Clinical manifestations

Presentation and course

The clinical manifestations of germinomas depend on the location of the tumor. Pineal region tumors generally present with evidence of increased intracranial pressure secondary to ventricular obstruction at the aqueduct of Sylvius and are accompanied by signs related to compression or infiltration of the dorsal midbrain. Parinaud syndrome (loss of upward gaze, abnormal pupillary reaction to light, and convergence-retraction nystagmus) is often present, and isolated Argyll Robertson pupils or diplopia are also common. Precocious puberty is seen occasionally both with pineal (more commonly) and neurohypophyseal tumors. When the tumor is located in the neurohypophyseal region, the most frequent presenting signs and symptoms are endocrinologic and visual. In a large series, 86% of patients presented with diabetes insipidus, 93% with primary or secondary amenorrhea, and 30% with growth retardation (76). Panhypopituitarism and elevations in prolactin may occur, and psychiatric disturbance as the sole presenting symptom has also been described. Chiasmal visual field defects and reduced visual acuity due to infiltration or compression of the anterior visual pathways is common (85% of patients). Diplopia is most often due to fourth or sixth nerve palsies. These symptoms (particularly the endocrinopathies) may precede the diagnosis of tumor by months or years, especially with neurohypophyseal tumors (54; 44). Basal ganglia and thalamic tumors may present with hemiparesis or choreoathetotic movements, personality changes, ocular disturbances, and hemisensory changes in speech (61; 14). Amnesia and hearing loss have also been reported. In rare cases, patients can present with symptoms of myelopathy when the spinal cord is the primary site of involvement. When the tumor spreads in the CSF, a variety of symptoms can occur as seen with other leptomeningeal metastases (60; 43).

Clinical vignette

An 18-year-old man was involved in a motor vehicle accident and sustained a compression fracture in his midthoracic spine. As part of his trauma workup, he underwent a CT of the head, which incidentally showed calcification and enlargement of his pineal gland. His alpha-fetoprotein was 5 ng/mL and his beta HCG was less than 2 IU/L. As he was asymptomatic, the decision was made to follow him with serial imaging. Approximately 8 months later, a head MRI showed enlargement of the mass as well as evidence of early hydrocephalus. He was started on acetazolamide and underwent a third ventriculostomy with endoscopic biopsy of the pineal tumor. Pathology confirmed a germinoma.

The patient enrolled in a clinical trial and received 2 cycles of carboplatin + etoposide and external beam irradiation consisting of 23.4 Gy in 13 fractions to the whole ventricles followed by a boost of 30.6 Gy in 17 fractions to the tumor. He unfortunately suffered a relapse in the spinal region approximately 1 year later. At this point, he was enrolled in another clinical trial and was treated with 4 cycles of cisplatin + etoposide, which resulted in significant hearing loss. Subsequent imaging showed marked improvement in the spinal tumor with no evidence of intracranial recurrence, so he underwent conditioning with carboplatin + thiotepa followed by an autologous peripheral blood stem cell transplant. He then received radiation 20.4 Gy to the whole brain and spine in 12 fractions and a 23.4 Gy spinal boost.

In the posttreatment setting, the patient struggled with nausea, malnutrition, and failure to thrive to the point where he had to be admitted for nasogastric tube feeds. He eventually had a gastrostomy tube placed, which was then converted to a jejunal tube. Other complications included a varicella zoster infection with resulting neuropathic pain and significant depression. All of these issues resolved over time.

At the time of this report, the patient was 7 years out from completing therapy, and he remained disease-free. He continued to have bilateral hearing loss for which he wore hearing aids. With regard to endocrine issues, he had several low testosterone levels that subsequently normalized. Cognitively, he continued to struggle with attention and memory issues but was in the process of finishing college.

Biological basis

Etiology and pathogenesis

The etiology of germinoma is unknown. The striking association between the onset of puberty and the incidence of intracranial germ cell tumors suggests a possible relation between increased neuroendocrine activity in diencephalic structures, and the induction of malignant change in previously dormant primordial germ cell rests (54). A similar relationship is seen with mediastinal germ cell tumors.

During early embryogenesis, primordial germ cells migrate from the fetal yolk sac to the developing gonadal ridge. Whether by chance (“misinvolvement-misenfoldment” hypothesis) (110) or ontologic design (33), primordial germs cells also disseminate widely throughout the developing embryonic tissues, but in most instances are destroyed either through immune mechanisms or apoptosis. It remains unclear why these cells persist in certain extragonadal midline structures, including the pineal gland and neurohypophysis. The pineal gland and hypothalamus or neurohypophysis are themselves linked neuroanatomically for reasons that may relate to orchestration of gonadotrophin release.

Chromosomal abnormalities may also play a role in the development of CNS germ cell tumors (03; 27; 133). Genetic profiles of germ cell tumors overall include imbalances with gains of 12p, 8q, 1p, 2p, 7q, 10q, and X, as well as losses of 11q, 13, 15q, 10q, and 18q (105). An international study of 62 cases of CNS germ cell tumors showed that over 90% of pure germinomas exhibited chromosomal abnormalities with frequent amplification of chromosomes X, 21q, 12p, 1q, and 14q and deletions of 11q, 10q, 17p, and 13q (135). The presence of 12p abnormalities is independent of histology and does not seem to influence prognosis, but the average number of chromosomal abnormalities does correlate with the histologic aggressiveness of the germ cell tumor. Overrepresentation of 12p is hypothesized to act as an initiating event in tumor development through overexpression of cyclin D2. Pineal region germ cell tumors also frequently show an increased copy number of chromosome X or chromosome 21, suggesting that relevant oncogenes reside in both of these locations as well. Interestingly, patients with Klinefelter syndrome (excess copies of the X chromosome) and Down syndrome (trisomy 21) have an increased incidence of CNS germ cell tumors (see Epidemiology section).

In 2 studies, Iwato and colleagues evaluated various molecular changes in intracranial germ cell tumors (51; 52). They found that intracranial germ cell tumors overexpress p53 and MDM2 and that an incidence of MDM2 gene alteration higher than TP53 gene mutation is seen. This finding, also seen in testicular germ cell tumors, suggests a common origin. The researchers also found that pure germinomas have a higher frequency (90%) of INK4a/ARF gene deletions than nongerminomatous germ cell tumors (55%), a finding not seen in testicular germ cell tumors. Germinomas also exhibit increased expression of Bax, which inhibits Bcl-2 (an antiapoptotic protein) and may account for an increased sensitivity to treatment (81). Miyanohara and colleagues (80) found that germinomas express c-kit on their cell surface, a finding not seen in teratomas and syncytiotrophoblastic giant cell germinomas. Co-expression of c-kit and stem cell factor has been reported in germinomas (131). Analysis of c-kit mutational hotspots in germinomas revealed mutations in 4 of 16 cases (108). Asgharzadeh and colleagues, however, found no correlation between expression of CD117 (c-kit) and mutational analysis. They reported strong positive immunostaining for CD117 in 100% of cases whereas their mutational analysis revealed only 9% c-kit mutation rate (exon 17 [N822Y]) (10). In an international study led by Wang, 53% of CNS germ cell tumors exhibited somatic mutations in at least one of the genes involved in the KIT/RAS or AKT/mTOR pathways. CBL, which encodes a RING finger ubiquitin E3 ligase, was the third most frequently mutated gene (135). Another study that examined 124 CNS germ cell tumors observed alterations in the MAPK pathway in 64.3% of germinomas. The PI3K pathway was the second most common target of alteration, with the most frequently mutated gene being MTOR (47).

Epidemiology

Intracranial germ cell tumors comprise about 3% of all childhood brain tumors, with a higher incidence of approximately 11% in Japan and other Asian countries (29). Germ cell tumor incidence has the highest rate in the 10- to 14-year-old patient population (99). Extragonadal germ cell tumors, including the intracranial germ cell tumors, are seen with increased frequency in males with Klinefelter syndrome (08; 102) and in patients with Down syndrome (where the tumors are disproportionately located in the basal ganglia) (85; 112; 23).

Germinomas, like the other intracranial germ cell tumors, are overwhelmingly tumors of the first 3 decades, with a median age at diagnosis of about 12 years; and approximately 90% of cases occur before 20 years (54). Although this age distribution is similar for both males and females, there is a consistent male predominance of about 2.5:1 (54; 106; 44; 117; 76). Interestingly, among suprasellar germinomas, the sex ratio slightly favors females. In contrast, germinomas of the pineal region, basal ganglia, and other locations are rare in females, accounting for the strong male predominance overall.

Prevention

No means of prevention are known.

Differential diagnosis

The visual, and especially the endocrinologic signs and symptoms of neurohypophyseal tumors, may be subtle and may remain unrecognized for months or even years after their onset. In contrast, diplopia and the signs and symptoms of increased intracranial pressure (headache, nausea and vomiting, gait impairment, papilledema), which commonly arise from pineal region tumors, are rarely neglected. Ultimately, both sets of symptoms generally lead to cranial imaging. If a mass is seen in the pineal or suprasellar region, the differential diagnosis includes germinoma and nongerminomatous germ cell tumors in addition to the other lesions that may arise in these regions: glioma, meningioma, primitive neuroectodermal tumor, craniopharyngioma, ependymoma, pineocytoma and pineoblastoma, and metastatic tumors (most commonly breast, lung, kidney and melanoma).

Diagnostic workup

Reliable differentiation of germinomas from nongerminomatous germ cell tumors or other CNS tumors cannot strictly be characterized by imaging alone. Evaluation of the patient with a suspected germinoma includes an MRI of the brain and spinal cord, assays for tumor markers (AFP, beta-HCG) in both serum and CSF, hypothalamic-pituitary hormonal function evaluation, and a large volume CSF cytology. Note that a full endocrine evaluation is also important in pineal tumors to help rule out occult multifocal disease. If possible, the spinal MRI and CSF studies should be performed prior to any surgical intervention. If the presence of obstructive hydrocephalus precludes the safe performance of a lumbar puncture, ventricular CSF sampling following the placement of a ventriculostomy or ventriculoperitoneal shunt, or intraoperatively during an internal shunting procedure, can be performed. Of critical importance for prognosis and treatment decisions is distinguishing pure germinomas from nongerminomatous germ cell tumors. The diagnosis and classification of germ cell tumors can be made on the basis of tumor markers, histology, or both. It is important to note that efforts are underway to determine a tumor marker threshold for beta-HCG-secreting germinomas, as Europe more commonly uses a threshold of 50 mIU/mL compared to 100 mIU/mL in North America. See Table 1 for classification of germ cell tumors according to tumor markers and immunohistochemical markers.

Table 1. Classification of Germ Cell Tumors According to Markers

Beta-HCG= human chorionic gonadotropin; AFP= alpha-fetoprotein; PLAP= placental alkaline phosphatase; += positive; - = negative; +/- = equivocal

Adapted from (29).

Unless there are characteristic increased tumor markers found in the serum, or CSF, or both, a diagnosis of germ cell tumors often requires a biopsy. When the tumor markers are negative or only mildly elevated, or if there is any noncharacteristic finding, a tumor biopsy for histologic diagnosis should be performed. Surgery can be avoided for patients with compelling clinical and radiographic evidence of nongerminomatous germ cell tumors. Aggressive removal of tumors prior to chemotherapy or radiotherapy is generally not recommended or required due to the effectiveness of therapy as well as concern for increasing endocrine, cognitive, and neurologic toxicity for tumors in the regions most commonly affected by germinoma.

Radiological findings characteristic of the various CNS germ cell tumors have been reported (34; 128; 69).

MRI images of germinomas are isointense to slightly lower intensity than grey matter, especially solid components, on T1-weighted images and are isointense or (more commonly) hyperintense on T2-weighted images. The high intensity components on T1 correspond to hemorrhage, fat, or high protein and may assist in differentiating germinomas from nongerminomas (69). Earlier reports suggested that germinomas showed homogenous contrast enhancement, but heterogenous enhancement is seen in more than half of germinomas (69). Large suprasellar tumors with a “dural tail” may be suggestive of a neurohypophyseal germ cell tumor (69). In most cases, there is little, if any, peritumoral edema. Tumors show low apparent diffusion coefficient (ADC) values, which appear to normalize following treatment; robust normalization of low ADC may be predictive of improved response to chemotherapy (92). Calcifications are seen in slightly less than half of cases by CT scanning. Intratumoral cysts are identified by MRI in about half of patients with germinomas (128; 83; 69). MRI can also demonstrate teratomatous tissue components (especially fat) suggestive of mixed germ cell tumors (34). A few case reports have noted cerebral hemiatrophy by MRI as a presenting feature in cases where germinomas were localized to the basal ganglia, thalamus, or brainstem. (61; 94).

Multifocal intracranial disease is seen in 18% of patients with germinomas in modern series and is important to document for radiotherapeutic target delineation. Sites of involvement include the pineal and suprasellar region, hypothalamus, optic chiasm, cavernous sinus, third ventricle wall, and anterior half of the lateral ventricle wall. Some cases are radiographically occult at diagnosis but become more evident after chemotherapy response. Hypophyseal size and pineal “cyst” sizes are important to follow after induction chemotherapy as decrease in size of these initially normal-appearing structures suggests occult involvement. Disseminated disease is diagnosed in cases with a positive CSF cytology or cases in which MRI scanning reveals metastases within the spinal canal. This later situation occurs in about 15% of patients and effects treatment planning and disease monitoring.

The distinctive biochemical profiles of these tumors can be exploited diagnostically by evaluation of CSF and serum. AFP and beta-HCG remain the most useful tumor markers in CNS germ cell tumors, although other markers, such as placental alkaline phosphatase (PLAP) and c-kit, are being investigated. Pure germinomas frequently display increased levels of placental alkaline phosphatase in the CSF (50; 124; 96). Alpha-fetoprotein is absent such that any detectable alpha-fetoprotein rules out the diagnosis. Markedly elevated levels of alpha-fetoprotein (greater than 700 ng/ml) are diagnostic of yolk sac tumors, and more modest elevations are seen in embryonal carcinomas, immature teratomas, and mixed germ cell tumors containing a predominance of these histologic types. Modest elevations of human chorionic gonadotrophin (usually less than 100 mIU/ml) and beta-human chorionic gonadotrophin (usually less than 50 mlU/ml) may be seen in the serum, CSF, or both in about one quarter of patients with pure germinomas (04; 13; 76; 114; 18; 48) and have been attributed to the presence of syncytiotrophoblastic giant cells within the tumor (123; 48), though serum elevations greater than 500 mIU/L and CSF elevation greater than 2000 mIU/L have been reported in a child with a recurrent, disseminated, pathologically confirmed pure germinoma (31). Markedly increased levels of human chorionic gonadotrophin (greater than 2000 mIU/ml) are seen in choriocarcinomas. Mature teratomas demonstrate no marker elevations. These markers are useful both as diagnostic tools and as surrogates of disease progression and response to therapy.

Implementation of appropriate treatment modalities and ultimate prognostication are dependent on the accurate histological classification of CNS germ cell tumors. Macroscopically, germinomas may appear ill-defined due to a propensity to infiltrate neighboring brain structures. On cut surface, the tumors are generally pink-gray and solid, although these may occasionally display cystic change. Microscopically, intracranial germ cell tumors display similar histologic features and immunohistochemical staining properties as testicular seminomas and ovarian dysgerminomas. The architecture of the tumor typically shows a lobular arrangement with fibrovascular septa dividing aggregates of tumor cells. The tumor cells generally appear uniform, with large round nuclei, vesicular chromatin, and single prominent nucleoli. The presence of glycogen renders the cytoplasm clear, which can be demonstrated by strong positive Periodic Acid Schiff (PAS) staining, which is sensitive to diastase digestion. Mitoses are frequently present, and necrosis is occasionally encountered. The T-cell rich lymphocytic infiltrates are typically intermingled to variable degrees, often associated with the fibrovascular septa, and may be so prominent as to mask the underlying tumor cells. Both neoplastic cells and reactive astrocytes can be seen in intraoperative smear preparations. Multinucleated syncytiotrophoblast-like giant cells may also be present and may lead to consideration of a granulomatous inflammatory process such as sarcoidosis or tuberculosis. Immunohistochemistry for placental alkaline phosphatase and c-kit produce cytoplasmic and membrane staining in germinomas (62).

OCT3/4, a transcription factor necessary for maintaining pluripotency in embryonic stem cells and primordial germ cells and a member of the POU class of homeobox genes, has emerged as a robust nuclear marker for gonadal and extragonadal (eg, CNS) germ cell tumors, specifically germinomas and embryonal carcinomas, and may be useful in establishing a germ cell origin for tumors that otherwise may remain unidentified (42; 71; 21; 88). Additionally, podoplanin, a transmembrane sialoglycoprotein, was found to be a sensitive marker for CNS germinomas and absent in nongerminomatous germ cell tumors except for some immature teratomas (79). NANOG, a homeodomain transcription factor, has been shown to be expressed in more than 90% of germinoma cells, even in cases of inconsistent expression of OCT3/4 and placental alkaline phosphatase, any may help tumor identification in small biopsy specimens (111). Additionally, immunohistochemistry is employed in the evaluation of intracranial germ cell tumors for additional elements such as embryonal carcinoma (CD30 positive), yolk sac tumor (alpha-fetal protein positive), choriocarcinoma (beta-human chorionic gonadotropin and human placental lactogen positive), and teratoma. SALL4 has been identified as a sensitive and specific immunohistochemical marker for primary CNS germ cell tumors (77).

Because germ cell tumors arising outside the CNS may metastasize to brain, some authors recommend whole-body [18F]-FDG PET to rule out a primary extracranial germ cell tumor (45).

Management

With success rates approaching 100% for curing CNS germinomas with radiation, it has been long been considered standard treatment; however, the required dose and tumor volume remain controversial. Today, standard treatment of localized germinomas typically consists of whole ventricular field irradiation plus a boost to the primary tumor with or without neoadjuvant chemotherapy. In patients with metastatic disease, craniospinal irradiation is indicated. The most pressing therapeutic challenge today is to reduce the long-term side effects of treatment without compromising disease-specific survival. CNS irradiation is associated with frequent and significant late consequences, and this observation provides the rationale for innovative combined modality approaches to therapy employing surgery, radiation, and chemotherapy. Nevertheless, treatment innovations that reduce the likelihood of cure below that seen with conventional therapy are not acceptable.

Surgery. The role of surgery in the treatment of CNS germinomas has evolved over time. In the past, because of the high morbidity and mortality associated with pineal surgery, a nonhistologic diagnosis, based on the exquisite radiosensitivity of germinomas, was frequently accepted. In patients with suggestive epidemiologic and radiographic characteristics, a “test dose” of cranial irradiation of 2000 cGy was often administered. If complete or near-complete tumor shrinkage was documented, the lesion was presumed to be a germinoma, and the full course of radiation therapy was then completed (44). Because of the variety of tumors that can present in suprasellar or pineal locations, many with variable radio- and chemosensitivities, radiographic and biochemical characteristics, and natural histories, this approach is no longer considered acceptable. In experienced hands, stereotactic, endoscopic, and craniotomy procedures for biopsy entail minimal risk (1% to 5% morbidity and mortality) (103; 13; 76; 115; 12; 138). For pure germinomas, for which curative nonsurgical therapy is available, open, stereotactic, or endoscopic biopsy is sufficient. Compared with biopsy alone, long-term follow up suggests that patients undergoing more extensive surgical resection may experience worse cognitive function and quality of life without improving tumor outcome (75). If CSF diversion is necessary, endoscopic “internal shunting” (third ventriculostomy) should be considered, obviating the need for ventriculoperitoneal shunting, which itself may lead to extraneural dissemination of tumor (54; 101). This approach also permits the identification of surgically curable but radio-resistant tumors such as mature teratomas and pineocytomas, which can then be fully resected, sparing patients additional toxic therapy. Metastasis, along the route of endoscopy, has been reported (132).

A second-look surgery may be considered in patients with partial response or stable disease with greater than 1.5 cm residual tumor (32; 55; 19). Mature teratomas and pineocytomas are radio-resistant and should be ruled out because they can be cured surgically. A second-look surgery would also help identify occult mixed germ cell tumors, which may require more intense therapy than pure germinomas. With regard to safety, it has been demonstrated that the risk of complications from a second-look surgery are significantly less due to decreased vascularity after induction chemotherapy (17).

Radiation. Modifications are occurring in radiotherapy treatment strategies for the management of localized CNS germinoma. Historically, radiotherapy doses have included 4000 to 5500 cGy to the involved field, and 3000 to 3600 cGy of craniospinal irradiation. Over the last decade, due to the attainment of high survival and cure rates as well as the desire to reduce long-term side effects resulting from radiotherapy, a trend toward reducing field size and total dose of radiation has emerged. Cranial irradiation fields have been narrowed to include the local tumor plus the ventricular system generally to 24 Gy based on success for microscopic disease using this dose for CSI without chemotherapy (19), and doses to primary tumor were initially reduced to 4000 cGy to 4500 cGy, dependent, in some institutions, on tumor size (122; 40; 125). One study showed a 92% 5-year survival rate and 88.8% relapse-free survival rate using craniospinal irradiation (3000 cGy in 150 cGy fractions) with a 1500 cGy boost to the tumor volume (12). Ogawa and colleagues studied the effects of further localization of radiotherapy by showing that selective radiotherapy alone in 70 patients with intracranial germinomas led to 7 relapses, 3 of which occurred in 6 patients who received focal irradiation that involved the primary tumor plus a margin but did not include the ventricles (90). In another study of 35 patients with localized germinomas, none of the 21 patients who received whole-ventricle irradiation experienced a relapse whereas 5 of the 14 patients who received focal irradiation developed recurrent tumors within the ventricular system (38). As a result, whole-ventricle irradiation is considered the current standard of care for radiation-only treatment of localized germinoma. Shibamoto’s group proposes further dose reduction and substratification based on tumor size. Utilizing 36 Gy after gross total resection, 40 Gy for tumors smaller than 2.5 cm, 45 Gy for tumors 2.5 to 4.0 cm, and 40 Gy for larger tumors, they report a 10-year, relapse-free survival rate of 95%, equivalent to historical rates cited for uniform use of 50 Gy. Ogawa and colleagues conclude that 40 to 45 Gy is adequate to cure germinomas 4 cm or smaller (120).

One concern about omitting the craniospinal field of radiation is the perceived risk of cerebrospinal fluid relapse in patients with initially localized disease. In older series (where pre-treatment staging procedures were often incomplete), relapse in the CSF was seen in 23% of cases versus 8% when prophylactic craniospinal irradiation was used (70; 117). Other studies, however, suggest an incidence of CSF relapse of approximately 7.6% (11/145), somewhat greater in patients receiving only involved field radiation (6/33, 18%), but less in patients treated to the whole brain or ventricular system (2/56, 3.8%) and in those who received only involved field radiation with pre-radiation chemotherapy (3/56, 5.4%) (70; 25; 127; 13; 39; 07; 115; 18). A comprehensive review of published cases since 1988 evaluated disease relapse and cure rates between craniospinal radiotherapy and reduced volume irradiation alone (whole brain or whole ventricular followed by boost) in the setting of localized intracranial germinomas (104). Recurrence rates were found to be 7.6% after reduced volume radiotherapy and 3.6% after craniospinal radiotherapy with isolated spinal relapses of 2.9% versus 1.2% respectively. Based on these data, the authors argue that replacing craniospinal radiotherapy with reduced volume whole-brain or whole-ventricular radiotherapy plus boost is warranted for the management of localized disease and that such a modification can have significant impact on the late toxic effects that arise due to irradiation (104). Shibamoto reviewed data from 2 earlier studies published by his group and reported no significant difference in 10-year survival rates between patients receiving craniospinal radiation and focal radiation therapy, 95% and 88% respectively (120). Based on these results, some authors recommend reserving craniospinal irradiation for positive CSF cytology or clinical or radiographic evidence of CSF dissemination (120). However, reduced volume radiotherapy remains controversial, and a retrospective review of 21 patients treated with either focal irradiation plus chemotherapy or craniospinal radiotherapy suggested higher rates of failure in the focal radiation group (89).

Certain subgroups of CNS germinoma are at higher risk for spinal dissemination and, therefore, may benefit from craniospinal irradiation. Ogawa and colleagues showed that large intracranial disease (larger than 4 cm) is associated a higher rate of spinal recurrence whereas smaller tumors (smaller than 4 cm) are associated with a better prognosis (91). Ogawa and colleagues also showed that multifocal intracranial disease is an independent risk factor for spinal recurrence. In a retrospective analysis of 170 patients with intracranial germinomas, the spinal cord metastasis rate was 3.4% in patients with localized disease who did not undergo spinal cord irradiation compared to 16.7% in patients with bifocal disease who were treated with whole-ventricular or whole-brain irradiation (68).

In nonlocalized CNS germinoma, when CSF cytology or neuraxis MRI scanning reveals disseminated disease at diagnosis, craniospinal radiation is used. The optimal dose of craniospinal irradiation remains controversial. Long-term results of SIOP96 support 24 Gy CSI +16 Gy boost to primary site (with no benefit to additional chemotherapy) as event-free survival was 100% and overall survival was 98% (19). Similar results have been reported from other groups using doses of 2000 to 2400 cGy (119; 120; 55). In addition, advances in radiotherapy techniques such as proton beam irradiation, which eliminate exit dose associated with radiation beams, are expected to significantly reduce toxicities associated with both localized and nonlocalized germinoma treatment (74). Early reports suggest similar event-free and overall survival rates with significant reductions in dose to temporal lobes and eliminating radiation exposure of the organs and structures anterior to the spine. Lastly, radiosurgery has been used to treat intracranial germinomas, but the role of this type of radiotherapy is not clear (141; 41; 30; 84; 139).

Chemotherapy. A number of multi-institutional and multinational collaborative groups have studied the possibility of using postoperative systemic chemotherapy for patients with CNS germinomas, primarily in the pediatric setting, in order to further reduce the CNS radiation dose and treatment volume. CNS germinomas, like their extraneural counterparts, are extremely sensitive to multiple chemotherapeutic agents and regimens, including single-agent cisplatin; carboplatin; ifosfamide; cyclophosphamide; and etoposide; as well as combination regimens including cisplatin or carboplatin plus etoposide; carboplatin, etoposide, and cyclophosphamide or ifosfamide; and carboplatin, etoposide, and bleomycin. Radiographic, cytologic, and biochemical complete responses to a variety of chemotherapeutic regimens are seen in 74% of patients (range: 47% to 100%), including those with leptomeningeal dissemination (05; 04; 140; 20; 11; 13; 116; 16; 18; 126) (Tables 1 and 2). A clinical trial through Children’s Oncology Group (COG-ACNS1123, NCT01602666) is currently assessing neoadjuvant chemotherapy (4 cycles of carboplatin and etoposide) followed by response-based radiation therapy for localized germinomas. For patients with complete response or residual disease of less than 1.5cm, patients undergo 18Gy whole-ventricle irradiation plus 12Gy boost to the primary site; it is too early to assess the outcome of this study as the germinoma stratum is closed to accrual and patient follow-up is ongoing.

Subsequent studies and long-term follow-up have shown, however, that most relapses with focal irradiation tend to occur in the ventricular system and that whole-ventricle irradiation may help prevent these relapses. In their final update of patients with localized CNS germinomas treated in the SIOP CNS GCT 96 trial, Calaminus and colleagues reported that of their 65 patients who received chemotherapy followed by focal irradiation, 7 relapses were noted, and 6 of these were ventricular recurrences outside the primary radiotherapy field (19). Of note, although there was no difference in 5-year event-free or overall survival between patients treated with craniospinal irradiation alone versus combined therapy, there was a difference in progression-free survival (0.97 +/- 0.02 vs 0.88 +/- 0.04; P=0.04). In addition, late relapses are seen after chemoradiation therapy and long-term follow-up. A series of patients treated with chemotherapy plus focal radiation therapy demonstrates event-free survival of 88% at 5 years that declined to 80% at 10 years (55). A review of 10 relapsed CNS germinomas from a group of 60 patients initially treated with neoadjuvant chemotherapy followed by focal irradiation showed that 9 patients had local or locoregional recurrence, 8 of which were periventricular (01). This is consistent with other similar reports noting the pattern of marginal relapses suggested that chemotherapy was not able to sterilize the ventricular CSF (28; 55). These findings suggest that expanding the radiation field to include the ventricles may help prevent the majority of relapses in patients with localized germinoma treated with chemotherapy followed by radiation.

The role of chemotherapy in disseminated patients is controversial. Historically, chemotherapy has been used with the rationale to allow radiation-dose reduction. Historic craniospinal doses were 36 Gy, and many modern chemoradiotherapy trials demonstrated doses can be lowered to 20 to 24 Gy in good responders without compromising outcomes. Aoyama and colleagues (06) published a promising report in which patients with pure germinomas treated with a combination of etoposide and cisplatin followed by a radiation dose of only 2400 cGy. They demonstrated that the actuarial relapse free survival at 5 years was 90%, and the overall survival rate at 5 years was 100%. Patients with B-HCG secreting tumors had rates of 75% and 44% (chemotherapy used-ifosfamide, cisplatin, and etoposide); hence, it was felt that they should be treated with higher doses of radiation therapy than 2400 cGy. The observation that elevated B-HCG levels may indicate treatment-resistant disease is supported by data from Inoue and colleagues (49). In their study of 21 patients with CNS germinomas treated with chemoradiation therapy, 19 had a complete response, and 17 of these patients did not have elevated B-HCG levels in their CSF or tumor tissue. The 2 patients who were refractory to treatment both had elevated B-HCG in their CSF and tumor tissue. Kortmann and colleagues additionally reviewed in abstract form the 24 Gy CSI + 16 Gy boost arm of SIOP GCT 96 to the 30 Gy CSI + 15 Gy boost employed in MAKEI 89. The 5-year event-free survival was 90% in MAKEI 89 and 91% in SIOP 96, again supporting the efficacy of reduced-dose craniospinal radiation therapy (63).

Kretschmar and colleagues reported on the results of a Phase II study employing chemotherapy and response-based radiotherapy in children with CNS germ cell tumors (64). Subjects were stratified as either low-risk (ie, pure germinoma with normal alpha-fetoprotein and HCG less than 50 mIL/ml in serum and CSF) or high-risk (ie, mixed germinoma or other malignant variant or elevated HCG or alpha-fetoprotein). Subjects were treated with alternating courses of chemotherapy etoposide plus cisplatin alternating with cyclophosphamide plus vincristine for 4 courses. High-risk subjects received double doses of cisplatin and cyclophosphamide. Radiotherapy was initiated after chemotherapy with the delivered dose determined by risk group, dissemination, and response status. Kretschmar and colleagues documented 91% objective radiographic response in their subjects with germinomas, and all germinoma patients were alive at a median 66 months after radiation therapy (64).

However, the long-term results of a prospective multinational trial (SIOP96) call the role of chemotherapy into question again (19). For patients with metastatic tumors (n = 45), progression-free survival was estimated to be 1.00. Event-free survival at 5 years was 0.98+0.02 among all patients, of whom 28 received craniospinal radiotherapy alone and 17 had received additional chemotherapy before craniospinal radiotherapy. This indicates that there is no difference in outcomes with chemotherapy. Overall survival at 5 years was 0.98+0.02 among all patients. Although it appears that the dose of radiation can be reduced safely without the addition of chemotherapy, the decision to utilize chemotherapy was not randomized, and it is certainly possible that patients with bulky disease may have been more likely receive neoadjuvant chemotherapy.

Studies utilizing neoadjuvant, inductive chemotherapy followed by dose- and volume-reduced irradiation appear to yield comparable cure rates to traditional full-dose, large volume radiation; however, longer follow-up will be required to determine if these newer regimens avoid the neurocognitive sequelae of radiation and improve the quality of life of long-term survivors. The primary goals of this approach are to reduce cognitive and endocrine toxicity after radiation therapy. The current recommended radiation dose without chemotherapy is 20 to 24 Gy for either CSI or WVFI. Both modalities are associated with less than 10% risk of long-term endocrine dysfunction with the exception of growth hormone (78), so this is primarily a risk for skeletally immature children. Chemotherapy allows for reduction of the primary tumor boost dose from 40 to 45 Gy down to 30 to 36 Gy for good responders, which is likely meaningful for suprasellar tumors, especially in young children for whom vascular or cognitive issues are well-defined and are generally dose-dependent. Emerging data regarding hippocampal radiotherapy dose and cognitive outcomes would suggest that reducing the dose to the medial temporal lobes (as would be expected in suprasellar tumors) may also reduce short-term memory difficulties in adults (36; 72). In the pineal region there would be no impact on endocrine risks, and it is likely there would be less benefit from a cognitive and neurovascular toxicity standpoint as less of the frontal or temporal lobes would be in the boost field. Alapetite and colleagues addressed this concern in his report of results of SFOP-90 in which 60 nondisseminated, histologically proven germinomas were treated with 4 alternating courses of etoposide-carboplatin and etoposide-ifosfamide followed by 49 Gy to the tumor bed and followed for a median of 76 months (range 7 to 136 months) (02). The 5- and 8-year survival was 98%. Twenty percent of subjects had a neuropsychological evaluation that showed good preservation. A study of 15 patients by Tseng and colleagues utilized chemotherapy and low-dose radiation and found that most patients had favorable outcomes over a median follow-up period of 45 months (134). Cheng and colleagues showed that of the 8 children in their study with CNS germinomas who were treated with chemotherapy and whole-ventricle irradiation, all exhibited stable full-scale IQ, albeit with a decline in processing speed, through a median follow-up period of 61 months (22).

One trial was designed to test the feasibility of entirely avoiding the use of radiation therapy, in favor of an intensive chemotherapy regimen consisting of high-dose carboplatin, etoposide, and bleomycin, with the addition of high-dose cyclophosphamide in patients without an initial complete response (11). Of 71 patients (45 with germinomas and 26 with nongerminomatous germ cell tumors) enrolled in this trial, 68 were evaluated and 39 (57%) demonstrated a complete response to initial chemotherapy. Of the remaining 29 patients, 16 patients became complete responders following re-operation or cyclophosphamide-intensified chemotherapy, for an overall complete response rate of 77% (55 patients out of 71 patients), including 38 patients of 45 patients with germinomas (84%). Histology did not influence the complete response rate but did impact survival. Unfortunately, 28 of the 55 complete responders (51%), and 26 of the 45 (58%) with germinomas have relapsed, with a median follow-up time of 18 months. The 2-year survival probability for patients with germinomas was only 84%. Chemotherapy-related morbidity and mortality were also substantial (7 patients died from chemotherapy-related complications). The Third International CNS Germ Tumor Study stratified subjects into low risk (n=11), intermediate risk (n=2), and high risk (n=12) based on serum/CSF levels of HCGbeta, alpha-fetoprotein, and the histologic presence of nongerminomatous malignant elements (24). This study also utilized a chemotherapy-only strategy, with low-risk subjects treated with 4 to 6 alternating cycles of carboplatin/etoposide and cyclophosphamide/etoposide. The intermediate- and high-risk groups received 4 to 6 cycles of carboplatin/cyclophosphamide/etoposide. This approach yielded a response rate of 76% after 4 cycles of chemotherapy and a 6-year event-free and overall survival of 46% and 75%, respectively. The authors concluded that this approach yielded results that were inferior to regimens utilizing some form of radiotherapy.

The results of the Second International CNS Germ Cell Tumor Study Group Protocol (the successor to the Balmaceda study) (59) largely mirror the findings of the original study. Patients received 2 cycles of cisplatin, etoposide, cyclophosphamide and bleomycin, followed (for complete responders) by 2 cycles of carboplatin, etoposide, and bleomycin, followed (for those whose complete responses came only after 4 courses of chemotherapy) by an additional cycle of each regimen. Twenty patients with germinomas were enrolled, and 16 out of 17 evaluable patients had a partial or complete response after the first 2 cycles of treatment. At a median follow-up of 6.3 years, 14 out of 20 patients were disease-free; 8 of these patients had no relapse or progression whereas the other 6 are in second or third remission after additional chemotherapy and radiation therapy. A second paper by this group on the same regimen suggests that this regimen is effective in achieving remissions but this approach had a high morbidity and mortality suggesting this approach may not be prudent (58). In both International Studies, relapses occurred predominantly at distant neuraxis sites (including the CSF). Also, in both studies, patients with diabetes insipidus tolerated chemotherapy less well than their unaffected counterparts.

Modak and colleagues evaluated the use of high dose chemotherapy with autologous stem cell rescue in patients with relapsing or progressive disease (82). They gave a high dose escalation regimen of various thiotepa-based regimens to 9 patients with germinomas and 12 patients with nongerminomatous germ cell tumors. This was followed by autologous stem cell rescue. Of the 9 patients with relapsed or progressive germinomas, 7 had a median survival of 48 months whereas only 4 of the patients with nongerminomatous germ cell tumors survived with a median of 33 months. All patients developed grade 4 hematologic toxicity, but no patients died from toxic causes. Kubota and colleagues analyzed the data from 6 patients with recurrent CNS germinoma (65). All patients received between 1 and 7 cycles of platinum-based chemotherapy and achieved a complete response prior to undergoing high-dose chemotherapy that consisted of 1 to 3 cycles of carboplatin, etoposide, and melphalan. This was followed by autologous stem cell transplant and resulted in 5 of 6 patients being alive and disease-free with a median survival of 65 months. These results warrant further investigation to see if high-dose chemotherapy followed by autologous stem cell transplant is a reasonable approach for patients with relapsing and progressive germinomas.

In summary, despite remarkable initial complete response rates, chemotherapy-alone treatment regimens have not translated into acceptable long-term survival rates for patients with germinomas, and the toxicity associated with aggressive regimes is substantial. High-dose chemotherapy without radiation cannot currently be considered a treatment option for patients outside of the protocol setting, but can be an effective salvage regimen with or without salvage radiation.

Summary of chemoradiation therapy. Optimal chemoradiation therapy in patients with CNS germinomas entails individual risk assessment and consultation with both radiotherapy and chemotherapy to determine optimal treatment based on tumor risk factors, age, and location of tumor. Pediatric patients most commonly receive neoadjuvant chemotherapy with an active regimen, followed by whole-ventricle irradiation and radiation tailored to the pretreatment extent of disease and response to chemotherapy. Patients with localized disease are safely treated with 2400 cGy WVFI with boost to the primary tumors at doses around 3000 to 3600 cGy for complete chemotherapy responders and somewhat higher doses (4000 to 5400 cGy) for less than complete responders. For patients treated without chemotherapy (generally low-risk young adults) 2400 cGY WVFI with boost to 4000 tp 4500 cGy to the primary tumor can be administered. Prophylactic craniospinal irradiation can be omitted in most patients with individualized risk assessment for patients with bifocal or multifocal disease. If CSF dissemination has been documented prior to treatment, craniospinal irradiation with or without neoadjuvant chemotherapy should be considered based on patient age, bulk of disease, and clinical risk factors. Doses of about 2000 to 2400 cGy for CSI appear adequate with or without chemotherapy. For boost volumes (both primary site as well as metastatic sites) response-based tumor boost volumes and dose are the same as localized guidelines. These guidelines should permit decreases in the dose and extent of nervous system irradiation and will hopefully translate into a reduction in the late consequences of CNS irradiation. Nevertheless, substantial additional research is necessary to validate the preliminary results on which these suggestions are based to determine what chemotherapeutic regimens are most active, to evaluate novel markers of disease responsiveness, and to see whether further reductions in CNS irradiation are possible.

Importantly, patients whose germinomas recur after chemotherapy and reduced-dose radiation may still be cured with additional chemotherapy and radiotherapy (140; 20; 11; 13; 76; 78; 116; 16; 18). The true recurrence rate of germinomas after initial therapy may be higher than traditionally reported. In one study with a median follow-up of 134 months, the median time to recurrence was 50 months, with 36% of subjects having recurrence at 60 months or later, and the latest occurrence occurred at 230 months (57). Platinum-based chemotherapy followed by wide-field, low-dose radiation, 18 to 25.2 Gy, has shown some efficacy (57). Moreover, myeloablative chemotherapy regimens with stem cell transplant have demonstrated some benefit in recurrent disease (56; 37). In addition, studies have emphasized the importance of second-look surgery for patients treated with chemotherapy and radiation in which CSF and serum markers, if any, have returned to normal, but follow-up MRI scans continue to show minimal residual enhancing tissue. Usually in these studies, necrosis, fibrosis, or mature teratoma has been identified (11; 18). In any case, further chemotherapy and radiotherapy are not indicated, and in the latter case, curative gross total resection can be undertaken.

A more novel approach to treating patients with recurrent GCT is the use of dendritic cells, in which a patient had a radiographic response and a decrease in serum beta-HCG before succumbing to septic shock from a urinary tract infection (97). Future approaches may consider the role of the c-kit proto-oncogene in the pathogenesis of CNS germinomas and will need to evaluate the use of molecular therapies, such as ST1571, that target c-kit (108; 136).

Dasatinib has been shown to be feasible in children with central nervous system germinoma as it is a potent inhibitor of c-KIT (98). Anti-PD-1/PD-L1 agents may also represent a potential therapeutic modality given that the expression of PD-1 in tumor infiltrating lymphocytes and PD-L1 in CNS germinoma tumors were 96% and 92% respectively in a study of 25 tumor samples by Liu and colleagues (67).

Table 2. Chemo-radiotherapy Trials in Patients with CNS Germinomas

IF: involved field radiation; CS: craniospinal irradiation
* 57-month median (range 10-76); no patient has yet recurred
** CS irradiation for patients who had documented CSF dissemination at diagnosis

Outcomes

The prognosis for patients with CNS germinoma has improved dramatically over the last 3 decades with refinements in neurosurgical, neuroradiologic, and oncologic techniques. Five-year and 10-year survival rates in most studies have been 90% or greater for patients with germinomas treated with conventional radiotherapy regimens (70; 26; 44; 20; 122; 137; 46; 13; 40; 76; 125; 115a; 12). Survival rates for patients treated prior to 1980 are difficult to interpret because surgical verification of tumor histology was often not undertaken and tumor staging was sometimes incomplete. Earlier series, therefore, probably include tumors of various histology, extent, and prognosis. In addition to the presence of nongerminoma histologic features, some studies have suggested that elevated serum or CSF HCG levels (11; 115a; 48), the presence of syncytiotrophoblastic giant cells (140; 76; 123; 115a), or the presence of a cystic component on pre-treatment MRI (83), are poor prognostic factors in patients with germinomas. Germinomas with HCG-producing syncytiotrophoblastic giant cells (STGCs) have been implicated as poor prognostic indicators in some early studies. However, Fujimaki and colleagues reported a study of 131 germinomas and 39 STGC- or HCG-producing germinomas (HCG 2 to 200 mIU/ml). After induction chemotherapy, the relapse-free rate for germinoma was 90.1% and 84.6% for HCG germinoma, and the 5-year survival rates were 98.3% and 100% respectively. There was no clinically significant difference in outcome between germinomas and HCG germinomas (35).

Despite the excellent survival of patients with CNS germinomas, tumor-related deficits frequently do not fully resolve (40; 93). Residual endocrine abnormalities in patients with neurohypophyseal tumors are extremely common, as are persistent eye movement deficits and visual field disturbances. Preoperative hormonal assessment can aide in determining the postoperative need for hormone replacement in patients with suprasellar germinomas (86). Of even greater concern are the long-term consequences of chemotherapy and CNS irradiation. Both adults and children are at risk, but late effects are more common, and more pronounced at both ends of the age spectrum. Chemotherapy alone can produce lasting deficits in physical, psychosocial, and intellectual functioning, all of which are more prominent when cranial irradiation is added to the treatment approach (109). Even low doses of irradiation can produce significant late effects (53; 93). In children and adolescents, hypothalamic and pituitary dysfunction, hearing loss, growth retardation, and progressive cognitive dysfunction are especially common. One study of neurocognitive function showed significant decline in working memory, processing speed, and visual memory over time (73). O’Neil and colleagues report that an approach utilizing chemotherapy followed by reduced-dose irradiation may lead to better neurocognitive outcomes in children and adolescents (95). If craniospinal irradiation is used, bone marrow suppression, gonadal and thyroid hypofunction, and growth deficiencies become additional concerns (115; 16; 118). Infertility and second malignancies are also a concern in long-term survivors who receive alkylating agent or topoisomerase II inhibitor-containing chemotherapy regimens. The exact incidence of these complications is difficult to determine, and published studies vary widely in reported complication rates, based partly on differing definitions and durations of follow-up, but reduced educational and occupational attainment, the need for special education, and compromised social and personal functioning have been reported in 30% to 100% of long-term survivors. Data from Sutton and colleagues showed that the long-term outcome of 22 patients with marker negative suprasellar germinomas treated with neuro-axis RT had no significant long-term effects, except short stature in females and the need for hormone replacement (129). There was no correlation with age at the time of radiation treatment and any adverse effects.

Another complication of intracranial germinomas is their ability to spread via ventriculo-peritoneal shunts. This suggests that abdominal imaging may be required for surveillance purposes. Dissemination into the abdominal cavity may occur months to years after shunt placement and necessitates abdominal therapies such as radiation or chemotherapy.

Special considerations

Pregnancy

Following cranial radiotherapy for CNS germinoma, acquired reproductive dysfunction in the form of hypogonadotropic hypogonadism can arise. Menstrual irregularities can result in up to 70% of cases. Because the functional capacity of the ovaries is not altered by CNS radiotherapy, the recreation of normal reproductive cycles and the retention of reproductive potential can be can be achieved through the administration of exogenous gonadotropins (113).