The intracranial compartment is a common site of metastatic cancer. Of the patients projected to die of systemic cancer approximately 25 percent, or over are expected to have intracranial metastasis. The importance of intracranial metastasis is, however, not primarily due to its frequent occurrence but to the fact that the large majority of the new foci become symptomatic. Compared with other organs, such as the lung and liver, in which the incidence of metastasis is even higher, the manifestations of metastases affecting the brain are usually more overt and disabling and if untreated, tend to be rapidly lethal. For these reasons brain metastasis demands prompt diagnostic and therapeutic attention.
A sense of frustration is justifiably inherent in the treatment of patients suffering from disseminated cancer. With few exceptions, even the eradication of a presumed solitary metastasis is followed sooner or later by the discovery of metastases elsewhere or at the primary site. Nevertheless, treatment of metastases that are associated with high morbidity and mortality rates, such as those within the brain, can be, and often is, rewarded by meaningful palliation. The best results are not achieved by rigidly standardized methods of therapy. Rather, the task of the oncologist is to select the most rational treatment based on criteria such as general health, histology of the primary cancer, extent of the disease, and expected or proven response of that particular neoplasm to various modes of therapy. In some instances, the most appropriate course of action may consist of corticosteroid therapy alone or even no treatment at all, whereas in others, radical multidisciplinary therapy, including surgery directed at the brain metastasis or systemic metastases and primary tumor, may prove most beneficial.
Incidence and Classification
Autopsy studies of large numbers of cancer patients dying in the hospital, although imperfect, provide useful guides to the frequency and distribution within the cranium of metastases from various primary neoplasms. The overall frequency of intracranial metastases reported from autopsy series carried out over the past several decades ranges from 12 to 35 percent. These studies differ in respect to the population and time period surveyed and, undoubtedly, the thoroughness of examination of the central nervous system. Series reporting lower figures generally cover earlier time periods and therefore do not reflect the steep rise in deaths from lung cancer, which have doubled in the past 30 years. They also tend not to include leukemia, lymphoma, or dural or pituitary metastases. Taking such factors into consideration, one may reasonably assume that 25 to 30 percent of cancer patients now develop intracranial metastases in the course of their disease. Two large series both report a frequency of about 25 percent.
The propensity of tumors of different primary origin to metastasize to the cranial contents, as well as to the various intracranial compartments, differs widely. Among other factors, ready access to the arterial circulation of the head and an environment necessary to sustain the growth of tumor emboli are of importance. For example, primary tumors of the lung and tumors that commonly metastasize to the lung early in the disease, such as breast carcinoma and melanoma, have a very high incidence and wide distribution of intracranial metastases. Lymphoma metastasizes almost exclusively to the leptomeninges; and prostatic cancer, which has a much lower frequency of intracranial metastases, has a distinct predilection for the skull and dura. Intracranial prostate and breast metastases may be dural-based, mimicking the growth pattern of meningiomas. Primary tumors originating in the pelvis, such as prostate, bladder, and uterine cancer, frequently metastasize to the cerebellum, as do gastrointestinal tumors.
Individual intracranial metastases are conveniently classified as to their probable site of origin within the skull, dura, leptomeninges, or brain. Although the opposing surface of the dura is commonly secondarily invaded by tumors that begin in the skull and not infrequently by those that metastasize to the parenchyma, this structure generally acts as a barrier to further invasion. Occasionally a single tumor (usually originating in the skull or dura) involves all four structures.
Skull and Dura
Although surgically of less importance than parenchymal tumors, those metastatic to the skull or dura occasionally reach considerable size within the intracranial space and may warrant excision. Those located at the vertex or in the low occiput may produce neurological dysfunction by compression of the sagittal or lateral sinuses, and those at the skull base may do so by compression of cranial nerves. Skull or dural metastases are common in the following malignancies: prostate carcinoma, lymphoma, breast carcinoma, melanoma, neuroblastoma, and osteogenic sarcoma.
Although infrequently included in autopsy series, metastases to the pituitary are not rare. In the series of Takakura et a!., metastases to the pituitary were present in 6 percent of all patients with cancer and in 20 percent of patients dying with breast cancer.
Metastases to the leptomeninges and spread within the cerebrospinal pathways (neoplastic meningitis), virtually all malignant neoplasms have been reported to have produced this entity, which can result in an extremely variable constellation of neurological symptoms and signs by invasion of the brain, cranial nerves, spinal cord, and spinal nerve roots. Obliteration of the subarachnoid spaces with consequent hydrocephalus is more prevalent in patients with carcinomatous, as opposed to leukemic, meningitis; perhaps this reflects the better response to treatment of the latter. Neoplastic meningitis is most common in patients who have leukemia especially the acute lymphocytic variety, non-Hodgkin's lymphoma, and breast carcinoma; there is a significant incidence in patients with lung cancer and melanoma. Timely diagnosis requires a high degree of suspicion regarding patients with these cancers and familiarity with the often subtle symptoms and signs that appear early in the disease.
Because the neurosurgeon may be called on to provide longterm access to the cerebrospinal fluid for delivery of intrathecal chemotherapy, it is important to recognize that these patients often have an abnormally small ventricular system as a result of diffuse cerebral swelling. With few exceptions, safe, accurate placement of an indwelling ventricular cannula requires preoperative computed tomography (CT); or MRI.
Approximately 16 to 18 percent of cancer patients develop brain metastases, and in about 9 percent, intracranial metastases represent the only site of cancer. To get to the central nervous system, metastatic cells leave the primary tumor by local invasion, enter the circulation, circulate in the blood, adhere to brain microvessels. penetrate the blood-brain endothelial barrier, and multiply in the brain. Most often the metastatic cells are trapped at the site of acute arterial narrowing near the brain surface and grow in a soft, yielding matrix without significant tissue planes. Thus metastases to the brain tend to be peripherally located (at the gray-white matter junction) and roughly spherical. Their distribution among the cerebrum, cerebellum, and brain stem corresponds to the relative weight of the subdivision. Overall, 45 to 50 percent of solid parenchymal metastases, as determined by current diagnostic techniques, are single; of these, roughly 80 percent are located in the cerebrum, 16 percent in the cerebellum, and 3 percent in the brain stem. However, the presence of a truly solitary metastasis in the brain, as borne out by long follow-up after total removal, is rare. Most frequently reported solitary lesions originate from renal cell carcinomas; but others, including metastases from lung and colon carcinomas, certainly also occur.
As is the case with metastases to other sites, the interval between the diagnosis of the primary cancer and the diagnosis of brain metastases varies with the tumor type. The median interval in lung cancer is notably short. but in breast cancer it is frequently protracted. The range for each type of cancer, however, varies considerably. Brain metastases may be synchronous. that is, present at the time of the diagnosis of the primary tumor, or conversely, may occur more than a decade later (metachronous). Histologic verification is necessary to distinguish between metastases from a neoplasm believed to be long-cured and a new primary tumor. Multiple primary tumors occur in about 15 percent of patients with cancer.
Because of their prevalence or unusual frequency, brain metastases originating from cancer of the lung, breast, colon, and kidney and from melanoma are of special interest. Carcinoma of the lung is presently responsible for approximately 45 percent of all intracranial metastases. Of the histologic varieties of lung cancer, squamous cell carcinoma is less likely to metastasize to the brain than is adenocarcinoma or undifferentiated carcinoma. The frequency of brain metastases in patients with small cell carcinoma varies from 49 percent in older publications to 10 to 20 percent in more recently published studies. Approximately 45 percent of non-small cell brain metastases are single, and 35 percent are synchronous. Breast carcinoma, the second most common source of metastatic brain tumor, is also the most widely distributed throughout the intracranial contents and is apt to involve several compartments simultaneously. Fortunately, carcinoma of the breast is more chemo- and radiosensitive than are most other tumors that metastasize to the brain. Both colon and kidney metastases have a marked predilection for the brain as compared with other intracranial compartments, tend to be single and are radioresistant. Of all malignant tumors, melanoma has the highest incidence of metastasis to the brain and the greatest tendency to bleed spontaneously. The lesions are usually multiple and commonly occur in large numbers.
The neurological symptoms and signs of metastatic brain tumors are indistinguishable from those of many other expanding intracranial mass lesions, and without the presence or history of cancer, metastatic brain tumors cannot be diagnosed on clinical grounds alone. Most metastatic brain tumors are subcortical in location, grow rapidly, and, even when relatively small, produce extensive oedema. Neurological deterioration commonly proceeds at a rapid pace and can be measured in terms of days or weeks. In most patients it is the spread of oedema through the white matter, not the increase in size of the tumor per se, that accounts for the relatively rapid onset and progression of symptoms and signs and the fact that the latter are generally of limited value in localizing the site of the metastatic neoplasm with precision. An abrupt stroke-like onset of neurological deficit occurs in about 10 percent of patients and may result from tumor haemorrhage or compromise of local blood supply by the neoplasm.
As might be expected, symptoms of increased intracranial pressure are common in patients with metastatic brain tumors. Headache is the initial complaint in 50 to 60 percent of patients. Decreases in cognitive function and nausea and vomiting are less frequent. Papilloedema is observed in about 10 percent of patients. Raised intracranial pressure is usually caused by cerebral oedema, but it also may result from ventricular obstruction secondary to cerebellar and brain stem metastases, obstruction of the venous sinuses by neoplasm. and, occasionally, concomitant carcinomatous meningitis. Focal weakness is the presenting symptom in about 40 percent of patients. but it is apparent on the initial examination about 60 percent of the time. Ataxia is the first symptom noted in 20 percent of patients, and seizures (predominantly focal) in approximately 15 to 25 percent. Seizures are predictably more common in patients who have multiple brain metastases, especially from melanoma.
Most patients with metastasis to the pituitary gland are asymptomatic. When symptoms do occur they are fairly constant and include a characteristic triad of headache. visual disturbances, and diabetes insipidus. The clinical features reflect the fact that most (33 to 100 percent) pituitary metastases are found in the posterior lobe of the gland or within the pituitary stalk. The predilection for the posterior lobe is simply explained by the more abundant and direct blood supply to this structure from the inferior hypophyseal arteries.
Primary Brain Tumor
Most cancer patients who develop a full-blown syndrome of an expanding intracranial mass lesion have one or more metastatic brain tumors. In some, a second primary neoplasm is found. Meningioma is the most common primary intracranial tumor encountered in this setting and may occur in association with breast carcinoma. Presumably because of the high blood flow to these tumors, meningiomas may themselves be the recipient of metastases. Malignant gliomas are occasionally present in patients with systemic cancer, but it is uncertain whether the association is mere chance.
Brain abscesses are quite uncommon in cancer patients but occur in them with a greater frequency than in the general population. Patients who are immunosuppressed as a result of either their disease (e.g., lymphoma) or its treatment are prone to infection by opportunistic organisms. Abscesses caused by a fungus or Toxoplasma are not uncommon. Surgically significant bacterial abscesses may occur in patients in whom a communication between the intracranial space and the body surface develops following radical surgery and/or radiation therapy for malignant tumors near the skull base. They may also originate from septic emboli in patients who have lung abscesses secondary to bronchial obstruction by primary or metastatic tumors.
Infarction and Hemorrhage
Cerebrovascular lesions are common in patients with cancer; they are found at autopsy in about 15 percent of cases. Hemorrhage and infarction occur with equal frequency, and about onehalf of each are symptomatic. Both entities, at times, produce clinical symptoms and radiologic findings that may be confused with those of metastatic tumors. In addition to their potential for creating diagnostic problems and the fact that they most frequently occur during the end stages of disease, several cerebrovascular lesions, especially intracerebral and subdural hematomas, are of surgical importance. Intracerebral hematoma is encountered in about 4 to 5 percent of cancer patients at autopsy, and subdural hematoma in 1 to 2 percent. The most common cause of haemorrhage is a coagulopathy, usually thrombocytopenia. In some patients, a transient thrombocytopenia may be responsible for a chronic subdural hematoma. In most instances, acute, coagulopathy-induced haemorrhage into the subdural space or brain is massive and attempts at salvage are futile.
Intra- or peritumoral haemorrhage, usually spontaneous, is responsible for about 25 percent of intracerebral haemorrhages. The majority are associated with melanoma and choriocarcinoma but can be produced by any metastatic tumor. Bleeding usually emanates from relatively small vessels and dissects along fiber tracts in the centrum, displacing rather than destroying significant amounts of neural tissue. The neurological deficit after surgical removal is, therefore, often insignificant even with very large clots. In a neurologically deteriorating patient, the most effective treatment is prompt, thorough removal of the clot and tumor. This not only may be lifesaving but, in patients with no other intracranial disease and limited or treatable disease elsewhere, can result in prolonged and meaningful survival. Since delay in the removal of large clots usually results in death or a profound deficit in such patients, the frequent practice of "watchful waiting" should be condemned.
The presence of brain metastases is critical information in staging the cancer patient and planning treatment. Under most circumstances, the presence of brain metastases is synonymous with poor prognosis, and conservative or palliative treatment is administered. However, the patient with a solitary brain metastasis may derive benefit, in both quality of life and survival, from an aggressive approach consisting of surgical resection or stereotactic irradiation of the lesion. Therefore, it is essential that such a lesion be detected at the earliest-possible stage. Current CT and magnetic resonance imaging (MRI) scanners are capable of detecting tumors less than 5 mm in diameter and provide a simple method for early, accurate diagnosis of metastatic brain tumors and meaningful follow-up after treatment.
MRI is currently the diagnostic test of choice for brain metastases. Contrast (gadolinium) enhanced T1-weighted images are preferred for the diagnosis of all types of intracranial tumors. They have been found to be more sensitive than double-dose contrastenhanced CT images in detecting metastatic deposits, particularly those located in the posterior fossa. Besides the inherently superior resolution of MRI, detection of metastatic lesions is facilitated by the fact that the tumor and surrounding oedema have similar intensities. This summation effect, seen best on T2-weighted images, causes an apparent increase in lesion size, thereby improving detectability. The contrast-enhanced MRI is also the preferred method of assessing the presence of leptomeningeal metastases. A characteristic nodular pattern of enhancement is often seen within the basal cisterns, sylvian fissures, and cortical sulci and along the tentorium. The usual dose of the contrast agent [gadolinium diethylenetriamine pentaacetic acid (Gd-DTPA)] is 0.1 mmol/kg. In some cases, a higher dose, double dose, or delayed scanning may improve tumor visualization.
If MRI is not readily available or there is a contraindication to its use, as in patients with a pacemaker or ferromagnetic implant, CT remains a highly useful tool for detecting cerebral metastases.
The typical parenchymal metastasis on both CT and MRI is a discrete, rounded mass with extensive surrounding oedema . Multiple lesions are seen in approximately 50 percent of patients. Most metastatic tumors are hypodense but can appear isodense or hyperdense before contrast administration. Hyperdensity or a nonenhanced scan is usually related to haemorrhage and is commonly seen with melanoma and choriocarcinoma. Acute haemorrhage within and surrounding a metastatic tumor may obscure the presence of the tumor nodule. The majority of metastatic lesions (90 percent) show enhancement following the administration of a contrast medium. Heterogeneous enhancement is common in large tumors (more than 2 cm in diameter) and is usually caused by central necrosis. Metastases that are largely cystic show ringlike enhancement with a clear center, a pattern that is also typical of an abscess. CT, unlike MRI, is able to visualize bone and thus show lytic, expanding calvarial or skull base lesions and intracranial extension of such a mass. Blastic or sclerotic lesions of the skull may be more subtle.
Diverse pathologic processes, such as neoplasia, radiation necrosis, abscess, haemorrhage, infarction, and other inflammatory processes, may share similar radiologic features on contrastenhanced CT and MRI and must be considered in the differential diagnosis. Especially in patients with breast cancer, if radiotherapy is contemplated, it is wise, but not always possible, to exclude meningioma by cerebral angiography when a juxtadural tumor with dense, homogeneous uptake of contrast is discovered. In patients who have depressed cellular immunity and fungus or Toxoplasma abscesses, the only radiologic clue may be the atypical location of the intracranial mass (e.g., basal ganglia). Malignant gliomas, although usually demonstrating irregularity in shape and uptake of contrast, sometimes appear as single or multiple (10 to 15 percent) discrete, round lesions. If such patients do not have other evidence of cancer, a biopsy is usually necessary for a definitive diagnosis, although angiographic findings may sometimes be conclusive. Cerebral infarction may produce discrete areas of enhancement, which, however, resolve on repeat sequential scans.
Delayed brain necrosis as a sequela of therapeutic radiation may resemble a recurrent metastatic tumor by appearing as a discrete mass, which occasionally shows progressive enlargement on repeat scans.
In patients considered for surgery, preoperative localization scans have been found to be extremely useful for the surgeon in the accurate placement of scalp flaps, bone openings, and cortical incisions, and in the exposure of relatively minute subcortical lesions. We currently use MRI localization based on a grid localizer; the scalp is marked with a staple. The advantages of MRI localization over CT localization include accuracy, in that MRI avoids the error of parallax by its ability to localize the lesion from two perpendicular planes. MRI localization is done with the patient in the supine position, whereas CT localization in the coronal plane may require the prone position. The scalp can be marked after the procedure is completed without loss of scanning time. If CT localization is used, scans should be done in both axial and coronal planes to provide a two-dimensional view of the tumor. To plan the surgical approach, localization markers are positioned on the scalp over the point at which the tumor is most superficial.
In the immediate postoperative period, both CT and MRI are extremely useful in distinguishing between increasing oedema and postoperative clot. Contrast-enhanced MRI or CT in the first 4 postoperative days can be helpful in distinguishing residual tumor from postoperative changes. Varying degrees of enhancement at the margins of the tumor bed frequently occur in the first few weeks after surgery, presumably from breakdown of the bloodbrain barrier and neovascularization. Such abnormal ring enhancement may be difficult to differentiate from residual tumor without sequential scans.
CT and MRI have replaced cerebral angiography in the evaluation of metastases. However, cerebral angiography or magnetic resonance angiography (MRA) is occasionally used preoperatively to determine the magnitude of increased vascularity in such tumors as those of thyroid and kidney origin or to demonstrate the position of a tumor with respect to major blood vessels, such as those in the sylvian fissure. Angiography and MRA may also be useful in distinguishing hemorrhagic metastases from haemorrhages of other causes and in providing information about the patency of the major venous sinuses.
Positron emission tomography (PET) with fluorine-18 fluorodeoxyglucose has been found to be useful in neurooncology for its ability to allow distinction between high-grade and low-grade tumors, and between recurrent tumor and radionecrosis, chemonecrosis, or postsurgical change. However, PET does not distinguish between primary and secondary central nervous system neoplasms. Its main role and advantage may prove to be in assessing tumor response after treatment, especially following radiation therapy, to distinguish between tumor progression and radiation necrosis.
The options for treatment of metastatic intracranial neoplasms include corticosteroids, surgery, radiation therapy, chemotherapy, and stereotactic radiosurgery, either alone or in various combinations. At present, therapy in the vast majority of these patients is palliative because most have or will develop widely disseminated disease. There exists, however, a small but significant group of patients with no evidence of cancer elsewhere, perhaps 8 to 10 percent of the total, in whom eradication of intracranial disease carries the possibility of cure. For this reason, the effectiveness of various treatment modalities should be compared for their potential for eliminating specific intracranial lesions rather than on the basis of survival statistics alone.
Corticosteroids are unique in this armamentarium because they not only may effect significant palliation of neurological symptoms and signs when used alone, but also are of immense value when used in conjunction with other treatment modalities. The primary role of these compounds is in the reduction of tumor-induced oedema in the white matter, but by decreasing permeability of normal and oedematous brain, they also retard oedema formation resulting from surgical trauma, ionizing radiation, and chemotherapy. A direct oncolytic action of corticosteroids, although reported for some tumors, is rare. We have, however, observed shrinkage of metastatic lymphomas with corticosteroid therapy.
Dexamethasone, 10 mg initially and 4 mg q 6 h, or equivalent dosages of an analogue, usually results in noticeable clinical improvement within 12 hours in most patients. If practical, treatment with corticosteroids should begin 3 to 5 days before surgery (or other specific therapy) to achieve maximal clinical benefit. This not only ensures a significant reduction of oedema before operation and reduces oedema resulting from surgery but also may provide an indication of whether or not major neurological deficits are fixed or potentially remediable. The success of the surgeon or oncologist in avoiding increased deficit should be gauged in relation to the neurological condition of the patient following a maximum response to corticosteroid therapy and not on the basis of the presenting neurological signs.
Patients with symptomatic brain metastases who are preterminal or in whom specific therapy has failed often receive significant palliation from corticosteroid therapy. Even in the patient who obviously has only days or a few weeks to live, such treatment may bring welcome relief from headache or incapacitating neurological deficit. Although an increase in the median survival of 1 month for patients treated with corticosteroids alone is widely quoted, extension of a tolerable existence for many months is quite common, especially for patients in whom radiation therapy has failed to ablate the metastases. Some metastases from relatively radioresistant tumors, such as kidney and colon cancer and melanoma, may respond to radiation therapy by shrinkage and very occasionally by disappearance, but it is from the concomitant use of corticosteroids and not the radiation therapy that the majority of these patients benefit.
Much larger dosages of corticosteroids than those noted above may be necessary in some patients to achieve palliation or to control cerebral oedema following treatment. However, to minimize side effects such as myopathy, diabetes, and immunosuppression associated with the long-term use of corticosteroids, continued efforts should be made to reduce the dosage to the lowest level that prevents recurrence of major neurological symptoms. Prolonged corticosteroid dependency (more than 4 to 6 weeks) following surgery or radiation therapy usually indicates the presence of residual tumor and has been used as a criterion of treatment failure.
The primary role of surgery is largely confined to the treatment of patients with a single brain metastasis who do not have widespread or rapidly progressive cancer. This group, unfortunately, represents only about 20 to 25 percent of patients with parenchymal brain metastases; and ideal candidates for surgery constitute an even smaller percentage. Occasionally, patients with multiple brain metastases may benefit from surgery, for example, those with several radioresistant but surgically accessible tumors who are apparently free of disease elsewhere and those with potentially radiosensitive tumors in whom a single large tumor is life-threatening.
A secondary role of surgery in patients with intracranial metastases includes excision of some metastases to the skull or dura, biopsy of lesions that are clinically and radiologically obscure, removal of subdural and intracerebral hematomas, insertion of indwelling catheters and reservoirs for the delivery of intrathecal chemotherapy, and insertion of cerebrospinal fluid shunts for the treatment of hydrocephalus.
The majority of metastatic brain tumors are superficial in location, moderate in size, and relatively avascular and can be easily and cleanly separated from the surrounding brain by gentle dissection. For these reasons, the risk of increased neurological deficit as a result of extirpation is usually small. In contradistinction to a seemingly widely held belief, it is noticeable that no major differences in this regard in the removal of single lesions from the dominant hemisphere or cerebellum as opposed to those in the nondominant hemisphere. The risk of death within a 30-day period following craniotomy (i.e., standard operative mortality) in these patients is primarily a function of their neurological and general physical condition prior to operation rather than of complications directly related to the operation. Although it is certainly not desirable to restrict the use of surgery to patients who are at least risk, since many others may benefit from operation, analysis of the cause of death in patients undergoing the removal of a brain metastasis reveals that it would be entirely possible to reduce overall operative mortality to considerably under 5 percent by the selection of patients.
Two major factors influence survival in patients undergoing surgery and radiation therapy for single metastases. These are (1) the extent of systemic disease, perhaps the most important variable, because the major cause of death is progression of cancer outside the nervous system and (2) the patient's neurological condition prior to craniotomy, which at the extremes of the scale is predictably reflected in surgical mortality. The interval between the date of diagnosis of the primary neoplasm and that of the brain metastasis has no statistically significant impact on patient survival following the neurosurgical procedure. Any valid comparison between series or various modes of therapy of metastatic brain tumors must take these variables into account, in addition to those usually considered, such as age, sex, and histologic diagnosis.
Because cancer or its treatment commonly impairs function of many organs and systems, laboratory evaluation prior to surgery in patients undergoing craniotomy must be especially thorough. Aside from studies routinely used to determine the presence of metastases, such as chest roentgenograms and bone and liver - spleen scans, extensive cardiopulmonary evaluation may be required in patients who have received cardiotoxic and pneumotoxic chemotherapy, who have had pulmonary resection, or who have existing primary or metastatic disease of the lung. Elective surgery in patients undergoing chemotherapy must be timed so that the operation and early postoperative phase will not correspond to the nadir in the platelet and white blood cell counts. Although the timing of major depression in bone marrow function can be fairly well predicted for most chemotherapeutic agents, if possible, it is best to carry out the surgery after the nadir is past and a stable rising platelet count is documented by daily determination. A count of at least 100,000 normally functioning platelets is necessary to ensure haemostasis in oedematous brain. A bleeding time within the normal range with lower platelet counts should not be accepted as safe, because the test is carried out in an organ, that is, skin, in which the vessels are normal and the physical properties are markedly different from those of the tumor and brain.
Surgery for all intracranial tumors, including metastatic brain tumors, should be carried out with magnified vision by using microsurgical techniques. Even the very large intracerebral tumors can usually be removed via small, well-planned cortical incisions. Sacrifice of large areas of cortex ("uncapping") or lobes of the brain is necessary only if they are involved with tumor.
Although some metastatic tumors present on the surface of the brain, most are entirely subcortical, and, even if quite superficial, rarely produce reliable signs of their location on inspection and palpation of the brain. Therefore, in addition to radiographic localization on the scalp, it is often essential to have intraoperative ultrasonography available. Ultrasonography is helpful in choosing the most appropriate placement of the cortical incision and the direction of the transparenchymal approach. At some centers, surgery on metastatic lesions is done stereotactically either with the stereotactic frame or with frameless stereotaxy using a navigational wand.
The ideal cortical incision, and the one that is often appropriate, follows the precise center of a single gyrus perpendicular to its transverse diameter. This incision minimizes the risk of major damage to the large arteries in the adjacent sulci as well as to their branches, which run transversely across the surface of the intervening gyrus to end or anastomose at its center. The lack of major deficit following removal of small tumors from immediately beneath primary motor, sensory, and speech cortex suggests that careful splitting of a gyrus in this manner is compatible with its continued function.
For tumors below the surface of the brain, a small incision over the surface of the tumor is first made, and the wound is thereafter enlarged in the appropriate direction and only to the extent needed for removal of the neoplasm. Incisions through white matter should be carried out by careful blunt dissection parallel to the major tracts. A well-defined plane is usually present between the surface of a metastatic tumor and the surrounding brain. If the tumor is of a moderately firm consistency, it can and should be removed in one piece if this can be accomplished without major injury to critical areas of the brain. Large and deep tumors, and those with ill-defined margins, should be dealt with by progressively reducing the center and carefully dissecting, folding in, and removing the adjacent margins of the tumor. Surgery for this type of neoplasm is greatly facilitated by use of the ultrasonic aspirator. As a source of recurrence, the relative danger of seeding the wound with viable cells compared with that of incompletely removing tumor in the margins of the cavity is unknown. The use of the laser in attempts at total removal of extremely friable tumors may help provide an answer.
Results of Surgery
The treatment of metastatic tumors by surgical resection followed by radiation therapy is highly effective. The overall median survival calculated by the Meier-Kaplan method was 9.2 months from the time of craniotomy. The mean survival is about 24 months. The age at time of surgery ranging from 15 to 85 years with a median of 56 years. Age had no impact on survival. The male/ female ratio is 1 : 1. Women had a statistically significant longer median survival than did men, 11 versus 8.4 months, respectively (p < .02, log-rank test).
The 30-day surgical mortality is 5 percent. Altogether, (37.2 percent) died within 6 months of brain surgery. One-year survival is 39.4 percent; the percentages of patients who surviving 2 and 3 years were 16.3 and 11.5 percent, respectively. Among longterm survivors (>5 years) are children with sarcoma. patients with non-small cell lung cancer, renal cancer, melanoma, testicular cancer. and breast cancer. Only 25% patients survived to years. Rare patients are alive and well 15 years after the resection of a metastatic testicular tumor.
Carcinoma of the lung is not only the leading cause of death from cancer, but also is responsible for the majority of brain metastases. Factors like patient age, tumor histology, synchronous or metachronous diagnosis of primary and brain lesions, and presence of a single or multiple tumors did not affect survival in a Cox multivariate analysis. Extent of resection (complete or partial resection of brain lesion), location of tumor (supratentorial or infratentorial) and size of brain lesion were statistically significant factors in a univariate log-rank test and approached significance in Cox analysis: (p = .to, p = .05. P = .06. respectively). The most significant factors influencing survival in patients with non-small cell lung cancer were the extent of primary lung tumor resection (favourable correlation p = .0002), presence of active systemic disease (adverse correlation. p = .008), and male gender (adverse correlation p = .008). Patients undergoing curative resection of the primary lung tumor had a median survival of 14.5 months, which was significantly different from that of patients undergoing palliative resection or no surgical treatment of the primary lesion. Among one series of patients. there was a higher incidence of tumor recurrence in the brain in a subgroup of 11% patients who had previously failed whole brain radiation therapy than in a subgroup of 40% patients who had received radiation therapy following surgery (70.2 versus 46.8 percent). In addition, the patients who had previously failed radiation therapy had a much shorter median survival of 7.5 months from date of craniotomy and tended to be corticosteroid-dependent even after the apparently successful surgical removal of the tumor.
The site of the metastasis in the posterior fossa (cerebellum) has been found to influence survival adversely. Among patients operated on for brain metastases. the median survival of 17% patients with tumors in the cerebellum was 7 months, compared with a median survival of 10 months (p = .002, log-rank test) for patients with supratentorial tumors. A similar tendency was observed among the patients with non-small cell lung cancer (p = .04, log-rank test).
Patients with multiple brain metastases are rarely considered surgical candidates. Exceptions exist, however. These include patients whose systemic disease is either limited or controlled, and (1) whose metastases are resectable in one or two operations, or (2) who have one or two lesions that pose a life-threatening situation and who have exhausted radiation therapy and have a reasonable performance status. In a minority of patients this aggressive, compassionate approach may provide an increased life span. More importantly, it may maintain the quality of life beyond that achievable by radiation therapy alone. Among one series of patients, (7.3 percent) who had multiple metastases were operated on. There was no statistical difference in the median survival between patients with a resection of a single metastasis and those with the resection of multiple metastases: 9.2 versus 9.0 months (p < .27, log-rank test), respectively. Small number of patients had more than three operations to remove all (maximum, five) metastatic lesions. A similar observation was seen in the subgroup of patients with non-small cell lung cancer, in whom no difference in median survival after resection of single and multiple tumors was found. Patients harbouring multiple metastases is not unique, and several reports of surgical resection of multiple metastases have appeared.
The local recurrence rate and/or the appearance of a new metastasis in another region of the brain following surgery is as high as 49 percent in patients with-small cell carcinoma.
Should this occur while the systemic disease is limited and the patient's performance is good, it is better to offer to the patient the possibility of further surgery. (8.4 percent) have a second operation for recurrence. The median survival of this subgroup from the time of first surgery is around 15 months. In one series with non-small cell lung cancer, 50% patients had either a local recurrence or a new lesion. The median survival of patients who underwent a second resection is 17 months, compared with 11 months (p < .0002, log-rank test) in patients who did not have second surgery. Median survival calculated from second surgery in those patients usually 10 months, and the median time between first and second operation is 5 months.
Postoperative Radiation Therapy
Important issues related to the treatment of single brain metastases are whether radiation therapy should be given following surgery and whether focal or whole brain radiation therapy should be used. The latter question also applies to patients with single lesions treated by radiation therapy alone. The use of postoperative whole brain radiation therapy is predicated on the assumption that, even in lesions cleanly removed, microscopic foci may be left in the tumor bed and that undetected metastases reside elsewhere in the brain. Although seemingly logical, neither assumption is based on extensive evidence or controlled trials. Nevertheless, most published reports suggest that postoperative radiation therapy decreases relapses both at the site of the surgical extirpation and elsewhere in the brain. In one small series of patients who underwent resection of single brain metastases, no difference in survival between patients who received radiation therapy after surgery and those who underwent surgery alone. A similar trend demonstrating no difference in median survival or rate of brain recurrence was noted in nonsmall cell lung cancer, comparing patients who were treated with postoperative radiation therapy and those who did not receive it. This controversial issue can be solved only by a prospective randomized study.
The relative efficacy of whole brain versus focal radiation therapy in patients with apparently single metastases has not been determined. Whether the high sensitivity of enhanced MRI is sufficient to rule out the presence of additional metastatic deposits to enable safe withholding of radiation therapy is unknown. The fact that the incidence of single metastasis observed by CT scanning is essentially the same as that found at autopsy suggests that micrometastases are uncommon in these patients. Apart from preventing the untoward consequences of whole brain radiation therapy, such as dementia, the use of focal therapy affords the possibility of radiation therapy for subsequent metastases in long-lived patients.
Radiation Therapy and Chemotherapy
The use of radiation therapy and chemotherapy as primary treatment modalities or for palliation is discussed elsewhere. Response to ionizing radiation, of course, varies with tumor type; reduction of the mass lesion, even in patients in whom there is complete tumor kill, is relatively slow because it is dependent on mechanisms such as phagocytosis. Nevertheless, combined with corticosteroids, radiation therapy has been shown to extend survival of significant numbers of patients with multiple brain metastases and should be used in patients who have a life expectancy of more than 2 to 3 months. Demonstration by the Radiation Therapy Oncology Group of the effectiveness of a higher-dose fraction delivered over a shorter period of time (e.g., 2000 cGy in I week or 3000 cGy in 2 weeks) has improved the socioeconomic impact of whole brain radiation therapy.
The role of chemotherapy in patients with brain metastases is limited. However, the concept that it is almost always ineffective is being challenged by recent studies. In some tumors, such as germ cell neoplasms (especially choriocarcinoma), small cell lung carcinoma and possibly some breast carcinomas, the combined use of chemotherapy and radiation therapy may enhance the therapeutic response. Combination chemotherapy using chloroethylnitrosoureas and tegafur has an additive effect on radiation therapy in reducing the size of metastatic brain tumors from lung carcinoma.
Stereotactic radiosurgery is a radiotherapeutic technique that involves precisely focused ionizing radiation to destroy an intracranial target. This concept of focused radiation to treat different diseases was conceived by Lars Leksell in 1951 and was the origin of modern radiosurgery. During the last 2 decade, stereotactic radiosurgery has been used to treat a wide variety of intracranial lesions, from arteriovenous malformations to benign and malignant neoplasms. The commercial availability of stereotactic radiosurgical technology has greatly expanded the utilization and applications of radiosurgery.
Metastatic brain tumors, by virtue of being well demarcated, generally spherical, and small in diameter (<3 cm), are well suited to stereotactic radiosurgery. Although there are no studies to date comparing the results of radiosurgery with those of surgery, radiosurgery seems to be utilized more frequently for the treatment of brain metastases and is likely to supplant surgery as the standard treatment of the solitary brain metastasis of small size. Surgery remains favoured for patients with large lesions, hemorrhagic lesions, and those with significant mass effect. Today, results of radiosurgical treatment of brain metastases are encouraging. The local control rate using a single fraction (session) of 1600 to 3500 cGy is reported to be around 88 percent, almost twice the 45 percent local control rate achieved with standard whole brain irradiation. Response rates of relatively radioresistant tumors, such as colon carcinoma and melanoma, seem to be good. Moreover, peritumoral oedema, a major source of morbidity in patients with metastatic brain tumors, is reduced after treatment, and in the majority of patients the steroid requirements are lessened. Up to three metastases can be treated in one session and the hospital stay is extremely short. The cost of the procedure is lower than that of neurosurgical treatment. The potential advantages of stereotactic radiosurgery are clear for patients with an overall dismal prognosis; it provides a safe alternative to neurosurgery.
Intracranial metastases are the most common neurological complication of cancer. Brain metastases originate most commonly from carcinomas of lung, breast, and colon and from melanoma. The most frequent symptoms and signs include headache, seizures, focal weakness, and behavioural and cognitive changes. The neurodiagnostic test of choice is contrast-enhanced MRI. For most patients with brain metastases, treatment is palliative and consists of steroid administration and whole brain radiation therapy. Palliative treatment is appropriate in patients with unresectable multiple brain metastases or a single brain metastasis associated with widespread systemic disease. Only a minority of patients are candidates for interventional therapy, such as surgery or stereotactic radiosurgery. The best candidates for either neurosurgery or stereotactic radiosurgery are the patients with limited systemic disease and good performance status, which are the most important prognostic criteria.
Surgical extirpation is appropriate in the patient with a solitary metastasis, with a local recurrent or a remote metastasis after surgery or radiation therapy, or with surgically accessible multiple metastases. In experienced hands, current morbidity and mortality from craniotomy are low enough that no patients should be denied the possible benefits of surgery.
Stereotactic radiosurgery is being used increasingly in the management of brain metastases. Treatment is noninvasive, is associated with low morbidity and zero mortality and demonstrates excellent local control rates-better than those found with surgery and whole brain radiation therapy. Radiosurgery treatment can be applied for up to three lesions in one session and either is done on an outpatient basis or requires an ultrashort hospitalization. Patients unsuitable for treatment with stereotactic radiosurgery include those with lesions larger than 3 cm in diameter and lesions producing a significant mass effect.
Clearly, the prognosis of most patients with brain metastases remains poor. However, a small but increasing percentage of patients can be "cured." The goal of therapy is to maintain an acceptable quality of life while extending the length of survival. This objective can be reached only through a collaborative effort of various specialists who are fully cognizant of the advantages and limitations of the various treatment options.