Setsuko K. Chambers

Tubal origin of 'ovarian' low-grade serous carcinoma.

Ovarian low-grade serous carcinomas are thought to evolve in a stepwise fashion from ovarian epithelial inclusions, cystadenomas, and borderline tumors. The current study was designed to gain insight into the origins of low-grade serous carcinomas (tubal versus ovarian) by comparatively evaluating the morphologic (secretory and ciliated cell distribution) and immunophenotypic (using antibodies to PAX8, tubulin, calretinin, and Ki67) attributes of its putative precursor lesions, the normal tubal epithelium, and the overt malignancy. A total of 226 adnexal tissues from 178 patients were studied, including 98 adnexae removed for non-neoplastic indications, 48 serous cystadenomas, 42 serous borderline tumors, and 38 low-grade serous carcinomas. Normal distal tubal epithelium comprised an admixture of PAX8+/tubulin− secretory cells and PAX8−/tubulin+ ciliated cells with a proliferative index of ∼3%. The vast majority of ovarian surface epithelia displayed a mesothelial phenotype (calretinin+/PAX8−/tubulin−) and low proliferative index (0% (12 per 1000)), although 4% of cases also displayed foci with tubal phenotype (calretinin−/PAX8+/tubulin+). In contrast, most (78%) of the ovarian epithelial inclusions displayed a tubal phenotype and had a significantly higher proliferative index (1%) than ovarian surface epithelium, indicating that in most cases, the ovarian surface epithelium and ovarian epithelial inclusions are of different lineages. There was a progressive decrease in the population of ciliated cells, as evidenced by increasing secretory/ciliated cell ratio, from ovarian epithelial inclusions/cystadenomas to borderline tumors to low-grade serous carcinoma, indicating that the latter is a clonal expansion of secretory cells. Overall, the findings make a strong argument that the ovarian epithelial inclusions with a tubal phenotype is likely derived from fallopian tube through an intraovarian endosalpingiosis rather than through Mullerian metaplasia from ovarian surface epithelium. Genetic and molecular studies are needed to further confirm this finding as tubal origination of ovarian serous cancers will have a significant impact on ovarian cancer prevention and management.

Precursors of endometrial clear cell carcinoma

The recognition of morphologically identifiable lesions which may confer an increased risk for subsequent development of an invasive malignancy offers an opportunity to investigate and better understand the molecular-genetic etiopathogenesis of the well-developed tumor, and potentially, to administer a therapeutic intervention before its development. In contrast to uterine endometrioid and serous carcinomas, very little is known about the potential precursor lesions of endometrial clear cell carcinoma (ECCC). In our routine practice, we have noted the presence of a spectrum of atypical glandular changes in the endometria adjacent to ECCC or endometrial carcinomas with a clear cell component, which on the basis of current criteria, would not qualify for any specific designation. We hypothesize that these lesions represent the earliest morphologically recognizable precursor lesions of ECCC and systematically characterize their clinicopathologic features herein. Thirty archived cases of pure ECCC (n=14) or mixed endometrial carcinomas with a >10% clear cell component (n=16) were retrieved and the “normal” endometria adjacent to the malignancies were evaluated in detail. Thirty-eight benign uteri and 30 uteri with classic endometrial endometrioid carcinoma (EEC) served as controls. All cases were reviewed in a blinded fashion. Putative precursor lesions (PPL) were searched for and identified microscopically. The lesions were typically isolated glands or surface epithelium (within an otherwise normal endometrial region) that displayed cytoplasmic clarity and/or eosinophilia with varying degrees of nuclear atypia. Twenty-seven (90%) of the 30 cases had at least 1 PPL. In contrast, PPL were identified neither in the benign uteri nor in endometrioid carcinoma control groups (P<0.001). A total of 67 foci of PPL were identified in the 27 cases with an average of 2.5 foci per case. The immunohistochemical expression of p53, mib-1, estrogen receptor (ERs), and progesterone receptor in the benign endometria, ECCC, and the PPL were evaluated on all 27 cases. The mean p53 scores for the benign endometria, PPL, and ECCC were 0, 4.5, and 6.2, respectively. Parallel values for mib-1 were 15%, 45%, and 63%. ER/progesterone receptor indices for benign endometria, PPL, and carcinoma were 90/80, 21.52/4.61, and 11/4, respectively.The PPL described herein have a morphologic and immunophenotypic profile which seems to be distinct from both the benign endometria in which they reside and the adjacent areas of ECCC. The high frequency of association of these lesions with ECCC, their frequent occurrence as isolated lesions within otherwise “benign-appearing” endometria, and their continuous spectrum of nuclear atypia from minimum (grade 1, cytologic atypia falls short of ECCC cells) to maximum (grade 3, cytologically identical to ECCC cells), argues in favor of our hypothesis that these may represent precursor lesions of ECCC. Further studies are required to conclusively define the nature of these lesions. However, such studies can only be performed if diagnostic surgical pathologists recognize, highlight, and segregate these lesions for further analysis.

Occurrence of endometrial glandular dysplasia precedes uterine papillary serous carcinoma.

Endometrial glandular dysplasia (EmGD) is a newly defined entity that is commonly and specifically associated with serous endometrial intraepithelial carcinoma and uterine papillary serous carcinoma (UPSC). Endometrial glandular dysplasia has been proposed as a true precancerous lesion of UPSC based on our recent studies showing morphological and molecular linkages between these 2 lesions. The present report is to examine if EmGD occurs before UPSC development and to define the period from the occurrence of EmGD to a full-blown UPSC by studying their clinicopathologic features in a retrospective setting. A total of 250 UPSC and 258 benign cases were used as initial study source. To identify if EmGD existed before the development of UPSC, we blindly reviewed all available endometrial biopsies from a period of 3 months or earlier before hysterectomies. These included an available pool of 27 biopsy specimens from UPSC group and 29 samples from benign control group. Any endometrial abnormalities, which morphologically qualified as EmGD as defined previously in preceding biopsies were recorded. Among all endometrial biopsies before hysterectomies, we morphologically identified a total of 10 EmGD cases; 9 (33%) of 27 were from UPSC group and 1 (3.5%) of 29 were from benign control group. All 10 morphologically diagnosed EmGD cases showed a high p53 staining score (≥5) except 1 noncontributory from UPSC group and 1 from the benign control group with a score of 0. A high MIB-1 index score was seen in all EmGD cases, whereas low index was found in morphologically benign biopsies. The main purpose of this study is to report these retrospectively identified EmGD cases. The period from identifying EmGD to the presence of either a serous endometrial intraepithelial carcinoma or a full-blown UPSC ranged from 16 to 98 months with an average of 33 months. We conclude that occurrence of EmGD precedes the development of UPSC. The findings support our recently proposed UPSC development model, in which EmGD is likely to be a precursor lesion of UPSC. Further studies are needed to address issues in regard to molecular and cellular mechanisms, reversibility, risk of UPSC development, and clinical management of EmGD.

Regulation of non-AU-rich element containing c-fms proto-oncogene expression by HuR in breast cancer

The role of RNA-binding proteins in cancer biology is recognized increasingly. The nucleocytoplasmic shuttling and AU-rich RNA-binding protein HuR stabilizes several cancer-related target mRNAs. The proto-oncogene c-fms, whose 3′untranslated region (3′UTR) is not AU-rich, is associated with poor prognosis in breast cancer. Using a large breast-cancer tissue array (N=670), we found nuclear HuR expression to be associated with nodal metastasis and independently with poor survival (P=0.03, RR 1.45), as well as to be co-expressed with c-fms in the breast tumors (P=0.0007). We described c-fms mRNA as a direct target of HuR in vivo, and that HuR bound specifically to a 69-nt region containing ‘CUU’ motifs in 3′UTR c-fms RNA. Overexpressing or silencing HuR significantly up- or down-regulated c-fms RNA expression, respectively. We also found that known glucocorticoid stimulation of c-fms RNA and protein is largely dependent on the presence of HuR. HuR, by binding to the 69-nt wild type, but not mutant, c-fms sequence can regulate reporter gene expression post-transcriptionally. We are the first to describe that HuR can regulate gene expression by binding non-AU-rich sequences in 3′UTR c-fms RNA. Collectively, our findings suggest that HuR plays a supportive role for c-fms in breast cancer progression by binding a 69-nt element in its 3′UTR, thus regulating its expression.

Correlation of tumor phenotype with c-fms proto-oncogene expression in an in vivo intraperitoneal model for experimental human breast cancer metastasis

Although proto-oncogene expression has been shown to correlate with clinical outcome in breast carcinoma, an experimental model has not been proposed to study this phenomenon in vivo. In addition, the ability to modulate this proto-oncogene in vivo to correlate with phenotypic behavior has not been determined. Utilizing an intraperitoneal model for metastatic spread with BT20 human breast carcinoma cells, clonally expanded cells expressing five fold higher c-fms protein were compared with parent BT20 cells as well as an underexpressing clone using intrasplenic injection following left flank cut-down in female nude and Severe combined immunodeficient (SCID) mice. Athymic BALB/c nude and SCID animals were observed for clinical evidence of tumorigenicity with necropsy performed at either 50 or 80 days unless compromised earlier. Immunohistochemistry (IHC) of the harvested tumors was performed to correlate c-fms expression from its original in vitro culture to the in vivo model. At day 50, differences in primary tumor take and spread to the pelvis were already evident favoring the c-fms over-expression group with IHC of these tumors revealing significantly higher intensity of staining for c-fms, (mean H score of 205 vs. 43 in the over-expression and parent groups, respectively). At day 80, tumor take and spread was comparable; however, tumor size in the over-expression group was significantly larger than the parent and under-expressing group in both the BALB/c and SCID experiments. Modulation of c-fms proto-oncogene expression was also achieved using the anti-glucocorticoid, RU-486, via oral administration to SCID mice with subsequent correlation to IHC staining. This model thus provides tumors of significant size and organ diversity which retain their phenotype early in tumorigenesis allowing an early endpoint to assess efficacy of novel treatments.

Inhibition of the c-fms proto-oncogene autocrine loop and tumor phenotype in glucocorticoid stimulated human breast carcinoma cells.

The c-fms proto-oncogene encoded CSF-1 receptor and its ligand represent a feedback loop, which in a paracrine manner, is well known to promote spread of breast cancers. The role of the autocrine feedback loop in promotion of breast tumor behavior, in particular in vitro, is less well understood. The physiologic stimulation of c-fms expression by glucocorticoids (GCs) in vitro and in vivo magnifies the tumor promoting effect seen in these cells from activated c-fms signaling by CSF-1. Targeted molecular therapy against c-fms could therefore abrogate both complementary feedback loops. Using breast cancer cells endogenously co-expressing receptor and ligand, we used complementary approaches to inhibit c-fms expression and function within this autocrine pathway in the context of GC stimulation. Silencing RNA (shRNA), antisense oligonucleotide therapy (AON), and inhibition of c-fms signaling, were all used to quantitate inhibition of GC-stimulated adhesion, motility, and invasion of human breast cancer cells in vitro. shRNA to c-fms downregulated GC-stimulated c-fms mRNA by fourfold over controls, correlating with over twofold reduction in cellular invasiveness. AON therapy was also able to inhibit GC stimulation of c-fms mRNA, and resulted in threefold less invasiveness and 1.5 to 2-fold reductions in adhesion and motility. Finally, the small-molecule c-fms inhibitor Ki20227 was able to decrease in a dose–response manner, breast cancer cell invasion by up to fourfold. Inhibition of this receptor/ligand pair may have clinical utility in inhibition of the autocrine as well as the known paracrine interactions in breast cancer, thus further supporting use of targeted therapies in this disease.