Microscopic Benign and Invasive Malignant Neoplasms and a Cancer ...

ARTICLES Microscopic Benign and Invasive Malignant Neoplasms and a Cancer-Prone Phenotype in Prophylactic Oophorectomies Hernando Salazar, Andrew K. Godwin, Mary B. Daly, Paul B. Laub, W. Michael Hogan, Norman Rosenblum, Matthew P. Boente, Henry T. Lynch, Thomas C. Hamilton*

Background: The occurrence of approximately 5% of common epithelial malignant tumors of the ovary can be traced to inheritance of risk. One prophylactic strategy to decrease the probability of development of disease in individuals within families where this mendelian-dominant pattern of occurrence is apparent is to remove the ovaries of individuals at risk for ovarian cancer. The procedure, when done for this purpose, is recommended soon after completion of childbearing. Purpose: Our goal was to compare the histologic features of the ovaries of women at increased risk for ovarian cancer to those at no known increased risk for the disease. Methods: Ovaries removed for prophylaxis from 20 women considered to be at increased risk for developing ovarian cancer were examined histologically. During the course of this work, it seemed apparent that these ovaries contained numerous atypical features compared with the expected appearance of normal ovaries. Hence, we expanded the study to include a control group whose ovaries were removed for reasons unrelated to cancer. The study, therefore, was not blinded. The increased risk in the cancer-prone individuals was determined by family history, specifically the presence of at least one first-degree relative and one second-degree relative with ovarian and/or breast cancer and positive linkage or mutational analysis of BRCA1 in some. The difference in mean ages of patients in the control and high-risk groups was not statistically significant The difference among both groups with respect to the number of atypical features as well as the intensity of those features was ascertained by computing probabilities using Fisher's exact test (two-sided) for rows x columns contingency tables. Results: Two unanticipated microscopic or near-microscopic malignant neoplasms and other benign and borderline tumors were discovered in the ovaries of the high-risk individuals. Of substantial interest was the finding that among the ovaries of high-risk women, 85% presented two or more and 75% presented three or more of the following histologic features: surface epithelial pseudostratification; surface 1810 ARTICLES

papfllomatosis; deep cortical invaginatioas of the surface epithelium, frequently with multiple papillary projections within small cystic spaces (microscopic papillary cystadenomas); epithelial inclusion cysts, frequently with epithelial hyperplasia and papillary formations; cortical stromal hyperplasia and hyperthecosis; increased folUcular activity; corpus luteum hyperplasia; or hilar cell hyperplasia. Two or more or three or more such changes were observed in a lesser percentage (30% or 10%, respectively) of ovaries obtained from the control individuals, with a statistically significant difference (P = .001 or P = .00007, respectively), particularly considering that a detailed determination of a family history of cancer in the control group was not possible. Conclusions: The frequency of these changes in the high-risk ovaries compared with control ovaries suggests a characteristic histologic preneoplastic phenotype defined by an increased frequency and intensity of the above-described histologic features in the high-risk ovaries. Limited access to numerous small (stage I) ovarian cancers or cancer-prone ovaries by any one pathologist may explain the failure to identify the phenotype in the past Implications: We suggest that the ovaries removed from ovarian cancer-prone individuals as a preventative measure should be thoroughly examined histologically to identify or rule out microscopic or near-microscopic invasive neoplasms. [J Natl Cancer Inst 1996; 88:1810-20] It has been estimated that in 1995 approximately 27 000 women in the United States were diagnosed and 15 000 died as

*Affiliations of authors: A. K. Godwin, M. B. DaJy, P. B. Laub, W. M. Hogan, M. P. Boente, T. C. Hamilton, Fox Chase Cancer Center, Philadelphia, PA; H. Salazar, N. Rosenblum, The Reading Hospital Medical Center, PA; H. T. Lynch, Creighton University School of Medicine, Omaha, NE. Correspondence to: Thomas C. Hamilton, Ph.D., Fox Chase Cancer Center, 7701 Burholme Ave., Philadelphia, PA 19111, or Hemando Salazar, M.D., M.P.H., Department of Pathology, Reading Hospital Medical Center, Sixth Ave. and Spruce St., Reading, PA 19611. See "Notes" section following "References."

Journal of the National Cancer Institute, Vol. 88, No. 24, December 18, 1996

a result of complications associated with ovarian cancer (1). The disease also accounts for a similar number of deaths in the European community (2). Hence, ovarian cancer is an important contributor to cancer mortality in the industrialized western world (3). The vast majority (-90%) of the malignant ovarian neoplasms are classified as common epithelial or epithelialstromal tumors of the ovary (4,5). Histopathologic observation strongly suggests that these tumors arise from the specialized mesothelial (celomic) cells that cover the ovarian surface (6-8), i.e., the surface epithelium. A study (9) showing that spontaneously transformed rat ovarian surface epithelial cells form tumors histologically similar to clinical ovarian cancer adds experimental support to this concept. The tumors that arise from the surface epithelium have histologic patterns resembling those of the various parts of the Miillerian genital system, also of celomic origin. Although there has been much speculation as to why the surface epithelial cells are the ovarian cells that most frequently undergo malignant transformation (10-17), in reality we are far from an unequivocal answer to this question or an understanding as to which genetic alterations are at the root of the process. Microscopic evaluation has revealed a spectrum of histologically similar ovarian neoplasms that range from overtly benign to those that have been found to have low malignant potential and to others that are clearly malignant (18,19). It has been rare, however, to see transition to cancer within the former two lesion types (7,18,20,21). This lack of evidence of clear progression from benign to malignant has caused some speculation as to whether there are precursor lesions for ovarian cancer (22) or, alternatively, if the disease arises de novo as a cancer with perhaps only a brief premalignant or subclinical phase. Support for the concept that there may be histologically identifiable entities that precede overt cancer comes from the histologic evaluation of polycystic ovaries, ovaries of individuals with cancer in the contralateral ovary, and those with endometrial cancer (12,23,24). At least some of these individuals may be at increased ovarian cancer risk, and the studies have shown changes in the features and growth pattern of the surface epithelial cells in their ovaries. In this study, we have compared the histologic features of ovaries from individuals with a high probability of developing ovarian cancer with ovaries from individuals with no known increased risk for the disease. Although our study was not blinded, we suggest that the features of the cancer-prone ovaries may create an identifiable milieu from which common epithelial tumors of the ovary will most likely arise.

Materials and Methods

covery within the patient history portion of the pathology report of an indication that the individual was a member of a cancer-prone kindred. Another group of ovaries from the Fox Chase Cancer Center and several other institutions throughout the United States consisted of those removed for prophylaxis from 20 women (mean age, 42.1 years) as a measure to help minimize the development of ovanan cancer because they were considered at increased risk. No other consideration, e.g., the presence of any specific histopathologic characteristics, influenced in any way their inclusion in the study. The assessment of increased risk is defined below. The difference in the mean ages of these two groups is not significantly different (P - .23; two-sided Student's / test).

Assessment for Increased Risk of Developing Ovarian Cancer The minimum criterion for inclusion of patients in the high-risk category was that the woman had at least one first-degree relative and one second-degree relative with breast and/or ovarian cancers. In some cases, individuals were determined to be obligate BRCA1 mutant allele carriers on the basis of linkage analysis (25,26) or mutational studies (25,27).

Linkage Analysis To establish the probability that a family was linked to BRCA1, linkage analysis was performed using five polymorphic CA-repeat markers flanking the BRCA1 gene, including the intragenic marker D17S855. The other markers were D17S250, D17S588, D17S579, and THRA1. Multipoint linkage analysis was performed with the use of the LINKAGE program under a previously described model (25,26).

Single-Strand Conformational Polymorphism (SSCP) Analysis Gcnomic DNA was isolated from ovary specimens and/or blood samples. DNA was amplified by polymerase chain reaction (PCR) with the use of the pnmer sequences described previously (27). PCR was carried out in a reaction volume of 10 |lL containing 100 ng of genomic DNA as template, 10 mM TnsHC1 (pH 8.3), 50 nW KC1, 1.5 mM MgCl2, 0.001% gelatin, 1 [iM of both the forward and reverse primer, 60 |KWeach of deoxyguanosine tnphosphate, deoxycytidine tnphosphate, and thymidine tnphosphate, and 12 ji/W deoxyadenosine triphosphate (dATP), 1.0 uCi [a-32P]dATP (Du Pont, NEN, Boston, MA), 5% dimethyl sulfoxide, and 0.5 U Amplitaq DNA polymerase (The Perkin-Elmer Corp., Foster City, CA). Following an initial denaturation step at 94 'C for 4 minutes, DNA was amplified through 20 cycles consisting of a 5-second denaturing at 94 "C, a 1-minute annealing at 65 'C-0.5 "C per cycle, and a 1minute extension at 72 *C. The samples were then subjected to an additional 25 cycles, consisting of a 5-second denaturation at 94 "C, a 1-minute annealing at 55 'C, a 1-minute extension at 72 'C, and a final extension at 72 'C for 5 minutes. Annealing temperatures were optimized for each pnmer pair. Because the size of PCR products from exon 20 were too large to perform SSCP analysis with high efficacy, products were directly sequenced as described below. PCR reaction products were diluted 1:10 in denaturing loading dye (95% formamide, 10 mM NaOH, 0.25% bromophenol blue, and 0.25% xylene cyanol), heated at 94 "C for 5 minutes, and flash-cooled on ice. Four microliters were loaded onto a 0.5 x MDE gel (AT Biochem, Malvem, PA), prepared according to the manufacturer's specifications, and run at 6 W for 12-16 hours at room temperature in 0.6 x TBE (lx = 0.09 M Tris, 0.09 M boric acid, and 0.002 M EDTA). Following electrophoresis, the gel was dried on a vacuum gel dryer and exposed to autoradiography film at -80 'C for 4-12 hours. Variant and normal SSCP bands were cut out from the gels after alignment with the autoradiograph, and the DNA was eluted in 100 |iL of double-distilled H2O at 37 "C for 3 hours.

Specimens The ovaries from 20 women (mean age, 44.7 years) who underwent total abdominal hysterectomy/bilateral salpingo oophorectomy at the Lankenau Hospital, Wynnewood, PA, for various reasons, without any ovarian or uterine cancers, served as the control specimens for this study. The control subjects were selected subsequently to the selection of cancer-prone individuals described below to approximate the mean age of this latter group. These control subjects included 17 cases of leiomyomatosis (two with accompanying menorrhagia [cases 3 and 13]), two cases of adenomyosis (cases 6 and 17, the latter with menorrhagia), and one case of endometriosis (case 8). The only basis for exclusion of one individual from those submitted as control subjects was the dis-

DNA Sequencing of Variant SSCP Bands Variant SSCP bands were excised, eluted, and amplified by PCR, and the product separated from primers with the use of Wizard resin (Promega Corp., Madison, WI) according to the manufacturer's specifications. The purified DNA was subjected to cycle sequencing with the use of an automated fluorescencebased cycle sequencer (Model 373A Automated Sequencer, Applied Biosystems, Foster City, CA) and Taq dye terminator chemistry. Sequencing primers were the same as those used to amplify the template. Each mutant allele carrier was tested by SSCP and direct sequencing a second time on a fresh sample to verify the mutation.


ARTICLES 18J1

Histopathologic Analysis Ovaries were processed in serial transverse sections 2-3 mm thick, fixed in buffered 10% formalin, embedded in paraffin, sectioned at 4-5 |im thick, and stained with standard hematoxylin—eosin stain. Examination of specimens was not blinded. Every case included the right and left ovaries, and all the sections from each ovary were examined histologically.

Immunohistochemical Detection of Estrogen Receptors and Keratin Detection of estrogen receptors was done in some cases by immunohistochemistry with the use of a mouse monoclonal antibody (1D5) directed against the human estrogen receptor (Dako Corp., Carpinteria, CA). Incubation was overnight at a 1:400 dilution (4 'C). Visualization of the primary antibody was with avidin-biotin technology (Vector Laboratories, Burlingame, CA), as directed by the manufacturer. Briefly, sections were washed and incubated for 30 minutes at room temperature with biotinylated secondary antibody. This was followed by washing, incubation (30 minutes at room temperature) with the Vectastain avidin-biotin complex reagent, and washing as specified by the manufacturer. Visualization was accomplished by incubation in peroxidase substrate, i.e., diaminobenzidine tetrahydrochloride (as per manufacturer's instructions) (2-10 minutes). Washing in all cases was done two times, with 5 minutes per wash in phosphate-buffered saline at room temperature. Before viewing, the sections were counterstained with Gill's hematoxylin No. 3 stain. In the case of analysis of keratin, the primary antibody was polyclonal pancytokeratin (BioGenex Laboratories, San Ramon, CA) used directly as supplied. Incubations and visualization were as described for the estrogen receptor.

Statistical Analysis Probabilities were computed by contingency table analysis (28). Risk status, that is, membership in either the control or high-nsk groups, forms columns. Intensity of features (ie., -, +, ++, or +++) or the number of positive features compose rows. The null hypothesis of the independence of features (i.e., the rows) and risk status (i.e., the columns) was tested. Consequently, P values of less than .05 suggest statistically significant association. Conventionally, probability is computed from the chi-squared statistic measuring the difference between observations and the corresponding values expected by the null hypothesis. Because expected frequencies of less than 5 were present in some cells in many of the contingency tables, estimation of probability was instead computed with the use of Fisher's exact test for rows and columns (28,29). The statistical computations were performed by the use of the S-Plus 3.1 statistical software package (Statistical Sciences, Seattle, WA). All statistical tests were two-sided.

Results This study was based on the histopathologic analysis of two groups of ovaries: 1) those from individuals assessed to be at increased risk of ovarian cancer and who underwent oophorectomy for prophylaxis and 2) those from individuals with no known increased risk of the disease. Within each group, the two ovaries from each woman revealed similar types of changes, with some variation in magnitude and extent. In the cases with invasive tumors, which were present in one ovary only, the contralateral ovary also presented features characteristic of the cancer-prone phenotype, as described below. In the control subjects, the sporadic changes were similar in both ovaries. The laterality of the specific changes was not specifically documented. Characteristics That Defined Individuals as Ovarian Cancer Prone Table 1 summarizes the documented cases of ovarian and breast cancers within the families of the women from whom this 1812 ARTICLES

first group of ovaries was removed. As can be seen in Table 1, the frequency of ovarian cancers in these families varies from 1 to 11 and of breast cancer from 0 to 22. It should be noted, however, that the size of the pedigrees and relative capacity to document cases is variable between families. Additionally, Table 1 shows that, of the 20 individuals in the increased-risk category, seven have been documented to have inherited an increased risk of developing ovarian and/or breast cancer on the basis of linkage analysis or mutational studies, and six already have breast cancer. In the cases where mutations or linkage is known, ovarian DNA has also been examined and found to be heterozygous for the BRCA1 alleles (data not shown). The histopathologic findings in the ovaries of the cancerprone individuals included two microscopic or near-microscopic invasive common epithelial tumors (Table 2), which were not noted prior to or during the surgery. Histopathologic Features of Ovaries From Case Patient 2-1 The first case of cancer involved the right ovary of the case patient 2-1. A poorly differentiated invasive papillary serous adenocarcinoma measuring 0.6 cm in maximum diameter was uncovered in this ovary. It was not anticipated clinically nor was it noted during surgery (laparoscopic salpingo oophorectomy) or by the gross pathology prosector. It was observed only during the microscopic examination of the specimen. The tumor was located in the cortex of the ovary and was invasive. It was formed by several closely related nests of glands and papillary strands of pleomorphic columnar and cuboidal epithelial cells with few mitoses. These structures penetrated deeply, infiltrating the ovarian cortical stroma and eliciting a fibroblastic and lymphocytic reaction (Fig. 1, A and B). No psammoma bodies were seen. The tumor cells were large, with irregular nuclei, distinct nucleoli, and a high nucleus/cytoplasmic ratio. They showed a reduced expression of keratin compared with the surface epithelial cells in the specimen and were positive for nuclear estrogen receptors (data not shown). In close proximity, i.e., superficial to the malignant tumor, was an apparently distinct microscopically detected cystic papillary tumor, measuring about 0.3 cm in diameter, that presented all of the characteristics of a tumor of low malignant potential or "borderline" (Fig. 1, C and D). It was composed of a cystic cavity at the surface of the cortex that was lined by a layer of columnar pseudostratified epithelium with intercalated "peg" cells. The interior of the cyst was studded by numerous slender papillary excrescences composed of fibrovascular cores covered by hyperplastic serous epithelium forming up to three layers of cells. Cellular tufts, with cellular atypia in the form of nuclear pleomorphism, variation in polarity, irregular nuclear contours, enlarged nucleoli, and very rare mitoses, were also seen. There was, however, no evidence of direct stroma] invasion by the epithelium of this tumor. Although the adjacent invasive tumor was in close proximity to this well-circumscribed cystic structure, it was deeper in the cortex, and there was no apparent continuity between the epithelial elements of the two structures. There was an additional superficial microscopic cystic structure of similar size in this ovary, but it was lined by a simple pseudostratified columnar epithelium with a serous pattern of differentiation. Within the cyst were broad papillae with dense


Table 1. Family history of ovarian and breast cancers Family 1 2 3 4 5 6 6 7 8 9 10 11 12 12 13 14 15 16 17 18

Individual

No. of breast cancers* (mean age at diagnosis, y)

No. of ovarian cancers* (mean age at diagnosis, y)

Age, yt

11 (44) 2(34) 9(52) 3(56)

1(47) 4(44) 3(43) 2(49) 2(57) 3(48) 3(48) 11(51) 5(53) 3(69) 2(42) 5(56) 3(55) 3(55) 3(50) 6(54) 7(52) 4(66) 2(60) 4(45)

66 50§ 29 4411 35 44 441 38# 46 41 51 42** 41 37*. 34 41 39 52tt 38 29

0 ;>

5(35) 5(35) 22 (43) 6(39) 0

>

2(68) 10(46) 20(46) 20(46) 6(49) 3(40) 4(55) 8(44) 1(68) 7(46)

BRCA1 statust + N.D. + N.D. N.D. + + ++ + N.D. N.D. N.D. + N.D. N.D. N.D. N.D. N.D. N.D.

•Number of verified cases of breast and ovarian cancers in the family. A patient with both breast and ovarian cancers was counted in each total. Bilateral breast cancers were counted as two separate cancers. tAge of the individual at the lime oophorectomy was performed for prophylaxis. $+ = that the individual is an obligate mutant allele carrier based on mutation studies; ++ = that the individual is an obligate mutant allele carrier based on linkage studies; - = not linked to BRCA1 locus; N.D. = not determined. §Individual was diagnosed with ovarian cancer at age 50 years. Illndividual was diagnosed with breast cancer at age 40 years. Individual was diagnosed with breast cancer at age 33 years and ovarian cancer at age 44 years. individual was diagnosed with breast cancer at age 38 years. ••Individual was diagnosed with breast cancer at age 36 years, ttlndividual was diagnosed with bilateral breast cancer at age 44 years.

fibrovascular cores that were also covered by a single layer of pseudostratified-tall epithelial cells, similar to those lining the wall of the cyst. There was no hyperplastic proliferation or piling up of the epithelium, no tufts, no atypia, and no evidence of invasion (Fig. 1, E and F). As noted in Table 2, this specimen also contained other histologic features not typical of a normal ovary. Histopathologic Features of Ovaries From Case Patient 6-2 The second invasive adenocarcinoma was found in the left ovary of individual 2 in family 6 (case patient 6-2). Similar to the previously described case, this tumor was not identified clinically or at the time of surgery. It was recognized by the pathologist on gross examination. This tumor was classified as an invasive endometrioid adenocarcinoma that measured 1.0 cm in diameter and involved the ovarian cortex close to the surface. It was formed by numerous well-differentiated glands, closely packed with a few papillary structures but mainly exhibiting features consistent with the endometrioid designation. It invaded the cortical stroma and presented focal necrosis as well as a desmoplastic and inflammatory lymphocytic reaction. The tumor cells were arranged in single or double layers forming cords and glandular structures and presented angular contours with scant eosinophilic cytoplasm and large ovoid or irregular nuclei with prominent nucleoli. A moderate number of mitotic figures was observed (Fig. 1, G). There was a relatively well-circumscribed, firm, white-tan benign Brenner's tumor measuring approxi-

mately 0.7 cm in maximum diameter adjacent to this invasive endometrioid carcinoma and also involving the cortex. This synchronous lesion was also unrecognized until gross pathologic examination. The tumor was composed of multiple distinct islands or nests of epithelial neoplastic cells, usually solid but some containing minute cystic spaces. The cells in each island were closely packed against each other in a pseudostratified transitional manner, each cell having a polygonal or spindly elongated shape with pale eosinophilic cytoplasm and a small rounded nucleus, sometimes with a discrete nucleolus. The stroma was dense and cellular around the epithelial nests, forming a distinct nodule (Fig. 1, H). Other Atypical Histologic Features Commonly Seen in Ovaries of Cancer-Prone Individuals The cancer-prone ovaries, in addition to containing occasional unanticipated microscopic benign, borderline, or malignant tumors, also contained a range of histologic features not usually seen in such magnitude, combination, and complexity in ovaries of control women (Tables 2 and 3; Figs. 2 and 3). A significant number of cases (70% of high-risk ovaries versus 20% of the control group ovaries, P = .005; Table 4) presented multifocal surface papillomatosis, ranging from a few to markedly extensive foci (Fig. 2, A). These projections were usually short and stubby, with a fibrous core covered by cuboidal or pseudostratified epithelium (Fig. 2, A). These superficial papillae are similar to those commonly seen in association with deep-surface


ARTICLES 1813

Table 2. Histopathologic findings in cancer-prone ovaries*,t Tumor

Case No.

Malignant invasive

Non-neoplastic changes

LMP

Papillomatosis

Berugn

Inclusion cysts

Invaginations

Epithelial pseudostratification

Strom al activity

1-1 2-1

Papillary serous

Papillary serous

Papillary serous cystadenoma

3-1 4-1 5-1

Mature teratoma; microscopic papillary serous cystadenoma

6-1 6-2

Endometnoid

Brenner's tumor

7-1 8-1 9-1 10-1

Microscopic papillary serous cystadenoma; cystadenofibroma

11-1 12-1 12-2 13-1

Microscopic papillary serous cystadenoma

14-1 15-1 16-1 17-1 18-1 * 1 st number designates a family and the second, an individual. tLMP = low malignant potential; + = presence; - = absence.

invaginations (Fig. 2, B). The features of these ovaries, including papillomatosis and deep invaginations, often resulted in lesions that, except for size, fit the criteria to be designated as benign papillary cystadenomas (Fig. 2, C and D). For example, a frequently observed change, seen in 19 of 20 high-risk ovaries (Table 2), was the presence of invaginations of the surface epithelial cells into the stroma, usually associated with short papillary excrescences (Fig. 2, B). These invaginations were often very deep into the cortex, sometimes with bifurcations or branches (Fig. 2, B). Similar but fewer and less-deep invaginations were also observed in 11 of 20 control ovaries. The difference is statistically significant {P = .0004; Table 4). In the cancer-prone ovaries, the minute papillary cystic lesions were often formed at the end of the deep invaginations of the surface epithelium, with distention of the deepest portion and papillary proliferation within the lumen (Fig. 2, B-D). The papillae were 1814 ARTICLES

of various sizes and presented with a fibrovascular core lined by a single layer of columnar or cuboidal epithelial cells but did not show hyperplasia, tufting, atypia, or any suggestion of invasiveness. Furthermore, as suggested above, the epithelium often could be seen to be continuous, via the deep invaginations, with the cells covering the ovarian surface. Another frequent finding in the cancer-prone ovaries was the presence of cortical superficial inclusion cysts in 70% of highrisk case patients against 25% of control subjects (P = .006; Table 4). The cysts had the appearance of "disturbed glands." They were of variable size and shape, usually lined by a pseudostratified epithelium of serous or tubal type and usually devoid of any contents within the space (Fig. 2, E and F). In some cases, groups of the cysts of varying complexity occurred, creating the appearance of microscopic cystadenomas or adenofibromas (Fig. 2, F-H). Occasionally, a solid epithelial


Fig. 1. A) Microscopic papillary serous cystadenocarcinoma, (case patient 2-1), 0.6 cm in maximum diameter. Note papillary and invasive character of the tumor infiltrating the fibrous stroma, eliciting a lymphocytic and dcsmoplastic reaction (hematoxylin-eosin [H & E], original magnification x4O). B) Higher-power view of same tumor in A. Note the pseudostratified serous type of epithelium, with pleomorphic cellula-ity and mitotic activity, invading the stroma (H & E, original magnification x400). C) Papillary serous cystadenoma, "borderline" type, located at the surface of the cortex, exhibiting papillae with fibrovascular cores and hyperplastic epithelium but the absence of stromal invasion (H & E, original magnification x40). D) Higher-power view of "borderline" tumor in C. Note the papillary fibrovascular stroma and the hyperplastic serous epithelium forming multiple layers and epithelial tufts with cellular atypia (H & E, original magnification x250). E) Papillary serous cystadenoma. This microscopic tumor was adjacent to the previous tumors and measured 0.3 cm in diameter. It is formed by broad papillae lined by a single layer of pseudostratified serous epithelium without piling up or tufting (H & E, original magnification xlOO). F) Higher-power view of same tumor in E. Note the single layer of serous epithelium lining the papillae and a diffuse lymphocytic infiltrate present in this area (H & E, original magnification x250). G) Invasive endometrioid adenocarcinoma (case patient 6-2), near-microscopic 1.0 cm in diameter, with the characteristic glandular epithelium moderately differentiated, eliciting an inflammatory stromal reaction (H & E, original magnification xlOO). H) Benign Brenner's tumor (case patient 6-2), also nearmicroscopic 1.0 cm in diameter, with characteristic nests of transitional-type epithelium embedded within a dense fibrous stroma (H & E, original magnification x 100).

connection with the surface was seen. Although epithelial hyperplasia and papillomatosis were present in several cases, no mitoses or atypia were observed in any of the lesions credited to this category. It is noteworthy, however, that in one case the inclusion cyst epithelium showed mucinous differentiation along endocervical lines (Fig. 2, H). Occasional mitoses, as well as shrunk-degenerated, probably apoptotic cells, were observed in the surface epithelium of cancer-prone ovaries associated with an increase in height and pseudostratification of the tall columnar cells. A significantly increased number of cells or real hyperplasia, however, was not detected, but the surface epithelial cells of these ovaries appeared to be "active" (Fig. 2). This term is used to denote the apparent activation of some functional capability such as secretion. For example, the cells were often enlarged, tall, columnar, and pseudostratified with elongated or ovoid nuclei containing dispersed or dense chromatin and occasional small nucleoli and

had increased granular cytoplasm compared with the smaller, cuboidal, or flat ordinary superficial cells with small round nuclei and scant cytoplasm. Furthermore, examination of estrogen receptor status by immunohistochemistry revealed the "more-lush"-appearing pseudostratified portions of the surface epithelium to stain more intensely than that of adjacent or neighboring flat or cuboidal epithelium, a finding consistent with previous reports {30,31). This regional pattern of relative estrogen receptor expression was also apparent in the papillary lesions described above. In addition to the described differences in the appearance and growth pattern of the cells that may become malignant if cancer develops in a cancer-prone ovary, the stroma of these organs also appears to be uncharacteristically active. Activity is noted in 75% of high-risk ovaries against only 10% of control ovaries (P = .00017; Table 4). A range of alterations was often present that may generally be described as stromal hyperplasia (Fig. 3).


ARTICLES 1815

Table 3. Histopathologic findings in control ovaries* Non-neoplastic changes Case No.

Age,y (reason for oophorectomy)

1 2 3 4 5 6 7 g 9 10 11 12 13 14 15 16 17 18 19 20

Endometriosis

Inclusion cysts

Papillomatosis

Invaginations

Epithelial hyperplasia

Stromal activity

39 (L) 44 (L) 48(L,M) 48 (L) 40 (L) 41 (A) 49 (L) 45 (E) 48 (L) 48 (L) 49 (L) 48 (L) 49 (L, M ) 49 (L) 48 (L) 38 (L) 44 (A, M ) 39 (L) 44 (L) 35 (L)

•L = leiomyomatosis; M = menorrhagia; E = endometnosis; A = adenomyosis; + = presence; - = absence.

These changes included increased follicular activity, stromal hyperthecosis and luteinization (Fig. 3, A-C), hylar cell hyperplasia (Fig. 3, D), and corpus luteum hyperplasia. Considered together, papillomatosis, inclusion cysts, invaginations, epithelial hyperplasia, and stromal activity permit strong differentiation between control and high-risk ovaries on the basis of statistical analysis. Table 5 describes the differences between the high-risk and control groups as a function of the number of these five histologic markers having grade +, ++, or +++ found in each ovary. For example, 15 of 20 high-risk ovaries had three or more positive markers versus only two of the 20 control ovaries (P = .00007).

Discussion In the development of many types of cancer, an increasingly well-defined series of morphologic changes are recognized to occur as the benign epithelium becomes malignant. This is perhaps most notable in the colon, where this realization has guided increasingly successful efforts to unravel the molecular genetic basis of the disease (32,33). In contrast, there is controversy as to whether there are morphologic precursors in the pathway to development of clinical ovarian cancer (22). This in large part may be due to the fact that ovarian cancer is most frequently diagnosed at a late stage. Hence, the opportunity for any one pathologist to examine a large number of early stage ovarian cancers where it might be possible to repetitiously see and document a consistent pattern of transition between benign and malignant ovarian surface epithelium has been rare. This, in tum, has obviously constrained our understanding as to the most frequent sequence of morphologic changes that occur as clinical ovarian cancer develops. If a series of morphologic precursors of clinical ovarian cancer could be defined, their study using currently available molecular biologic strategies could help to 1816 ARTICLES

clarify which of the morass of genetic alterations seen in late stage disease are causal. Previous investigations aimed at defining the types of lesions that lead to ovarian cancer have involved a variety of approaches, including the following: 1) examination of the contralateral ovaries in patients with unilateral ovarian cancer (23,24) and 2) examination of ovaries that contain stage I tumors (34). We believe our current study complements these previous efforts. Through access to ovaries of individuals with a strong family history of ovarian cancer, we have had the opportunity to examine tissue where there is a very high probability that malignant transformation will occur sometime during the individual's lifetime. This is evidenced by the unanticipated discovery of microscopic and near-microscopic adenocarcinomas in the ovaries of two of these individuals. It should be noted, however, that the exact risk of ovarian neoplasia, although high, varies in this study group. For example, in individuals where family history is the sole criterion for inclusion in the group, there is only a 50% probability that the individual has inherited the predisposition. Furthermore, even when inheritance is known, i.e., by linking studies or DNA sequencing, penetrance apparently varies on the basis of several factors, including the exact mutation within BRCA1. To date, more than 80 unique germline mutations have been identified in the BRCA1 gene; however, specific phenotypic changes resulting from a given mutation have yet to be clearly demonstrated (25-27,35-44). Berman et al. (27) found that, in families carrying the 185delAG mutations, a higher incidence of ovarian cancer (42%) was observed compared with the reported incidence associated with the 4184del4 (21%) and the 5382insC (26%) mutations. Gayther et al. (45) found a statistically significant correlation between the location of the mutation and the ratio of breast to ovarian cancer incidence. They observed that a higher incidence of ovarian cancer was evident in families with BRCA1 mutations in the first


Fig. 2. Examples of altered ovarian surface epithelial cell growth patterns seen in cancer-prone ovaries (selected from case patients presented in Table 1). A) The ovarian surface is studded by a large number of micropapillary projections. This is a frequent finding but usually is more focal, and the papillae are in smaller groups but sometimes with larger papillary excrescenses (hematoxylineosin [H & E], original magnification x20). B) Convoluted ovarian cortex with deep invagjnations of the surface epithelium, with focal micropapillary projections (H & E, original magnification x20). C) The deepest recess of an ovarian surface epithelial invagination appears dilated and presents a multilobulated papillary projection within the lumen, forming a microscopic papillary cystadenoma within the cortical stroma (H & E, original magnification x40). D) Another microscopic papillary cystadenoma, measuring 1.1 mm in diameter, presenting a more complex papillary component lined by a simple epithelium continuous with the invaginated surface ovarian epithelium (H & E, original magnification x40). E) Superficial simple epithelial inclusion cyst of the most common variety, displaying a portion of increased stratification and height of the lining epithelium. (These more stratified or hyperplastic areas usually express estrogen receptors more intensely than the flat or cuboidal areas, not illustrated here.) Note the absence of surface epithelium that has been mechanically stripped for culture (H & E, original magnification x40). F) Example of a very complex group of epithelial inclusion cysts within the ovarian cortical stroma, showing a hyperplastic epithelial network, sometimes only present focally within a given cystic structure. This is probably another form of microscopic cystadenoma, superficial and nonpapillary (H & E, original magnification x40). G) Cortical inclusion cyst with rich epithelial pseudostratification and numerous papillary projections. Note the plump cells in the surrounding fibrous stroma suggestive of focal luteinization and presumptive hormonal activity (H & E, original magnification xlOO). H) Cortical inclusion cyst with single papillary projection lined by simple mucous epithelium with endocervical pattern, which is not apparent in the rest of the cyst. Mucoid material is present in the lumen. This finding is uncommon. Most cortical inclusion cysts are lined by a flat, cuboidal, or serous type of epithelium (H & E, original magnification x400).

two thirds of the gene compared with its last third. Furthermore, Holt et al. (46) reported that ovarian cancer formed a significantly lower proportion (2%) of the cancers in individuals with mutant proteins that would include the granin motif at codons 1214-1223 compared with the proportion (25%) of ovarian cancers in patients with more severely truncated proteins. Experimentally, they demonstrated in culture that near full-length truncated BRCA1 proteins do not inhibit breast cancer cell growth, but do inhibit ovarian cancer cell growth (47). Consistent with these results, we have found that the mutations detected in BRCA1 in the constitutional DNA of the women in our study who underwent oophorectomy for prophylaxis all occur prior to the granin motif (data not shown). Even though the exact risk of carcinogenesis within the ovaries we have studied is not known, they displayed a remarkable number of differences compared with control specimens. These differences

predominantly involved variation in features and growth patterns of the surface epithelium, but there was also evidence of hyperactive stroma. Such changes are consistent with previous studies (7,12,23,24,34) aimed at identifying precursor lesions for ovarian cancers in other types of ovarian specimens. The major difference between the previous and present investigations is that, in our study, only the genetic risk for cancer was present, whereas in the others, overt ovarian cancer, although "early" in some cases, had already occurred in most individuals. On the basis of recent work from Dubeau's laboratory (48), these earlier studies must be viewed with some caution, since there is the possibility that the "apparent" benign precursors could bear the genetic constitution of the nearby carcinomas, based on similar outcome of molecular genetic analysis of the closely associated lesions. The same must be said for the changes we describe in the ovaries incidentally discovered to contain

Journal of the National Cancer Institute, Vol. 88, No. 24, December 18, 19%

ARTICLES 1817

Fig. 3. Examples of altered ovarian surface epithelial cell growth patterns and stromal activity seen in cancer-prone ovaries (selected from cases presented in Table 1). A) Multiple cortical epithelial inclusion cysts lined by cuboidal or pseudostratified epithelium and separated by a cortical stroma with plump rounded cells with clear cytoplasm that represent dispersed stromal luteinization. Note the absence of surface ovarian epithelium that has been stripped for study (hematoxylin-eosin [H & E], original magnification xlOO). B) Higher-power view of complex cortical inclusion cysts with a luteinized fibrous stroma, suggestive of active steroid production (H & E, original magnification x250). O Stromal hyperthecosis with irregular nests of luteinized stromal cells that become rounded with abundant eosinophilic or clear cytoplasm and central round nuclei, characteristic of steroid-producing cells (H & E, original magnification X250). D) Hilar (Leydig) cell hyperplasia at the hilum of one ovary, formed by a large conglomerate of uniform polygonal cells with round nuclei and granular eosinophilic cytoplasm, closely associated with nerve bundles and blood vessels. The contralateral ovary had similar changes at the hilum (H & E, original magnification x250).

small adenocarcinomas. Therefore, it is possible that the apparently benign lesions thought to be precursors are, in fact, malignant lesions masquerading as benign, perhaps because of some locally active paracrine regulatory mechanism. Limited previous studies have used specimens somewhat similar to ours. These have been reviewed by Scully (7). The results have been variable and in most cases involved the study of only a few ovaries, with family history being the only basis for examination. The one clear exception is the study by Gusberg and Deligdisch (49) of the ovaries of three sets of identical twins, where inherited genetic risk is indisputable. The authors of this study concluded that precancerous changes were present in the ovaries of each unaffected twin. Hence, this and another early-limited study (50) support the work presented here. Additionally, limited experimental evidence supports our findings. For example, it has been shown that surface epithelial cells from the ovaries of the cancer-prone individuals in this study (data not shown) and others have different growth features in vitro

and have divergent functional characteristics on the basis of CA125 production compared with control cells (57). We believe the results presented here raise several questions. The discovery of two microscopic malignant tumors in high-risk ovaries was of interest. These tumors were not anticipated and in one case not uncovered until extensive microscopic examinations were performed. This suggests that a cursory analysis of prophylactically removed ovaries could, in some cases, fail to detect small malignant tumors. If such were the case and these tumors had acquired early metastatic potential, cells that metastasized from them could explain the discovery of peritoneal carcinomatosis subsequent to prophylactic oophorectomy (52). This idea is supported by one case report, where careful retrospective examination of the ovaries of a woman who developed peritoneal carcinomatosis 3 years after prophylactic oophorectomy revealed the primary malignant tumor (55). The idea would gain more potential validity by invoking the concept that the early metastatic cells could still have growth regulatory con-

Table 4. Correlation between risk category and histologic marker grade Histologic grade Marker

Risk category

Papillomatosis

Control High risk

16

6

3 11

2

Inclusion cysts

Control High risk

15 6

5 9

o. 5

Invaginations

Control High risk

9 1

10 7

6

Control High risk

20 12

0 6

Control High risk

18 5

1

1

8

4

Pseudostratification Stromal activity

1

1

D 1

0 1

.005

0 0 0 6

.006

0 1

.003

0 3

.00017

.0004

"P is the probability that correlation between risk status and the grade (-, +, ++, or +++) of a histologic marker could arise by chance alone.

1818 ARTICLES


TaWe 5. Number of positive histologic markers as a function of risk status No. of subjects (%) No. of positive histologic markers

Control (n = 20)

High risk (n = 20)

P*

One or more •Two or more Three or more Four or more

13(65) 6(30) 2(10) 1(5)

20(100) 17(85) 15 (75) 11(55)

.008 .001 .00007 .001

*P is the probability that correlation between risk status and the number of positive markers observed could arise by chance alone. Five histologic markers are considered: papillomatosis, inclusion cysts, invaginations, epithelial hyperplasia, and stromal activity.

straints intact. Over time, it is possible that these could become defective through mutation, inactive because of changes in hormonal status associated with the surgical menopause, or perhaps overridden by exogenous hormones (55). A second question arises on the basis of the observation of hyperactive stroma in prophylactic oophorectomy specimens. It is accepted that the malignant component of a carcinoma is the epithelium. Therefore, it is the surface epithelium that gives rise to a cancer, should it occur in a high-risk ovary. This does not rule out the possibility, however, that the stroma, by nature of an abnormal genetic constitution, could either lack a growthrestraining capacity characteristic of the normal tissue or could, on the contrary, provide an abnormal growth stimulus. Alternatively, the abnormal genetic constitution of the surface epithelial cells could result in the inappropriate production of a signal to the stroma to produce substances mitogenic to the surface epithelium, i.e., the paracrine loop. If any of these possibilities are correct, proliferation of the surface epithelium would be increased. There, in fact, has been much speculation (10,11,16,54) and some experimental data (9,55) to support a role for proliferation in malignant transformation of these cells. The last and perhaps most provocative issue our data raise is in regard to whether early steps in the process of malignant transformation of surface epithelial cells from the ovaries in cancer-prone individuals involve dominant or recessive-acting genes. It is generally accepted that the BRCA1 gene, the vehicle through which most inherited ovarian cancer is transmitted, functions as a classic tumor suppressor gene. This suggests that a mutant BRCA1 protein could be acting in a dominant-negative way, similar to mutant TP53 (56) or, alternatively, perhaps another gene(s) cooperates with BRCA1 to modify the incidence of inherited ovarian cancer. If this latter, highly speculative concept were the case, we would propose that the gene would have limited distribution in the population and on its own might confer only a minimally increased risk of disease. However, if inherited in common with a mutant allele for BRCA1, it could produce a substantial risk. In this aspect, Phelan et al. (57) have found that the risk of ovarian cancer is increased in BRCA1 carriers by the presence of rare HRAS alleles. They observed that the risk for ovarian cancer was 2.11 times greater for BRCA1 carriers harboring one or two rare alleles of the HRAS1 variable number of tandem repeat compared with carriers with only common alleles. Another interesting candidate gene would be one whose product produced slightly increased fertility. This

feature would enhance its distribution in the population. Furthermore, if relative ovulatory activity with subsequent growth of the surface epithelium increases the risk of mutations within the target cells, this could provide a mechanism for inactivation/mutation of BRCA1 as well as other critical genes. Related to this possibility, Jensen et al. (47) have shown that BRCA1 expression is increased in ovarian cancer cells exposed to estrogen or tamoxifen and that there is a dose-response effect. One might speculate that women producing lower amounts of estrogen may express lower levels of BRCA1 and therefore fail to tightly regulate proliferation of ovarian surface epithelial cells following ovulation. In addition to the several questions our study raises with regard to the process of malignant transformation of the surface epithelium of cancer-prone ovaries, we suggest that it provides evidence that a characteristic histopathologic phenotype is prevalent in these specimens—i.e., they contain multiples and increased intensity of the histologic features described that are rarely seen in normal ovaries. As we have indicated in the "Materials and Methods" section, these findings were not obtained through a classic blinded study design. Clearly, our data indicate that a larger-scale investigation in the suggested format is warranted to confirm these observations. Additionally, our results suggest that ovaries removed from ovarian cancer-prone individuals should be thoroughly examined histologically to identify or rule out microscopic or near microscopic invasive neoplasms.

References (/) Wingo PA, Tong T, Bolden S. Cancer statistics, 1995. CA Cancer J Clin 1995;45:8-30. (2) van der Berg ME, van Lent M, Buyse M, Kobierska A, Colombo N, Favalli G, et al. The effect of debulking surgery after induction chemotherapy on the prognosis in advanced ovarian cancer. Gynecological Cancer Cooperative Group of the European Organization for Research and Treatment of Cancer [see comment citations in Medline]. N Engl J Med 1995;332:62934. (3) Parkin DM, Laara E, Muir CS. Estimates of the worldwide frequency of sixteen major cancers in 1980. Int J Cancer 1988;41:184-97. (4) Scully R. Histological typing of female genital tract tumours. New York: Springer-Verlag, 1994. (5) Stalsberg H, Bjamason O, dfi Carvalho A, Correa P. International comparison of histologic types of ovarian cancer in cancer registry material. UICC Tech Rep Ser 1983;75:247-8O. (6) Bell DA, Scully RE. Early de novo ovarian carcinoma. A study of fourteen cases. Cancer 1994;73:1859-64. (7) Scully R. Ovary. In: Henson D, Albores-Saavedra J, editors. Pathology of incipient neoplasia. Philadelphia: Saunders, 1993:283-300. (8) Nicosia S, Nicosia R. Neoplasms of the ovarian mesothelium. In: Azar H, editor. Pathology of human neoplasms. New York: Raven Press, 1988:43586. (9) Testa JR, Getts LA, Salazar H, Liu Z, Handel LM, Godwin AK, et al. Spontaneous transformation of rat ovarian surface epithelial cells results in well to poorly differentiated tumors with a parallel range of cytogenetic complexity. Cancer Res 1994^4:2778-84. (Iff) Fathalla MF. Incessant ovulation—a factor in ovarian neoplasia? Lancet 1971;2:163. (//) Fathalla MF. Factors in the causation and incidence of ovarian cancer. ObstetGynecolSurv 1972;27:751-68. (12) Fox H. Pathology of early malignant change in the ovary. Int J Gynecol Pathol 1993; 12:153-5. (13) Godwin AK, Testa JR, Hamilton TC. The biology of ovarian cancer development. Cancer 1993;71(2 Suppl):53O-6. (14) Lowry S, Russell H, Hickey I, Atkinson R. Incessant ovulation and ovarian cancer [letter] [see comment citation in Medline]. Lancet 1991;337:15445.


ARTICLES 1819

(J5) Woodruff JD. History of ovarian neoplasia: facts and fancy. Obstet Gynecol Annu 1976;5:331-44. (16) Hamilton TC. Ovarian cancer, Part I: biology. Curr Probl Cancer 1992; 16:1-57. (17) Cramer DW, Xu H, Harlow BL. Does "incessant" ovulation increase risk for early menopause? Am J Obstet Gynecol 1995; 172:568-73. (18) Scully R, Bell D, Abu-Jawdeh G. Update on early ovarian cancer and cancer developing in benign ovarian tumors. In: Sharp F, Mason P, Blackett T, et al, editors. Ovarian cancer 3. London: Chapman & Hall, 1995:139-44. (19) Ozols R, Rubin S, Dembo A, et al. Epithelial ovarian cancer. In: Hoskins W, Perez C, Young R, editors. Gynecologic oncology: principles and practice. Philadelphia: Lippincott, 1992:731-81. (20) Powell DE, Puls L, van Nagell J Jr. Current concepts in epithelial ovarian tumors: does benign to malignant transformation occur? [see comment citations in Medline]. Hum Pathol 1992;23:846-7. (21) Puls LE, Powell DE, DePriest PD, Gallion HH, Hunter JE, Kryscio RJ, et al. Transition from benign to malignant epithelium in mucinous and serous ovarian cystadenocarcinoma. Gynecol Oncol 1992;47:53-7. (22) Cheng PC, Gosewehr JA, Kim T, Velicescu M, Wan M, Zheng J, et al. Potential role of the inactivated X chromosome in ovarian epithelial tumor development [see comment citation in Medline]. J Natl Cancer Inst 1996;88:510-8. (23) Mittal KR, Zeleniuch-Jacquotte A, Cooper J, Demopoulos RJ. Contralateral ovary in unilateral ovarian carcinoma: a search for preneoplastic lesions. Int J Gynecol Pathol 1993; 12:59-63. (24) Resta L, Russo S, Colucci GA, Prat J. Morphologic precursors of ovarian epithelial tumors [see comment citation in Medline]. Obstet Gynecol 1993; 82:181-6. (25) Serova O, Montagna M, Torchard D, Narod SA, Tonin P, Sylla B, et al. A high incidence of BRCA1 mutations in 20 breast-ovarian cancer families. Am J Hum Genet 1996;58:42-51. (26) Simard J, Tonin P, Durocher F, Morgan K, Rommens J, Gingras S, et al. Common origins of BRCA1 mutations in Canadian breast and ovarian cancer families. Nat Genet 1994;8:392-8. (27) Berman DB, Wagner-Costalas J, Schultz DC, Lynch HT, Daly M, Godwin AK. Two distinct origins of a common BRCA1 mutation in breast-ovarian cancer families: a genetic study of 15 185delAG-mutation kindreds. Am J Hum Genet 1996;58:1166-76. (28) Dawson-Saunders B, Trapp R. Basic and clinical biostatistics. Norwalk (CT): Appleton and Lange, 1994. (29) Armitage P. Statistical methods in medical research. New York: John Wiley & Sons, 1971. (30) Karlan BY, Jones J, Greenwald M, Lagasse LD. Steroid hormone effects on the proliferation of human ovarian surface epithelium in vitro. Am J Obstet Gynecol 1995;173:97-104. (31) Hamilton T, Davies P, Griffiths K. Steroid-hormone receptor status of the normal and neoplastic ovarian surface germinal epithelium. In: Greenwald G, Terranova P, editors. Factors regulating ovarian function. New York: Raven Press, 1983:81-5. (32) Fearon ER, Vogelstein B. A genetic model for colorectal tumorgenesis. Cell 1990:61:759-67. (33) Parsons R, Li GM, Longley MJ, Fang WH, Papadopoulos N, Jen J, et al. Hypermutability and mismatch repair deficiency in RER+ tumor cells. Cell 1993;75:1227-36. (34) Deligdisch L, Einstein AJ, Guera D, Gil J. Ovarian dysplasia in epithelial inclusion cysts. A morphometric approach using neural networks. Cancer 1995;76:1027-34. (35) Castilla LH, Couch FJ, Erdos MR, Hoskins KF, Calzone K, Garber JE, et al. Mutations in the BRCA1 gene in families with early-onset breast and ovarian cancer. Nat Genet 1994;8:387-91. (36) Friedman LS, Ostermeyer EA, Szabo CI, Dowd P, Lynch ED, Rowell SE, et al. Confirmation of BRCA1 by analysis of germline mutations linked to breast and ovarian cancer in ten families. Nat Genet 1994;8:399-404. (37) Friedman LS, Szabo CI, Ostermeyer EA, Dowd P, Butler L, Park T, et al. Novel inherited mutations and variable expressivity of BRCA1 alleles, including the founder mutation 185delAG in Ashkenazi Jewish families. Am J Hum Genet 1995;57:1284-97. (38) Futreal PA, Liu Q, Shattuck-Eidens D, Cochran C, Harshman K, Tavtigian S, et al. BRCA1 mutations in primary breast and ovarian carcinomas. Science 1994,266:120-2. (39) Hogervorst FB, Comelis RS, Bout M, van Vliet M, Oosterwijk JC, Olmer R, et al. Rapid detection of BRCA1 mutations by the protein truncation test Nat Genet 1995;10:208-12. (40) Inoue R, Fukutomi T, Ushijima T, Matsumoto Y, Sugimura T, Nagao M. Germline mutation of BRCA1 in Japanese breast cancer families. Cancer Res 1995:55:3521-4.

1820 ARTICLES

(41) Johannsson O, Ostermeyer EA, Hakansson S, Friedman LS, Johansson U, Sellberg G, et al. Founding BRCA1 mutations in hereditary breast and ovarian cancer in southern Sweden. Am J Hum Genet 1996;58:441 -50. (42) Miki Y, Swensen J, Shattuck-Eidens D, Futreal PA, Harshman K, Tavtigian S, et al. A strong candidate for the breast and ovarian cancer susceptibility gene BRCA1. Science 1994;266:66-71. (43) Shattuck-Eidens D, McClure M, Simard J, Labrie F, Narod S, Couch F, et al. A collaborative survey of 80 mutations in the BRCA1 breast and ovarian cancer susceptibility gene. Implications for presymptomatic testing and screening JAMA 1995;273:535-41. (44) Struewing JP, Brody LC, Erdos MR, Kase RG, Giambarresi TR, Smith SA, et al. Detection of eight BRCA1 mutations in 10 breast/ovarian cancer families, including 1 family with male breast cancer. Am J Hum Genet 1995;57:l-7. (45) Gayther SA, Warren W, Mazoyer S, Russell PA, Harrington PA, Chiano M, et al. Germline mutations of the BRCA1 gene in breast and ovarian cancer families provide evidence for a genotype-phenotype correlation. Nat Genet 1995;11:428-33. (46) Holt JT, Thompson ME, Szabo C, Robinson-Benion C, Arteaga CL, King MC, et al. Growth retardation and tumour inhibition by BRCA1 [see comment citation in Medline]. Nat Genet 1996; 12:298-302. (47) Jensen RA, Thompson ME, Jetton TL, Szabo CI, van der Meer R, Helou B, et al. BRCA1 is secreted and exhibits properties of a granin [see comment citations in Medline]. Nat Genet 1996; 12:303-8. (48) Zheng J, Benedict W, Xu HJ, Hu SX, Kim TM, Velicescu M, et al. Genetic disparity between morphologically benign cysts contiguous to ovarian carcinomas and solitary cystadenomas [see comment citation in Medline]. J Natl Cancer Inst 1995;87:1146-53. (49) Gusberg SB, Deligdisch L. Ovarian dysplasia. A study of identical twins. Cancer 1984;54:1-4. (50) Graham J, Graham R, Scheuller E. Preclinical detection of ovarian cancer. Cancer 1964; 17:1414-32. (5/) Auersperg N, Maines-Bandiera S, Booth JH, Lynch HT, Godwin AK, Hamilton TC, et al. Expression of two mucin antigens in cultured human ovarian surface epithelium: influence of a family history of ovarian cancer. Am J Obstet Gynecol 1995; 173:558-65. (52) Tobacman JK, Greene MH, Tucker MA, Costa J, Kase R, Fraumeni JF Jr. Intra-abdominal carcinomatosis after prophylactic oophorectomy in ovarian-cancer-prone families. Lancet 1982;2:795-7. (53) Chen KT, Schooley JL, Flam MM. Peritoneal carcinomatosis after prophylactic oophorectomy in familial ovarian cancer syndrome. Obstet Gynecol 1985;66:93s-94s. (54) Iingeman CH. Environmental factors in the etiology of carcinoma of the human ovary: a review. Am J Ind Med 1983;4:365-79. (55) Godwin AK, Testa JR, Handel LM, Liu Z, Vanderveer LA, Tracey PA, et al. Spontaneous transformation of rat ovarian surface epithelial cells: association with cytogenetic changes and implications of repeated ovulation in the etiology of ovarian cancer [see comment citation in Medline]. J Natl Cancer Inst 1992;84:592-601. (56) Godwin A, Hamilton T, Knudson A. Oncogenes and anti-oncogenes in gynecologic malignancies. In: Hoskins W, Perez C, Young R, editors. Gynecologic oncology: principles and practice. Philadelphia: Lippincott, 1992:87-116. (57) Phelan CM, Rebbeck TR, Weber BL, Devilee P, Ruttledge MH, Lynch HT, et al. Ovarian cancer risk in BRCA1 carriers is modified by the HRAS1 variable number of tandem repeat (VNTR) locus. Nat Genet 1996;12:309-ll.

Notes Supported by Public Health Service grants CA60643, CA09035 (P. B. Laub), and CA569I6 (T. C. Hamilton) from the National Cancer Institute, National Institutes of Health, Department of Health and Human Services; a Hoxie Harrison Smith grant (A. K. Godwin); The Evy Lessin Fund for Ovarian Cancer Research; DAMD17-94-J-4340 (H. T. Lynch); State of Nebraska Cancer and Smoking Disease Research Program grant LB595; and Council for Tobacco Research grant 1297ERI. We acknowledge the pathologists and surgeons at hospitals throughout the United States who provided ovary specimens collected during oophorectomy procedures performed on women for prophylactic purposes, and Teresa Conway and Carrie Snyder (Creighton University), David Berman and Josephine Costolos (Fox Chase Cancer Center), and Francis X. McBrearty (Lankenau Hospital) for providing histologic sections of control ovaries. Manuscript received April 5, 1996; revised September 6, 1996; accepted October 9, 1996.
