Krukenberg tumor Robert H.Young Professor of Pathology Director of Anatomic Pathology Massachusetts General Hospital Harvard Medical School Boston, Massachusetts, U.S.A.

Clinical History: A 58-year-old woman with abdominal pain was found to have bilateral ovarian tumors with involvement of the bowel serosa. One ovary, from which slide obtained, measured 11 cm and the other 5 cm. The largest neoplasm had a mottled appearance with tan-yellow areas alternating with purple foci. The consistency varied from firm to soft. Figure 1 Click to view with ImageScope Click to view with a Web-Based Viewer Diagnosis: Krukenberg tumor Discussion: The designation Krukenberg tumor should be reserved for tumors with an appreciable component (arbitrarily defined as >10% of the tumor) of signet ring cells and no evidence of another specific diagnosis such as clear cell carcinoma or mucinous carcinoid, or any other primary neoplasm that rarely has signet ring cells. As recently emphasized [1] many Krukenburg tumors have a dominant microscopic appearance that is not characterized by signet-ring cells lying in a cellular stroma, the "textbook" morphology of the Krukenberg tumor. Summarizing the diverse spectrum of the Krukenberg tumor and contrasting it with the much rare metastases of intestinal-type gastric cancer [2] will be the focus of this case presentation. The frequency of the Krukenberg tumor varies with that of gastric carcinoma in the population and accordingly it is common in Japan but uncommon in North America and many other parts of the world. The average age of the patients is about 45 years. This obviously indicates that many patients are relatively young and accordingly, there should not be, as there sometimes is, reluctance to diagnose the tumor in the reproductive years. This age distribution is due to the relative youth of many patients with gastric signet-ring cell carcinomas and the greater vascularity of the ovary in young women, which facilitates hematogenous spread. The symptoms that patients with a Krukenberg tumor have are extremely variable. Most have symptoms related to ovarian involvement similar to that of any patient with ovarian cancer. Notably, the Krukenberg tumor is one of the ovarian tumors most often associated with stromal luteinization and endocrine manifestations as a result ("ovarian tumor with functioning stroma"). This is seen disproportionately in patients who are pregnant. This can be treacherous particularly if the Krukenberg tumor has a tubular pattern, the combined clinical and pathological picture resulting in the tumor being easily misconstrued as a Sertoli-Leydig cell tumor. This was a common error in a series of 13 tubular Krukenberg tumors reported some years ago [3] and is an ever present trap for the pathologist; this issue is considered further in the section on differential diagnosis. Although symptoms related to the ovarian tumors usually dominate (only 25-30% of patients are initially known to have a primary elsewhere) in some cases they are overshadowed by symptoms related to the primary tumor at its original site or at other sites of metastatic spread. Examples of the latter are as diverse as pulmonary symptoms due to pleural effusion or massive embolization of tumor in the lungs, to gastric symptoms not related to primary gastric cancer but rather breast cancer that has spread there as well as to the ovaries, to ureteral obstruction due to narrowing of the ureteral lumen due to metastasis, and pain due to bone metastases. Diffuse infiltration of the marrow may result in thrombocytopenia purpura, anemia and leukoerythroblastosis. In these various cases the ovarian tumors may be "incidental" findings found during evaluation of the varied symptoms. At operation ascites is common and liver metastases are rare. If not already known, the carcinoma that has resulted in the Krukenberg tumor is usually either found intraoperatively (particularly if a frozen section provokes a search for the primary) or becomes evident within the next six months. The primary neoplasm, however, is hard to find because of its small size in some cases and is not detected until 5 or more years after the metastatic tumor in the ovary in rare instances. About 76% of Krukenberg tumors originate in the stomach, 11% in the intestines (usually colon or rectum), 4% in the breast, 3% in the biliary system, 3% in the appendix, and the remaining 3% in miscellaneous sites such as pancreas, uterine cervix, urinary bladder (including urachus) and renal pelvis. Krukenberg tumors, at least 80% of which are bilateral (in some series all have been bilateral at least microscopically), are typically solid with external surfaces that are characteristically bosselated, sometimes strikingly so, and devoid of appreciable adhesions [1, 4, 5]. The features of their sectioned surfaces are quite variable. They vary from soft, even spongy or gelatinous, to firm, depending on the degree of edema or mucin spillage into the stroma, or stromal hypercellularity and fibrous transformation. The sectioned surface may be uniform but ill-defined nodularity or sometimes discrete nodules may be seen. The color is equally variable, white to yellow-tan to dusky red-purple with, in some cases, extensive hemorrhage. Cystic degeneration is occasionally seen and rarely thin-walled cysts are present. The gross appearance contrasts with the typically softer more necrotic appearance of typical metastatic colon cancer and the rare metastatic cancers from intestinal-type adenocarcinoma of the stomach [2]. The size is variable; most are over 5 cm., with an average size of about 10 cm, but they are typically not huge neoplasms, uncommonly exceeding 20 cm. Microscopic examination may show various quite different appearances as recently highlighted in what is the largest experience published to date [1]. The nature of the stroma and the prominence of signet ring cells are the first two variables that affect the morphology of the tumor. The stroma may be cellular, the feature that led to the misdiagnosis of the tumor in its first description as a "mucinous sarcoma." The cellularity is only rarely pronounced enough to provide a true sarcoma-like look despite the historically interesting trap this presented. In addition to densely cellular areas there may be regions with the cellularity of the usual ovarian fibroma, with in some cases a storiform pattern. In yet other cases there is a relatively acellular often edematous stroma which when it separates cellular areas may result in a pseudolobular appearance on low power examination. Finally, in some other cases the stromal reaction is unimpressive or absent and the profile is that of a glandular and solid epithelial neoplasm, albeit one, by definition, with signet-ring cells at least focally. Although mucin-filled signet-ring cells are a requirement for the diagnosis their frequency is variable. In many cases much of the tumor may be free of signet ring cells, or they are at least scant and not a striking feature such that a Krukenberg tumor diagnosis does not immediately come to mind. In other cases signet-ring cells are numerous and closely packed or pools of mucin may contain variable, sometimes innumerable, numbers of them. The signet-ring cell is such a typical cell of this tumor that it is sometimes forgotten that not only may it vary in appearance, but other cell types are usually present. The cytoplasm of the pathognomonic signet-ring cells is typically pale and vacuolated but occasionally it is dense and eosinophilic; the cytoplasm sometimes has a bull's-eye appearance, containing a large vacuole with a central eosinophilic body. Some mucin-rich cells have central rather than eccentric nuclei and small mucin-poor or mucin-negative cells with non-specific features are usually present. Rarely sheets of clear cells are present and exceptionally even squamous cells or cells with a transitional appearance are seen. The aforementioned stromal aspects and signet-ring cells represent the aspects generally emphasized when these tumors are reported but other features, which in aggregate are almost equally distinctive, are common. Gland differentiation is present in most of the tumors. The glands are typically small to medium sized and often have flattened cells that result in a rather characteristic microcystic appearance in many cases. When these glands lined by innocuous-appearing epithelium lie in a paracellular fibrous stroma an adenofibroma-like appearance results. Intestinal-type glands may also be seen but a picture reminiscent of metastatic colon cancer is in rare part because "dirty necrosis" is not a feature, with rare exceptions, of the glands within a Krukenberg tumor. Occasionally, small or large cysts lined by minimally atypical-appearing mucinous epithelium or completely flattened benign-appearing epithelium are present and merging with obvious neoplastic glands, which is almost invariably seen, helps to indicate that the cysts are due to the Krukenberg tumor, rather than being an independent primary tumor. When considering the diversity of morphology noted above it is important to reflect that many primary gastric signet-ring cell carcinomas have focal gland differentiation and indeed a sizable minority of gastric cancers are of mixed signet-ring cell (diffuse) and intestinal types. Accordingly, when such tumors spread to the ovary and fall in the Krukenberg group by having signet-ring cells, their glandular component is also reflected in the ovarian morphology. Small regular round acini, larger hollow tubules and elongated solid tubules are seen less often but treacherous when conspicuous, so-called tubular Krukenberg tumor. Entirely non-specific patterns of carcinoma such as growth as cords, masses, and single cells are present to at least a minor degree in most cases. Blood vessel and lymphatic invasion is common [4]. It is seen at the periphery of the ovary, in its substance in cases in which the parenchyma is not effaced, and particularly in the adjacent tissue such as around the ovarian hilus, the tube, and within the uterine wall. Another characteristic feature of metastases, surface involvement either observed grossly or microscopically, is less common than with other gastrointestinal metastases presumably because the spread to the ovary is dominantly hematogenous or by lymphatics, rather than transperitoneal, in cases of Krukenberg tumor. Lutein cells are often present in the stroma, particularly if the patient is pregnant. Some Krukenberg tumors, perhaps most often those that originate in the appendix, contain nests reminiscent of those seen in mucinous carcinoids but generally lack neuroendocrine cells. However, the latter may be seen but their focal presence in an otherwise typical Krukenberg tumor is acceptable as is their presence in diverse other neoplasms. Particularly when there is no known extra-ovarian primary the Krukenberg tumor causes numerous issues in differential diagnosis. For various reasons Sertoli-Leydig cell tumor is a frequent misdiagnosis. Causes of Confusion Between Krukenberg Tumor and Sertoli-Leydig Cell Tumor The tumors often occur in the third to fifth decades. Each tumor may be androgenic, the Krukenberg tumor being one of the commonest "non-endocrine tumors" of the ovary with this feature (so-called ovarian tumor with functioning stroma). Both may be soft and edematous on gross inspection. Both tumors may have a striking pseudolobular pattern on low power. Both may have a striking tubular pattern. Signet-ring cells may be seen in SLCTs (those of heterologous type with small nests of mucinous carcinoid). Both tumors have usually morphologically indistinguishable lutein-Leydig cells in their stroma, (only the finding of crystals of Reinke in such cells distinguishes them and they are rarely found). The above similarities not withstanding, the pathologic differences between the two tumors are such that distinguishing them, particularly given good sampling, should not be difficult. Although, as noted above, they may be grossly similar in some cases and microscopy is ultimately definitive, the solid, fibroma-like gross appearance of some Krukenberg tumors is almost unheard of for a Sertoli-Leydig tumor. Alternatively, although some Krukenberg tumors may be yellow, at least in part, the striking uniform yellow cut surface of many SLCTs would be most unusual for a Krukenberg tumor. The signet-ring cells in SLCTs invariably occur in association with neuroendocrine cells in clusters that typify the mucinous carcinoid component of occasional heterologous SLCTs. Signet-ring cells in most Krukenberg tumors have a much more random distribution than those in the typical little clusters of cells that typify the mucinous carcinoid elements just noted. Many Krukenberg tumors have at least minor foci of non-specific adenocarcinoma inconsistent with a diagnosis of Sertoli-Leydig cell tumor. Cytologic atypia is generally more striking in Krukenberg tumors although some SLCTs have bizarre nuclei of degenerative type. Two stromal tumors with signet-ring-like cells may cause confusion. The sclerosing stromal tumor may contain cells resembling signet cells as well as a proliferating fibroblastic component, but the former cells contain lipid rather than mucin. The diagnosis should be readily made on high-power microscopic examination, especially with the aid of mucin stains, which are negative. Holtz and Hart [4] rightly emphasized the potential for Krukenberg tumors with a prominent relatively acellular fibroblastic stroma to be misdiagnosed as a fibroma, even on permanent sections, and example of this error can be found in the older literature. This can be particularly treacherous at the time of frozen section particularly given gross similarities between the two neoplasms. The rare signet-ring stromal tumor also may enter the differential diagnosis but its cells also are negative with mucin stains. The unusual luteinized thecomas that are associated with sclerosis peritonitis have a cellular stromal component similar to the cellular stroma of some Krukenberg tumors but they lack signet-ring cells. Surface epithelial carcinomas can be in the differential with a Krukenberg tumor [2, 5], albeit rarely. Clear cell carcinoma has small tubules in the tubulo-cystic pattern of that neoplasm which are to a degree simulated by the "microcystic" tubules of Krukenberg tumors. Additionally, rare clear cell carcinomas have signet-ring cells. In the clear cell carcinoma, mucin, when present, is typically luminal and extracellular. In portions of the tumor that contain signet-ring cells the presence of other characteristic features of the tumor, such as papillae (never seen in Krukenberg tumors) permit its recognition. To complicate the matter, however, misdiagnosing a Krukenberg tumor as clear cell carcinoma is possible when, rarely, a Krukenberg tumor contains numerous clear cells. The co-existence of typical foci of Krukenberg tumor is diagnostic. Primary mucinous tumors may contain signet-ring cells but rarely in great number and confusion with a Krukenberg tumor is almost never a realistic problem. Even when the latter contains mucinous cysts, non-cystic solid areas of typical Krukenberg tumor are present. High-grade non-cystic carcinoma in mucinous cystic tumors is usually conventional non-signet ring cell mucinous adenocarcinoma or in some cases anaplastic carcinoma with differing morphology (large cells with abundant eosinophilic cytoplasm) from that of a Krukenberg tumor. Rarely serous carcinomas and less often endometrioid and undifferentiated carcinoma can have signet-ring cells as documented by Che et al [6]. In their 15 cases they were diffusely distributed in nine and focal in six. They were all seen in tumors with microcystic patterns. All the cases had other patters of surface epithelial neoplasia when well sampled and lacked stromal and other features of a Krukenberg tumor. In our experience a microcystic pattern of the type described by Che et al. is commonest (in the surface epithelial family) in transitional cell carcinoma but we have not been struck to date by the presence of signet cells. Primary mucinous carcinoid tumors that contain large numbers of signet-ring cells are distinguished from Krukenberg tumors by the distinctive morphology of the former in differentiated areas and presence of plentiful neuroendocrine cells, which can be confirmed by special stains. It is likely that at least some of the primary Krukenberg tumors reported are high-grade primary mucinous carcinoids. The majority of metastatic mucinous carcinoids to the ovary reported (perhaps all of them) in my opinion are best placed in the category of a Krukenberg tumors so the differential of metastatic mucinous carcinoid with a Krukenberg tumor is moot. Mesothelial neoplasms may be a rare diagnostic consideration. The small vacuoles of an adenomatoid tumor [7] may mimic signet-ring cells but always are associated with larger, more typical, adenomatoid formations producing an overall distinctive morphology in aggregate markedly diffuse from that of a Krukenberg tumor. Malignant mesothelioma may involve the ovary and although signet-ring-like foci may be seen, other patterns, particularly the tubulo-papillary pattern of mesothelioma should be sufficiently distinctive as to solve any problem that might arise. I will conclude by making a few remarks on the other variant of ovarian metastasis of gastric cancer, that of intestinal type, which is uncommon and has been the subject of limited coverage in the literature. A recent small series of four cases [2] does, however, point out differences between this ovarian metastatic neoplasm and the Krukenberg tumor. The intestinal-type metastases are typically seen in patients older than those with Krukenberg tumors and are usually seen in patients with a known gastric carcinoma in contrast to the situation in many cases of Krukenberg tumor. Grossly the resemblance is to the familiar metastatic colorectal carcinoma and on microscopic examination some cases have the familiar pseudoendometrioid morphology of ovarian spread of that cancer. Indeed, two of the four cases I recently reported with Dr. Melinda Lerwill had such an appearance and their differential diagnosis is similar to that which pertains to metastatic colorectal carcinoma versus primary endometrioid carcinoma. The remaining two cases in our series were more reminiscent of mucinous primary neoplasms and were distinguished from them using conventional criteria for distinction of primary and metastatic mucinous neoplasia, including of course due consideration of the history and overall clinical background. It should be noted that inevitably there is some overlap inasmuch as just as Krukenberg tumors may show gland differentiation as recently emphasized [1], so some metastatic intestinal adenocarcinomas in the ovary may have a minor component of signet ring cells. Nonetheless, the overall morphology is markedly different in the two variants of ovarian spread of gastric cancer. References Kiyokawa T, Young RH, Scully RE. Krukenberg tumors of the ovary. A clinicopathologic analysis of 120 cases with emphasis on their variable pathologic manifestations. Am J Surg Pathol 30:277-299, 2006. Lerwill MF, Young RH. Ovarian metastases of intestinal-type gastric carcinoma. A clinicopathologic study of 4 cases with contrasting features to those of the Krukenberg tumor. Am J Surg Pathol. In press. Bullon A, Arseneau J, Prat J, Young RH, Scully RE. Tubular Krukenberg tumor. A problem in histopathologic diagnosis. Am J Surg Pathol 5: 225-232, 1981. Holtz F, Hart WR. Krukenberg tumors of the ovary. A clinicopathologic analysis of 27 cases. Cancer 50: 2438-2447, 1982. Hart WR. Diagnostic challenge of secondary (metastatic) ovarian tumors simulating primary endometrioid and mucinous neoplasms. Pathology International 55:321-243, 2005. Che M, Tornos C, Deavers MT, Malpica A, Gershenson D, Silva EG. Ovarian mixed-epithelial carcinomas with a microscopic pattern and signet-ring cells. Int J Gynecol Pathol 20:333-328, 2001. Phillips V, McCluggage WG, Young RH. Oxyphilic adenomatoid tumour of the ovary: a case report with discussion of the differential diagnosis of ovarian tumours with vacuoles and related spaces. Int J Gynecol Pathol In press.