Translation research on testicular germ cell tumors

Germ cell tumors (GCTs) are neoplasms that can be diagnosed in the gonads (testes, ovaries and dysgenetic gonads). They originate from early embryonal stem cells or germ cells in various stages of their maturation. In addition, they can also be identified in extra-gonadal locations, like the mediastinum and the brain, that correspond to the migration route of primordial germ cells, i.e., the stem cell of gametogenesis, during embryogenesis [1]. Various subtypes of GCTs can be recognized. These show striking similarities within the defined subtypes, both in clinical behavior, pathogenesis, and molecular constitution, being irrespective of sex of the patient (male, female or indetermined) as well as anatomical localization (gonadal or extragonadal). The focus of this paper will be on testicular GCTs. Related to translational research, this paper will relate to three topics, being: 1) classification, 2) defined microRNAs as liquid biopsy biomarker, and 3) the underlying mechanism of treatment resistance. These are selected based on recently reported promising results, with potential significant impact on clinical management of this population of relatively young patients. The overall aim is to have the optimal tools for (early) primary diagnosis, informative follow up as well as (more) effective treatment, specifically in case of presence or development of refractory disease.

In fact, testicular GCTs can be divided in three subtypes [1, 2]. Type II GCTs are the most frequently subtype and occur in post-pubertal men with a mean age in the 3rd and 4th decade of life. These, by definition, malignant tumors can present themselves as two clinically relevant types of GCTs: seminomas (mean age 35 years) and non-seminomas (mean age 25 years), based on their histological appearance. The latter can consist of various histological subtypes: embryonal carcinoma, yolk sac tumor, choriocarcinoma and teratoma, either as one pure subtype or mixed within one tumor (schematically illustrated in Fig. 1). Moreover, a seminoma component can be mixed with the various types of non-seminomas (mean age 30 years). These different histological components mirror cell types found during intra-uterine embryonal development. All histological elements of type II GCTs develop from a common precursor lesion called Germ Cell Neoplasia In Situ (GCNIS), previously referred to as CIS, IGCNU and TIN ([3] for a historical overview). The invasive elements almost universally show increased chromosomal copy numbers of the short arm of chromosome 12 (12p), which is therefore not related to initiation but progression [4]. Without further specification, this paper will deal with the type II testicular GCTs (TGCTs).

Type I GCTs of the testis occur in prepubertal boys, rarely above 6 years of age [5]. They can consist of only two histological subtypes, i.e., yolk sac tumor and teratoma, with most tumors only presenting with teratoma, and as such showing an overall benign course. There is no association with GCNIS. In fact, the cell or origin is likely a cell in the transition between embryonal stem cell and primordial germ cell [6], although this is still not proven so far. These tumors also show no gain of 12p, nor any other universal genetic feature. Lastly, Type III GCTs, also known as spermatocytic tumors, historically spermatocytic seminoma, happen in older men and are benign, and are characterized by gain of chromosome 9 [7]. The cell of origin is most likely either a spermatogonium or spermatocyte [8, 9].

As indicated, TGCTs are derived from in principle pluripotent cells and therefore also express certain marker genes corresponding to this cell type, like OCT3/4 (also known as POU5F1), SOX2 and NANOG1. These markers are very specific and have proven to be useful in histological TGCT diagnosis, especially OCT3/4 [10]. Interestingly, several genes related to pluripotency, including NANOG, are located on the short arm of chromosome 12, suggesting that gain of this region could contribute to retaining pluripotency in TGCTs that is normally lost during development.

Retention of embryonal characteristics is also the source of some of the main clinically informative biomarkers used for detection, follow-up as well as risk classification. Serum levels of Alpha Feto Protein (AFP, an embryonic serum protein) and β‑Human Chorionic Gonadotropin (β-HCG, a placental hormone) are used together with levels of LDH to screen for patients with TGCTs. Based on the origin, non-seminomas show overall higher levels of these markers. In addition, levels above a pre-defined threshold are informative to identify patients with a poor prognosis, forming the basis for the IGCCCG risk classification which is widely used to classify metastatic (T)GCT patients into low, intermediate or poor risk categories [11]. However, these markers have several significant limitations, including the observations that their levels are elevated in only 30–40% of patients at initial diagnosis and that they are not specific for (T)GCTs but can also be elevated in different cancer types and other conditions [12], indicating the need for more accurate biomarkers. One of the expected targets is a defined set of so-called microRNAs (miRNA). This refers to a family of non-coding RNAs, relevant for normal development and cellular maintenance by messenger RNA translation regulation (i.e., protein formation), which might become mis-regulated during the processes of disease development, including cancer (including (T)GCTs).

Unbiased miRNA profiling has shown that there are several miRNAs that show differential expression between healthy tissue and TGCTs [13]. This group of miRNAs also shows different expression levels within the different TGCT histologies, with the highest expression in embryonal carcinoma, intermediate expression in seminoma and no expression in teratoma. Follow-up studies suggest that miRNA 371-3p is the most sensitive and specific target to identify patients with malignant GCTs (Type II as well as yolk sac tumor part of Type I). This miRNA is part of a cluster with miRNA-372 and miRNA-373, which are also highly expressed in (T)GCTs, but whose levels are less specific in identifying patients with this tumor type [14]. MiRNA-371-3p has a sensitivity of 90% and a specificity of 94% [15], respectively, consistently outperforming the serum protein markers described above that reach a sensitivity of only ~60%12. In summary, it is informative both in type I and type II GCT non-teratomatous elements. Moreover, GCNIS can be identified using miRNA371-3p expression in approximately 50% of cases, while classic protein markers are not informative for this lesion. The most significant downside is that miRNA371-3p detection is not informative for pure teratomas because this miRNA is not expressed in this histology (i.e., due to loss of the embryonic characteristics). However, the sensitivity for teratoma detection using the forementioned classical serum markers is also < 25%, stressing the need for sensitive biomarkers for the identification of patients with mature teratoma (both during the process of initial diagnosis as well as follow up). Levels of miRNA375-3p have been suggested to be an informative marker for the detection of teratoma [16], however, other studies have not been able to reproduce these results [17, 18]. An interesting alternative biomarker for teratoma could be the methylated fragment of the RASSF1A promotor [19], although this requires confirmatory studies, specifically focusing on the more modern detection methods, like digital droplet based polymerase chain reaction (ddPCR).

In addition, detection of miRNA371-3p has also been shown to be superior to protein markers in relapse detection (83% vs < 50%) [15, 20]. Moreover, several cases have been described where increased miRNA371-3p expression was detected months to years before clinical (including radiological) detection, indicating its potential application in surveillance as well [21, 22]. Interestingly, a health economic analysis has shown implementation of miRNA371-3p detection could lead to a reduction of the cost of surveillance of TGCTs of up to 44% due to the reduced need for imaging procedures [23], suggesting that this method could have clinical as well as economic benefits.

In spite of all the effort to determine their clinical value, not much is known about the functional role of miRNAs in (T)GCTs, however, it has been established that miRNA372 and miRNA373 can have an oncogenic role in primary (i.e., non-cancerous) cells. When an oncogenic RAS mutant is introduced into these cells, they normally enter senescence, however, expression of miRNA372 or miRNA373 prevents this and enables the cells to keep growing. This effect is functionally similar to P53 inactivation, which led the authors to investigate the link between these miRNAs and P53. They found that miRNA372-373 overexpression inhibits LATS2 expression leading to increased CDK2 activity and cell cycle progression, similar to observed when TP53 is mutated [24]. Interestingly, TGCTs show a very low frequency of TP53 mutations compared to most other types of cancer, suggesting that mir372-373 expression obviates the need for P53 inactivation. Concordantly, TP53 mutations were only discovered in TGCTs that show lower levels of miRNA-372-373. However, this is likely not the only mechanism involved, because lack of functional P53 does not result in absence of expression of this set of miRNAs. In addition, cisplatin refractory GCTs (with the exception of teratoma) still express this cluster, which is relevant in de context of the use of this liquid biopsy biomarker in this clinical setting.

Besides the functional similarities between overexpression of the microRNA371-373 cluster TP53 mutations there is also one clear difference. TP53 mutations are associated with poor therapy response and outcome in patients and were shown to induce cisplatin resistance when induced in GCT cell lines [25, 26]. For the miRNA371‑3 cluster it was reported that higher expression was correlated with higher tumor mass, however, no association with outcome and therapy response was reported. This suggests that mutation of TP53 has a broader impact in GCTs than only the re-activation of the cell cycle described before. Possibly related to this, TP53 mutations are preferentially observed in mediastinal (Type II) GCTs [25], which could suggest that the difference in microenvironment (mediastinum versus testis) requires a more potent oncogenic driver (TP53 mutation versus miRNA371-373 overexpression) [27].

Besides TP53 mutations and aberrations in associated genes, like MDM2 amplifications [25], also aberrations in WNT/CTNNB1 signaling, the absence of KIT mutations and increased tumor mutational burden have been associated with cisplatin resistance in TGCTs based on the Next Generation Sequencing (NGS) analysis of resistant tumor material [28]. Moreover, we recently established a relationship between copy number gain of chromosome 3p cytoband p25.3 and cisplatin resistance in a set of independently generated cell line models, as well as a relationship between this gain of this region and poor outcome in multiple publicly available datasets [29]. The inclusion of this biomarker improves significantly upon the IGCCCG classification and could therefore be used to better assess patients risk of poor treatment response. Interestingly, the 3p25.3 copy number gain described before occurs throughout all histologies but is only significantly associated with poor prognosis in non-seminomas. Moreover, it occurs significantly more often in yolk sac tumors then in other histological subtypes. Interestingly, yolk sac tumor histology is often enriched in refractory or relapsed tumors [30, 31], suggesting that certain histologies may be more amenable to acquiring genetic aberrations that induce cisplatin resistance. However, the driving mechanisms behind the association between 3p25.3 and cisplatin resistance remain to be determined and are the goal of currently ongoing studies. In this context the preliminary results will be extended regarding the role of 3p25.3 copy number gain in Type I and II GCTs beyond the testis including the ovary, mediastinal, and intracranial localizations. These results might facilitate a unique starting point to develop alternative treatment options for GCT patients with refractory disease, both intrinsic as well as acquired (with the exception of teratoma).

Apart from gain of chromosome 3p25.3 and genetic aberrations in the TP53/MDM2 axis also micro satellite instability and decreased expression of DNA mismatch repair proteins have been associated with cisplatin resistance [30, 32], suggesting that DNA repair, possibly as a result from separate underlying mechanisms, could play a general role in (T)GCT cisplatin resistance. A recent in vitro study supports this hypothesis by showing that the balance between the activation of different DNA repair pathways contributes to cisplatin resistance in this cancer indeed [33].

Other than cisplatin resistance that is acquired through genetic aberrations or other molecular changes, also inherent cisplatin resistance can play an important role in the variation in clinical behavior of (T)GCTs. The different histological subtypes show very different sensitivities to cisplatin. Seminomas are very sensitive and are therefore classified as low risk, irrespective of other criteria [11]. Teratoma, on the other hand is completely insensitive, and therefore, residual teratoma often remains after treatment. The teratoma is not intrinsically malignant but can cause problems because of its localization in respect to vital organs and in some cases can grow post chemotherapy [34], and is, therefore, required to be surgically resected. Moreover, teratoma also can progress to non-germ cell malignancies that are generally more aggressive and therapy resistant than TGCTs [35]. There is currently no good biomarker to identify these teratomas, since the miRNA-371-3p is not expressed, and therefore not detectable in liquid biopsies, so it is difficult to identify (residual) teratoma. However, combining miRNA371-3p expression with methylation analysis of the tumor marker RASSF1AM does enable the detection of teratoma, enabling easier identification of patients that require surgical resection [19]. Although of significant interest, based on the forementioned improvements of detection methods, it requires further evaluation related to its clinical impact.