High-density array comparative genomic hybridization (CGH) showed amplification of chromosome 1q22 centered on the RAB25 small GTPase, which is implicated in apical vesicle trafficking, in approximately half of ovarian and breast cancers. RAB25 mRNA levels were selectively increased in stage III and IV serous epithelial ovarian cancers compared to other genes within the amplified region, implicating RAB25 as a driving event in the development of the amplicon. Increased DNA copy number or RNA level of RAB25 was associated with markedly decreased disease-free survival or overall survival in ovarian and breast cancers, respectively. Forced expression of RAB25 markedly increased anchorage-dependent and anchorage-independent cell proliferation, prevented apoptosis and anoikis, including that induced by chemotherapy, and increased aggressiveness of cancer cells in vivo. The inhibition of apoptosis was associated with a decrease in expression of the proapoptotic molecules, BAK and BAX, and activation of the antiapoptotic phosphatidylinositol 3 kinase (PI3K) and AKT pathway, providing potential mechanisms for the effects of RAB25 on tumor aggressiveness. Overall, these studies implicate RAB25, and thus the RAB family of small G proteins, in aggressiveness of epithelial cancers.

Purpose: We aimed to find key molecules associated with chemoresistance in ovarian cancer using gene expression profiling as a screening tool. Experimental Design: Using two newly established paclitaxel-resistant ovarian cancer cell lines from an original paclitaxel-sensitive cell line and four supersensitive and four refractory surgical ovarian cancer specimens from paclitaxel-based chemotherapy, molecules associated with chemoresistance were screened with gene expression profiling arrays containing 39,000 genes. We further analyzed 44 genes that showed significantly different expressions between paclitaxelsensitive samples and paclitaxel-resistant samples with permutation tests, which were common in cell lines and patients' tumors. Results: Eight of these genes showed reproducible results with real-time reverse transcription-PCR, of which indoleamine 2,3-dioxygenase gene expression was the most prominent and consistent. Moreover, by immunohistochemical analysis using a total of 24 serous-type ovarian cancer surgical specimens (stage III, n = 21; stage IV, n = 7), excluding samples used for GeneChip analysis, the Kaplan-Meier survival curve showed a clear relationship between indoleamine 2,3-dioxygenase staining patterns and overall survival (log-rank test, P = 0.0001). All patients classified as negative survived without relapse.The 50% survival of patients classified as sporadic, focal, and diffuse was 41, 17, and 11months, respectively. Conclusion: The indoleamine 2,3-dioxygenase screened with the GeneChip was positively associated with paclitaxel resistance and with impaired survival in patients with serous-type ovarian cancer.

BACKGROUND The authors investigated whether the presence of matrix metalloproteinase‐2 (MMP‐2) and its inducer, CD147, in cancerous esophageal lesions and surrounding tissue might help to predict patient prognosis. METHODS Tissue samples from 101 patients with esophageal squamous cell carcinoma were stained with anti‐CD147 and anti–MMP‐2 antibodies for immunohistochemical analysis. RESULTS CD147 was expressed in cancerous and dysplastic lesions, but not in normal tissue. In contrast, MMP‐2 was detected mainly in normal interstitial tissue adjacent to cancerous lesions, but it was detected also in cancerous lesions in some patients. Pathologic findings demonstrated that the intensity of MMP‐2 staining in normal tissue was associated positively with the depth of tumor infiltration and the stage of disease, whereas MMP‐2 staining in cancerous tissue was associated positively with vascular and lymphatic vessel invasion as well as with immature differentiation of cancer cells. Using a proportional hazard model, including information on CD147 staining patterns within cancerous lesions along with clinical cancer staging, improved the accuracy of predicting patient prognosis. CONCLUSIONS These results suggested that measurement of CD147 and MMP‐2 expression with simple immunohistochemical staining may enhance further the understanding of the pathophysiology of invading tumor cells and, when used in combination with cancer staging, may increase the ability of investigators to predict prognosis in patients with esophageal squamous cell carcinoma. Cancer 2004. © 2004 American Cancer Society.

Purpose: We aimed to develop an ovarian cancer-specific predictive framework for clinical stage, histotype, residual tumor burden, and prognosis using machine learning methods based on multiple biomarkers.Experimental Design: Overall, 334 patients with epithelial ovarian cancer (EOC) and 101 patients with benign ovarian tumors were randomly assigned to "training" and "test" cohorts. Seven supervised machine learning classifiers, including Gradient Boosting Machine (GBM), Support Vector Machine, Random Forest (RF), Conditional RF (CRF), Na€ ve Bayes, Neural Network, and Elastic Net, were used to derive diagnostic and prognostic information from 32 parameters commonly available from pretreatment peripheral blood tests and age.Results: Machine learning techniques were superior to conventional regression-based analyses in predicting multiple clinical parameters pertaining to EOC. Ensemble meth-ods combining weak decision trees, such as GBM, RF, and CRF, showed the best performance in EOC prediction. The values for the highest accuracy and area under the ROC curve (AUC) for segregating EOC from benign ovarian tumors with RF were 92.4% and 0.968, respectively. The highest accuracy and AUC for predicting clinical stages with RF were 69.0% and 0.760, respectively. High-grade serous and mucinous histotypes of EOC could be preoperatively predicted with RF. An ordinal RF classifier could distinguish complete resection from others. Unsupervised clustering analysis identified subgroups among early-stage EOC patients with significantly worse survival.Conclusions: Machine learning systems can provide critical diagnostic and prognostic prediction for patients with EOC before initial intervention, and the use of predictive algorithms may facilitate personalized treatment options through pretreatment stratification of patients.NOTE: There were too few early-stage EOC patients with residual tumor. A definition for the significance of bold is P value of < 0.05.

Abstract. Ovarian clear cell carcinoma (OCCC) has several significant characteristics based on molecular features that are distinct from those of ovarian high-grade serous carcinoma. Cellular glycogen accumulation is the most conspicuous feature of OCCC and in the present study its metabolic mechanism was investigated. The amount of glycogen in cells cultured under hypoxia increased significantly and approximately doubled after 48 h (P<0.01) compared to that under normoxic conditions. Periodic acid-Schiff positive staining also demonstrated intracellular glycogen storage. Western blot analysis revealed that HIF1α, which was overexpressed and stabilized under hypoxic conditions, led to an increase in the levels of cellular glycogen synthase 1, muscle type (GYS1), and conversely to a decrease in inactive phosphorylated GYS1 at serine (Ser) 641. Additional increases were observed in both protein phosphatase 1, which dephosphorylates and thereby induces GYS1 enzyme activity, and glycogen synthase kinase 3 beta (GSK3β) phosphorylated at Ser9, which is inactive on phosphorylation of GYS1 and subsequently induces its enzyme activity. By contrast, the level of PYGM-b decreased. These results indicated that the glycogen accumulation under a hypoxic environment resulted in the promotion of glycogen synthesis, but did not lead to inhibition of glycogen degradation and/or consumption. Under hypoxic conditions, HAC2 cells showed activation of the PI3K/AKT pathway caused by a mutation in exon 20 of PIK3CA, encoding the catalytic subunit p110α of PI3K. The resulting activation of AKT (phosphoSer473) also plays a role as a central enhancer in glycogen synthesis through suppression of GSK3β via phosphorylation at Ser9. Hypoxia decreased the cytocidal activity of cisplatin and doxorubicin to various degrees. In conclusion, the hypoxic conditions together with HIF1 expression and stabilization increased the intracellular glycogen contents and resistance to the anticancer drugs.

To cite this article: Kearon C, Comp P, Douketis J, Royds R, Yamada K, Gent M. Dose-response study of recombinant human soluble thrombomodulin (ART-123) in the prevention of venous thromboembolism after total hip replacement. J Thromb Haemost 2005; 3: 962-8.

Cytokine expression in a tumor microenvironment can impact both host defense against the tumor and tumor cell survival. In this study, we sought to clarify whether the cytokine gene expression profile could have clinical associations with ovarian cancer. We analyzed the expression of 16 cytokine genes (IL-1α, IL-1β, IL-2, IL-4, IL-5, IL-8, IL-10, IL-12p35, IL-12p40, IL-15, IFN-γ, TNF-α, IL-6, HLA-DRA, HLA-DPA1 and CSF1) in 50 ovarian carcinomas. Hierarchical clustering analysis of these tumors was carried out using Cluster software and differentially expressed genes were examined between clear cell carcinoma (CCC) and other subtypes. Following this examination we evaluated the biological significance of IL-6 knockdown in CCC. Unsupervised hierarchical clustering analysis of cytokine gene expression revealed two distinct clusters. The relationship between the two clusters and clinical parameters showed statistically significant differences in CCC compared to other histologies. CCC showed a dominant Th-2 cytokine expression pattern driven largely by IL-6 expression. Inhibition of IL-6 in CCC cells suppressed Stat3 signaling and rendered cells sensitive to cytotoxic agents. The unique cytokine expression pattern found in CCC may be involved in the pathogenesis of this subtype. In particular, high IL-6 expression appears likely to be driven by the tumor cells, fueling an autocrine pathway involving IL-6 expression and Stat3 activation and may influence survival when exposed to cytotoxic chemotherapy. Modulation of IL-6 expression or its related signaling pathway may be a promising strategy of treatment for CCC.

Human epididymis protein 4 (HE4) levels and the Risk of Ovarian Malignancy Algorithm (ROMA) have recently been shown to improve the sensitivity and specificity of epithelial ovarian cancer (EOC) diagnosis. We evaluated HE4 levels and ROMA as diagnostic tools of type I and type II EOC in Japanese women. Women who had a pelvic mass on imaging and were scheduled to undergo surgery were enrolled as ovarian mass patients. Serum levels of carbohydrate antigen 125 (CA125) and HE4 were tested in 319 women (131 benign, 19 borderline, 75 malignant, and 94 healthy controls). CA125, HE4, and ROMA were evaluated for sensitivity and by receiver operating characteristics (ROC) in type I and type II EOC. The results showed that, at 75 % specificity, the sensitivity of CA125 and HE4 for type II was 92.1 % for both markers and for type I was 51.5 % and 78.8 %, respectively. The sensitivities of ROMA (type I, 84.8 % and type II, 97.4 %) were better than those of CA125 and HE4. CA125, HE4, and ROMA were all highly accurate markers for type II. For type I, HE4 and ROMA showed better sensitivity than CA125. ROMA displayed the best diagnostic power for type I and type II including for the early stage of type I. In conclusion, HE4, CA125, and ROMA are valuable markers for type II EOC diagnosis. HE4 and ROMA analyses may improve differentiation between type I EOC and a benign mass. Measurement of combined HE4 and CA125 levels provides a more accurate method for EOC diagnosis.