Introduction

Metastases are the most common brain tumors and may occur up to 10 times more frequently than primary brain tumors in adults. They reportedly affect 9–17 % of cancer patients in large population- and autopsy-based studies [1], although the true incidence and prevalence tends to be significantly underestimated. The incidence continues to increase with advances in diagnostic imaging and systemic disease control as well as with increased use of routine staging tests that access the brain. Brain metastasis can also be the initial manifestation of extracranial, clinically unsuspected cancers.

Although they are generally considered a single disease entity which is invariably fatal in patients with advanced cancer, brain metastases represent a remarkably heterogeneous group of neoplasms with variable histologies and clinical factors which impact prognosis. They include status of the primary tumor, presence of extracranial metastases, patient performance status, location, volume, and number of metastases, and therapeutic response. The incidence of brain metastases in patients with otherwise well-controlled systemic cancers has been increasing impacting the management of these patients.

Magnetic resonance imaging (MRI) with contrast enhancement is the currently most commonly used modality in evaluating for the presence of brain metastases and, if present, for the accurate assessment of number and size of the tumors, their exact locations, and impact on nearby structures. Often the diagnosis of intracranial metastasis is made radiologically without histological confirmation. However, microscopic tissue diagnosis should be performed for brain mass(es) of uncertain etiology, especially when there is ambiguity in radiological features, since all brain masses in cancer patients are not necessarily metastatic (e.g., abscess). In addition, excisional biopsy is advocated for a solitary lesion in an accessible area of the brain. On the other hand, for leptomeningeal carcinomatosis, detection of malignant cells on cytologic examination of the cerebrospinal fluid (CSF) is the diagnostic gold standard. This underscores the role of the pathologist in evaluating these lesions.

The mainstay of treatment includes a combination of steroids, radiation (whole-brain and/or stereotactic), and surgery. Conventional systemic chemotherapy, in general, has limited efficacy for brain metastases, with few notable exceptions such as small cell lung cancer and germ cell tumors. However, molecularly targeted therapies have recently shown some promise against metastatic brain tumors especially those with ‘druggable’ mutations [2–4].

In the new era of evolving molecular medicine, as a critical member of the multidisciplinary team in the diagnosis and management of metastatic brain tumors, we pathologists should provide not only an accurate pathologic diagnosis of metastatic brain tumor (including its probable site of origin) but also information on prognostic and targetable molecular biomarkers.

This review discusses the epidemiology, general pathologic features of brain metastasis and differential diagnosis from primary brain tumors, practical immunohistochemical approach to identification of the primary site, unique pathologic patterns of metastasis with their clinical and radiological features, and specifically, two major brain metastatic tumors (breast cancer and non-small cell lung cancer) with emphasis on the molecular alterations related to therapy and their clinicopathologic significance.

Epidemiology

In general, brain metastases are believed to result from hematogenous spread mostly through arteries. Primary lung carcinoma, breast carcinoma, and melanoma, in decreasing order of reported frequency, are the most common sources of brain metastasis [5–8]. Despite its small fraction of cancer diagnoses, melanoma has the highest propensity to metastasize to the brain of all primary tumors in adults. Over one-third of patients with advanced-stage melanoma have brain metastases [9, 10]. Other tumors, including renal carcinoma, gastrointestinal carcinomas, and cancer of unknown primary (CUP), reportedly show high incidence of brain metastasis [5–8]. Despite their relatively higher overall incidence, brain metastases from prostate, ovary and thyroid cancers are uncommon with a reported metastatic incidence of less than 1 % [11–13]. Brain metastases in sarcomas are rather rare with the reported incidence of 1–8 % [14]. In pediatric population, osteosarcoma, germ cell tumors, neuroblastoma, and Ewing’s sarcoma are the major primary solid tumors that can develop brain metastasis [15, 16].

Another well-known and much less common route of brain metastasis is retrograde perineural tumor spread (PNS) of head and neck malignancies. The tumor migration occurs along the endoneurium and/or perineurium in a contiguous fashion and does not involve perineural lymphatics. This may enable complete surgical resection of the perineural tumor along with the main primary tumor, leading to a possible cure [17]. The facial nerve and the maxillary and mandibular divisions of trigeminal nerve are most commonly affected, and head and neck squamous cell carcinoma and adenoid cystic carcinoma of major and minor salivary glands are the two most common tumor types to disseminate in this fashion, although any head and neck malignancies may cause PNS [17], which is best evaluated with MRI.

Pathological features

Gross findings

Most brain metastases occur in the cerebrum (around 80 %), followed by the cerebellum (15 %) and brainstem (5 %) [18, 19]. Metastases to the pineal gland, choroid plexus, and parasellar/sellar regions are rarer. The overall distribution of metastases corresponds roughly to the relative blood flow to each area of the brain given their hematogenous nature of dissemination through the arteries. Within the brain parenchyma, metastatic foci are known to be most commonly seen at the cerebral gray–white junction, primarily in the territory of middle cerebral artery (MCA), probably due to sudden luminal narrowing and coiling architecture of arterial vessels at this point, which may serve as a trap for tumor cells passing through the blood stream [20]. Brain metastases also frequently involve the terminal “watershed” areas of arterial circulations (i.e., vascular border zones), especially in the territory between MCA and posterior cerebral artery [21]. Brain involvement due to advanced PNS may be seen in the middle or posterior cranial fossa, cavernous sinus, orbit, or meninges, the latter resulting in leptomeningeal carcinomatosis [22].

The gross appearance is largely non-specific with the exception of its tendency to have sharp borders (Fig. 1a) in contrast to primary brain tumors, most of which typically have infiltrating borders. Softening of the surrounding brain parenchyma due to edema is often prominent and sometimes disproportional to the size of the lesions. Metastatic melanoma can display black discoloration due to melanin pigment, and metastatic mucin-producing adenocarcinoma may show a shiny mucoid appearance [23]. Hemorrhagic brain metastasis is not uncommon and requires the differential diagnosis from primary hemorrhagic brain tumors (most commonly, glioblastoma). Most commonly known hemorrhagic tumors are melanoma, renal cell, chorio, thyroid, breast and bronchogenic carcinomas (mnemonic MR CT BB). Massive hemorrhage may be seen with small foci of metastasis, especially in melanoma and choriocarcinoma. Rarely, metastases such as squamous cell carcinoma, are cystic and may resemble an abscess.

Fig. 1
figure 1

Metastatic brain tumor originating from lung: a postmortem brain shows a well-circumscribed mass lesion in the hypothalamic area (arrowheads). b Poorly differentiated adenocarcinoma with a relatively well-defined border, surrounded by gliotic brain parenchyma (H&E stain)

Full size image

Microscopic findings and immunohistochemistry

Microscopically, brain metastases are typically sharply demarcated from the surrounding brain parenchyma, which is usually gliotic and rarefied (Fig. 1b). This is a very helpful feature in the differentiation of primary diffuse gliomas. In the later stages, however, they may infiltrate into the surrounding tissue. In most cases, ancillary studies are not necessary as the metastatic nature of the lesion is histologically evident as the lesions retain their inherent morphology. There are exceptions, however, where the metastasis is less differentiated and the primary site is less obvious. In such cases, the patient’s clinical history and review of any prior surgical material is very helpful.

Immunohistochemistry (IHC) is particularly useful in the following settings:

  1. 1.

    Differential diagnosis between primary neoplasms and poorly differentiated or undifferentiated metastasis.

    1. (a)

      Epithelioid glioblastoma (E-GBM) vs. metastatic amelanotic melanoma vs. metastatic undifferentiated carcinoma.

    2. (b)

      Gliosarcoma vs. metastatic sarcoma vs. metastatic sarcomatoid carcinoma: interestingly, a significant proportion of metastatic melanoma in the brain is amelanotic/pauci-melanotic, mimicking epithelioid malignancies, including E-GBM. Glial fibrillary acidic protein (GFAP) labels E-GBM and glioma component of gliosarcoma, with no expression in sarcomatous or carcinomatous components, and is thus a very useful marker for the differential diagnosis. Glioblastoma frequently demonstrates obvious immunoreactivity for pancytokeratin (AE1/AE3) probably due to cross reactivity in intermediate filaments (i.e., cytokeratin and GFAP), which can be confusing. In this setting, low-molecular-weight cytokeratin cocktail (e.g., CAM5.2) is useful to confirm the epithelial nature of metastasis. In cases suggestive of metastatic melanoma, specific melanocytic markers such as HMB-45, Melan-A, and Microphthalmia-associated transcription factor (MITF) are needed for the diagnosis and differentiation from its mimickers. SOX10 is a relatively newly introduced pan-schwannian and melanocytic marker, with high sensitivity and specificity for malignant melanomas, including desmoplastic and spindle cell variants. It is important to note, however, that a significant proportion of primary central nervous system (CNS) neoplasms, primarily astrocytomas, are immunoreactive for SOX10 [24].

    3. (c)

      Anaplastic meningioma vs. dural-based metastatic poorly differentiated/undifferentiated carcinoma: both tumors are composed of mitotically active, malignant epithelioid cells without obvious glandular or squamous differentiation. Since there are no specific IHC markers for meningioma, the IHC differential is also difficult. This is further confounded by the fact that both tumors show epithelial membrane antigen (EMA) immunoreactivity and that cytokeratin is usually expressed in anaplastic meningiomas unlike grade 1 meningiomas. Vimentin IHC may be helpful since it is diffusely and strongly positive in meningiomas but is generally not expressed in carcinomas, with some notable exceptions (e.g., clear cell renal cell carcinoma, adrenocortical carcinoma).

    4. (d)

      Central primitive neuroectodermal tumor (PNET)/medulloblastoma vs. metastatic small cell carcinoma of the lung: these tumors are histologically indistinguishable from each other and are characterized by a diffuse cellular proliferation of undifferentiated malignant blue cells with a high nuclear cytoplasmic ratio. Cases with vague nodularity, if present, favor PNET/medulloblastoma. Although these typically have a different age distribution (children vs., older adults, respectively), however, there is some overlap and cases of medulloblastoma in older patients do occur. TTF-1 and pancytokeratin immunoreactivity is consistent with metastatic pulmonary small cell carcinoma and excludes PNET/medulloblastoma, if present.

    5. (e)

      Hemangioblastoma vs. metastatic clear cell renal cell carcinoma: hemangioblastoma (HB; WHO grade I) occurs both sporadically and in association with von Hippel-Lindau (VHL) disease, a familial autosomal dominant syndrome. HB is a cardinal feature of VHL disease, affecting 60–80 % of patients. Histologically, HB is highly cellular and highly vascular neoplasm composed of haphazardly oriented small capillaries and sinusoidal channels separated by larger, vacuolated, lipid-containing cells (‘stromal cells’). Stromal cells represent the neoplastic component of HB and commonly demonstrate degenerative nuclear atypia. The histological differential diagnosis from metastatic clear cell renal cell carcinoma (CCRCC) can occasionally be difficult, especially in cases with significant nuclear atypia in stromal cells. Immunohistochemically, inhibin-A and aquaporin-1 (a widely expressed water channel protein), respectively, show strong cytoplasmic positivity and plasma membrane positivity of stromal cells in the majority of HB cases. Carney et al. proposed the combined use of PAX2, PAX8 and inhibin-A to differentiate between these tumors as HB is typically PAX2(−), PAX8(−), inhibin-A(+), whereas metastatic CCRCC is usually PAX2(+), PAX8(+), inhibin-A(−) [25]. Of note, inhibin-A and EMA immunoreactivity is present in a significant percentage of RCC and HB cases, respectively.

  2. 2.

    Identification of the primary site for undifferentiated/poorly differentiated malignant tumors: brain metastasis can be the initial manifestation of extracranial malignancies. In this setting, identification of the primary site is important particularly in tumors which have the potential of benefiting from site-specific targeted therapy. In most instances, the primary site can be determined by histologic examination, IHC and detailed clinical and radiological information. IHC is most helpful and a basic panel of antibodies is necessary to determine the broad category of the neoplasm. The first line of antibodies that are commonly used includes pancytokeratin for carcinoma, melanoma cocktail for malignant melanoma, GFAP for primary glial tumors, and leukocyte common antigen for malignant lymphoma. For metastatic carcinoma, in addition, combined use of cytokeratin (CK)-7 and CK-20 is helpful in narrowing down the site of origin. Judicious use of relatively organ-specific markers, listed in Table 1, is indicated to confirm the site of origin [26].

    Table 1 Immunohistochemical profile of metastatic brain tumors (ref. [26])
    Full size table
  3. 3.

    Confirmation of presence or absence of metastatic malignant cells in an obscuring background: occasionally, relatively small hemorrhagic metastasis causes massive bleeding that is heavy enough to obscure the tumor cells (most notably, choriocarcinoma and malignant melanoma). Given the relative paucity of tumor cells in hemorrhagic lesions, the diagnosis can be difficult. The most common etiologies of intraparenchymal hemorrhagic lesion, particularly in the elderly, are amyloid angiopathy and hypertensive hemorrhage. If there is a high index of suspicion, IHC can be helpful in highlighting isolated tumor cells.

  4. 4.

    IHC biomarkers for metastatic breast cancers: estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor (Her)-2 IHC status in metastatic breast cancer provides useful clinical and therapeutic information (see details below).

Unique pathological patterns of brain metastasis

Dural metastasis

Gadolinium-enhanced brain MRI is the modality of choice in the detection of dural metastasis which is much less common than intraparenchymal metastasis and reportedly occurs in up to 9–10 % of patients with primary extracranial malignancy according to a review of autopsy cases from 1950 to 1976 [27, 28]. In one half of those, it is the only site of intracranial tumor involvement [27]. The exact incidence of pure dural metastasis is unknown since most cases occur in combination with leptomeningeal involvement. Dural metastases arise either from a direct extension of skull metastases or from hematogenous spread to the dura, the former being more common [29, 30]. The two most common primary sites are breast and prostate, but they are not exclusive [29, 30]. Possible explanation of the high frequency of skull and dural metastases from prostatic cancer is retrograde seeding by the valveless vertebral venous system (Batson plexus) [31]. There are two clinically significant forms of dural metastasis: diffuse and localized forms. A significant proportion of dural metastases are asymptomatic particularly in tentorial metastases, whereas headache and cranial neuropathy are the two most frequent complaints in symptomatic patients [29]. Cranial neuropathy is more frequently seen in the diffuse form, with dysphagia being common in the terminal phase. The main differential diagnosis in the localized form is meningioma, particularly in women with a history of breast cancer presenting with a single mass in the cranial convexity or falx (Fig. 2), since both can present as a dural-based enhancing tumor commonly with ‘dural tail’ sign. The intense and homogeneous contrast enhancement of dural metastasis can be explained by the absence of blood–brain barrier (BBB). Since there are no definite MRI or angiographic distinguishing features [30], microscopic examination of the tumor is essential.

Fig. 2
figure 2

Dural (falx) metastasis from lung cancer: MRI (T1-weighted image) with contrast enhancement shows an enhancing nodular mass attached to the falx, with prominent peritumoral edema. This patient has a history of lung cancer treated. (Photo: courtesy of Dr. Tejin Ri, Kushiro Kojinkai Hospital, Japan)

Full size image

A well-known complication of dural metastases is a subdural hematoma/hemorrhage, which does not include spontaneous hematomas associated with coagulation disorders that are commonly seen in cancer patients. The possible mechanisms include (1) a rupture of fragile neo-vessels within the tumor and (2) dilatation and eventual breakdown of capillaries of the inner dural layer due to tumor emboli in the dural vessels and/or to mechanical obstruction of external dural vessels secondary to expanding skull metastases [32–34].

Leptomeningeal carcinomatosis

Leptomeningeal carcinomatosis (LC), or leptomeningeal metastasis, is a devastating complication that represents wide spread of tumor cells to the subarachnoid space, and occurs in 4–7 % of patients with solid tumors [35, 36]. Autopsy studies show much higher rates of involvement [37]. The clinical manifestation is highly diverse and includes symptoms related to increased CSF pressure, cranial and/or spinal nerve deficits, psychiatric symptoms, and meningeal irritation symptoms. On MRI (contrast-enhanced T1-weighted), focal or diffuse contrast enhancement on meninges, ependyma, tentorium, basal cistern and sulci in the brain and spinal cord is considered to be positive for a diagnosis of LC (Fig. 3a); however, cytologic identification of malignant cells in the CSF is required for the definitive diagnosis, which has very high specificity. This is a diagnostically challenging disease, especially in the early stages, given that CSF cytology has low sensitivity and the specificity of MRI is not optimal for LC. Adenocarcinoma is the most frequent histology. Breast cancer (triple-negative biological subtype and lobular histological type), lung cancer (adenocarcinoma and small cell carcinoma), and melanoma are the most common primary solid malignancies causing LC [38–42]. Overall, breast carcinoma is the most common solid tumor associated with this disease, with the incidence of 5 % [43, 44]. Although rare, LC is known to be a complication of poorly differentiated gastric adenocarcinoma, especially in south-eastern Asia. Early detection and start of therapy may prevent the onset of irreversible symptoms (e.g., cranial and spinal nerve deficits) and prolong survival with an acceptable quality of life [45].

Fig. 3
figure 3

Leptomeningeal carcinomatosis from pulmonary adenocarcinoma: a MRI (T1-weighted image) with contrast enhancement demonstrates diffuse subtle enhancement of the leptomeninges. b Postmortem brain shows focal slightly opaque leptomeninges with congestion. c Microscopically, the subarachnoid space between cerebral sulci is filled with abundant discohesive cancer cells. (H&E stain). d Immunohistochemically, Napsin-A positivity is seen in the tumor cells

Full size image

For cytological diagnosis of CSF, presence of any large cells and/or mitotically active cells in CSF should be considered to be metastatic carcinoma, regardless of degree of cytologic atypia, until proven otherwise. This is especially true for patients with a history of malignancy. Metastatic cancer cells can look deceptively bland in CSF. It is reported that the use of IHC does not significantly increase the sensitivity over cytology alone [46–48]. Given the low sensitivity, repeated CSF cytologic examination is strongly recommended to increase the sensitivity. Conversion from positive to negative cytology is considered a response to the therapy.

Postmortem brain examination in LC may be grossly unremarkable but often shows subtle non-specific gross findings such as congestion, mild thickening, and/or slightly increased opacification of the leptomeninges (Fig. 3b). Hydrocephalus due to CSF flow obstruction may be seen. Microscopic examination shows cancer cells diffusely infiltrating the leptomeninges, particularly at the base of the brain (Fig. 3c, d). The perivascular spaces (Virchow–Robin spaces) are also usually involved.

Miliary metastasis

Miliary metastasis (MM), also termed as carcinomatous encephalitis, which was originally documented by Madow and Alpers in 1951, is a very uncommon form of brain metastasis [49]. This condition is usually seen as an advanced stage of the primary malignancy, and pulmonary adenocarcinoma is the most frequent primary tumor. The initial neurologic manifestations vary, but psychiatric symptoms, hemiparesis, and convulsion are common [50, 51]. Interestingly, progressive dementia as the presentation has also been documented [50]. Metastatic foci are usually detected by MRI with contrast enhancement. The radiological differential diagnosis includes infectious etiologies such as miliary brain tuberculosis, neurocysticercosis, or toxoplasmosis [52]. The definitive diagnosis of MM requires histologic examination, which is usually performed after death.

The gross autopsy findings are usually unremarkable or non-specific. Histologically, MM is characterized by innumerable widespread metastatic foci located in the perivascular spaces (Virchow–Robin spaces) of gray and white matter, without forming intraparenchymal mass lesions. Tumor cells can extend to the subpial spaces. Of note, the perivascular spaces have been shown to be confluent with the subpial space, but not with the subarachnoid space [53, 54].

Intravascular carcinomatosis

We have recently described an extremely rare form of metastatic breast carcinoma at autopsy (originally, inflammatory breast carcinoma), characterized by the presence of single tumor cells diffusely present within capillaries in all brain regions, without parenchymal or perivascular invasion, akin to intravascular lymphoma (Fig. 4a) [55]. Pancytokeratin IHC is very helpful to highlight intravascular single tumor cells (Fig. 4b). The patient presented with strokes, and multifocal acute ischemic changes/infarcts with widespread occlusion of significant numbers of capillaries by single tumor cells were found at autopsy. This unique pattern of metastasis is considered to represent ‘intravascular carcinomatosis (IVC)’. This patient’s initial presentation was mental status change. ‘IVC’ is distinct from miliary metastasis in that the tumor cells are confined solely to intravascular spaces and radiologically, the metastatic foci are not detected by MRI with contrast enhancement. In addition, miliary metastasis is not typically associated with ischemic infarcts [50].

Fig. 4
figure 4

“Intravascular carcinomatosis” from inflammatory breast cancer: a two large cancer cells (arrowheads) within a capillary in the edematous cerebral cortex. b Immunohistochemically, pancytokeratin (AE1/AE3) highlights intravascular single cancer cells that are widely disseminated, without involving the perivascular spaces or surrounding brain parenchyma

Full size image

Tumor-to-tumor metastasis

Tumor-to-tumor metastasis (TTM) is an uncommon phenomenon and should be separated from collision tumors, which are defined as two neighboring neoplasms that infiltrate one another. The strict criteria for “true” TTM were proposed by Campbell et al. [56] and Pamphlett [57]. In general, TTM most commonly occurs from an aggressive high-grade malignancy to a slow-growing, indolent tumor in both intra- and extracranial settings. In the CNS, the most common recipient is a meningioma, with breast and lung carcinomas reported to be the most common “donor metastases” [58]. Renal and prostatic carcinomas and malignant melanoma are also reported as donors [58]. Pituitary adenoma, schwannoma, and hemangioblastoma (HB) can also serve as the recipient CNS tumors. Metastasis to gliomas has been uncommonly recorded, with breast and lung carcinoma being the most common donors [58]. Given the rapid progression with short survival of high-grade gliomas, cases of glioblastoma harboring metastatic carcinoma practically do not exist, and if present, careful pathological examination is needed to exclude other possibilities (e.g., collision tumors).

Patients with hereditary cancer syndrome may have a higher risk of the development of TTM. In the brain, cases of HB harboring metastatic RCC have been reported in the setting of VHL disease [59, 60]. Identification of metastatic RCC within HB is very challenging and usually requires IHC (refer to the IHC section above in detail).

Given that there are no specific or reliable clinicoradiological features for CNS TTM (Fig. 5a, b), pathologic examination is crucial for the diagnosis (Fig. 5c, d) and a metastatic tumor can be unexpectedly found within a surgically resected meningioma, especially in patients with a history of malignancy. It is reported that TTM was much more frequently found in postmortem material than in surgical material [61]. At autopsy, the donor tumor is usually disseminated widely in the systemic organs. Pathologists should be aware of this uncommon phenomenon of TTM primarily in patients with advanced extracranial malignancy in both surgical and autopsy settings.

Fig. 5
figure 5

Tumor-to-tumor (pulmonary adenocarcinoma-to-meningioma) metastasis: a MRI (fluid-attenuated inversion recovery image) reveals a high intensity tumor with prominent peritumoral edema in the temporal lobe. b MRI (T1-weighted image) with contrast enhancement shows diffuse enhancement of the tumor. c Microscopically, a focus of adenocarcinoma (lower left) is present within a microcystic meningioma. (H&E stain). d Immunohistochemically, cytokeratin-7 labels cancer cells with unlabeled meningioma cells in the background

Full size image

Major metastatic brain tumors

Metastatic non-small cell lung cancer

Non-small cell lung cancer (NSCLC) is the leading cause of brain metastases. Approximately 15 % of patients already have brain metastasis at the diagnosis of NSCLC, and in about one-third, brain metastasis is the initial presentation [62]. Molecular marker analysis discloses oncogene-driven subtypes of NSCLC with distinct responsiveness to currently available, specific targeted therapies: epidermal growth factor receptor (EGFR)-mutated and anaplastic lymphoma kinase (ALK)-rearranged NSCLCs. Adenocarcinoma represents the most common histology for both of these subtypes.

EGFR mutation is seen in approximately 10 % of patients with NSCLC in the US and 35 % in East Asia. In the recent retrospective study using MRI in Japan, analyzing the largest NSCLC patient cohort (1127 patients), Iuchi et al. reported that patients with EGFR-mutated NSCLC had a significantly greater risk of developing brain metastasis than those with EGFR wild-type NSCLC (31.4 and 19.7 %, respectively) [2]. Li et al. reported similar predisposition [62]. Nearly 25 % of patients with EGFR-mutated NSCLC are found to have brain metastasis at the initial diagnosis [63]. Iuchi et al. also reported that patients with EGFR-mutated NSCLC survived significantly longer after brain metastasis than those with EGFR wild-type, although EGFR-mutated brain metastases were smaller but tended to be multiple and disseminated. Multiple current studies suggest remarkable efficacy with rapid response and mild toxicity of the EGFR tyrosine kinase inhibitors (TKIs) such as gefitinib and erlotinib for metastatic NSCLCs in the brain in addition to primary and extracranial sites, and this benefit is probably limited to NSCLCs with EGFR mutation. Brain metastasis from EGFR-mutated NSCLC in patients who were previously treated with TKIs (i.e., CNS failure) with and without progression/relapse at extracranial sites is fairly common (the latter is specifically called ‘isolated CNS failure’). This CNS failure might not be due to secondary resistance mutations even with the acquisition of such mutations in the extracranial recurrent/metastatic foci. Although the mechanisms of CNS failure remain largely unknown, poor penetrability of TKIs through the BBB (i.e., low CSF to plasma concentration ratio) might be a plausible explanation. This information is therapeutically important since strategies to boost CSF exposure of TKIs (e.g., dose escalation, pulsatile dosing, switch from gefitinib and erlotinib) can be effective [64]. Given its prognostic significance and therapeutic implications, the mutation analysis of the tumor, ideally specimens from brain metastasis if feasible, is crucial, since it remains to be elucidated whether there is inconsistency in EGFR mutation status between primary tumor and metastasis [64]. The mutation analysis is currently being performed using various molecular techniques from formalin-fixed paraffin-embedded (FFPE) tissue. No histological characteristics in brain metastasis associated with EGFR-mutated NSCLC (adenocarcinoma) have been described.

Activation of the ALK oncogene results from fusion of the ALK and the echinoderm microtubule-associated protein-like 4 (EML4) genes and occurs in about 4–5 % of patients with NSCLC. ALK rearrangements confer sensitivity to treatment with the multitargeted ALK TKI, crizotinib. Analysis of these gene rearrangements is currently, most commonly performed by fluorescent in situ hybridization (FISH), using FFPE tissue. It is reported that the brain is the most common site of new lesions during crizotinib treatment in patients with ALK-rearranged NSCLC [65, 66]. As with EGFR-mutated NSCLCs, brain metastasis found at initial diagnosis (about 25 % of patients) and CNS failure (including isolated CNS failure) after crizotinib treatment are fairly common in patients with ALK-rearranged NSCLC [63, 67]. One of the plausible explanations is low CSF to plasma ratio of crizotinib due to poor BBB penetration. Histologically, it is well known that signet-ring cell type primary NSCLCs have significant association with ALK rearrangement; however, there have been no reported histological characteristics in brain metastasis associated with these rearrangements. Gainor et al. reported two distinct rare metastatic forms from their cohort: intramedullary spinal cord metastasis and leptomeningeal carcinomatosis [68].

Information on the molecular alterations, especially druggable mutations (EGFR mutations and ALK rearrangements), is crucial for the patients and clinicians in NSCLCs. Pathologists should make every effort to save enough tissue for the molecular study, especially for small biopsy specimens, in addition to the histologic examination.

Metastatic breast cancer

An estimated 10–30 % of all breast cancer patients develop brain metastases during the course of their disease [69]. Young patient age, negative hormone receptor status, and Her-2 positivity are the main risk factors for the development of brain metastasis [70, 71]. For Her-2-positive breast cancers, brain metastasis is known to frequently occur as an isolated event after anti-Her-2 treatment (i.e., isolated CNS failure). According to the most recent study of IHC biomarkers, comparing the surgically resected brain metastases to the original breast cancers, Shen et al. reported that overexpression of Her-2 was highly concordant between these paired tumors, whereas changes of ER and PR status were seen in a substantial proportion of patients [72]. They also reported that Her-2 overexpression was an independent predictor of better survival after brain metastasis provided that the patients were treated with Her-2 targeted medications.

According to the most recent American Society of Clinical Oncology (ASCO) clinical practice guideline, patients with accessible, newly diagnosed metastases from primary breast cancer should be offered biopsy for confirmation of disease process and testing ER, PR and Her-2 status given that the status may have changed from the primary tumor [73]. In cases of discordance, the Panel consensus is to preferentially use the ER, PR, and Her-2 status from the metastasis to direct therapy if clinically indicated [73]. IHC for ER, PR and Her-2 should be performed in the surgical specimens of brain metastasis from breast. Given the reported high concordance rate of Her-2 status between the primary and metastasis, a brain biopsy only for the purpose of determining Her-2 status might not always be required [72]. To improve accuracy of the IHC testing for breast cancer, according to the most recent ASCO/College of American Pathologists (CAP) guidelines, it is recommended in preanalytical variables that the cold ischemia time (i.e., time from tissue acquisition to fixation) should be 1 h or less and that the fixation time in neutral buffered formalin should be 6–72 h regardless of size of the specimen [74]. This recommendation is also applicable to specimens of metastatic brain tumors from breast.