Molecular tumor profiling for prediction of response to anticancer therapies

Sequential feature selection and inference using multi-variate random forests

Motivation

Random forest (RF) has become a widely popular prediction generating mechanism. Its strength lies in its flexibility, interpretability and ability to handle large number of features, typically larger than the sample size. However, this methodology is of limited use if one wishes to identify statistically significant features. Several ranking schemes are available that provide information on the relative importance of the features, but there is a paucity of general inferential mechanism, particularly in a multi-variate set up. We use the conditional inference tree framework to generate a RF where features are deleted sequentially based on explicit hypothesis testing. The resulting sequential algorithm offers an inferentially justifiable, but model-free, variable selection procedure. Significant features are then used to generate predictive RF. An added advantage of our methodology is that both variable selection and prediction are based on conditional inference framework and hence are coherent.

Results

We illustrate the performance of our Sequential Multi-Response Feature Selection approach through simulation studies and finally apply this methodology on Genomics of Drug Sensitivity for Cancer dataset to identify genetic characteristics that significantly impact drug sensitivities. Significant set of predictors obtained from our method are further validated from biological perspective.

Availability and implementation

https://github.com/jomayer/SMuRF.

Contact

souparno.ghosh@ttu.edu.

Supplementary information

Supplementary data are available at Bioinformatics online.

Integration of phytochemicals and phytotherapy into cancer precision medicine

Concepts of individualized therapy in the 1970s and 1980s attempted to develop predictive in vitro tests for individual drug responsiveness without reaching clinical routine. Precision medicine attempts to device novel individual cancer therapy strategies. Using bioinformatics, relevant knowledge is extracted from huge data amounts. However, tumor heterogeneity challenges chemotherapy due to genetically and phenotypically different cell subpopulations, which may lead to refractory tumors. Natural products always served as vital resources for cancer therapy (e.g., Vinca alkaloids, camptothecin, paclitaxel, etc.) and are also sources for novel drugs. Targeted drugs developed to specifically address tumor-related proteins represent the basis of precision medicine. Natural products from plants represent excellent resource for targeted therapies. Phytochemicals and herbal mixtures act multi-specifically, i.e. they attack multiple targets at the same time. Network pharmacology facilitates the identification of the complexity of pharmacogenomic networks and new signaling networks that are distorted in tumors. In the present review, we give a conceptual overview, how the problem of drug resistance may be approached by integrating phytochemicals and phytotherapy into academic western medicine. Modern technology platforms (e.g. “-omics” technologies, DNA/RNA sequencing, and network pharmacology) can be applied for diverse treatment modalities such as cytotoxic and targeted chemotherapy as well as phytochemicals and phytotherapy. Thereby, these technologies represent an integrative momentum to merge the best of two worlds: clinical oncology and traditional medicine. In conclusion, the integration of phytochemicals and phytotherapy into cancer precision medicine represents a valuable asset to chemically synthesized chemicals and therapeutic antibodies.

Metabolomics Applications in Precision Medicine: An Oncological Perspective

Nowadays, cancer therapy remains limited by the conventional one-size-fits-all approach. In this context, treatment decisions are based on the clinical stage of disease but fail to ascertain the individual´s underlying biology and its role in driving malignancy. The identification of better therapies for cancer treatment is thus limited by the lack of sufficient data regarding the characterization of specific biochemical signatures associated with each particular cancer patient or group of patients. Metabolomics approaches promise a better understanding of cancer, a disease characterized by significant alterations in bioenergetic metabolism, by identifying changes in the pattern of metabolite expression in addition to changes in the concentration of individual metabolites as well as alterations in biochemical pathways. These approaches hold the potential of identifying novel biomarkers with different clinical applications, including the development of more specific diagnostic methods based on the characterization of metabolic subtypes, the monitoring of currently used cancer therapeutics to evaluate the response and the prognostic outcome with a given therapy, and the evaluation of the mechanisms involved in disease relapse and drug resistance. This review discusses metabolomics applications in different oncological processes underlining the potential of this omics approach to further advance the implementation of precision medicine in the oncology area.

Emerging applications of metabolomics in drug discovery and precision medicine

Metabolomics is an emerging 'omics' science involving the comprehensive characterization of metabolites and metabolism in biological systems. Recent advances in metabolomics technologies are leading to a growing number of mainstream biomedical applications. In particular, metabolomics is increasingly being used to diagnose disease, understand disease mechanisms, identify novel drug targets, customize drug treatments and monitor therapeutic outcomes. This Review discusses some of the latest technological advances in metabolomics, focusing on the application of metabolomics towards uncovering the underlying causes of complex diseases (such as atherosclerosis, cancer and diabetes), the growing role of metabolomics in drug discovery and its potential effect on precision medicine.

A Copula Based Approach for Design of Multivariate Random Forests for Drug Sensitivity Prediction

Modeling sensitivity to drugs based on genetic characterizations is a significant challenge in the area of systems medicine. Ensemble based approaches such as Random Forests have been shown to perform well in both individual sensitivity prediction studies and team science based prediction challenges. However, Random Forests generate a deterministic predictive model for each drug based on the genetic characterization of the cell lines and ignores the relationship between different drug sensitivities during model generation. This application motivates the need for generation of multivariate ensemble learning techniques that can increase prediction accuracy and improve variable importance ranking by incorporating the relationships between different output responses. In this article, we propose a novel cost criterion that captures the dissimilarity in the output response structure between the training data and node samples as the difference in the two empirical copulas. We illustrate that copulas are suitable for capturing the multivariate structure of output responses independent of the marginal distributions and the copula based multivariate random forest framework can provide higher accuracy prediction and improved variable selection. The proposed framework has been validated on genomics of drug sensitivity for cancer and cancer cell line encyclopedia database.

Predictive value of ex vivo biodynamic imaging in determining response to chemotherapy in dogs with spontaneous non-Hodgkin's lymphomas: a preliminary study

Biodynamic imaging (BDI) is a novel phenotypic cancer profiling technology which optically characterizes changes in subcellular motion within living tumor tissue samples in response to ex vivo treatment with cancer chemotherapy drugs. The purpose of this preliminary study was to assess the ability of ex vivo BDI to predict in vivo clinical response to chemotherapy in ten dogs with naturally-occurring non-Hodgkin's lymphomas. Pre-treatment tumor biopsy samples were obtained from all dogs and treated ex vivo with doxorubicin (10 μM). BDI measured six dynamic biomarkers of subcellular motion from all biopsy samples at baseline and at regular intervals for 9 h following drug application. All dogs subsequently received doxorubicin to treat their lymphomas. Best overall response to and progression-free survival time following chemotherapy were recorded for all dogs. Receiver operating characteristic (ROC) curves were used to determine accuracy and identify possible cut-off values for the BDI-measured biomarkers which could accurately predict those dogs' cancers that would and would not respond to doxorubicin chemotherapy. One biomarker (designated 'MEM') showed 100% discriminative capability for predicting clinical response to doxorubicin (area under the ROC curve = 1.00, 95% CI 0.692-1.000), while other biomarkers also showed promising predictive capability. These preliminary findings suggest that ex vivo BDI can accurately predict treatment outcome following doxorubicin chemotherapy in a spontaneous animal cancer model, and is worthy of further investigation as a technology for personalized cancer medicine.

An Integrated Approach to Anti-Cancer Drug Sensitivity Prediction

A framework for design of personalized cancer therapy requires the ability to predict the sensitivity of a tumor to anticancer drugs. The predictive modeling of tumor sensitivity to anti-cancer drugs has primarily focused on generating functions that map gene expressions and genetic mutation profiles to drug sensitivity. In this paper, we present a new approach for drug sensitivity prediction and combination therapy design based on integrated functional and genomic characterizations. The modeling approach when applied to data from the Cancer Cell Line Encyclopedia shows a significant gain in prediction accuracy as compared to elastic net and random forest techniques based on genomic characterizations. Utilizing a Mouse Embryonal Rhabdomyosarcoma cell culture and a drug screen of 60 targeted drugs, we show that predictive modeling based on functional data alone can also produce high accuracy predictions. The framework also allows us to generate personalized tumor proliferation circuits to gain further insights on the individualized biological pathway.

Inference of dynamic biological networks based on responses to drug perturbations

Drugs that target specific proteins are a major paradigm in cancer research. In this article, we extend a modeling framework for drug sensitivity prediction and combination therapy design based on drug perturbation experiments. The recently proposed target inhibition map approach can infer stationary pathway models from drug perturbation experiments, but the method is limited to a steady-state snapshot of the underlying dynamical model. We consider the inverse problem of possible dynamic models that can generate the static target inhibition map model. From a deterministic viewpoint, we analyze the inference of Boolean networks that can generate the observed binarized sensitivities under different target inhibition scenarios. From a stochastic perspective, we investigate the generation of Markov chain models that satisfy the observed target inhibition sensitivities.

An ensemble based top performing approach for NCI-DREAM drug sensitivity prediction challenge

We consider the problem of predicting sensitivity of cancer cell lines to new drugs based on supervised learning on genomic profiles. The genetic and epigenetic characterization of a cell line provides observations on various aspects of regulation including DNA copy number variations, gene expression, DNA methylation and protein abundance. To extract relevant information from the various data types, we applied a random forest based approach to generate sensitivity predictions from each type of data and combined the predictions in a linear regression model to generate the final drug sensitivity prediction. Our approach when applied to the NCI-DREAM drug sensitivity prediction challenge was a top performer among 47 teams and produced high accuracy predictions. Our results show that the incorporation of multiple genomic characterizations lowered the mean and variance of the estimated bootstrap prediction error. We also applied our approach to the Cancer Cell Line Encyclopedia database for sensitivity prediction and the ability to extract the top targets of an anti-cancer drug. The results illustrate the effectiveness of our approach in predicting drug sensitivity from heterogeneous genomic datasets.

The lignan, (-)-sesamin reveals cytotoxicity toward cancer cells: pharmacogenomic determination of genes associated with sensitivity or resistance

(-)-Sesamin is a lignan present in sesam oil and a number of medicinal plants. It exerts various pharmacological effects, such as prevention of hyperlipidemia, hypertension, and carcinogenesis. Moreover, (-)-sesamin has chemopreventive and anticancer activity in vitro and in vivo. Multidrug resistance (MDR) of tumors leads to fatal treatment outcome in many patients and novel drugs able to kill multidrug-resistant cells are urgently needed. P-glycoprotein (MDR1/ABCB1) is the best known ATP-binding cassette (ABC) drug transporter mediating MDR. ABCB5 is a close relative to ABCB1, which also mediates MDR. We found that the mRNA expressions of ABCB1 and ABCB5 were not related to the 50% inhibition concentrations (IC50) for (-)-sesamin in a panel of 55 cell lines of the National Cancer Institute, USA. Furthermore, (-)-sesamin inhibited ABCB1- or ABCB5-overexpressing cells with similar efficacy than their drug-sensitive parental counterparts. In addition to ABC transporter-mediated MDR, we attempted to identify other molecular determinants of (-)-sesamin resistance. For this reason, we performed COMPARE and hierarchical cluster analyses of the transcriptome-wide microarray-based mRNA expression of the NCI cell panel. Twenty-three genes were identified, whose mRNA expression correlated with the IC50 values for (-)-sesamin. These genes code for proteins of different biological functions, i.e. ribosomal proteins, components of the mitochondrial respiratory chain, proteins involved in RNA metabolism, protein biosynthesis, or glucose and fatty acid metabolism. Subjecting this set of genes to cluster analysis showed that the cell lines were assembled in the resulting dendrogram according to their responsiveness to (-)-sesamin. In conclusion, (-)-sesamin is not involved in MDR mediated by ABCB1 or ABCB5 and may be valuable to bypass chemoresistance of refractory tumors. The microarray expression profile, which predicted sensitivity or resistance of tumor cells to (-)-sesamin consisted of genes, which do not belong to the classical resistance mechanisms to established anticancer drugs.

A new approach for prediction of tumor sensitivity to targeted drugs based on functional data

Background

The success of targeted anti-cancer drugs are frequently hindered by the lack of knowledge of the individual pathway of the patient and the extreme data requirements on the estimation of the personalized genetic network of the patient’s tumor. The prediction of tumor sensitivity to targeted drugs remains a major challenge in the design of optimal therapeutic strategies. The current sensitivity prediction approaches are primarily based on genetic characterizations of the tumor sample. We propose a novel sensitivity prediction approach based on functional perturbation data that incorporates the drug protein interaction information and sensitivities to a training set of drugs with known targets.

Results

We illustrate the high prediction accuracy of our framework on synthetic data generated from the Kyoto Encyclopedia of Genes and Genomes (KEGG) and an experimental dataset of four canine osteosarcoma tumor cultures following application of 60 targeted small-molecule drugs. We achieve a low leave one out cross validation error of <10% for the canine osteosarcoma tumor cultures using a drug screen consisting of 60 targeted drugs.

Conclusions

The proposed framework provides a unique input-output based methodology to model a cancer pathway and predict the effectiveness of targeted anti-cancer drugs. This framework can be developed as a viable approach for personalized cancer therapy.

Transcriptional shift identifies a set of genes driving breast cancer chemoresistance

Background

Distant recurrences after antineoplastic treatment remain a serious problem for breast cancer clinical management, which threats patients’ life. Systemic therapy is administered to eradicate cancer cells from the organism, both at the site of the primary tumor and at any other potential location. Despite this intervention, a significant proportion of breast cancer patients relapse even many years after their primary tumor has been successfully treated according to current clinical standards, evidencing the existence of a chemoresistant cell subpopulation originating from the primary tumor.

Methods/Findings

To identify key molecules and signaling pathways which drive breast cancer chemoresistance we performed gene expression analysis before and after anthracycline and taxane-based chemotherapy and compared the results between different histopathological response groups (good-, mid- and bad-response), established according to the Miller & Payne grading system. Two cohorts of 33 and 73 breast cancer patients receiving neoadjuvant chemotherapy were recruited for whole-genome expression analysis and validation assay, respectively. Identified genes were subjected to a bioinformatic analysis in order to ascertain the molecular function of the proteins they encode and the signaling in which they participate. High throughput technologies identified 65 gene sequences which were over-expressed in all groups (P ≤ 0·05 Bonferroni test). Notably we found that, after chemotherapy, a significant proportion of these genes were over-expressed in the good responders group, making their tumors indistinguishable from those of the bad responders in their expression profile (P ≤ 0.05 Benjamini-Hochgerg`s method).

Conclusions

These data identify a set of key molecular pathways selectively up-regulated in post-chemotherapy cancer cells, which may become appropriate targets for the development of future directed therapies against breast cancer.

Hotspot oncomutations: implications for personalized cancer treatment

Understanding the extent to which specific tumor mutations impact or mediate patient response to particular cancer therapies has become a rapidly increasing area of research. Recent research findings regarding four predominant mutational targets (KRAS, BRAF, EGFR and PIK3CA) show that these tumor mutations have predictive power for identifying which patients are likely to respond to particular therapies, and have prognostic significance irrespective of treatment. However, in this regard, the literature is frequently nuanced and sometimes contradictory. This lack of clarity may be due, at least in part, to the utilization of mutation detection methods with varying sensitivities across studies of different patient populations. Nevertheless, considerable evidence suggests minor tumor subpopulations may be contributing to inappropriate patient stratification, development of resistance to treatment, and the relapse that often follows treatment with molecularly targeted therapies. Consequently, mutant tumor subpopulations need to be considered in order to improve strategies for personalized cancer treatment.

Systems pathology: a critical review

The technological advances of the last twenty years together with the dramatic increase in computational power have injected new life into systems-level thinking in Medicine. This review emphasizes the close relationship of Systems Pathology to Systems Biology and delineates the differences between Systems Pathology and Clinical Systems Pathology. It also suggests an algorithm to support the application of systems-level thinking to clinical research, proposes applying systems-level thinking to the health care systems and forecasts an acceleration of preventive medicine as a result of the coupling of personal genomics with systems pathology.

Role of peroxisome proliferator-activated receptor gamma coactivator 1alpha in AKT/PKB-mediated inhibition of hepatitis B virus biosynthesis

Hepatitis B virus (HBV) transcription and replication are essentially restricted to hepatocytes because liver-enriched transcription factors govern viral RNA synthesis. The level of transcription from the HBV promoters depends on both the transcription factors binding to these regulatory sequence elements and their ability to recruit coactivators capable of mediating assembly of the transcription preinitiation complex containing RNA polymerase II. Nuclear receptors are a primary determinant of HBV pregenomic RNA synthesis and, hence, viral replication. Peroxisome proliferator-activated receptor γ coactivator 1α (PGC1α) enhances the activity of nuclear receptors and, consequently, HBV biosynthesis. PGC1α is also an important target of signal transduction pathways involved in hepatic glucose and lipid homeostasis, suggesting that this coactivator may have an important role in modulating HBV biosynthesis under various physiological conditions. Consistent with this suggestion, v-akt murine thymoma viral oncogene homolog/protein kinase B (AKT/PKB) is shown to modulate PGC1α activity and, hence, HBV transcription and replication in a cell line-specific manner. In addition, AKT can modulate HBV replication in some but not all cell lines at a posttranscriptional step in the viral life cycle. These observations demonstrate that growth and nutritional signals have the capacity to influence viral production, but the magnitude of these effects will depend on the precise cellular context in which they occur.