At present, the clinical diagnosis of intracranial GCTs relies on a combination of imaging characteristics and the presence of tumor markers, namely alpha-fetoprotein (AFP) and beta subunit of human chorionic gonadotropin (β-HCG), in the serum and/or cerebrospinal fluid (CSF). For cases where tumor markers are negative, surgical biopsy is recommended for histopathological confirmation (5). In addition to the characteristic morphological appearance on histology, common immunohistochemical (IHC) analysis used for the diagnostic work-up for GCT include CD117/KIT (germinoma), POU5F1/OCT4 (germinoma), Placental alkaline phosphate (PLAP) (germinoma), AFP (yolk sac tumor), CD30 (embryonal carcinoma), and HCG (choriocarcinoma or syncytiotrophoblast in germinoma).

Although these measures have been the standard of diagnostics for decades, they are imperfect. For instance, conventional tumor markers have low sensitivity and specificity, with some studies reporting only one-third of patients with intracranial GCT being tumor marker positive (6, 7). This low frequency is in part related to the predominance of germinomas within intracranial GCTs, where the majority of germinomas do not secrete tumor markers. For the minority of germinomas that do secrete β-HCG, they generally have low-level marker elevation and is likely related to the presence of syncytiotrophoblastic elements. Importantly, while β-HCG secreting germinomas is a widely accepted entity, there remains a lack of consensus on the cut-off level of β-HCG for the diagnosis of germinoma versus NGGCT.

For example, in Children’s Oncology Group (COG) trials, β-HCG cut-offs of up to ≤ 100 IU/L have been used to indicate pure germinoma histology, however, European SIOP trials have used a more conservative cut-off of ≤ 50 IU/L (8, 9). In Japan, a histopathologic-based diagnosis is generally preferred and used in their clinical trials, except for extreme instances such as β-HCG levels of >2,000 IU/L, which would indicate NGGCT, such as choriocarcinoma (10). Similarly, there are different consensus cut-off for AFP levels. AFP > 10ng/ml (or > upper limit of normal) is used in COG trials, whereas in European trials a level >25 ng/ml has been used as an indicator of a NGGCT (8, 9). Like germinomas, teratomas are often tumor marker negative; while this holds true for pure mature teratomas (MT), immature teratomas (IT) may secrete AFP. The AFP level that indicates an IT has not been well established, with examples of extracranial pure ITs having mean AFP levels of approximately 30-80 ng/ml (11).

Importantly, even in instances where histopathologic diagnosis is obtained through tissue biopsy, sampling error remains a significant concern. This is particularly challenging, especially with the known predilection of these tumors to have mixed histology. For instance, a marker negative tumor is often a mixed tumor containing numerous distinct components. However, a biopsy may capture only the germinoma component, leading to inadequate treatment. This is of critical clinical significance, as the treatment regimens for germinoma and NGGCT vary significantly, and with differing survival outcomes. Finally, other non-Intracranial GCT entities can mimic marker negative intracranial GCTs, such as Langerhans Cell Histiocytosis (LCH) and lymphocytic hypophysitis. Given these imperfect means of diagnosing these tumors, efforts to enhance accuracy of diagnosis, identify potential biomarkers that are predictive and prognostic, are imperative.

Despite considerable variation in treatment regimens commonly used in North America, Europe, and Asia, the general strategy for intracranial GCT involves surgery for diagnosis and/or CSF diversion, and the combination of chemotherapy and radiation therapy.

In recent years, there have been substantial advancement in the understanding of the molecular basis of intracranial GCTs. However, due to the rarity of these tumors and lack of adequate tissue samples, molecular profiling has been challenging. In recent years, utilizing blood and/or CSF as an alternative has been evaluated by various groups. Given that CSF collection is a standard component of the diagnostic workup and evaluation of response to therapy, several groups have sought to evaluate CSF for novel biomarkers of intracranial GCTs.

One such example is with MicroRNAs (miRNAs), which has been emerging as a novel biomarker for several malignancies, including GCTs. MiRNAs have been studied extensively in extracranial GCTs, particularly in testicular GCTs (26–28). In extracranial GCTs, the miRNA clusters (miR-371-373 and miR-302/367) have been identified as biomarkers of malignant GCTs, but are notably not expressed in benign teratomas (26, 29). Specifically, miR-371a-3p has emerged as a highly sensitive and specific marker of malignant testicular GCTs (30). In patients with intracranial GCTs, two small case series have recently demonstrated the feasibility of detecting these two miRNA clusters (31, 32). Despite promise, larger validation studies will be needed to demonstrate reproducibility of this methodology, as well as evaluate the sensitivity and specificity of these miRNA clusters in the setting of intracranial GCTs. If miRNAs prove to be a sensitive diagnostic tool for detecting intracranial GCTs, this could potentially be beneficial to the group of patients who present with neuroendocrine dysfunction and slowly growing suprasellar lesions, who often have a delay in diagnosis due to negative tumor markers and insufficient mass for biopsy (33, 34).

In addition to miRNA, circulating tumor DNA (ctDNA) is another evolving field in oncology that holds immense promise. Particularly in CNS tumors, various researchers have looked at the utility of CSF to identify recurrent molecular alteration, both at diagnosis and for disease monitoring (35–37). Recurrent somatic mutations in the KIT/RAS/MAPK pathways and AKT/mTOR pathways have been well documented in intracranial GCTs, with KIT/KRAS/MAPK alterations known to be enriched in germinomas. Takayasu et al., analyzed 8 germinomas and 4 NGGCTs for the presence of ctDNA in CSF of patients, utilizing a next generation sequencing (NGS) panel that covered 52-genes. In this cohort, they identified five genetic alterations, including two KIT mutations, two NRAS and one MAPK2K1 mutation (38). Recently, Zhang et al. published a cohort of 17 NGGCT patients, where they were able to detect ctDNA in the CSF of 13 of the 17 patients at initial diagnosis. Importantly in this study, the investigators found that presence of ctDNA in the CSF after chemotherapy treatment to be prognostic. The NGS panel used to assess for ctDNA in this study covered 86 genes, and all CSF ctDNA found were copy number alterations in genes such as AKT2 and MAPK1, among others (39). These studies show proof-of-concept and the feasibility of evaluating CSF for ctDNA. However, larger cohorts (ideally with paired tissue) will be needed to determine the true frequency and reliability of capturing genomic alterations by ctDNA in CSF.

Recurrent chromosomal alterations, such as copy number gains, losses, and structural variants are the most common somatic alterations identified in GCTs. In a recent study, tumor analysis of intracranial GCTs showed that gain of 12p (a common alteration in testicular GCTs) is enriched in NGGCTs. Additionally, investigators from this study reported that the presence of 12p gain is associated with worse progression-free survival (PFS) and overall survival (OS), making this a potentially useful prognostic biomarker (40). Additionally, a gain of 3p25.3 has recently been reported as an independent poor prognostic factor for some extracranial and intracranial GCTs (41, 42). Given the potential prognostic value of these two chromosomal gains, one could consider ctDNA analysis for the presence of these alterations as a component for risk stratification.

Lastly, DNA methylation profiling is rapidly emerging as a valuable tool for the diagnosis of pediatric brain tumors. Currently, tissue samples have been utilized to create classifiers to diagnosis brain tumors, down to the level of genetic alteration subclassifications (43). Lack of robust intracranial GCT tissue samples representing all the various histology subtypes has made classifier challenging for this tumor type, but the German Cancer Research Center (DKZF) (https://www.molecularneuropathology.org) has incorporated some intracranial GCT histologic types, including germinoma, yolk sac and teratoma. Classification of the other NGGCT histologies has yet to be developed, and therefore the ability to classify mixed NGGCTs is still to be determined. Although further refinement is needed for intracranial GCT classification, differentiating germinoma from NGGCT can be distinguished by assessing the global DNA methylation patterns. Broadly, DNA methylation profiling of intracranial GCT tissue samples has shown that germinomas have global hypomethylation, while NGGCTs are globally hypermethylated (44).

As with other emerging molecular technologies, profiling intracranial GCTs has been hindered by the paucity of sufficient tissue samples for analysis. As the availability of tissue for patients can vary, the development of a liquid biopsy platform with ctDNA would be of great interest. Of note, the DKFZ methylation platform was developed based off the Illumina Infinium MethylationEPIC array platform, which calls for 250 ng of DNA input. The feasibility of obtaining 250 ng of ctDNA from CSF is unclear, as this would require large quantities of CSF. As such, other methylation sequencing platforms such as methylation DNA immunoprecipitation sequencing (MeDIP-seq) and enzymatic methyl sequencing (EM-seq) are being explored for DNA methylation profiling of lower inputs of DNA, such as cfDNA from CSF or plasma (45, 46). These technologies hold potential promise for developing cfDNA methylation profiling of CSF.

Taking all these emerging technologies and biomarkers into consideration, we are moving towards better means of diagnosing and stratifying IGCTs, which would be immensely helpful for treatment planning, risk stratification and in clinical trial design. These emerging technologies and biomarkers are summarized in Table 1 .

The advancement in our understanding of the molecular drivers of cancer has led to the development of biologic agents and targeted therapy for various malignancies. For intracranial GCTs, activating alterations in the MAPK pathway, including KIT, RAS, and PI3K/mTOR pathway, are known to be commonly seen in intracranial GCT (47, 48). KIT expression is of particular interest, as it is seen in the majority of pure germinoma and not seen among NGGCT without germinomatous component. In recent years, various inhibitors of KIT have been developed, with several gaining FDA approval for gastrointestinal stromal tumor (GIST) (49). With the success of targeted therapy in other pediatric indications, KIT inhibition has recently emerged as an intriguing potential treatment approach for CNS germinoma. Several trials have been proposed, both for recurrence disease as well as for upfront treatment (to potentially decrease the dose of chemotherapy needed for cure). These trials are actively under development.

Immunotherapy has also emerged as an effective treatment modality for a variety of cancers. Various immune checkpoint inhibitors have been approved for many malignancies, especially in adults. The role of immune checkpoint inhibitors in primary pediatric CNS malignancies, however, is unclear. One exception is for patients with constitutional mismatch repair deficiency syndrome (cMMRD) and high tumor mutational burden, where there is a clear indication and improved outcomes with immune checkpoint inhibition (50). In GCTs, there have been several case reports suggesting that this treatment modality may be of therapeutic potential in these tumors. This is evidenced by the durable responses reported in these cases with multiply recurrent/refractory disease (51–53). This includes a case of a multiply recurrent and refractory CNS NGGCT, who was treated with nivolumab and ipilimumab, resulting in a complete response and durable remission for over five years (51). Additionally, several recent studies have also demonstrated robust presence of tumor infiltrating lymphocytes and/or expression of immune checkpoint markers in both CNS germinoma and a subset of CNS NGGCT, further supporting the potential of this treatment modality in this patient population (54, 55).

For patients with recurrent CNS GCTs, trials with these innovative approaches are critically important to potentially expand therapeutic options and possibly augment the contemporary treatment paradigm, especially for recurrent disease. Additionally, if deemed effective, these treatments could be incorporated into the upfront treatment regimens, potentially decreasing the need for/dose of cytotoxic chemotherapy and radiation therapy, thereby reducing treatment related short- and long-term side effects.