Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Brain metastasis is common in patients with advanced solid tumors, occurring in roughly 15% of all cancer patients. According to autopsy analyses, the incidence of brain metastasis is as high as 30% in patients with breast cancer, 40% in patients with lung cancer, and 75% in patients with melanoma ( ). With 170 000 new cases diagnosed annually in the United Sates alone, brain metastases are 10 times more common than primary intracranial tumors and can be associated with substantial morbidity and mortality.
Although central nervous system (CNS) metastases can develop from any primary cancer, the predilection for distant spread varies by cancer type. Roughly 50% of all CNS metastases arise from primary lung cancer, 20% from breast cancer, 15% from melanoma, and 5–10% from unknown primary cancers; CNS metastases from renal cell carcinoma, colorectal cancer, gynecologic cancers, and other miscellaneous cancers account for an additional 5–10% ( ). Interestingly, prostate, oropharyngeal, and non-melanoma skin cancers rarely spread to the CNS ( ).
Although new, more effective anticancer therapies have been developed over the past several decades, multiple studies report that the incidence of brain metastasis is rising. One hypothesized reason for this rise is that brain metastases are sequelae of newly developed highly efficacious selective therapies for systemic extracranial metastases. Because patients receiving these therapies live longer, they may have more time (and thus are more likely) to develop brain metastases: the CNS is considered a sanctuary site, protecting tumor cells from exposure to full-dose systemic agents. In addition, technologic advances in diagnostic imaging have likely helped increase detection of brain metastasis.
Although brain metastases are typically a late manifestation of disease, primary cancers can spread to the brain at various times in the course of the illness; some studies have shown that synchronous brain metastases (those found within 1 month of the primary cancer diagnosis) occur in almost one-third of patients ( ). More commonly, however, brain metastases are diagnosed after a primary cancer is known to have spread to other systemic organs first. The average time between primary diagnosis and detection of brain metastasis is less than 1 year in patients with lung cancer and 2–3 years in patients with breast cancer, melanoma, or renal cell carcinoma. Overall, the average time between primary cancer diagnosis and diagnosis of metastatic brain disease is approximately 12 months.
This chapter focuses on brain metastases from primary lung cancer, breast cancer, and melanoma, which are by far the most common malignancies associated with brain metastases. Additional information about other solid organ tumors will be included as appropriate, for comparison.
The blood–brain barrier (BBB) is located at the level of the cerebral capillaries. It is instrumental in protecting the CNS by restricting the movement of solutes and cellular elements between the systemic circulation and neuronal tissue. The endothelial cells, astrocytes, and pericytes that form the neurovascular unit are critical to the function of the BBB. Endothelial cells are thin, flat cells that course along the cerebral capillaries. They are interconnected by a continuous line of tight junctions and thereby limit the movement of particles. Pericytes are contractile cells that synthesize biologically active substances and lie close to endothelial cells. They have been shown to contribute to the regulation of blood flow, endothelial cell proliferation, angiogenesis, and inflammatory processes. Researchers have found that, without pericytes, endothelial cells undergo hyperplasia, and abnormal vasculogenesis then occurs, allowing the BBB to become more permeable ( ). Astrocytes ensheath the capillary walls, almost fully covering endothelial cells and pericytes. Where this coverage is not complete, nerve endings have direct contact with the basement membrane. Astrocytes are known to maintain the homeostasis of the brain’s microenvironment and protect metastatic tumor cells from cytotoxicity induced by chemotherapy via mechanisms that upregulate survival genes in tumor cells.
As such, transport across the BBB is highly regulated. Molecules must penetrate a fourfold defense mechanism consisting of a paracellular barrier (maintained by the interendothelial tight junctions), a transcellular barrier (assured by the presence of endothelial cells, pericytes, and astrocytes), an enzymatic barrier that degrades numerous neurotransmitters, and a multitude of efflux transporters that expel chemicals from the CNS ( Table 1.1 ). Small gaseous molecules, such as oxygen and carbon dioxide, along with lipophilic agents, such as barbiturates, nicotine, and ethanol, can freely diffuse through the BBB, but specific influx transporters are required for nutrients such as glucose and amino acids to enter the CNS.
Tight junctures |
Intercellular pathways: water-soluble molecules |
Transcellular lipophilic pathways: lipid-soluble molecules |
Basement membrane |
Endothelium (low pinocytic activity) |
Receptor-mediated transcytosis: insulin, transferrin |
Absorptive transcytosis: albumin |
Astrocytes |
Pericytes |
Microglia (tumor-associated macrophages) |
High IL-10 and low IL-2 |
TNFα → phosphorylation of JNK and NFκB |
Wnt gene |
Fibroblasts |
Drug transporters |
Influx: LRP1 |
Efflux: MRP, PgP, ABCG2 |
Markers of integrity of BBB |
GLUT1, BCRP (correlation with triple-negative vs. HER2-positive tumors) |
Not only does the BBB protect the CNS from certain molecules, but it can also limit the transmigration of whole cells. This phenomenon is highly relevant with regard to metastatic cells, and researchers have found that, in certain cases, the BBB plays a supportive role in the spread of metastatic disease to the brain. Endothelial cells actively take part in the transmigration of some metastatic cells, allowing them to penetrate the defenses of the BBB. The BBB then serves to protect these metastatic growths by allowing them to evade immune surveillance.
Because the CNS lacks lymphatic drainage, brain metastases arise predominantly from hematogenous spread. As a result, migrating tumor cells become trapped at gray/white matter junctions and watershed zones, where the vessel diameter approaches the size of the metastatic cell. At that site, the role of the BBB has great significance; it either promotes or hinders the seeding of metastatic cells. Studies have shown that the extravasation of malignant cells into brain parenchyma can take several days, and intravascular proliferation preceding transendothelial migration seems to characterize cell lines with a high affinity for metastasis to the brain ( ).
After diapedesis, endothelial cells remain central to the proliferation and survival of metastatic cells both by aiding in the formation of the tumor vasculature through angiogenesis or vessel co-option and by maintaining the blood–tumor barrier, which significantly impairs drug delivery ( ). Of note, the distribution of brain metastases seems to be proportional to the blood flow to different regions of the brain; roughly 80% of brain metastases are found in the cerebral hemispheres, 10–15% in the cerebellum, and 1–5% in the brain stem. Interestingly, however, found that posterior fossa tumors arise disproportionately more commonly from pelvic or abdominal primary tumors, regardless of blood flow volume. One suggested hypothesis for this phenomenon is the presence of Batson’s spinal epidural venous plexus, which is a direct extension of the cerebral dural sinuses. Thus, with increased abdominal pressure and compression of the vena cava owing to tumor growth, a primary abdominal or pelvic neoplasm could seed the spine directly and metastasize through retrograde flow to the posterior fossa ( ). The authors pointed out, however, that if Batson’s plexus were the likely mechanism by which primary abdominal or pelvic tumors metastasized to the CNS, higher rates of spine and skull metastatic lesions would be expected with abdominal or pelvic primary tumors than with tumors at other sites. However, this was not found to be the case ( ).
In a follow-up retrospective study, challenged the general conclusion that primary abdominal and pelvic tumors more commonly metastasize to the posterior fossa. Using contrast-enhanced magnetic resonance imaging (MRI) and computed tomography (CT), compared the location of brain metastases in 100 subjects with primary abdominal and pelvic tumors with the location of brain metastases in 100 subjects with primary tumors at other sites and found no difference in the distribution of brain metastases between the two patient populations. The difference in findings may be related to the fact that used only CT to diagnose CNS lesions, whereas used both MRI and CT. Furthermore, argued that the conclusions made by were based on a small number of patients (15) with a single brain metastasis from primary pelvic and abdominal tumors, and did not report the distribution of multiple brain metastases from these pelvic and abdominal tumors. reported a ratio of 4:1 supratentorial to infratentorial single metastases from primary abdominal and pelvic tumors, which is in accordance with the distribution of cerebral blood flow, and they concluded that the predominance of posterior fossa metastases from primary abdominal and pelvic tumors could not be verified.
The metastatic process consists of three main principles. First, the malignant cells must detach from the primary tumor and spread to distant sites. Second, these cells must be able to invade the target structure and undergo initial growth. Finally, biochemical processes must develop to allow these metastatic cells to grow in the new or modified microenvironment. Several molecular mediators for invasion have been identified. Among them are matrix metalloproteinases, serine proteases, and heparanases. Furthermore, metastatic cells are known to express integrins, cadherins, selectins, and proteoglycans such as CD44, among others, that enable these cells to adhere to tissue surfaces and undergo initial growth. Once initial seeding has taken place, angiogenesis must occur to establish an energy source for the metastatic growths. Vascular endothelial growth factor and its receptors, as well as hypoxia-inducible factors and other matrix metalloproteinases, have been found to play key roles in these processes ( ).
Brain metastasis should be considered when any cancer patient develops new neurologic symptoms, but occasionally brain metastases are asymptomatic. Historical data showed that brain metastasis occurs in as many as one-third of patients with breast cancer, but asymptomatic brain metastases are less common, occurring in roughly 10–15% of patients, although variability between studies may exist owing to differences in CNS imaging modalities ( ).
The most common symptoms at presentation include headache (50%), focal weakness (40%), confusion or altered mental status (30%), seizures (15%), and ataxia (10%), and these symptoms tend to worsen with time as the tumor grows and the surrounding edema exerts a mass effect on nearby structures. Development of such neurologic symptoms is most often a slow process, but hemorrhage into a metastatic lesion should be suspected when acute neurologic symptoms, like seizure, develop, especially in patients with melanoma. This finding was attributed to the high prevalence of multiple metastases in melanoma, as well as to the tendency of melanoma to be hemorrhagic ( ).
MRI is the diagnostic study of choice for the detection of intracranial metastases. When clinical suspicion is high, MRI is often the first line of imaging modality used. MRI works well because the breakdown of the BBB results in contrast enhancement of metastases. The two most common patterns observed are solid enhancement and rim enhancement with a central cystic nonenhancing region. These cystic areas can arise owing to necrosis, keratin deposits in squamous cell carcinoma, or mucin secretion in adenocarcinoma ( ). Therefore, for lesions to be characterized in terms of their anatomic location, size, and number, or for the amount of associated edema to be estimated, various MRI sequences must be used.
T1 precontrast images are useful for detecting subacute hemorrhage, which is evident as a hyperintense signal. Melanin, fat, and protein can also demonstrate bright signal on noncontrast T1-weighted images. In contrast, edema surrounding metastases is best evaluated on T2-weighted images, especially the fluid-attenuated inversion recovery sequence, in which the cerebrospinal fluid signal is suppressed, resulting in increased conspicuity of hyperintensity adjacent to ventricles and sulci.
T2-weighted sequences can detect hemorrhage or melanin, which manifests as a decreased signal and is occasionally the only abnormality that brain metastasis from melanoma demonstrates on MRI. Susceptibility-weighted imaging is a high-resolution gradient echo MRI sequence that has an increased ability to detect blood products and venous structures, and this technique is currently being explored for its ability to identify additional internal characteristics of brain tumors.
A novel MRI sequence, the motion-sensitized driven-equilibrium, is capable of selectively suppressing signals from flowing blood. Application of this technique can be useful in cases in which end-on vessel enhancement on images from routine postcontrast MRI is mistaken for punctate metastases. However, although this technique can improve detection of brain metastases, it may also increase the rate of false positives.
Distinguishing brain metastases from primary brain tumors is of great clinical importance, yet can be difficult radiographically. Findings that aid in identifying metastatic lesions include multiplicity, well-defined contrast-enhancing margins, and the location of gray/white matter. Although primary brain neoplasms are infiltrative more often than are metastatic lesions, no pathognomonic imaging findings can distinguish a metastatic brain lesion from a primary tumor, particularly in the setting of a solitary lesion. Thus, tissue biopsy confirmation is often needed before appropriate therapy can be initiated. Advanced MRI sequences such as diffusion, perfusion, and spectroscopy can also provide complementary information and aid in differentiating metastatic lesions from primary brain tumors or other mimickers such as abscesses and ischemia. Furthermore, these techniques can be of value in differentiating changes that occur after irradiation, which demonstrate heterogeneous enhancement and might mimic recurrent disease on routine diagnostic MRI.
CT is often used as a screening examination in patients with acute symptoms, as well as to elucidate life-threatening sequelae of brain metastasis such as herniation, hemorrhage, and hydrocephalus. Metastases are usually isodense or hypodense compared with brain tissue on noncontrast studies, and metastases demonstrate enhancement following administration of contrast in CT studies, although the tissue contrast resolution and sensitivity are lower than on MRI studies. Acute hemorrhage demonstrates increased density on noncontrast CT studies. Although bony erosion is better characterized on CT, meningeal involvement by metastatic disease is best evaluated by MRI. The combination of CT with positron emission tomography (PET) improves the spatial resolution of lesions. However, the accuracy of PET-CT in differentiating residual or recurrent brain tumors from necrosis caused by radiotherapy is dependent on the time interval between the radiotherapy and the PET study, the type of radiotherapy administered, and the type of tumor.
In a comparison of fluorine-18 fluorodeoxyglucose ( 18 F-FDG) PET with MRI for the detection of brain metastases, found that 18 F-FDG PET identified, at most, 61% of MRI-diagnosed brain metastases. This limited detection rate was attributed to the fact that gray matter has physiologically high 18 F-FDG uptake, impairing PET imaging from accurately diagnosing brain metastases, particularly when the lesions are small. The likelihood of detecting a 1-cm lesion with 18 F-FDG PET was observed to be only 40%, and the lesion had to be approximately 1.8 cm before the mean detection rate increased to 90%. Recently, however, dual-phase 18 F-FDG PET was reported to have improved precision in separating recurrent tumors from posttreatment necrosis in brain metastases.
Carbon-11-methionine PET, which is currently gaining popularity for its high detection rate of brain neoplasms and for its good lesion delineation, is being studied to assess its diagnostic accuracy for brain metastasis. Studies have shown that 11 C-methionine, as an amino acid tracer, is avidly taken up by many tumor types as the rapidly proliferating cells synthesize proteins. However, because the spatial resolution of PET cameras is not high enough to identify subcentimetric brain metastases, 11 C-methionine PET is used more for guiding therapy and follow-up than for detecting lesions ( ).
With the incidence of brain metastasis on the rise, multiple studies have been conducted to assess the utility of certain biologic markers, as well as nomograms and risk-stratification techniques, to improve treatment and understanding of patient outcomes with brain metastasis. Currently, the Radiation Therapy Oncology Group (RTOG) retrospective recursive partitioning analysis (RPA) is among the most widely used prognostic indices for patients with brain metastasis. The RPA, developed by , is based on a study population of roughly 1200 patients who received external beam radiotherapy in three consecutive RTOG brain metastasis trials between 1973 and 1993. The RPA classifies patients into one of three prognostic groups (“classes”) that are based on age, performance status, and extracranial tumor control; outcomes between these groups are significantly different ( Table 1.2 ).
Variable | Class 1 | Class 2 | Class 3 |
---|---|---|---|
Karnofsky performance status score | ≥70 | ≥70 | ≤70 |
Primary tumor status | Controlled | Uncontrolled | Uncontrolled |
Age | ≤65 years | >65 years | >65 years |
Extracranial metastases | None | Present | Present |
For RPA Class 1 patients, the median survival time was 7.1 months; for RPA Class 2 patients, the median survival time was 4.2 months; and for RPA Class 3 patients, the median survival time was 2.3 months. As evidenced by this classification model, RPA Class 2 tends to be a largely heterogeneous population, more so than RPA Class 1 or 3. For this reason, other researchers sought to modify the RPA classification technique or develop new prognostic indices to stratify patients more accurately according to expected survival time. Unfortunately, many of these prognostic indices fail to stratify patients according to primary tumor type, bringing their value into question.
To further assess prognostic factors, studied 1292 patients with CT-diagnosed brain metastases treated at Daniel den Hoed Cancer Center in Rotterdam from 1981 through 1990. In this patient cohort, the median survival time was 3.4 months, and, as many other studies have corroborated, characteristics associated with improved prognosis included high performance status, limited systemic tumor burden, normal serum lactate dehydrogenase levels, age younger than 70 years, and no more than two brain metastases. Of note, this study also documented additional improvement in prognosis in patients with brain metastasis from breast cancer compared with those with brain metastases from other primary cancer types.
While studying the results of reoperation in 48 patients with recurrent brain metastases, developed a prognostic grading system that was based on five characteristics. They found that presence of systemic disease, age, preoperative Karnofsky performance status (KPS), time to recurrence, and type of primary tumor all significantly affected survival. The grading method assigned scores of 1 or 0 for each of these five characteristics, for a maximum possible total of 5 points. Patients were then assigned a disease grade ranging from I to IV, depending on the number of points (Grade I: 1 point, Grade II: 2 points, Grade III: 3 points, Grade IV: 4 or more points). This grading system was correlated with overall survival; patients with Grade I disease had a 5-year overall survival rate of 57%, compared with a 1-year overall survival rate of <20% for patients with Grade III and IV disease. None of the patients with Grade III or IV disease survived 5 years. The authors did note, however, that this grading scheme is valid only in patients with previously resected brain metastases.
introduced the score index for radiosurgery (SIR) as an alternative to the RPA for assessing survival prognosis in patients with brain metastases previously treated with radiosurgery. Age, KPS, systemic disease status, volume of the largest brain lesion, and total number of lesions were found to be highly prognostic of survival outcomes. In this classification system, each prognostic variable is given a score of 0–2, for a combined total score ranging from 0 to 10. According to this classification system, patients with a total SIR less than 3 had a median survival time of 2.91 months, those with an SIR between 4 and 7 had a median survival time of 7 months, and those with an SIR between 8 and 10 had a median survival time of 31.38 months.
In response to elaborate and complex stratification systems, introduced the basic score for brain metastases (BSBM), which is based on three main prognostic factors that could be assessed more easily. For BSBM, KPS, presence of extracranial metastases, and control of the primary tumor are given scores of either 0 or 1. A patient with a KPS >80 who does not have extracranial metastases and whose primary tumor is controlled receives 3 points. A patient with a KPS <70 whose primary tumor is not under control and who has extracranial metastases receives a score of 0. For patients with a BSBM of 0, the median survival time was 1.9 months; for a BSBM of 1, it was 3.3 months; and for a BSBM of 2, it was 13.1 months. Median survival time for patients with a BSBM of 3 was undefined; 55% of patients were alive after 32 months.
After noting several limitations with the RPA, SIR, and BSBM, which are the three most commonly used prognostic indices for patients with brain metastases, introduced the graded prognostic assessment (GPA). The GPA was found to be as accurate as the RPA, but more accurate than the SIR or BSBM. Furthermore, it was found to be the least subjective, most quantitative, and based on the most current data from randomized trials. The GPA was modified in 2010 and updated in 2012 to construct diagnosis-specific GPA classes to predict survival in patients with brain metastasis from various primary tumors, including lung cancer, renal cancer, melanoma, gastrointestinal cancer, and breast cancer ( ). As research on brain metastasis continues to expand, studies are documenting the significant influence of primary tumor type and subtype on prognosis, and interest in disease-specific prognostic models has taken the forefront in the development of new stratification strategies.
In spite of the development of new disease-specific prognostic models, more modern imaging techniques, and more effective cancer treatments, brain metastases remain an indicator of poor prognosis for patients; studies have failed to show any significant increase in survival times over recent years. Before the advent of modern imaging modalities to diagnose brain metastasis early, patients with symptomatic brain metastasis that could not be treated survived more than 2 months. However, with corticosteroids, median survival times improved to 2–2.5 months, and with whole-brain radiotherapy (WBRT), median survival times improved to 3–6 months. Unfortunately, most patients with brain metastasis will or have already developed widespread, disseminated systemic disease, for which overall survival is largely determined by progression of the extracranial metastases ( ). In this regard, the goal of treatment for brain metastasis is palliative and to treat symptoms, whereas more aggressive therapies are reserved for patients with longer expected overall survival times.
When taking primary tumor sites into consideration, different studies report different median survival times and prognostic factors. The median survival time for patients diagnosed with brain metastasis from breast cancer was reported to be 7.5 months ( ). However, survival time after identification of brain cancer metastasis in patients with breast cancer varies significantly with molecular prognostic factors, including cancer subtype (i.e., luminal [hormone-receptor positive], HER2-positive, and basal or triple-negative [hormone-receptor and HER2 negative]; Dawood 2013). Time to brain metastasis was found to differ according to subtype; reported median brain metastasis-free survival times of 14 months for patients with triple-negative breast cancer, 18 months for patients with HER2-positive breast cancer, and 34 months for patients with luminal breast cancer. Moreover, the cause of death has also been shown to be highly influenced by tumor subtype. Nearly half of all patients with brain metastasis from HER2-positive breast cancer die from CNS progression, but these patients have longer median survival times than patients with other subtypes. In contrast, those with triple-negative breast cancer die more quickly and more commonly from extracranial metastatic disease progression, most likely owing to limited systemic treatment options ( ).
Studies are now concluding that although treatment with trastuzumab, an HER2-specific monoclonal antibody, significantly improves survival in patients with HER2-positive breast cancer, trastuzumab also increases the incidence of brain metastasis. This finding is attributed to better extracranial disease control and the fact that the large molecular size of trastuzumab inhibits its penetration through the BBB, making the CNS a sanctuary site for metastatic cells ( ). showed that 1.6% of women with HER2-positive, early-stage breast cancer who did not receive trastuzumab developed brain metastasis, whereas 10.5% of patients with HER2-postive breast cancer who received the monoclonal antibody developed brain metastasis. Nonetheless, continued trastuzumab-based therapy may improve survival.
Nomograms, which are unique to breast cancer, allow physicians to estimate the likelihood of the occurrence of brain metastasis on the basis of selected clinical and pathologic variables, thereby providing personalized risk estimates to guide treatment options. showed in a multivariate analysis that brain metastasis was independently associated with age, histologic grade, interval between diagnosis and first metastasis, number of nonbrain metastatic sites, and hormone receptor and HER2 status. Using such clinical data, the researchers were able to construct a nomogram capable of identifying a subgroup of patients with proven metastatic breast cancer who were most likely to develop brain metastasis, thus allowing these high-risk patients to be selected for more intensive prophylactic therapy.
According to , this nomogram could also be used to plan clinical trials to test the efficacy of prophylactic radiation (or drugs that may be shown to cross the BBB) in preventing brain metastases. Their research models showed that, if a trial were to include all patients without any selection criteria, all brain metastases could potentially be prevented with prophylactic radiation. This approach, however, would be undertaken at the expense of irradiating many patients who may otherwise never develop brain metastasis. In contrast, if only patients with a >25% chance of developing brain metastasis were included, half of patients would receive prophylactic radiation and roughly 89% of all potential brain metastases would be treated.
In keeping with published data on risk stratification for patients with brain metastases, several studies of patients with lung cancer further validate the RPA and diagnosis-specific GPA. One such study conducted at Duke University Medical Center evaluated the development of brain metastasis in 975 patients with resected early-stage non-small cell lung cancer (NSCLC). Using multivariate analysis, the researchers identified four factors that were independently associated with an increased risk of developing brain metastasis. These included young age, large tumor size, and the presence of lymphovascular invasion and hilar lymph node involvement ( ). Importantly, investigations conducted on patients with lung cancer have not only validated known prognostic measures, but have also identified tumor histologic subtypes that are associated with increased rates of brain metastasis. Studies have found that, as in patients with HER2-positive breast cancer, patients with NSCLC adenocarcinoma have significantly longer median overall survival times but are also more likely to develop brain metastasis compared with patients with other NSCLC histologic subtypes ( ).
Small cell lung cancer (SCLC), which accounts for only 15% of all lung cancers, is associated with very poor prognosis because of its highly aggressive nature, and many patients diagnosed with SCLC are found to have distant metastasis at presentation. However, SCLC can be highly responsive to chemotherapy, especially when detected early. Recent evidence suggests a correlation between circulating tumor cell (CTC) levels and progression-free survival and overall survival times in patients with metastatic breast cancer, colorectal cancer, castration-resistant prostate cancer, and NSCLC. However, the prognostic impact of CTCs and their relationship with brain and nonbrain metastases in patients with SCLC remains poorly understood. evaluated the link between CTC levels and disease burden and prognosis, as well as the optimal CTC cutoff level for predicting outcomes in patients with SCLC. They found that a CTC cutoff level of 8 CTCs per 7.5 mL of blood discriminated between groups with favorable and unfavorable prognoses. This cutoff level is higher than that reported in previous studies for other tumor types.
Become a Clinical Tree membership for Full access and enjoy Unlimited articles
If you are a member. Log in here