain metastases have some neurologic symptoms throughout the course of their illness. The clinical presentation of brain metastases is similar to that of other mass lesions in the brain and is determined by the size, location, and number of lesions. Symptoms can progress quickly and evolve over a few weeks. Most patients have a combination of "generalized" nonlocalizable symptoms (eg, headache, altered mental status, cognitive impairment) as well as "focal" signs and symptoms determined by the anatomic location. Generalized symptoms occur in patients with mass effect and increased intracranial pressure or in patients with multiple bilateral metastases. Headache is the single most common symptom of brain metastases, although a minority of patients never develop headache. Approximately 15% of patients have seizures as a presenting symptom of the metastasis, with another 10% of patients developing seizures subsequently. This is a lower seizure incidence than that which is observed with most primary brain tumors. Anywhere from 5% to 10% of patients present with acute neurologic symptoms caused by hemorrhage of, or sudden expansion of, the metastasis. Hemorrhage of a metastasis is particularly common with choriocarcinoma, renal cell carcinoma, and melanoma. Because brain metastases present varied signs and symptoms, their presence should be suspected in all patients with known systemic cancer in whom any new neurologic findings develop (46). As central nervous system imaging is more frequently performed, it is likely that asymptomatic brain metastases are being detected with increasing frequency.

Prognosis and complications

Brain metastases are generally associated with a poor prognosis (as are most other manifestations of stage IV cancer), regardless of treatment. Historically, patients who receive no treatment for brain metastases or those treated only with corticosteroids generally survive 1 to 2 months and die as a direct result of the brain lesions. Historically, overall median survival is 3 to 8 months among patients who receive some type of anti-tumor treatment for brain metastases (156; 126). More recent data, however, have shown improvement in survival with advances in systemic therapy and the median ranges from 8 to 16 months and beyond (159; 160). For most patients, treatment of the brain metastases prevents death due to neurologic causes, so the overall survival prognosis becomes largely dependent on control of the systemic cancer. However, uncontrolled or recurrent brain metastases remain the sole or contributing cause of death in a significant proportion of patients with melanoma, non-small cell lung cancer, and breast cancer.

Several key factors play a role in patient outcomes, including age, extent of primary disease control, rate of brain metastases development (ie brain metastases velocity), presence of extracranial metastases or leptomeningeal disease, performance status (typically measured by the Karnofsky Performance scale), extent of intracranial disease, and histology of the primary tumor. To more reliably estimate prognosis for patients with brain metastases and to help guide decisions regarding the appropriate aggressiveness of therapy, prognostication indices have been developed. These include the recursive partitioning analysis (RPA) score, graded prognostic assessment (GPA), and disease-specific GPA (dsGPA), among others.

In 1997, Gaspar and colleagues reported a recursive partitioning analysis of survival data from more than 1200 patients with brain metastases treated with whole brain radiation in Radiation Therapy Oncology Group (RTOG) protocols. Three RPA classes with significantly different survival were identified (see Table 1). The patient population utilized to create these RPA classes may not reflect the contemporary brain metastases population encountered in oncology and neuro-oncology clinics.

Table 1. Recursive Partitioning Analysis (RPA) Classes for Persons with Brain Metastases

Median survival for patients in RPA Classes 1, 2, and 3 was 7.1 months, 4.2 months, and 2.3 months, respectively (63). The predictive value of this RPA classification scheme was validated in a subsequent RTOG study of whole brain radiation therapy (62), as well as in several series of patients treated with stereotactic radiosurgery (37; 142).

A newer prognostic scheme for patients with brain metastases eliminates the subjective judgment of "controlled" or "uncontrolled" systemic tumor and incorporates data correlating overall survival with the number of brain metastases. Patients are assigned a subscore for each of four key prognostic factors and then a total graded prognostic assessment (GPA) score ranging from 0.0 to 4.0 (see Table 2).

Table 2. Graded Prognostic Assessment (GPA) for Persons with Brain Metastases

Retrospectively applying the GPA to nearly 2000 patients from RTOG studies defined four patient groups with significantly different median survivals (GPA 0.0 to 1.0, 2.6 months; GPA 1.5 to 2.5, 3.8 months; GPA 3.0, 6.9 months; GPA 3.5 to 4.0, 11.0 months) (155). The majority of patients in these studies had lung cancer. Unfortunately, up to one half of patients with newly diagnosed brain metastases fall into a poor prognostic category (total GPA score 0.0 to 1.0).

The predictive value of the individual prognostic factors in the GPA varies somewhat among different primary tumor types, leading to "disease-specific GPA" (156; 157). Molecular markers such as EGFR mutations and ALK rearrangements have been incorporated to create a more specific lung cancer GPA (161). This is similar to the prior incorporation of hormone receptor and HER-2 status in the GPA for breast cancer patients (158). Disease-specific GPA is currently used in several clinical trials for patient eligibility and stratification.

Another method for estimating individual patient prognosis is the use of nomograms (13; 112), which in some studies outperformed the disease-specific GPA. Finally, the use of brain metastases velocity, the annualized rate at which a patient develops new brain metastases, can be used to predict the likelihood of death due to neurologic causes (56; 115). As would be assumed, patients who develop numerous metastases quickly have a higher likelihood of having their neurologic CNS disease burden drive mortality.

Clinical vignette

After presenting with a syncopal episode, a 55-year-old man was diagnosed with metastatic small cell lung cancer, involving lung, lymph nodes, bone, and adrenal gland. Initial staging showed no evidence of brain involvement, and he began treatment with cisplatin/etoposide and planned for prophylactic cranial irradiation. After four months of treatment and a systemic response, he underwent an MR brain for radiation planning, which demonstrated at least 20 supra and infratentorial metastases, some with intratumoral hemorrhage.

He experienced no major neurologic symptoms and completed 3000 cGy whole brain radiation therapy. His therapy was subsequently changed to ipilimumab and nivolumab for progressive systemic disease. However, after 2 months, he developed superior vena cava syndrome from worsening disease in the chest and tumor encasement of the pulmonary artery. MRI brain also showed a mixed response, with growth of two previously treated metastases and a new one, which were targeted with stereotactic radiosurgery. He remained neurologically asymptomatic and continued treatment with single agent paclitaxel. Within 2 months, repeat imaging of the brain showed numerous new and increased metastases, and treatment was changed to temozolomide. However, after one month, the disease progression in the brain continued, which systemically prompted a change to etoposide, carboplatin, and bevacizumab. He experienced minimal neurologic symptoms aside from fatigue related to treatment and some mild difficulty with concentration, and an MR brain demonstrated an initial response. Five months later, however, he experienced a generalized seizure after which he had progressively worsening confusion, lethargy, and gait instability. MR brain confirmed progression of multiple brain metastases with peritumoral edema, and CT for staging showed systemic progression as well. The decision was made to discontinue tumor directed therapies and only continue steroids and seizure medications. He died 2 years after his initial diagnosis.

Biological basis

Etiology and pathogenesis

Brain metastases are caused by the spread of tumor cells to the brain from tissues located outside the central nervous system.

Most tumor cells reach the brain by hematogenous spread, nearly always via the arterial circulation. Within the brain, metastases are most commonly found in the area directly beneath the gray-white junction. The predominance of metastases here is due to a change in the size of blood vessels at this point; the narrowed vessels act as a trap for emboli. Brain metastases also tend to be more common at the terminal "watershed areas" of arterial circulation (the zones on the border of or between the territories of the major cerebral vessels). The distribution of metastases among the large subdivisions of the central nervous system follows roughly the relative weight of and blood flow to each area. Approximately 80% of brain metastases are located in the cerebral hemispheres, 15% in the cerebellum, and 5% in the brainstem.

There are large differences in the propensity of various neoplasms to metastasize to the brain parenchyma. The highest prevalence of brain metastases occurs in patients with germ cell tumors, melanoma, and small cell lung carcinoma. Breast carcinoma, non-small cell lung cancers, and renal carcinoma have an intermediate propensity to metastasize to the brain. Among patients with breast carcinoma, a higher risk of developing brain metastases is associated with overexpression of the epidermal growth factor receptor or of the HER2 (ErbB-2) growth factor receptor, or with absent expression of HER2, estrogen receptors, and progesterone receptors ("triple-negative tumors") (88; 44). Mutations in EGFR and KRAS in non-small cell lung cancer are associated with a high prevalence of brain dissemination in non-small cell lung cancer (178; 117). Brain metastases are disproportionately uncommon among patients with carcinoma of the prostate and GI tract. Sarcomas do not commonly spread to the brain, despite their high propensity to metastasize to the lungs.

Studies are elucidating the molecular pathophysiology of brain metastases and the molecular basis for the differences in "neurotropism" among various systemic tumors. To successfully form brain metastases, tumor cells must attach to and penetrate microvessel endothelium, degrade the extracellular matrix, and respond to autocrine and brain-derived survival and growth factors. Complex interactions between the primary tumor site, metastatic cells, and the tumor microenvironment are crucial for growth before and after seeding (175; 133; 26). The role of astrocytes in metastases has been elucidated more recently, and coupling of metastatic cells with astrocytes through gap junctions can confer chemoresistance and enhance tumor growth (35). Immunosuppressive elements of the metastatic microenvironment through tumor associated macrophages and regulatory T cells, also play a key role (91). In addition, cancer cells metastasizing to the brain are often molecularly distinct from the primary tumor, harboring driver mutations absent in their respective primary tumors. Some of these genetic alterations include mutations in CDKN2A and PIK3CA, loss of PTEN, amplification of ERBB2, and activating proto-oncogene KRAS mutations (20).

Epidemiology

Approximately 150,000 patients in the United States develop symptomatic brain metastases each year, making them the most common intracranial malignancies. Approximately 60% to 80% of patients that present with brain metastases also have synchronous systemic metastatic disease, with pulmonary involvement being the most frequent (170). Any neoplasm is capable of metastasizing to the brain, although two thirds of all adult patients have lung cancer, breast cancer, or melanoma. Bronchogenic carcinoma is the single most common source of brain metastases in adults, accounting for 30% to 50% of all cases (46). Lung carcinomas are also the tumors most likely to spread to the brain in the absence of other systemic metastases. In most series of brain metastases breast carcinoma and melanoma each account for 10% to 20% of cases. In up to 10% of patients presenting with brain metastases, no primary tumor can be initially identified despite investigation. The primary tumor site, if eventually detected, is usually the lung (124).

For some tumors, there is an increasing incidence of brain metastases as the sole or early site of relapse due to more effective therapies for systemic disease. This is especially true for non-small cell lung cancer and for breast carcinomas that overexpress HER-2. Women with breast carcinoma treated with the anti-HER-2 monoclonal antibody trastuzumab have been shown to have a 25% to 35% cumulative lifetime incidence of brain metastases, and about a 10% incidence of isolated brain metastases as the sole site of tumor progression (96; 24). This may at least in part reflect an "unmasking" of occult brain metastases that are protected by the inability of trastuzumab to cross the blood-brain barrier (103). However, data are conflicting as three large, randomized phase 3 clinical trials did not show a significant difference in the incidence of brain metastases in HER-2 positive patients treated with or without trastuzumab (138; 151; 96).

Brain metastases in children occur in approximately 5% of patients with solid tumors, and most often arise from sarcomas (osteogenic sarcoma, rhabdomyosarcoma, and Ewing sarcoma) and neuroblastoma (19; 87).

Prevention

Prevention of brain metastases has mainly been focused on patients with small cell lung cancer, because of the high prevalence of brain metastases after systemic treatment and the high degree of sensitivity of this tumor to radiation. For patients with limited-stage small cell lung carcinoma who achieve a complete remission after initial systemic therapy, the consensus results of several randomized studies indicate that prophylactic whole brain radiation therapy significantly reduces the cumulative incidence of brain metastases (from 50% to 60% to 25% to 30%) and significantly improves overall survival (05; 72; 08). Prophylactic cranial irradiation may also be beneficial for a subset of patients with initially extensive small cell lung carcinoma who have a good response to initial chemotherapy (150). Deferring prophylactic whole brain radiation therapy until after completion of induction systemic chemotherapy is believed to reduce the risk of long-term neurotoxicity. The most commonly used dose for prophylactic whole brain radiation therapy is 2500 cGy in 10 daily fractions. Higher doses are not more effective.

Although there has been an increasing incidence of metachronous brain metastases from non-small cell lung cancer with improvements in multimodality treatment of newly diagnosed patients, several studies have shown that prophylactic whole brain radiation for patients with non-small cell lung cancer does not prolong overall patient survival and can have significant neurotoxicity that can impact quality of life (69). Prophylactic whole brain radiation therapy is not currently considered standard care for patients with non-small cell lung cancer.

Differential diagnosis

Confusing conditions

The differential diagnosis in a patient presenting with a solitary or multiple enhancing brain lesions includes primary brain tumors, abscesses, cerebral infarction, hemorrhage, and demyelinating disease. Contrast-enhanced MRI is usually able to differentiate among these possibilities and the finding of multiple lesions is strongly suggestive of metastases, particularly in a patient with a known diagnosis of cancer. However, not all brain lesions in patients with systemic cancer are metastases. In a study of patients known to have systemic cancer who had single lesions diagnosed by contrast MRI, 11% had lesions other than brain metastases when tissue was obtained (128). Half of the nonmetastatic lesions were primary brain tumors; the other half were infections. The false-positive rate for diagnosis of multiple metastases is unknown but is certainly much less than the 11% quoted for single metastases. Biopsy is the only truly reliable method of establishing the diagnosis, if there is any doubt after contrast-enhanced imaging studies.

Associated or underlying disorders

There are no specific conditions or genetic predisposition syndromes specifically associated with the development of brain metastases from a systemic malignancy.

For patients presenting with suspected or proven brain metastases and no previous cancer diagnosis, workup for a systemic cancer should be pursued with a focus on the lungs. If the primary tumor is not bronchogenic, it has probably spread to the lungs before seeding into the arterial circulation and reaching the brain. More than 60% of patients with brain metastases will have a mass demonstrated on chest radiograph, CT scan, or MR scan that is caused by either a primary lung cancer or a lung metastases from the primary located elsewhere (113). It is reasonable to perform a mammogram in women and to do a careful skin examination in all patients. A CT or MR scan of the abdomen occasionally will show an unsuspected renal or colon cancer. Whole-body FDG-PET scanning may reveal a primary tumor that goes undetected by other imaging techniques (90). The practical utility and cost effectiveness of PET scanning in this setting are not clearly known. In the few patients with brain metastases but no identifiable lesions in the lung, the pathogenesis of the brain metastases may be spread through Batson vertebral venous plexus, tumor emboli through a patent foramen ovale (paradoxical embolus), or tumor filtered through the lungs with only local or microscopic growth.

Diagnostic workup

The best diagnostic test for brain metastases is contrast-enhanced MRI, showing more than one lesion in up to 75% of patients. Findings that favor metastases include multiplicity of lesions, a gray-white junction location, a lesion in the border zone between two major arterial distributions, and a small tumor nidus with a large amount of associated vasogenic edema. Contrast-enhanced MRI is more sensitive than either enhanced CT scanning (including double-dose delayed contrast) or unenhanced MRI in detecting lesions. Up to one third of patients with an apparently single lesion on contrast-enhanced CT have one or more other small lesions when a contrast-enhanced MRI is obtained (143).

The number of brain metastases is important for treatment planning (eg, surgical resection or stereotactic radiosurgery) and, to some extent, for prognostic predictions. Techniques such as 1.5-tesla MRI with double-dose gadolinium and 2 mm cuts (55), 3.0-tesla MRI with thin cuts (139), high-dose gadolinium, or "high resolution" MRI (75) can detect additional brain metastases in approximately 20% of patients compared with "standard" 1.5-tesla MRI scanning. In one retrospective study, 3.0-tesla MRI was slightly more sensitive than 1.5-tesla MRI in detecting small brain metastases prior to stereotactic radiosurgery, but patients who had 1.5-tesla scans did not have a higher incidence of subsequent "distant brain failure" (105).

Advanced neuroimaging techniques are also being used in clinical evaluation of suspected brain metastases as MR imaging cannot always reliably distinguish a single brain metastasis from an anaplastic glioma or an abscess. MR spectroscopy and MR perfusion may be helpful in this situation. Brain metastases are generally well circumscribed, whereas anaplastic gliomas nearly always have tumor cells infiltrating beyond the limits of contrast enhancement. These differences are reflected in the "peritumoral" choline peaks on MR spectroscopy and on cerebral blood volumes (95).

Although neuroimaging can be used to identify brain lesions, it is not specific enough for definitive diagnosis, and histopathologic analysis of tissue remains the gold standard for diagnosis. As a general rule, patients should always have histologic verification of cancer, whether from an extraneural tumor or from the brain lesion itself. For patients with a previous known cancer diagnosis, biopsy of the brain lesion is usually not necessary. Factors that raise the level of uncertainty for the diagnosis of brain metastases include: systemic tumors (eg, prostate carcinoma), which do not often metastasize to the brain parenchyma; a long interval since systemic tumor diagnosis; minimal or no systemic tumor burden; and a single versus multiple brain lesions. Imaging in these circumstances is often not reliable enough to substitute for histologic verification. A role for “liquid biopsy” of cerebrospinal fluid has been suggested as a minimally invasive approach to assess the presence of characteristic genetic alterations of brain metastases (129; 17). This approach, however, still requires validation.

Management

Treatment options for patients with brain metastases include a combination of surgical resection, whole brain radiotherapy, stereotactic radiosurgery, systemic medical therapy, and/or supportive measures including anticonvulsants and corticosteroids. Several factors must be considered when determining the ideal treatment for each patient, including the primary tumor histology, extent of systemic disease, anticipated survival, neurologic status at diagnosis, and the location, number, and size of metastases. There are several possible measures of treatment efficacy for patients with brain metastases, including improvement in neurologic signs and symptoms, improvement or maintenance of functional performance status, corticosteroid requirements, and radiographic tumor response. Survival after treatment of brain metastases is an easy endpoint to measure but is confounded by the fact that systemic tumor is the cause of death for most patients with brain metastases. Measuring survival is, therefore, more meaningful when the cause of death is determined to be "neurologic" or "non-neurologic." Because brain metastases in most patients arise in the setting of disseminated tumor, the goal of treatment is not necessarily to eradicate the brain lesions. Rather, treatment should be aimed at "controlling" the brain lesions, and at maintaining patients' neurologic function for as long as possible.

Corticosteroids. In the contemporary era it is no longer necessary to initiate steroids in the majority of newly diagnoses brain metastases patients. Careful consideration should be made prior to initiating steroids as they may abrogate the efficacy of immunotherapies, oft used treatments for systemic cancers. In addition, they may obfuscate the diagnosis if the patient has CNS involvement of lymphoma.

Anticonvulsants. Seizures occur in approximately 25% of patients with brain metastases and are among the presenting symptoms in 10% to 15% of patients. To date, there is no definite evidence that any particular antiepileptic drug is differentially effective for tumor-related seizures versus epilepsy caused by other structural brain lesions. In addition to their direct side effects, anticonvulsants may cause unfavorable drug interactions with dexamethasone and several chemotherapy agents. For patients taking dexamethasone or receiving chemotherapy agents metabolized by the liver, the nonenzyme-inducing antiepileptic drugs (eg, levetiracetam or lamotrigine) may offer fewer drug interactions than enzyme-inducing drugs (eg, phenytoin or carbamazepine). Valproate may inhibit the hepatic metabolism of some cancer chemotherapy drugs. For patients who do not have seizures at the time of diagnosis of brain metastases, the current (limited) information indicates that prophylactic anticonvulsants do not reduce the incidence of subsequent seizures (65).

Surgical resection. The role of surgery in the treatment of brain metastases has been well established and can be useful for histologic confirmation, genetic profiling, cerebral decompression, reducing symptomatic mass effect/edema, and prolonging survival.

Three prospective randomized trials assessed the value of aggressive local management through surgical removal of single brain metastases (128; 169; 116). In each study, patients with a single brain metastasis were treated either with complete surgical resection plus whole brain radiation therapy or with whole brain radiation therapy alone. In the study by Patchell and colleagues, patients in the surgery group had significantly longer overall survival, longer time to recurrence of brain metastases, longer duration of functional independence, and lower risk of death due to neurologic causes (128). The results of the other two studies were less decisive, but taken together, these studies have generally been interpreted to show that surgical resection is beneficial in select patients.

The value of surgery in the management of patients with multiple metastases is more controversial. In some patients with multiple metastases, there is one lesion that is large or is in a life-threatening location (eg, in the posterior fossa). For these patients, resection of one lesion can improve neurologic function and "buy time" for therapy of the other metastases. Another approach is to resect more than one brain metastasis, either as a single operation or as a stage procedure. Retrospective studies suggest that the postoperative morbidity and mortality rates for patients who undergo resection of multiple metastases are relatively low and are comparable with that reported in patients with single surgically resected metastases (15; 174). The impact of resection of multiple metastases on patients' neurologic function, quality of life, tumor control, and survival is not clear. It is not known how this treatment approach compares to "standard" whole brain radiation therapy (WBRT) or to stereotactic radiosurgery.

For most patients with brain metastases, surgical resection is not an option because of the presence of multiple lesions, extensive systemic disease and associated medical complications, or the involvement of deep-seated eloquent areas of the brain. Furthermore, surgery alone has been shown to be insufficient in achieving long-term local control (110).

Radiation therapy: whole brain radiation. Historically, whole brain radiation therapy was the standard treatment for patients with brain metastases and remains quite common. The use of whole brain radiation therapy is due in large part to a phase III study that showed whole brain radiation therapy following surgical resection reduced the rate of local and distant metastatic recurrence compared to postoperative observation alone (127).

Whole brain radiation therapy is usually given as 3000 to 4000 cGy in 10 to 15 daily fractions. Several randomized studies of altered dose/fractionation regimens have not shown any significant advantage in terms of radiographic response rates, local tumor control rates, or patient outcomes compared with "standard" 3000 cGy in 10 fractions or a biologically equivalent dose. There are surprisingly few good studies looking at the radiographic response of brain metastases to whole brain radiation therapy. Among solid tumors, metastases from breast and small cell lung carcinoma show the highest rate of radiographic response, whereas metastases from melanoma and renal carcinoma may stabilize but do not often shrink after whole brain radiation therapy (125). Other solid tumors show an intermediate radioresponsiveness. For any tumor histology, smaller metastases are more likely to regress after whole brain radiation therapy than larger lesions.

Although whole brain radiation therapy is still preferred in patients with numerous brain metastases, leptomeningeal metastasis, or primary histologies prone to micrometastatic disease such as small-cell lung cancer, treatment paradigms are changing as late delayed radiation induced neurotoxicity is being increasingly recognized. Whole brain radiation therapy may cause clinically significant diffuse brain injury. At least one half of patients with brain metastases who survive more than one year after 3000 to 4000 cGy of whole brain radiation therapy develop changes on serial CT or MRI scans, including diffuse cerebral atrophy, ventricular enlargement, and hyperintense T2-weighted or FLAIR signal abnormalities in the hemispheric white matter (45; 147; 120). A small proportion of long-term survivors with brain metastases develop progressive dementia, psychomotor retardation, gait disturbance, and urinary incontinence, appearing 6 to 18 months after whole brain radiation therapy (45; 48). The incidence of "radiation therapy dementia" is not clearly known; a disproportionately high percentage of reported patients received daily dose fractions greater than 300 cGy. The clinical similarity of this syndrome to normal pressure hydrocephalus has led some patients to undergo ventriculoperitoneal cerebrospinal fluid shunting, which can produce partial improvement in some patients. To date, however, there is no reliable way to predict which patients would benefit from a shunt.

Current data suggest that inadequate control or recurrence of brain metastases is a more common cause of deteriorating neurocognitive function than is the toxicity of whole brain radiation, and whole brain radiation remains an important part of treatment for some patients (97). As a result, strategies to minimize long-term neurocognitive deficits include the use of intensity-modulated radiation therapy to deliver "whole-brain" treatment while sparing the hippocampi (68), as well as neuroprotective agents such as memantine given during the course of whole brain radiation therapy and for a total of six months (23). The contemporary NRG cooperative group CC001 randomized phase 3 trial combined both approaches with additive benefit in preventing neurocognitive deterioration (21). In more recent years, radiotherapy has also moved toward more localized techniques such as stereotactic radiosurgery.

Stereotactic radiosurgery. Stereotactic radiosurgery as a monotherapy has become increasingly popular in recent years for patients with limited brain disease as a way to largely spare normal brain. The technique uses a linear accelerator or multiple Cobalt-60 sources ("gamma knife") to deliver a highly focused single dose of radiation (usually 1000 to 2000 cGy) to a circumscribed target. Brain metastases are theoretically well suited for radiosurgery because they are usually roughly spherical and well circumscribed.

Radiosurgery has been administered to a large number of patients in a number of settings: as primary treatment of single metastases instead of surgical resection; as the sole treatment of newly diagnosed single or multiple brain metastases; in combination with whole brain radiation therapy for newly diagnosed metastases; and as treatment of recurrent brain metastases. Numerous retrospective series of stereotactic radiosurgery have reported "local control" rates as high as 75% to 95%, ie, radiographic shrinkage or stabilization of treated lesion(s) for the lifetime of the patient (07; 146; 100). Tumors such as melanoma and renal carcinoma, which are considered "radioresistant," have a better local control rate from radiosurgery than from whole brain radiation (94; 30; 99). Larger metastases generally have lower radiographic control rates than smaller lesions, which is partly related to the ability to safely administer a higher tumor margin dose for metastases smaller than 2cm (29; 30). For patients with multiple metastases, the local control rate is probably related more to the total volume of treated tumor than to the number of metastases (99; 100). In most centers the maximum tumor size for single-fraction radiosurgery is 3 cm. Large retrospective cohort studies have demonstrated excellent local control for tumors less than or equal to 2 cm treated with 2400 cGy single-296 fraction stereotactic radiosurgery alone. However, metastases of 2 cm or greater treated with single-fraction stereotactic radiosurgery doses of 1500 to 1800 cGy 297 have been associated with poor local control and multifractionated stereotactic radiosurgery is recommended for tumors 3 cm or greater (40).

Several studies have compared stereotactic radiosurgery alone versus stereotactic radiosurgery and whole brain radiation therapy in patients with oligometastatic disease (up to three brain metastases) and showed no significant difference in overall survival with the omission of whole brain radiation therapy (03; 04; 92). Although local and distant control improved with the combination, neurocognitive function is retained with stereotactic radiosurgery monotherapy (31; 22). As a result, it is recommended whole brain radiation therapy not be routinely added to stereotactic radiosurgery for the treatment of limited numbers of brain metastases.

The low risk of complications associated with stereotactic radiosurgery and the ability to perform stereotactic radiosurgery on an outpatient basis offers potential cost savings and safety advantages over conventional surgical resection for the treatment of cerebral metastases. Most comparisons of patients with a single, newly diagnosed surgically resectable brain metastasis showed that stereotactic radiosurgery produced local tumor control rates generally equivalent to those seen after surgical resection (07; 134; 135; 122). In one retrospective study, surgical resection yielded better local control and survival than radiosurgery (14). Surgical resection does offer the advantage of immediate debulking and relief of symptoms due to mass effect, and it can deal with metastases too large to be suitable for radiosurgery.

An increasingly utilized approach is a resection of the single brain metastasis and then give stereotactic radiosurgery to the margins of the resection cavity. In several series, this combined treatment resulted in 1-year local control rates of 70% to 90% (85; 38; 110). Local control rates after single-fraction radiosurgery are lower if the resection cavity is large or its margins are indistinct. Hypofractionated radiosurgery can be used for postoperative treatment of a large (eg, greater than 3 cm) resection cavity; however, specific patient selection criteria and treatment dosimetry are not well established.

Several studies have also looked at the efficacy of stereotactic radiosurgery for patients with more than four brain metastases and suggest that stereotactic radiosurgery alone may be a reasonable approach for patients with multiple metastases. A large prospective observational study by Yamamoto and colleagues for patients undergoing stereotactic radiosurgery for 1 to 10 metastases found similar survival outcomes for patients with two to four versus five to 10 metastases (177), and the aggregate volume of lesions being treated may be more important in predicting the efficacy of stereotactic radiosurgery (101). A phase 3 randomized control trial comparing stereotactic radiosurgery versus whole brain radiation in patients with five to 15 intact brain metastases was completed and found in nonmelanoma patients with four to 15 metastases, and stereotactic radiosurgery was associated with reduced cognitive deterioration compared to whole brain radiation without compromising OS (98). However, the recommendation for stereotactic radiosurgery for patients with brain metastases is determined not only by the number of brain metastases but by other patient-related factors. These include tumor size, location, brain metastasis velocity, histology, age, functional status, extent of extra cranial disease, molecular profile, systemic treatment options, and overall prognosis.

The acute toxicity of radiosurgery includes transient worsening of neurologic deficits and increased peritumoral edema, occurring in approximately 5% of patients within several days after treatment (146; 132). The symptoms improve with corticosteroids. Five percent to 10% of patients develop symptomatic or asymptomatic radiographic worsening (increased contrast enhancement and increased edema) 3 to 6 months after radiosurgery. Most patients subsequently show clinical and radiographic improvement over the subsequent few months. Intratumoral hemorrhage may occur following radiosurgery, particularly with melanoma (30; 99). Delayed symptomatic focal brain necrosis occurs in 2% to 5% of treated patients (146; 131). The incidence of symptomatic radiation necrosis may be higher after treatment of larger versus smaller metastases (16). Symptomatic cyst formation occasionally occurs in long-term survivors after radiosurgery. In a long-term follow-up study of patients who survived at least 3 years after radiosurgery for brain metastases, solid mass lesions or cysts developed at the treatment site with an actuarial incidence of 4% at five years and 21% at 10 years (176).

The practical difficulty in the patients treated with stereotactic radiosurgery is distinguishing radiation injury from tumor progression or from the common coexistence of necrosis and viable tumor using standard diagnostic MRI scans. Fluorodeoxyglucose-PET scanning (171; 81), MR spectroscopy, or perfusion MRI (78; 118) have fairly good reliability, particularly in cases of “pure” necrosis or “pure” tumor, but all have false positives and false negatives.

Chemotherapy. Cytotoxic chemotherapy has generally not been highly successful in treating brain metastases due to several factors: (1) several common sources of brain metastases are not sensitive to currently available agents; (2) many patients who develop brain metastases have already "failed" one or more chemotherapy regimens; (3) subclones of tumor cells that metastasize may be inherently less chemosensitive than the cells in the primary tumor; and (4) the blood-brain barrier may limit penetration of water-soluble agents and macromolecules into brain metastases. These difficulties were highlighted in a study of patients with newly diagnosed small cell lung cancer, considered a "chemosensitive" tumor: 73% of patients had a systemic response to initial chemotherapy, but only 29% of patients with asymptomatic brain metastases showed a neuroimaging response to the same chemotherapy (144). Similarly, for patients with brain metastases from melanoma, conventional systemic chemotherapeutic options have shown very limited efficacy with radiographic response rates of 0% to 10% and tumor stabilization rates of 10% to 30% (93; 06).

Despite these serious limitations, chemotherapy may have a role in the treatment of selected patients. Some patients with chemosensitive tumors (eg, breast, small cell lung cancer, germ cell tumors) respond to systemic chemotherapy, either as initial therapy for previously untreated brain metastases (57; 33; 73) or for recurrence of brain metastases after prior whole brain radiation therapy (18; 54). For patients with breast cancer, the choice of systemic therapy is often driven by the subtype: with brain metastases from triple negative breast cancer, capecitabine has shown benefit (54), and for estrogen or progesterone receptor positive breast cancer, hormonal therapies are primarily utilized (141; 162; 53). For most patients with germ cell tumors, chemotherapy without cranial irradiation is effective in treating brain metastases, but a small minority of patients have isolated relapse of brain metastases despite eradication of systemic tumor (09). Carboplatin plus pemetrexed has shown modest activity as first-line therapy for brain metastases from lung adenocarcinoma, as well as combinations of vinorelbine, cisplatin, and paclitaxel (137; 42; 11; 47).

Molecular targeted agents. As targeted agents are becoming an important part of standard initial therapy for the management of systemic cancer, the utility of these agents for the treatment of brain metastases has also become increasingly evident.

Several targeted agents have become first-line standard of care for patients with non-small cell lung cancer (NSCLC) harboring epidermal growth factor receptor (EGFR), anaplastic lymphoma kinase (ALK), or ROS-1 mutations and have shown significant benefit in CNS disease management. Specifically in epidermal growth factor receptor-mutant non-small cell lung cancer brain metastases, first generation tyrosine kinase inhibitors (TKIs), such as gefitinib and erlotinib, have demonstrated response rates of 50% to 80%, overall survival of 15 to 22 months, and increased efficacy with pulsatile high dose administration (28; 82; 148; 89; 74; 172; 83). In one nonrandomized retrospective series, patients with advanced-stage epidermal growth factor receptor-mutant non-small cell lung cancer treated with gefitinib or erlotinib had a lower risk of developing brain metastases than similar patients treated with cytotoxic chemotherapy (76). More recently due to improved CNS penetration, durable response, and efficacy in patients with progression after treatment with first generation tyrosine kinase inhibitors, the second-generation tyrosine kinase inhibitor osimertinib is now considered first-line therapy for patients with EGFR-mutant NSCLC, particularly for those with brain metastases. It can also be considered for the primary prevention of brain metastases compared with first-generation tyrosine kinase inhibitors (119; 154). For anaplastic lymphoma kinase (ALK)-positive, anaplastic lymphoma kinase inhibition is a viable option. The first-generation anaplastic lymphoma kinase inhibitor crizotinib has limited blood-brain barrier penetration. For patients with brain metastases, second and third-generation anaplastic lymphoma kinase inhibitors, such as alectinib, brigatinib, ceritinib, and lorlatinib, have shown superior activity even in an anaplastic lymphoma kinase-inhibitor-pretreated patient, with response rates up to 81%, and they are preferred (152; 153; 27; 60). For patients with ROS1 translocations, entrectininb has shown CNS activity in crizotinib-naïve disease with overall response rates of 79% (52). For RET fusion-positive NSCLC, selpercatinib has shown benefit in patients with brain metastases (50; 163).

For patients with breast cancer-related brain metastases, treatment is dependent on the presence of HER-2 and/or hormone receptor expression. The occurrence of isolated brain metastases in patients with HER-2-expressing breast cancer treated with the monoclonal antibody trastuzumab is at least partly due to the inability of trastuzumab to cross the blood-brain barrier. Small molecule inhibitors of the HER-2 tyrosine kinase, such as lapatinib, have shown modest activity against breast cancer brain metastases (104; 164). Cyclin-dependent kinase (CDK) inhibitors specifically targeting CDK4/6 have become an important part of treatment for hormone receptor-positive breast cancer. Abemaciclib, with the highest CNS penetration, has shown promise in the treatment of brain metastases with intracranial benefit of 24% (166).

CNS penetrant chemotherapy-targeted drug combinations and drug conjugates are increasingly showing efficacy in the management of breast cancer-related brain metastases in patients with breast cancer and brain metastases. A phase 2 study of lapatinib plus capecitabine (LANDSCAPE) for previously untreated brain metastases from breast cancer showed a 66% radiographic response rate (10). Although neratinib alone demonstrated limited efficacy, the combination with capecitabine yielded a 49% response rate (58; 59). Similar results have been seen with TDM1, tucatinib/trastuzumab/capecitabine (HER2CLIMB), fam-trastuzumab deruxtecan (DEBBRAH), and sacituzumab govitecan for triple negative breast cancer (ASCENT) (102; 121; 12; 130). In the phase 3 HER2CLIMB study, for patients with HER2-positive breast cancer with brain metastases treated with trastuzumab/capecitabine and the addition of tucantinib, intracranial overall response rates doubled, the risk of intracranial progression or death was reduced by 68%, and the risk of death was reduced by 42% when compared to trastuzumab and capecitabine alone (102; 121).

Targeted therapies have also altered the treatment paradigm for metastatic melanoma. Vemurafenib is a signal transduction inhibitor that acts on BRAF kinase and has shown effectiveness in some patients with brain metastases from melanoma (114). Dabrafenib is another BRAF inhibitor that produced an objective radiographic response in one third of melanoma brain metastases in patients whose tumors had the Val600Glu BRAF mutation (107). The combination of dabrafenib and the mitogen-activated protein kinase (MAPK) inhibitor trametinib has yielded improved intracranial response rates, as have combinations of vemurafenib and cobimetinib or encorafenib and binimetinib. Intracranial overall response rates with these agents ranges between 33% to 43%, intracranial disease control rate between 60 and 79%, and mOS between 9.5 and 11.2 months (64; 49; 79). However, duration of response remains a problem (43). Combination with immunotherapy has not shown improved duration of response, and a phase 2 study of atezolizumab, vemurafenib, and cobimetinib (TRILOCET) resulted in an intracranial response rate of 42%, but PFS was five months, and duration of response was seven months (51). In addition, combination of these targeted therapies with stereotactic radiosurgery may have additional benefit (108).

In patients with renal cell carcinoma, the VEGFR tyrosine kinase inhibitor cabozantinib has shown early signs of efficacy in patients with brain metastases with a reported intracranial response rate of 61% (77). There are also reports of radiographic response of brain metastases from renal carcinoma to the oral multikinase inhibitors sorafenib (168) or sunitinib (70). However, only one prospective study, a phase 2 trial of sunitinib in patients with untreated brain metastases, has examined the CNS efficacy of VEGFR tyrosine kinase inhibitors in patients with mRCC. Among 16 evaluable patients, the CNS response rate was 0% and the percentage of patients with stable disease was 31%. The median time to progression was 2.3 months and the median overall survival was 6.3 months (36). In one study of patients with renal carcinoma and brain metastases treated with stereotactic radiosurgery, those who were also treated with a tyrosine-kinase inhibitor, mTOR inhibitor, or bevacizumab had a better rate of local control of the brain lesions compared to patients who did not receive targeted therapy (39).

Immunotherapy. To date, immunotherapy for brain metastases mainly applies to melanoma and non-small cell lung cancer. Immune checkpoint inhibitors targeting CTLA-4 (ipilimumab), PD-1 (nivolumab, ipilimumab), and PD-L1 (atezolizumab), all of which act to augment T-lymphocyte activation, have shown benefit in the treatment of melanoma and non-small cell lung cancer-related brain metastases (109). Disease control rates of melanoma brain metastases with single agent ipilimumab or pembrolizumab have ranged from 10% to 22% with sustained response, and combination therapy has shown even higher response rates near 60% with a marked increase in survival (111; 67; 106; 165). In patients with non-small cell lung cancer treated with single-agent pembrolizumab, intracranial response rates reached 33% and were durable in most patients, with CNS PFS of 10.7 months and mOS of 8.9 months (67; 66). Single agent atezolizumab in patients with asymptomatic treated non-small cell lung cancer brain metastases was associated with improved survival when compared to traditional chemotherapy (61). It is not yet clear how ipilimumab should be integrated into the overall treatment strategy of patients with brain metastases, and exploration of combined regimens of radiation therapy and immunotherapy have been pursued and are suggestive of a synergistic benefit (02; 123). Alternative approaches include treatment with autologous tumor-infiltrating lymphocytes engineered to express a T-lymphocyte receptor that recognizes a melanocytic differentiation antigen has also caused regression of brain metastases in some patients with melanoma brain metastases (80). Studies are ongoing to better explore and better understand the optimal combinations, timing of treatment, and mechanisms involved in immune activation.

Recurrent brain metastases. A difficult and frequently encountered clinical problem is the treatment of recurrent brain metastases. The reappearance of brain metastases is often complicated by the fact that many of these patients also have extensive systemic disease. In general, the same types of treatment used for newly diagnosed brain metastases are also available for recurrent tumors. However, the type of previous therapy often limits the therapeutic options available at recurrence.

Surgical resection of recurrent brain metastases is an option for selected patients with a single site of brain recurrence, limited systemic disease, and a fairly good performance status. This includes patients who received prior resection, whole brain radiation, or radiosurgery (167; 86).

For patients with recurrence of brain metastases after whole brain radiation therapy, the amount of additional radiation therapy that can be given safely is usually 1500 to 2500 cGy. Repeat courses of whole brain radiation therapy can improve or stabilize neurologic function in a significant proportion of patients (173; 140). Reirradiation may be somewhat more beneficial in the subset of patients who showed an initial favorable response to radiation therapy and who remain in good general condition when the cerebral recurrence develops (41). Systemic tumor status is the main determinant of survival for these patients.

Several series of patients with recurrent brain metastases following whole-brain radiation and subsequent radiosurgery have shown high rates of local control over the patients' lifetimes (34; 32; 25). Longer overall survival is correlated with good performance status, "controlled" primary tumor, a single brain metastasis versus multiple metastases, and the total volume of brain lesions treated with radiosurgery (25). Despite control of the treated lesions, patients may die of leptomeningeal dissemination or "polyfocal intraparenchymal" metastases (34; 131).

Novel approaches: laser interstitial therapy (LITT). Stereotactic laser ablation, or laser interstitial therapy, is one of the newer and innovative techniques that has emerged for recurrent brain metastases and radiation necrosis. It involves a small burr hole through which a biopsy is simultaneously obtained and treatment is delivered through using a laser probe. Although there are limitations on lesional size and posttreatment related edema, this technique allows for treatment of deep-seated lesions that may not be accessible for resection, as well previously irradiated areas. Small prospective studies have shown some benefit, but more rigorous prospective randomized studies are still needed (01; 136; 145).

Follow-up. There is no set standard for the follow-up of brain metastases after treatment. Clearly, MRI or CT scans are indicated at any time after therapy where patients develop new neurologic symptoms. How frequently asymptomatic patients need follow-up scans is controversial. For patients treated with surgery, a contrast MRI scan should be performed within five days after surgery to detect residual disease. This is especially important if consideration is being made to forgo postoperative whole brain radiation therapy. A reasonable schedule of follow-up scanning would be to get a scan 2 to 3 months after completion of last therapy (either whole brain radiation therapy, surgery, or radiosurgery), and then about every 2 to 3 months for the first year following treatment. The length of time between scans can then be gradually stretched out so that asymptomatic patients are scanned only once per year.

Outcomes

Median survival for patients with brain metastases varies widely for patients with non-small cell lung, breast, melanoma, gastrointestinal, and renal cancers (see Prognosis and complications). Outcomes are largely dependent on a number of variables that include age, functional status, extent of systemic and CNS disease, CNS disease velocity, and molecular profiling. Improved radiation techniques and developments in targeted therapies and immunotherapies for patients with non-small cell lung cancer, breast cancer, and melanoma have dramatically transformed the treatment and survival of patients with brain metastases. As patients live longer with CNS metastatic disease, however, in addition to side effects such radiation necrosis discussed earlier, cognitive impairment as a direct effect of metastatic disease as well as an effect of therapy remains a challenge. Up to 30% of patients with cancer exhibit cognitive impairment prior to treatment, 75% might have measurable cognitive impairment during treatment, and 35% of cancer survivors will continue to exhibit cognitive difficulties in the months to years that follow treatment (84).