Clinicopathological Characteristics of Gastric-type Endocervical Adenocarcinoma Misdiagnosed as an Endometrial, Ovarian or Extragenital Malignancy, or Mistyped as Usual-type Endocervical Adenocarcinoma

Discussion

Extension of endocervical adenocarcinoma to the uterine corpus is rarely encountered in daily practice. In addition to the shared histological features between endocervical and endometrial adenocarcinomas, extensive endomyometrial involvement of endocervical adenocarcinoma can lead to a misdiagnosis of primary endometrial adenocarcinoma with cervical stromal extension. An accurate and prudent differential diagnosis of uterine adenocarcinomas is crucial for treatment decision making, including the appropriate therapeutic option (e.g., radical vs. total hysterectomy), extent of lymph node sampling or dissection, and postoperative radiation therapy or chemotherapy (12).

GEA has a predilection to extend widely and deeply into the cervical stroma and myometrium and is usually associated with prominent stromal desmoplasia. However, in endometrial biopsy or curettage specimens, the amount of stromal tissue is often too small to evaluate whether the tumor is accompanied by desmoplastic reaction. Furthermore, because endometrial endometrioid carcinoma often shows an intracytoplasmic and intraglandular mucin, it is challenging to distinguish GEA from endometrial adenocarcinoma. Low-grade endometrioid carcinoma with mucinous differentiation typically displays mild nuclear pleomorphism, and it shows uniform and strong nuclear immunoreactivity for estrogen and progesterone receptors.

In addition to the hormone receptor positivity, aberrant nuclear β-catenin expression and loss of phosphatase and tensin homolog deletion on chromosome 10 (PTEN) expression favor a diagnosis of endometrial endometrioid carcinoma. High-grade endometrioid carcinoma with mucinous differentiation is frequently associated with microcystic, elongated, and fragmented (MELF) pattern of invasion; deep myometrial involvement; and LVSI. Prominent stromal desmoplasia and inflammatory infiltrates in association with MELF invasion pattern can raise the strong suspicion of GEA. However, high-grade endometrioid carcinoma of the endometrium consists of hypercellular, solid sheets of tumor cells with or without co-existing low-grade endometrioid carcinoma components.

The presence of stromal histiocytic aggregations and intraepithelial neutrophils as well as endocervical-type intracytoplasmic mucin favor endometrial endometrioid carcinoma. Detection of mismatch repair (MMR) deficiency can also be useful to distinguish endometrioid carcinoma from GEA, as the latter does not exhibit an MMR-deficient phenotype but frequently harbors a TP53 mutation. According to the latest WHO Classification of Tumors of Female Reproductive Organs (17), the most frequent type of endocervical adenocarcinoma is UEA, followed by GEA.

Although cervicovaginal cytology screening and HPV vaccination have reduced the incidence rate of squamous cell carcinoma and HPVA, GEA has become relatively frequent. GEA shows significantly worse oncogenic behavior and clinical outcome in comparison to UEA, and it also has a different histogenesis and immunophenotype (9, 11, 31, 32). The IECC documented that apical mitotic figures and frequent apoptotic bodies that can be easily identified in UEA (2). In general, UEA consists of tumor cells with enlarged, elongated nuclei and mucin-poor, eosinophilic cytoplasm. It also contains less than 50% of cells with appreciable intracytoplasmic mucin (2, 33). In contrast, GEA consists of tumor cells with abundant clear or pale eosinophilic cytoplasm, distinct cytoplasmic borders, and low nuclear-cytoplasmic ratios. Further, it exhibits prominent stromal desmoplasia. As UEA is much more prevalent than GEA and an intracytoplasmic mucin within 50% can also be observed in UEA, it is sometimes challenging for pathologists to differentiate endocervical adenocarcinomas of UEA from those of GEA.

Wada et al. (32) analyzed diagnostically confusing cases of endocervical adenocarcinoma for reclassifying based on the latest WHO classification. They reported that the UEA-like pseudoendometrioid morphology of GEA can be induced by intracytoplasmic mucin depletion. Further, mucin-rich, GEA-like components can also exist in UEA (32). In the current study, patients 4-6 exhibited some mucin-poor glands and subtle desmoplastic stromal reaction, causing a confusion with UEA. Accordingly, an awareness of histological features and careful interpretation of immunostaining results should be used to ensure proper classification and avoid mistyping GEA as UEA. Clinicians should consider the possibility of GEA in cases presenting with voluminous, clear to pale eosinophilic cytoplasm; prominent stromal desmoplasia; non-block p16 positivity; and mutant p53 immunostaining pattern.

Considering the small size, the ovary is a relatively common site of metastasis (34). Metastatic ovarian neoplasms are thought to account for 10-30% of all ovarian malignancies, and the common primary sites include gastrointestinal tract, breast, pancreatobiliary organ, and other gynecological organs (34, 35). Most patients with metastatic ovarian neoplasms have known history of primary malignancy. However, some of them manifest only symptoms associated with an ovarian mass without any clinical history of malignancy. Although metastatic ovarian neoplasms usually present as bilateral ovarian masses that can provide a hint for distinguishing between metastasis and primary origin, imaging sometimes detects only unilateral ovarian mass or rather several masses in multiple organs, such as in patient 8.

It is well-known that primary ovarian epithelial neoplasms, especially those showing mucinous differentiation, can be grossly and histologically mimicked by metastatic carcinoma, leading to misdiagnosis (36, 37). Grossly, primary ovarian mucinous tumors are typically large, unilateral, multilocular cystic lesions with a smooth capsule. Further, they histologically display an expansile/confluent glandular pattern or destructive/infiltrative stromal invasion (36). There are some histological features favoring a metastatic rather than a primary ovarian neoplasm. These include bilaterality, a nodular pattern of ovarian involvement, an infiltrative pattern of stromal invasion, microscopic surface deposits of tumor, and marked LVSI (38, 39). However, these features are not specific nor exclusive for metastasis.

The presence of benign and borderline components within the ovarian tumor can be a clue for the diagnosis of primary ovarian tumor. This is because primary ovarian carcinoma usually arises in association with a benign or borderline tumor. However, metastatic tumors can also sometimes include bland-appearing findings. Particularly, metastatic endocervical adenocarcinoma involving the ovary often has a deceptively bland-looking tumor area consisting of well-formed glands with mucinous epithelium, overlapping with those of primary ovarian benign or borderline mucinous tumor (35). This happens in other metastatic adenocarcinomas involving the ovary, especially those with mucinous differentiation, and is referred to as a maturation phenomenon (35, 38).

In addition to the above-mentioned difficulties, immunostaining is of no value to differentiate between cervical and ovarian origin, making it even harder for a pathologist to make a correct diagnosis. Both are expected to show cytokeratin (CK) 7 positivity and CK 20 negativity. In case of GEA, the most common HPV-independent adenocarcinoma, p16 immunostaining and HPV DNA detection are also not helpful. Despite these difficulties, it is crucial to correctly diagnose ovarian tumors as either primary or metastatic neoplasms because their clinical management and prognoses of may be significantly different from each other (36). Metastatic adenocarcinoma involving the ovary can morphologically mimic primary ovarian tumor; thus, pathologists should consider both the clinical and pathological context comprehensively.

As mentioned above, pancreatobiliary adenocarcinoma can also metastasize to the ovary. Notably, patients with pancreatobiliary neoplasm may not have apparent symptoms related to the primary site until late stage. As such, symptoms related to a metastatic ovarian mass or imaging findings of multicystic ovarian mass could be the first manifestation (38).

We previously reported four cases of ovarian metastasis of pancreatobiliary origin (36). In one of the four cases, the primary tumor was diagnosed by follow-up investigation only 6 months after ovarian resection. Similar to metastatic endocervical adenocarcinoma involving the ovary, pancreatobiliary adenocarcinoma occasionally has a bland appearance and is deceptively similar to a primary ovarian mucinous tumor (35). In an ambiguous clinical setting, it is difficult to determine the primary origin as the ovary, pancreatobiliary tract, and cervix among ovarian mucinous tumors. In addition, the limited amount of specimen during the intraoperative frozen section examination makes it even harder to make a correct diagnosis of metastatic ovarian mass.

A previous study suggested an algorithm based on tumor size and laterality to determine if an ovarian tumor is primary or metastatic. Using the algorithm, they correctly classified all cases of ovarian metastasis originating from the pancreatobiliary tract (40). The study showed that bilaterality and smaller size favors a diagnosis of metastatic disease. Immunostaining is also of limited value and should be interpreted with caution only as an ancillary test. Both pancreatobiliary and gynecological neoplasms are commonly positive for CK 7 and stain variably for CK 20. However, recent studies reported that some tumors originating from the pancreatobiliary tract demonstrate loss of staining with deletions in pancreatic carcinoma 4 and over-expression of CK17 (41, 42). Extensive sampling for clues such as foci of destructive stromal invasion, LVSI, and bilaterality is the first step toward an accurate diagnosis. Further, a comprehensive clinicopathological evaluation with consideration of the possibility of metastasis is required.

Interestingly, we found a case of GEA with tubal intraepithelial metastases that were multifocal but were also microscopic. Although adnexal metastasis of cervical carcinoma is very uncommon, some studies have reported a secondary involvement of the fallopian tube in endocervical adenocarcinomas (29, 43). Reyes et al. (43) found eight cases of endocervical adenocarcinoma with metastatic tubal involvement. They reported that the tubal metastatic tumors often showed mucosal colonization mimicking a primary tubal lesion. Further, six of the eight tumors were unilateral and microscopically colonized pre-existing tubal epithelium simulating STIC.

We have also previously demonstrated that three endocervical adenocarcinomas (i.e., GEA, UEA, and mucinous-type adenocarcinoma) had tubal intraepithelial metastases only as an adnexal involvement (29). In this study, five of the eight patients with GEA had adnexal metastases, and one patient showed tubal intraepithelial metastasis alone. This patient showed no metastatic lesion in other anatomical locations including the ovary, bilateral pelvic lymph nodes, abdominal and pelvic peritoneum. This finding is in line with previous reports of tubal metastasis in early-stage cervical carcinoma (44, 45). Our observation of the endomyometrial involvement and intraluminal microscopic tumor emboli supports the idea of superficial luminal spread of tumor cells as one of the routes for tubal metastasis.

The most important differential diagnosis of tubal intraepithelial metastasis is STIC. STIC is histologically characterized by cellular crowding, stratification, loss of polarity, and enlarged nuclei with hyperchromasia, all of which can be observed in cases of tubal intraepithelial metastasis (29). In the presence of mucinous differentiation, the diagnosis of intraepithelial metastasis is favored. However, diagnosis based on morphological features can be difficult in the absence of these findings. In these cases, careful consideration of clinical history and the use of immunohistochemistry would be helpful. STIC can present as multifocal lesions but has a predilection for arising in the fimbrial portion. Thus, STIC is unlikely to involve both fimbrial and non-fimbrial portions. The absence of WT1 immunoreactivity, a serous cell lineage marker, rules out the possibility of STIC.

In summary, we described the detailed clinicopathological characteristics of eight cases of GEA misdiagnosed as endometrial, ovarian, or extragenital malignancies and mistyped as UEA. As already documented in the literature, GEA is an aggressive malignancy that presents frequently as an advanced-stage disease with various manifestations. When GEA involves the uterine corpus extensively and displays mucin-depleted glands in a pseudoendometrioid architecture, it closely resembles endometrial endometrioid carcinoma. A mucin-poor GEA having hyperchromatic, elongated nuclei, and apical mitotic figures and frequent apoptotic bodies makes it difficult to distinguish GEA from UEA based solely on histological features. A non-block p16 positivity and mutant p53 immunostaining pattern will be helpful in providing the diagnosis of GEA. Lastly, GEAs metastatic to the ovary mimic both primary ovarian mucinous carcinoma and metastatic adenocarcinoma of pancreatobiliary origin. Deceptively bland-looking areas within an ovarian tumor exacerbate the difficulty in identifying the primary origin. Given that accurate diagnosis in these cases is very important owing to its clinical implications in pathological staging, management, and prognosis, pathologists should comprehensively understand the clinical and histological findings of this entity.