Ovarian metastases are well known to both clinicians and pathologists and account for a significant subgroup of ovarian neoplasms, posing various problems in diagnosis and management. Secondary tumors of the rest of the gynecologic tract are rather infrequent during routine practice, and they can be easily overlooked. Pathogenesis of secondary gynecologic tumors remains unclear with lymphogenous, hematogenous, and transcoelomic pathways being proposed 1. Detailed histologic data with regard to secondary tumors of the gynecologic tract are largely lacking. The type of primary tumors and the type and time of secondary lesions are data that could help delineate the mechanisms and the nature of metastatic disease in gynecologic organs. Thus, the aim of this study was to investigate the detailed clinicopathologic features of secondary tumors involving the gynecologic tract.
MATERIALS AND METHODS
The pathology files from our institution were retrospectively reviewed for all tumors secondarily involving the gynecologic tract for a period of 20 yr. In addition to originally diagnosed metastases, all malignant lesions were reviewed to find cases diagnosed as tumors of uncertain origin. All available slides, reports, and clinical files were evaluated to achieve the correct diagnosis. History of prior cancer was recorded from clinical files or pathology reports. Immunohistochemical study had often been performed at the time of diagnosis or was currently completed, according to the morphology and prior clinical history; it included mostly CK7, CK20, ER, PR, and WT1. Criteria to make the differential diagnosis between a metastatic mucinous tumor from a primary ovarian mucinous neoplasm (Fig. 1) included tumor size, bilaterality, surface involvement, pseudomyxoma ovarii presence, CK20/CK7 immunohistochemical profile, and most importantly, the clinical proof of another primary tumor 2–4.
Cases in which a definite distinction could not be made because of insufficient morphologic, immunohistochemical, or clinical data were excluded. Cases with multiple metastases to the gynecologic tract either simultaneously or at different time points were all separately recorded. Intragenital tumors with spread to contiguous sites, such as between ovary and tube, between adnexa and uterine serosa, or between endometrium and cervix, were also excluded. However, extension from adnexa to myometrium/endometrium/cervix, and from cervix/endometrium to adnexa or vagina were included.
After applying the aforementioned criteria, 196 secondary tumors were detected among almost 30,000 gynecologic specimens assessed at the same time period. Eight cases were excluded after not having sufficient data to properly classify these tumors; they included undifferentiated tumors heavily infiltrating the pelvis, in which it was impossible to determine their site of origin. Sixteen tumors had metastasized to >1 site in the genital tract or at different time points and each metastatic occasion was recorded in these cases. Consequently, a total of 196 secondary tumors originated from 180 different primary neoplasms.
The following parameters were recorded: (a) the primary site, (b) the histologic type, (c) the grade of the original tumor, if applicable, that is, for gastrointestinal, breast, endometrial adenocarcinomas, and urothelial carcinomas, (d) the presence of excessive mucous production for gastrointestinal adenocarcinomas (with mucus consisting >50% of tumor mass in the primary tumor) in the form of colloid adenocarcinomas or signet-ring adenocarcinomas 5, (e) the type of secondary involvement, defined as distant metastasis, as direct extension from an adjacent tumor or as locoregional recurrence, (f) the clinical information of a tumor prior to or concurrent with the metastasis, (g) the type of the secondary tumor as synchronous or metachronous with the primary tumor, and (h) if metachronous, the time to metastasis.
Vaginal secondary tumors with surgical excision were separately evaluated for the pattern of invasion as only superficial (epithelial and superficial stroma) or deep into the vaginal wall.
Secondary tumors involved the ovary (Figs. 1, 2) in 98 cases (50%), the vagina (Fig. 3) in 43 cases (22%), the myometrium in 21 cases (10.7%), the cervix in 20 cases (10.2%), the endometrium in 7 cases (3.6%), the vulva in 4 (2%), and the Fallopian tube in 3 (1.5%) (Table 1). Primary tumors included in descending order colorectal tumors (39.8%), endometrial (15.3%), breast (12.7%), ovarian (10.7%), and gastric tumors (5.6%), whereas other localizations were rare (Table 1). Most tumors were adenocarcinomas (126, 64.2%); there were also 22 serous carcinomas (11.2%), 7 urothelial carcinomas (3.6%), 4 carcinosarcomas (2%), 3 clear cell carcinomas (1.5%), and 3 neuroendocrine carcinomas (1.5%). Among breast tumors, 13 were ductal and 11 were lobular. Other tumor types were very rare. Of the 98 gastrointestinal adenocarcinomas, 31 (31.6%) showed excessive mucus production.
In 177 of the 196 cases wherein the information was available, the primary tumor was diagnosed previous to or concurrent with the metastasis; in 21 cases (9.7%) the metastasis was found first, resulting in a search for the primary neoplasm. In 86 cases (43.9%) the secondary tumor was metachronous to the primary tumor with a mean time to recurrence of 55.5 mo, whereas in 112 cases (56.1%) the lesion was synchronous to the primary tumor.
The site of involvement (Table 2) was significantly associated with the origin of the primary tumor (P<0.0001), as gastrointestinal tumors involved mostly the ovaries, endometrial tumors the vagina, ovarian tumors the myometrium, and urothelial tumors the cervix/vagina. Among gastrointestinal tumors, there was a strong tendency (P=0.06) for those with excessive mucus production to mostly affect the ovary, whereas those without excessive mucus production involved the uterus and the vagina.
As for the type of secondary involvement, distant metastases were the most common type (64.8%), followed by direct extension (19.9%) and local recurrence (15.3%). Type of secondary involvement differed with the organ affected, as ovaries and endometrium were affected mostly by distant metastases, vagina by local recurrences, and myometrium/cervix by direct extension (P<0.0001). The type of secondary involvement also differed by the primary tumor’s features (Table 3). Distant metastases were more frequent with gastrointestinal and breast tumors, whereas local recurrences were more frequent with gynecologic tumors (P<0.0001). Lower grade adenocarcinomas (grades 1 and 2) presented commonly with local recurrences, whereas higher grade adenocarcinomas presented with distant metastases and direct extension (P=0.01). Adenocarcinomas with excessive mucus production gave rise more frequently to distant metastases than to direct extension or local recurrences (P=0.01).
Metachronous metastases mostly came from breast and endometrial tumors, whereas synchronous metastases originated from gastrointestinal and ovarian ones (P=0.02). Time to recurrence (Table 4) for metachronous lesions significantly differed between breast and all gastrointestinal tumors (0.007), between breast and colorectal tumors (0.012), between breast and all gynecologic tumors (0.043) and between breast and endometrial tumors (P = 0.044).
In 29 cases of vaginal involvement, the surgical excision specimen was available (Fig. 3); it showed only superficial invasion in 10 cases (34.5%) while deep invasion was found in 19 cases (65.5%). This invasion significantly correlated with the type of primary tumor, as endometrial carcinomas showed mostly superficial invasion (10/15) in contrast with ovarian, cervical, or colorectal carcinomas involving the vagina (n=14), which all showed deep wall invasion (P=0.0002).
Although ovarian metastases is a well known and frequent condition, the detailed features of secondary lesions affecting gynecologic tract organs overall have not been adequately studied. In the current study we show that almost half of secondary tumors involving the gynecologic tract are of gastrointestinal origin, followed by intragenital tumors, and breast and urothelial neoplasms, whereas other sites are very rare. Carcinomas and specifically adenocarcinomas account for the vast majority of metastases, with other tumor types being exceedingly rare.
Type of secondary involvement differs between tumors of different origin, as primary endometrial tumors most often give rise to local recurrences, whereas gastrointestinal tumors present mostly as distant metastases. It is also associated with the tumor grade, as lower grade tumors present more often with local recurrence than with distant metastasis. The opposite is true for excessive mucus production by gastrointestinal adenocarcinomas: these tumors present in their majority as distant metastases, whereas adenocarcinomas without mucous production show more often direct extension or local recurrences. Time to recurrence also differs between ovarian, myometrial, and cervical lesions, which were often synchronous, while vaginal ones were mostly metachronous to the primary tumor. As for the time to recurrence for metachronous lesions, breast cancer showed the longest time to relapse.
Disease features also differed by the organ affected. Thus, the ovary was mostly affected by distant metastases, whereas the vagina was affected by local recurrences. This was also reflected in the histologic type, as vaginal involvement was seen mostly with endometrioid adenocarcinomas, whereas ovarian involvement was more frequent with gastrointestinal adenocarcinomas, but also with tumors showing excessive mucus production. Metastatic organotropism, defined as the propensity of tumors to preferentially metastasize in a particular secondary site, is not well studied for gynecologic organs as it is for lung, bone, or brain, and its overlying mechanisms remain unknown for these organs 1. Here we show that a primary tumor’s characteristic, that of excessive mucous production, is a major feature in tumors that metastasize to the ovary, suggesting a possible role of mucous in this organotropism. A large autopsy study examining the metastatic spread of colorectal carcinoma showed that mucinous carcinomas and signet-ring cell carcinomas metastasize more frequently on the peritoneal surface in comparison with common adenocarcinomas 6. Similarly, despite the ovary being a rare metastatic localization in this study, comprising only 2.1% of all metastases, it was more often found with these carcinomas 6.
Available studies on secondary tumors of the gynecologic system mostly report on ovarian metastases; a study of 131 ovarian metastases showed that the gastrointestinal tract was the most frequent site of origin; 7 similarly, a study of 255 nongenital ovarian metastases showed that 57% come from the gastrointestinal tract, followed by breast cancer in 29% of the cases 8. Data for the rest of the sites are sparse and often consist of case reports. In a large retrospective study, 46% of metastases were from extragenital sites, whereas the remaining 54% were metastases from primary tumors of the genital tract; the most common primary tumors were endometrial (34.1%), colorectal (17.2%), breast (16%), and ovarian (9.5%) ones 9. As for the site of involvement, 44% involved the ovaries, 36.7% the vagina, 5.2% the vulva, 4.6% the endometrium, 3.7% the cervix, and 2.8% the salpinx 9. In 86% the primary tumor was diagnosed previous to or concurrent with the discovery of the metastasis 9. Neto et al. 10 reported in a series of 66 vulvar metastases diagnosed over a period of 57 yr that almost half of them arose from gynecologic tumors—22.7% were cervical, 12.1% ovarian, 9% endometrial, and 3% vaginal primaries. From the other half, most were gastrointestinal (18.2%) and breast (6%) primaries 10. Lemoine and Hall 11 studied 33 cervical metastases diagnosed in a 65-yr period; most cases were ovarian primaries (36.3%), followed by colon (30.3%), stomach (15.1%), and breast (12.1%) primaries. The rarity of metastasis to the cervix is attributed to its low vascularity and fibrous stroma as well as its lymphatic drainage, which is centrifugal, requiring excessive blockage, so as to facilitate retrograde flow to occur 11. As for the uterine corpus, a study of 63 cases showed that most of them involve the myometrium 12. Most primaries are breast (42.9%), followed by colon (17.5%), gastric (11.1%), and pancreatic (11.1%) tumors 12. An important primary site of secondary cervical/vaginal tumors in our study was urothelial carcinomas. One previous study described the features of 6 urothelial carcinomas involving the gynecologic tract; it showed that time to involvement was 2 to 8 yr, and the first site of involvement was the vagina or the cervix 13. The presence or absence of invasion in the original urothelial carcinoma did not correlate with invasive disease of the gynecologic tract, neither with presentation as continuous lesions nor as “skip” lesions 13.
These studies provide important epidemiological data, but they do not investigate the factors that may influence the nature of these metastases. Our results delineate a probable pattern of tumor extension: gastrointestinal adenocarcinomas with excessive mucus production will preferentially affect the ovaries in a metachronous and distant metastasis manner, whereas gastrointestinal adenocarcinomas without mucus production can affect the whole gynecologic tract, mostly through direct extension or local recurrence. Endometrial tumors preferentially affect the vagina by local recurrence, and urothelial tumors affect the uterine cervix and vagina mostly by direct extension. Breast carcinoma metastases can affect all gynecologic tract organs, and they will present much later than gastrointestinal or gynecologic recurrences. Breast carcinoma was the third cause after gastrointestinal and intragenital tumors involving the gynecologic tract. Ductal and lobular carcinomas were equally represented. Breast cancer showed a mean time to recurrence of almost 8 yr. In a study of 102 breast cancer metastases, only 10 included the gastrointestinal and gynecologic tract but without further distinction between the 2 organ systems 14. Lobular carcinomas (11/102) affected more frequently (n=5) these sites than other carcinoma types 14, a finding not reproduced in our series.
A major issue of metastatic disease in the gynecologic tract is the presence of simultaneous endometrial and ovarian tumors that occurs in 5% to 10% of endometrial or ovarian cancers 15. This has long been a major concern for pathologists, as the prognosis and treatment differ if the lesions are considered as 2 different primaries or as metastasis from the one to the other. For decades, the classic criteria of Ulbright and Roth 16 and Scully 17 have been used to define which tumors are ovarian metastases from an endometrial primary: multinodular ovarian pattern, small ovaries, bilateral ovarian involvement, large endometrial tumor, deep myometrial invasion, vascular invasion, tubal lumen involvement, atypical endometrial hyperplasia, and absent ovarian endometriosis. However, molecular analysis of simultaneous endometrial and ovarian cancers showed that, actually, the vast majority has a common ancestral clone; they are therefore metastases, whereas independent primary tumors seem to be restricted only in patients with Lynch syndrome 15. The better prognosis seen with some of these metastases, actually those with the original histologic criteria of independent primary tumors, may be explained by the different mechanism of metastasis—that of detachment and spreading through the tubal lumen 15.
Vaginal recurrence, which is the most common site of failure for an endometrial cancer, ranges from 0% to 19% and is more likely to occur with more advanced age at diagnosis, and in tumors of higher grade, stage, and with lymphovascular invasion 18. To avoid this recurrence, adjuvant external beam radiotherapy has been originally applied, replaced nowadays mostly by vaginal cuff brachytherapy, as it shows less toxicity but the same efficacy 18. In the current series, vaginal recurrence of an endometrial cancer was the second most common cause of all gynecologic secondary tumors. Vaginal involvement was seen more often with endometrioid carcinoma, and to a lesser extent with serous carcinomas or carcinosarcomas; endometrioid adenocarcinomas were mostly of grades 1 and 2. Furthermore, we show that, in most endometrial carcinomas, vaginal recurrences, in contrast to other tumor type recurrences, showed superficial invasion, if any, of the cervical stroma. The mechanism for vaginal recurrence in endometrial cancer is not known, but our data support an implantation mechanism of tumor cells into the vagina, which could occur either during preoperative uterine bleeding or during surgery.
To conclude, our study shows that the primary tumor’s origin and histology dictate a different pattern of secondary involvement of the gynecologic tract, as endometrial tumors produce mostly superficial vaginal recurrences, mucus-producing gastrointestinal tumors present mostly as distant ovarian metastases, whereas breast tumors affect the whole gynecologic tract and present tumors with the most late recurrences.
The authors thank Philippe Cosmo from the Tumorothèque/Centre de Ressources Biologiques de CHU Saint-Etienne (BRIF no. BB-0033-00041) for his assistance.
1. Kubeček O, Laco J, Špaček J, et al. The pathogenesis, diagnosis, and management of metastatic tumors to the ovary: a comprehensive review. Clin Exp Metastasis 2017;34:295–307.
2. Karpathiou G, Venet M, Chauleur C, et al. Eosinophilic cells, autoimplants, degree of cellular proliferation, and adenofibroma pattern are important histologic findings in ovarian borderline tumors. Int J Gynecol Pathol 2017;36:447–52.
3. Karpathiou G, Venet M, Mobarki M, et al. FOXA1 is expressed in ovarian mucinous neoplasms. Pathology 2017;49:271–6.
4. Karpathiou G, Chauleur C, Corsini T, et al. Seromucinous ovarian tumor A comparison with the rest of ovarian epithelial tumors. Ann Diagn Pathol 2017;27:28–33.
5. Chu PG, Chung L, Weiss LM, et al. Determining the site of origin of mucinous adenocarcinoma: an immunohistochemical study of 175 cases. Am J Surg Pathol 2011;35:1830–6.
6. Hugen N, van de Velde CJ, deWilt JH, et al. Metastatic pattern in colorectal cancer is strongly influenced by histological subtype. Ann Oncol 2014;25:651–7.
7. Sal V, Demirkiran F, Topuz S, et al. Surgical treatment of metastatic ovarian tumors from extragenital primary sites. Int J Gynecol Cancer 2016;26:688–96.
8. Skirnisdottir I, Garmo H, Holmberg L. Non-genital tract metastases to the ovaries presented as ovarian tumors in Sweden 1990–2003: occurrence, origin and survival compared to ovarian cancer. Gynecol Oncol 2007;105:166–71.
9. Mazur M, Hsueh S, Gersell D. Metastases to the female genital tract
analysis of 325 cases. Cancer 1984;53:1978–84.
10. Neto AG, Deavers MT, Silva EG, et al. Metastatic tumors of the vulva: a clinicopathologic study of 66 cases. Am J Surg Pathol 2003;27:799–804.
11. Lemoine NR, Hall PA. Epithelial tumors metastatic to the uterine cervix. A study of 33 cases and review of the literature. Cancer 1986;57:2002–5.
12. Kumar NB, Hart WR. Metastases to the uterine corpus from extragenital cancers. A clinicopathologic study of 63 cases. Cancer 1982;50:2163–9.
13. Reyes MC, Park KJ, Lin O, et al. Urothelial carcinoma involving the gynecologic tract: a morphologic and immunohistochemical study of 6 cases. Am J Surg Pathol 2012;36:1058–65.
14. St Romain P, Madan R, Tawfik OW, et al. Organotropism and prognostic marker discordance in distant metastases of breast carcinoma: fact or fiction? A clinicopathologic analysis. Hum Pathol 2012;43:398–404.
15. Anglesio MS, Wang YK, Maassen M, et al. Synchronous endometrial and ovarian carcinomas: evidence of clonality. J Natl Cancer Inst 2016;108:djv428.
16. Ulbright TM, Roth LM. Metastatic and independent cancers of the endometrium and ovary: a clinicopathologic study of 34 cases. Hum Pathol 1985;16:28–34.
17. Scully R, Young R, Clement P. Tumors of the Ovary, Maldeveloped Gonads, Fallopian Tube, and Broad Ligament. Washington DC: Armed Forces Institute of Pathology; 1998.
18. Harkenrider MM, Block AM, Siddiqui ZA, et al. The role of vaginal cuff brachytherapy in endometrial cancer. Gynecol Oncol 2015;136:365–72.