|Home | About | Journals | Submit | Contact Us | Français|
Correspondence to: Gozde Kir, MD, Department of Pathology, Umraniye Education and Research Hospital, Umraniye, Istanbul 34766, Turkey. moc.oohay@rikzog
Telephone: +90-216-4494543 Fax: +90-216-4494842
Metastasis of colorectal adenocarcinoma of the ovary is not an uncommon occurrence and ovarian metastases from colorectal carcinoma frequently mimic endometrioid and mucinous primary ovarian carcinoma. The clinical and pathologic features of metastatic colorectal adenocarcinoma involving the ovary is reviewed with particular focus on the diagnostic challenge of distinguishing these secondary ovarian tumors from primary ovarian neoplasm. Immunohistochemical stains that may be useful in the differential diagnosis of metastatic colorectal tumors to the ovary and primary ovarian tumors are detailed.
The ovary is a common site of metastases. Secondary tumors account for 17.4%-30% of all ovarian malignancies[2-4] of which nongenital cancer metastases to the ovaries constitute 9%-14.6%[3-6]. Primary colon cancer has been identified in 10%-33% of metastatic ovarian tumors in various series[2-10]. In the literature, patients with metastatic colon cancer to the ovary (MCCO) range between 19 and 87 years (median 51 years) with 24%, younger than 40 years[11,12]. 1.2%-14% of women with intestinal cancer have ovarian metastases at sometime during the course of their disease[13-16]. Estimates of true incidence of ovarian metastases from a colorectal primary varies depending on whether autopsy data or clinical series are examined. At autopsy of women dying of colorectal cancer, 6%-14 % are found to have ovarian metastases[17,18]. Up to 45 % of MCCO are thought to be clinically primary ovarian tumors, even though most of the colonic tumors are of Dukes stage B or C[11,19-21]. These tumors may spread to the ovary via blood-borne or lymphatic routes, transperitoneal or by direct extension[13,22,23].
In this review, clinical and pathological status of MCCO are discussed with special focus on the diagnostic challenge of distinguishing these secondary ovarian tumors from primary ovarian neoplasms. Studies on useful immunohistochemical stains for the differential diagnosis are also discussed in detail.
Common presenting symptoms are usually related to ovarian involvement. Pelvic mass, abdominal and pelvic pain are the most common presenting symptoms[11,12,24]. Most patients have changes in bowel habits, rectal bleeding, feeling of abdominal fullness or bloating. Less frequently, patients present with abnormal vaginal bleeding, nausea, vomiting and constitutional symptoms such as fatique or weight loss. Stromal luteinization is most frequently found in MCCO and increased steroid hormone production in these patients often results in endocrine manifestations. However significantly younger age of the women, uniform presentation as pelvic masses with few bowel symptoms, elevated CA-125 levels, and occasional presentation as large clinically unilateral tumor can all contribute to misclassification of these metastases as primary ovarian neoplasms[26,27]. The frequency of metastatic colorectal carcinoma in the ovary relative to primary ovarian neoplasms is sufficient to justify colonoscopy as a preoperative test in women younger than 50 years, even in pregnant women with adnexal masses lacking clinical symptoms referable to the lower intestinal tract[26,28].
Colorectal tumors with ovarian metastases are predominantly distal lesions and most of them originate from the rectosigmoid colon[11,13,29]. Transvers colon, ascending colon, cecum and descending colon are affected with decreasing frequency[11,13]. Mean size of the primary tumor is less than that of the ovarian metastatic lesion[16,30]. MCCO are usually associated with advanced metastatic disease[11,12,15,20]. In a study, out of 19 colorectal tumors with ovarian metastases the stage of the primary tumors was as follows : Dukes stage : B1: 2 ; B2: 7 (Stage B: 47%); C2: 8 (Stage C: 42%); D: 2(Stage D:11%). In a report by
Lash et al; none of the intestinal primary tumors were Dukes stage A, 32% were B and 68% were C. In a study by Lewis et al, 86 cases of MCCO were reviewed. Primary tumor invaded the full thickness of the bowel wall in 58 cases (pT3), while in 17 cases, perforation of visceral peritoneum or direct invasion of other structures was noted (pT4). Nodal status was reported in 62 cases, of which 87% nodal involvement was documented at the time of resection; 29 (47%) had involvement of four or more nodes (pN2). Eighty one percent of patients had metastatic involvement of nonovarian sites either at or after the time of colectomy, with omental and/or other peritoneal involvement being the most common. Twenty five patients (40%) had liver metastases at some point during the course of their disease.
MCCO may form solid or, more commonly, partially or predominantly cystic masses[11,15,22,29,31,32]. They are often friable due to extensive necrosis[15,32] and tend to be associated with surface implants. Careful gross and microscopic inspection of the external ovarian surface for fibrous plaques containing infiltrating carcinoma is helpful in recognition of metastatic colorectal neoplasms. In the study by Lewis et al, among 46 cases for which data were available, 21 featured an ovary with evidence of surface involvement by tumor and 7 showed evidence of surface rupture. Nodular growth pattern (Figure (Figure1)1) and hilar involvement are less frequently seen but also highly correlated with metastatic carcinoma. In different series, the size of tumors ranges from 1 to 27 cm with a median of 10-11 cm[11,15,25,31,33-36]. MCCO are bilateral in more than 80% of the reported cases[11,12,15,29,31,33-36]. Unilateral metastasis is more frequent in the right ovary[11,29]. Most commonly, MCCO mimic primary ovarian mucinous or endometrioid adenocarcinoma[10-12,15,22,29,32-36]. Mucinous carcinomas are reported to comprise 6%-25% of ovarian carcinomas (mean 12%), although recent regimens in the interpretation of histologic features of noninvasive and metastatic mucinous carcinomas suggest that this may be an overestimate. Mucinous carcinomas in the ovaries are commonly metastatic, but the proportion of primary versus metastatic mucinous carcinoma in unselected patients is unknown. In Seidman’s report, among 52 cases of mucinous carcinoma in the ovaries, 40 (77%) were metastatic and 12 (23%) were primary.
In another report on 74 cases of mucinous carcinomas, 16 were primary ovarian; 52 metastatic, and 6 of indeterminate origin. An algorithm has been proposed to assist diagnosis in which all bilateral mucinous and those unilateral tumors < 10 cm are classified as metastatic carcinomas whereas unilateral tumors ≥ 10 cm are classified as primary ovarian mucinous carcinomas. In Khuramornpong’s series, when 6 tumors of indeterminate primary site were excluded, the proposed algorithm correctly classified primary and metastatic tumors in 84% of 68 cases. Of 21 unilateral mucinous adenocarcinomas ≥ 10 cm, 62% were primary ovarian. Of 5 unilateral tumors < 10 cm, 80 % were metastatic. Of 42 bilateral mucinous carcinomas, 95% were metastatic. By adjusting the size criteria to 12 cm, performance of the algorithm is both maintained for primary ovarian tumors and improved for metastases, giving correct classification of 86% of tumors overall including 100% primary tumors and 80% of metastases.
In general, features which assist in distinguishing metastatic colorectal adenocarcinoma from primary ovarian adenocarcinoma are similar to those which apply to other metastatic adenocarcinomas (e.g. bilaterality, nodular pattern of ovarian involvement, surface tumour deposits and extensive lymphovascular permeation especially in hilar and paraovarian vessels, single cell invasion, signet-ring cells) all favour metastatic rather than primary neoplasm. However none of these features is specific[22,33].
Metastatic colorectal carcinoma involving the ovary may closely mimic primary ovarian endometrioid or mucinous neoplasm[7,11,12,15,20,22,29,31,32,36,38]. Young et al introduced a classification of histological aspects of metastatic ovarian carcinomas. They observed the prevalence of glandular endometrioid-like pattern and mucinous-like pattern[11,15,22,31,32,36].
The following histological features are classic and characteristic of pseudoendometrioid metastases from large intestine; garland and cribriform histologic growth patterns, abundant intraluminal “dirty” necrosis, segmental destruction of glands and absence of mullerian features (squamous differentiation, adenofibromatous components or association with endometriosis)[11,15,22,30,32,36]. The garland pattern is typified by multiple large cystic glandular structures containing necrotic debris encircled by an array of rounded glands, often with segmental necrosis of their walls (Figure (Figure2).2). Typically, dirty necrosis consists of densely eosinophilic, coarsely granular necrotic debris containing abundant karyorrhectic material of sloughed carcinoma cells. Necrosis and intraluminal cellular debris also may occur in primary ovarian carcinoma[32,33] but often the intraluminal debris consists of thin secretions and degenerating neutrophils. Thus although dirty necrosis is characteristic of but not specific for colorectal adenocarcinoma, additional histologic features may be helpful in arriving at correct diagnosis. Classic cytological criteria such as marked cytologic atypia (2+ or 3+) and high mitotic index may be considered helpful in the diagnosis of colorectal ovarian metastases[15,22,29,36].
Other metastatic colorectal adenocarcinomas involving the ovary may mimic ovarian mucinous neoplasm. Some may be cystic, closely mimicking the gross appearance of a primary ovarian neoplasm[7,11,15,22,29,31]. Histologically there may also be a close resemblance to primary ovarian mucinous cystadenoma as well as obviously malignant areas with destructive stromal invasion. Analogous to the situation with other metastatic mucinous carcinomas, those morphologically bland foci (maturation phenomenon) have been erroneously interpreted as evidence of primary ovarian neoplasm[11,15,22]. In a recent study, frequent findings strongly favoring metastatic mucinous adenocarcinoma were; bilaterality, surface implants and an infiltrative pattern of stromal invasion. Findings that strongly favor primary ovarian mucinous carcinoma were; an “expansile” pattern of invasion and complex papillary pattern. Stromal luteinization that may occur with any mass lesion in the ovary appears to be more common in metastatic colorectal adenocarcinoma than other metastatic neoplasms. Rare adenocarcinomas metastatic from the intestine may contain cells with abundant clear cytoplasm, simulating either clear cell carcinoma or the secretory variant of endometrioid carcinoma. In a recent study the clinical and pathological features of 86 cases of metastatic colorectal adenocarcinoma involving the ovary were reviewed. Glandular and papillary architecture, “dirty necrosis”, desmoplasia, garland pattern, surface involvement, single infiltrative cells, extracellular mucin, “incomplete glands”, infiltrative nests of cells, cystic glandular dilatation, small glands, low malign potential -like areas, multimodularity and goblet cells were observed in decreasing frequency. In 19% of cases, foci with benign or low malignant potential appearance were seen. One point worthy of emphasis is the anecdotal experience of Hart who observed “microscopic nests of carcinoma within corpora lutea or corpora albicantia also points the metastasis”.
When the characteristic gross and microscopic distinguishing features are lacking between primary ovarian adenocarcinoma and metastatic colorectal adenocarcinoma, immunohistochemistry may be very useful[15,40-46]. Tumors with pseudoendometrioid histological pattern are most readily identified by immunophenotyping. However when the tumor is of mucinous type immunostains are less useful. This is due to the high frequency of intestinal differentiation in most primary ovarian mucinous neoplasms which results in considerable overlap in immunophenotype with metastatic mucinous neoplasms. Use of a panel of antibodies provides the most accurate immunophenotype and can usually assist in correct identification of the site of origin (Table (Table1).1). Based on the immunohistochemistry results obtained from recent studies a decision flow chart has been constructed (Figure (Figure33).
Use of coordinate expression of cytokeratins 7 and 20 (CK 7/20) for distinguishing primary ovarian tumors from colorectal metastases has been evaluated in large number of studies[39,42,45,46,48-51]. Combined use of CK 7 and 20 allows discrimination of most metastatic colorectal carcinoma from nonmucinous adenocarcinoma of the ovary.
Nonmucinous ovarian adenocarcinomas are almost always diffusely CK 7 positive and CK 20 negative whereas the majority of colorectal carcinomas are usually negative for CK 7 and uniformly positive for CK 20[42,45,46,48-50,52]. A caveat is that a small percentage of colorectal carcinomas, particularly those that are right-sided and high grade, have CK 7 positive and CK 20 negative immunophenotype[48,53]. Ovarian mucinous tumors are almost always positive for CK 7 but show variable positivity for CK 20 which is often patchy in distribution[42,45,46,48,49,54]. However, mucinous tumors arising in ovarian mature cystic teratomas which were morphologically and immunohistochemically familiar with gastrointestinal tract-type mucinous tumors were negative for CK 7 and positive for CK 20[51,52,55]. Intestinal-type mucinous epithelial neoplasm of low malignant potential, intraepithelial carcinomas and invasive adenocarcinomas more frequently had a CK 7-/CK 20+ phenotype (56%, 50% and 100%) respectively. A CK 7+/CK 20- phenotype was rare in these later 3 morphologic groups (6%).
Carcinoembryonic antigen (CEA) is a highly glycosylated cell surface protein overexpressed in a variety of tumors such as colorectal, ovarian, pancreatic, breast and nonsmall cell carcinomas. When evaluating ovarian tumors its greatest utility is in distinguishing metastatic colorectal carcinoma with pseudoendometrioid pattern from primary endometrioid carcinoma. These pseudoendometrioid metastatic tumors typically demonstrate strong and diffuse staining for CEA especially along the glycocalyceal border, apical cytoplasm or throughout the cytoplasm[20,58]. However several studies have demonstrated that CEA immunostaining is of no value in the differentiation between secondary ovarian tumors showing a mucinous pattern and primary ovarian mucinous adenocarcinomas because both show equally strong staining[20,54]. In recent studies 67%-85% of primary ovarian mucinous carcinomas were CEA positive compared to 95%-100% of metastatic colorectal carcinomas[44,50,54]. It is, however, possible to use CEA in combination with other tumor markers.
Cancer antigen (CA) 125 shows strong and diffuse staining of serous and endometrioid ovarian carcinomas with positivity ranging from 0-50% in mucinous tumors[40,50,54,59]. 4%-15 % of colorectal carcinomas are immunoreactive for CA 125 although staining may be weak and focal and the pattern may be cytoplasmic rather than membranous[15,50,54]. CA 125 may be helpful as a part of a diagnostic panel but its’ use as a single test in identifying ovarian adenocarcinoma is not reliable.
In recent studies, 100% of primary mucinous adenocarcinomas were shown to express MUC5AC whereas 0-33% of metastatic colorectal carcinomas expressed the same marker[45,60]. MUC2 reactivity was found in 90% of metastatic colorectal adenocarcinomas, in 70% of primary mucinous cytadenocarcinomas and all borderline tumors of intestinal type but in none of the cystadenomas or endocervical-like borderline tumors.
The CDX2 gene encodes an intestine-specific transcription factor belonging to the homeobox family that plays an important role in the regulation of proliferation and differentiation of intestinal epithelial cells. Diffuse and strong CDX2 protein expression is reported in almost all MCCO[41,43,54,59,62-64]. CDX2 expression is observed in up to 70% of primary ovarian tumors[41,54,62,64,65]. Gaggero et al
reported that all 47 endocervical-type mucinous cystadenomas stained negative for CDX2 and two of three intestinal-type cystadenomas stained positive. They concluded that the expression of CDX2 in mucinous tumors is likely to be dependent on the cell type (endocervical or intestinal). In view of the findings of these recent studies, positive CDX2 expression in a tumor involving the ovary should be interpreted with caution as although it may be a primary ovarian adenocarcinoma the possibility of metastatic tumor needs to be carefully excluded. Conversely, none-expression of CDX2 in a tumor would strongly support an ovarian primary.
A frequently observed genetic change in colorectal carcinoma is an inactivating mutation of the adenomatosis coli gene. This leads to the accumulation of the protein β-Catenin in the nucleus. In several recent studies, 59%-83% of MCCO showed nuclear expression of β-Catenin[41,44]. In one study, nuclear expression of β-Catenin was noted in only 9% of primary ovarian mucinous carcinomas. In Logani et al nuclear expression was found in 19% of 23 primary ovarian adenocarcinomas with only one tumor (5%), an endometrioid carcinoma, having a diffuse pattern of expression. Two endometrioid adenocarcinomas had strong nuclear immunoreactivity in foci of squamous metaplasia.
α-methyl-coenzyme a racemase
α-methyl-coenzyme a racemase, also known as P504S, is a mitochondrial and proximal enzyme involved in the metabolism of fatty acids. Overexpression of P504S has been observed in several tumors, most notably prostate and colorectal carcinoma[69,70]. Some authors have found that the frequency of P504S expression is decreased in poorly differentiated colonic adenocarcinomas[60,71]. In Logani’s study, 32% of MCCO showed diffuse expression of P504S as versus none in primary ovarian tumors. Although there seems to be value of using P504S in the differential diagnosis of primary ovarian and MCCO, its’ expression in primary ovarian tumors requires further evaluation.
In the current literature some controversy exist about Vimentin expression in mucinous carcinomas. Van Niekerk et al described rather high expression. Viale et al could only show it at a low level in some samples while Moll et al did not find it at all. In a recent publication, Vimentin was found to be substantially present (except for one sample) in colonic carcinomas but was present in only 18% of mucinous ovarian carcinomas. Therefore it combines a low sensitivity with high specificity.
Estrogen receptor (ER) and progesterone receptor (PR) expression in primary ovarian mucinous tumors and the utility of these markers for distinguishing MCCO from primary ovarian mucinous tumors have not been extensively investigated. In one study, all atypical proliferative mucinous tumors of gastrointestinal type, primary ovarian mucinous carcinomas and metastatic mucinous carcinomas (including 24 metastasis from colorectal primary) were negative for ER and PR with the exception of three metastatic endocervical adenocarcinomas. The authors concluded that immunohistochemical assesment of hormone receptor expression is of no value in distinguishing the common types of primary ovarian mucinous tumors from the vast majority of mucinous tumors metastatic to the ovary.
In premenopausal women ovaries have rich blood supply owing to both direct origin of ovarian arteries from the aorta and anastomosis through ovarian branches of the uterine arteries. Approximately ten arterial branches from this arcade penetrate the ovarian hilus, becoming markedly coiled and branched as they course through the medulla. It is also reported that ovarian metastases from primary colorectal cancer may involve haematogenous spread as the main pathway. This vascular-rich histology of the ovaries may be the main reason why colorectal tumor metastasis to the ovaries often expand rapidly, resulting in significant size increase compared to the mean size of the primary tumour. In Judson’s study, mean size of the colorectal tumor and ovarian metastasis were 5.1/5.0 cm and 12.8/14.1 cm in undiagnosed and known colorectal adenocarcinoma groups respectively. The large metastatic sizeis the main reason why patients need resection for this disease in spite of poor prognosis[24,76,77].
Ovarian metastasis are associated with advanced metastatic disease. In a report of 624 patients with colorectal adenocarcinoma, 19 (7.7%) had ovarian metastasis. They were divided into two groups according to the diagnostic time; Group A: syncronous (9 patients), Group B: metachronous (10 patients ). 5 year survival in group A was 16% while it was 0 in goup B. The authors concluded that resection of primary tumor plus bilateral oophorectomy is suitable for syncronous ovarian metastasis and as palliative treatment for metachronous disease. Miller et al reported on 23 patients with MCCO at the time of initial diagnosis. Surgical treatment consisted of colon resection in all but one patients, bilateral or unilateral salpingoophorectomy in 22 patients and hysterectomy in nine patients. Only one patient survived 5 years. Sixteen patients died of colon cancer. The median survival time was 17.8 mo, with a range from 1 to 86 mo. Tumor size was of no prognostic importance. Median survival time of patients with peritoneal disease (10.8 mo) was significantly shorter than for patients without peritoneal disease (25.2 mo). In the presence of liver metastasis , the median survival time was,likewise, significantly reduced from 20.1 to 8.1 mo. Some recent reports show that bilateral oophorectomy for MCCO has a good impact on disease-free and overall survival (OS) for patients with isolated ovarian metastasis. McCormick et al observed that patients with metastatic disease confined to the ovaries had a median OS of 61 mo (range 15-120 mo) compared to 17 mo (range 0.5-73 mo) for those with more extensive metastases (P = 0.0428). The observation that optimal cytoreduction was associated with prolonged progression-free survival and OS in both patients with localized ovarian and widespread metastases of colon cancer suggests a role for the management of metastatic colon cancer in women. Chung et al analysed the clinicopathological and follow-up data on 34 patients who underwent surgical resection of metastatic tumors originating from colorectal cancer. They concluded that surgical resection may be beneficial in selected patients with ovarian metastasis limited to the pelvis. Another report suggested that ovarian metastases are less responsive to chemotherapy compared to other sites and that surgical resection should always be considered for ovarian metastases even in the case of associated extraovarian metastases. In conclusion, macroscopic metastatic disease to the ovary is a poor prognostic factor in colon cancer. In selected patients who can be rendered disease-free by surgery, prolonged survival is possible and an aggresive aproach is recommended.
Controversies exist regarding the role of prophylactic oophorectomy for improving outcome following colorectal cancer surgery. While clinical series report rates of 1.2% to 14% of patients with colorectal carcinoma having ovarian metastasis[13-16] it is impossible to speculate how many, if any, could have been cured by oophorectomy on the premise that ovarian metastases were initially isolated and therefore treatable. These figures would suggest that a proportion of women with colorectal cancer would benefit from prophylactic oophorectomy. However recent studies have been unable to confirm this hypothesis[82-84]. Tentes et al divided the patients into two groups; in 70 of 124 (56.6 %) patients, the ovaries were preserved during surgery and in 54 (43.4%), synchronous prophylactic oophorectomy during primary tumor resection was performed. Seilezneff et al offered bilateral oophorectomy to all postmenopausal women in a consecutive series of 92 patients. Forty-one agreed to undergo oophorectomy. In both studies comparison of both groups revealed no difference in overall survival.
From 1.2% to 14% of women afflicted by colorectal adenocarcinoma are found to have ovarian involvement at some point in the course of their disease[13-16]. In many of these cases diagnosis of colorectal carcinoma has been established prior to the recognition of the ovarian lesion although in a minority of cases, ovarian mass is the initial manifestation of the disease[19,24]. In such cases preoperative differential diagnosis centers on the primary ovarian neoplasm. If the possibility of metastatic carcinoma is not raised either at the time of intraoperative consultation or in the final pathology report, clinical management may be adversely affected. Correct classification is important from both therapeutic and prognostic point of view. Unfortunately there is significant overlap between metastatic colorectal adenocarcinoma and those of primary epithelial ovarian neoplasms especially endometrioid and mucinous adenocarcinomas regarding the gross and histologic features[11,12,15,22,30,32,34-38]. Bilaterality, high-stage disease, multimodularity, surface implants, infiltrative pattern of invasion, invasion of hilar structures and vascular invasion are strong markers for metastatic ovarian tumors[11,33,35]. Prominent intraluminal dirty necrosis with a garland and cribriform pattern is characteristic of metastatic colorectal carcinomas. Features favoring primary ovarian endometrioid or mucinous neoplasm include; unilateral involvement, large size, an expansile pattern of invasion, complex papillary pattern and presence of Mullerian features[11,22,33]. A recently reported adjusted algorithm for mucinous carcinomas in which bilateral tumors and those unilateral tumors smaller than 12 cm are classified as metastatic has proved to be correct in 86% of cases[35,37].
Selected immunostains may be helpful in identifying MCCO. Tumors with a pseudoendometrioid histologic pattern are most readily identified by immunophenotyping. However when the tumor is of mucinous type, immunostains are less useful. A panel comprising of CDX2, β-Catenin and P504S is helpful in distinguishing primary mucinous or endometrioid adenocarcinoma from colorectal metastasis to the ovary in the majority of cases and is a useful adjunct to the already established role of differential staining with CK 7/CK 20, CA 125, CEA in this differential diagnosis[15,41].
Ovarian metastases are associated with advanced stage disease. Median survival time is 17.8 mo (1-86 mo) and 5 years survival rates are 16% and 0% in groups with oophorectomy and without oophorectomy respectively.
It is important to reemphasize that both gynecologist and pathologist should have a high level of suspicion of metastasis from another organ when they encounter a mucinous tumor in the ovary in order to prevent misdiagnosing a metastatic neoplasm as primary tumor.
Peer reviewer: Tsuyoshi Konishi, MD, PhD, Department of Gastroenterological Surgery, Cancer Institute Hospital, 3-10-6 Ariake, Koto-ku, Tokyo 135-8550, Japan
S- Editor Li LF L- Editor Hughes D E- Editor Yang C