Predicting survival outcomes in ovarian cancer using gene expression data

ObjectivesCytoreductive surgery is the cornerstone of ovarian cancer (OVCA) treatment. Detractors of initial maximal surgical effort argue that aggressive tumor biology will dictate survival, not the surgical effort. We investigated the role of biology in achieving optimal cytoreduction in serous OVCA using microarray gene expression analysis.MethodsFor the initial model, we used a gene expression signature from a microarray expression analysis of 124 women with serous OVCA, defining optimal cytoreduction as removal of all disease greater than 1 cm (with 64 women having optimal and 60 suboptimal cytoreduction). We then applied this model to 2 independent data sets: the Australian Ovarian Cancer Study (AOCS; 190 samples) and The Cancer Genome Atlas (TCGA; 468 samples). We performed a second analysis, defining optimal cytoreduction as removal of all disease to microscopic residual, using data from AOCS to create the gene signature and validating results in TCGA data set.ResultsOf the 12,718 genes included in the initial analysis, 58 predicted accuracy of cytoreductive surgery 69% of the time (P= 0.005). The performance of this classifier, measured by the area under the receiver operating characteristic curve, was 73%. When applied to TCGA and AOCS, accuracy was 56% (P= 0.16) and 62% (P= 0.01), respectively, with performance at 57% and 65%, respectively. In the second analysis, 220 genes predicted accuracy of cytoreductive surgery in the AOCS set 74% of the time, with performance of 73%. When these results were validated in TCGA set, accuracy was 57% (P= 0.31) and performance was at 62%.ConclusionGene expression data, used as a proxy of tumor biology, do not predict accurately nor consistently the ability to perform optimal cytoreductive surgery. Other factors, including surgical effort, may also explain part of the model. Additional studies integrating more biological and clinical data may improve the prediction model.

Download Full-text

Gene expression data reveal common pathways that characterize the unifocal nature of ovarian cancer

American Journal of Obstetrics and Gynecology ◽

10.1016/j.ajog.2013.08.004 ◽

2013 ◽

Vol 209 (6) ◽

pp. 576.e1-576.e16 ◽

Cited By ~ 6

Author(s):

Douglas C. Marchion ◽

Yin Xiong ◽

Hye Sook Chon ◽

Entidhar Al Sawah ◽

Nadim Bou Zgheib ◽

...

Keyword(s):

Gene Expression ◽

Ovarian Cancer ◽

Gene Expression Data ◽

Expression Data

Download Full-text

Creation of a Human Secretome: A Novel Composite Library of Human Secreted Proteins: Validation Using Ovarian Cancer Gene Expression Data and a Virtual Secretome Array

Clinical Cancer Research ◽

10.1158/1078-0432.ccr-14-3173 ◽

2015 ◽

Vol 21 (21) ◽

pp. 4960-4969 ◽

Cited By ~ 29

Author(s):

Vinod Vathipadiekal ◽

Victoria Wang ◽

Wei Wei ◽

Levi Waldron ◽

Ronny Drapkin ◽

...

Keyword(s):

Gene Expression ◽

Ovarian Cancer ◽

Gene Expression Data ◽

Secreted Proteins ◽

Cancer Gene ◽

Expression Data

Download Full-text

Exploration of Potential miRNA Biomarkers and Prediction for Ovarian Cancer Using Artificial Intelligence

Frontiers in Genetics ◽

10.3389/fgene.2021.724785 ◽

2021 ◽

Vol 12 ◽

Author(s):

Farzaneh Hamidi ◽

Neda Gilani ◽

Reza Arabi Belaghi ◽

Parvin Sarbakhsh ◽

Tuba Edgünlü ◽

...

Keyword(s):

Gene Expression ◽

Ovarian Cancer ◽

Gene Expression Data ◽

Prediction Models ◽

Gynecologic Cancer ◽

External Validation ◽

Small Sample ◽

Validation Dataset ◽

Expression Data ◽

Embedded Algorithms

Ovarian cancer is the second most dangerous gynecologic cancer with a high mortality rate. The classification of gene expression data from high-dimensional and small-sample gene expression data is a challenging task. The discovery of miRNAs, a small non-coding RNA with 18–25 nucleotides in length that regulates gene expression, has revealed the existence of a new array for regulation of genes and has been reported as playing a serious role in cancer. By using LASSO and Elastic Net as embedded algorithms of feature selection techniques, the present study identified 10 miRNAs that were regulated in ovarian serum cancer samples compared to non-cancer samples in public available dataset GSE106817: hsa-miR-5100, hsa-miR-6800-5p, hsa-miR-1233-5p, hsa-miR-4532, hsa-miR-4783-3p, hsa-miR-4787-3p, hsa-miR-1228-5p, hsa-miR-1290, hsa-miR-3184-5p, and hsa-miR-320b. Further, we implemented state-of-the-art machine learning classifiers, such as logistic regression, random forest, artificial neural network, XGBoost, and decision trees to build clinical prediction models. Next, the diagnostic performance of these models with identified miRNAs was evaluated in the internal (GSE106817) and external validation dataset (GSE113486) by ROC analysis. The results showed that first four prediction models consistently yielded an AUC of 100%. Our findings provide significant evidence that the serum miRNA profile represents a promising diagnostic biomarker for ovarian cancer.

Download Full-text

Unsupervised Machine Learning for Data Encoding applied to Ovarian Cancer Transcriptomes

10.1101/855593 ◽

2019 ◽

Author(s):

Tom M George ◽

Pietro Lio

Keyword(s):

Gene Expression ◽

Machine Learning ◽

Ovarian Cancer ◽

Gene Expression Data ◽

Clustering Algorithm ◽

Machine Learning Algorithms ◽

Data Sets ◽

Dimensional Manifold ◽

Expression Data ◽

Wide Range

AbstractMachine learning algorithms are revolutionising how information can be extracted from complex and high-dimensional data sets via intelligent compression. For example, unsupervised Autoen-coders train a deep neural network with a low-dimensional “bottlenecked” central layer to reconstruct input vectors. Variational Autoencoders (VAEs) have shown promise at learning meaningful latent spaces for text, image and more recently, gene-expression data. In the latter case they have been shown capable of capturing biologically relevant features such as a patients sex or tumour type. Here we train a VAE on ovarian cancer transcriptomes from The Cancer Genome Atlas and show that, in many cases, the latent spaces learns an encoding predictive of cisplatin chemotherapy resistance. We analyse the effectiveness of such an architecture to a wide range of hyperparameters as well as use a state-of-the-art clustering algorithm, t-SNE, to embed the data in a two-dimensional manifold and visualise the predictive power of the trained latent spaces. By correlating genes to resistance-predictive encodings we are able to extract biological processes likely responsible for platinum resistance. Finally we demonstrate that variational autoencoders can reliably encode gene expression data contaminated with significant amounts of Gaussian and dropout noise, a necessary feature if this technique is to be applicable to other data sets, including those in non-medical fields.

Download Full-text

Inferring Differential Networks by Integrating Gene Expression Data With Additional Knowledge

Frontiers in Genetics ◽

10.3389/fgene.2021.760155 ◽

2021 ◽

Vol 12 ◽

Author(s):

Chen Liu ◽

Dehan Cai ◽

WuCha Zeng ◽

Yun Huang

Keyword(s):

Gene Expression ◽

Prostate Cancer ◽

Ovarian Cancer ◽

Gene Expression Data ◽

Gene Networks ◽

Biological Knowledge ◽

Platinum Resistance ◽

Expression Data ◽

Androgen Resistance ◽

Differential Networks

Evidences increasingly indicate the involvement of gene network rewiring in disease development and cell differentiation. With the accumulation of high-throughput gene expression data, it is now possible to infer the changes of gene networks between two different states or cell types via computational approaches. However, the distribution diversity of multi-platform gene expression data and the sparseness and high noise rate of single-cell RNA sequencing (scRNA-seq) data raise new challenges for existing differential network estimation methods. Furthermore, most existing methods are purely rely on gene expression data, and ignore the additional information provided by various existing biological knowledge. In this study, to address these challenges, we propose a general framework, named weighted joint sparse penalized D-trace model (WJSDM), to infer differential gene networks by integrating multi-platform gene expression data and multiple prior biological knowledge. Firstly, a non-paranormal graphical model is employed to tackle gene expression data with missing values. Then we propose a weighted group bridge penalty to integrate multi-platform gene expression data and various existing biological knowledge. Experiment results on synthetic data demonstrate the effectiveness of our method in inferring differential networks. We apply our method to the gene expression data of ovarian cancer and the scRNA-seq data of circulating tumor cells of prostate cancer, and infer the differential network associated with platinum resistance of ovarian cancer and anti-androgen resistance of prostate cancer. By analyzing the estimated differential networks, we find some important biological insights about the mechanisms underlying platinum resistance of ovarian cancer and anti-androgen resistance of prostate cancer.

Download Full-text