A personalized probabilistic approach to ovarian cancer diagnostics.

OBJECTIVE: The identification/development of a machine learning-based classifier that utilizes metabolic profiles of serum samples to accurately identify individuals with ovarian cancer. METHODS: Serum samples collected from 431 ovarian cancer patients and 133 normal women at four geographic locations were analyzed by mass spectrometry. Reliable metabolites were identified using recursive feature elimination coupled with repeated cross-validation and used to develop a consensus classifier able to distinguish cancer from non-cancer. The probabilities assigned to individuals by the model were used to create a clinical tool that assigns a likelihood that an individual patient sample is cancer or normal. RESULTS: Our consensus classification model is able to distinguish cancer from control samples with 93% accuracy. The frequency distribution of individual patient scores was used to develop a clinical tool that assigns a likelihood that an individual patient does or does not have cancer. CONCLUSIONS: An integrative approach using metabolomic profiles and machine learning-based classifiers has been employed to develop a clinical tool that assigns a probability that an individual patient does or does not have ovarian cancer. This personalized/probabilistic approach to cancer diagnostics is more clinically informative and accurate than traditional binary (yes/no) tests and represents a promising new direction in the early detection of ovarian cancer.

The Clinical Significance of Genetic Variation in Ovarian Cancer.

Genetic variation is a well-known contributor to the onset and progression of cancer. The goal of this study is to provide a comprehensive examination of the nucleotide and chromosomal variation associated with the onset and progression of serous ovarian cancer. Using a variety of computational and statistical methods, we examine the exome sequence profiles of genetic variants present in the primary tumors of 432 ovarian cancer patient samples to compute: (1) the tumor mutational burden for all genes and (2) the chromosomal copy number alterations associated with the onset/progression of ovarian cancer. Tumor mutational burden is reduced in the late vs. early stages, with the highest levels being associated with loss-of-function mutations in DNA-repair genes. Nucleotide variation and copy number alterations associated with known cancer driver genes are selectively favored over ovarian cancer development. The results indicate that genetic variation is a significant contributor to the onset and progression of ovarian cancer. The measurement of the relative levels of genetic variation associated with individual ovarian cancer patient tumors may be a clinically valuable predictor of potential tumor aggressiveness and resistance to chemotherapy. Tumors found to be associated with high levels of genetic variation may help in the clinical identification of high-risk ovarian cancer patients who could benefit from more frequent monitoring.

Correlating mechanical and gene expression data on the single cell level to investigate metastatic phenotypes.

Stiffness has been observed to decrease for many cancer cell types as their metastatic potential increases. Although cell mechanics and metastatic potential are related, the underlying molecular factors associated with these phenotypes remain unknown. Therefore, we have developed a workflow to measure the mechanical properties and gene expression of single cells that is used to generate large linked-datasets. The process combines atomic force microscopy to measure the mechanics of individual cells with multiplexed RT-qPCR gene expression analysis on the same single cells. Surprisingly, the genes that most strongly correlated with mechanical properties were not cytoskeletal, but rather were markers of extracellular matrix remodeling, epithelial-to-mesenchymal transition, cell adhesion, and cancer stemness. In addition, dimensionality reduction analysis showed that cell clustering was improved by combining mechanical and gene expression data types. The single cell genomechanics method demonstrates how single cell studies can identify molecular drivers that could affect the biophysical processes underpinning metastasis.

Ethnic disparities in mortality and group-specific risk factors in the UK Biobank.

Despite a substantial overall decrease in mortality, disparities among ethnic minorities in developed countries persist. This study investigated mortality disparities and their associated risk factors for the three largest ethnic groups in the United Kingdom: Asian, Black, and White. Study participants were sampled from the UK Biobank (UKB), a prospective cohort enrolled between 2006 and 2010. Genetics, biological samples, and health information and outcomes data of UKB participants were downloaded and data-fields were prioritized based on participants with death registry records. Kaplan-Meier method was used to evaluate survival differences among ethnic groups; survival random forest feature selection followed by Cox proportional-hazard modeling was used to identify and estimate the effects of shared and ethnic group-specific mortality risk factors. The White ethnic group showed significantly worse survival probability than the Asian and Black groups. In all three ethnic groups, endoscopy and colonoscopy procedures showed significant protective effects on overall mortality. Asian and Black women show lower relative risk of mortality than men, whereas no significant effect of sex was seen for the White group. The strongest ethnic group-specific mortality associations were ischemic heart disease for Asians, COVID-19 for Blacks, and cancers of respiratory/intrathoracic organs for Whites. Mental health-related diagnoses, including substance abuse, anxiety, and depression, were a major risk factor for overall mortality in the Asian group. The effect of mental health on Asian mortality, particularly for digestive cancers, was exacerbated by an observed hesitance to answer mental health questions, possibly related to cultural stigma. C-reactive protein (CRP) serum levels were associated with both overall and cause-specific mortality due to COVID-19 and digestive cancers in the Black group, where elevated CRP has previously been linked to psychosocial stress due to discrimination. Our results point to mortality risk factors that are group-specific and modifiable, supporting targeted interventions towards greater health equity.

A computational approach to generate highly conserved gene co-expression networks with RNA-seq data.

We describe a consensus approach for network construction based on fully conserved gene-gene interactions from randomly downsampled data subsets for an unbiased differential analysis of gene co-expression networks. The pipeline allows users to identify network nodes lost, conserved, and acquired in cancer as well as interpret the functional significance of these network changes. For proof of concept, the protocol is used to leverage RNA-seq data of tumor samples from TCGA and healthy tissue samples from the GTEx database. For complete details on the use and execution of this protocol, please refer to Arshad and McDonald (2021).

High throughput, label-free isolation of circulating tumor cell clusters in meshed microwells.

Extremely rare circulating tumor cell (CTC) clusters are both increasingly appreciated as highly metastatic precursors and virtually unexplored. Technologies are primarily designed to detect single CTCs and often fail to account for the fragility of clusters or to leverage cluster-specific markers for higher sensitivity. Meanwhile, the few technologies targeting CTC clusters lack scalability. Here, we introduce the Cluster-Wells, which combines the speed and practicality of membrane filtration with the sensitive and deterministic screening afforded by microfluidic chips. The >100,000 microwells in the Cluster-Wells physically arrest CTC clusters in unprocessed whole blood, gently isolating virtually all clusters at a throughput of >25 mL/h, and allow viable clusters to be retrieved from the device. Using the Cluster-Wells, we isolated CTC clusters ranging from 2 to 100+ cells from prostate and ovarian cancer patients and analyzed a subset using RNA sequencing. Routine isolation of CTC clusters will democratize research on their utility in managing cancer.

Association of Genetic Ancestry and Molecular Signatures with Cancer Survival Disparities: A Pan-Cancer Analysis.

While overall cancer mortality has steadily decreased in recent decades, cancer health disparities among racial and ethnic population groups persist. Here we studied the relationship between cancer survival disparities (CSD), genetic ancestry (GA), and tumor molecular signatures across 33 cancers in a cohort of 9,818 patients. GA correlated with race and ethnicity but showed observable differences in effects on CSD, with significant associations identified in four cancer types: breast invasive carcinoma (BRCA), head and neck squamous cell carcinoma (HNSCC), kidney renal clear cell carcinoma (KIRC), and skin cutaneous carcinoma (SKCM). Differential gene expression and methylation between ancestry groups associated cancer-related genes with CSD, of which, seven protein-coding genes [progestin and adipoQ receptor family member 6 (PAQR6), Lck-interacting transmembrane adaptor 1 (LIME1), Sin3A-associated protein 25 (SAP25), MAX dimerization protein 3 (MXD3), coiled-coil glutamate rich protein 2 (CCER2), refilin A (RFLNA), and cathepsin W (CTSW)] significantly interacted with GA and exacerbated observed survival disparities. These findings indicated that regulatory changes mediated by epigenetic mechanisms have a greater contribution to CSD than population-specific mutations. Overall, we uncovered various molecular mechanisms through which GA might impact CSD, revealing potential population-specific therapeutic targets for groups disproportionately burdened by cancer. SIGNIFICANCE: This large-cohort, multicancer study identifies four cancer types with cancer survival disparities and seven cancer-related genes that interact with genetic ancestry and contribute to disparities.

Changes in gene-gene interactions associated with cancer onset and progression are largely independent of changes in gene expression.

Recent findings indicate that changes underlying cancer onset and progression are not only attributable to changes in DNA structure and expression of individual genes but to changes in interactions among these genes as well. We examined co-expression changes in gene-network structure occurring during the onset and progression of nine different cancer types. Network complexity is generally reduced in the transition from normal precursor tissues to corresponding primary tumors. Cross-tissue cancer network similarity generally increases in early-stage cancers followed by a subsequent loss in cross-tissue cancer similarity as tumors reacquire cancer-specific network complexity. Gene-gene connections remaining stable through cancer development are enriched for "housekeeping" gene functions, whereas newly acquired interactions are associated with established cancer-promoting functions. Surprisingly, >90% of changes in gene-gene network interactions in cancers are not associated with changes in the expression of network genes relative to normal precursor tissues.

Leveraging TCGA gene expression data to build predictive models for cancer drug response.

BACKGROUND: Machine learning has been utilized to predict cancer drug response from multi-omics data generated from sensitivities of cancer cell lines to different therapeutic compounds. Here, we build machine learning models using gene expression data from patients' primary tumor tissues to predict whether a patient will respond positively or negatively to two chemotherapeutics: 5-Fluorouracil and Gemcitabine. RESULTS: We focused on 5-Fluorouracil and Gemcitabine because based on our exclusion criteria, they provide the largest numbers of patients within TCGA. Normalized gene expression data were clustered and used as the input features for the study. We used matching clinical trial data to ascertain the response of these patients via multiple classification methods. Multiple clustering and classification methods were compared for prediction accuracy of drug response. Clara and random forest were found to be the best clustering and classification methods, respectively. The results show our models predict with up to 86% accuracy; despite the study's limitation of sample size. We also found the genes most informative for predicting drug response were enriched in well-known cancer signaling pathways and highlighted their potential significance in chemotherapy prognosis. CONCLUSIONS: Primary tumor gene expression is a good predictor of cancer drug response. Investment in larger datasets containing both patient gene expression and drug response is needed to support future work of machine learning models. Ultimately, such predictive models may aid oncologists with making critical treatment decisions.

Cancer Exacerbates Chemotherapy-Induced Sensory Neuropathy.

For the constellation of neurologic disorders known as chemotherapy-induced peripheral neuropathy, mechanistic understanding and treatment remain deficient. Here, we present the first evidence that chronic sensory neuropathy depends on nonlinear interactions between cancer and chemotherapy. Global transcriptional profiling of dorsal root ganglia revealed differential expression, notably in regulators of neuronal excitability, metabolism, and inflammatory responses, all of which were unpredictable from effects observed with either chemotherapy or cancer alone. Systemic interactions between cancer and chemotherapy also determined the extent of deficits in sensory encoding and ion channel protein expression by single mechanosensory neurons, with the potassium ion channel Kv3.3 emerging as one potential contributor to sensory neuron dysfunction. Validated measures of sensorimotor behavior in awake, behaving animals revealed dysfunction after chronic chemotherapy treatment was exacerbated by cancer. Notably, errors in precise forelimb placement emerged as a novel behavioral deficit unpredicted by our previous study of chemotherapy alone. These original findings identify novel contributors to peripheral neuropathy and emphasize the fundamental dependence of neuropathy on the systemic interaction between chemotherapy and cancer. SIGNIFICANCE: These findings highlight the need to account for pathobiological interactions between cancer and chemotherapy as a major contributor to neuropathy and will have significant and immediate impact on future investigations in this field.

The ability of miRNAs to induce mesenchymal-to-epithelial transition (MET) in cancer cells is highly dependent upon genetic background.

Understanding of the molecular basis of host cell-miRNA interactions is prerequisite to the successful application of miRNAs as potential therapeutic agents. We studied the morphological and molecular consequences of over expression of three sequence divergent miRNAs previously implicated in the mesenchymal-to-epithelial transition process (MET) in three distinct mesenchymal-like cancer cell lines. The ability of miRNAs to induce morphological changes characteristic of MET positively correlated with induced changes in the expression of genes previously implicated in the process. Variability in the responses of different mesenchymal-like cells to over expression of the same miRNAs was attributable to inherent differences in trans-regulatory profiles pre-disposing these cells to miRNA-induced MET. Collectively our results indicate that miRNA-mediated regulation of MET is a highly integrated process that is significantly modulated by the molecular background of individual cells.

MEDICASCY: A Machine Learning Approach for Predicting Small-Molecule Drug Side Effects, Indications, Efficacy, and Modes of Action.

To improve the drug discovery yield, a method which is implemented at the beginning of drug discovery that accurately predicts drug side effects, indications, efficacy, and mode of action based solely on the input of the drug's chemical structure is needed. In contrast, extant predictive methods do not comprehensively address these aspects of drug discovery and rely on features derived from extensive, often unavailable experimental information for novel molecules. To address these issues, we developed MEDICASCY, a multilabel-based boosted random forest machine learning method that only requires the small molecule's chemical structure for the drug side effect, indication, efficacy, and probable mode of action target predictions; however, it has comparable or even significantly better performance than existing approaches requiring far more information. In retrospective benchmarking on high confidence predictions, MEDICASCY shows about 78% precision and recall for predicting at least one severe side effect and 72% precision drug efficacy. Experimental validation of MEDICASCY's efficacy predictions on novel molecules shows close to 80% precision for the inhibition of growth in ovarian, breast, and prostate cancer cell lines. Thus, MEDICASCY should improve the success rate for new drug approval. A web service for academic users is available at http://pwp.gatech.edu/cssb/MEDICASCY.

An atlas of transposable element-derived alternative splicing in cancer.

Transposable element (TE)-derived sequences comprise more than half of the human genome, and their presence has been documented to alter gene expression in a number of different ways, including the generation of alternatively spliced transcript isoforms. Alternative splicing has been associated with tumorigenesis for a number of different cancers. The objective of this study was to broadly characterize the role of human TEs in generating alternatively spliced transcript isoforms in cancer. To do so, we screened for the presence of TE-derived sequences co-located with alternative splice sites that are differentially used in normal versus cancer tissues. We analysed a comprehensive set of alternative splice variants characterized for 614 matched normal-tumour tissue pairs across 13 cancer types, resulting in the discovery of 4820 TE-generated alternative splice events distributed among 723 cancer-associated genes. Short interspersed nuclear elements (Alu) and long interspersed nuclear elements (L1) were found to contribute the majority of TE-generated alternative splice sites in cancer genes. A number of cancer-associated genes, including MYH11, WHSC1 and CANT1, were shown to have overexpressed TE-derived isoforms across a range of cancer types. TE-derived isoforms were also linked to cancer-specific fusion transcripts, suggesting a novel mechanism for the generation of transcriptome diversity via trans-splicing mediated by dispersed TE repeats. This article is part of a discussion meeting issue 'Crossroads between transposons and gene regulation'.

Tumor suppressor genes and allele-specific expression: mechanisms and significance.

Recent findings indicate that allele-specific expression (ASE) at specific cancer driver gene loci may be of importance in onset/progression of the disease. Of particular interest are loss-of-function (LOF) of tumor suppressor gene (TSGs) alleles. While LOF tumor suppressor mutations are typically considered to be recessive, if these mutant alleles can be significantly differentially expressed relative to wild-type alleles in heterozygotes, the clinical consequences could be significant. LOF TSG alleles are shown to be segregating at high frequencies in world-wide populations of normal/healthy individuals. Matched sets of normal and tumor tissues isolated from 233 cancer patients representing four diverse tumor types demonstrate functionally important changes in patterns of ASE in individuals heterozygous for LOF TSG alleles associated with cancer onset/progression. While a variety of molecular mechanisms were identified as potentially contributing to changes in ASE patterns in cancer, changes in DNA copy number and allele-specific alternative splicing possibly mediated by antisense RNA emerged as predominant factors. In conclusion, LOF TSGs are segregating in human populations at significant frequencies indicating that many otherwise healthy individuals are at elevated risk of developing cancer. Changes in ASE between normal and cancer tissues indicates that LOF TSG alleles may contribute to cancer onset/progression even when heterozygous with wild-type functional alleles.

Hybrid negative enrichment of circulating tumor cells from whole blood in a 3D-printed monolithic device.

Isolation and analysis of circulating tumor cells (CTCs) from blood samples present exciting opportunities for basic cancer research and personalized treatment of the disease. While microchip-based negative CTC enrichment offers both sensitive microfluidic cell screening and unbiased selection, conventional microchips are inherently limited by their capacity to deplete a large number of normal blood cells. In this paper, we use 3D printing to create a monolithic device that combines immunoaffinity-based microfluidic cell capture and a commercial membrane filter for negative enrichment of CTCs directly from whole blood. In our device, stacked layers of chemically-functionalized microfluidic channels capture millions of white blood cells (WBCs) in parallel without getting saturated and the leuko-depleted blood is post-filtered with a 3 µm-pore size membrane filter to eliminate anucleated blood cells. This hybrid negative enrichment approach facilitated direct extraction of viable CTCs off the chip on a membrane filter for downstream analysis. Immunofluorescence imaging of enriched cells showed ∼90% tumor cell recovery rate from simulated samples spiked with prostate, breast or ovarian cancer cells. We also demonstrated the feasibility of our approach for processing clinical samples by isolating prostate cancer CTCs directly from a 10 mL whole blood sample.

Bioinformatics analysis of circulating miRNAs related to cancer following spinal cord injury.

Patients with spinal cord injury (SCI) have an increased risk of developing esophageal, bladder and hematologic malignancies compared with the normal population. In the present study, we aimed to identify, through in silico analysis, miRNAs and their target genes related to the three most frequent types of cancer in individuals with SCI. In a previous study, we reported a pattern of expression of miRNAs in 17 sedentary SCI males compared with 22 healthy able-bodied males by TaqMan OpenArray. This list of miRNAs deregulated in SCI patients was uploaded to miRWALK2.0 to predict the target genes and pathways of selected miRNAs. We used Cytoscape software to construct the network displaying the miRNAs and their gene targets. Among the down-regulated miRNAs in SCI, 21, 19 and 20 miRNAs were potentially associated with hematological, bladder and esophageal cancer, respectively, and three target genes (TP53, CCND1 and KRAS) were common to all three types of cancer. The three up-regulated miRNAs were potentially targeted by 18, 15 and 10 genes associated with all three types of cancer. Our current bioinformatics analysis suggests the potential influence of several miRNAs on the development of cancer in SCI. In general, these data may provide novel information regarding potential molecular mechanisms involved in the development of cancer among individuals with SCI. Further studies aiming at understanding how miRNAs contribute to the development of the major cancers that affect patients after SCI may help elucidate the role of these molecules in the pathophysiology of the disease.

Sequence diverse miRNAs converge to induce mesenchymal-to-epithelial transition in ovarian cancer cells through direct and indirect regulatory controls.

Epithelial-to-mesenchymal transition (EMT) has been shown to be similarly regulated by multiple miRNAs, some displaying little or no sequence identity. While alternate models have been proposed to explain the functional convergence of sequence divergent miRNAs, little experimental evidence exists to elucidate the underlying mechanisms involved. Representative members of the miR-200 family of miRNAs and the sequence divergent miR-205 miRNA were independently over expressed in mesenchymal-like ovarian cancer (OC) cells resulting in mesenchymal-to-epithelial transition (MET). The miR-205 and the miR-200 family of miRNAs were found to coordinately induce MET in mesenchymal-like OC cells by affecting both direct and indirect changes in the expression of genes previously associated with EMT/MET. Only two direct targets of these miRNAs (ZEB 1 and WNT5A) are commonly down regulated in response to over-expression of miR-205 and/or the miR-200 family of miRNAs. Down-regulation of these genes, alone or in combination, only partially recapitulates the changes induced by the miRNAs indicating an additional contribution of indirect changes regulated by the miRNAs. Combined gene expression analyses and phylogenetic comparisons suggest an evolutionarily more recent involvement of miR-205 in the EMT/MET process.

Machine learning predicts individual cancer patient responses to therapeutic drugs with high accuracy.

Precision or personalized cancer medicine is a clinical approach that strives to customize therapies based upon the genomic profiles of individual patient tumors. Machine learning (ML) is a computational method particularly suited to the establishment of predictive models of drug response based on genomic profiles of targeted cells. We report here on the application of our previously established open-source support vector machine (SVM)-based algorithm to predict the responses of 175 individual cancer patients to a variety of standard-of-care chemotherapeutic drugs from the gene-expression profiles (RNA-seq or microarray) of individual patient tumors. The models were found to predict patient responses with >80% accuracy. The high PPV of our algorithms across multiple drugs suggests a potential clinical utility of our approach, particularly with respect to the identification of promising second-line treatments for patients failing standard-of-care first-line therapies.