Introduction

Brain metastases (BM) from solid tumors represent a treatment challenge. Although cytotoxic chemotherapy and or molecular-targeted agents are the primary therapeutic approaches for disseminated cancer, they play a limited role in the treatment of brain metastases; whole brain radiotherapy (WBRT), stereotactic radiosurgery (SRS), and surgical resection are the primary modalities. Routine use of cytotoxic chemotherapy following WBRT for BM from a variety of solid tumor primaries failed to improve survival in multiple studies [1]. However, this lack of data does not prove that chemotherapy is ineffective in all patients. At present, the majority of the existing data is limited to patients with typically chemotherapy-resistant non–small cell lung cancer (NSCLC). Clinical benefit in other, more chemosensitive histologies such as breast and small cell lung cancer (SCLC) may be possible. Further support comes from studies that added chemotherapy to WBRT with an improvement in the radiographic response rates, though response was not the primary study end point [1]. BM patients are a heterogeneous group, further complicating the interpretation of many clinical trials of systemic chemotherapy for BM. The number and size of metastases, patient age and performance status, clinical symptoms, extent of systemic disease, relative radiosensitivity and chemosensitivity of the underlying tumor histology, and anatomic localization of the metastasis are all important factors that affect treatment of an individual patient. As such, treatment strategy may need to be individualized. Primary systemic therapy may be considered in those with small, multiple (> 3), asymptomatic BM, especially in those with a chemosensitive primary tumor who require a change in therapy for systemic disease. Systemic BM therapy should also be considered as combination therapy with WBRT, or as a potential palliative measure for those who are not candidates for surgery or radiotherapy.

This review discusses the challenges of treating BM with systemic therapy because of the unique physiology of the brain vasculature, interactions with supportive care medications, and issues with treatment resistance. In addition, recent insights from the medical literature are highlighted for specific tumor types that have the greatest incidence of BM.

Blood–Brain Barrier

When determining the optimal treatment for a patient with BM, the specialized physiology of the brain vasculature must be considered. The blood–brain barrier (BBB) is a physiologic barrier between the vasculature and brain parenchyma formed by specialized endothelial cells lining the cerebral microvasculature, pericytes, and astrocytic perivascular endfeet [2, 3]. Charge, lipophilicity, binding affinity to plasma proteins, and molecule size are determinants of a substance’s BBB permeability. Large hydrophilic molecules, which includes many cytotoxic and molecular-targeted agents, are excluded from the central nervous system (CNS) unless they can be actively transported by receptor-mediated transcytosis [3]. In addition, the endothelial cells that form the BBB highly express P-glycoprotein, a drug efflux protein associated with resistance to several chemotherapeutic agents [4]. Even if an agent transverses the BBB, reactive astrocytes surrounding the BM may actively protect the tumor [5]. As such, the CNS may serve as a sanctuary site for metastases of patients with successfully treated systemic disease. Thus, as more effective systemic therapies are developed, the incidence of BM may increase and more frequently be the primary site of treatment failure.

Although the BBB may hinder treatment of micrometastases, it may not be the primary cause of treatment failure in patients with established metastases. Experimental BM models reveal that the BBB becomes structurally and functionally compromised when a metastatic tumor grows larger than 1–2 mm [3]. BM require angiogenesis for growth and the newly formed capillaries in BM are fenestrated, lack tight junctions, have an increased number of pinocytotic vesicles and an irregular basilar membrane and thus do not possess a normal BBB. Preclinical studies have shown that hydrophilic drugs such as paclitaxel may achieve higher concentration in BM than in the surrounding parenchyma [6]. Furthermore, following a single 2 Gray dose of WBRT there is substantial disruption of the BBB in experimental animals 2 hours after a radiation dose. The BBB is reconstituted after 24 hours, but 30 Gray in 10 doses leads to progressive increase in BBB permeability for up to 4 weeks after treatment [2]. In the clinical setting, computerized tomography scanning and magnetic resonance imaging have shown that the BBB is disrupted in and near most brain metastases as evidenced by enhancement of tumor following intravenous contrast administration. Thus, the most important and perhaps only factor in the use of systemic therapy for BM may be the efficacy of the therapeutic agent against the primary tumor histology.

Interactions with Supportive Care Medications

The supportive care medications used in BM patients may compromise the effectiveness of systemic anti-tumor therapy. Upon diagnosis of BM, treating physicians often reflexively start an anticonvulsant for “seizure prophylaxis” and corticosteroids “to treat edema.” Many of the anticonvulsants, including phenytoin, carbamazepine, phenobarbital, and oxcarbazepine, are strong inducers of the hepatic P450 enzyme system which may lead to increased metabolism of chemotherapeutic agents that are metabolized by this system. An inclusive list of interactions are highlighted in the review by Kamar and Posner [7]. In addition, antiepileptic drugs that are histone deacetylase inhibitors (carbamazepine, topiramate, valproic acid, and metabolites of levetiracetam) may interact with chemotherapeutic deacetylase inhibitors such as vorinostat to cause thrombocytopenia [7, 8]. Although controversial, practice guidelines from the American Academy of Neurology recommend against the prophylactic use of anticonvulsants in patients with BM unless they have experienced a seizure [9]. If a BM patient has a history of seizures, the use of a non-enzyme inducing drug such as levetiracetam, lamotrigine, lacosamide, valproic acid, gabapentin, topiramate, tiagabine, or zonisamide should be considered.

The mechanism of action of corticosteroids on edema is incompletely understood, but thought to involve restoration of the disrupted BBB. The use of corticosteroids in the management of BM should be avoided unless brain edema is symptomatic. Otherwise, reconstitution of the BBB may impair entry of water-soluble systemic agents that may have efficacy against BM.

Sensitivity of BM to Systemic Therapy

As discussed above, perhaps the most important factor in the treatment of BM is the sensitivity of the underlying tumor histology to the systemic therapy. Two criticisms of large clinical trials that have been conducted in BM patients are:

  1. 1.

    Systemic treatments explored in BM clinical trials are often selected based upon ability to transverse the BBB rather than efficacy against the underlying tumor histology.

  2. 2.

    Most series include patients that have been heavily pretreated with multiple systemic therapies and thus have treatment-resistant disease.

Treatment of breast cancer BM with temozolomide and capecitabine illustrates the first point. Temozolomide, an oral alkylating agent with excellent CNS penetration, has shown disappointing efficacy for breast cancer BM. The results are not surprising since temozolomide has limited activity in systemic breast cancer. In contrast, capecitabine, an agent with systemic activity, has shown radiographic responses of BM even though the drug has limited penetration through an intact BBB.

The second point is less straightforward as the sensitivity of BM to systemic therapy may be similar or different to the primary tumor and other metastatic sites. The sensitivity of two separate brain metastases may be different. Generally, agents that are effective against the primary tumor should be used first for brain metastases. Theoretically, patients with isolated CNS recurrence can be re-challenged with therapies that were previously effective for the systemic disease, but are now resistant. In this setting micro-metastatic BM may not have been exposed to the prior treatment because of sequestration behind an intact BBB.

Non–Small Cell Lung Cancer

NSCLC is the most common systemic tumor to metastasize to the brain [10–12]. BM are present at the time of diagnosis in 10% and later in the disease course in 40%. A variety of systemic agents have been explored in clinical trials for primary treatment of NSCLC BM. Most of these studies have demonstrated modest, short lived responses, but nothing to advocate their routine use.

Motexafin gadolinium (MGd) and temozolomide have been explored as radiation-sensitizing agents in NSCLC [13]. An effective radiation-sensitizing agent enhances tumor cell death, while sparing the effects of radiotherapy on normal cells. MGd is a redox modulator that is selectively taken up by tumor cells [14]. MGd catalyzes the oxidation of intracellular reducing metabolites in a process known as futile redox cycling. The depletion of these reducing metabolites removes the substrate necessary in a cell to repair oxidative damage induced by radiation. A randomized, phase 3 study of 554 patients was conducted to determine the efficacy of MGd in combination with WBRT for the treatment of BM from NSCLC [15]. The primary end point was interval to neurologic progression. MGd improved the interval to neurologic progression compared with WBRT alone (15 vs 10 months; P = 0.12). In North America, where treatment was more prompt after diagnosis of BM as compared to Europe, there was a statistically significant prolongation of the interval to neurologic progression for MGd and WBRT vs WBRT alone (24.2 vs 8.8 months; P = 0.004).

Temozolomide increases survival in glioblastoma patients when used in conjunction with radiotherapy [16]. Preclinical studies suggest that temozolomide is a radiation sensitizer [17, 18]. The combination of temozolomide with WBRT for NSCLC BM has been promising in early non-randomized studies [19–21]. Validation is being explored in several ongoing, large, randomized clinical trials [1].

Recent studies have explored the use of molecular-targeted agents for the treatment of NSCLC BM. Studies using bevacizumab have excluded BM patients because of concern for intracranial hemorrhage. An open-label, multicenter trial for first-line and second-line treatment of NSCLC enrolled patients with history of treated BM [22]. First-line patients received bevacizumab every 3 weeks with platinum-based doublet therapy or erlotinib, and second-line patients received bevacizumab with single-agent chemotherapy or erlotinib. Of the 115 patients enrolled, with a median of five bevacizumab cycles there were no episodes of grade 2 or worse CNS hemorrhage. Bevacizumab’s safety and efficacy in patients with active BM is unknown; no episodes of grade 2 or worse CNS hemorrhage were reported in a small series of six patients, suggesting that the issue warrants further investigation [23].

NSCLC with activating mutations of the epidermal growth factor receptor (EGFR) may respond to erlotinib, a small-molecule EGFR inhibitor. Multiple case reports have described complete and partial BM responses to EGFR inhibitors, although the ability of the agents to accumulate in the brain is unclear. Weber and colleagues [24] used dynamic positron emission tomography with [11C]-erlotinib to demonstrate that erlotinib accumulated in the BM of a treatment responder. In another study of 69 NSCLC BM patients, 17 were identified as having EGFR mutations [25]. The objective response rate to erlotinib was 82.4%; no responses were observed in unselected patients. Even patients that develop BM during treatment with conventional erlotinib dosing could be considered for “pulsatile” high-dose erlotinib [26]. A retrospective review identified nine patients with EGFR-mutant lung cancer that were treated with erlotinib 1500 mg weekly for CNS metastases that occurred despite conventional erlotinib or other EGFR tyrosine kinase inhibitors. The radiographic response rate was 67% and 11% had stable disease; a prospective study is planned by the authors.

If BM are present at diagnosis of NSCLC, it may be acceptable to delay WBRT until after initiating chemotherapy. Barlesi and colleagues [27] conducted a multicenter phase 2 trial of pemetrexed and cisplatin as first-line chemotherapy for metastatic NSCLC with asymptomatic, inoperable brain metastases. Patients were treated with up to six cycles of chemotherapy followed by WBRT at disease progression or chemotherapy completion. A total of 43 patients were enrolled and 41.9% of patients achieved an objective response. The authors did not specify whether best response occurred during chemotherapy or after radiotherapy. Median survival time and time to progression were 7.4 and 4 months, respectively.

Breast Cancer

Patients with metastatic breast cancer have the second highest incidence of BM, with frequency between 10% and 30% of patients [28]. Standard chemotherapy regimens including cisplatin or carboplatin, etoposide, anthracyclines, cyclophosphamide, high-dose methotrexate, and 5-fluorouracil have achieved CNS response rates of 50–59% [28–31]. The fact that most of these agents poorly cross an intact BBB supports the notion that the chemosensitivity of the tumor to the treatment agent may be the most important factor in determining the response to therapy. It has been argued that the utility of these regimens against BM in the current era may be limited because most women will have already been treated with chemotherapy in the adjuvant setting [28]. However, because microscopic BM may have been sequestered behind the BBB at the time the water-soluble chemotherapy was administered, they may not have been exposed to the treatment. Thus, it is possible that repeating prior chemotherapy would be effective once the BM become clinically apparent.

Capecitabine and hormonal agents have also shown efficacy against BM. Several authors have reported CNS response in breast cancer patients with BM following capecitabine therapy [29, 32–34]. BM in estrogen receptor–positive disease may respond to tamoxifen and megestrol acetate [28, 35–38].

BM from HER2-positive breast cancer represents a special situation. HER2-positive breast cancer is thought to carry a biological predisposition to metastasize to the CNS [39, 40]. In addition, there is concern that successful systemic treatment of HER2-positive breast cancer with trastuzumab, a monoclonal antibody that does not cross an intact BBB, leads to increased incidence of BM as well as increased frequency of the CNS being the first site of relapse [40]. BM usually maintain the HER2 status of the primary tumor [41]. Lapatinib, a small-molecule dual inhibitor of HER2 and EGFR, has shown efficacy against CNS metastases in phase 2 studies [42]. Lin and colleagues [42] prospectively treated 242 HER2-positive breast cancer patients, with progressive BM following prior trastuzumab and WBRT, with lapatinib. The primary end point was objective response, defined as ≥ 50% volumetric reduction of CNS lesions in the absence of increasing steroid use, progressive neurologic signs and symptoms, or progressive extra-CNS disease. The study was later amended to allow patients who progressed on lapatinib the option of receiving lapatinib plus capecitabine. Objective CNS responses were observed in 6% of the 242 patients enrolled. Twenty-one percent of patients experienced a ≥ 20% volumetric reduction in CNS lesions. Additional responses were observed with the combination of lapatinib and capecitabine.

More recently, Lin and colleagues [43] completed a phase 2 randomized study of lapatinib plus capecitabine or lapatinib plus topotecan for patients with HER2-positive breast cancer and BM. Patients were required to have HER2-positive breast cancer with prior trastuzumab exposure and unequivocal radiographic evidence of new and/or progressive BM despite prior treatment with WBRT and/or SRS. The primary end point was objective response, defined as ≥ 50% volumetric reduction of CNS lesions in the absence of new or progressive CNS or non-CNS lesions, or increasing steroid requirement. The objective response rate in the lapatinib plus capecitabine arm was 38% (95% CI 13.9–68.4). The study was closed early because of excess toxicity and lack of efficacy of the lapatinib plus topotecan arm.

Melanoma

Although melanoma is relatively uncommon, it is the third most frequent cause of BM due to its high propensity to metastasize to the brain. Melanoma is relatively radioresistant and chemoresistant; treatment of BM is mostly ineffective. Temozolomide and fotemustine, the most commonly used chemotherapeutic agents against BM, have limited efficacy.

Ipilimumab, a T-cell potentiator that works by blocking cytotoxic T-lymphocyte antigen-4 (CTLA-4), is a critical negative regulator of the antitumor T-cell response [44]. In a randomized phase 3 trial, treatment with ipilimumab significantly increased overall survival both as a single agent and in combination with a g100 vaccine as compared to a vaccine control, leading to its approval by the Food and Drug Administration in March 2011. Patients with known BM were excluded from the trial. A recent phase 2 study investigated the use of ipilimumab as a single agent for patients with brain metastases. The study separated patients into two groups based on whether corticosteroids were required. Of 51 patients in the steroid-free group, there were 5 with partial response and 6 with stable disease at week 12. The response duration was 3–12+ months. The authors’ conclusion was that ipilimumab had similar activity for brain and non-CNS lesions [45]. Further work needs to be done before ipilimumab can be recommended as standard treatment for patients with brain metastases.

Renal Cell Carcinoma

Renal cell carcinoma (RCC) BM are chemotherapy resistant. Sunitinib is approved as first-line therapy for patients with metastatic RCC. Patients with BM were excluded from the randomized trial that led to its approval. Results from a sunitinib open-label, expanded access program were published in 2010 [46]. Three hundred and twenty-one of the 3464 had BM. Of the 213 BM patients evaluable for response, 92% had clear cell histology, 88% had undergone prior nephrectomy, and 12% had received prior anti-angiogenic therapy. Data regarding prior CNS irradiation were not reported. Toxicity of patients with BM was comparable to the overall population. Of the 213 patients with BM, one achieved a complete response and 25 had partial responses for an overall response rate of 12%. Median progression-free survival and overall survival in patients with brain metastases was 5.6 months (95% CI: 5.2–6.1) and 9.2 months (95% CI: 7.8–10.9), respectively. Median progression-free survival and overall survival for the total population was 10.9 months (95% CI: 10.3–11.2) and 18.4 months (95% CI: 17.4–19.2), respectively. The study demonstrated that the safety profile of sunitinib in BM patients was comparable to that of the overall population. The efficacy results suggest clinical activity of sunitinib in BM patients, a subgroup with limited treatment options. Compared to an extended access program of sorafenib, the overall response rate with sunitinib was higher (4% vs 12%).

In the sorafenib extended access program, of the 1891 evaluable patients with advanced RCC, 70 had brain metastases [47]. The efficacy assessment performed on 50 patients revealed no complete responses, 2 (4%) partial responses, and 34 (68%) stable disease for at least 8 weeks. Thus the overall response rate was 4%. Survival data were not reported.

Treatment of RCC BM with sunitinib or sorafenib following WBRT or SRS has been associated with a radiologic appearance and clinical symptoms that mimic tumor progression [48, 49]. With corticosteroid treatment and/or time, the imaging and clinical changes improve. Recognition of this pseudo-progression phenomenon is important to avoid misinterpretation of disease progression that could lead to an erroneous change in treatment.

Conclusions

Although there is no class I evidence to suggest routine use of systemic treatment for BM, each patient should be considered individually based on factors such as number and size of metastases, age, performance status, clinical symptoms, extent of disease, site of metastases, and chemosensitivity of the primary tumor. Those with good performance status and multiple, asymptomatic, small metastases could be considered for systemic therapy prior to WBRT, especially if the primary tumor is chemosensitive and a change in treatment is warranted for non-CNS disease. Otherwise, the use of systemic therapy is primarily for patients with BM that have progressed despite surgery, WBRT, and/or SRS.

Future considerations include identification of agents that enhance WBRT and consideration of alternative end points for BM clinical trials. End points that assess the maintenance of neurologic and neurocognitive function may be more important than the traditional end point of overall survival, especially since survival is also influenced by progression of non-CNS disease. As additional small-molecule inhibitors that readily pass the BBB are developed and more effective regimens against the primary tumor are identified, the use of systemic therapy for BM will continue to evolve and play a more prominent role.