INTRODUCTION — Germ cell tumors (GCTs) are classified as extragonadal if there is no evidence of a primary tumor in either the testes or the ovaries. Extragonadal GCTs typically arise in midline locations, and specific sites vary with age. In adults, the most common sites are the anterior mediastinum, retroperitoneum, and the pineal and suprasellar regions. In infants and young children, intracranial GCTs and sacrococcygeal teratomas are more common than other locations.

Intracranial GCTs are discussed here. Throughout the text we generally do not distinguish between children and adults. It is important to note, however, that the literature on intracranial GCTs is largely based on children under the age of 15 years, and historical data and outcomes are being extrapolated to young adults based on a small proportion of these patients in most series.

Sacrococcygeal teratomas and extragonadal GCTs arising in the mediastinum and retroperitoneum are discussed elsewhere. (See "Sacrococcygeal germ cell tumors" and "Extragonadal germ cell tumors involving the mediastinum and retroperitoneum".)

EPIDEMIOLOGY — In North America and Europe, intracranial GCTs represent 0.5 to 3 percent of pediatric central nervous system (CNS) tumors [1]. By contrast, these tumors are substantially more frequent in Asia, with a reported incidence of up to 11 percent of pediatric CNS tumors [1]. Even in the United States, in data from Surveillance, Epidemiology, and End Results (SEER) and the Central Brain Tumor Registry (CBTRUS), Asian/Pacific Islanders have a two- to threefold higher risk of intracranial GCT compared with White Americans, suggesting that genetic factors may be more important than environmental factors in the etiology of GCT [2].

The peak incidence of intracranial GCT is during the second decade of life, with a median age at diagnosis of 10 to 14 years [3,4]. There is a male preponderance of between 2:1 to 3:1, especially with tumors in the pineal region [5,6].

ETIOLOGY

Histogenesis — Various theories have been proposed to explain the origin of extragonadal GCTs. These are discussed elsewhere. (See "Extragonadal germ cell tumors involving the mediastinum and retroperitoneum", section on 'Pathogenesis and risk factors'.)

Molecular biology — Collaborative efforts have begun to shed light on the molecular pathogenesis of intracranial GCTs. Prior to these efforts, the best described abnormalities were isochromosome 12p [7-11] and gain-of-function mutations of KIT [12,13].

In an international collaborative study, 62 intracranial GCTs were analyzed using next-generation sequencing techniques [14]. The analysis confirmed many previously described alterations and identified several novel abnormalities that may have therapeutic potential. Overall, 53 percent of tumors harbored somatic mutations in at least one of the genes involved in the KIT/Ras or AKT/mechanistic target of rapamycin (mTOR) signaling pathways.

●KIT mutations were identified in 16 tumors (26 percent), exclusively in germinomas. Mutations were clustered in exon 17 and 11, similar to mutations described in testicular seminoma, but distinct from those described in gastrointestinal stromal cell tumors (GIST). Such mutations may be of therapeutic significance, as multiple tyrosine kinase inhibitors that target KIT are in development or already in use in other malignancies.

●Mutations in KRAS or NRAS, which encode downstream targets of the c-kit receptor, were found in 19 percent of tumors, mutually exclusive with KIT mutations. Ras mutations are well described in non-small cell lung cancer and colon cancer, but no effective targeted therapies have yet been identified. (See "Personalized, genotype-directed therapy for advanced non-small cell lung cancer" and "Personalized, genotype-directed therapy for advanced non-small cell lung cancer", section on 'RAS mutations'.)

●Mutations in casitas B-lineage lymphoma (CBL), which encodes a RING finger ubiquitin E3 ligase and is a negative regulator of KIT expression, were identified in seven tumors. CBL mutations have been described in some hematologic cancers and are associated with KIT overexpression. Copy number analysis of CBL showed that 13 out of 28 intracranial GCTs in this series had either clonal or subclonal 11q loss of heterozygosity spanning the CBL locus, strongly suggesting a key role of CBL in the pathogenesis of intracranial GCTs. (See "Overview of the myeloproliferative neoplasms", section on 'Other mutations'.)

●The combined frequency of KIT, KRAS/NRAS, and CBL mutation/genetic alteration in this series of intracranial GCTs exceeds 50 percent, which is supportive of this molecular pathway's contribution to tumorigenesis and highlights the therapeutic possibility of targeting this pathway.

●Nineteen percent of the intracranial GCTs of this series showed AKT1 amplification/copy number gain, associated with elevated mRNA expression. This discovery also suggests that use of AKT1/mTOR inhibitors may be worthy of clinical investigation in patients with intracranial GCTs.

Subsequent molecular studies further confirm involvement of the MAPK (KIT/Ras) and PI3K/mTOR pathways in intracranial GCTs [4]. One study found mutually exclusive mutations of KIT or Ras in 60 percent of germinomas [15]. High expression of KIT mRNA was also associated with severe chromosomal instability. In another study, KIT and Ras mutations were observed in germinomas and all subtypes of nongerminomatous GCT (NGGCT); combining KIT, Ras, and other less frequently mutated genes such as CBL, alterations in the MAPK pathway were observed in 48 percent of all GCTs, including 64 percent of germinomas and 20 to 50 percent of other subtypes of NGGCT [16]. Mutations in the PI3K/mTOR pathway were observed in 12 percent of intracranial GCTs, with mTOR mutations (6.5 percent) being the most common, and these mutations correlated highly with basal ganglia tumors [4]. In vitro treatment with an mTOR inhibitor in tumors with PI3K pathway mutations led to tumor suppression.

This emerging genomic information offers promise that therapies targeting the MAPK and PI3K pathways could prove useful in refractory intracranial GCTs or in the newly diagnosed setting as part of a strategy to further reduce or eliminate radiation in some patients.

HISTOLOGIC CLASSIFICATION — In the World Health Organization classification system, intracranial GCTs are divided into germinomas and nongerminomatous GCTs (NGGCTs) (table 1) [17]. NGGCTs include embryonal carcinoma, endodermal sinus tumor (also known as yolk sac tumor), choriocarcinoma, teratoma (immature and mature teratoma, teratoma with malignant transformation), and mixed tumors with more than one element. Germinomas comprise 60 to 65 percent of all pediatric intracranial GCTs. Approximately 25 percent of NGGCTs are mixed and contain more than one histologic component [5,6].

Intracranial GCTs can be further defined by tumor markers secreted into the cerebrospinal fluid (CSF) and serum, as well as by the presence of histochemical markers on tumor cells (table 2). Secreted tumor markers measured in the CSF and serum include alpha-fetoprotein (AFP) and beta-human chorionic gonadotropin (beta-hCG), and immunohistochemistry is used to detect placental alkaline phosphatase (PLAP) and c-kit on tumor cells.

Histologically, pure germinomas are composed of large polygonal undifferentiated cells with abundant cytoplasm arranged in nests separated by bands of connective tissue. The histologic appearance of NGGCTs varies depending upon the specific cell types present [17]. Infiltrating small lymphocytes are often present and can obscure the diagnosis, especially in small biopsy specimens.

Intracranial GCTs can also be divided into "secreting" and "nonsecreting" tumors. Secreting tumors are commonly defined as GCTs with CSF AFP >10 microg/L (or above the institutional normal range) and/or a CSF beta-hCG level >50 international units/L. Secreting GCTs are generally considered to behave more aggressively and carry a poorer prognosis than nonsecreting GCTs.

Pure germinomas generally are associated with absent AFP and beta-hCG levels in both CSF and serum. Although an elevated AFP in either the serum or CSF effectively rules out a pure germinoma, a minority of germinomas are associated with elevated beta-hCG levels in the CSF and/or serum [18,19]. The source of the elevated beta-hCG is thought to be syncytiotrophoblasts that are associated with germinomas. (See 'Beta-hCG secreting germinomas' below.)

Another classification system commonly used in Japan separates intracranial GCTs into "good," "intermediate," and "poor prognosis" groups [20]. Pure germinomas and mature teratomas are included in the "good prognosis" group. Choriocarcinoma, yolk sac tumor, embryonal carcinoma, and mixed NGGCTs composed mainly of these three histologies are included in the "poor prognosis" group, with all other tumors included in the "intermediate prognosis" group. Patients in the "good prognosis" have overall survival (OS) exceeding 90 percent, while patients in the "intermediate" and "poor prognosis" groups have OS rates of approximately 70 and 40 percent, respectively [20,21].

CLINICAL PRESENTATION

Location — Intracranial GCTs arise almost exclusively from midline locations. The two most frequent sites are the pineal gland and the suprasellar regions, with pineal tumors occurring nearly twice as often as suprasellar GCTs. Intracranial GCTs can also arise in the basal ganglia, thalamus, cerebral hemisphere, and cerebellum [22,23].

In 5 to 10 percent of cases, patients present with tumors at both pineal and suprasellar locations (sometimes referred to as bifocal disease) [5,6]; these tumors are usually pure germinomas [4,24]. Tumor seeding or multiple tumor nodules along the lateral and third ventricles are observed in approximately 10 percent of patients.

Symptoms — Presenting symptoms of patients with intracranial GCTs depend upon the location of the tumor. Delays in diagnosis are common and are associated with a higher incidence of disseminated disease [25]. In particular, symptoms related to endocrinopathy (delayed vertical growth, diabetes insipidus, etc) are associated with delays of greater than 12 months.

Pineal tumors — Pineal tumors typically cause obstructive hydrocephalus. Patients present with signs of increased intracranial pressure (headache, vomiting, papilledema, lethargy, somnolence) in 25 to 50 percent of cases. Other symptoms associated with pineal GCTs and obstructive hydrocephalus include ataxia, behavioral changes, and decline in academic performance. (See "Pineal gland masses".)

Neuroophthalmologic abnormalities (especially paralysis of upward gaze and convergence) are present in up to 50 percent of cases. (See "Ocular gaze disorders", section on 'Parinaud syndrome'.)

Endocrinopathies are rarely associated with pineal tumors at diagnosis, although diabetes insipidus is sometimes observed and may indicate occult tumor involvement of the floor of the fourth ventricle and the suprasellar area [26].

Suprasellar tumors — Suprasellar GCTs most commonly present with hypothalamic/pituitary dysfunctions, including diabetes insipidus, delayed pubertal development or precocious puberty, isolated growth hormone deficiency, or other aspects of hypopituitarism (central hypothyroidism, adrenal insufficiency). (See "Causes, presentation, and evaluation of sellar masses".)

Suprasellar GCTs can also cause ophthalmologic abnormalities such as decreased visual acuity from chiasmic or optic nerve compression or visual field deficit (classically, bitemporal hemianopsia). Patients with suprasellar GCTs often have chronic subtle symptoms, and their tumors are diagnosed incidentally on imaging studies performed for unrelated reasons.

DIAGNOSIS AND STAGING — Histologic examination is needed to establish a definitive diagnosis of an intracranial GCT and to ascertain the histologic subtype. A tissue sample should be obtained unless surgery cannot be performed safely. Surgery to obtain tissue for diagnosis is mandatory for patients with normal cerebrospinal fluid (CSF) and serum alpha-fetoprotein (AFP) and beta-human chorionic gonadotropin (beta-hCG), as a pure germinoma or a mature teratoma must be distinguished from other benign and malignant lesions, including pineal primitive neuroectodermal tumor (PNET), ependymoma (pineal location), craniopharyngioma, Langerhans cell histiocytosis (suprasellar location), low-grade glioma, hamartoma, or metastatic disease from extracranial tumors (either location). (See 'Surgery' below.)

An elevated level of AFP in either the CSF or serum is sufficient to classify a tumor as a nongerminomatous GCT (NGGCT), although tumor tissue is useful for prognostic classification and biologic studies. Patients with an elevated beta-hCG (>50 international units/L) but normal AFP should undergo surgery if possible to distinguish a beta-hCG secreting germinoma from an immature teratoma or a choriocarcinoma, as the last two are considered NGGCTs and need more aggressive treatment.

Neuroimaging — Magnetic resonance imaging (MRI) is the preferred imaging technique for diagnosis and staging, although computed tomography (CT) is also very sensitive in detecting suprasellar and pineal GCTs.

On MRI, intracranial GCTs appear isointense or hypointense on T1 sequences and hyperintense on T2 sequences. These tumors typically show homogeneous enhancement with gadolinium or heterogeneous enhancement if cysts are present. Imaging characteristics of the histologic subtypes are similar, and MRIs cannot reliably distinguish germinomas from NGGCTs [27-29].

MRI of the entire spine is imperative for adequate staging of intracranial GCTs, since 10 to 15 percent of patients will have leptomeningeal spread at the time of diagnosis [5,26].

Tumor markers — The distinction between germinomas and NGGCTs is critical, since patients with germinomas have a more favorable prognosis and require less intensive therapy than those with NGGCTs. Tumor markers, such as AFP and beta-hCG, are helpful in making this distinction, although histologic examination is required for a definitive diagnosis.

Pure germinomas and mature teratomas typically present with normal levels of AFP and beta-hCG in both serum and CSF. Some histologically confirmed germinomas have elevated beta-hCG levels that are presumed to be due to beta-hCG secreting syncytiotrophoblasts. In such cases, elevations of serum beta-hCG are generally limited (<50 international units/L), although some tumors have levels >100 international units/L [18,19]. (See 'Beta-hCG secreting germinomas' below.)

Any tumor with an elevated AFP (>10 microg/L or higher than the institutional normal range) can be assumed to contain elements of endodermal sinus tumor and/or immature teratoma. If a histologic diagnosis is not possible because surgery is contraindicated, such a patient should be treated as having an NGGCT. Pure endodermal sinus tumor or pure choriocarcinomas are often associated with dramatic elevations in AFP (>500 microg/L) or beta-hCG (>1000 international units/L), whereas immature teratomas have less dramatic elevations of AFP and/or beta-hCG. A serum AFP >1000 microg/L has been identified as a poor prognostic indicator [30,31], but since a significant proportion of these tumors has mixed components, it is not yet feasible to use tumor markers for risk stratification without a tissue diagnosis.

Measurement of AFP and beta-hCG in the CSF is more sensitive than serum levels in detecting abnormalities. However, discordant serum and CSF tumor marker results have been observed, and both serum and CSF tumor markers should be obtained in the absence of clinical contraindications. If a lumbar puncture can be safely performed, lumbar CSF is considered more accurate for tumor markers and cytology than ventricular CSF. However, if a lumbar puncture is contraindicated, then tumor markers from ventricular CSF can be used for diagnostic purpose.

CSF cytology — CSF cytology should be obtained during staging of an intracranial GCT whenever a lumbar puncture can be safely performed. Even if MRI of the spine does not show evidence of tumor involvement, patients with positive CSF cytology are considered to have metastatic intracranial GCTs and should receive craniospinal irradiation (CSI) as part of their treatment. (See 'Treatment' below.)

Surgery — Obtaining tissue to establish a histologic diagnosis should be strongly considered for patients with a suspected intracranial GCT, unless the morbidity of the procedure outweighs the benefit. In addition, immediate neurosurgical intervention is indicated for obstructive hydrocephalus from a pineal mass or for acute visual deterioration from a suprasellar mass. (See 'Diagnosis and staging' above.)

Surgical biopsies often yield only a small sample, and this can lead to an inaccurate tissue diagnosis. As an example, in a mixed GCT that contains both germinoma and nongerminomatous components, a small biopsy may only include an area of pure germinoma. When the tissue diagnosis is discordant from the CSF and/or serum markers, treatment should be based upon the result that is associated with the most malignant histology and worst prognosis. As an example, a tissue diagnosis of a pure germinoma would be inconsistent with an elevated AFP, and such a patient should be considered as having an NGGCT. Conversely, if the tissue diagnosis reveals elements of an NGGCT, despite normal AFP and/or beta-hCG levels, the patient should be treated as having an NGGCT and not a pure germinoma.

Gross total resection at diagnosis is indicated only for patients with mature teratoma confirmed by histology and normal tumor markers, since surgery is curative and no further interventions are required [20,21,32].

A gross total resection of localized germinomas is generally not recommended because of the risk of surgical complications and because pure germinomas are exquisitely sensitive to radiation therapy (RT). (See 'Localized germinoma' below.)

The benefit of gross total tumor resection in localized NGGCTs has not been established; several large series have not confirmed that macroscopic tumor resection at the time of diagnosis improves the final outcome of children with intracranial NGGCT [21,30,31]. Instead of pursuing a macroscopic resection of an NGGCT at the initial surgery, such a procedure may be performed more safely as a "second-look" surgery after tumor reduction by chemotherapy and/or RT. (See 'Nongerminomatous GCTs' below.)

With continued evolution of chemotherapy and RT for intracranial GCTs, the role of neurosurgery for diagnosing and treating these tumors also continues to be discussed and reexamined [33].

TREATMENT

Localized germinoma — Intracranial germinomas are exquisitely sensitive to radiation. Most contemporary series have reported long-term progression-free survival (PFS) rates >90 percent for patients with localized, pure germinomas after radiation therapy (RT) alone [34-41].

Historically, patients with localized germinomas received 36 Gy craniospinal irradiation (CSI) and a boost to the primary tumor for a total of 50 to 54 Gy. Subsequent studies demonstrated that replacing CSI with whole brain or whole ventricle irradiation in patients with localized germinomas resulted in a spinal failure rate of less than 10 percent [20,36,42-48]. The patterns of relapse following whole brain or whole ventricle RT compared with CSI were not significantly different, suggesting that recurrent germinomas after RT are unlikely to be related to a volume reduction in RT. Whole ventricle RT with an additional boost to the tumor therefore replaced CSI in the treatment of localized germinoma. (See 'Whole-ventricular RT' below.)

The remaining issue with RT, even when limited to whole ventricle, however, is the incidence of clinically significant late neurocognitive and endocrine complications. Thus, clinical research has focused on reducing both the dose and volume of radiation, without compromising the excellent survival rate. Germinomas are also highly sensitive to chemotherapy, and more recent studies have added chemotherapy and reduced the dose of RT in an effort to further minimize late complications. (See 'Neoadjuvant chemotherapy' below.)

Whole-ventricular RT — The current standard of care for radiation alone (without neoadjuvant chemotherapy) in localized germinoma is 21 to 24 Gy to the whole ventricle and an additional boost to the tumor for a total dose of 40 to 45 Gy.

Although CSI is not required in patients with localized germinoma, retrospective data suggest that eliminating whole ventricle irradiation and limiting RT to the tumor increases the recurrence rate in patients with localized germinomas.

The importance of whole ventricle RT was demonstrated in a series of 35 patients with localized germinomas who did not receive CSI [46]. In this cohort, 21 patients were treated with whole ventricle irradiation, none of whom developed recurrent disease. By contrast, 5 of 14 (36 percent) who received only focal tumor irradiation had recurrent tumors within the ventricular system but outside the primary treatment field. Similarly, in a second series, the recurrence rate for patients receiving localized irradiation without ventricular coverage was higher than in those receiving CSI (28 versus 2 percent) [34]. Additional studies in support of whole-ventricular RT for localized germinomas are discussed below.

Neoadjuvant chemotherapy — Platinum-based chemotherapy regimens have a high level of activity against extracranial GCTs in children. As in adults with advanced GCTs, the most widely used combinations are bleomycin, etoposide, and either cisplatin (BEP) or carboplatin (BEJ). (See "Initial risk-stratified treatment for advanced testicular germ cell tumors".)

Based upon the excellent response and survival outcome of children and adults with extracranial GCTs, neoadjuvant chemotherapy has been explored in patients with localized intracranial germinomas, followed by a reduced dose and volume of RT, in an effort to minimize toxicity. Several series have shown excellent tumor response to chemotherapy (partial response [PR] and complete response [CR] rates nearing 100 percent), suggesting that neoadjuvant chemotherapy allows for the reduction of both the dose and volume of RT in patients with localized germinomas without compromising PFS [49-56].

While it is clear that the addition of chemotherapy may allow reduction of radiation volume (from CSI to whole-ventricular radiation), larger studies with longer follow-up have observed that further elimination of whole ventricle RT increases the tumor recurrence rate [30,57-60]:

●In a series of 60 patients with localized germinomas, neoadjuvant chemotherapy followed by 40 Gy focal RT to the tumors resulted in an eight-year event-free survival (EFS) of 83 percent [57]. Eight of 10 recurrences occurred outside the RT field, in the periventricular area [59].

●In the International Society of Paediatric Oncology (SIOP) central nervous system (CNS) GCT 96 prospective nonrandomized study, 190 patients with localized germinomas received either chemotherapy plus 40 Gy focal RT or 24 Gy CSI with a 16 Gy tumor boost without chemotherapy [60]. The five-year EFS for patients receiving chemotherapy and focal RT was less than for those receiving RT to a larger field without chemotherapy (88 versus 94 percent). In the patients who received chemotherapy plus focal RT, six of seven recurrences (86 percent) were ventricular, either alone or in combination with local tumor recurrence. In the patients who received CSI, all four relapses were at the original tumor site.

●A preliminary report of a phase II study from the Japanese CNS GCT Study Group described outcomes in 123 patients with localized germinomas, most of whom were treated with chemotherapy followed by focal RT [30]. The recurrence rate was higher in patients who received chemotherapy plus focal RT compared with those who received chemotherapy plus whole ventricle RT (28 versus 6 percent).

As discussed above, the current standard of care for radiation alone (without neoadjuvant chemotherapy) in localized germinoma is 21 to 24 Gy to the whole ventricle and an additional boost to the tumor for a total dose of 40 to 45 Gy. Despite the promise of neoadjuvant chemotherapy, further reduction of the total RT dose below 40 Gy and/or elimination of whole ventricle RT should only be done within the context of a prospective randomized trial.

Stratum 2 of the Children's Oncology Group (COG) trial for localized germinoma (ACNS1123; NCT01602666) is examining the efficacy of reducing the dose of whole ventricle irradiation (to 18 Gy) and local tumor boost (to 12 Gy) in patients whose tumors had CRs to chemotherapy and compare to historical outcomes. An interim analysis determined that the recurrence rate did not exceed expectation from historical series, and the trial is continuing accrual. A Korean prospective study showed that, in 30 patients with histology-proven localized germinomas who received neoadjuvant chemotherapy followed by 18 Gy to the whole ventricle and 12 Gy to the tumor, no recurrence was observed at a median follow-up of 3.4 years [61].

Chemotherapy only — Although nearly all patients with localized germinomas respond to chemotherapy, responses are not durable, and chemotherapy alone has resulted in unacceptable tumor recurrence rates. In two series that included a total of 64 patients with pure germinomas, recurrent disease eventually developed in 48 and 58 percent [62,63]. The Third International CNS Germ Cell Tumor Study also confirmed that a chemotherapy-only approach led to inferior EFS compared with radiation-containing regimens [64].

Beta-hCG secreting germinomas — Pure germinomas can contain syncytiotrophoblasts that produce and secrete beta-human chorionic gonadotropin (beta-hCG). Early reports suggested that beta-hCG secreting germinomas had a higher relapse rate than nonsecreting pure germinomas [39,58]. However, subsequent reanalysis of the data found that the increased rate of recurrence of beta-hCG secreting germinomas was primarily related to elimination of whole brain and/or whole ventricle RT [18,19,21,30,65,66], and no recurrences were observed in patients with beta-hCG secreting germinomas who received at least whole ventricle RT.

The major concern in patients with presumed beta-hCG secreting germinomas is whether or not the intracranial GCT actually is a pure germinoma, or whether there also is an element of choriocarcinoma or immature teratoma present. Although there is no consensus on the optimal approach for children with histologically proven germinoma and a beta-hCG >50 international units/L, a repeat surgery for a more generous tumor sample may be warranted to exclude a nongerminomatous GCT (NGGCT) and avoid overtreatment.

Disseminated germinoma — While CSI is no longer considered necessary for children with localized germinomas as long as they are treated with a combined-modality approach that uses chemotherapy and irradiation based upon the response to chemotherapy, CSI continues to be recommended for patients with disseminated germinomas based upon magnetic resonance imaging (MRI) and/or cerebrospinal fluid (CSF) findings. Current recommendations include treating bifocal tumors as localized tumors if the MRI of the spine and the CSF cytology are negative.

A change from the historical approach of using 36 Gy CSI was initially supported by a series of 49 patients, in which the dose of CSI was reduced to 30 Gy and the dose of radiation to the primary tumor was decreased from 50 to 45 Gy [34].

Additional data supporting the efficacy of a reduced dose of CSI come from the SIOP CNS GCT 96 study [60]. For patients with metastatic germinoma, the five-year PFS was 98 percent after chemotherapy, 24 Gy CSI, and a 16 Gy tumor boost (total RT dose to the primary tumor 40 Gy), suggesting that the dose of CSI can be further reduced when given in combination with chemotherapy, even for patients with metastatic germinomas.

Recurrent germinoma — In contemporary series of patients with pure germinomas, recurrences are most commonly encountered in patients who received neoadjuvant chemotherapy and a reduced dose and/or volume of RT. The majority of these patients can be salvaged with radiation with or without chemotherapy [30,57,59,60,64,67]. Salvage PFS and overall survival (OS) are reported as high as 80 percent.

Most recurrent germinomas remain sensitive to chemotherapy, and retreatment with the original regimen is a viable strategy to reinduce a complete remission. For patients who received RT to a reduced volume and for whom disease recurs outside of the radiation field, CSI alone is a reasonable salvage strategy. If disease recurs in the radiation field, then additional chemotherapy to reinduce a CR/near-CR is warranted before consolidating with additional RT.

For patients who have already received CSI, salvage is still possible, with either standard chemotherapy and reirradiation, or an aggressive approach of myeloablative high-dose chemotherapy with autologous stem cell rescue with or without additional RT, if it can be safely tolerated [68-73].

Nongerminomatous GCTs — NGGCTs are less common than germinomas, include several histologic subtypes (table 2), and have a worse overall prognosis compared with germinomas. Multimodality therapy is therefore required in all patients. Based on the available data, the standard of care includes neoadjuvant multiagent chemotherapy and RT that consists of both CSI and a tumor boost. Whether a reduced field of RT can be used for certain subgroups of NGGCTs remains controversial, as discussed below.

Chemotherapy plus craniospinal irradiation — NGGCTs are relatively insensitive to radiation alone. In historical series, patients with NGGCTs who were treated with RT alone (CSI and a boost to the tumor) had OS rates of 20 to 40 percent [5,6,21,74]. Combining neoadjuvant chemotherapy with CSI significantly improves outcomes, with OS rates exceeding 60 to 70 percent [21,75-78], especially for localized NGGCT, although results remain inferior to that for patients with germinoma.

Almost all the chemotherapy regimens for NGGCT contain a platinum compound (either cisplatin or carboplatin) and etoposide [21,76,77], and some groups add ifosfamide or cyclophosphamide to this combination [75,78]. The most common approach uses four to six cycles of chemotherapy prior to second-look surgery. This is followed by RT, including CSI (30 to 36 Gy) and a tumor boost (for a total of 54 to 60 Gy).

Results of this approach are illustrated by the COG ACNS0122 trial, which enrolled 102 children with newly diagnosed NGGCT (median age, 12 years) [78]. The majority of patients had either pineal or suprasellar tumors (54 and 24 percent, respectively). After initial surgery, patients were treated with neoadjuvant carboplatin-etoposide alternating with ifosfamide-etoposide for six cycles, followed by 36 Gy CSI and local tumor boost to 54 Gy. A second-look surgery was permitted prior to RT and performed in 15 patients. With a median follow-up of 5.1 years in the entire cohort, five-year PFS and OS were 84 and 93 percent, respectively. Among the 49 patients with localized disease who achieved a CR or PR after chemotherapy with or without second-look surgery, five-year PFS and OS were 92 and 98 percent, respectively. These results led to the subsequent COG NGGCT trial (ACNS1123), which examined the feasibility of reducing the field of radiation from CSI to whole ventricular field with tumor boost. (See 'Potential role of reduced radiation' below.)

The presence of residual disease upon completion of neoadjuvant chemotherapy is a risk factor for worse outcomes. In ACNS0122, those with no residual tumors before radiation had five-year PFS and OS of 100 percent, compared with 81 and 92 percent among those with residual tumors; of those patients with residual masses after chemotherapy, 15 patients underwent second-look surgery, and five patients had residual elements of NGGCT (one embryonal carcinoma, one mixed GCT, three malignant teratomas) [78]. In the SIOP CNS GCT 96 trial, the prognostic significance of residual disease after neoadjuvant chemotherapy was even more pronounced; PFS was 85 percent for those with no residual tumors versus 48 percent for those with residual tumors [31]. These data suggest that, while a gross surgical resection at the time of diagnosis may not be necessary, a second-look surgery in those patients with residual tumors after neoadjuvant chemotherapy (before starting radiation) may significantly improve their long-term outcome. (See '"Second-look" surgery' below.)

The magnitude of alpha-fetoprotein (AFP) elevation is also an adverse prognostic indicator in children with intracranial NGGCTs. In the SIOP CNS GCT 96 study, patients with AFP >1000 microg/L had PFS of 32 percent versus PFS of 76 percent in patients with AFP levels <1000 microg/L [31]. By contrast, the degree of beta-hCG elevation in children with intracranial NGGCTs does not appear to have an impact on survival [31,54,78].

Potential role of reduced radiation — In selected patients with localized NGGCT, the feasibility of reducing the total field of RT has been explored in an effort to reduce long-term toxicities of treatment [31,79-81]. However, results have been mixed, and decisions to omit CSI in patients with NGGCT should continue to be individualized for patients being treated outside of a clinical trial.

In the SIOP CNS GCT 96 study, 116 patients with localized NGGCT received neoadjuvant chemotherapy followed by local RT (omitting ventricular or spinal irradiation) [31]. This approach showed inferior five-year PFS (72 percent) compared with results from ACNS0122 (neoadjuvant chemotherapy followed by CSI), in which five-year PFS was 92 percent [78]. In one retrospective series, local RT was associated with a 32 percent distant recurrence rate at 10 years [80]; in a second series, local RT with or without ventricular radiation was associated with a four-year PFS of 81 percent [81]. In ACNS1123, 66 (62 percent) of 107 patients with localized NGGCT achieved a CR or PR after chemotherapy with or without second-look surgery and were eligible for reduced RT, which consisted of whole-ventricular RT plus tumor boost [79]. Among these patients, three-year PFS and OS were 88 and 92 percent, respectively. All eight recurrences involved distant relapses (six spinal alone, two combined local and spinal). Although the preponderance of spinal relapses prompted early closure of the study and was most likely attributable to the elimination of CSI, the majority of children with localized NGGCT after achieving a PR or CR with neoadjuvant chemotherapy appeared to be cured without CSI. Neurocognitive outcomes have not yet been reported and will be important to analyze to confirm the presumed benefit with reduced RT.

Pending further data, a definitive recommendation that CSI is not required for children with localized NGGCT who achieve a PR or CR after neoadjuvant chemotherapy may be premature. ACNS1123 had a relatively short median duration of follow-up of three years, so late recurrences beyond three years are possible. In addition, there were 18 patients enrolled on ACNS1123 who had either germinomas or germinomas mixed with mature teratoma, which are associated with a better prognosis [18,19] and could have contributed to the favorable survival outcomes. A more detailed breakdown of the patient subgroups in ACNS1123, based on tumor markers and histology, is needed to confirm whether patients classified as "intermediate prognosis" (eg, patients with predominantly germinoma and/or immature teratoma, with minor elements of more aggressive entities), which have shown favorable survival without CSI [21,77], can be safely treated with reduced RT. A future trial mandating up-front surgery and tissue diagnosis may be required to accurately determine whether CSI can be eliminated for certain subgroups of NGGCT.

"Second-look" surgery — Although more definitive data are needed, strong consideration should be given for "second-look" surgery in patients who have residual tumors after chemoradiotherapy [33].

There are no definitive data that suggest that gross total resection of NGGCT at the time of diagnosis improves either PFS or OS. However, resection of residual tumors after chemotherapy and/or RT may have a role, with a few small series suggesting that gross total resection may improve survival [77,82,83]. At resection, the majority of residual masses in patients with intracranial NGGCTs after chemotherapy and/or RT are mature teratoma and/or necrotic/scar tissue [33], although viable tumor cells have also been observed [77,78,82-86].

In SIOP CNS GCT 96 trial, five-year PFS was 85 percent for those with no residual tumors versus 48 percent for those with residual tumors [31]. In a study in patients with intermediate-prognosis NGGCTs from the Japanese GCT Study Group, the tumor recurrence rate was worse for those with residual masses after chemoradiotherapy compared with those without a residual mass (32 versus 5 percent) [30]. In ACNS0122, those with no residual tumors before radiation had five-year PFS and OS of 100 percent, compared with 81 and 92 percent for those with residual tumors [78].

Growing teratoma syndrome — The "growing teratoma syndrome" is defined as a solitary enlarging tumor, with normal or declining AFP and/or beta-hCG, which upon resection proves to be composed entirely of a mature teratoma. This is a well-defined event that occurs in up to 10 percent of patients with extracranial NGGCTs [87]. This phenomenon is rare in patients with intracranial NGGCTs, occurring in less than 10 percent of patients [88-95]. A contemporary review of 777 cases of pediatric intracranial GCTs identified only 39 cases of growing teratoma syndrome [96]. (See "Posttreatment follow-up for men with testicular germ cell tumors", section on 'Growing teratoma syndrome' and "Approach to surgery following chemotherapy for advanced testicular germ cell tumors", section on 'Rationale for resection of residual masses in patients with NSGCT'.)

Surgical resection of a probable growing mature teratoma should be attempted in patients with intracranial NGGCTs who experience an enlarging solitary tumor during or shortly after chemotherapy or RT despite normalization of tumor markers. This approach avoids subjecting patients to more intensive treatments due to a mistaken diagnosis of tumor recurrence or progression.

Surgical resection for a growing mature teratoma is the only curative intervention, since these lesions do not respond to chemotherapy or RT.

Recurrent NGGCTs — The prognosis for patients with recurrent intracranial NGGCTs is poor. Nearly all will have received chemotherapy and RT as their initial treatment.

High-dose chemotherapy with autologous stem cell rescue has been attempted with variable success [68-73], and prognosis for recurrent NGGCT is much worse compared with recurrent germinoma. Identification of novel effective agents for intracranial NGGCTs is urgently needed to improve clinical outcome for children with these heterogeneous and challenging tumors.

SUMMARY AND RECOMMENDATIONS — Intracranial germ cell tumors (GCTs) are rare brain tumors that typically arise in the pineal or suprasellar regions. Intracranial GCTs include a number of histologic tumor types (table 1); the primary distinction is between germinomas and nongerminomatous GCTs (NGGCTs), since NGGCTs require more intensive chemotherapy and craniospinal irradiation (CSI). (See 'Clinical presentation' above and 'Histologic classification' above.)

Initial evaluation

●Unless clinically contraindicated, the initial evaluation of a patient with a suspected intracranial GCT should include:

•Neuroimaging with magnetic resonance imaging (MRI) with and without contrast to include the head and the entire spine (see 'Neuroimaging' above)

•Measurement of alpha-fetoprotein (AFP) and beta-human chorionic gonadotropin (beta-hCG) in both serum and cerebrospinal fluid (CSF) (see 'Tumor markers' above)

•Cytology from CSF (see 'CSF cytology' above)

•Biopsy of the tissue mass to establish a histologic diagnosis, especially in patients with normal tumor markers (see 'Surgery' above)

Treatment of germinoma

●For patients with localized pure germinoma, standard treatment with radiation alone consists of whole ventricle radiation therapy (RT) with a boost field to the primary tumor (Grade 1B). (See 'Localized germinoma' above.)

●For radiation-only treatment for a localized germinoma, the current standard of care is 21 to 24 Gy to the whole ventricles, followed by a boost to the tumor for a total tumor dose of 40 to 45 Gy. (See 'Whole-ventricular RT' above.)

●Several series have demonstrated that in patients with localized germinomas who achieved a complete remission after neoadjuvant chemotherapy, radiation dose to the tumor and whole ventricle can be further reduced without compromising the excellent progression-free survival (PFS) in these patients. The ongoing Children's Oncology Group (COG) study will prospectively determine whether neoadjuvant chemotherapy followed by reduced radiation to the tumor and whole ventricle will maintain the excellent historical outcome. (See 'Neoadjuvant chemotherapy' above.)

●Although the majority of germinomas are sensitive to chemotherapy, a chemotherapy-alone approach has been associated with an unacceptable rate of relapse. (See 'Chemotherapy only' above.)

●CSI is indicated only for patients with multifocal germinomas or metastatic tumors, as identified by MRI and/or CSF cytology or those tumors that do not demonstrate either a complete response (CR) or a partial response (PR) to chemotherapy.

Treatment of NGGCTs

●For patients with NGGCTs, we recommend neoadjuvant chemotherapy followed by CSI, rather than either RT or chemotherapy alone (Grade 1B). This approach has resulted in long-term survival in 60 to 70 percent of cases, and in the most recent COG series, five-year PFS exceeded 80 percent. (See 'Nongerminomatous GCTs' above.)

●The optimal chemotherapy regimen has not been defined. The available data indicate that platinum-based regimens, such as those used in other gonadal and extragonadal GCTs, are effective. (See 'Nongerminomatous GCTs' above.)

●Available data indicate that RT is an essential component of initial treatment. The current standard of care is to use CSI plus tumor boost, as demonstrated by the excellent survival from ACNS0122. Although early data from the follow-up ACNS1123 trial suggest that the vast majority of patients with localized NGGCT who achieve a CR or PR with neoadjuvant chemotherapy appear to be cured with whole-ventricular RT and tumor boost only, further data maturation is needed. The role of reduced field RT in patients with localized NGGCT therefore remains controversial. (See 'Chemotherapy plus craniospinal irradiation' above and 'Potential role of reduced radiation' above.)

●In patients with residual masses after chemotherapy and RT, second-look surgery should be strongly considered. (See '"Second-look" surgery' above.)

●Patients with both germinomas and NGGCTs should be encouraged to participate in prospective clinical trials whenever possible.