Tracking the origin of simultaneous endometrial and ovarian cancer by next-generation sequencing – a case report

Endometrioid adenocarcinoma of the uterus and ovarian endometrioid carcinoma share many morphological and molecular features [1, 2]. Coincidence of both primary tumours is a relatively common event which occurs in about 10% of patients with ovarian carcinoma and in about 5% with endometrial carcinoma [3]. When simultaneous uterine and ovarian carcinomas show the same histology and differentiation, determination of the primary origin may be difficult. In case of independent primary tumours, this might be due to the several common risk factors shared by the two malignancies or due to germline alterations that predispose to the development of simultaneous cancers. In most cases, however, one of the endometrioid type carcinomas is the primary tumour that gives rise to a metastatic lesion. It is also not excluded that both carcinomas might originate from a common endometrial epithelial precursor cell of origin, a theory that can explain the histological and genetic similarities of the lesions [4]. Distinguishing between two independent lesions versus one primary and one metastatic lesion is extremely difficult, nevertheless important for the prognosis and treatment of the patient [5].

A novel powerful tool to identify common genetic alterations specific for a cancer entity and to provide information on the tumour clonal evolution is next-generation sequencing (NGS) [6]. To date, the knowledge of the mutational landscape of many different tumours including endometrial and ovarian cancer has increased throughout the last 10 years. TCGA has published data from whole exome sequencing of a large number of both gynaecological malignancies [1, 2]. Furthermore, studies have been reported on endometrial and ovarian carcinoma using two different cohorts – one including patients with endometrial endometrioid carcinomas and one consisting of ovarian endometrioid carcinomas [4]. The authors suggest that although the same genes are affected in both cancers the mutation patterns differ possibly due to the different microenvironment in the ovary and the uterus. This offers a possibility to determine the origin of simultaneously identified endometrial and ovarian endometrioid lesions. While these reports contribute to the big picture of what are the most common aberrations in the corresponding tumour type and beyond doubt profit from the large number of cases investigated, single patients with synchronously diagnosed endometrial and ovarian cancer can shed light on the direct tumour evolution. Although NGS is broadly used in preclinical research, thus providing the possibility to unravel genetic alterations and potential therapeutic targets, in the clinical setting current diagnostics and treatment of metastatic endometrial and ovarian cancer is still mainly based on radiological and conventional histological results as well as chemotherapy combination regimens.

In the present case report we investigated the mutation pattern in a 33-year-old female patient diagnosed with synchronous endometrioid adenocarcinoma of the corpus uteri and the left ovary. The patient developed a lung metastasis of an endometrioid type adenocarcinoma four years after laparoscopic hysterectomy, vaginal brachytherapy and tibolon therapy. Based on histological findings, it is difficult to decide whether the gynaecological carcinomas are two independent primary carcinomas occurring in parallel. Furthermore, it is almost impossible to recognize the origin of the lung mass. Thus, we employed targeted NGS of 409 tumour-related genes (Ion Ampliseq Comprehensive Cancer Panel, Thermo Fisher) in order to identify the relation between the two carcinomas and the source of the distant metastasis. Moreover, we were interested in the genetic fingerprint of the metastasis to clarify the recurrence-free survival of the patient after its removal and megestrol treatment for seven years. Finally, we looked for possible germline mutations from blood samples of the patient that could explain the early onset of the disease.

In the present study, we investigated the molecular profile of a simultaneous low-grade endometrioid endometrial and ovarian carcinoma as well as of a lung metastasis in a young female patient. From the histology alone, the primary origin is sometimes difficult to determine. The identification of precursor lesions, the pattern of spread and tumour size are valuable parameters for this differentiation. Immunohistochemical analysis of hormone receptors, PTEN and beta-Catenin may show overlapping staining patterns and is therefore not always helpful. The simultaneous occurrence of an endometrial and ovarian carcinoma in women below 50 years of age is a common combination [3], however, the pattern of pulmonary metastasis in this low grade and low stage cancer disease is unusual.

In our molecular analysis, we detected and verified mutations in genes such as ARID1A, CTNNB1, PIK3CA, and PTEN which are frequently affected in these tumour entities [7, 8]. However, no striking alterations in mismatch repair genes known to cause early development of endometrial cancer were detected.

We further showed that some mutations were lost during the metastasis formation, either due to the treatment or because of redundancy with other alterations affecting the same pathway. Last but not least, we identified enrichment of mutations causing upregulation of PI3K signalling that could explain the prolonged complete remission under megestrol acetate.

Somatic mutations were identified with targeted PCR-based NGS and verified with Sanger sequencing. The fact that 67% (8/12) of the detected alterations were shared by the ovarian and endometrial tumour with similar variant frequencies suggests a common origin of the two lesions. PTEN is more frequently mutated in low-grade endometrial endometrioid carcinomas (67%) compared to low-grade ovarian endometrioid carcinomas (17%). In contrast, CTNNB1 mutations are significantly different in low-grade ovarian endometrioid carcinomas (53%) compared to 28% of the low-grade endometrial endometrioid carcinomas [4]. In the present case, both genes were mutated in both tumours, hampering the unambiguous identification of the primary lesion. However, all shared mutations showed a slightly higher variant frequency in the endometrial adenocarcinoma. Based on this observation it is tempting to speculate that the endometrial adenocarcinoma gave rise to the ovarian adenocarcinoma. The clones that populated the ovary seemed to be of mixed origin and the cells containing the mentioned mutations could have lost their selection advantage in the ovarian microenvironment. The origin of the lung metastasis, however, could not be specified unambiguously.

The early onset of the malignant disease in this young woman at age 33 suggests possible germline mutations that accelerated the accumulation of genetic aberrations promoting cancer development. A common genetic defect that leads to the development of endometrial cancer in the context of Lynch syndrome is aberrations in the mismatch repair (MMR) proteins. Interestingly, neither mutations in the MMR genes nor increased double strand breaks could be detected. Another possible cause of the early disease onset is Cowden’s syndrome. Targeted PCR-based re-sequencing, however, is not the ideal method to detect large gene deletions. The uniformity blot (sequencing depth across amplicons) for the PTEN gene showed no inconsistencies. Therefore, larger deletions on chromosome 10 were unlikely. In addition, germline PTEN mutations could be ruled out.

The one base insertion detected in gene ARID1A causes a frame shift at codon 1053 of the protein. Since the percentage of the mutation suggested a heterozygous alteration, the sustained immunohistochemical staining is probably due to wild-type ARID1A expressed in the tumour samples. However, our study does not provide information on the expression status of the mutated protein or possible dominant negative interference with wild-type ARID1A. The mutation was not present in the metastasis or present at a low percentage from a marginal subclone indicating that this mutation was not substantially contributing to the tumour spread.

In contrast, the somatic mutations in CTNNB1 and PIK3CA were detected not only in the endometrial and ovarian tumours, but also in the lung metastasis. Both mutations are reported in COSMIC, providing experimental evidence that both mutations affect the function of the each protein and are pathogenic. The amino acid change p.Thr41Ile caused by the SNV c.122C > T in CTNNB1 increases protein stability of beta-catenin and leads to its relocalisation to the nucleus [9]. The SNV c.1637A > G in PIK3CA results in a mutant PI3K displaying increased substrate phosphorylation [10]. Hyperactivation of the PI3K pathway is a known cancer driving event in endometrial cancer [8].

Another key protein that is a part of the same signalling cascade and was affected by gene mutations in the described case is the tumour suppressor PTEN. The PTEN SNV c.370 T > A results in replacement of the cysteine in the active site against serine (p.C124S) leading to an enzymatically inactive protein. Although, as shown by immunohistochemistry, the protein level was not affected and the mutation was likely to be heterozygous, it has been recently reported that this enzymatically dead mutant inhibits wild-type PTEN in a dominant negative manner [11]. Additionally, we found a second PTEN mutation (c.376G > A) that leads to a threonine replacing the alanine at the adjacent position 126 of the protein, likely to distort the tertiary structure of the active site. None of the PTEN mutations could be detected in the lung metastasis. Instead, another PTEN mutation (p.Asp92Tyr) was detected in the lung metastasis that is also annotated in COSMIC. A possible explanation for the loss of the dominant negative PTEN mutation is the biological redundancy of the PTEN inactivating and PIK3CA activating mutations that both lead to an elevated signalling through the PI3K pathway. Indeed, there is evidence that single clones from the primary tumour are able to colonise distant organs and that the metastases carry rather less mutations than the primary tumour due to natural selection process that sustains only the aberrations indispensable for tumour cell survival and proliferation [12]. Interestingly, four years after surgery the patient presented with a lung metastasis that was excised and a therapy with megestrol acetate was initialised. To date, the patient is recurrence-free. Megestrol acetate is reported to act through de-phosphorylation of the PI3K downstream target Akt [13], which might be a possible reason for the complete response of the patient to the treatment.

It has been shown very recently by two independent groups that a majority of synchronous endometrial and ovarian endometrioid carcinomas show clear evidence of clonality [14, 15]. The findings of these two studies strongly indicate that a considerable part of clinically synchronous endometrial and ovarian endometrioid carcinomas constitute in fact dissemination from one site to the other which is also very likely in our case. In our study we extend these results by next generation sequencing analysis of a distant metastasis and demonstrated that not all mutations found in the primary tumour were conserved in the metastasis. This is illustrated e.g. by the SNVs in PTEN (p.Cys124Ser and p.Ala126Thr) and ARID1A (p.Arg1053fs) which were found in both, the endometrial and the ovarian endometrioid carcinoma, but were not conserved in the distant lung metastasis. Co-occurrence of mutations in ARID1A and PTEN have been shown as potential cancer driving mutations in endometrioid ovarian carcinomas in animal models [16]. Interestingly, these two mutations were not conserved in the distant metastasis, indicating another clonal event. Neither the PTEN nor the ARID1A mutation has given rise to the distant metastasis. To our knowledge this is the first case demonstrating the lack of an ARID1A mutation in a distant metastasis of an ARID1A mutated primary tumour. This demonstrates that even very frequent and potential cancer driving mutations such as PTEN and ARID1A in endometrioid endometrial and ovarian cancer may not be found in distant metastases, a fact that is essential for potential future targeted therapies [17‐19]. However, this also demonstrates the importance of sequencing metachronous distant metastases since its mutational profile may differ from the profile of the primary tumour.

In conclusion, our results demonstrate that next generation sequencing is an important tool in the differentiation of simultaneous primary tumours and metastases, which has an important impact for clinical practice. This may not only be valuable information for prognostic considerations, but also may be of increasing importance for future targeted therapies.