Clinical manifestations

Presentation and course

The clinical manifestations of intracranial germ cell tumors, irrespective of histological tumor type (germinoma vs. nongerminomatous germ cell tumor), depend both on the age at presentation and whether tumor location is in the suprasellar region, the pineal region, or in both regions simultaneously.

The teratoma subtype of intracranial nongerminomatous germ cell tumor is frequently diagnosed antenatally by ultrasound or soon after birth and has a different presentation from that of older children and adults, regardless of the tumor site (45). A congenital teratoma is suggested by a history of polyhydramnios and hydrocephalus, and by a heterogeneous echogenic mass with cystic and solid components; this is associated with an almost uniformly fatal prognosis.

In young infants, germ cell tumors (most often teratoma or choriocarcinoma subtypes of nongerminomatous germ cell tumor) frequently present with protean manifestations that include irritability, listlessness, failure to thrive, macrocephaly, and a bulging fontanelle, representing less specific signs of increased intracranial pressure.

After infancy, germ cell tumors of any type in a suprasellar location and when diagnosed earlier in their course usually present with endocrine deficits. These may include anterior pituitary dysfunction with signs and symptoms of thyroid or cortisol deficiency or both; children may have growth failure from growth hormone deficiency and delayed puberty from gonadotrophin deficiency. This may also cause signs and symptoms of regression of sexual development or dysfunction in adults (84; 04). A high proportion of suprasellar germ cell tumor patients also present with posterior pituitary dysfunction (vasopressin deficiency) that causes polyuria or polydipsia of diabetes insipidus. In Jennings’ series, 33% of patients with suprasellar tumors presented with some type of anterior pituitary dysfunction, and 41% presented with diabetes insipidus (84). In a small British series of suprasellar germ cell tumors (both germinoma and nongerminomatous germ cell tumors), diabetes insipidus occurred in all 11 patients, as did partial or complete anterior pituitary failure (82). Isolated diabetes insipidus was the presenting symptom that led to diagnosis in all 12 intracranial germ cell tumors (both germinoma and nongerminomatous) and had been present for a median of 1.1 years before diagnosis in a single-center retrospective review of hypothalamic tumors (170). Occasionally, precocious puberty may develop in a prepubertal child with a suprasellar germ cell tumor as well as with a pineal or basal ganglia tumor, due either to tumor-induced hypothalamic injury or from a tumor that is producing human chorionic gonadotrophin (84; 144).

The symptoms of endocrine dysfunction from a suprasellar germ cell tumor can sometimes go undiagnosed for months or even years until visual symptoms (33%), hydrocephalus, or both, and signs of increased intracranial pressure (21%) call attention to the tumor (84). Diagnosis was delayed more than six months from the onset of presenting symptoms in 54% of 70 children with intracranial germ cell tumors, and there was a higher incidence of disseminated disease in these patients compared to those with a more prompt diagnosis (158). The visual signs and symptoms with which suprasellar germ cell tumors present can include deficits of visual acuity and visual fields (classically with a bitemporal hemianopsia because of frequent involvement of the optic chiasm). Suprasellar tumors can occasionally grow larger, invading the thalamus or basal ganglia and producing hemiparesis or movement disorder.

Pineal region germ cell tumors tend to present more acutely than do suprasellar tumors, with hydrocephalus and signs of increased intracranial pressure (headache, nausea, vomiting, lethargy, papilledema) due to obstruction of the ventricular system at the third ventricle and cerebral aqueduct (41%). Situated at the posterior aspect of the third ventricle, pineal germ cell tumors also compress the dorsal midbrain (tectum), interfering with cortical input to oculomotor nuclei and gaze centers and producing a Parinaud syndrome (34%). This syndrome consists of: (1) vertical upward-gaze paralysis, (2) decreased pupillary response to light, and (3) retractory or convergence nystagmus. Sometimes, more extensive growth of pineal tumors into the cerebellum or cerebellar peduncles can cause ataxia and tremor (19%), and growth into the brainstem may produce other cranial nerve deficits or long tract signs (01; 05; 84; 48; 74). Diabetes insipidus and anterior hypopituitarism can sometimes be found in association with germ cell tumors that appear to be isolated to the pineal region, suggesting that microscopic disease involving the pituitary axis is involved.

Patients with simultaneous pineal and suprasellar germ cell tumor masses can have the above-noted signs of either or both, depending on the size and sequence of development of the lesions.

Clinical vignette

A 14-year-old boy presented to the emergency room because of symptoms of progressive headache that had been getting worse over 3 weeks. The headaches became more frequent and occurred daily, although they were not continuous. They were frequently present in the morning, and over the previous 3 to 4 days, he had headaches along with vomiting. He felt fatigued, even during the daytime, and developed double vision that was worse when he looked toward the left. His parents had noticed that his left eye drifted inward. On examination, his mental status was notable for a mild degree of lethargy but was otherwise normal. His visual acuity and fields were normal, but papilledema was present bilaterally on funduscopic examination. His pupils were 5 mm and sluggishly reactive to light, with better response to accommodation bilaterally. He had a partial left sixth nerve palsy with decreased vertical upward gaze, and he developed convergence nystagmus when he attempted upgaze. An MRI of his brain showed a 3 cm by 2.5 cm gadolinium-enhancing mass centered in the pineal region at the posterior aspect of the third ventricle. It was obstructing the cerebral aqueduct and was associated with a moderate degree of hydrocephalus, with enlargement of the third and lateral ventricles. Preoperative evaluation revealed an elevated serum level of alpha-fetoprotein (486 IU/L) and of chorionic gonadotrophin (625 IU/L). After placement of a ventriculostomy, he underwent an open craniotomy by a supracerebellar approach. A subtotal resection of the tumor was accomplished, and a permanent ventriculoperitoneal shunt was placed several days after the resection. The histopathologic diagnosis of the tumor was a mixed tumor that had components of germinoma and embryonal carcinoma. Postoperatively, his headaches and sixth nerve deficit resolved, but the upward-gaze paresis, though improved, was still present. Staging evaluation of the tumor revealed no spinal metastases and negative cytologic examination of the CSF. The ventricular CSF levels of alpha-fetoprotein and chorionic gonadotrophin obtained just before tumor resection was 689 IU/dl and 900 IU/dl, respectively. Following surgery, he was initially treated with four cycles of chemotherapy containing cisplatin, etoposide, and cyclophosphamide, and his tumor responded with a 90% reduction of residual postoperative volume and normalization of the serum markers. His upward gaze paresis had resolved by the time of completion of the chemotherapy. He then received 36 Gy of craniospinal irradiation with a 20 Gy boost to the tumor volume. Eight weeks after completion of the radiation therapy, a surveillance MRI showed a decrease in the residual tumor volume to just a 2 to 3 mm area of enhancement in the pineal region, and it had decreased to 1 to 2 mm by his next surveillance scan. Serum and CSF markers, also monitored at the time of the follow-up imaging, remained negative as well. The residual small amount of gadolinium enhancement has remained stable, as have serum and CSF tumor markers, for the five years since diagnosis.

Biological basis

Etiology and pathogenesis

There are two theories on the pathogenesis of CNS germ cell tumors. One theory, for which evidence is most compelling for development of germinomas, is that they arise from primordial germ cells (PGC) that have migrated aberrantly from the yolk sac endoderm to midline diencephalic brain instead of their normal migration path to the genital ridge for incorporation into developing gonads (116). This migration process is orchestrated by complicated molecular events that could predispose to defects in the intricate process (60; 76). However, how these aberrantly migrated primordial germ cells become neoplasms requires further explanation. Indeed, many otherwise normal embryos have been demonstrated to have extragonadal primordial germ cells, but few develop into tumors. Thus, these cells must either be stimulated to become neoplastic or released from normal controls on cell division or cell death. One such process known in other biological systems and necessary for normal embryonic and postnatal development is termed apoptosis or programmed cell death. Increasing data support a theory of derangement of cell death, rather than uncontrolled cellular proliferation in the development of some tumors (111). Ultrastructural changes characteristic of apoptosis are reported in mouse primordial germ cells (140). Because pathways that regulate apoptosis are complex, an error anywhere in the molecular system could lead to the development of a germ cell tumor. Several lines of evidence support primordial germ cells as the direct cell of origin of germinomas. First, primordial germ cells are large with plump, round nuclei and prominent nucleoli, histologically resembling germinoma tumor cells (59); second, primordial germ cells and germinoma cells share protein expression patterns of PLAP and c-KIT, which are not found on other pluripotent stem cells or on non-germinomatous germ cell tumors; third, from methylation profiling, the methylation status of intracranial germinomas closely resembles that of migrating primordial germ cells, with the unique characteristic feature of global DNA hypomethylation versus the hypermethylation of nongerminomatous germ cell tumor and more primitive embryonic stem cells (129; 61). In this primordial germ cell theory of origin of germ cell tumors, nongerminomatous germ cell tumors could develop from transformed primordial germ cells into more differentiated tissue-like tumors through epigenetic reprogramming (134).

The second theory is the so-called “pluripotent stem cell theory” in which germinomas may originate from primordial germ cells, but nongerminomatous germ cell tumors develop from pluripotent embryonic stem cells that reside in both gonadal and extra-gonadal tissue and express pluripotency markers, including OCT4, NANOG, and SOX2, which are also expressed robustly in embryonal carcinomas (64). As well, the methylation profile of intracranial nongerminomatous germ cell tumors displays a global hypermethylation pattern, which is like that of the more primitive embryonic stem and not like the hypomethylated primordial germ cells (61). Thus, embryonal carcinoma is the prototype of all nongerminomatous germ cell tumors and is regarded as the neoplastic counterpart of embryonic stem cells. Subsequently, more differentiated subtypes: yolk sac tumors, choriocarcinoma, and immature teratoma evolve from pluripotent stem cells with or without the intervening form of embryonal carcinoma (134).

There may be a connection between the pathogenesis of intracranial germ cell tumors and the neuroendocrinologic mechanisms that regulate the complex physiological functions of reproduction. This is suggested by both the predominance of the location of germ cell tumors in the diencephalon where this regulation is centered, and by the frequent temporal association of intracranial germ cell tumors with puberty. It is possible that the hormonal surges of puberty may sometimes induce malignant changes in these normally quiescent cells (84; 164).

The cytogenetic abnormalities of germ cell tumors and how they may play a role in their pathogenesis have been studied most extensively in germ cell tumors of testicular origin. The most characteristic chromosome anomaly found in testicular germ cell tumors of all histological types is the presence of an isochromosome 12p in more than 80% of cases (151).

Less had been known about the cytogenetic pathogenesis of intracranial germ cell tumor and whether they are similar to their testicular counterparts. Frequent gains on chromosome 12p and the characteristic isochromosome 12p of testicular germ cell tumors have been seen in intracranial germ cell tumors, but the majority of 12p gains in intracranial germ cell tumors are different from that of testicular tumors, with either whole 12p arm gain or association with various other complex structural abnormalities (19). From a study of a large number (27) of intracranial germ cell tumors, 10 regions of significant copy number gain and 11 regions of significant copy number loss were identified, different from gonadal germ cell tumors. The overall pattern of genomic aberration was similar between germinoma and nongerminomatous germ cell tumors, although a few subtype peak regions were identified (171). The most significant region of copy number gain was 12p13 where multiple candidate genes, including CCND2, are located. Gain of gene copies of CCND2, involved in the inactivation of the tumor suppressor gene RB1, were found in 52% of 27 cases with no differences between the germinomas and NGGCT. This aberration and those in RB1 (13 q14) suggest that the cycIin/CDK-RB-E2F pathway may play a critical role in the pathogenesis of intracranial germ cell tumors. These findings taken together with those of a phase II study of the favorable responses of pRB-expressing refractory tumors to treatment with the cyclin-dependent kinase (CDK) 4 and 6 inhibitor, palbociclib, support the possible use of such cyclin-dependent kinase inhibitors in upfront therapy for central nervous system (CNS) germ cell tumors (GCTs) with activated cyclin-dependent kinase/retinoblastoma pathways (174).

Mutations of the p53 gene on chromosome 17 were found in 26% of 19 intracranial germ cell tumors, predominantly in the mixed nongerminomatous tumors (94). Although these TP53 tumor suppressor gene mutations do not occur at a high frequency in intracranial germ cell tumors, some types of intracranial germ cell tumors have been shown to carry the MDM2 gene amplification and protein overexpression. The MDM2 protein has important interaction with p53 protein, inactivating and targeting it for rapid proteolysis; this finding suggests that the disruption of the balance of p53-MDM2 interaction might play an important role in the tumorigenesis of intracranial germ cell tumors (80). Genetic alterations in another tumor suppresser gene at the INK4a/ARF locus, whose gene product is known to have a major influence on p53-MDM2 interactions, were also found in 71% of 21 intracranial germ cell tumors, further supporting that the p53-MDM2 gene interactions may be important in tumorigenesis of intracranial germ cell tumors (81).

Wang and colleagues reported on a large and comprehensive genetic analysis of 62 intracranial germ cell tumors, including germinomas and nongerminomatous germ cell tumors, by next-generation sequencing, single nucleotide polymorphism array, and expression array (175). In germinomas, there were mutations in the KIT pathway, as well as in its downstream mediators, KRAS and NRAS and its negative regulator, CBL. Nongerminomatous germ cell tumors did not have KIT mutations but did have mutations in the downstream RAS mediator and in its negative regulator, CBL. Nongerminomatous germ cell tumors also had mutations in CBK, a gene that encodes RINGfiner ubiquitin E3 ligase, which is a negative regulator of receptor tyrosine kinases (RTK) including CKIT. Also found were alterations in the oncogene AKT/m TOR pathway, including copy number gains of the AKT1 locus, with corresponding up-regulation of AKT1 expression (175). The MEK inhibitor, dasatinib was reported to be effective against myeloid leukemia with CBL mutations and could be beneficial for the treatment of CBL-mutated nongerminomatous germ cell tumors (110).

A number of approved tyrosine kinase inhibitors target the KIT pathway and could be of therapeutic benefit in targeting relapsed nongerminomatous germ cell tumors that have an altered KIT component. There are not yet any approved targeted therapies for cancers with KRAS mutation, but MEK inhibitors have been effective in KRAS-mutated small cell lung cancer (83). As well, ERK inhibitors have been effective in MEK inhibitor-resistant cells (70). However, the role of targeted inhibition in CNS nongerminomatous germ cell tumors remains unclear, given their intra-tumor heterogeneity and lack of consistent molecular targets (172).

Epidemiology

Intracranial germ cell tumors are relatively uncommon in western countries, making up only 1% and 3% to 4% of brain tumors in adults and children, respectively (23; 74; 62; 135). However, intracranial germ cell tumors are much more common in Japan, where they constitute 4% of all brain tumors (169), and they are thought to represent 12% to 16.5% of brain tumors in children (99; 109).

A 1985 retrospective analysis of 398 informative cases of all intracranial germ cell tumors in the literature reported that 65% of the tumors were germinomas and 35% were nongerminomatous germ cell tumor, including 18% teratomas (both malignant and benign), 5% embryonal carcinoma, 7% endodermal sinus (yolk sac) tumors, and 5% embryonal carcinomas (84). Some of the nongerminomatous tumors were of mixed histology and classified in terms of their most malignant tissue component. In a Japanese series, mixed germ cell tumors with multiple histologic components, including malignant nongerminomatous germ cell tumors, were the most common variants, comprising 44% of the cases (153). A higher fraction of nongerminomatous germ cell tumors than is known in western series was also reported in another Japanese series of germ cell tumors (133). Nongerminomatous germ cell tumors appear to be more frequent than germinomas in younger children between birth and nine years of age, where teratomas and choriocarcinomas predominated (84).

Primary intracranial germ cell tumors, including nongerminomatous germ cell tumors, can occur at any age but have a peak age at diagnosis in the second and third decade, with the majority (68%) diagnosed between 10 and 21 years (84; 115; 71). In a retrospective analysis of reported intracranial germ cell tumors, the peak age of occurrence for both males and females was 10 to 12 years (84).

Primary intracranial germ cell tumors arise predominantly in midline central nervous system locations, most frequently in the pineal and the suprasellar regions. In Jenning’s review, considering all types of germ cell tumors together, 95% of these tumors originated near the third ventricle along an axis between the suprasellar cistern (37%) and the pineal gland (48%), 6% occurred sequentially or simultaneously in both sites, and the remaining 8% occurred in other sites. Among nongerminomatous germ cell tumor, the primary site of origin was suprasellar alone (18%), pineal region alone (65%), both simultaneously (3%), and at other locations (14%). Dissemination of nongerminomatous germ cell tumor in other nonmutually exclusive sites at diagnosis included metastases in spinal cord (7%), in other CNS locations (23%), and in systemic locations (6%). There was extension into the hypothalamus (presumably in the suprasellar primary tumors) in 15%, and into the third ventricle in 31% (84).

Overall, there is a definite male predominance for intracranial germ cell tumors, reported in Jennings’ analysis as 1.88:1 for germinomas and as an even greater 3.25:1 for nongerminomatous germ cell tumors. However, in younger children, they seem to occur with equal frequency (71).

Prevention

There are currently no known effective preventative measures or treatment for primary intracranial germ cell tumors, including the nongerminomatous germ cell tumor type. A timely diagnosis and awareness of potential complications may contribute to a better outcome with regard to morbidity and mortality.

Differential diagnosis

The differential diagnosis of an intracranial nongerminomatous germ cell tumor depends on its location, but it also varies with age and geography of the patient population. A spectrum of tumors arises in the pineal region with variation in the incidence of the different histological types between series. In a large pediatric series from Japan, germ cell tumor comprised 73% of pineal region tumors, the majority of which were of mixed histology (152). In contrast, from a single-institution North American series of 25 biopsied pediatric pineal region tumors, 32% were pineal parenchymal tumors (pineocytomas and pineoblastomas), 32% were glial tumors (astrocytomas or gangliomas), and only 32% were germ cell tumors, including both germinoma (4%) and nongerminomatous germ cell tumor (28%) (137). This incidence was similar to the 30% incidence of germ cell tumor reported by Bruce and Allen in another pineal region pediatric surgical series from the United States, in which 7% were germinomas and 23% were malignant nongerminomatous germ cell tumors. Pineal cell tumors, gliomas, or primitive neuroectodermal tumors constituted the remainder (27). Similarly, in a large surgical series from France that included adults and children, only 23% of pineal region masses were germ cell tumors (100). In another single institution, large surgical experience consisted of 154 pineal region tumors in adult and pediatric patients at Columbia University in New York; considerable histologic diversity was found among them. Only 37% were germ cell tumor (17% germinoma and 20% nongerminomatous germ cell tumor), 28% were glial tumors, and 23% were pineal cell tumors (28). In a series of 36 patients with pineal tumors at the University of California at San Francisco, 56% had germ cell tumors (31% germinoma), whereas the rest were pineal parenchymal or glial tumors (48).

The differential diagnosis of a germ cell tumor that arises in suprasellar location is different from that of the pineal region and also varies within the patient population with respect to age and geographical distribution. In children, suprasellar masses can be diencephalic or hypothalamic gliomas or hamartomas or craniopharyngiomas, in addition to the germ cell tumors that constitute about one third of tumors in this location (165). In adults, this list is expanded to also include suprasellar meningiomas and large pituitary tumors that extend into the hypothalamus.

Previously, the presence of synchronous, bifocal suprasellar and pineal region tumors in the setting of detectable hCG (5 to 100 mIU) and normal AFP in serum and CSF was thought pathognomonic for germinoma and that such bifocal tumors did not require tissue diagnosis based on reports that bifocal nongerminomatous germ cell tumors did not exist (139; 39). But a report of 14 synchronous bifocal tumors disclosed three nongerminomatous germ cell tumors, two of which had no elevation of AFP or b-HCG above the normal limits for a germinoma, suggesting that biopsy of bifocal tumors is advisable (03).

Diagnostic workup

Neuroradiologic imaging is the first step in identifying a tumor of the pineal or suprasellar region that may prove to be a nongerminomatous germ cell tumor. CT will demonstrate a suspected tumor, although MRI with and without gadolinium is probably the best method for delineation of tumor anatomy, and may suggest specific histopathology, but MRI does not substitute for a tissue diagnosis (71; 179).

Among nongerminomatous germ cell tumors, benign teratomas tend to have mixed densities on MRI and CT, often with large cysts and areas of calcification but with distinct tumor margins (183; 58). Occasionally, teeth can be identified within a benign teratoma (184). Malignant teratomas are also heterogeneous, but they have fewer and smaller cystic areas, more irregular tumor margins (sometimes with peri-tumor edema), and are less likely to be calcified than mature teratomas (58). Endodermal sinus tumors tend to be irregular in shape and to be lower density than germinomas on CT; they have more heterogeneous enhancement and sometimes edema (58). However, pineal parenchymal tumors and pineocytomas are also commonly calcified on CT (184), and on MRI they have mixed T1 signal and high T2 signal, so they are not readily distinguished from germinoma. Similar features are seen with pineoblastoma (183). Thus, it is apparent that although these imaging modalities are good for defining tumor anatomy, neither CT nor MRI can predict specific tumor histology with a sufficient degree of certainty to be used alone when making a diagnosis (183; 179).

Histologic diagnosis. Because the differential diagnosis of nongerminomatous germ cell tumors is fairly broad for both of the common tumor locations, the suprasellar and pineal regions, tissue confirmation of tumor histopathology is nearly always critical for optimal tumor management, with the one exception of a secreting nongerminomatous germ cell tumor. In the pineal region, in addition to all types of intracranial germ cell tumors, the differential includes pineocytomas, pineoblastomas, and astrocytomas; in the suprasellar region, it includes craniopharyngioma and hypothalamic glioma.

For suprasellar tumors, which are more easily reached than pineal tumors, an open surgical procedure has been considered justified to obtain a tissue diagnosis for some time (165).

However, because the surgical approach to the pineal region is more difficult, prior to the 1980s, a standard diagnostic approach for pineal region tumors was to observe their radiographic response to a small dose of irradiation (42; 133). This approach was based on the relatively high incidence of germinomas in this location (especially in Japan), the known radiosensitivity of germinomas, and the relative inaccessibility of tumors in the pineal region. A prompt and marked reduction in tumor volume after a trial of 20 Gy of irradiation was taken as evidence that the tumor was a radio-sensitive germinoma, and a full course of radiation therapy was then completed. Tumors that were radio-resistant were managed in a number of ways. A biopsy confirming a glioma would have dictated appropriate involved-field irradiation or higher dose and larger field irradiation may have been given for a presumed nongerminomatous germ cell tumor.

More widely, the diagnosis of pineal tumors has changed since the 1980s, and a general consensus now exists among most neurosurgeons, radiation oncologists, and neuro-oncologists that tissue diagnosis is nearly always necessary for optimal management of both pineal and suprasellar lesions, except for germ cell tumors that secrete markers into CSF or serum. Moreover, since the development of endoscopic neurosurgical techniques, endoscopically guided biopsy of pineal region tumors is usually the best and least morbid way of obtaining diagnostic tissue, often in conjunction with treatment of hydrocephalus by the performance of a third ventriculo-cisternotomy (21; 163). The exception to the need for tissue diagnosis by any means for either pineal or suprasellar tumors is in the setting of elevated serum or CSF tumor markers. Surgery or endoscopic biopsy is not necessary for the sole purpose of obtaining tissue for histologic diagnosis of tumors that secrete alpha-fetoprotein or beta HCG at levels greater than 50 to 100 IU/ml, as such tumors are considered virtually certain to be nongerminomatous germ cell tumors and may be treated without tissue diagnosis (08; 11; 09). In fact, upfront surgical resection for secreting germ cell tumors may be contraindicated, based on the finding of death from postoperative complications in four of 21 patients who underwent primary resection of their secreting tumors on the German MAKEI 89 study (33).

The change in approach toward surgical intervention for diagnosis or treatment of nonsecreting pineal tumors has come about, in part, because of improvements in neurosurgical techniques that have decreased morbidity and mortality and facilitated surgery in this location that had previously been considered inaccessible. Operative morbidity for pineal region tumors is currently reported as less than 2% to 5% (137; 48; 73; 28). There is a greater appreciation for the limitations of empiric diagnostic irradiation, especially in pediatric and western populations where the incidence of radio-responsive germinomas can be as low as 4% to 30% (137; 100; 48; 28). Awareness of the limitations of diagnostic irradiation also increased with the recognition that nongerminomatous pineal region tumors such as pineal parenchymal tumors and nongerminomatous germ cell tumors may also initially respond to radiation therapy, but invariably progress thereafter without other therapy (137; 103). Indeed, the prognosis of nongerminomatous germ cell tumor and pineoblastoma treated with various more intensive combined irradiation and chemotherapy regimens is better than previously reported for these tumors, and the prognosis is critically dependent on making a correct initial histologic diagnosis (31; 145). Finally, irradiation of more the benign tumors that do occur in the pineal region can also be inappropriate therapy as curative surgical resection may be possible with less morbidity (62; 28).

The best method for obtaining tissue for histological diagnosis of pineal and suprasellar tumors, whether by stereotactic biopsy or at open resection, is debated among neurosurgeons. Stereotactic biopsy for both suprasellar and pineal tumors has been endorsed by some because of low morbidity (104; 42). However, an open procedure is advocated by most, probably for two main reasons (15; 163). The first reason concerns the sampling error that can occur with small needle biopsies, especially germ cell tumors that are frequently heterogeneous. The misdiagnosis of a nongerminomatous germ cell tumor as a germinoma because of such an error would result in significant undertreatment of the tumor, with a high likelihood for a poor outcome. The second reason for an open surgical approach is that it allows for the possibility of more radical surgery for a tumor type that might benefit from a greater degree of resection.

It has been thought that biopsy/tissue diagnosis was not required for synchronous, bifocal suprasellar and pineal region tumors in the setting of detectable hCG (5-100 mIU) and normal AFP in serum and CSF because this presentation was thought pathognomonic for germinoma. This was based on reports that nongerminomatous germ cell tumors presenting like this did not exist (139; 39). However, in a report of 14 synchronous bifocal tumors, three tumors were nongerminomatous germ cell tumors, two of which had no elevation of AFP or b-HCG above the normal limits for a germinoma, suggesting that biopsy of bifocal tumors is advisable (03).

Tumor staging and tumor markers. A suspected intracranial germ cell tumor in a suprasellar, pineal, or other midline location that produces and secretes the tumor markers, alpha-fetoprotein (AFP) and/or above-threshold level beta-human chorionic gonadotrophin (b-HCG) into cerebrospinal fluid and serum, is presumed to be a nongerminomatous germ cell tumor. Sufficient elevation of these markers in serum or CSF confirms the tumor histology as a nongerminomatous germ cell tumor and may make surgery to obtain a tissue diagnosis unnecessary. Thus, both serum and CSF marker levels of alpha-fetoprotein and human chorionic gonadotrophin should be done preoperatively whenever possible because the finding of any elevation of alpha-fetoprotein or a sufficiently high human chorionic gonadotrophin may obviate surgery (125). Both serum and at least lumbar CSF marker levels should be obtained because serum AFP levels are often greater than in CSF, and CSF human chorionic gonadotropin levels are often greater than in serum; moreover, lumbar CSF chorionic marker levels are almost always greater than or equal to ventricular levels (142; 101). Postoperative levels also should be measured as a baseline for following therapy response.

If AFP is elevated (above laboratory norm) in serum or CSF, it is confirmation that the tumor has elements of either pure endodermal sinus tumor (yolk sac), embryonal carcinoma, or malignant teratoma (08; 48; 71). Endodermal sinus tumors appear to secrete only alpha-fetoprotein, the levels of which may be high, whereas embryonal carcinoma and malignant teratoma can produce less elevated levels of alpha-fetoprotein (less than 700 IU/L), human chorionic gonadotrophin, or both. Human chorionic gonadotrophin is a placental glycoprotein, of which the beta chain is present in fetal serum and in pregnant women (71).

When human chorionic gonadotrophin alone is elevated, the tumor may contain choriocarcinoma or germinoma elements, but the degree of the elevation of this marker distinguishes them. High levels of chorionic gonadotrophins, somewhat arbitrarily set at 50 to 100 mIU/L are considered to predict the presence of choriocarcinoma tissue, whereas lower values can occur with germinoma in the absence of any nongermiomatous germ cell tumor elements (08). However, there is currently no international consensus on the threshold level of chorionic gonadotrophin permitted in a germinoma versus a secreting NGGCT with choriocarcinoma elements (125; 52). Japanese investigators have identified subgroups of germ cell tumors that have “intermediate prognosis,” including germinomas with histologically defined syncytiotrophoblastic elements that may secrete higher levels of chorionic gonadotrophin, although the absolute maximum level is not yet established (113). However, at the 2nd international CNS Germ Cell Tumor Symposium in November 2005, Fujimaki and Matsutani, on behalf of the Japanese Brain Tumor Study Group, reported the largest experience assessing this question. They compared 131 nonsecreting germinomas with 39 germinomas that secreted chorionic gonadotrophin to levels in serum or CSF of up to 200 mIU/L and found no difference in the progression-free or overall survivals between these two groups (52).

Thus, in summary, any elevation of alpha-fetoprotein is confirmation that the tumor contains nongerminomatous elements, either endodermal sinus tumor or, if human chorionic gonadotrophin is also elevated, embryonal carcinoma or malignant teratoma. Elevated levels of human chorionic gonadotrophin alone may predict the presence of a choriocarcinoma nongerminomatous germ cell tumor, although the current threshold levels set at 50 to 100 mIU/L in most North and South American and European studies may include some chorionic gonadotrophin-secreting germinomas (08; 11; 09; 52). These associations are sufficiently strong that presurgical detection of any elevation of alpha-fetoprotein or a high level (greater than 50 to 100 IU/L) of human chorionic gonadotrophin in serum or CSF of a patient with a pineal or suprasellar tumor means that a tissue diagnosis may not be required. If an endoscopic biopsy of a secreting tumor is considered to have more than minimal morbidity risk, treating it as a presumed nongerminomatous germ cell tumor, without histologic confirmation, is likely the best approach. However, in view of the evidence that the threshold level of chorionic gonadotrophin at 50 to 100 mIU/L may be too low and may include secreting germinomas that do not require therapy as intense as that for malignant nongerminomatous germ cell tumors, any patient with a chorionic gonadotropin level between 50 and 200 mIU/L might be better served by obtaining a biopsy for histologic confirmation of their diagnosis. Nevertheless, attempts at extensive surgical tumor resection in this setting are probably contraindicated, in view of the finding of death from postoperative complications in four of 21 patients who underwent primary resection of their secreting germ cell tumors (33). CSF marker levels usually exceed those of serum, and lumbar fluid levels frequently exceed ventricular levels (08). Therefore, when serum levels are higher than those of the CSF, the possibility of metastases from a systemic germ cell tumor should be considered, although, this expectation has been contradicted for AFP in at least two series. In a retrospective study of contemporaneous serum and CSF levels of tumor markers in patients with CNS germ cell tumors (germinomas and nongerminomatous germ cell tumors), serum AFP levels were greater than those in the CSF in all six nongerminomatous germ cell tumors, whether lumbar or ventricular CSF, though lumbar AFP levels were greater than or equal to ventricular in most. Lumbar CSF chorionic gonadotropin levels were also greater than or equal to ventricular levels. (101). In a series of 67 patients with intracranial germ cell tumors (42 germinomas and 25 nongerminomatous germ cell tumors), when tumor markers were measured simultaneously in serum and CSF, AFP levels were concordant for diagnostic utility in 98%; but again, AFP levels were higher in serum than in CSF in most; chorionic gonadotropin levels were concordant for diagnostic utility in only 72% and were more often higher in CSF than serum (142). In another series, when tumor markers were measured simultaneously in serum and CSF, marker elevation was sometimes found only in the CSF (11). Thus, staging evaluations should include marker levels obtained from both serum and lumbar CSF to prevent missing the diagnosis of a secreting tumor.

Circulating tumor DNA and CSF metabolites are promising biomarkers for CNS germ cell tumors and could differentiate germinomas from nongerminoma germ cell tumors, but will still require validation before clinical adoption (167).

An emerging biomarker tool for the diagnosis of CNS germ cell tumors is microRNAs (miRNA) (181). Serum and CSF miR-371a-373pmir-302/367 clusters found in patients with testicular nongerminomatous germ cell tumors have shown higher sensitivity and specificity than standard tumor markers, with potential for diagnostic and prognostic benefit (123). Among 27 miRNAs candidates with differential expression among various nongerminomatous germ cell tumors, serum and CSF overexpression of miRNA-214-3p was associated with cisplatin resistance through repression of BCL2-like pro-apoptotic proteins (75). With the largest data set, to date, the diagnostic benefit of four miRNAs: miR-371a, miR-372, miR-367, and miR-302d, particularly in biomarker-negative germinoma, and as serum, can be readily obtained. This may also allow monitoring of response to therapy and detection of early relapse (156).

Radiologic and CSF cytology staging. A staging evaluation to determine the amount of residual tumor at the primary site after surgery and the extent of metastatic disease is critical in the treatment planning of nongerminomatous germ cell tumors. It is best accomplished by postoperative gadolinium-enhanced MRI of the brain and by gadolinium-enhanced MRI of the entire spine, either preoperatively or postoperatively. This type of staging is particularly important with germ cell tumors because the incidence of CNS tumor spread via ventricular and subarachnoid pathways to third ventricles and the spinal cord is high (up to 42%) (84). In a retrospective survey of all Canadian intracranial germ cell tumors diagnosed and treated between 1990 and 2004, 32% of the nongerminomatous germ cell tumors were disseminated at diagnosis (89). Information about the extent of the disease is needed to determine the volume of radiotherapy required (involved-field vs. craniospinal); this information can also help to determine whether a boost of irradiation may be needed at the primary site or for any bulky metastatic deposits. Also, the presence or absence of tumor spread may help to address the question of whether an intracranial germ cell tumor could possibly represent metastasis from a primary systemic germ cell tumor. An isolated suprasellar or pineal mass would virtually exclude this possibility, whereas the presence of leptomeningeal metastases would not and further evaluation would be necessary.

After histologic confirmation of a nongerminomatous germ cell tumor (or germinoma), postoperative staging should also include CSF cytological examination for malignant cells. Evaluation for spreading of the disease outside the CNS is usually unnecessary unless the patient’s symptoms suggest it, given the rarity of systemic metastases from primary intracranial germ cell tumors. Less than 4% of 233 patients with pineal tumors had pulmonary or abdominal metastases, and these were all at relapse and in association with a ventriculoperitoneal shunt (62).

An isolated metastatic relapse in the abdomen was reported in a 7-year-old boy with an initial third ventricular nongerminomatous germ cell tumor and ventriculoperitoneal shunt, but without leptomeningeal metastases at diagnosis (16).

Management

Surgical issues. When hydrocephalus is present at diagnosis of a suspected intracranial germ cell tumor, the initial management decision is whether early treatment with CSF diversion is indicated. This decision should be informed by the knowledge that nongerminomatous germ cell tumors can have a very prompt response to upfront adjuvant therapy, either radiation therapy or chemotherapy, which could fairly rapidly reduce tumor size and relieve obstructive hydrocephalus.

There is general agreement that a tissue confirmation of diagnosis is nearly always necessary in the management of patients with intracranial pineal and suprasellar tumors (with the exception of secreting nongerminomatous germ cell tumors). More controversial is the therapeutic role for more radical surgery with histologically confirmed or marker-secreting nongerminomatous germ cell tumors.

Nongerminoma germ cell tumors are much less radio-responsive than their germinoma counterparts, and with conventional radiotherapy alone, usually after a small biopsy, long-term survival had been poor with virtually no chance for survival with some of the histologic subtypes (84). For such marginally responsive tumors, the prevailing wisdom in oncology is that decreasing tumor burden by surgical means should increase the curative potential of subsequent radiotherapy and chemotherapy (04). A few studies suggest that an extensive resection improves survival. In a Japanese series of histologically confirmed intracranial germ cell tumors treated between 1963 and 1994, among the 30 pineal malignant nongerminomatous germ cell tumors treated with surgery and radiotherapy alone (60%), radiation therapy plus chemotherapy (22%), or chemotherapy alone (18%), there was significantly better 5-year survival among 14 patients whose tumors were extensively resected (70%) compared to 16 patients who had only partial removal or biopsy (10%) (114). From a retrospective review of intracranial germ cell tumors at seven institutions, improved survival was associated with the extent of resection of 43 nongerminomatous germ cell tumors, of which 45% also received adjuvant radiation therapy and chemotherapy and 15% received adjuvant radiation therapy alone (155). A literature review of 66 cases of intracranial nongerminomatous germ cell tumors secreting high levels of beta HCG, including pure choriocarcinoma and mixed tumors with choriocarcinoma elements, demonstrated total or subtotal resection to be a significant independent good prognostic factor (160). Another factor in the surgical decision-making process is the possible benefit of a more extensive tumor resection to obviate placement of a ventriculoperitoneal shunt (15).

However, the combined data from several European studies contradict therapeutic benefit of more extensive upfront surgery, demonstrating no positive correlation between extent of resection and event-free survival of 48 patients with intracranial nongerminomatous germ cell tumor (31). Moreover, among more than 250 patients recruited (mainly from the UK, Germany, France, and Italy) to the international intracranial germ cell tumor study, (SIOP CNS GCT 96), 48 tumor resections of secreting nongerminomatous germ cell tumors were considered unjustified, according to the protocol, because the marker secretion was sufficient for diagnosis of nongerminomatous germ cell tumor. The toxicity from surgery is likely underreported, as there was no specific protocol form for that purpose, nevertheless, was severe in 10 patients, three of whom died from their complications (131). Similar severe toxicity was also seen in the German MAKEI 89 study in which four of 21 patients with secreting germ cell tumors died from postoperative complications of primary tumor resection (33).

Another surgical consideration is that of so-called “second-look surgery”. Because nongerminomatous germ cell tumors have shown moderately high response rates to chemotherapy, it has been proposed that more radical resection might better be reserved for those tumors that require second-look surgery because of an incomplete response to initial chemotherapy (11). This approach may be useful not only as a therapeutic measure after incomplete response to initial chemotherapy but also as a diagnostic one, to verify the histology of any residual tissue in this setting. In an increasing number of cases where this is done, such tissue has been determined to be nonmalignant. In the context of a French Society of Pediatric Oncology study of 27 patients with intracranial nongerminomatous germ cell tumors, all 14 patients who underwent second-look surgery for residual radiographic abnormalities after chemotherapy were found to have only fibrosis, mature teratoma, or in a few cases, immature teratoma (12). Similar findings at second-look surgery for intracranial germ cell tumors have been confirmed by others (25; 56; 163; 132). Clearly, most of such patients would not likely benefit from or require further chemotherapy, and this would not have been possible to determine without the second surgery. However, second-look surgery may also sometimes be therapeutically beneficial (as discussed in the Chemotherapy section), such as in a series of nongerminomatous germ cell tumors where gross total resection of tumor was accomplished at second-look surgery in nine of 11 patients who had residual tumor after some chemotherapy/radiation therapy (96). Also, in the reported outcomes of the SIOP-CNS-GCT-96 trial, there was a significantly increased rate of relapse among 53 patients with residual tumor at completion of therapy than among 93 patients without, supporting a possible role for resection of residual tumor at that time (34).

The so-called “growing teratoma syndrome” is another complication of intracranial pineal region mixed germ cell tumors with surgical implications. This is a mixed germ cell tumor with a secreting portion that responds to chemotherapy and a nonsecreting portion of usually mature teratoma that continues to grow during or after chemotherapy. This syndrome was first observed among systemic gonadal or extragonadal non-CNS mixed germ cell tumors, usually with metastases, but is reported as intracranial nongerminomatous germ cell tumors and may occur in up to 6% to 7% of these tumors (92). The treatment of this type of tumor should include chemotherapy and radiation therapy for the secreting portion and surgery for the mature teratoma, although complete resection may not always be possible (69; 180; 18; 139; 173). A rapidly progressive mature pineal region “growing teratoma” that was unresectable was reported to have stabilized on treatment with the cyclin-dependent kinase 4 and 6 inhibitor palbociclib (157).

Radiation therapy. Malignant central nervous system nongerminomatous germ cell tumors are much less sensitive to irradiation than are germinomas. In a review of patients from published reports before 1985, long-term disease-free survival rates of less than 25% were reported for malignant nongerminomatous germ cell tumors after conventional radiation therapy, with virtually no survival reported for some of the histological subtypes (84). In a later literature analysis of pineal region tumors treated with radiotherapy, Fuller reported 30% long-term survival among 17 nongerminomatous germ cell tumor patients (62). Several small series reported outcomes with radiotherapy alone, ranging from 21% to as high as 60% 5-year survival (115; 48; 74; 126; 38; 68). However, even considering the best results with radiation therapy alone, the prognosis could be improved. Thus, it is generally agreed that chemotherapy should be a part of therapy for intracranial nongerminomatous germ cell tumors (74; 177; 11; 30; 145).

Moreover, because of the relative radio-resistance of nongerminomatous germ cell tumors compared to germinomas in addition to concerns about the late effects of irradiation, attempts have been made to treat nongerminomatous germ cell tumors with chemotherapy alone. However, combined data from several European intracranial nongerminomatous germ cell tumor studies showed that nine of 11 patients (81%) treated with various platinum-based chemotherapy regimens but no irradiation, died from their disease. However, 20 of 27 patients (76%) treated with chemotherapy plus irradiation were alive and free of disease four years after diagnosis (31). In an international CNS germ cell tumor study, almost 50% of 26 nongerminomatous germ cell tumors treated initially with a carboplatin-based chemotherapy regimen alone had a relapse, and afterward there was salvage with irradiation, but lower survival than protocols using radiation therapy upfront (11). The same basic approach was used in a later study of 12 nongerminomatous germ cell tumors, with similar results of 6-year 50% event-free survival and 58.3% overall survival (41). An attempt to treat 13 secreting nongerminomatous germ cell tumors with carboplatin-based chemotherapy alone also resulted in 12 relapses; unfortunately, only six of them could then be salvaged (13). These data support the view that nongerminomatous germ cell tumors need a combined initial approach that includes both irradiation and chemotherapy.

The optimal volume of irradiation for nongerminomatous germ cell tumors has been controversial. It is generally agreed that craniospinal irradiation should be given when there is evidence of CNS dissemination, including cases with positive CSF cytology with MRI evidence of cranial or spinal metastases and usually in cases with multiple midline tumors such as simultaneous pineal and suprasellar sites of bulky tumor (04; 177).

The optimal irradiation volume in cases of nongerminomatous germ cell tumors with localized disease is still debated. Several studies show a trend supporting the need for a volume larger than involved field for these tumors. In a study of 26 nongerminomatous germ cell tumor patients treated with multimodal therapy, including neoadjuvant cisplatin-based chemotherapy and radiation therapy, among 16 M0 patients (without metastatic disease at diagnosis) who received whole ventricular field irradiation, four of the six relapses were in the spine outside the radiation field, without local recurrence (146). In a retrospective analysis of 13 malignant nongerminomatous germ cell tumors (excluding immature teratomas), spinal relapses did not occur in any of the five patients who received craniospinal irradiation but did occur in three of eight who did not (10). Yoshida and colleagues concluded from their study of 30 newly diagnosed germ cell tumors that craniospinal axis irradiation was necessary in the treatment of nongerminomatous germ cell tumors (182). Moreover, the high rate of ventricular relapses among the 13 of 26 relapsed nongerminomatous germ cell tumors after chemotherapy without initial radiotherapy (even among those with complete responses) argues for at least whole ventricular radiation volumes in this group (11). Most recently, there was an attempt to reduce the radiation volume from CSI to whole ventricular for patients with localized nongerminomatous germ cell tumors who achieved a complete or partial response to chemotherapy. This resulted in early closure of the reduced radiation arm of the study due to eight relapses, all distant in the spine and out of the radiation field, among 49 eligible patients who received reduced whole ventricular volume (51).

The best outcomes reported for nongerminomatous germ cell tumors have been with multimodality therapy that included both chemotherapy and craniospinal irradiation. These outcomes include the 86% 5-year relapse-free survival among 14 patients in a German study and the 79% 5-year event-free survival among another 14 patients in a Children’s Oncology Group study (31; 98). There was also an excellent 84% 5-year relapse-free survival reported for the Children’s Oncology Group study employing craniospinal irradiation and carboplatin-based chemotherapy (65). Taken together, these findings support a role for craniospinal irradiation volume for nongerminomatous germ cell tumor patients, even those with localized disease at diagnosis.

Conversely, a British study of 21 nongerminomatous germ cell tumors (excluding malignant teratomas) reported that only one of the ten relapses was from outside the irradiated field. Among six patients with intracranial localized nongerminomatous germ cell tumors who did not receive craniospinal irradiation, from a combined series from UCSF and Stanford universities, one patient experienced an isolated spinal relapse but was salvaged (68). A report of the outcomes of a Society of Paediatric Oncology trial, SIOP-CNS-GCT-96, in which 116 nonmetastatic patients received four cycles of chemotherapy (cisplatin/etoposide/ifosfamide) and focal radiation therapy only (54 Gy), resulted in a 5-year progression-free survival of 72% and overall survival of 82%; seven of the 27 (26%) relapses after treatment were outside the radiation therapy field, leading to the conclusion that focal radiotherapy is adequate for localized nongerminomatous tumors (34).

In a study from the French Society of Pediatric Oncology, the abstracted data of 27 secreting nongerminomatous germ cell tumors treated with chemotherapy and focal radiotherapy, except for patients with metastatic disease at diagnosis, showed there was a 4-year event-free survival of 70%, although the pattern of relapse was not reported (12). Sixteen nongerminomatous germ cell patients with nonmetastatic disease who received reduced volume radiation therapy after a response to six cycles of chemotherapy (three focal radiation therapy after complete response and 13 whole ventricular radiation therapy with residual tumor after completion of chemotherapy) had 4-year failure-free survival of 81% and overall survival of 92% (43). These findings support radiation volume reduction, at least for some patients with focal disease at diagnosis.

Extensive radiation dose-response data for nongerminomatous germ cell tumors are lacking. Most patients treated with radiation therapy have received involved-field doses between 40 and 60 Gy, with an estimate of the majority receiving 50 to 54 Gy, but with broad variability of irradiation volume and other adjuvant therapy (143; 42; 31; 177; 145; 10; 154). Among 13 nongerminomatous germ cell tumor patients who received involved-field radiotherapy at a median dose of 43 Gy (± chemotherapy, ± craniospinal), local failure was seen in three of 10 patients who received less than 54 Gy but in none of the three who received greater than or equal to 54 Gy (maximum 60 Gy) (10). Among 17 nongerminomatous germ cell tumor patients treated with multimodality “sandwich” therapy, there were no local relapses among the seven patients who received 54 to 55 Gy irradiation to the involved field, but there was locally progressive disease in one in nine patients who received 50 Gy to the involved field (145). In the two studies with the best 5-year relapse-free survival rates of 80% and 79%, respectively, all patients received greater than or equal to 50 Gy to the involved field plus craniospinal irradiation and cisplatin-based chemotherapy (30; 98).

Chemotherapy. Systemic germ cell tumors are highly chemosensitive, and cisplatin-based regimens containing vinblastine, bleomycin, or etoposide have greatly improved the survival of high-risk disseminated gonadal germ cell tumors (49). Moreover, brain metastases from systemic germ cell tumors have been responsive to chemotherapy. Rustin reported that all 10 patients suffering from germ cell tumor metastatic to brain responded to combination sequential multidrug chemotherapy (149). Speculation that this kind of therapy might be effective for primary CNS germ cell tumor led to the demonstration of responses of recurrent CNS tumors (germinoma and nongerminomatous germ cell tumor) to salvage chemotherapy (05; 95). The feasibility of using neoadjuvant chemotherapy to allow radiation therapy dose-reduction for newly diagnosed CNS germinomas was then demonstrated by Allen and confirmed by Jereb in two separate single-agent trials with cyclophosphamide, and in a trial with carboplatin (07; 06; 85).

Multidrug chemotherapeutic regimens effective in the treatment of systemic germ cell tumors have proven useful in the treatment of the relatively radio-resistant intracranial nongerminomatous germ cell tumors for which the prognosis is poor with radiation therapy alone. Responses of both recurrent and newly diagnosed nongerminomatous germ cell tumors to various combinations of these agents are well-documented (05; 07; 90; 112; 118; 95; 79; 78; 138; 32; 50; 141; 162; 182; 72; 37; 11; 145; 146).

Moreover, a trend toward improved survival rates was suggested among newly diagnosed nongerminomatous germ cell tumor patients treated with multimodality therapy in a number of small series. Kida and colleagues reported a 2-year progression-free survival of approximately 40% among 10 newly diagnosed nongerminomatous germ cell tumor patients treated with either cisplatin alone or combined with bleomycin and vinblastine, plus irradiation. There was an initial 100% response rate and 50% complete response rate to the chemotherapy (90). Results from a Japanese intracranial germ cell tumor study showed a 2-year survival rate of 67.7% among 30 nongerminomatous germ cell tumor patients who received cisplatin, bleomycin, and vinblastine in addition to irradiation, compared to only 46.5% 2-year survival in the control arm that received only irradiation (112). Yoshida and colleagues achieved a 2-year survival of 48% among several newly diagnosed nongerminomatous germ cell tumor patients using cisplatin and etoposide with radiation therapy (182). Five patients with alpha fetoprotein-producing intracranial nongerminomatous germ cell tumors treated with cisplatin and etoposide and radiation therapy had complete responses of their tumors with no recurrence at a mean of 53 months after diagnosis (78). From a literature review of 66 patients with high beta HCG-secreting tumors, treatment with chemotherapy was an independent good prognostic factor, as were radiation therapy and more extensive surgery (160).

Robertson and colleagues reported 5-year actuarial event-free survival of 67% (± 9%) (median follow-up 4 years) among 18 newly diagnosed intracranial nongerminomatous germ cell tumor patients treated on a multicenter study, with a “sandwich” multimodality therapy regimen. There was a 75% response rate to the neoadjuvant chemotherapy (four cycles of cisplatin and etoposide) followed by treatment with irradiation and additional adjuvant chemotherapy (four cycles of carboplatin, etoposide, bleomycin, and vinblastine) (145). A later study that employed four neoadjuvant chemotherapy cycles of cisplatin, etoposide, and ifosfamide achieved a 55% complete response rate in 22 evaluable patients; this increased to 67% after conversion to complete response in three additional patients who had initial partial responses, with two cycles of stem cell-supported, dose-intensified chemotherapy with carboplatin and cyclophosphamide (146). Six-year relapse-free survival was 63% +/- 10%.

The best earliest studies with very good progression-free and overall outcomes employed cisplatin-based chemotherapy. A critical effect of the dose of cisplatin on the outcome of intracranial nongerminomatous germ cell tumor was seen in the combined data from several different European studies wherein 17 patients who received greater than or equal to 400 mg/m2 cumulative cisplatin dose in a multiagent regimen (plus irradiation) had an 86% event-free survival when followed for nearly four years, compared to only 56% in those who received 200 mg/m2 cumulative cisplatin dose and were followed for 5.5 years (31). Also, 14 patients (12 with localized disease) treated in a children’s oncology group study, which employed cisplatin-based chemotherapy (400 mg/m2/cumulative dose) and irradiation that included craniospinal field, resulted in 79% 5-year progression-free survival (98). Equally good outcomes were achieved with the 400 mg/m2 cumulative cisplatin dosage in the most recent SIOP trial; 116 patients with nongerminomatous germ cell tumors with localized tumors who received chemotherapy plus 54 Gy focal radiotherapy demonstrated a 72% 5-year progression-free survival, and 33 patients with metastatic disease who received chemotherapy plus craniospinal radiation therapy demonstrated a 68% 5-year progression-free survival (34).

Although cisplatin has been used in the majority of platinum-based regimens for nongerminomatous germ cell tumors, a French group reported 70% 4(+)-year event-free survival among 27 intracranial nongerminomatous germ cell tumors, employing carboplatin rather than cisplatin in their multiagent regimen, along with etoposide and ifosfamide (12). A Children’s Oncology Group clinical trial using carboplatin-based chemotherapy and craniospinal irradiation for 102 intracranial nongerminomatous germ cell tumors resulted in an excellent 84% 5-year event-free survival (65).

The same carboplatin-based chemotherapy regimen was employed in the subsequent COG trial, ACNS1123, in which patients with localized nongerminomatous germ cell tumors who achieved response to chemotherapy (CR or PR) received whole ventricular radiation volume instead of craniospinal. Despite the distal spinal pattern of failure of all eight relapses, the 3-year progression-free survival was 87.8% and overall survival 92% (51)

A novel approach, employing neoadjuvant therapy that consisted of combined chemo- and radiotherapy administered before complete excision of tumor, was successfully used among 11 intracranial nongerminomatous germ cell tumor patients. Two of the 11 patients achieved a complete response and six a partial response before tumors were gross totally resected in 9. Ten of the patients are alive and free of progressive disease a mean of 96 months from diagnosis (96).

As previously noted, the possibility of withholding radiation therapy altogether in favor of chemotherapy alone proved to be an ineffective strategy for nongerminomatous germ cell tumor. This approach was explored in an international trial where both germinomas and nongerminomatous germ cell tumors were treated with fairly intensive chemotherapy consisting initially of four cycles carboplatin, etoposide, and bleomycin, followed by two more identical cycles if there was a complete response after the first four cycles, and by chemotherapy intensified by cyclophosphamide if there was less than a partial response after the first four cycles (11). Although there was a nearly 80% eventual complete response rate of the 26 nongerminomatous germ cell tumors treated in this manner, 13 of 26 had a relapse and only half were salvaged with irradiation or other therapy.

A Japanese group reported that a rapid logarithmic rate of decline of very high initial serum markers (AFP > 1000 ng/ml or HCG > 2000mIU/ml) in response to chemotherapy was of significant prognostic value for improved survival; the absence of such a decline may indicate the need for more aggressive therapy (88).

Chemotherapy may be effective in the particular setting of the relatively rare intracranial malignant teratoma type of nongerminomatous germ cell tumors, which are most common in infants. Optimal therapy is not certain, but complete resection, where possible, has generally been thought to be most beneficial. However, in a single case of an infant with a large intracranial malignant teratoma that recurred after a gross total resection, the tumor was completely responsive to eight cycles of platinum-based chemotherapy, and the child remained progression-free at least two years off-therapy (63).

A review of the role of high-dose chemotherapy with autologous stem cell support in patients with recurrent or refractory nongerminoma germ cell tumors confirms that durable tumor control can be achieved using this modality (35; 119; 44; 54; 24; 124; 36).

Therapy summary. From these observations, it may be concluded that the most effective treatment for patients with nongerminomatous germ cell tumor consists of combination platinum-based chemotherapy that includes a minimum dose of cisplatin (400 mg/m2, total dose or the equivalent of carboplatin) as the cornerstone agent, plus radiotherapy. However, the optimal volume of radiotherapy for patients with localized disease is still debated. There are data supporting at least whole-ventricular volume, and possibly craniospinal volume, even in patients with localized disease (98; 65; 51); but data from an SIOP study of a large number of nongerminomatous germ cell tumor patients with localized disease argue for focal radiation therapy in these patients (34). It remains unclear why patients with nongerminomatous germ cell tumors show some significant differences in event-free survival after treatment on similar therapy regimens as noted above, even those that included high dose platinum-based chemotherapy and comparable radiotherapy doses and volumes. Possible explanations include the small number of patients treated on the various protocols, differences in efficacy between cisplatin and carboplatin, unrecognized synergistic effects of the various drug combinations, and the influence of timing and the specific order of components of the multimodality therapy. How survival may be further improved and therapy-related toxicity mitigated is not yet certain, but the question would seem to be posed best in a cooperative group setting where the number of patients treated for this rare tumor could be sufficient to answer it. The currently ongoing Children’s Oncology Group trial for nongerminomatous germ cell tumors is one such study, aiming to decrease therapy-related late effects without sacrificing survival by reducing radiation dose and volume based on response to chemotherapy.

The chemosensitivity of systemic germ cell tumors supports the possibility that some poor outcomes of intracranial nongerminomatous germ cell tumors with prior chemotherapy regimens may have resulted from a failure to achieve adequate drug levels at tumor sites in the brain. Dose-intensification of chemotherapy may be a way to address this issue. Bone marrow ablative chemotherapy supported by autologous bone marrow transplant or peripheral blood stem cell transplant has been shown to have curative potential for some relapsed systemic germ cell tumors, and may have promise as a consolidation of first-line therapy for some high-risk gonadal germ cell tumors (122; 121). A few relapsed or progressive intracranial germ cell tumors have also shown responses to such an approach (108). An international pilot study of high-dose chemotherapy using etoposide (1.5 g/m2) and thiotepa (900 mg/m2) followed by peripheral blood stem cell support put six of 12 patients into complete remission who had relapsed germ cell tumor (nine nongerminomatous germ cell tumors, four germinomas) and put four of 12 patients into partial remission, with a median follow-up of 16 months (14). The questions of the optimal regimen for relapsed intracranial nongerminomatous germ cell tumors, as well as which substrata of nongerminomatous germ cell tumor patients might benefit from more intensive chemotherapy regimens earlier in their course of treatment, should be addressed in the cooperative group setting.

Outcomes

The prognosis in terms of long-term disease-free survival with nongerminomatous germ cell tumors is a function of the approach to therapy. Generally, outcome for nongerminomatous germ cell tumors treated mainly with radiation therapy (at various dosages and volumes), has been poor, with a 7% to 30% long term survival rate (84; 67; 126; 38; 62). Modestly better outcomes in nongerminomatous germ cell tumor patients treated with radiation therapy alone were reported in a few single-institution series. These were 5-year survival rates of 45% in 13 nongerminomatous germ cell tumor patients (74), 34% in 29 nongerminomatous germ cell tumor patients (115), and 37% in eight nongerminomatous germ cell tumor patients (48).

In most series, all subtypes of malignant nongerminomatous germ cell tumors, including mixed tumors with any malignant nongerminomatous germ cell tumor component and all of the pure subtypes of nongerminomatous germ cell tumor (choriocarcinoma, embryonal carcinoma, endodermal sinus, malignant teratoma) are considered together for prognosis. However, several Japanese groups have singled out nongerminomatous germ cell tumor subtypes that appear to differ prognostically from the rest, at least in their patient populations. One series of 153 histologically verified germ cell tumors determined 10-year survival rates of approximately 93% for both pure germinomas and mature teratomas (which they currently classify for treatment as “good prognosis”), 71% for malignant teratomas (“intermediate prognosis”), and 23% for other pure malignant nongerminomatous tumors (“poor prognosis”) (114). Based on some of their initial observations, the Japanese Pediatric Brain Tumor Study Group conducted a multi-institutional phase II study for primary intracranial germ cell tumors in which treatment was stratified according to these prognostic groups (113). In the “good prognosis” group, 75 pure germinomas received three cycles of chemotherapy with carboplatin and etoposide, followed by 24 Gy focal radiotherapy, with an 88% 4-year progression-free survival. Ten patients with germinoma with syncytiotrophoblastic components (STGCs) and 18 patients with malignant teratomas classified as “intermediate prognosis” were treated with three cycles of carboplatin or cisplatin plus etoposide followed by 30 Gy generous focal field and 50 Gy boost to primary tumor site, as well as five additional cycles of the same chemotherapy over 1.5 years; the STGCs had 100% 4-year progression-free survival, and the malignant teratomas had 89% 4-year progression-free survival. (Mixed tumors with mainly germinoma or malignant teratoma components have also been included in the “intermediate prognosis” group, but none of these were treated on this protocol.) Nine patients with pure malignant nongerminomatous germ cell tumors (or mixed with predominant malignant nongerminomatous germ cell tumor component) classified as “poor prognosis” were treated with ICE chemotherapy (ifosfamide, carboplatin, and etoposide) with 30 Gy craniospinal radiation therapy and a 50 Gy primary site boost, followed by five additional cycles of ICE and had 33% progression-free survival. The pattern of a better “intermediate prognosis” for malignant teratomas or mixed tumors with predominant malignant teratoma has been seen in several other studies stratified according to this same schema (10; 154; 93). Thus, all these studies demonstrated a significantly better prognosis for malignant teratomas than for any other malignant nongerminomatous germ cell tumor. In a 1985 review by Jennings of 398 germ cell tumors, teratomas (both malignant and benign) treated “conventionally,” had slightly better 5-year progression-free survival (32%) than the other nongerminomatous germ cell tumors, although survival rates did not approach that of the Japanese studies. Likewise, 12 malignant teratomas treated with radiation therapy in Britain between 1962 and 1987 had 5-year progression-free survival of only 18% (42). The reason for this discrepancy is not clear, but it may lie in the specific therapy as well as inherent differences between the populations.

The generally poor outcome of malignant nongerminomatous germ cell tumors treated only with radiation therapy has led to other approaches, including treatments with chemotherapy alone. These have not been as successful as hoped, with a high percentage of relapses that frequently could not be salvaged with irradiation (11; 13). The best outcomes for nongerminomatous germ cell tumors have been with multimodality therapy that included both chemotherapy and irradiation, with cures being achieved in approximately two thirds of patients (31; 13; 128; 146). A series of 18 nongerminomatous germ cell tumor patients initially treated with four cycles of cisplatin-based chemotherapy followed by irradiation and further chemotherapy had 4-year progression-free survival of 67% (145). Very good outcomes were reported in two series: one reported 14 patients from a German study (80% survival) and the other reported 14 patients (12 with localized disease) treated in a Children’s Oncology Group study (79% survival), both of which employed cisplatin-based chemotherapy and irradiation that included craniospinal fields (31; 98). These good outcomes taken together with the excellent 84% 5-year relapse-free survival among 102 patients in the children’s oncology group study, ACNS0122, that employed craniospinal irradiation and carboplatin-based chemotherapy support the use of craniospinal irradiation for nongerminomatous germ cell tumors (65).

However, a report of outcomes from the SIOP-96 study disclosed a 72% 5-year progression-free survival among 116 nongerminomatous patients with nonmetastatic disease and a response to six cycles of chemotherapy who received only focal radiotherapy (34). And despite the spinal relapse pattern of all eight of 49 patients who received whole ventricular radiotherapy after complete or partial response to chemotherapy on COG trial ACNS1123, the 3-year- progression-free survival was 87.8%, also supporting that radiation volume reduction for at least some patients with focal disease at diagnosis (51).

Morbidity and sequelae of tumor. Patients with nongerminomatous germ cell tumors can have persistence of the neurologic or endocrinological presenting symptoms and signs even after successful therapy for the tumor, and just as at initial presentation, these would be tumor site-dependent. Some abnormalities of visual acuity or visual fields caused either by tumor compression of optic pathway structures, as in the case of suprasellar germ cell tumor, or by prolonged elevation of increased intracranial pressure with suprasellar or pineal region tumors, may not return to normal even after tumor eradication. Endocrinological deficits associated with suprasellar nongerminomatous germ cell tumors almost always persist after the tumor is treated. Therapy given for these tumors also has potential to produce similar neurologic and endocrinological problems that may be difficult to distinguish from residual effects of the tumor (72).

Surgical morbidity. The overall surgical morbidity among the 85 survivors of intracranial germ cell tumors from a Japanese series was 19% (16 of 85), and included primarily minor neurologic deficits such as upward gaze palsy and visual field defects (154). The significant postoperative mortality in four of 21 patients who underwent primary surgical resection for marker-secreting nongerminomatous germ cell tumors serum or CSF in the German MAKEI 89 study underscores the probable contraindication to major primary surgery for secreting tumors in which the diagnosis can be made without histologic confirmation (33). In other series, operative morbidity for pineal region tumors has been reported as less than 2% to 5%, with primarily transient abnormalities of eye movement, ataxia, and cognitive function (137; 48; 28). Hemorrhage into tumor appears to be a particularly common occurrence among choriocarcinoma-containing tumors (high-level beta HCG-secreting tumors), particularly just after biopsy of the tumor or after biopsy during radiotherapy. From a literature review of 66 cases, tumor hemorrhage was confirmed in 22, leading the authors to recommend an early, more radical operation to prevent it (160).

Therapy late-effects and cognitive-neurodevelopmental-functional sequelae. Successful treatment of intracranial germ cell tumors with radiation therapy, as with other primary CNS tumors, raises concerns about cognitive sequelae, particularly in children because the deleterious effects on intelligence are related to radiation dose and volume and are inversely proportional to age at the time of therapy (40; 47). Radiation therapy is also not completely without cost among older patients (150; 91; 161). Significant late neurocognitive sequelae were reported in five of 27 evaluable germinoma patients treated mainly with whole-brain irradiation at a median age of approximately 16 years (161). Sakai reported “intellectual retardation” in four of 30 patients whose mean age was 14.9 years with intracranial germ cell tumor (23 germinoma, seven other germ cell tumor) and who were mostly treated with radiation therapy (150). Learning difficulties of a mild degree were noted informally in 10 of 16 CGT survivors under 16 years of age at diagnosis, and difficulties of a moderate degree were noted in another three of these patients (91). A Canadian study describing long-term toxicities of intracranial germ cell tumors in the adolescent and young adult population found that physician-reported neurocognitive impairment was the most common late toxicity, with a 10-year cumulative incidence of 38.5% (178).

A Japanese study reported the outcomes of 111 patients with all types of intracranial germ cell tumor and median age at diagnosis of 14 years (range 2 to 37 years); the patients were treated primarily with radiation therapy (154). The Karnofsky performance status at follow-up (median 8 years) was less than 80% in 26 (31%) of the 84 survivors, the majority of whom suffered from cognitive dysfunction and from some motor paresis or visual disturbances. Radiological evidence of possible radiation injury to brain parenchyma was seen in 19 patients (ie, atrophy in 14 patients; multifocal encephalomalacia in three patients; focal necrosis in two patients). Neurocognitive deficits with IQs between 65 and 83 were seen in seven patients (8% of survivors).

In another study of therapy-related late effects after treatment of nine patients with intracranial germ cell tumors (mean age 14 years), all five patients who underwent psychometric testing had neurocognitive deficits, with full scale and verbal IQs in the normal range but with performance IQs that were consistently below average and seemed to lead to considerable limitation of daily life activities (17).

A longitudinal evaluation of neurocognitive function after treatment of CNS germ cell tumors in children found that although general cognitive abilities were intact and remained stable, a significant decline in working memory, processing speed, and visual memory was evident over time (107). Tumor location appeared to be important in the trajectory of these deficits. In a retrospective study of cognitive function and memory in 26 patients treated for intracranial nongerminomatous germ cell tumors, there was a surprisingly high incidence (42%) of significant memory deficits that were unrelated to overall cognitive function as measured by IQ (176). As well, in a retrospective review of 56 Taiwanese patients less than 20 years of age with germinoma or nongerminomatous germ cell tumors, location in basal ganglia versus pineal or suprasellar region was associated with lower full-scale IQ, and treatment with craniospinal irradiation versus whole ventricular irradiation was associated with a worse outcome (102).

Audiologic and vision deficits. Among the 84 survivors in Sawamura’s series, late-onset hearing loss requiring a hearing aid was detected in five patients who received only radiation therapy (154). However, chemotherapy, especially cisplatin (a useful agent in the treatment of nongerminomatous germ cell tumor), can potentially cause dose-dependent hearing loss. This late effect is well documented from its use for other tumors, including brain tumors in children and adults (22; 55). Chemotherapy with cisplatin was associated with high-tone hearing loss in a large proportion of germ cell tumor patients in a Japanese study but was mild due to dose modification with careful otological monitoring during therapy (154). Among 112 adolescent and young adult patients in a Canadian multicenter study of intracranial germ cell tumors, the cumulative incidence of ototoxicity was 39.2% for patients who received cisplatin, compared to 3.6% for those who received carboplatin but no cisplatin (178).

Although vision deficits are frequently associated with the tumor itself, two survivors in this same series also developed postradiotherapy deterioration of visual function without evidence of tumor progression. Both had moderate visual disturbances at diagnosis and received radiation therapy to the suprasellar region (154).

Cerebral vascular occlusion. Radiation-induced cerebrovascular injury is a known complication of cranial irradiation and is reported after irradiation for brain tumors of various types and for CNS prophylaxis of hematologic cancers in adults and children, although the exact incidence is uncertain (53; 77; 117; 120). Specifically, among intracranial germ cell tumors, radiation-induced occlusive disease was reported in three of 85 germ cell tumor survivors irradiated in the parasellar region at 6, 14, and 16 years of age (154). Two patients had strokes 2 and 14 years after diagnosis, respectively; they had no other cerebrovascular risk factors at the time. The other developed a large but asymptomatic dural arteriovenous malformation 11 years after the irradiation.

Endocrine and growth sequelae. Generally, the impaired pituitary functions present before treatment of intracranial germ cell tumor persist or worsen after tumor remission (72). These include posterior pituitary dysfunction with diabetes insipidus, anterior pituitary deficits with hypothyroidism, adrenal insufficiency, gonadotrophin deficiency causing delayed puberty or sexual dysfunction, and growth failure from growth hormone deficiency. Pituitary hormone replacement was required in 68% of Sawamura’s 111 intracranial germ cell tumor patients. An additional five of 31 patients with no deficits before irradiation developed late-onset pituitary dysfunction, considered an effect of therapy (154). In the series of Kiltie and colleagues, 14 of 17 patients required some kind of hormone replacement therapy, eight of which had panhypopituitarism (91). All nine intracranial germ cell tumor patients in a study were found to have growth hormone deficiency after therapy, six of whom did not have deficits at diagnosis; several developed the deficiency many years after therapy (17). Failure to achieve predicted height may also be the result of spinal irradiation in children who are still growing at the time of therapy (159; 46).

Secondary neoplasms. Both radiation therapy and chemotherapy of CNS tumors, including intracranial germ cell tumors, have the potential to produce secondary neoplasms (105; 130; 86). Chemotherapy-related leukemias, most frequently of a myeloid type, are among the most common second malignancies in both adult and pediatric populations (02). However, the incidence of such leukemias was low (only 0.37%) in a large series of 538 systemic germ cell tumor patients treated on clinical trials with etoposide and platinum-based regimens similar to that used for nongerminomatous germ cell tumor (130). Radiation-induced secondary brain neoplasms occurred in four of the 84 germ cell tumor survivors (4.8%) of Sawamura’s series (154). Two patients, 14 and 6 years of age at diagnosis, developed fatal glioblastomas 10 and 12 years after craniospinal and local irradiation for solitary germ cell tumors. Two boys, 10 and 12 years of age at diagnosis, developed meningiomas in the radiation field 15 and 19 years after treatment.

Special considerations

Pregnancy

Fertility. Because of the high frequency of endocrine deficits (including gonadotrophin deficiencies) particularly in patients with suprasellar region germ cell tumor, it would not be surprising if there were some decrease in eventual fertility among both male and female patients who have had these tumors. Moreover, irradiation of the hypothalamic and pituitary axis, as well as some of the chemotherapy used to treat nongerminomatous germ cell tumors (especially alkylating agents such as cyclophosphamide), may contribute further to infertility in treated patients, although the specific incidence of this late-effect among intracranial germ cell tumors has not been reported.

Pregnancy complications. The risk of fetal death or malformation is higher for a woman undergoing treatment of an intracranial germ cell tumor with chemotherapy during the pregnancy. The risk would be similar to that for treatment of any cancer during pregnancy with the same chemotherapy and is dependent on the specific agents used. Risk is greatest during the first trimester, during the period of fetal organogenesis (29).

Anesthesia

The primary complication of nongerminomatous germ cell tumors that could influence anesthesia is the diabetes insipidus with which tumors in the suprasellar region frequently present. This endocrine abnormality makes the maintenance of normal fluid and electrolyte balance complicated during the administration of general anesthesia. The potential for development of hypernatremia with inadequate fluid replacement (or hyponatremia with fluid overload during treatment with replacement antidiuretic hormone) could produce life-threatening consequences (26). Increased intracranial pressure from the obstructive hydrocephalus that is frequently produced by intracranial germ cell tumors, particularly by those in the pineal region where nongerminomatous germ cell tumor is predominate, can cause increased anesthetic risk; this must be managed accordingly during surgery.