Clonal Origin of Metastatic Testicular Teratomas

Testicular germ cell tumors (GCT) are the most common solid malignancies in men between the ages of 20 and 45 years (1) and frequently metastasize to retroperitoneal lymph nodes. Teratomas show tremendous histologic diversity, containing a variety of tissue elements derived from all three embryonic germ cell layers. Mature teratomas are typically composed of differentiated tissues and may occur in the testes of both children and adults. However, the pathogenesis and behavior of these tumors are quite different in these two age groups with postpubertal patients having the potential for metastasis. Previous studies have shown that 22% to 43% of adult patients with “pure” teratoma in the testis may have metastases (2–4) and these metastases often are associated with other GCT components (4, 5). Likewise, the absence of teratomatous components in the primary tumor does not preclude its presence in retroperitoneal metastases (6, 7). Mixed testicular GCTs with a teratomatous component are much more common than pure teratomas of the testis in adult patients. These tumors frequently have a component of teratoma in their metastases.

Teratomas of the adult testis are thought to be derived from the same malignantly transformed germ cell that gives rise to other GCT elements, and in many cases, teratomas are thought to arise directly from other GCT components, such as yolk sac tumor or embryonal carcinoma. Indeed, Kernek et al. (8) have shown that mature teratoma shares a common clonal origin with other GCT components within the primary testicular tumor. It is unclear whether metastatic mature teratomas result from a direct metastasis of a primary testicular teratoma or whether these metastatic lesions result from teratomatous differentiation of a more primitive GCT component following metastasis. In either case, it is clear that mature teratomas should be regarded as malignant neoplasms despite a deceptively benign histologic appearance based on the fact that primary pure teratomas can be associated with retroperitoneal lymph node metastases and the fact that mature teratomas can show cytologic atypia (5), chromosomal aneuploidy (9, 10), and chromosome 12p abnormalities (11). These tumors should be treated with complete surgical excision of both the primary and metastatic tumors (12–15). Because the molecular genetic relationships among mature teratoma and other GCT components within metastatic lesions have not been previously studied, we analyzed metastatic mature teratomas in 31 patients with one or more other coexisting metastatic GCT elements using loss of heterozygosity (LOH) analysis in an effort to assess the clonal relationships of these different components. Because the clonal relationship among primary and metastatic GCTs of different components has not been well established, we compared the patterns of allelic loss in metastatic mature teratoma with each GCT component of the primary testicular tumor in six cases to define the clonal relationships among these lesions.

Patients. Thirty-one men with a history of mixed GCTs of the testis and with retroperitoneal lymph node metastases underwent primary retroperitoneal lymph node dissection from 1991 to 2004. The mean age of patients was 29 years (range, 18-54 years). Retroperitoneal histology revealed metastatic teratoma in all patients. On pathologic examination, the retroperitoneal metastases all contained teratoma but also included embryonal carcinoma in 12 cases, yolk sac tumor in 12 cases, choriocarcinoma in 9 cases, and seminoma in 8 cases (Table 1). All metastatic GCT components were either within the same lymph node or within the same regional lymph node group. The number of GCT components present in addition to mature teratoma ranged from one to three. In six patients with available tissue from primary testicular tumor, the metastatic lesions were compared with the primary testicular neoplasm. Each of the six primary tumors contained mature teratoma. In addition, yolk sac tumor, embryonal carcinoma, choriocarcinoma, and seminoma were present in five, two, two, and one of the primary mixed GCTs, respectively.

This research was approved by the Indiana University Institutional Review Board.

Tissue samples and microdissection. This study included a total of 72 metastatic GCT components and 16 primary GCT components from 31 patients (Table 1). Histologic sections were prepared from formalin-fixed, paraffin-embedded tissue and were stained with H&E for microscopic evaluation. From these slides, the different GCT components were identified (Fig. 1). Laser-assisted microdissection of the neoplastic cells was done (Fig. 1) on H & E stained sections using a PixCell II laser capture microdissection system (Arcturus Engineering, Mountain View, CA) as previously described (8, 16, 17). Approximately 400 to 1,000 cells of each component were microdissected from the 5-μm histologic sections. Normal tissue (lymphoid tissue) from each case was microdissected as a control.

Detection of LOH. PCR was used to amplify genomic DNA at seven specific loci on four different chromosomes: 1p36 (D1S1646), 9p21 (D9S171 and IFNA), 9q21 (D9S303), 13q22-q31 (D13S317), 18q22 (D18S543), and 18q21 (D18S60). Previous studies have shown that LOH at these loci frequently occurs in GCTs (8, 18–22). D1S1646, D9S303, D13S317, D18S543, and D18S60 contain putative tumor suppressor genes that may play a role in the development of a number of human malignancies. D9S171 and IFNA include regions of the putative tumor suppressor gene p16. Chromosome 18q21 corresponds to the DCC gene. PCR amplification and gel electrophoresis were done as previously described (8, 16, 23). The criterion for allelic loss was complete or nearly complete absence of one allele in tumor DNA (24–26). PCRs for each polymorphic microsatellite marker were repeated at least twice from the same DNA preparations and the same results were obtained.

Analysis of allelic loss pattern. When the genetic material in a patient was found to be homozygous for the polymorphic markers (i.e., showing only one allele in the normal control tissue), the case was considered noninformative. Patients with genetic material that was informative (i.e., showing two alleles in the normal control tissue) were divided into two categories: cases showing no allelic deletions in the tumor, retaining two different alleles of similar intensity on autoradiographs, and those with absence of one allele (27–29). DNA sampled from the cells of a metastatic mature teratoma and from the cells of another coexisting GCT component showing identical allelic loss patterns is compatible with a common clonal origin, whereas different patterns of allelic deletions are compatible with independent clonal origins.

The overall frequency of allelic loss was 94% (29 of 31) in metastatic mature teratomas and 93% (38 of 41) in other metastatic GCT components (Table 1). The number of specific loci lost in a single metastatic mature teratoma ranged from one to five. The number of specific loci lost in the other metastatic GCT components also ranged from one to five. The frequency of allelic loss in all of the informative samples of metastatic mature teratoma was 27% (8 of 30) with D1S1646, 34% (10 of 29) with D9S171, 37% (10 of 27) with IFNA, 27% (8 of 30) with D9S303, 46% (13 of 28) with D13S317, 26% (7 of 27) with D18S543, and 36% (10 of 28) with D18S60.

In every patient with LOH in their metastatic mature teratoma, the other coexisting GCT components also showed allelic loss (Fig. 2). An identical or nearly identical allelic loss pattern was seen in 29 of 29 (100%) cases in which LOH was shown at one or more loci. Completely concordant patterns of allelic loss between mature teratoma and all other coexisting metastatic GCT components were present in 26 of the 29 patients in which the mature teratoma component showed LOH. The remaining three cases showed nearly identical allelic loss patterns with only one locus in each case showing a difference in LOH. In each of these three cases, an identical pattern of allelic loss was seen at all other loci. Two cases did not show LOH at any of the seven loci examined.

In six cases, we compared the pattern of LOH seen in each metastatic lesion with that seen in each GCT component of the primary testicular tumor (Table 1, cases 26-31). In all six cases, an identical allelic loss pattern was seen in each component of both the metastatic and the primary tumors, confirming the common clonal origin of these lesions (Fig. 3).

GCTs of the adult testis often show a mixed histology, having multiple components, such as teratoma, yolk sac tumor, embryonal carcinoma, choriocarcinoma, and seminoma. Approximately 70% of testicular GCTs metastasize; however, treatment with surgical excision and cisplatin-based chemotherapy provides cures in >90% of cases (30). Whereas the molecular genetic relationships among different, coexisting GCT components in primary testicular neoplasms have previously been examined, these relationships have never been studied in mixed metastatic lesions. In addition, the precise genetic interrelation among primary and metastatic GCTs consisting of multiple different components is unclear. In this investigation, we examined 31 metastatic GCTs, each consisting of a component of mature teratoma as well as other GCT elements, by loss of heterozyosity analysis in an effort to assess tumor clonality. We also compared LOH patterns among different GCT components of primary and metastatic lesions. We found evidence for a common clonal origin for metastatic mature teratoma, the other metastatic GCT components, and each component of the corresponding primary testicular tumor.

Teratoma is a common component of testicular GCTs in adults. These teratomas are histologically diverse, consisting of differentiated tissues derived from any or all three embryonic germ cell layers, and often show both epithelial and mesenchymal differentiation. Teratoma is thought to derive from the same malignantly transformed germ cell that gives rise to other GCT components, possibly by transformation of such components (31). This means that despite a deceptively benign histologic appearance, mature teratomas are malignant in postpubertal males. Evidence in support of their proposed pathogenesis includes the relative rarity of pure teratomas in the adult testis, the presence of intratubular germ cell neoplasia, unclassified in association with pure teratomas of the testis, and the demonstration of chromosomal aneuploidy and chromosome 12p abnormalities in teratoma and other GCT components. Molecular and cytogenetic studies provide additional support for this pathogenesis (8, 9, 20, 32). Kernek et al. (8) analyzed 16 mixed testicular GCTs, each having mature teratoma as one of the components, by LOH analysis and found an identical pattern of allelic loss among the coexisting GCT components in the vast majority of cases. Similarly, Rothe et al. (20) examined 20 mixed testicular GCTs by LOH analysis and found similar patterns of allelic loss among the different components. Gillis et al. (32) used in situ hybridization for centromeric regions of chromosome 15 to show the common clonality of the different components of mixed testicular GCTs. Similar karyotype abnormalities were shown in coexisting mature teratoma and choriocarcinoma in one case (9).

Whereas the clonal relationships among GCT components in primary testicular tumors have been established, similar analyses have not been done on metastases. Metastatic mature teratoma is frequently present in retroperitoneal lymph nodes following chemotherapy in patients with stage II or III testicular GCTs. A meta-analysis review of the histology of surgical specimens from 24 publications (996 patients) found that residual mature teratoma was present after chemotherapy in 36% of cases (33). Other GCT components were much less frequent. It makes sense that different GCT components, if clonally related in the testis, should likewise be clonally related in metastases from these primary tumors. However, human cancers are often genetically heterogeneous neoplasms and the differences in responsiveness to chemotherapy between mature teratoma and other components suggest that there could be genetic differences between these categories of GCT. In a case report by de Graaff et al. (34), mature teratoma and immature teratoma components of a metastatic testicular GCT were karyotyped and shown to have similar patterns of chromosomal abnormalities, suggesting that phenotypic differences among these components may be epigenetically determined. The results of the current study are compatible with the findings of de Graaf et al. and also show that nonteratomatous elements are clonally related to both the mature teratoma component and to each other. Brandli et al. (16) found that stromal cells adjacent to metastatic mature teratoma in postchemotherapy lymph node specimens frequently have genetic abnormalities similar to the metastatic teratoma, suggesting that both are derived from germ cells.

In a study by van Echten et al. (35), cytogenetic analysis of the residual metastatic teratoma after chemotherapy and of the primary testicular GCT was undertaken. They showed that there were no significant chromosomal differences between 31 residual metastatic teratomas and 70 testicular GCTs. Our results are consistent with those of van Echten et al. in that we showed identical LOH patterns in all GCT components of both primary and metastatic tumors in six of six patients. Residual mature teratomas after chemotherapy are often composed of fully differentiated tissue. This high degree of differentiation in the postchemotherapy state has been postulated to be secondary to several possible causes: therapy-related induction of tumor cells to fully differentiated cells; the selective destruction of components other than mature teratoma; or the selection of cells with an inherent capacity of spontaneous differentiation or therapy-related differentiation (35–38). Allelic losses at the seven loci examined in the current study, however, are unlikely to be involved in therapy-related differentiation or chemoresistance of metastatic GCTs as identical or nearly-identical patterns of LOH were seen in all components regardless of their level of differentiation and LOH was not present at any single locus in all tumors or even in the majority of tumors. It is unlikely that DNA alterations at chromosome 1p36 (D1S1646), 9p21 (D9S171, IFNA), 9q21 (D9S303), 13q22-q31 (D13S317), 18q22 (D18S543), or 18q21 (D18S60) play a critical role in tumor metastasis because LOH at any of these loci was not observed in all or even the majority of metastatic lesions. Our data show that each component within a metastatic GCT is clonally related and that each metastasis is clonally related to all components of the primary testicular tumor. Because all primary and metastatic components showed identical patterns of allelic loss, these data do not answer the question of whether the teratoma component metastasized directly from the primary testicular tumor or whether it arose from one of the other metastatic GCT components. Our data are compatible with both of these scenarios. As the seven polymorphic markers analyzed in this study represent only a very small fraction of the genome, additional genetic aberrations not detected by our methods could help to define genes important in metastasis and could potentially clarify the mechanism by which metastatic mature teratomas arise.

In conclusion, concordant genetic alterations were seen in both metastatic mature teratomas and other coexisting metastatic GCT components. Similar genetic concordance was shown in both metastatic and primary GCTs. These results provide further evidence that all GCT components are derived from a common malignantly transformed precursor cell and that mature teratoma, although often histologically bland, is a malignant tumor displaying identical genetic derangements as other GCT components, including yolk sac tumor, embryonal carcinoma, choriocarcinoma, and seminoma.