Prediction of chemotherapeutic response in ovarian cancer with DNA microarray expression profiling

Prediction of Ovarian Cancer Response to Therapy Based on Deep Learning Analysis of Histopathology Images

BACKGROUND

Ovarian cancer remains the leading gynecological cause of cancer mortality. Predicting the sensitivity of ovarian cancer to chemotherapy at the time of pathological diagnosis is a goal of precision medicine research that we have addressed in this study using a novel deep-learning neural network framework to analyze the histopathological images.

METHODS

We have developed a method based on the Inception V3 deep learning algorithm that complements other methods for predicting response to standard platinum-based therapy of the disease. For the study, we used histopathological H&E images (pre-treatment) of high-grade serous carcinoma from The Cancer Genome Atlas (TCGA) Genomic Data Commons portal to train the Inception V3 convolutional neural network system to predict whether cancers had independently been labeled as sensitive or resistant to subsequent platinum-based chemotherapy. The trained model was then tested using data from patients left out of the training process. We used receiver operating characteristic (ROC) and confusion matrix analyses to evaluate model performance and Kaplan-Meier survival analysis to correlate the predicted probability of resistance with patient outcome. Finally, occlusion sensitivity analysis was piloted as a start toward correlating histopathological features with a response.

RESULTS

The study dataset consisted of 248 patients with stage 2 to 4 serous ovarian cancer. For a held-out test set of forty patients, the trained deep learning network model distinguished sensitive from resistant cancers with an area under the curve (AUC) of 0.846 ± 0.009 (SE). The probability of resistance calculated from the deep-learning network was also significantly correlated with patient survival and progression-free survival. In confusion matrix analysis, the network classifier achieved an overall predictive accuracy of 85% with a sensitivity of 73% and specificity of 90% for this cohort based on the Youden-J cut-off. Stage, grade, and patient age were not statistically significant for this cohort size. Occlusion sensitivity analysis suggested histopathological features learned by the network that may be associated with sensitivity or resistance to the chemotherapy, but multiple marker studies will be necessary to follow up on those preliminary results.

CONCLUSIONS

This type of analysis has the potential, if further developed, to improve the prediction of response to therapy of high-grade serous ovarian cancer and perhaps be useful as a factor in deciding between platinum-based and other therapies. More broadly, it may increase our understanding of the histopathological variables that predict response and may be adaptable to other cancer types and imaging modalities.

Gene expression profile association with poor prognosis in epithelial ovarian cancer patients

Ovarian cancer (OC) is the eighth most common type of cancer for women worldwide. The current diagnostic and prognostic routine available for OC management either lack specificity or are very costly. Gene expression profiling has shown to be a very effective tool in exploring new molecular markers for patients with OC, although association of such markers with patient survival and clinical outcome is still elusive. Here, we performed gene expression profiling of different subtypes of OC to evaluate its association with patient overall survival (OS) and aggressive forms of the disease. By global mRNA microarray profiling in a total of 196 epithelial OC patients (161 serous, 15 endometrioid, 11 mucinous, and 9 clear cell carcinomas), we found four candidates-HSPA1A, CD99, RAB3A and POM121L9P, which associated with OS and poor clinicopathological features. The overexpression of all combined was correlated with shorter OS and progression-free survival (PFS). Furthermore, the combination of at least two markers were further associated with advanced grade, chemotherapy resistance, and progressive disease. These results indicate that a panel comprised of a few predictors that associates with a more aggressive form of OC may be clinically relevant, presenting a better performance than one marker alone.

KLF6 Induces Apoptosis in Human Lens Epithelial Cells Through the ATF4-ATF3-CHOP Axis

Background

Many studies have confirmed that high myopia is related to the high prevalence of cataracts, which results from apoptosis of lens epithelial cells (LECs) due to endoplasmic reticulum stress. Krüppel-like factor 6 (KLF6) is a tumor suppressor that is involved in the regulation of cell proliferation and apoptosis.

Purpose

In this study, our purpose was to find the relationship between KLF6-induced apoptosis in LECs and ATF4 (activating transcription factor 4)-ATF3 (activating transcription factor 3)-CHOP (C/EBP homologous protein) signaling pathway.

Methods

KLF6, ATF4, ATF3, and CHOP were ectopically expressed using cDNAs subcloned into the pCDNA3.1+ vector. ATF4, ATF3, and CHOP knockdown were performed by small interfering RNA (siRNA). Expression of relative gene was tested using QT-PCR and western-blot. Then, accompanied by UVB stimulation, cell viability was measured by CCK-8 assay; The cell damage was examined by live & dead staining; The apoptotic markers Bax and Bcl-2 were detected by immunoblotting; Quantitative apoptotic levels were measured with the Apoptosis Detection Kit; The expression level of reactive oxygen-free radical (ROS) was analyzed by DCFH-DA` probe.

Results

Ectopically expressed ATF4, ATF3, and CHOP-induced apoptosis in cells, whereas ATF4, ATF3, and CHOP knockdown by small interfering RNA (siRNA) blocked KLF6-induced apoptosis. In addition, we determined that ATF4 regulates ATF3 and CHOP expression and that ATF3 silencing reduces CHOP upregulation without changing ATF4 levels; however, ATF4 and ATF3 expression was unaffected by blockade of CHOP, suggesting that KLF6 triggers endoplasmic reticulum stress in LECs by mediating the ATF4-ATF3/CHOP axis. Besides, KLF6 overexpression significantly induced LEC apoptosis under UV radiation, as demonstrated by the elevated Bax/Bcl-2 ratio.

Conclusion

The ATF4-ATF3-CHOP pathway plays an important role in KLF6-induced apoptosis in HLECs. Our results increase our understanding of the mechanisms that regulate LEC apoptosis and contribute to the development of a new preventative strategy for cataract.

Association of Somatic Mutations of ADAMTS Genes With Chemotherapy Sensitivity and Survival in High-Grade Serous Ovarian Carcinoma

IMPORTANCE

Chemotherapy response in the majority of patients with ovarian cancer remains unpredictable.

OBJECTIVE

To identify novel molecular markers for predicting chemotherapy response in patients with ovarian cancer.

DESIGN, SETTING, AND PARTICIPANTS

Observational study of genomics and clinical data of high-grade serous ovarian cancer cases with genomic and clinical data made public between 2009 and 2014 via the Cancer Genome Atlas project.

MAIN OUTCOMES AND MEASURES

Chemotherapy response (primary outcome) and overall survival (OS), progression-free survival (PFS), and platinum-free duration (secondary outcome).

RESULTS

In 512 patients with ovarian cancer with available whole-exome sequencing data, mutations from 8 members of the ADAMTS family (ADAMTS mutations) with an overall mutation rate of approximately 10.4% were associated with a significantly higher chemotherapy sensitivity (100% for ADAMTS-mutated vs 64% for ADAMTS wild-type cases; P < .001) and longer platinum-free duration (median platinum-free duration, 21.7 months for ADAMTS-mutated vs 10.1 months for ADAMTS wild-type cases; P = .001). Moreover, ADAMTS mutations were associated with significantly better OS (hazard ratio [HR], 0.54 [95% CI, 0.42-0.89]; P = .01 and median OS, 58.0 months for ADAMTS-mutated vs 41.3 months for ADAMTS wild-type cases) and PFS (HR, 0.42 [95% CI, 0.38-0.70]; P < .001 and median PFS, 31.8 for ADAMTS-mutated vs 15.3 months for ADAMTS wild-type cases). After adjustment by BRCA1 or BRCA2 mutation, surgical stage, residual tumor, and patient age, ADAMTS mutations were significantly associated with better OS (HR, 0.53 [95% CI, 0.32-0.87]; P = .01), PFS (HR, 0.40 [95% CI, 0.25-0.62]; P < .001), and platinum-free survival (HR, 0.45 [95% CI, 0.28-0.73]; P = .001). ADAMTS-mutated cases exhibited a distinct mutation spectrum and were significantly associated with tumors with a higher genome-wide mutation rate than ADAMTS wild-type cases across the whole exome (median mutation number per sample, 121 for ADAMTS-mutated vs 69 for ADAMTS wild-type cases; P < .001).

CONCLUSIONS AND RELEVANCE

ADAMTS mutations may contribute to outcomes in ovarian cancer cases without BRCA1 or BRCA2 mutations and may have important clinical implications.

Identification of Cathepsin K in the Peritoneal Metastasis of Ovarian Carcinoma Using In-silico, Gene Expression Analysis

Ovarian carcinomas (OC) are often found in the advanced stage with wide peritoneal dissemination. Differentially-expressed genes (DEGs) between primary ovarian carcinoma (POC) and peritoneal metastatic ovarian carcinomas (PMOC) may have diagnostic and therapeutic values. In this study, we identified 246 DEGs by in-silico analysis using microarrays for 153 POCs and 57 PMOCs. Pathway analysis shows that many of these genes are associated with lipid metabolism. Microfluidic, card-based, quantitative PCR validated 19 DEGs in PMOCs versus POCs (p<0.05). Immunohistochemistry confirmed overexpression of MMP13, CTSK, FGF1 and GREM1 in PMOCs (p<0.05). ELISA detection indicated that serum CTSK levels were significantly increased in OCs versus controls (p<0.001). CTSK levels discriminated between OCs and healthy controls (ROC 0.739; range 0.685-0.793). Combining CA125 and HE4 with CTSK levels produced an improved specificity in the predictive of OCs (sensitivity 88.3%, specificity 92.0%, Youden's index 80.3%). Our study suggests that CTSK levels may be helpful in the diagnosis of primary, ovarian carcinoma.

A distinct pre-existing inflammatory tumour microenvironment is associated with chemotherapy resistance in high-grade serous epithelial ovarian cancer

Background:

Chemotherapy resistance is a major determinant of poor overall survival rates in high-grade serous ovarian cancer (HGSC). We have previously shown that gene expression alterations affecting the NF-κB pathway characterise chemotherapy resistance in HGSC, suggesting that the regulation of an immune response may be associated with this phenotype.

Methods:

Given that intrinsic drug resistance pre-exists and is governed by both tumour and host factors, the current study was performed to examine the cross-talk between tumour inflammatory microenvironment and cancer cells, and their roles in mediating differential chemotherapy response in HGSC patients. Expression profiling of a panel of 184 inflammation-related genes was performed in 15 chemoresistant and 19 chemosensitive HGSC tumours using the NanoString nCounter platform.

Results:

A total of 11 significantly differentially expressed genes were found to distinguish the two groups. As STAT1 was the most significantly differentially expressed gene (P=0.003), we validated the expression of STAT1 protein by immunohistochemistry using an independent cohort of 183 (52 resistant and 131 sensitive) HGSC cases on a primary tumour tissue microarray. Relative expression levels were subjected to Kaplan–Meier survival analysis and Cox proportional hazard regression models.

Conclusions:

This study confirms that higher STAT1 expression is significantly associated with increased progression-free survival and that this protein together with other mediators of tumour–host microenvironment can be applied as a novel response predictive biomarker in HGSC. Furthermore, an overall underactive immune microenvironment suggests that the pre-existing state of the tumour immune microenvironment could determine response to chemotherapy in HGSC.

Prediction of resistance to chemotherapy in ovarian cancer: a systematic review

Background

Patient response to chemotherapy for ovarian cancer is extremely heterogeneous and there are currently no tools to aid the prediction of sensitivity or resistance to chemotherapy and allow treatment stratification. Such a tool could greatly improve patient survival by identifying the most appropriate treatment on a patient-specific basis.

Methods

PubMed was searched for studies predicting response or resistance to chemotherapy using gene expression measurements of human tissue in ovarian cancer.

Results

42 studies were identified and both the data collection and modelling methods were compared. The majority of studies utilised fresh-frozen or formalin-fixed paraffin-embedded tissue. Modelling techniques varied, the most popular being Cox proportional hazards regression and hierarchical clustering which were used by 17 and 11 studies respectively. The gene signatures identified by the various studies were not consistent, with very few genes being identified by more than two studies. Patient cohorts were often noted to be heterogeneous with respect to chemotherapy treatment undergone by patients.

Conclusions

A clinically applicable gene signature capable of predicting patient response to chemotherapy has not yet been identified. Research into a predictive, as opposed to prognostic, model could be highly beneficial and aid the identification of the most suitable treatment for patients.

Electronic supplementary material

The online version of this article (doi:10.1186/s12885-015-1101-8) contains supplementary material, which is available to authorized users.

Using distance covariance for improved variable selection with application to learning genetic risk models

Variable selection is of increasing importance to address the difficulties of high dimensionality in many scientific areas. In this paper, we demonstrate a property for distance covariance, which is incorporated in a novel feature screening procedure together with the use of distance correlation. The approach makes no distributional assumptions for the variables and does not require the specification of a regression model and hence is especially attractive in variable selection given an enormous number of candidate attributes without much information about the true model with the response. The method is applied to two genetic risk problems, where issues including uncertainty of variable selection via cross validation, subgroup of hard-to-classify cases, and the application of a reject option are discussed.

POSTN/TGFBI-associated stromal signature predicts poor prognosis in serous epithelial ovarian cancer

OBJECTIVE

To identify molecular prognosticators and therapeutic targets for high-grade serous epithelial ovarian cancers (EOCs) using genetic analyses driven by biologic features of EOC pathogenesis.

METHODS

Ovarian tissue samples (n = 172; 122 serous EOCs, 30 other EOCs, 20 normal/benign) collected prospectively from sequential patients undergoing gynecologic surgery were analyzed using RNA expression microarrays. Samples were classified based on expression of genes with potential relevance in ovarian cancer. Gene sets were defined using Rosetta Similarity Search Tool (ROAST) and analysis of variance (ANOVA). Gene copy number variations were identified by array comparative genomic hybridization.

RESULTS

No distinct subgroups of EOC could be identified by unsupervised clustering, however, analyses based on genes correlated with periostin (POSTN) and estrogen receptor-alpha (ESR1) yielded distinct subgroups. When 95 high-grade serous EOCs were grouped by genes based on ANOVA comparing ESR1/WT1 and POSTN/TGFBI samples, overall survival (OS) was significantly shorter for 43 patients with tumors expressing genes associated with POSTN/TGFBI compared to 52 patients with tumors expressing genes associated with ESR1/WT1 (median 30 versus 49 months, respectively; P = 0.022). Several targets with therapeutic potential were identified within each subgroup. BRCA germline mutations were more frequent in the ESR1/WT1 subgroup. Proliferation-associated genes and TP53 status (mutated or wild-type) did not correlate with survival. Findings were validated using independent ovarian cancer datasets.

CONCLUSIONS

Two distinct molecular subgroups of high-grade serous EOCs based on POSTN/TGFBI and ESR1/WT1 expressions were identified with significantly different OS. Specific differentially expressed genes between these subgroups provide potential prognostic and therapeutic targets.

Identification of the IGF1/PI3K/NF κB/ERK gene signalling networks associated with chemotherapy resistance and treatment response in high-grade serous epithelial ovarian cancer

BACKGROUND

Resistance to platinum-based chemotherapy remains a major impediment in the treatment of serous epithelial ovarian cancer. The objective of this study was to use gene expression profiling to delineate major deregulated pathways and biomarkers associated with the development of intrinsic chemotherapy resistance upon exposure to standard first-line therapy for ovarian cancer.

METHODS

The study cohort comprised 28 patients divided into two groups based on their varying sensitivity to first-line chemotherapy using progression free survival (PFS) as a surrogate of response. All 28 patients had advanced stage, high-grade serous ovarian cancer, and were treated with standard platinum-based chemotherapy. Twelve patient tumours demonstrating relative resistance to platinum chemotherapy corresponding to shorter PFS (< eight months) were compared to sixteen tumours from platinum-sensitive patients (PFS > eighteen months). Whole transcriptome profiling was performed using an Affymetrix high-resolution microarray platform to permit global comparisons of gene expression profiles between tumours from the resistant group and the sensitive group.

RESULTS

Microarray data analysis revealed a set of 204 discriminating genes possessing expression levels which could influence differential chemotherapy response between the two groups. Robust statistical testing was then performed which eliminated a dependence on the normalization algorithm employed, producing a restricted list of differentially regulated genes, and which found IGF1 to be the most strongly differentially expressed gene. Pathway analysis, based on the list of 204 genes, revealed enrichment in genes primarily involved in the IGF1/PI3K/NF κB/ERK gene signalling networks.

CONCLUSIONS

This study has identified pathway specific prognostic biomarkers possibly underlying a differential chemotherapy response in patients undergoing standard platinum-based treatment of serous epithelial ovarian cancer. In addition, our results provide a pathway context for further experimental validations, and the findings are a significant step towards future therapeutic interventions.

The changing landscape of therapeutic strategies for recurrent ovarian cancer

Advanced epithelial ovarian cancer, cancer of the fallopian tube and primary peritoneal cancer have a poor prognosis and a high rate of disease recurrence following primary therapy. Recurrent ovarian cancer is currently classified according to sensitivity to platinum-based chemotherapy. Data on targeted therapy provide evidence of improvement with systemic treatment in addition to chemotherapy. Other strategies, although not proven in randomized trials, offer interesting options for future research and therapeutic development. In this review, the covered treatment modalities include surgery, chemotherapy and targeted therapy, immunological approaches and irradiation.

Integrated analysis of gene expression and tumor nuclear image profiles associated with chemotherapy response in serous ovarian carcinoma

Background

Small sample sizes used in previous studies result in a lack of overlap between the reported gene signatures for prediction of chemotherapy response. Although morphologic features, especially tumor nuclear morphology, are important for cancer grading, little research has been reported on quantitatively correlating cellular morphology with chemotherapy response, especially in a large data set. In this study, we have used a large population of patients to identify molecular and morphologic signatures associated with chemotherapy response in serous ovarian carcinoma.

Methodology/Principal Findings

A gene expression model that predicts response to chemotherapy is developed and validated using a large-scale data set consisting of 493 samples from The Cancer Genome Atlas (TCGA) and 244 samples from an Australian report. An identified 227-gene signature achieves an overall predictive accuracy of greater than 85% with a sensitivity of approximately 95% and specificity of approximately 70%. The gene signature significantly distinguishes between patients with unfavorable versus favorable prognosis, when applied to either an independent data set (P = 0.04) or an external validation set (P<0.0001). In parallel, we present the production of a tumor nuclear image profile generated from 253 sample slides by characterizing patients with nuclear features (such as size, elongation, and roundness) in incremental bins, and we identify a morphologic signature that demonstrates a strong association with chemotherapy response in serous ovarian carcinoma.

Conclusions

A gene signature discovered on a large data set provides robustness in accurately predicting chemotherapy response in serous ovarian carcinoma. The combination of the molecular and morphologic signatures yields a new understanding of potential mechanisms involved in drug resistance.

Meta-analysis of gene expression profiles associated with histological classification and survival in 829 ovarian cancer samples

Transcriptomic analysis of global gene expression in ovarian carcinoma can identify dysregulated genes capable to serve as molecular markers for histology subtypes and survival. The aim of our study was to validate previous candidate signatures in an independent setting and to identify single genes capable to serve as biomarkers for ovarian cancer progression. As several datasets are available in the GEO today, we were able to perform a true meta-analysis. First, 829 samples (11 datasets) were downloaded, and the predictive power of 16 previously published gene sets was assessed. Of these, eight were capable to discriminate histology subtypes, and none was capable to predict survival. To overcome the differences in previous studies, we used the 829 samples to identify new predictors. Then, we collected 64 ovarian cancer samples (median relapse-free survival 24.5 months) and performed TaqMan Real Time Polimerase Chain Reaction (RT-PCR) analysis for the best 40 genes associated with histology subtypes and survival. Over 90% of subtype-associated genes were confirmed. Overall survival was effectively predicted by hormone receptors (PGR and ESR2) and by TSPAN8. Relapse-free survival was predicted by MAPT and SNCG. In summary, we successfully validated several gene sets in a meta-analysis in large datasets of ovarian samples. Additionally, several individual genes identified were validated in a clinical cohort.

Microarray-Based Gene Expression Studies in Ovarian Cancer

Background

DNA microarray technology is a powerful genomic tool that has the potential to elucidate the relationship between clinical features of cancers and their underlying biological alterations.

Methods

We performed a systemic search in PubMed and Medline databases for recently published articles. The search terms used included “genome-wide,” “microarrays,” “ovarian cancer,” “prognosis” “gene expression profiling,” “molecular marker,” and “molecular biomarker.”

Results

Genome-wide expression profiling using DNA microarray technology has enhanced our understanding of the genes that influence ovarian cancer development, histopathologic subtype, progression, response to therapy, and overall survival.

Conclusions

Gene expression profiling has demonstrated its utility in ovarian cancer research. It is hoped that with technologic, statistical, and bioinformatic advances, the reliability and reproducibility of this technique will increase, spawning clinical applications that may enhance our understanding of the disease and our ability to care for patients in the future.

[DNA microarrays and prediction of clinical outcome in ovarian carcinoma patients]

Despite debulking surgery and taxane/platinum-based chemotherapy, ovarian cancer is the most lethal pelvic gynaecological cancer in western countries, with a 25% 5-years survival. Current histo-clinical prognostic factors are insufficient to capture the heterogeneous clinical outcome of patients. A better molecular characterization of the disease is crucial to refine the prognostic classifications and to identify new therapeutic targets. DNA microarrays, which allow the quantitative measurement of expression level of the whole genome simultaneously in a single tumor sample, have been recently used towards this objective with promising results. Here, we present and discuss the main published studies and the issues to address in the future to allow the expected transfer to clinical practice.

Emerging roles of Kruppel-like factor 6 and Kruppel-like factor 6 splice variant 1 in ovarian cancer progression and treatment

Epithelial ovarian cancer is one of the most lethal gynecologic cancers and the fifth most frequent cause of female cancer deaths in the United States. Despite dramatic treatment successes in other cancers through the use of molecular agents targeted against genetically defined events driving cancer development and progression, very few insights into epithelial ovarian cancer have been translated from the laboratory to the clinic. If advances are to be made in the early diagnosis, prevention, and treatment of this disease, it will be critical to characterize the common and private (personalized) genetic defects underlying the development and spread of epithelial ovarian cancer. The tumor suppressor Kruppel-like factor 6 and its alternatively spliced, oncogenic isoform, Kruppel-like factor 6 splice variant 1, are members of the Kruppel-like zinc finger transcription factor family of proteins, which have diverse roles in cellular differentiation, development, proliferation, growth-related signal transduction, and apoptosis. Inactivation of Kruppel-like factor 6 and overexpression of Kruppel-like factor 6 splice variant 1 have been associated with the progression of a number of human cancers and even with patient survival. This article summarizes our recent findings demonstrating that a majority of epithelial ovarian cancer tumors have Kruppel-like factor 6 allelic loss and decreased expression coupled with increased expression of Kruppel-like factor 6 splice variant 1. The targeted reduction of Kruppel-like factor 6 in ovarian cancer cell lines results in marked increases in cell proliferation, invasion, tumor growth, angiogenesis, and intraperitoneal dissemination in vivo. In contrast, the inhibition of Kruppel-like factor 6 splice variant 1 decreases cellular proliferation, invasion, angiogenesis, and tumorigenicity; this provides the rationale for its potential therapeutic application. These results and our recent demonstration that the inhibition of Kruppel-like factor 6 splice variant 1 can dramatically prolong survival in a preclinical mouse model of ovarian cancer are reviewed and discussed.

Targeting the epidermal growth factor receptor in epithelial ovarian cancer: current knowledge and future challenges

The epidermal growth factor receptor is overexpressed in up to 60% of ovarian epithelial malignancies. EGFR regulates complex cellular events due to the large number of ligands, dimerization partners, and diverse signaling pathways engaged. In ovarian cancer, EGFR activation is associated with increased malignant tumor phenotype and poorer patient outcome. However, unlike some other EGFR-positive solid tumors, treatment of ovarian tumors with anti-EGFR agents has induced minimal response. While the amount of information regarding EGFR-mediated signaling is considerable, current data provides little insight for the lack of efficacy of anti-EGFR agents in ovarian cancer. More comprehensive, systematic, and well-defined approaches are needed to dissect the roles that EGFR plays in the complex signaling processes in ovarian cancer as well as to identify biomarkers that can accurately predict sensitivity toward EGFR-targeted therapeutic agents. This new knowledge could facilitate the development of rational combinatorial therapies to sensitize tumor cells toward EGFR-targeted therapies.

Gene expression profiles as prognostic markers in women with ovarian cancer

The purpose was to find a gene expression profile that could distinguish short-term from long-term survivors in our collection of serous epithelial ovarian carcinomas. Furthermore, it should be able to stratify in an external validation set. Such a classifier profile will take us a step forward toward investigations for more individualized therapies and the use of gene expression profiles in the clinical practice. RNA from tumor tissue from 43 Danish patients with serous epithelial ovarian carcinoma (11 International Federation of Gynecology and Obstetrics [FIGO] stage I/II, 32 FIGO stage III/IV) was analyzed using Affymetrix U133 plus 2.0 microarrays. A multistep statistical procedure was applied to the data to find the gene set that optimally split the patients into short-term and long-term survivors in a Kaplan-Meier plot. A 14-gene prognostic profile with the ability to distinguish short-term survivors (median overall survival of 32 months) from long-term survivors (median overall survival not yet reached after a median follow-up of 76 months) with a P value of 3.4 x 10 was found. The prognostic gene set was also able to distinguish short-term from long-term survival in patients with advanced disease. Furthermore, its ability to classify in an external validation set was demonstrated. The identified 14-gene prognostic profile was able to predict survival (short- vs long-term survival) with a strength that is better than any other prognostic factor in epithelial ovarian cancer including FIGO stage. This stratification method may form the basis of determinations for new therapeutic approaches, as patients with poor prognosis could obtain the biggest advantage from new treatment modalities.

Integrated genomics of chemotherapy resistant ovarian cancer: a role for extracellular matrix, TGFbeta and regulating microRNAs

Epithelial ovarian cancer is the sixth most common cancer in women worldwide and the most important cause of death from gynaecological cancers in the Western world. Our explorative pathway analysis on seven published gene-sets associated with platinum resistance in ovarian cancer reveals TP53 and transforming growth factor beta as key genes. Furthermore, the extracellular matrix was associated with chemotherapy resistance in ovarian cancer as well as endocrine resistance in breast cancer. Pathway analysis again revealed transforming growth factor beta as a key gene regulating extracellular matrix gene expression. A model is presented based on literature linking transforming growth factor beta, extracellular matrix, integrin signalling, epithelial to mesenchymal transition and regulating microRNAs with a (bivalent) role in chemotherapy response.

Predicting the response of advanced cervical and ovarian tumors to therapy

The management of advanced cervical and ovarian cancers remains a significant challenge as many women fail to respond to recommended therapy, resulting in disease progression and ultimately patient death. Because of tumor heterogeneity, it is rare for all cancers of a particular type to respond to a specific therapy; and, as a result, many patients receive treatment from which they derive little or no benefit, leading to increased morbidity and undue costs. A marker that could rapidly predict or forecast disease outcome would clearly be beneficial in allowing the administration of a tailored regime for each patient while reducing toxicity and cost. Traditional prognostic factors of tumor size, grade, and stage are not ideal for predicting patient outcome, whereas the use of in vitro assays to detect chemosensitivity or resistance has not yet translated into routine clinical practice. Similarly, biomarkers and tumor markers vary in their predictive ability. DNA array technology offers great promise in predicting the response to therapy based on gene expression profiles, and can allow for targeted therapies against specific molecular alterations that cause disease. Imaging techniques, particularly those with the ability to characterize biological tissues and provide functional, structural, and molecular information, have the potential to noninvasively integrate physical and metabolic information. These include F-18-fluorodeoxyglucose positron emission tomography, dynamic contrast-enhanced magnetic resonance imaging, and diffusion weighted magnetic resonance imaging, all techniques that attempt to evaluate and predict therapy response and so influence clinical outcome. This review examines different methods of predicting the response to treatment in advanced cervical and ovarian tumors.

TARGET AUDIENCE

Obstetricians & Gynecologists, Family Physicians.

LEARNING OBJECTIVES

After completion of this article, the reader should be able to describe why prediction of response to therapy for cervical and ovarian cancers is important, describe obstacles to use of in vitro assays to predict outcomes for therapy for ovarian and cervical cancers, and explain potentially new predictive markers.

Gene expression profiling and prediction of clinical outcome in ovarian cancer

Epithelial ovarian cancer is the most lethal gynaecological cancer. Despite debulking surgery and platinum/taxane-based chemotherapy, the prognosis remains poor with approximately 25% 5-year survival. Current histo-clinical prognostic factors are insufficient to capture the complex cascade of events that drive the heterogeneous clinical behaviour of the disease. There is a crucial need to identify new prognostic subclasses of disease as well as new therapeutic targets. Today, DNA microarrays allow the simultaneous and quantitative analysis of the mRNA expression levels of thousands of genes in a tumour sample. They have been applied to ovarian cancer research for predicting initial surgical resectability, survival and response to first-line chemotherapy. The first results are promising. In this review, we describe recent applications of DNA microarrays in ovarian cancer research and discuss some issues to address in the near future to allow the technology to reach its full potential in clinical practice.

A snapshot of microarray-generated gene expression signatures associated with ovarian carcinoma

It was hypothesized that analysis of global gene expression in ovarian carcinoma can identify dysregulated genes that can serve as molecular markers and provide further insight into carcinogenesis and provide the basis for development of new diagnostic tools as well as new targeted therapy protocols. By applying bioinformatics tools for screening of biomedical databases, a gene expression profile databank, specific for ovarian carcinoma, was constructed with utilizable data sets published in 28 studies that applied different array technology platforms. The data sets were divided into four compartments: (i) genes associated with carcinogenesis: in 14 studies, 1881 genes were extracted, 75 genes were identified in more than one study, and only 4 genes (PRKCBP1, SPON1, TACSTD1, and PTPRM) were identified in three studies. (ii) Genes associated with histologic subtypes: in four studies, 463 genes could be identified, but none of them was identified in more than a single study. (iii) Genes associated with therapy response: in seven studies, 606 genes were identified from which 38 were differentially regulated in at least two studies, 3 genes (TMSB4X, GRN, and TJP1) in three studies, and 1 gene (IFITM1) in four studies. (iv) Genes associated with prognosis and progression: 254 genes were found in seven studies. From these genes, merely three were identified in at least two different studies. This snapshot of available gene expression data not only provides independently described potential diagnostic and therapeutic targets for ovarian carcinoma but also emphasizes the drawbacks of the current state of global gene expression analyses in ovarian cancer.

Transcriptome analysis of endocrine tumors: clinical perspectives

There is considerable interest in the application of DNA microarrays to the pathologic evaluation of endocrine neoplasms. Improvements in tumor classification and prognostication, prediction of response to therapy, and comprehensive assessment of tumoral hormone production represent the major anticipated benefits. Here, some of the microarray studies that support the clinical use of transcriptome profiling for endocrine tumors are reviewed. In addition, some of the barriers to clinical implementation are discussed.

Gene-expression profiling in epithelial ovarian cancer

DNA-microarray technology has made it possible to simultaneously analyze the expression of thousands of genes in a small sample of tumor tissue. In epithelial ovarian cancer, gene-expression profiling has been used to provide prognostic information, to predict response to first-line platinum-based chemotherapy, and to discriminate between different histologic subtypes. Furthermore, DNA-microarray technology might permit identification of novel markers for early detection of disease and provide insights into the mechanisms of cancer growth and chemotherapy resistance. In this Review, we summarize the contributions of gene-expression profiling to the diagnosis and management of epithelial ovarian cancer and discuss ways in which this technique could become a useful tool in clinical management.

Genome-wide studies in thyroid neoplasia

There is much interest in the application of genome biology to the field of thyroid neoplasia, despite the relatively low mortality rate associated with thyroid cancer in general. The principal reason for this interest is that the field of thyroid neoplasia stands to benefit from the application of genomic information to address a variety of pathologic and clinical issues. In addition to practical patient care issues, there is an excellent opportunity of expand the basic understanding of thyroid carcinogenesis. In this article, the most relevant genomic work on thyroid tumors performed to date is reviewed along with some general comments about the potential impact of genomic biology on thyroid pathology and the management of patients with thyroid nodules and cancer.

Whole genome expression analysis for biologic rational pathway modeling: application in cancer prognosis and therapy prediction

Using semi-quantitative microarray technology, almost every one of the approximately 30 000 human genes can be analyzed simultaneously with a low rate of false-positives, a high specificity, and a high quantification accuracy. This is supported by data from comparative studies of microarrays and reverse-transcription PCR for established cancer genes including those for epidermal growth factor receptor (EGFR), human epidermal growth factor receptor-2 (HER2/ERBB2), estrogen receptor (ESR1), progesterone receptor (PGR), urokinase-type plasminogen activator (PLAU), and plasminogen activator inhibitor-1 (SERPINE1). As such, semi-quantitative expression data provide an almost completely comprehensive background of biological knowledge that can be applied to cancer diagnostics. In clinical terms, expression profiling may be able to provide significant information regarding (i) the identification of high-risk patients requiring aggressive chemotherapy; (ii) the pathway control of therapy predictive parameters (e.g. ESR1 and HER2); (iii) the discovery of targets for biologically rational therapeutics (e.g. capecitabine and trastuzumab); (iv) additional support for decisions about switching therapy; (v) target discovery; and (vi) the prediction of the course of new therapies in clinical trials. In conclusion, whole genome expression analysis might be able to determine important genes related to cancer progression and adjuvant chemotherapy resistance, especially in the context of new approaches involving primary systemic chemotherapy. In this review, we will survey the current progress in whole genome expression analyses for cancer prognosis and prediction. Special emphasis is given to the approach of combining biostatistical analysis of expression data with knowledge of biochemical and genetic pathways.

The role of gene expression profiling in the clinical management of ovarian cancer

Several studies have addressed the clinical value of gene expression profiling in the field of ovarian cancer. This paper reviews the current status of knowledge that can be derived from such studies. Gene expression profiles can be used to reveal sets of genes that can distinguish normal ovarian tissue from invasive ovarian carcinomas. Independent validation of these sets may result in the identification of (a set of) markers valuable for the detection in an early stage. Microarray analysis has shown that different histological subtypes of ovarian cancer might be partly reflected by a different aetiology through the deregulation and activation of different pathways. In addition, this heterogeneity could therefore also lead to different tumour behaviours. Worldwide, the combination of paclitaxel and platinum chemotherapy has been incorporated in the standard protocol for the management of patients with advanced stage ovarian cancer, although the outcome in individual patients is uncertain. Gene expression profiling was found to be a prognostic tool with respect to chemosensitivity and had a predictive performance of 78-86%. With increasing numbers of data from published reports, access to these data for the reproducibility of its results and pooling becomes more and more important and will possibly lead to more individualisation of therapy.

Comparative proteomic analysis of cisplatin sensitive IGROV1 ovarian carcinoma cell line and its resistant counterpart IGROV1-R10

Ovarian cancer is one of the leading causes of mortality due to gynaecological cancer. Despite a good response to surgery and initial chemotherapy essentially based on cisplatin (cis-diamino-dichloro-platinum(II) (CDDP)) compounds, late tumour detection and frequent recurrences with chemoresistance acquisition are responsible for poor prognosis. Several mechanisms have been implicated in CDDP resistance but they are not sufficient to exhaustively explain this resistance emergence. We applied a proteomic approach based on 2-DE coupled with MS to identify proteins associated with the chemoresistance process. We first established a proteomic pattern of the CDDP sensitive ovarian cell line IGROV1 using MALDI-TOF-MS and PMF. We then compared this 2-D pattern with that of the CDDP-resistant counterpart IGROV1-R10. Among the 40 proteins identified, cytokeratins 8 and 18 and aldehyde dehydrogenase 1 were overexpressed in IGROV1-R10, whereas annexin IV was down-regulated. These observations have been confirmed by Western blotting. The characterization of such variations could lead to the development of new protein markers or to the establishment of new therapeutic strategies. Moreover, the identification of proteins involved in CDDP resistance in ovarian tumours would be useful in completing our understanding on this complex mechanism.

Molecular prognostic markers in ovarian cancer: toward patient-tailored therapy

In ovarian cancer the ceiling seems to be reached with chemotherapeutic drugs. Therefore a paradigm shift is needed. Instead of treating all patients according to standard guidelines, individualized molecular targeted treatment should be aimed for. This means that molecular profiles of the distinct ovarian cancer subtypes should be established. Until recently, most studies trying to identify molecular targets were single-marker studies. The prognostic role of key components of apoptotic and prosurvival pathways such as p53, EGFR, and HER2 has been extensively studied because resistance to chemotherapy is often caused by failure of tumor cells to go into apoptosis. However, it is more than likely that different ovarian cancer subtypes with extensive molecular heterogeneity exist. Therefore, exploration of the potential of specific tumor-targeted therapy, based on expression of a prognostic tumor profile, may be of interest. Recently, new profiling techniques, such as DNA and protein microarrays, have enabled high-throughput screening of tumors. In this review an overview of the current status of prognostic marker and molecular targeting research in ovarian cancer, including microarray studies, is presented.

Increased expression of insulin-like growth factor i is associated with Ara-C resistance in leukemia

Resistance to cytosine arabinoside (Ara-C) is a major problem in the treatment of patients with acute myeloid leukemia (AML). In order to investigate the mechanisms involved in Ara-C resistance, the gene expression profile of Ara-C-resistant K562 human myeloid leukemia cells (K562/AC cells) was compared to that of Ara-C-sensitive K562 cells (K562 cells) by using a cDNA microarray platform. Correspondence analysis demonstrated that insulin-like growth factor I (IGF-I) gene was upregulated in K562/AC cells. The biological significance of IGF-I overexpression was further examined in vitro. When K562 cells were incubated with IGF-I ligand, they were protected from apoptosis induced by Ara-C. In contrast, a significant inhibition of growth and increase of apoptosis of K562/AC cells were induced by IGF-I receptor neutralizing antibody, or suramin, a nonspecific growth factor antagonist. Moreover, from the analysis of 27 AML patients, we have shown that IGF-I expression levels are higher in patients at refractory stage, after Ara-C combined chemotherapy, than those in patients at diagnosis. These results suggest that the inhibition of IGF-I and its downstream pathway is a valuable therapeutic approach to overcome Ara-C resistance in AML.

Expression profiling correlates with treatment response in women with advanced serous epithelial ovarian cancer

The majority of epithelial ovarian carcinomas are of serous subtype, with most women presenting at an advanced stage. Approximately 70% respond to initial chemotherapy but eventually relapse. We aimed to find markers of treatment response that might be suitable for routine use, using the gene expression profile of tumor tissue. Thirty one women with histologically-confirmed late-stage serous ovarian cancer were classified into 3 groups based on response to treatment (nonresponders, responders with relapse less than 12 months and responders with no relapse within 12 months). Gene expression profiles of these specimens were analyzed with respect to treatment response and survival (minimum 36 months follow-up). Patients' clinical features did not correlate with prognosis, or with specific gene expression patterns of their tumors. However women who did not respond to treatment could be distinguished from those who responded with no relapse within 12 months based on 34 gene transcripts (p < 0.02). Poor prognosis was associated with high expression of inhibitor of differentiation-2 (ID2) (p = 0.001). High expression of decorin (DCN) and ID2 together was strongly associated with reduced survival (p = 0.003), with an estimated 7-fold increased risk of dying (95% CI 1.9-29.6; 14 months survival) compared with low expression (44 months). Immunohistochemical analysis revealed both nuclear and cytoplasmic distribution of ID2 in ovarian tumors. High percentage of nuclear staining was associated with poor survival, although not statistically significantly. In conclusion, elevated expression of ID2 and DCN was significantly associated with poor prognosis in a homogeneous group of ovarian cancer patients for whom survival could not be predicted from clinical factors.

Drug resistance in ovarian cancer: The emerging importance of gene transcription and spatio-temporal regulation of resistance

Resistance to carboplatin plus paclitaxel, one of the most active drug combinations in ovarian cancer, is the major barrier to the successful long-term treatment of this disease. Understanding the mechanisms involved is a first step towards rational strategies to overcome drug resistance and is an area of intense research effort. Recent work has identified several gene families which appear to contribute to the evolution of drug resistance and which are involved in regulating DNA damage, apoptosis and survival signalling. These genes may be co-ordinately regulated as part of a gene expression program that confers drug resistance through multiple pathways. The subcellular localisation of the gene products and their kinetic regulation following exposure to chemotherapeutic agents may also play a part in the development of drug resistance. This provides a more complex paradigm for drug resistance in which the steady-state expression of a single gene may not be predictive of response to therapy. Nevertheless, the identification of critical genes, most relevant to the development of clinical drug resistance, is now feasible through microarray analysis of tumour samples, and strategies aimed at the circumvention of resistance can be developed using these data.

Prediction of response to neoadjuvant chemotherapy by sequential F-18-fluorodeoxyglucose positron emission tomography in patients with advanced-stage ovarian cancer

PURPOSE

The aim of this study was to evaluate sequential F-18-fluorodeoxyglucose positron emission tomography (FDG-PET) to predict patient outcome after the first and third cycle of neoadjuvant chemotherapy in advanced-stage (International Federation of Gynecology and Obstetrics stages IIIC and IV) ovarian cancer.

PATIENTS AND METHODS

Thirty-three patients received three cycles of carboplatin-based chemotherapy, followed by cytoreductive surgery. Quantitative FDG-PET of the abdomen and pelvis was acquired before treatment and after the first and third cycle of chemotherapy. Changes in tumoral FDG uptake, expressed as standardized uptake values (SUV), were compared with clinical and histopathologic response; overall survival served as a reference.

RESULTS

A significant correlation was observed between FDG-PET metabolic response after the first (P = .008) and third (P = .005) cycle of chemotherapy and overall survival. By using a threshold for decrease in SUV from baseline of 20% after the first cycle, median overall survival was 38.3 months in metabolic responders compared with 23.1 months in metabolic nonresponders. At a threshold of 55% decrease in SUV after the third cycle median overall survival was 38.9 months in metabolic responders compared with 19.7 months in nonresponders. There was no correlation between clinical response criteria (P = .7) or CA125 response criteria (P = .5) and overall survival. There was only a weak correlation (P = .09) between histopathologic response criteria and overall survival.

CONCLUSION

Sequential FDG-PET predicted patient outcome as early as after the first cycle of neoadjuvant chemotherapy and was more accurate than clinical or histopathologic response criteria including changes in tumor marker CA125. FDG-PET appears to be a promising tool for early prediction of response to chemotherapy.