intracranial metastases.pdf

3601_e13_p320-340 2/15/02 4:57 PM Page 320

Part V

Cancer Metastaticto the

Central Nervous System

3601_e13_p320-340 2/15/02 4:57 PM Page 320a

3601_e13_p320-340 2/15/02 4:57 PM Page 320b

321

13

Intracranial Metastases

W.K. ALFRED YUNG, LARA J. KUNSCHNER, RAYMOND SAWAYA, ERIC L. CHANG, AND GREGORY N. FULLER

Tumors that metastasize to the brain and intracranialspace are very common, occurring in approximately10% to 15% of patients with systemic cancers (Pos-ner and Chernik, 1978; Cairncross and Posner, 1983;Walker et al., 1985). In contrast to the 17,500 newprimary brain tumors that are expected to occur an-nually, more than 150,000 new cases of metastaticbrain tumors occur each year in the United States(Posner, 1992). Moreover, the occurrence of metas-tasis to the brain is increasing as cancer patients livelonger because of improved treatment and becausethe incidences of lung cancer and malignant mela-noma continue to rise (Galicich et al., 1980a). Thefollowing are systemic cancers with a high incidenceof metastasis to the brain

Lung carcinoma Breast carcinoma Malignant melanoma Renal cell carcinoma Colorectal carcinoma

Lung carcinoma, especially small cell carcinomaand adenocarcinoma, is the most common type ofcancer to metastasize to the brain, often with no ev-idence of systemic relapse (Figlin et al., 1988). Theincrease in the number of women and teenagerswho smoke cigarettes and the increased rate of cig-arette consumption has undoubtedly accounted forthe increased number of young patients developinglung carcinoma and subsequent metastases to thebrain.

Approximately 50% of patients with tumors meta-static to the brain present with single lesions, whereasthe other 50% have two or more lesions (Delattre etal., 1988). A significant number of patients are foundto have asymptomatic metastasis to the brain at thetime of presentation with lung carcinoma (Salbeck etal., 1990).

Breast cancer is the second most common type ofcancer to metastasize to the brain, occurring in ap-proximately 10% of patients with advanced breastcancer (DiStefano et al., 1979). Some studies reportthat as many as 30% of patients experience concur-rent systemic metastasis involving other organs, in-cluding the lung, bone, and liver (Tsukada et al.,1983). Younger patients and premenopausal patientsare more prone to develop metastasis to the brainthan are older, postmenopausal patients.

Malignant melanoma represents the third mostcommon type of cancer having a high incidence ofmetastasis to the brain. An estimated 30% to 40% ofpatients with malignant melanoma will develop intracranial disease, including parenchymal andmeningeal metastases (Byrne et al., 1983). In thebrain, melanoma metastases are often hemorrhagicand multiple. The increasing incidence of melanomametastasis to the brain may be the result of bettercontrol of extracranial disease through aggressivechemotherapy and biologic therapies.

Renal cell carcinoma and colorectal carcinomaare the fourth and fifth most common sources of brainmetastases, respectively. Renal cell metastases, like

3601_e13_p320-340 2/15/02 4:57 PM Page 321

322 CANCER METASTATIC TO THE CENTRAL NERVOUS SYSTEM

melanoma metastases, tend to be highly vascular andprone to hemorrhage. They produce single lesionsmore frequently than do lung cancer and melanomaand exhibit a rare, but well-documented, potential forproducing late brain metastases that sometimes pres-ent more than a decade after resection of the primaryrenal tumor (Radley et al., 1993). The brain is oneof the least frequent sites of metastasis from colorectalcarcinoma; these metastases occur in approximately5% of patients (compared with approximately 30%to 40% of melanoma patients, 35% of lung carcinomapatients, and 10% to 30% of breast carcinoma pa-tients) (Cascino et al., 1983). Among patients whohave solitary colorectal carcinoma metastases to thebrain, it has also been shown that a disproportion-ately large percentage (approximately 50%) of metas-tases occur in the posterior fossa (cerebellar) (De-lattre et al., 1988).

These five types of metastatic tumors account forapproximately 85% of all metastases to the brain (De-lattre et al., 1988). A large number of other systemicneoplasms may produce central nervous system(CNS) metastases, including hematologic malignan-cies (leukemia, lymphoma), a wide spectrum of sys-temic carcinomas, and, in small numbers but with in-creasing frequency, some varieties of sarcoma (Lewis,1988).

PATHOLOGY

Intracranial metastatic disease may involve any of thethree principal morphologic compartments: the dura,leptomeninges (subarachnoid space), or brain pa-renchyma. Metastases that initially involve only onecompartment frequently invade other compartmentsas they grow. This is particularly true of parenchymallesions located either very superficially in the graymatter or subjacent to the ventricular system. Theselesions may secondarily breach the pia or ependyma,which allows them access to the ventricular systemor subarachnoid space, and subsequently to dissem-inate widely via the cerebrospinal fluid pathways. Al-ternatively, tumors located primarily in the sub-arachnoid space (leptomeningeal carcinomatosis)often track centrally via the perivascular Virchow-Robin spaces (Fig. 131), with ultimate expansioninto the brain parenchyma.

By far, the most commonly encountered site of me-tastasis from carcinoma and sarcoma is within thebrain parenchyma. In contrast, metastasis resultingfrom leukemia preferentially involves the lep-tomeninges. Breast carcinoma tends to produce iso-lated dural metastases (Fig. 132) in addition toparenchymal lesions. Prostate carcinoma more com-monly metastasizes to the skull and vertebrae than to

Figure 131. Leptomeningeal dissemination of tumor with secondary spread into brain parenchyma along perivascular (Virchow-Robin) spaces. (A) Melanoma cells fill the subarachnoid space. Once access to this strategic compartment has been gained, met-astatic tumors can rapidly disseminate throughout the cerebrospinal fluid pathways of the neuraxis and may secondarily involvevarious anatomic constituents of the subarachnoid space such as blood vessels and cranial nerve rootlets or (B) may spread intothe brain substance along the perivascular spaces of penetrating blood vessels. This is a common pattern of dissemination for mel-anoma and can also be seen with breast and other metastatic carcinomas. Diffuse leptomeningeal involvement is also frequentlyseen with the leukemias.

3601_e13_p320-340 2/15/02 4:57 PM Page 322

Intracranial Metastases 323

the brain parenchyma, although intra-axial lesionsmay also occur. As in other locations, prostatic metas-tases to bone may be osteoblastic; in the cranium,such lesions may stimulate meningioma with hyper-ostosis.

Although parenchymal lesions may occur in anyregion of the CNS, several generalizations can bemade:

1. Most metastases to the brain are supratentorial,the most common location being the cerebralhemispheres, where tumor emboli tend to lodgein the vascular gray matter, particularly at the graymatterwhite matter interface where penetratingvessels narrow in caliber (Fig. 133).

2. Cortical hemispheric metastases are most fre-quently found in the vascular distribution ter-ritory of the middle cerebral arteries, particu-larly in the arterial border zones.

3. Metastases also occur frequently in the deepcerebral gray nuclei and white matter, as wellas in the cerebellum.

4. Brain stem and spinal cord lesions occur farless frequently, but metastases may involve vir-tually any anatomic locus within the CNS, in-cluding such specialized organs as the choroidplexus, pineal gland, and pituitary.

In rare cases, systemic neoplasms may metastasizeto a pre-existing primary brain tumor. The vast ma-

jority of such cases involve metastasis of a primarylung or breast carcinoma to a meningioma; veryrarely, the host tumor is a schwannoma or glioma(Schmitt, 1984).

Microscopic Features

Intraparenchymal metastases tend to expand asroughly spherical masses and establish a well-definedinterface with the surrounding brain parenchyma(Fig. 134). This sharp circumscription stands incontradistinction to the diffusely infiltrating marginsof most primary brain tumors and can be of consid-erable practical importance to the surgical patholo-Figure 132. Dural metastasis. Breast carcinoma has a rec-

ognized tendency to produce dural metastases, as noted in thiscase in which cords and nests of infiltrating tumor cells areseen dissecting through the dense connective tissue of thedura. Such metastases from breast, prostate, lung, or othercarcinomas can on occasion produce dural-based masses thatmimic meningioma on neuroimaging studies.

Figure 133. Parenchymal metastasis. (A) Most metastasesoriginate by hematogenous dissemination of tumor emboli.(B) As illustrated in this case of metastatic lung adenocarci-noma, parenchymal metastases are commonly supratentorial,cortical in location, centered around the graywhite junction,and most frequently lie in the vascular distribution territory ofthe middle cerebral artery, particularly in the parasagittal ar-terial boundary zones.

3601_e13_p320-340 2/15/02 4:57 PM Page 323


Figure 134. Interface of metastatic tumor with brain pa-renchyma. Sharp macroscopic (as in Fig. 133B) and micro-scopic circumscription is the rule for most metastatic tumors;for the surgical pathologist this is a useful diagnostic feature,and for the neurosurgeon it often permits total excision of theneoplasm. A notable exception to this rule is metastatic lym-phoma, which, like primary CNS lymphoma, diffusely infiltratesnervous tissue from angiocentric foci.

Figure 135. Histology of metastases. Many metastatic tu-mors closely recapitulate the morphology of the primary neoplasm, as illustrated here with (A) metastatic breast carcinoma, (B) mucinous colon carcinoma, and (C) os-teosarcoma. Frequently, however, metastases are poorly dif-ferentiated and lack suggestive histologic features. Often, inthe clinical setting of an unknown primary tumor, only a di-agnosis of metastatic carcinoma or adenocarcinoma can berendered in such cases, along with a list of the most likely primary sites.

gist, especially when dealing with small biopsy spec-imens of tumors that occur in the absence of a knownprimary systemic neoplasm.

As discussed, the overwhelming majority of tumorsthat give rise to CNS metastases are carcinomas; how-ever, a variety of sarcomas may also metastasize tothe CNS (Lewis, 1988). For all types of tumors, themorphology of the metastatic lesion in many cases re-capitulates that of the primary neoplasm with great fi-delity (Fig. 135). The degree of differentiation ex-hibited ranges from the metastatic lesion being almostpathognomonic of the primary organ of origin to be-ing so poorly differentiated that only a diagnosis ofmetastatic carcinoma or metastatic malignant neo-plasm can be made based on morphologic criteria.The brain parenchyma surrounding most metastaticlesions typically exhibits a robust reactive astroglio-sis, a fact that must be considered when interpretingneedle biopsy specimens obtained from this vicinity.

Immunostaining may be helpful in the diagnosis ofsome cases, particularly in distinguishing metastaticcarcinoma from the other two major types of malig-nant neoplasms that involve the CNS parenchyma ofolder individuals: glioblastoma and primary CNS lym-

3601_e13_p320-340 2/15/02 4:57 PM Page 324


phoma. In small and/or poorly preserved biopsyspecimens, all three of these tumor types may appearquite similar on hematoxylin and eosin (H&E)stained tissue sections: they are all high-grade pleo-morphic tumors that show large atypical tumor cells,mitotic figures, and tumor necrosis. The basic panelof immunostains typically used in this situation is (1) glial fibrillary acidic protein (GFAP), which ispositive only in gliomas; (2) a mixture of low- andhigh-molecular-weight keratin antibodies (keratincocktail) for metastatic carcinoma; and (3) the lym-phoma markers CD45 (leukocyte common antigen),CD3 (a T-cell marker), and CD20cy (L26, a B-cellmarker), either separately or combined as a CNSlymphoma cocktail. This panel will usually permitseparation of the three tumor types and definitiveidentification of metastatic carcinoma when present.

One important caveat is that many glioblastomaswill show cross-reactivity with keratin antibodies;however, the GFAP immunostain will only be positivein glioblastomas, not in carcinomas or lymphoma.Once a diagnosis of metastatic carcinoma has beenmade, it is sometimes, but not always, possible to con-firm the primary site of origin. If the patient has aknown primary, a tentative diagnosis of metastaticcarcinoma consistent with the primary can be ren-dered and immunostaining may not be necessary. Ifthere is no known primary, a thorough clinical eval-uation of the patient is required. For metastatic car-cinomas in patients in whom no primary site can beidentified after clinical work-up, a number of anti-bodies can be used to help identify tumor-specificphenotypic features. Unfortunately, many of the anti-bodies currently used in immunopathology are notuniformly as specific as one would like, and, there-fore, both positive and negative antibody immuno-staining must be interpreted with caution and in thecontext of the patients clinical history. Nevertheless,antibody studies can be helpful. Useful antibodies in-clude estrogen and progesterone receptor antibodiesfor breast carcinoma; thyroid transcription factor 1(TTF-1) for lung and thyroid carcinomas; prostaticacid phosphatase (PAP) and prostate-specific antigen(PSA) for prostate carcinoma; and thyroglobulin (aswell as TTF-1) for thyroid carcinoma. Again, it mustbe stressed that all these antibodies may show posi-tivity in other carcinomas, and the results should notbe considered absolute.

Another approach that has been taken is the use ofseveral keratin antibodies of differing molecularweight together with other markers in large panels;

diagnostic algorithms are then used to give estimatesof the most likely primary site(s) based on the par-ticular pattern of reactivity seen. Again, as with the useof individual antibodies, panel results can be helpfulin directing attention to primary sites that should beinvestigated further clinically, but findings should beinterpreted with caution. Brief mention must also bemade of metastatic melanoma. Heavily pigmented le-sions are easy to diagnose; however, up to a one-thirdof brain metastases appear amelanotic on H&E-stainedslides, and melanoma may initially present as a CNSmetastasis from an unknown primary tumor. Im-munomarkers usually employed in this situation areS-100 protein, HMB-45, and MART-1.

PATHOPHYSIOLOGY

It is generally accepted that most metastatic brain tu-mors arise as a result of the hematogenous spread ofcancer cells to the brain. Because brain microvesselsare structured differently from systemic microvessels,cells destined to metastasize to the brain first adhereto and then penetrate the bloodbrain barrier (BBB)formed by endothelial cells. Cells must penetrate thebasement membrane and astrocytic foot processesbefore reaching the soil that allows their prolifer-ation in brain parenchyma (Nicolson, 1982).

This process is thought to be nonrandom, as pro-posed by the seeds and soil hypothesis for can-cer metastasis (Paget, 1989). The properties of thetumor cells, or seeds, that determine the preferen-tial development of brain metastasis from certaintypes of cancer include the quantitative expression ofspecific adhesive molecules that allow preferential ad-hesion to brain endothelial cells (Nicolson, 1988a,b)and the increased production of certain degradativeenzymes, such as type IV collagenase and heparinase,that may enable tumor cells to penetrate the endo-thelial junction and the basement membrane (Liottaet al., 1991). Once the metastatic cells have pene-trated the BBB, they proliferate in the appropriate mi-croenvironment, or soil, indigenous to brain pa-renchyma. Certainly, locally produced growth factorsmay stimulate the growth of specific metastatic tumorcells (Cavanaugh and Nicolson, 1991). In the brain,fibroblast growth factor, which has been found in highlevels in the normal brain, may have such an effect(Rodeck et al., 1991).

Metastatic brain tumors, as opposed to primarymalignant gliomas, usually grow as well-demarcated,

3601_e13_p320-340 2/15/02 4:57 PM Page 325


spherical masses that are often amenable to total ex-cision by surgery. Another difference between meta-static tumors and primary malignant gliomas is thatbrain metastases generally grow more rapidly and ex-hibit a high bromodeoxyuridine labeling index, oftenin the range of 20% to 25%, indicating a large pro-liferating fraction of tumor cells (Cho et al., 1988).Most gliomas, on the other hand, have lower label-ing indices. Metastatic tumors are solid or cystic asa result of central necrosis; some tumors, such asmetastases from melanoma, choriocarcinoma, andtesticular carcinoma, are often hemorrhagic and tendto invade the vascular wall (Pullar et al., 1985).

CLINICAL PRESENTATION

The clinical presentation of patients with metastasesto the brain depends on the number and location ofthe metastases. Multiple metastases frequently occurin lung carcinoma, breast carcinoma, and malignantmelanoma; single metastases more commonly occurin patients with colorectal and renal cell carcinoma(Delattre et al., 1988). The common presentingsymptoms in patients with metastases to the brain arelisted in Table 131. Most patients present with com-plaints secondary to an increase in intracranial pres-sure (headache, mental change, or somnolence) orwith focal (complex partial) or generalized seizures.

Patients with a single metastasis usually develop fo-cal symptoms and signs in addition to headache andchange in mental status. These may include (1) cra-nial nerve palsies, usually involving nerves VI and VII;(2) dysphasia; (3) visual deficits; (4) hemiparesis;or (5) hemisensory loss. Patients with multiple brain

metastases can also present with diffuse, nonlocalizedsymptoms, such as generalized weakness and boweland bladder incontinence. Surprisingly, many patientshave few or no obvious symptoms and signs; thus, aphysician should have a high index of suspicion forpatients at risk of developing brain metastases.

IMAGING

The diagnosis of brain metastasis is confirmed withcomputed tomography (CT) or magnetic resonanceimaging (MRI) when the patients history and neuro-logic examination raise the possibility of this diagno-sis. Magnetic resonance imaging with gadolinium con-trast enhancement represents the most sensitivediagnostic tool used to detect the presence of single ormultiple metastases. Cerebrospinal fluid examination isonly helpful in the presence of meningeal metastasis.

The presence of a single lesion seen on a CT or MRIscan in a patient with progressive systemic cancer isnot an unequivocal indication of brain metastasis. Thepossibility that the lesion may be a cerebral abscess, amalignant glioma, or a meningioma must be consid-ered and carefully ruled out. This will often requiresurgical biopsy or, preferably, excision of the lesion.

In patients who present with neurologic symptomsand whose CT or MRI results show a contrast-enhancing mass suggestive of a metastatic lesion, theprimary cancer must be located before the brain le-sion can be treated. The most common primary sitewill be the lung in men and the lung or breast inwomen. Once a brain metastasis has been discov-ered, the recommended techniques for screeningsystemic tumors are anteroposterior and lateralchest radiographs, chest CT scan if chest radiographresults are negative, mammogram (in women), bonescan, urinalysis, and stool guaiac test for occultblood (Voorhies et al., 1980). If these tests fail toidentify a primary tumor site, surgical resection ofthe cerebral lesion can often provide the pathologicdiagnosis. Frequently, however, metastatic adeno-carcinomas do not exhibit pathognomonic morpho-logic features, and only a diagnosis of metastatic ade-nocarcinoma can be rendered. Not surprisingly, inone recent study, 85% of adenocarcinomas of un-known origin were ultimately found to have origi-nated in the lung (Mrak, 1993). If there is more thanone lesion in the brain, the largest or the most symp-tomatic lesion should be chosen first for surgical resection.

Table 131. The Most Common Presenting Symptoms inPatients With Metastases to the Brain

Relative Frequency atSymptoms Diagnosis (%)

Headache 71

Seizures 54

Mental change 52

Hemiparesis 43

Vomiting 32

Dysphasia 27

Impaired consciousness 25

Visual change 25

3601_e13_p320-340 2/15/02 4:57 PM Page 326

TREATMENT

General Considerations

The first stage of treatment for patients with single ormultiple metastases to the brain is to stabilize acuteneurologic symptoms caused by increased intracra-nial pressure and, in some patients, status epilepti-cus. Headache, nausea and vomiting, and change inmental status are the most common signs and symp-toms of increased intracranial pressure. Patients withthese symptoms should begin steroid treatment im-mediately (Cairncross and Posner, 1981). The typi-cal starting bolus is 10 to 20 mg, followed by 16mg/day in divided doses. The antiedema effect ofsteroids can often provide dramatic relief of symp-toms such as headache and mental confusion. If theCT or MRI scan shows a marked increase in cerebraledema and evidence of impending herniation, the ste-roid dose can be increased to 24 or 30 mg/day withan initial loading dose as high as 100 mg, if neces-sary.

In patients showing signs of cerebral herniation,steroids usually do not take effect fast enough to re-verse the rapid progression of neurologic dysfunctionand impending death. Hyperventilation and intra-venous hyperosmotic agents such as mannitol mustbe initiated immediately in addition to steroid ther-apy. Endotracheal intubation may be required toachieve adequate hyperventilation in obtunded or co-matose patients. A pCO2 level of 20 to 25 should bemaintained to achieve optimal vasoconstriction. Man-nitol should be given as a 20% solution at a dose of1 to 2 g/kg body weight every 6 hours after an initialdose of 50 to 100 g administered intravenously over20 to 30 minutes. The subsequent dose of mannitolcan be adjusted according to the patients level ofconsciousness and be tapered off when the patientscondition stabilizes with steroids (Ravussin et al.,1988).

Generalized or partial seizures are the presentingsymptoms in 15% to 25% of patients with metastasisto the brain and are most common in patients withsuperficial lesions near the cortical gray matter. Sta-tus epilepticus can be lethal if allowed to continue fora prolonged period and should be eliminated with in-travenous diazepam, lorazepam, or phenytoin, de-pending on the preference of the treating physician.Maintenance therapy should be initiated for long-termtreatment. Phenytoin remains the most commonlyused anticonvulsant, but carbamazepine and sodium


valproate should also be considered. Unfortunately,there has been no study to compare the efficacy ofthese three drugs in this group of patients. However,several studies have reported fluctuation of phenytoinlevels in patients receiving systemic chemotherapy(Grossman et al., 1989). Thus, the necessary anti-convulsant level should be monitored more frequentlyto maintain adequate control of seizures. Moreover,the issue of prophylactic anticonvulsant therapy in pa-tients with metastasis to the brain or a primary braintumor without seizures as a presenting symptom hasnot been resolved (Cohen et al., 1988). Despite theuse of prophylactic anticonvulsants, the incidence ofseizures with a later onset is the same as in patientsnot receiving prophylactic anticonvulsants. Neverthe-less, it is advisable to start the patient on an anticon-vulsant if the lesion is located in an epileptogenic area.

Radiation Therapy

The benefit of administering external-beam radiationtherapy to patients afflicted with metastatic depositsof cancer in the brain was first reported by Chao andcolleagues (1954) and subsequently by Chu and Hi-laris (1961). Since that time, intermediate-dosewhole-brain irradiation delivered in daily fractiona-tion over 1 to 4 weeks has been considered the stan-dard therapeutic approach for such patients (Coia etal., 1988). In numerous prospective and retrospec-tive series, the value of cranial irradiation in pre-venting or delaying progression of neurologic deficits,restoring function, and decreasing steroid depen-dency has been well documented (Order et al., 1968;Borgelt et al., 1980; West and Maor, 1980; Kurtz etal., 1981).

The Radiation Therapy Oncology Group (RTOG)conducted several large phase III randomized trialsin the 1970s evaluating the efficacy of myriad cranialirradiation fractionation schedules that varied radia-tion fraction sizes, total radiation doses, and times oftreatment. The first two studies evaluated five one-fraction-per-day fractionation schemes: (1) 20.0 Gyin five 4.0-Gy fractions over 1 week; (2) 30.0 Gy in10 3.0-Gy fractions over 2 weeks; (3) 30.0 Gy in 152.0-Gy fractions over 3 weeks; (4) 40.0 Gy in 15 2.67-Gy fractions over 3 weeks; and (5) 40.0 Gy in 20 2.0-Gy fractions over 4 weeks (Borgelt et al., 1980). Nodifferences in survival time (less than 26 weeks), timeto neurologic progression (12 to 19 weeks), or fre-quency of neurologic improvement were observedamong these treatment arms.

3601_e13_p320-340 2/15/02 4:57 PM Page 327

From 1976 to 1979, the RTOG evaluated two ra-diation schedules for 255 patients whose distantmetastases were limited to the brain and whose pri-mary tumor was controlled or absent. The patientswere randomized to receive 30.0 Gy in 10 3.0-Gy frac-tions over 2 weeks or 50.0 Gy in 20 2.5-Gy fractionsover 4 weeks (Kurtz et al., 1981). No differences inpalliative results or survival were observed. As a re-sult of reports from these studies, 30.0 Gy in 10 3.0-Gy fractions has become the most commonly admin-istered radiation regimen in the United States forpatients with brain metastases, although the use ofmore protracted fractionation schemes should beconsidered in certain instances and is discussed later.

Patient factors are important in the evaluation oftreatment response and outcome in clinical trials. Di-ener-West and coworkers (1989) used multivariateanalysis to identify favorable subgroups of patients forfuture protocols and showed that four factors wereassociated with improved survival: having a Karnof-sky performance scale (KPS) score of 70 or more,having an absent or controlled primary tumor, beingyounger than 60 years old, and having metastatic dis-ease limited to the brain. Patients with all four favor-able characteristics had a predicted 200 day survivalof 52%. Those with none of the favorable factors hada predicted survival of 1.8 months (Diener-West etal., 1989). These prognostic factors were again iden-tified by recursive partitioning analysis of a databasefrom three consecutive RTOG trials. Three classes ofpatients were proposed: Class 1 included patients witha KPS score of 70 who were 65 years of age andwho had a controlled primary tumor but no ex-tracranial metastases; class 3 included those with aKPS score of 70; class 2 included all remaining pa-tients (Gaspar et al., 1997).

An RTOG phase I/II trial of accelerated fractiona-tion in patients with brain metastases prescribedwhole-brain radiation of 1.6 Gy twice daily separatedby 4 to 8 hours delivered for 5 days a week to 48 to70.4 Gy. Analysis of the results suggested that doseescalation was tolerated without excessive toxicity andmight improve survival in patients receiving 54 Gy orgreater (Sause et al., 1993). In the follow-up RTOGphase III study, 445 unresected patients were ran-domized to receive either accelerated hyperfraction-ation of 1.6 Gy b.i.d. to 54.4 Gy or accelerated frac-tionation of 30 Gy in 10 fractions. Unfortunately, thistrial failed to demonstrate any improvement in sur-vival with 54.4 Gy (Murray et al., 1997).

Radiation sensitizers such as misonidazole andbromodeoxyuridine have been investigated by theRTOG. Randomized studies of patients with brainmetastases have failed to show any statistically signif-icant survival benefit with the administration of eithercompound along with whole-brain radiation therapy(WBRT) (Komarnicky et al., 1991; Phillips et al.,1995). Motexafin gadolinium texaphyrin (XCYTRIN),a drug that comes from a family of ring-shaped mol-ecules adapted from porphyrin molecules, is beinginvestigated as a radiation sensitizer in the treatmentof brain metastases. The results from the phase I andII studies found that it was well tolerated when ad-ministered daily with WBRT (Viala et al., 1999; Mehtaet al., 2000). A randomized phase III trial has beeninitiated.

Patchell and coworkers (1990) demonstrated asurvival advantage for selected patients with re-sectable, single brain metastases randomized to re-ceive resection and cranial irradiation over patientsgiven cranial radiation alone, and this trial establishedsurgery and postoperative irradiation as the standardapproach for such patients. The study was limited topatients with a KPS score of 70 or greater and con-firmed the benefit of resecting solitary lesions that hadbeen previously suggested in nonrandomized studyreports (Galicich et al., 1980a; Sause et al., 1990).The postoperative cranial radiation was delivered in12 3.0-Gy fractions to a total dose of 36.0 Gy.

Several retrospective studies have examined therole of postoperative radiation therapy for patientswith brain metastases (Dosoretz et al., 1980; DeAn-gelis et al., 1989b; Hagen et al., 1990; Smalley et al.,1992; Armstrong et al., 1994; Skibber et al., 1996).The majority of these studies do not demonstrate asurvival benefit from adding WBRT after surgery. Aprospective randomized study by Patchell and co-workers (1998) is the only one that addresses thisissue. Ninety-five patients with a single metastasis tothe brain were treated with complete surgical resec-tion and were then randomized either to treatmentwith postoperative WBRT or to observation. Recur-rence of metastases in the brain was less frequent inthe radiation group (18%) than in the observationgroup (70%) (p 0.001). Postoperative radiother-apy prevented brain metastasis recurrence at both thesite of the original lesion (10% versus 46% for theuntreated group; p 0.001) and at other sites in thebrain (14% versus 37%, respectively; p 0.01).Deaths from neurologic causes were also reduced by


3601_e13_p320-340 2/15/02 4:57 PM Page 328

administration of WBRT (14% versus 44% for un-treated patients; p 0.003). This reduction in neu-rologic deaths was seen only in those patients re-ceiving WBRT immediately after surgery and not atthe time of recurrence. No significant difference inoverall survival was observed for untreated or irradi-ated patients. On the basis of randomized datademonstrating the reduction of neurologic deaths inpatients receiving postoperative WBRT, it is reason-able to recommend its routine use.

Stereotactic radiosurgery was introduced in 1951by the Swedish neurosurgeon Lars Leksell as a tech-nique designed to destroy the target lesion with asingle large dose of radiation delivered with a seriesof narrow radiation beams. Stereotactic radiationtechniques exhibit rapid dose fall-off at the targetedges, permitting significant sparing of normal braintissue (Phillips et al., 1994). Numerous reports fromdifferent institutions support the use and effectivenessof radiosurgery for brain metastases (Wen and Loef-fler, 1999). A multi-institutional outcome and prog-nostic factor analysis of radiosurgery for resectablesingle brain metastases showed that radiosurgery inconjunction with WBRT for a single brain metastasiscan produce a substantial functional survival time of56 weeks from the date of radiosurgery in patientswith good performance status and who lack ex-tracranial metastases. The comparability of these re-sults to surgical series suggested that a randomizedcomparison of radiosurgery with surgery for brainmetastasis treatment would be of great interest (Shaw,1999). Such a trial is concurrently open to patientswith a single brain metastasis at The University ofTexas M. D. Anderson Cancer Center (M. D. Ander-son). There are apparent advantages to either radio-surgery or surgery, depending on the clinical sce-nario. Surgery provides immediate resolution of masseffect and tissue for pathologic diagnosis if neededand poses no risk of radiation necrosis (Loeffler andAlexander, 1993). With radiosurgery, there are de-creased risks of hemorrhage, infection, and tumorseeding as well as reduced costs produced, in part,by not requiring hospitalization.

The question has been raised by the results of themulti-institutional radiosurgery series as to whetherradiosurgical treatment can improve survival beyondthat produced by WBRT alone. This question was ad-dressed by a study from the University of Pittsburghin which 27 patients who had 2 to 4 metastases (25mm in diameter) were randomized to initial man-

agement with WBRT or with WBRT plus radiosurgery(Kondziolka et al., 1999). Local failure of tumor con-trol was 100% in those receiving WBRT alone but wasonly 8% in those who received a radiosurgery boost.The median time to local failure was 6 months afterWBRT alone and 36 months after WBRT plus radio-surgery (p 0.0005). The median time to any localcontrol failure in the brain was improved in the ra-diosurgery arm of the study (34 months) relative tothe WBRT-alone arm (5 months; p 0.002). Pa-tients in the radiosurgery arm had a median survivaltime of 11 months versus 7.5 months in the WBRT-alone arm, but this difference was not statistically significant. The RTOG 95-08 trial, which is nearingcompletion, compares WBRT with or without a ra-diosurgery boost and stratifies patients as having asingle metastasis or two to three metastases.

The role of WBRT after radiosurgery has been ex-amined in a retrospective study by the group at theUniversity of California at San Francisco, which re-ported on 106 patients with single or multiple brainmetastases that were managed initially with radio-surgery or radiosurgery plus WBRT (Sneed et al.,1999). In the two treatment groups, both median sur-vival and 1 year freedom from progression were sim-ilar at 11.3 months versus 11.1 months and 71% ver-sus 79%, respectively. However, freedom from tumorprogression in the brain at 1 year was significantlyworse for the group treated by radiosurgery alone(28%) than for the group receiving radiosurgery plusWBRT (69%). The authors analyzed the results fur-ther by allowing for successful salvage of the first fail-ure and found that if this was allowed, local controlof tumor growth within the brain at 1 year was notsignificantly different for the two study arms (62%versus 73% at 1 year; p 0.56) (Sneed et al., 1999).The group at the University of Heidelberg reported ona series of 311 brain metastases in 236 patients, eachof whom had one to three brain metastases (Pirzkallet al., 1998). One hundred fifty-eight patients receivedradiosurgery only to a median dose of 20 Gy. The re-maining 78 patients received a median radiosurgicaldose of 15 Gy, followed by WBRT. Overall median sur-vival time for patients was 5.5 months, and CNS dis-ease control was achieved in 92% of treated brainmetastases. Results were not significantly different forpatients who did or did not receive WBRT, but therewas a trend toward improved local control at 2 years(86% with WBRT versus 72% without it; p 0.13).Interestingly, for patients without extracranial dis-


3601_e13_p320-340 2/15/02 4:57 PM Page 329

ease, there was a trend toward increased survival inpatients receiving WBRT (15.4 months) relative tothose not receiving it (8.3 months) (p 0.08)(Pirzkall et al., 1998). The role of WBRT after ra-diosurgery warrants further study in a randomizedtrial that should include patient survival, freedomfrom progression, and a validated quality of life ques-tionnaire component as endpoints to be evaluated.

Metastases from renal cell carcinoma and mela-noma are described as radioresistant because oftheir lack of response to conventional radiotherapy.These histologic types deserve special attention anddiscussion. A review of the experience at M. D. An-derson with brain metastases from renal cell carci-noma described 119 patients who were treated withWBRT alone (Wronski et al., 1997). Overall mediansurvival time of patients with the diagnosis of brainmetastases was 4.4 months, and the cause of deathwas neurologic in 76% of patients and systemic in16%. The authors suggest more aggressive ap-proaches that include surgery or radiosurgery be-cause of these unsatisfying results. A report from theUniversity of Pittsburgh described the treatment of 35patients who had 52 renal cell carcinoma metastasesthat were treated with radiosurgery over a 9 year pe-riod (Mori et al., 1998b). WBRT was given to 28 pa-tients, and their median survival was 11 months afterradiosurgical treatment. A 90% local control rate wasachieved (21% disappearance, 44% tumor regres-sion, 26% stable disease). The addition of WBRT didnot improve survival or reduce distant failure (de-velopment of remote tumors) within the brain. How-ever, local failure was only observed in the radio-surgery-alone arm of the study, leading the authorsto hypothesize that WBRT might contribute to localcontrol of tumor growth. The number of local con-trol failures is small in this study, making it difficultto draw firm conclusions about the ability of WBRTto improve local control (Mori et al., 1998b).

The Harvard University group reported that mela-noma and renal cell carcinoma metastases can becontrolled just as easily with radiosurgery as tumorshaving radiosensitive histologies (Alexander et al.,1995). The University of California at San Franciscoreported the use of gamma knife radiosurgery to treatbrain metastases from melanoma in a series of 55 pa-tients, including 16 treated for recurrence, 11 ofwhom received radiosurgery as a boost to WBRT, and28 treated with radiosurgery alone as initial manage-ment (Seung et al., 1998). Median patient survivaltime was 35 weeks, and the only significant factor in

multivariate analysis of survival was target volume ofthe tumor. No significant difference was seen in ac-tuarial freedom from development of intracranialprogression by log-rank analysis (p 0.85), com-paring patients treated with radiosurgery alone, ra-diosurgery plus WBRT, or radiosurgery for recur-rence. Thus, the role of WBRT and how best tointegrate it with radiosurgery in the management ofmelanoma and renal cell carcinoma, remains to bedefined.

A series of patients who had single brain metas-tases from melanoma was also reported from the Uni-versity of Pittsburgh, and the authors concluded thatradiosurgery alone was appropriate because WBRTdid not improve survival or local tumor control. Newbrain metastases developed less frequently with theaddition of WBRT, but this was not statistically sig-nificant (Mori et al., 1998a). A multi-institutional re-port of the use of radiosurgery for single brain metas-tases demonstrated that tumor control improvedsignificantly for melanoma and renal cell carcinomarelative to other tumor types (p 0.0006)(Flickinger et al., 1994). Radiosurgery is thus an ap-parently effective modality for treating the so-calledradioresistant tumor histologies such as solitarybrain metastases from renal cell carcinoma and melanoma.

The subject of re-irradiation often arises when pa-tients who have already received WBRT develop new,persistent, or recurrent brain metastases. Loeffler andco-workers (1990) treated 18 patients who had 21recurrent or persistent brain metastases with radio-surgery. Patient eligibility requirements for treatmentwere having a KPS score that was 70 and havingstable systemic disease. With a reported median pa-tient follow up of 9 months, all tumors in the radio-surgery field were controlled, and no cases of symp-tomatic radiation necrosis occurred despite previousradiation treatment (Loeffler et al., 1990). A study todetermine the maximum tolerable dose of single-frac-tion radiosurgery in patients with previously irradi-ated primary brain tumors or brain metastases wascarried out by the RTOG study 90-05 (Shaw et al.,1996). There were 156 analyzable patients, 36% ofwhom had recurrent primary brain tumors (medianprior dose 60 Gy) and 64% of whom had recur-rent brain metastases (median prior dose 30 Gy),lesions that were all less than or equal to 40 mm inmaximum diameter. Initially, patients were enteredinto arms of the study based on the maximum di-ameter of their recurrent lesion: tumors of 20 mm


3601_e13_p320-340 2/15/02 4:57 PM Page 330


or less received 18 Gy; tumors ranging from 21 to30 mm received 15 Gy; and those ranging from 31to 40 mm received 12 Gy. Dose escalation was latercarried out such that tumors smaller than 20 mmreceived 21 Gy; tumors between 21 and 30 mm re-ceived 18 Gy; and those between 31 and 40 mm re-ceived 15 Gy. Unacceptable acute toxicity secondaryto cerebral edema was observed in 0%, 7%, and 5%of patients, respectively, in the first group. In thesecond, dose escalation, group, no unacceptableacute toxicity was seen. Multivariate analysis showedthat maximum tumor diameter was one variable as-sociated with significantly increased risk to patientswith grade 3, 4, or 5 neurotoxicity. Specifically, pa-tients with tumors of 21 to 40 mm in diameter were7 to 16 times more likely to develop grade 3 to 5neurotoxicity than those who had tumors less than20 mm in diameter. Radiosurgery dose to the tumorwas also associated with neurotoxicity. The actuar-ial incidence of radionecrosis was 5%, 8%, 9%, and11% at 6, 12, 18, and 24 months after radiosurgery,respectively (Shaw et al., 1996, 2000). These re-sults should be used as guidelines for dose se-lection to minimize unacceptable toxicities from radiosurgery.

Of note, in the final report of this study, a com-parison between linear-acceleratorbased radio-surgery and gamma knife radiosurgery results wasmade. The results, while interesting, cannot be mean-ingfully interpreted because of differences in the char-acteristics of the two groups of patients and the factthat this study was not originally designed to compareradiosurgical treatment units (Buatti et al., 2000).

The largest published series on external beam re-irradiation of brain metastases included 86 patientsfrom the Mayo Clinic (Wong et al., 1996). The firstcourse of radiation that was employed provided a me-dian dose of 30 Gy, followed by 20 Gy for re-treat-ment. Patient median survival time after re-irradia-tion was 4 months; 27% of patients showed totalsymptom resolution; 43% experienced partial reso-lution; and 29% showed no change or had worsen-ing of neurologic symptoms. There was no significanttoxicity related to re-irradiation in the majority of pa-tients. The only factor associated with improved sur-vival on multivariate analysis was the absence of ex-tracranial disease.

Another series, from New York University, included52 patients selected for re-irradiation of recurrentcerebral metastases who were in relatively good med-ical condition, had an interval of at least 4 months

from the initial course of radiation, and were expe-riencing a renewed deterioration in neurologic con-dition. Patients had initially been prescribed a doseof 30 Gy in 10 fractions over 2 weeks, whereas re-irradiation consisted of 25 Gy in 10 fractions. Therewas a 42% response rate to re-irradiation, and sur-vival after the second treatment averaged 5 months(Cooper et al., 1990). If eligible, patients with re-current brain metastases should be offered radio-surgery as the treatment of choice. If radiosurgery isnot feasible, then whole brain re-irradiation may becarefully considered for patients who are highly mo-tivated and have selection criteria of age, KPS status,absence of extracranial disease, and time interval be-tween treatments in their favor.

Prophylactic cranial radiation therapy (PCI)should be considered for patients with limited-stagesmall cell lung cancer in complete remission becausethere is a 20% to 25% incidence of developing brainmetastases subsequent to initial diagnosis (Ihde et al.,1997). Debate over PCI relates to concern that thetreatment may itself directly cause neurologic deficits.The endpoints to be measured are quality of life andsurvival. The Prophylactic Cranial IrradiationOverview Collaboration Group published a meta-anal-ysis of 987 patients with small cell lung cancer incomplete remission based on seven trials that ran-domized patients to receive PCI or no PCI (Auperinet al., 1999). The relative risk of death in the treat-ment group compared with the control groups was0.84, corresponding to a 5.4% increase in the rate ofsurvival at 3 yearsfrom 15.3% in the control groupto 20.7% in the treatment group. Prophylactic cranialirradiation decreased the cumulative incidence ofbrain metastasis with a relative risk of 0.46 (p 0.001). Addressing the concern of neurocognitive im-pairment in patients who have undergone PCI, the in-vestigators of the two largest trials included in themeta-analysis performed neuropsychological testingon most patients before and after treatment. Neu-rocognitive impairment was often detected at diag-nosis, but no deterioration was found after PCI (Ar-riagada et al., 1995; Gregor et al., 1997). Themeta-analysis makes a strong case that PCI should beincluded as standard care for all patients with smallcell lung cancer in complete remission. Prophylacticcranial irradiation should not be given concurrentlywith chemotherapy, to avoid increased neurotoxicity.

The acute side effects of WBRT can include mildfatigue, reversible hair loss, and mild scalp erythemaas well as hyperpigmentation. Of greater concern is

3601_e13_p320-340 2/15/02 4:57 PM Page 331


the development of somnolence syndrome, describedas persistent fatigue, anorexia, and irritability (espe-cially in children), which may occur 3 to 10 weeksafter treatment but may resolve within 6 weeks(Littman et al., 1984). Long-term survivors may be atrisk of developing the late effects of WBRT. Progres-sive dementia, ataxia, and urinary incontinence werereported 5 to 36 months after WBRT in a series of 12patients (DeAngelis et al., 1989a). Correlative CT find-ings identified cortical atrophy and hypodense whitematter changes in these patients. Analysis of this studyreveals that large radiation fractions (3 to 6 Gy) mayhave contributed to an increased incidence of latetoxicities associated with WBRT. Based on this report,smaller fraction sizes of 1.8 to 2.5 Gy should be con-sidered for patients who are expected to live longerthan average because of favorable prognostic factors.

In conclusion, the effectiveness of WBRT in palli-ating brain metastases has been established by mul-tiple RTOG trials. Prognostic factors identified by theRTOG that predict a greater life expectancy should beused to select patients for more aggressive treatmentsthan WBRT alone, such as radiosurgery and surgery.This is important because control of neurologic dis-ease is likely to impact overall survival. Also, long-term survivors are at increased risk of realizing thesequelae of radiation therapy, so that a patient be-longing to the most favorable risk group should begiven a more protracted course of radiation usingsmaller fraction sizes. Additional randomized trialsthat address quality of life in addition to traditionalendpoints will be necessary to further evaluate the re-spective roles of radiosurgery and surgery and to de-termine the best way to combine these therapies withWBRT to maximize neurologic survival and quality oflife. We hope that the data presented will help thereader make evidence-based decisions in the com-plex management of brain metastasis, the most com-mon tumor type afflicting the brain.

Surgery

Although radiation therapy is frequently employed inthe treatment of brain metastases, surgical removalof the tumor mass, whether single or multiple, maybe the most effective palliation, especially for tumorsfrom radioresistant diseases such as melanoma andcarcinomas of the kidney and colon (Galicich and Ar-bit, 1990; Lang and Sawaya, 1996; Lang and Sawaya,1998; Sawaya et al., 2000). Modern neurosurgicaltechniques and perioperative care have changed sur-

geons perception of surgery for brain metastasesover the past 30 years. With the potential for increasedbenefit that surgical resection offers, this treatmenthas become a routine consideration for certain pa-tients.

Series examining heterogeneous groups of patientshave revealed median survival times extending from10 to 14 months for patients treated surgically for asingle metastasis (Decker et al., 1984; Sundaresanand Galicich, 1985b; Ferrara et al., 1990; Patchell etal., 1990; Bindal et al., 1993). Previously, surgery formultiple metastases had not been considered as anoption (Young and Patchell, 1990; Patchell, 1991),but one study of patients treated surgically for multi-ple metastases showed that the median survival ex-tended to 10 months (Bindal et al., 1993). Becauseresection eliminates the neoplasm and the source ofbrain edema, surgery should be considered for large,symptomatic tumors, whereas removal of smallerasymptomatic lesions may be delayed until they be-come symptomatic or show rapid growth that wouldpredict the onset of symptoms.

Numerous retrospective studies have confirmedthat surgery with WBRT is more effective than WBRTalone. With the combined treatment, recurrence atthe original site of metastasis occurs significantly lessfrequently and functional independence is signifi-cantly longer. Reports show that for patients with noother detectable evidence of disease at the time ofcraniotomy, the median survival after surgery is sig-nificantly increased (Galicich et al., 1980b). In astudy of 33 patients who underwent surgical resec-tion and postoperative WBRT for solitary brain metas-tases, Galicich and co-workers (1980a) found a lowincidence of recurrence, a median survival time of 8months, and a 1 year survival rate of 44%. A largerstudy of 78 patients conducted by the same groupshowed a median survival of 6 months, with a 1 yearsurvival rate of 29% (Galicich et al., 1980b).

In two other studies, patients with cancer and asingle brain metastasis were prospectively random-ized to receive either surgery followed by WBRT orWBRT alone (Patchell et al., 1990; Vecht et al., 1993).In the study by Patchell and co-workers, the groupundergoing surgery had fewer instances of recur-rence (20% versus 52%, respectively), a significantlylonger median survival (40 versus 15 weeks, respec-tively), and longer functional independence (38 ver-sus 8 weeks, respectively). Similarly, Vechts groupdemonstrated that surgery plus WBRT was superiorto WBRT alone for the treatment of patients with sin-

3601_e13_p320-340 2/15/02 4:57 PM Page 332

gle brain metastases (Vecht et al., 1993). More re-cently, Patchell and co-workers (1998) revealed thatpatients undergoing complete surgical resection ofsingle brain metastases who also receive postopera-tive radiotherapy are subject to fewer tumor recur-rences in the brain and die less frequently of neuro-logic causes than similar patients not receiving suchradiotherapy; however, survival was not extended forpatients receiving radiotherapy in addition to surgery.

Surgery is also beneficial for (1) controlling neu-rologic symptoms for patients with a functional sta-tus that dictates further treatment, (2) confirming adiagnosis that is questionable, (3) providing imme-diate palliation, (4) offering increased tolerance toand physician flexibility in using alternative treat-ments, (5) providing more durable local control, and(6) confirming or ascertaining the presence of metas-tases.

Patchell and co-workers (1990) noted that 6 of54 patients initially considered to have brain metas-tases based on results from CT or MRI scans werefound at surgery (or biopsy) to have other lesions,and 3 had no neoplastic disorder at all. These errorscannot always be precluded by diagnostic imagingand can be detected only by microscopic examina-tion of tissue (Patchell et al., 1990).

Surgical considerations are based mainly on ac-cessibility and resectability. Accessibility has been de-fined as the risk and extent of neurologic injury thepatient is willing to accept (Moser and Johnson,1989). The location of the lesion affects potentialpostoperative complications: lesions located in ornear the motor cortex and Brocas speech area re-quire particular care to avoid paresis or dysphasia;lesions located in the visual cortex can produce tem-porary or permanent visual deficits. For some of theselesions, techniques such as cortical mapping can beuseful in minimizing damage to motor, sensory, andspeech areas (Landy and Egnor, 1991), but carefulpre- and intraoperative localization is vital, and eachlesion must be considered individually. Lesions lo-cated deep within the brain parenchyma have tradi-tionally been considered unresectable, but moderntechniques, mainly intraoperative ultrasonographyand stereotactic approaches, now allow access tothese lesions (Kelly et al., 1988; Lange et al., 1990).

A factor that plays a significant role in consideringresectability is the number of metastases. As late as1990, most neuro-oncologists and neurosurgeonsconsidered surgery for multiple metastases justifiedin only rare instances. The consensus was that the

presence of multiple lesions strongly contraindicatedsurgery and that the circumstances precipitating therare decisions to operate on patients with multiplemetastases were limited to (1) a life-threatening masseffect on the brain stem (associated with an asymp-tomatic or relatively radiosensitive supratentorial le-sion); (2) a large, life-threatening radiosensitivesupratentorial lesion; or (3) two or more lesions thatcould be removed through a single cranial opening(Young and Patchell, 1990). Modern technology,however, is expanding surgical options.

At M. D. Anderson, we evaluated the results of sur-gery in patients with multiple lesions and found thatsurgery can play a very important role in managingthese patients (Bindal et al., 1993). Fifty-six patientswho underwent surgery for multiple brain metastaseswere divided into two groups: group A, those patientswho had one or more lesions remaining after surgery(N 30); and group B, those patients who had alllesions removed (N 26). Patients in group B werematched by type of primary tumor, presence or ab-sence of systemic disease, and time from first diag-nosis of cancer to diagnosis of brain metastases to athird group of patients (group C, N 26), under-going surgery for a single lesion. Median survivaltimes were 6, 14, and 14 months for patients ingroups A, B, and C, respectively. Besides the signifi-cant difference in survival between groups A and B(p 0.003) and between groups A and C (p 0.012) there was a significant correspondence in re-currence or neurologic improvement rates betweengroups B and C, indicating that surgery for patientswith multiple metastatic lesions that can all be re-moved is as effective as surgery for a single lesion.Although the results appear intuitively correct, alarger study may be required to confirm and expandthese findings.

Patients with multiple brain metastases in whomall lesions cannot be surgically excised may also becandidates for surgery if resection of one or morehighly symptomatic, debilitating, or life-threateninglesions will result in greater, more rapid palliation ofsymptoms than might be achieved by irradiationalone.

Although the microsurgical techniques used for theremoval of brain metastases are largely the same asthose used for the removal of other intracranial le-sions, surgery is complicated by the generally smallsize of the tumors and the tendency of a cerebral me-tastasis to cause extensive edema with resulting neu-rologic symptoms, a factor that accounts for early di-


3601_e13_p320-340 2/15/02 4:57 PM Page 333

agnosis but contributes to the difficulty in locatingmetastases if they are not superficial. Modern imag-ing techniques allow for more efficacious treatmentof asymptomatic tumors that may be discovered at thetime that systemic cancers are first diagnosed andthat, consequently, may still be quite small. Computer-assisted stereotactic and/or intraoperative ultrasoundtechniques (1) allow surgeons to precisely locate thetumor before making the cortical incision; (2) pro-vide a direct route to the tumor that avoids eloquentareas of the brain; (3) aid surgeons in placing thebone flap precisely over the tumor location; and (4)eliminate the need to extensively probe for an elusivetumor, which can result in neurologic damage.

For patients experiencing considerable mass effectfrom the tumor, repeated preoperative CT scans helpmonitor the edema, which can be reduced in mostcases by administering steroids. The preoperative ad-ministration of corticosteroids for a minimum of 48hours helps prevent transdural herniation when thetumor is exposed. Administration of diuretics shouldbe undertaken with caution because they reduce theextracellular volume and induce hyponatremia andvasoconstriction. Effective control of edema duringsurgery can be achieved by using a proper, safe anes-thesiologic procedure (Gambardella et al., 1990).Surgery is limited by the functional importance of thebrain tissue to be traversed. Another major consid-eration is the expected quality of survival, a subjec-tively determined factor. For one patient, a fewmonths of restored neurologic function might be veryimportant, whereas it might be of minimal signifi-cance for another patient.

Surgery and imaging technology came together in1986 in the first successful removal of a lesion met-astatic to the tectum of the midbrain (Tobler et al.,1986). Although metastasis to this site is a very rareoccurrence (1% to 3% of brain metastases) and waspreviously considered an inoperable location, the tu-mor was vaporized with a carbon dioxide laser beam(up to 20 watts) and removed by a central coringtechnique. For patients with systemic cancer that can-not be controlled, management decisions for asymp-tomatic brain tumors are moot.

As noted previously, early neurosurgical attemptsmet with high rates of complications due to unso-phisticated radiographic methods and a limited abil-ity to control brain herniation. Modern advances, in-cluding the use of corticosteroids and modernanesthesia, the advent of CT and MRI, the use of the

surgical microscope, and the development of intra-operative ultrasound, stereotactic localization, andcortical mapping, have significantly reduced opera-tive mortalities and morbidities (Sawaya et al., 1998).Postoperative mortality is most often due to uncon-trolled systemic cancer, but comparisons of resultsfrom gross total removal and partial removal of brainmetastases indicate that the former yields the lowestrate of operative mortality and that the 30 day mor-tality risk may be doubled in cases of partial removal(Haar and Patterson, 1972). A recent study of neu-rosurgical outcomes in a series of 400 craniotomiesperformed for removal of brain metastases (48%) orgliomas (52%) (Sawaya et al., 1998) determined thatgross total resection of most tumors (73%) did notlead to more major neurologic deficits than were ob-served for patients undergoing subtotal or partial re-section. Nonfatal complications such as hematomas,wound infections, and pseudomeningocele forma-tions, which occur in 8% to 9% of all craniotomiesfor brain metastases (Bindal et al., 1993), as well assurgically induced neurologic impairments, are usu-ally transient events. Clinically evident thromboem-bolic complications, such as deep vein thrombosis orpulmonary embolism, occur in an estimated 10% ofpatients (Constantini et al., 1991; Sawaya et al.,1992). Mortality and morbidity rates have been re-duced to 3% or less and 5% or less, respectively (Sun-daresan and Galicich, 1985a; Brega et al., 1990;Patchell et al., 1990; Bindal et al., 1993).

Chemotherapy

Systemic chemotherapy has historically been consid-ered ineffective in the treatment of brain metastases,with several reasons being given for this presumedfailure (Buckner, 1991). The BBB has been assumedto be a major restriction to the CNS entry of many cy-totoxic drugs that are large polar or hydrophobiccompounds. Unlike normal brain capillaries, tumorcapillaries are variably disrupted in patients who havebrain metastases (and high-grade primary brain tu-mors). Evidence that the tumor capillaries in mostmetastatic tumors are disrupted is shown by the factthat virtually all metastatic tumors in the brain arecontrast enhancing, reflecting the leakage of contrastmaterial from the tumor vasculature to the intersti-tium. Drugs (cisplatin, etoposide, nimustine, and az-iquinone) administered before surgery and measured


3601_e13_p320-340 2/15/02 4:57 PM Page 334

in tumor samples removed at surgery have consis-tently demonstrated pharmacologically relevant levelsin these tissues, suggesting that these agents are ableto penetrate the tumor tissue in the brain when de-livered systemically (Stewart et al., 1979; Stewart etal., 1982; Savaraj et al., 1983, 1984). Most animalstudies, however, have shown lower capillary perme-ability and lower drug concentrations in brain tumorsthan in subcutaneous tumors for systemically admin-istered chemotherapy (Stewart, 1994).

Lipophilic drugs such as nitrosoureas or othersemisynthetic agents may be able to deliver evenhigher levels of drugs to the tumor periphery wherethe BBB may remain relatively intact (Levin et al.,1975, 1976). Some drugs may, however, inadequatelypenetrate tumor regions and will achieve subthera-peutic levels intracellularly (Levin et al., 1980). Theobservation of CNS relapse before relapse at othersites in acute leukemia and small cell lung cancer isoften cited as evidence of the importance of the BBBin inhibiting the effectiveness of chemotherapy.Bloodbrain barrier disruption with hyperosmolaragents, to increase chemotherapy delivery to brain,has been developed for primary brain tumors but hasnot yet been investigated for brain metastasis treat-ment.

The routine use of corticosteroids, which are ca-pable of re-establishing disrupted BBB function, forsymptomatic brain metastasis may provide an addi-tional protective effect against cytotoxic agents(Weller et al., 1997; Mariotta et al., 1999) and mayfurther limit drug delivery into CNS tumors (Naka-gawa et al., 1987).

Cancers that most frequently metastasize to thebrain, such as nonsmall cell lung cancer and ma-lignant melanoma, are often inherently insensitive tosystemic chemotherapy. Many patients develop brainmetastases in the face of widespread systemic relapseand/or after failure of several prior treatment regi-mens, including radiotherapy and chemotherapy.While long-duration chemotherapy treatment can in-crease the frequency of acquired resistance to che-motherapy agents for brain tumor metastases becauseof increased expression of efflux pumps such as theP-glycoprotein, encoded by the multidrug-resistance-1 gene, nonetheless, the intact BBB already has func-tioning efflux pumps such as the P-glycoprotein, rais-ing the question whether this is truly an acquiredmechanism or a common de novo mechanism fordrug failure of tumors in the brain.

Several lines of evidence suggest that a number ofdifferent chemotherapy regimens may be at least pal-liative with respect to some brain metastases (Greig,1984). An increasing number of clinical trials havedemonstrated response rates for brain metastases thatare in keeping with those seen for systemic metas-tases using the same regimens, especially for breastand lung carcinomas.

Several clinical trials have reported the activity ofsingle-agent and combination chemotherapy in sometypes of brain metastases. Breast carcinomas, in gen-eral, are considered chemosensitive tumors. Patientswith extracranial metastatic breast carcinomas whohave not had prior chemotherapy can achieve 50%to 70% response rates with combination chemother-apy, but only 20% to 30% of patients who failed priorchemotherapy will respond to second-line salvagechemotherapy. Rosner and coworkers (1986) dem-onstrated a response rate of 50% using cy-clophosphamide, 5-fluorouracil, and prednisone in100 breast cancer patients with cerebral metastasis.The best result reported to date has been with a five-drug combinationcyclophosphamide, methotrex-ate, fluorouracil, vincristine, and prednisone (CM-FVP) with which Rosner and colleagues (1986) found10% complete and 40% partial response rates for pa-tients not receiving prior chemotherapy. There weresome long-term survivors, but the median durationof response was only 7 months, and survival was sim-ilarly quite short. Several points should be empha-sized in this study: (1) patients with brain metastasesresponded to combination chemotherapy at a ratesimilar to that experienced by patients treated for ex-tracranial metastases; (2) patients who failed initialchemotherapy did respond, albeit at a lower rate, tosecond- or third-line regimens; (3) all patients re-ceived prednisone, an antiedema agent that may in-terfere with the response interpretation because of aprednisone-mediated decrease in edema and mass ef-fect as visualized on neuroimaging studies; and (4)the high response rates have not been reproduced byother investigators (Flowers and Levin, 1993).

Other drug regimens, including etoposide and cis-platin (Cocconi et al., 1990; Franciosi et al., 1999)and CAF (cyclophosphamide, doxorubicin, and fluo-rouracil) have shown similar activities for breast car-cinoma brain metastases with response rates in therange of 40% to 50%. A multidrug combination calledTPDC-FuHu (6-thioguanine, procarbazine, dibro-modulcitol, lomustine, fluorouracil, and hydrox-


3601_e13_p320-340 2/15/02 4:57 PM Page 335

yurea) that focused on overcoming nitrosourea re-sistance and potentiating nitrosourea tumor cell killdemonstrated similar results for patients with brainmetastases from breast and nonsmall cell lung car-cinoma and considerably better results for patientswith small cell lung carcinoma who failed prior ra-diation therapy for these metastases. For patients withbrain metastases from breast, nonsmall cell andsmall cell lung carcinoma, response rates were 36%,26%, and 67%, respectively. The corresponding dis-ease-free survival periods for these responding pa-tient subgroups were 27, 21, and 133 weeks, re-spectively (Kaba et al., 1997).

Among the various lung cancers, small cell carci-noma is the most sensitive to chemotherapy. Lee andco-workers (1989) reported the use of cyclophos-phamide, doxorubicin, vincristine, and etoposide asprimary chemotherapy for 15 patients with small celllung cancer who presented with brain metastases.Nine of 11 evaluable patients (82%) showed com-plete or partial responses. In another study, Twelvesand co-workers (1990) treated 14 patients who hadbrain metastases from small cell lung cancer at pre-sentation with cyclophosphamide, vincristine, andetoposide. Nine patients (64%) responded. In contrast,results from small studies in which chemotherapycombinations with etoposide and cisplatin (Croisileet al., 1992), fotemustine plus cisplatin (Cotto et al.,1996), and lomustine, carboplatin, vinorelbine, andfluorouracil (Colleoni et al., 1997) were used to treatbrain metastases from nonsmall cell lung carcinoma(a less chemosensitive tumor type) were much lessimpressive. Response rates for patients in these stud-ies were 0%, 14%, and 33%, respectively, to chemo-therapy. Robinet and colleagues (1991) reported a50% response rate using fluorouracil and cisplatin;however, this result has not yet been duplicated.These results suggest that the response of cerebralmetastases to chemotherapy may depend on the in-herent chemosensitivity of the primary cancer type.

Historically, malignant melanoma has been shownto be extremely insensitive to chemotherapy, withbrain metastases from melanoma being no exception.However, a meeting abstract in 1996 reported thatuse of cisplatin combined with interleukin-2 and in-terferon-2a to treat brain metastases from melanomahad shown improved results with a response rate of39% and a median survival time of 32 weeks(Mousseau et al., 1996).

Several studies of brain metastases from other sys-temic chemosensitive cancers, such as choriocarci-noma and germinoma, have demonstrated high re-sponse rates to combination chemotherapy regimensthat are deemed active for the systemic cancers. In onestudy, 13 of 18 (72%) patients with choriocarcinomaand brain metastasis responded to primary chemo-therapy with etoposide, methotrexate, dactinomycin,vincristine, cyclophosphamide, and cisplatin (Rustin etal., 1989), whereas in another study, 8 of 10 patients(80%) diagnosed with germinoma and brain metas-tases achieved complete response to cisplatin, vin-cristine, methotrexate, bleomycin, etoposide, dactino-mycin, and cyclophosphamide (Rustin et al., 1986).In gestational trophoblastic disease metastatic to thebrain, combination chemotherapy with etoposide,methotrexate, vincristine, actinomycin-D, and cyclo-phosphamide as sole therapy can be curative.

The role of chemotherapy in the overall manage-ment of patients with brain metastases remains un-der investigation. Chemotherapy should be consid-ered for chemosensitive tumors, keeping in mind thatsurgery and radiotherapy remain the primary treat-ment modalities. There is insufficient positive experi-ence to support the general use of chemotherapy inpatients with brain metastases; therefore, chemo-therapy should be considered as palliative and shouldbe given under the auspices of appropriately designedclinical trials. The choice of the drug or drug com-bination should be guided by the chemosensitivityprofile of the primary systemic cancer. Future treat-ment approaches may involve pre-radiation chemo-therapy for patients who have minimal neurologicsymptoms, which may reduce the tumor burden inthe brain, allowing more prolonged control after ra-diation therapy and, perhaps, decreased radiationneurotoxicity, if the radiation dose can be reduced.

The role of the BBB in restricting drug entry intobrain regions adjacent to a tumor probably playssome role in the efficacy of chemotherapy for brainmetastases. The use of corticosteroids by these pa-tients should be carefully limited as much as possi-ble to allow maximal benefit from chemotherapeuticagents while maintaining neurologic function. Ad-justments in steroid dose must be taken into consid-eration in clinical trial design, as the dose and thetiming of steroid administration may partially rectifya leaky tumor vasculature and falsely produce a re-sponse on CT or MRI scans.


3601_e13_p320-340 2/15/02 4:57 PM Page 336


REFERENCES

Alexander E 3rd, Moriarty TM, Davis RB, et al. 1995. Stereo-tactic radiosurgery for the definitive, noninvasive treatmentof brain metastases. J Natl Cancer Inst 87:3440.

Armstrong JG, Wronski M, Galicich J, Arbit E, Leibel SA, BurtM. 1994. Postoperative radiation for lung cancer metasta-tic to the brain. J Clin Oncol 12:23402344.

Arriagada R, Le Chevalier T, Borie F, et al. 1995. Prophylac-tic cranial irradiation for patients with small-cell lung can-cer in complete remission. J Natl Cancer Inst 87:183190.

Auperin A, Arriagada R, Pignon JP, et al. 1999. Prophylacticcranial irradiation for patients with small-cell lung cancerin complete remission. Prophylactic Cranial IrradiationOverview Collaborative Group. N Engl J Med 341:476484.

Bindal RK, Sawaya R, Leavens ME, Lee JJ. 1993. Surgical treat-ment of multiple brain metastases. J Neurosurg 79:210216.

Borgelt B, Gelber R, Kramer S, et al. 1980. The palliation ofbrain metastases: final results of the first two studies by theRadiation Therapy Oncology Group. Int J Radiat Oncol BiolPhys 6:19.

Brega K, Robinson WA, Winston K, Wittenberg W. 1990. Sur-gical treatment of brain metastases in malignant melanoma.Cancer 66:21052110.

Buatti JM, Friedman WA, Meeks SL, Bova FJ. 2000. RTOG 90-05: the real conclusion. Int J Radiat Oncol Biol Phys47:269271.

Buckner JC. 1991. The role of chemotherapy in the treatmentof patients with brain metastases from solid tumors. Can-cer Metastasis Rev 10:335341.

Byrne TN, Cascino TL, Posner JB. 1983. Brain metastasis frommelanoma. J Neurooncol 1:313317.

Cairncross JG, Posner JB. 1981. Neurological complicationsof systemic cancer. In: Yarbro JW, Bornstein RS (eds), On-cologic Emergencies. New York: Grune & Stratton, p 73.

Cairncross JG, Posner JB. 1983. The management of brainmetastases. In: Walker MD (ed), Oncology of the NervousSystem. Boston: Martinus Nijhof, p. 341377.

Cascino TL, Leavengood JM, Kemeny N, Posner JB. 1983. Brainmetastases from colon cancer. J Neurooncol 1:203209.

Cavanaugh PG, Nicolson GL. 1991. Lung-derived growth factorthat stimulates the growth of lung-metastasizing tumor cells:identification as transferrin. J Cell Biochem 47:261271.

Chao J, Phillips R, Nickson J. 1954. Roentgen-ray therapy ofcerebral metastases. Cancer 7:682689.

Cho KG, Hoshino T, Pitts LH, Nomura K, Shimosato Y. 1988.Proliferative potential of brain metastases. Cancer 62:512515.

Chu FCH, Hilaris BB. 1961. Value of radiation therapy in themanagement of intracranial metastases. Cancer 14:577.

Cocconi G, Lottici R, Bisagni G, et al. 1990. Combination ther-apy with platinum and etoposide of brain metastases frombreast carcinoma. Cancer Invest 8:327334.

Cohen N, Strauss G, Lew R, Silver D, Recht L. 1988. Shouldprophylactic anticonvulsants be administered to patientswith newly-diagnosed cerebral metastases? A retrospectiveanalysis. J Clin Oncol 6:16211624.

Coia LR, Hanks GE, Martz K, Steinfeld A, Diamond JJ, KramerS. 1988. Practice patterns of palliative care for the United

States 19841985. Int J Radiat Oncol Biol Phys 14:12611269.

Colleoni M, Graiff C, Nelli P, et al. 1997. Activity of combina-tion chemotherapy in brain metastases from breast and lungadenocarcinoma. Am J Clin Oncol 20:303307.

Constantini S, Kornowski R, Pomeranz S, Rappaport ZH. 1991.Thromboembolic phenomena in neurosurgical patients op-erated upon for primary and metastatic brain tumors. ActaNeurochir (Wien) 109:9397.

Cooper JS, Steinfeld AD, Lerch IA. 1990. Cerebral metastases:value of reirradiation in selected patients. Radiology174:883885.

Cotto C, Berille J, Souquet PJ, et al. 1996. A phase II trial offotemustine and cisplatin in central nervous system metas-tases from nonsmall cell lung cancer. Eur J Cancer32A:6971.

Croisile B, Trillet-Lenoir V, Catajar JF, et al. 1992. Cerebralmetastasis disclosing primary bronchogenic cancers. RevNeurol (Paris) 148:488492.

DeAngelis LM, Delattre JY, Posner JB. 1989a. Radiation-in-duced dementia in patients cured of brain metastases. Neu-rology 39:789796.

DeAngelis LM, Mandell LR, Thaler HT, et al. 1989b. The roleof postoperative radiotherapy after resection of single brainmetastases. Neurosurgery 24:798805.

Decker DA, Decker VL, Herskovic A, Cummings GD. 1984.Brain metastases in patients with renal cell carcinoma:prognosis and treatment. J Clin Oncol 2:169173.

Delattre JY, Krol G, Thaler HT, Posner JB. 1988. Distributionof brain metastases. Arch Neurol 45:741744.

Diener-West M, Dobbins TW, Phillips TL, Nelson DF. 1989.Identification of an optimal subgroup for treatment evalu-ation of patients with brain metastases using RTOG study7916. Int J Radiat Oncol Biol Phys 16:669673.

DiStefano A, Yong Yap Y, Hortobagyi GN, Blumenschein GR.1979. The natural history of breast cancer patients withbrain metastases. Cancer 44:19131918.

Dosoretz DE, Blitzer PH, Russell AH, Wang CC. 1980. Man-agement of solitary metastasis to the brain: the role of elec-tive brain irradiation following complete surgical resection.Int J Radiat Oncol Biol Phys 6:17271730.

Ferrara M, Bizzozzero L, Talamonti G, DAngelo VA. 1990. Sur-gical treatment of 100 single brain metastases. Analysis ofthe results. J Neurosurg Sci 34:303308.

Figlin RA, Piantadosi S, Feld R. 1988. Intracranial recurrenceof carcinoma after complete surgical resection of stage I,II, and III nonsmall-cell lung cancer. N Engl J Med318:13001305.

Flickinger JC, Kondziolka D, Lunsford LD, et al. 1994. A multi-institutional experience with stereotactic radiosurgery forsolitary brain metastasis. Int J Radiat Oncol Biol Phys28:797802.

Flowers A, Levin VA. 1993. Management of brain metastasesfrom breast carcinoma. Oncology 7:2126.

Franciosi V, Cocconi G, Michiara M, et al. 1999. Front-linechemotherapy with cisplatin and etoposide for patients withbrain metastases from breast carcinoma, nonsmall cell lungcarcinoma, or malignant melanoma: a prospective study.Cancer 85:15991605.

3601_e13_p320-340 2/15/02 4:57 PM Page 337


Galicich J, Arbit E. 1990. Metastatic brain tumors. In: YoumansJR (ed), Neurological Surgery. Philadelphia: WB Saunders,pp 32043222.

Galicich JH, Sundaresan N, Arbit E, Passe S. 1980a. Surgicaltreatment of single brain metastasis: factors associated withsurvival. Cancer 45:381386.

Galicich JH, Sundaresan N, Thaler HT. 1980b. Surgical treatmentof single brain metastasis. Evaluation of results by computer-ized tomography scanning. J Neurosurg 53:6367.

Gambardella G, Santamaria LB, Toscano SG, et al. 1990. Brainmetastases: surgical versus conservative treatment. J Neu-rosurg Sci 34:309314.

Gaspar L, Scott C, Rotman M, et al. 1997. Recursive parti-tioning analysis (RPA) of prognostic factors in three Radi-ation Therapy Oncology Group (RTOG) brain metastasestrials. Int J Radiat Oncol Biol Phys 37:745751.

Gregor A, Cull A, Stephens RJ, et al. 1997. Prophylactic cra-nial irradiation is indicated following complete response toinduction therapy in small cell lung cancer: results of a mul-ticentre randomised trial. United Kingdom CoordinatingCommittee for Cancer Research (UKCCCR) and the Euro-pean Organization for Research and Treatment of Cancer(EORTC). Eur J Cancer 33:17521758.

Greig NH. 1984. Chemotherapy of brain metastases: currentstatus. Cancer Treat Rev 11:157186.

Grossman SA, Sheidler VR, Gilbert MR. 1989. Decreasedphenytoin levels in patients receiving chemotherapy. Am JMed 87:505510.

Haar F, Patterson RH Jr. 1972. Surgery for metastatic in-tracranial neoplasm. Cancer 30:12411245.

Hagen NA, Cirrincione C, Thaler HT, DeAngelis LM. 1990. Therole of radiation therapy following resection of single brainmetastasis from melanoma. Neurology 40:158160.

Ihde DC, Pass HI, Glatstein E. 1997. Small cell lung cancer.In: DeVita VT, Hellman S, Rosenberg SA (eds), Cancer:Principles and Practice of Oncology. Philadelphia: Lippin-cott-Raven, pp 911949.

Kaba SE, Kyritsis AP, Hess K, et al. 1997. TPDC-FuHu chemo-therapy for the treatment of recurrent metastatic brain tu-mors. J Clin Oncol 15:10631070.

Kelly PJ, Kall BA, Goerss SJ. 1988. Results of computed to-mography-based computer-assisted stereotactic resectionof metastatic intracranial tumors. Neurosurgery 22:717.

Komarnicky LT, Phillips TL, Martz K, Asbell S, Isaacson S, Ur-tasun R. 1991. A randomized phase III protocol for theevaluation of misonidazole combined with radiation in thetreatment of patients with brain metastases (RTOG-7916).Int J Radiat Oncol Biol Phys 20:5358.

Kondziolka D, Patel A, Lunsford LD, Kassam A, Flickinger JC.1999. Stereotactic radiosurgery plus whole brain radio-therapy versus radiotherapy alone for patients with multi-ple brain metastases. Int J Radiat Oncol Biol Phys45:427434.

Kurtz JM, Gelber R, Brady LW, Carella RJ, Cooper JS. 1981. Thepalliation of brain metastases in a favorable patient popula-tion: a randomized clinical trial by the Radiation Therapy On-cology Group. Int J Radiat Oncol Biol Phys 7:891895.

Landy HJ, Egnor M. 1991. Intraoperative ultrasonography andcortical mapping for removal of deep cerebral tumors.South Med J 84:13231326.

Lang FF, Sawaya R. 1996. Surgical management of cerebralmetastases. Neurosurg Clin North Am 7:459484.

Lang FF, Sawaya R. 1998. Surgical treatment of metastatic braintumors. Semin Surg Oncol 14:5363.

Lange OF, Scheef W, Haase KD. 1990. Palliative radio-chemo-therapy with ifosfamide and BCNU for breast cancer patientswith cerebral metastases. A 5-year experience. CancerChemother Pharmacol 26:S7880.

Lee JS, Murphy WK, Glisson BS, Dhingra HM, Holoye PY, HongWK. 1989. Primary chemotherapy of brain metastasis insmall-cell lung cancer. J Clin Oncol 7:916922.

Leksell L. 1951. The stereotaxic method and radiosurgery ofthe brain. Acta Chir Scand 102:316.

Levin VA, Freeman-Dove M, Landahl HD. 1975. Permeabilitycharacteristics of brain adjacent to tumors in rats. ArchNeurol 32:785791.

Levin VA, Landahl HD, Freeman-Dove MA. 1976. The applica-tion of brain capillary permeability coefficient measure-ments to pathological conditions and the selection of agentswhich cross the blood-brain barrier. J Pharmacokinet Bio-pharm 4:499519.

Levin VA, Patlak CS, Landahl HD. 1980. Heuristic modeling ofdrug delivery to malignant brain tumors. J PharmacokinetBiopharm 8:257296.

Lewis AJ. 1988. Sarcoma metastatic to the brain. Cancer61:593601.

Liotta LA, Steeg PS, Stetler-Stevenson WG. 1991. Cancer me-tastasis and angiogenesis: an imbalance of positive and neg-ative regulation. Cell 64:327336.

Littman P, Rosenstock J, Gale G, et al. 1984. The somnolencesyndrome in leukemic children following reduced dailydose fractions of cranial radiation. Int J Radiat Oncol BiolPhys 10:18511853.

Loeffler JS, Alexander E, 3rd. 1993. Radiosurgery for the treat-ment of intracranial metastases. In: Alexander E, 3rd, Lo-effler JS, Lunsford LD (eds), Stereotactic Radiosurgery. NewYork: McGraw-Hill, pp 197206.

Loeffler JS, Kooy HM, Wen PY, et al. 1990. The treatment ofrecurrent brain metastases with stereotactic radiosurgery.J Clin Oncol 8:576582.

Mariotta M, Perewusnyk G, Koechli OR, et al. 1999. Dexam-ethasone-induced enhancement of resistance to ionizingradiation and chemotherapeutic agents in human tumorcells. Strahlenther Onkol 175:392396.

Mehta MP, Meyers CA, Curran WJ, et al. 2000. XCYTRIN(motexafin gadolinium) and whole brain radiation for pa-tients with brain metastases: lead-in phase to randomizedtrialfinal results. Int J Radiat Oncol Biol Phys 48:3S:204:A184.

Mori Y, Kondziolka D, Flickinger JC, Kirkwood JM, AgarwalaS, Lunsford LD. 1998a. Stereotactic radiosurgery for cere-bral metastatic melanoma: factors affecting local diseasecontrol and survival. Int J Radiat Oncol Biol Phys42:581589.

Mori Y, Kondziolka D, Flickinger JC, Logan T, Lunsford LD.1998b. Stereotactic radiosurgery for brain metastasis fromrenal cell carcinoma. Cancer 83:344353.

Moser RP, Johnson ML. 1989. Surgical management of brainmetastases: how aggressive should we be? Oncology3:123128.

3601_e13_p320-340 2/15/02 4:57 PM Page 338

Mousseau M, Nerson F, Khayat D, et al. 1996. Sequentialchemoimmunotherapy with cisplatin (CddP), interleukin 2(IL2) and interferon alpha 2 a (IFNa) for melanoma brainmetastasis (MBM). In: Sixth International Congress on Anti-cancer Treatment, p 108. Paris, France.

Mrak RE. 1993. Origins of adenocarcinomas presenting as in-tracranial metastases. An ultrastructural study. Arch PatholLab Med 117:11651169.

Murray KJ, Scott C, Greenberg HM, et al. 1997. A randomizedphase III study of accelerated hyperfractionation versusstandard in patients with unresected brain metastases: a re-port of the Radiation Therapy Oncology Group (RTOG)9104. Int J Radiat Oncol Biol Phys 39:571574.

Nakagawa H, Groothuis DR, Owens ES, Fenstermacher JD, Patlak CS, Blasberg RG. 1987. Dexamethasone effects on [125I]albumin distribution in experimental RG-2gliomas and adjacent brain. J Cereb Blood Flow Metab7:687701.

Nicolson GL. 1982. Cancer metastasis. Organ colonization andthe cell-surface properties of malignant cells. Biochim Bio-phys Acta 695:113176.

Nicolson GL. 1988a. Cancer metastasis: tumor cell and hostorgan properties important in metastasis to specific sec-ondary sites. Biochim Biophys Acta 948:175224.

Nicolson GL. 1988b. Organ specificity of tumor metastasis: roleof preferential adhesion, invasion and growth of malignant cellsat specific secondary sites. Cancer Metastasis Rev 7:143188.

Order SE, Hellman S, Von Essen CF, Kligerman MM. 1968. Im-provement in quality of survival following whole-brain ir-radiation for brain metastasis. Radiology 91:149153.

Paget S. 1989. The distribution of secondary growths in can-cer of the breast. Cancer Metastasis Rev 8:98101.

Patchell RA. 1991. Brain metastases. Neurol Clin 9:817824.Patchell RA, Tibbs PA, Regine WF, et al. 1998. Postoperative

radiotherapy in the treatment of single metastases to thebrain: a randomized trial. JAMA 280:14851489.

Patchell RA, Tibbs PA, Walsh JW, et al. 1990. A randomizedtrial of surgery in the treatment of single metastases to thebrain. N Engl J Med 322:494500.

Phillips MH, Stelzer KJ, Griffin TW, Mayberg MR, Winn HR.1994. Stereotactic radiosurgery: a review and comparisonof methods. J Clin Oncol 12:10851099.

Phillips TL, Scott CB, Leibel SA, Rotman M, Weigensberg IJ.1995. Results of a randomized comparison of radiotherapyand bromodeoxyuridine with radiotherapy alone for brainmetastases: report of RTOG trial 89-05. Int J Radiat OncolBiol Phys 33:339348.

Pirzkall A, Debus J, Lohr F, et al. 1998. Radiosurgery alone orin combination with whole-brain radiotherapy for brainmetastases. J Clin Oncol 16:35633569.

Posner JB. 1992. Management of brain metastases. Rev Neu-rol (Paris) 148:477487.

Posner JB, Chernik NL. 1978. Intracranial metastases from sys-temic cancer. Adv Neurol 19:579592.

Pullar M, Blumbergs PC, Phillips GE, Carney PG. 1985. Neo-plastic cerebral aneurysm from metastatic gestational chori-ocarcinoma. Case report. J Neurosurg 63:644647.

Radley MG, McDonald JV, Pilcher WH, Wilbur DC. 1993. Latesolitary cerebral metastases from renal cell carcinoma: re-port of two cases. Surg Neurol 39:230234.

Ravussin P, Abou-Madi M, Archer D, et al. 1988. Changes inCSF pressure after mannitol in patients with and without el-evated CSF pressure. J Neurosurg 69:869876.

Robinet G, Gouva S, Clavier J, et al. 1991. Chemotherapy withcisplatin and 5-fluorouracil in inoperable brain metastasesof bronchopulmonary cancers. Bull Cancer (Paris)78:831837.

Rodeck U, Becker D, Herlyn M. 1991. Basic fibroblast growthfactor in human melanoma. Cancer Cells 3:308311.

Rosner D, Nemoto T, Lane WW. 1986. Chemotherapy inducesregression of brain metastases in breast carcinoma. Can-cer 58:832839.

Rustin GJ, Newlands ES, Bagshawe KD, Begent RH, CrawfordSM. 1986. Successful management of metastatic and pri-mary germ cell tumors in the brain. Cancer 57:21082113.

Rustin GJ, Newlands ES, Begent RH, Dent J, Bagshawe KD.1989. Weekly alternating etoposide, methotrexate, andactinomycin/vincristine and cyclophosphamide chemother-apy for the treatment of CNS metastases of choriocarcinoma.J Clin Oncol 7:900903.

Salbeck R, Grau HC, Artmann H. 1990. Cerebral tumor stag-ing in patients with bronchial carcinoma by computed to-mography. Cancer 66:20072011.

Sause WT, Crowley JJ, Morantz R, et al. 1990. Solitary brainmetastasis: results of an RTOG/SWOG protocol evaluationsurgery RT versus RT alone. Am J Clin Oncol13:427432.

Sause WT, Scott C, Krisch R, et al. 1993. Phase I/II trial of ac-celerated fractionation in brain metastases RTOG 85-28. IntJ Radiat Oncol Biol Phys 26:653657.

Savaraj N, Lu K, Feun LG, et al. 1983.

intracranial metastases.pdf

Documents