Immunohistochemistry in the distinction between primary and metastatic ovarian mucinous neoplasms
- Correspondence to Professor W Glenn McCluggage, Department of Pathology, Royal Group of Hospitals Trust, Grosvenor Road, Belfast BT12 6BA, Northern Ireland;
- Accepted 27 June 2011
- Published Online First 18 July 2011
Although traditionally regarded as the second most common type of primary ovarian carcinoma following serous carcinoma, recent studies have illustrated that primary ovarian mucinous carcinomas are rather uncommon neoplasms representing approximately 3% of primary malignant ovarian epithelial tumours.1 2 Although some of the differences in prevalence between older and recent studies may be explained by different criteria used to distinguish between a mucinous borderline tumour at the upper end of the spectrum and a well-differentiated mucinous carcinoma with expansile invasion,3–5 the main reason for the decrease in incidence is that previously many metastatic mucinous carcinomas in the ovary were probably misinterpreted as primary ovarian neoplasms; clinical, gross and microscopic pathological features suggestive of a metastatic mucinous carcinoma in the ovary are discussed elsewhere in this issue. One point I wish to make is that although metastatic mucinous carcinomas in the ovary are still sometimes misdiagnosed as a primary ovarian mucinous carcinoma or even a mucinous borderline tumour due to the pronounced maturation effect seen with some secondary mucinous carcinomas in the ovary, we have to some extent come full circle in that, in my opinion, there is now a tendency to overplay the possibility of a secondary mucinous carcinoma even when the pathological features are obviously those of a primary ovarian neoplasm. I consider that in a large majority of cases, the distinction between a primary and a secondary mucinous carcinoma in the ovary can be achieved by careful pathological examination encompassing both the gross and microscopic findings and taking into account the distribution of the disease. It has been stated that when a mucinous carcinoma is diagnosed in the ovary, further investigations, such as colonoscopy and detailed imaging of the upper abdomen, should be undertaken to exclude a primary neoplasm elsewhere. I feel this is unnecessary in most cases since, as discussed, basic pathological examination is usually sufficient to distinguish between a primary and a secondary ovarian mucinous neoplasm.
Carcinomas arising in a wide range of primary sites may metastasise to the ovary; in this review, I largely restrict my comments to metastatic adenocarcinomas with abundant intracytoplasmic mucin, the so-called mucinous carcinomas. Metastatic carcinomas that do not typically contain significant intracytoplasmic mucin, such as breast carcinoma, are not discussed. Metastatic mucinous carcinomas in the ovary may be from a number of primary sites, chiefly including the colorectum, pancreas, biliary tract, appendix, stomach and cervix.6–8 If there is no known history of a primary neoplasm elsewhere (or even if there is), the likelihood of the ovary or one of the aforementioned sites being the primary site is largely initially formulated on the basis of the morphological features, as discussed elsewhere in this issue. However, immunohistochemistry may be of considerable assistance (table 1). The markers discussed may also be useful in helping to ascertain the primary site of a disseminated mucinous carcinoma. It is worth pointing out that a large majority of primary ovarian mucinous carcinomas are unilateral and stage 1 (confined to the ovary) at presentation.9 10 Therefore, a disseminated mucinous carcinoma in a woman is unlikely to be of ovarian origin.
At this point, I stress that most primary ovarian mucinous carcinomas (and borderline tumours) are of the so-called intestinal or enteric type (discussed in the next section). These intestinal-type neoplasms are almost invariably negative with oestrogen receptor (ER), progesterone receptor (PR) and WT1 and are almost always CA125 negative.9 Hormone receptor negativity means that ER and PR are of no value in distinguishing between a primary and a secondary ovarian mucinous neoplasm. Recent studies have shown that a proportion of primary ovarian mucinous carcinomas overexpress HER2 and exhibit HER2 amplification.11 A response to trastuzumab therapy has been shown in a small number of cases of recurrent ovarian mucinous carcinomas11 and future studies are warranted to assess the value of targeted therapies in this setting.
Metastatic colorectal adenocarcinoma
The most definitive use of immunohistochemistry in the distinction between a primary and a secondary mucinous carcinoma in the ovary is in helping to confirm or exclude a colorectal metastasis. However, most metastatic colorectal adenocarcinomas within the ovary, although they contain some intracytoplasmic mucin, are not overtly mucinous and more closely resemble a primary ovarian endometrioid than a mucinous carcinoma, a so-called pseudoendometrioid appearance.12 Features in favour of a primary ovarian endometrioid adenocarcinoma include the presence of an adenofibromatous growth pattern, squamous elements or endometriosis, this triad being characteristic of primary ovarian endometrioid neoplasms. Features in favour of a colorectal metastasis include the presence of dirty or segmental necrosis, an absence of the triad described and more significant nuclear atypia compared with the degree of glandular differentiation than seen in endometrioid adenocarcinomas. However, there may be morphological overlap, and in this diagnostic scenario, immunohistochemistry is almost definitive. The so-called differential cytokeratin (CK7 and CK20) staining can be used along with CA125, ER, carcinoembryonic antigen (CEA) and CDX2. Primary ovarian endometrioid adenocarcinomas are usually diffusely positive with CK7, CA125 and ER and negative with CK20, CEA and CDX213–24 (some primary ovarian endometrioid adenocarcinomas exhibit focal nuclear staining with CDX2 and squamous morules within these neoplasms are commonly diffusely positive25 26). By contrast, metastatic colorectal adenocarcinomas with a pseudoendometrioid appearance are usually diffusely positive with CK20, CEA and CDX2 and negative with CK7, CA125 and ER. CEA staining is sometimes difficult to interpret since there is typically staining of necrotic material and inflammatory cells such that it may be problematic to assess staining in the tumour cells.
In the distinction between a primary ovarian mucinous neoplasm (borderline or carcinoma) and those metastatic colorectal adenocarcinomas with an overt mucinous appearance, immunohistochemistry is less useful, although it may be helpful. As stated, most primary ovarian mucinous carcinomas and borderline tumours are of the so-called intestinal or enteric type.3–5 They are very commonly positive with the enteric markers CK20, CEA, CA19.9 and CDX2. CA19.9 is often diffusely positive while CK20, CEA and CDX2 are most commonly focally positive but may be negative or diffusely reactive such that in an individual case they may not assist due to overlap. Positivity with these markers may be reflected in increased serum levels of CEA and especially CA19.9 in primary ovarian mucinous neoplasms. For example, it has been shown that the serum CA19.9 may be massively elevated in primary ovarian mucinous neoplasms of intestinal type and that these levels are of no value in predicting whether a suspected ovarian mucinous neoplasm is benign, borderline or malignant.27 Helpfully, primary ovarian mucinous neoplasms of intestinal type are usually, although not always, diffusely CK7 positive, even when they exhibit staining with the enteric markers listed; as such, the distribution of CK7 and CK20 staining is more important than absolute positive or negative staining. Although most colorectal carcinomas, including those with an overt mucinous appearance, are CK7 negative or very focally positive, it should be borne in mind that rectal carcinomas are more likely to be CK7 positive than those at other sites in the large intestine28 29 and this is a potential pitfall. In one study, 74% of rectal adenocarcinomas were CK7 positive.28 Additionally, poorly differentiated colorectal adenocarcinomas may exhibit aberrant staining and are more likely to exhibit some degree of CK7 positivity than better differentiated neoplasms.18 30 It can be summarised that although immunohistochemistry may be useful, there is some degree of immunophenotypic overlap between primary ovarian mucinous carcinomas and borderline tumours of intestinal type and metastatic colorectal adenocarcinomas with a mucinous appearance.
A number of other markers such as villin, α-methylacyl-CoA racemase, meprin α, guanylyl cyclase C, MUC2, dipeptidase 1 and β-catenin have been reported to be of value since they are more likely to be expressed in colorectal than primary ovarian carcinomas31–36; some of these markers were chosen for immunohistochemical analysis after gene expression profiling showed them to be unregulated in colorectal adenocarcinomas compared with ovarian mucinous carcinomas.32 36 37 However, due to intestinal differentiation in primary ovarian mucinous neoplasms, there is considerable immunophenotypic overlap and these markers, most of which are not in widespread use, may not be of value in an individual case. In one study of primary ovarian mucinous and metastatic colorectal carcinomas, 83% of metastatic colorectal carcinomas exhibited nuclear positivity with β-catenin while only 9% of primary ovarian mucinous carcinomas did so.31 It should be noted that ovarian endometrioid adenocarcinomas may exhibit nuclear β-catenin positivity, especially within the squamous morules; this is secondary to β-catenin gene mutation.38
Some primary ovarian mucinous neoplasms are of teratomatous origin and exhibit an overt large intestinal immunophenotype
A small proportion of primary ovarian mucinous neoplasms of intestinal type are of teratomatous origin.39 40 Although other teratomatous elements are often present on microscopic examination, in some cases they are overgrown by the mucinous neoplasm. Some of these tumours exhibit a similar immunoprofile to the usual primary ovarian mucinous neoplasms of intestinal type in that they are diffusely CK7 and CA19.9 positive and focally positive with CK20, CEA and CDX2 (these may represent upper gastrointestinal-type mucinous neoplasms arising in a teratoma); others exhibit an overt large intestinal immunophenotype being negative with CK7 and diffusely positive with CK20, CEA and CDX2.
Metastatic appendiceal neoplasms
Immunohistochemical studies have been instrumental in establishing that in most cases of pseudomyxoma peritonei (PMP), the appendix is the site of primary neoplasm, with the low-grade mucinous epithelium within the appendiceal, ovarian, omental and peritoneal lesions being positive with the intestinal markers CK20, CEA, CDX2 and MUC2 and usually negative with CK7.41–45 One exception is that primary ovarian mucinous neoplasms arising in teratomas and exhibiting an overt large intestinal immunophenotype may uncommonly give rise to PMP40; it seems to be overt large intestinal-type mucinous epithelium with a low-grade cytology, which has the capacity to give rise to PMP.
Outside the setting of PMP, metastatic appendiceal adenocarcinomas within the ovary may be composed of signet ring cells and small mucinous glands and closely mimic a metastatic gastric adenocarcinoma rather than the more usual colorectal adenocarcinoma.46 It is my experience that with such morphology, the appendix is often overlooked or not considered as a likely site of primary neoplasm. In this scenario, immunohistochemistry is useful in that if the neoplasm is diffusely positive with CK20, CEA and CDX2 and negative with CK7, this is suggestive of a metastatic appendiceal rather than gastric adenocarcinoma.
Metastatic pancreatic, biliary tract and gastric adenocarcinomas
A relatively uncommon, but sometimes problematic, area is the distinction between a metastatic pancreatic or biliary tract mucinous carcinoma within the ovary and a primary ovarian mucinous carcinoma or borderline tumour.47 The immunophenotype of pancreatic and biliary tract carcinomas will be discussed together since these are similar. There is considerable immunophenotypic overlap between primary ovarian mucinous neoplasms of intestinal type and metastatic pancreatic or biliary tract mucinous carcinomas in that these neoplasms are most commonly diffusely positive with CK7 and CA19.9 and they may be negative, diffusely or focally positive with CK20, CEA and CDX2. It is doubtful whether any marker, at present, can reliably distinguish between a primary ovarian intestinal-type mucinous neoplasm and a metastatic pancreatic or biliary tract mucinous carcinoma since the former, on the basis of histochemical, immunohistochemical and ultrastructural studies, has been shown to contain pancreatic, biliary or gastric-type mucins (especially gastric type), which can collectively be referred to as upper gastrointestinal-type mucins.48 However, some markers may be of value. In my experience, pancreatic and biliary tract carcinomas are often CA125 positive49 50 and this may be useful in that most primary ovarian mucinous neoplasms of intestinal type are negative or at the most focally positive.9 Therefore, any appreciable CA125 staining in a mucinous neoplasm is against a primary ovarian carcinoma (an exception is primary ovarian Mullerian-type mucinous neoplasms, which may be CA125 positive—discussed later) and more in keeping with a pancreatic or biliary tract carcinoma. However, we have recently noticed an unusual phenomenon in that recurrent or metastatic primary ovarian mucinous carcinomas of intestinal type may be CA125 positive, even when the primary neoplasm was negative.51 As stated earlier, primary ovarian mucinous neoplasms of intestinal type are almost invariably negative with ER, PR and WT1. Pancreatic and biliary tract adenocarcinomas are also usually negative with these markers,52 although I have noticed that some pancreatic adenocarcinomas are focally PR positive. There is often non-specific cytoplasmic staining of mucinous carcinomas with PR. On low-power examination, this may erroneously be misinterpreted as a positive reaction and high power is required to confirm that the nuclei are negative. One marker that may be of use in the distinction between a primary ovarian mucinous neoplasm of intestinal type and a pancreatic or biliary tract adenocarcinoma is DPC4 (deleted in pancreatic carcinoma 4, MADH4, SMAD4). This is almost invariably positive in primary ovarian mucinous neoplasms while approximately 50% of pancreatic and some biliary tract adenocarcinomas are negative.53–57 This is because this tumour suppressor gene on chromosome 18q is genetically inactivated by allelic loss in approximately 50% of pancreatic and in some biliary adenocarcinomas. However, DPC4 is not in widespread use. Other markers that have been purported to be of use include mesothelin, fascin and prostate stem cell antigen since these are more likely to be expressed in pancreatic than ovarian mucinous carcinomas58; however, again there is overlap and these markers are not in widespread use.
Other upper gastrointestinal tract carcinomas, such as gastric carcinoma, when they metastasise to the ovary usually contain signet ring cells and the overall morphological features, in most cases, are not especially likely to be confused with a primary ovarian mucinous neoplasm. The immunophenotype of gastric carcinomas with regard to CK7, CK20, CA19.9, CEA, CA125 and CDX2 is essentially similar to that of pancreatic and biliary tract adenocarcinomas.
Metastatic cervical adenocarcinoma
Metastatic cervical adenocarcinoma within the ovary may mimic a primary ovarian mucinous or endometrioid neoplasm, either borderline or malignant; in such cases, the cervical primary may be extremely small or even not recognisably invasive.59–61 p16 may be useful in that most cervical adenocarcinomas are diffusely positive due to the presence of high-risk human papillomavirus (HPV) while most, but not all, primary ovarian mucinous and endometrioid neoplasms are negative or focally positive.62–64 More useful than p16 immunohistochemistry are molecular studies (PCR or in situ hybridisation) to demonstrate HPV, the presence of which is virtually diagnostic of a cervical primary. However, it should be remembered that unusual morphological types of primary cervical adenocarcinomas may be HPV negative, including overt mucinous neoplasms such as adenoma malignum.65 In the distinction between a primary ovarian endometrioid neoplasm and a metastatic cervical adenocarcinoma, ER and CEA are useful in that primary ovarian endometrioid neoplasms are usually ER positive and CEA negative while the converse applies to metastatic cervical adenocarcinomas. These markers are not useful in the distinction between a primary ovarian mucinous neoplasm of intestinal type and a metastatic cervical adenocarcinoma in that both tumour types are usually ER negative and often CEA positive.
Immunophenotype of primary ovarian Mullerian-type mucinous neoplasms
In discussing the immunophenotype of primary ovarian mucinous neoplasms, I have so far restricted my comments to those of intestinal or enteric type, by far the most common type of primary ovarian mucinous carcinomas and borderline tumours. A Mullerian- or endocervical-type primary ovarian mucinous neoplasm also exists; borderline variants are uncommon and carcinomas rare.66 The immunophenotype of ovarian Mullerian mucinous neoplasms has not been extensively investigated, and studies have mainly concentrated on borderline tumours, but there are distinct immunohistochemical differences between intestinal- and Mullerian-type mucinous neoplasms.67 68 Both are typically diffusely positive with CK7, although, as stated, intestinal-type neoplasms may be only focally positive or even negative on rare occasions. As discussed, the intestinal-type tumours typically express, focally or diffusely, enteric markers such as CK20, CEA, CA19.9 and CDX2. Mullerian-type mucinous neoplasms are usually negative with these markers (although they may be positive with CA19.9 and squamous elements, which can be present in these neoplasms, may be CEA positive) but typically express Mullerian markers, such as CA125, ER and PR. WT1 is usually negative. CA125, ER and PR are usually negative in intestinal-type mucinous neoplasms. One study described the presence of reserve cell-like cells, morphologically similar to reserve cells in the cervix, in Mullerian mucinous borderline tumours; similar to cervical reserve cells, these cells are positive with p63, 34βE12 and CK17.69
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.