Analysis In Cancer Research Research Articles

Stable Isotope Tracing Analysis in Cancer Research: Advancements and Challenges in Identifying Dysregulated Cancer Metabolism and Treatment Strategies.

Metabolic reprogramming is a hallmark of cancer, driving the development of therapies targeting cancer metabolism. Stable isotope tracing has emerged as a widely adopted tool for monitoring cancer metabolism both in vitro and in vivo. Advances in instrumentation and the development of new tracers, metabolite databases, and data analysis tools have expanded the scope of cancer metabolism studies across these scales. In this review, we explore the latest advancements in metabolic analysis, spanning from experimental design in stable isotope-labeling metabolomics to sophisticated data analysis techniques. We highlight successful applications in cancer research, particularly focusing on ongoing clinical trials utilizing stable isotope tracing to characterize disease progression, treatment responses, and potential mechanisms of resistance to anticancer therapies. Furthermore, we outline key challenges and discuss potential strategies to address them, aiming to enhance our understanding of the biochemical basis of cancer metabolism.

Integrated approach to generate artificial samples with low tumor fraction for somatic variant calling benchmarking

BackgroundHigh-throughput sequencing (HTS) has become the gold standard approach for variant analysis in cancer research. However, somatic variants may occur at low fractions due to contamination from normal cells or tumor heterogeneity; this poses a significant challenge for standard HTS analysis pipelines. The problem is exacerbated in scenarios with minimal tumor DNA, such as circulating tumor DNA in plasma. Assessing sensitivity and detection of HTS approaches in such cases is paramount, but time-consuming and expensive: specialized experimental protocols and a sufficient quantity of samples are required for processing and analysis. To overcome these limitations, we propose a new computational approach specifically designed for the generation of artificial datasets suitable for this task, simulating ultra-deep targeted sequencing data with low-fraction variants and demonstrating their effectiveness in benchmarking low-fraction variant calling.ResultsOur approach enables the generation of artificial raw reads that mimic real data without relying on pre-existing data by using NEAT, a fine-grained read simulator that generates artificial datasets using models learned from multiple different datasets. Then, it incorporates low-fraction variants to simulate somatic mutations in samples with minimal tumor DNA content. To prove the suitability of the created artificial datasets for low-fraction variant calling benchmarking, we used them as ground truth to evaluate the performance of widely-used variant calling algorithms: they allowed us to define tuned parameter values of major variant callers, considerably improving their detection of very low-fraction variants.ConclusionsOur findings highlight both the pivotal role of our approach in creating adequate artificial datasets with low tumor fraction, facilitating rapid prototyping and benchmarking of algorithms for such dataset type, as well as the important need of advancing low-fraction variant calling techniques.

Abstract 878: Enhancing single-cell RNA sequencing analysis in cancer research: A machine learning framework based on LightGBM for automated cell type annotation

Abstract Single-cell RNA sequencing (scRNA-seq) has been widely used in cancer research to understand the complex gene expression diversity and cancer heterogeneity. However, manual annotation of cell types in the scRNA-seq pipeline is time-consuming and depends on the expertise of analyzers, which can significantly influence the results of downstream analyses. To address this problem, we proposed a novel machine learning framework utilizing the LightGBM model for automated and efficient cell-type annotation of scRNA-seq. Two independent scRNA-seq datasets of non-small cell lung cancer (NSCLC) downloaded from the Gene Expression Omnibus (GEO) were used to train and test our model. A standard procedure is applied to both scRNA-seq datasets for quality control and preprocessing, in which poor-quality cells with low gene expressions or high scores for cellular stress/death were excluded. In addition, Harmony is applied to mitigate batch effects in scRNA-seq that could cause variability due to non-biological factors in experiments. Nine different cell types, including endothelial, epithelial, fibroblast, macrophages, mast, plasma, pulmonary alveolar, B, and T cells, were manually labeled in the two datasets by the providers, which were also examined using gene markers corresponding to different cell types from PanglaoDB and DAVID. These manually labeled cell types were used as the ground truth for training and testing our model. In the training stage, the training dataset (containing 85,000 cells from 44 NSCLC samples) of scRNA-seq was used to train the LightGBM model with its high-variable genes. Then, the model would be evaluated using an independent test dataset (containing 8,000 cells from 18 NSCLC samples) by comparing the automatically predicted and manually labeled cell types. The training result showed that our model could successfully specify the nine different cell types, achieving an overall average accuracy, F1 score, and precision of 0.86 each respectively. In the independent dataset test, the model demonstrated good generalization, showing high predictive performance across all cell types, with an average accuracy, F1 score, and precision of 0.8, 0.78, and 0.8, respectively. Specific to the predictions in the test dataset, we found that some epithelial cells were mistakenly identified as other cell types. This might be because of the complex gene expression patterns exhibited by tumor epithelial cells, making accurate predictions challenging. The proposed machine learning framework facilitates cell labeling and unravels the intricate heterogeneity within lung cancer datasets. The combination of LightGBM and standardized preprocessing establishes a benchmark for high-throughput, accurate single-cell analysis, paving the way for discoveries that are more targeted and have significant clinical impact. Citation Format: Tsung Hsien Chuang, Liang-Chuan Lai, Tzu-Pin Lu, Mong-Hsun Tsai, Hsiang-Han Chen, Eric Y. Chuang. Enhancing single-cell RNA sequencing analysis in cancer research: A machine learning framework based on LightGBM for automated cell type annotation [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 878.

Abstract 6257: Genome wide, high-throughput, high-resolution structural variation detection at low variant allele fraction for oncology samples

Abstract Structural variation detection remains an elusive undertaking. For decades, G-banded karyotyping has remained the standard practice for nearly all hematological malignancy samples. Cancers are diseases of the genome and structural variants (SVs) are a hallmark of nearly all cancers, are often drivers of oncogenicity, and can sometimes be targeted for treatment, as highlighted by the discovery of the Philadelphia chromosome and later the development of targeted kinase inhibitors. One of the most important limitations of karyotyping, its requirement for cells to divide in culture, has prevented its application for most solid tumors and therefore, there is considerably less understanding of the impact of SVs in solid tumors. Optical genome mapping is a technique that is revolutionizing cytogenomics and unlocking the understanding of SVs in the genome at a level never seen before. It not only can detect nearly all variants that a karyotype can, but it can provide 10,000x higher resolution and does not require cell culture. Because of this, it is being adopted for oncology studies across hematological malignancies and solid tumors. This study demonstrates the performance of a new high throughput Stratys instrument for generation of optical genome map data at extraordinarily high coverage depth. With this instrument, 12 samples can be loaded at once, each loaded onto its own cartridge and the instrument can collect 400x genome coverage for each of these samples in 24 hours. The instrument has random access to the cartridges so that completed samples can be swapped out to facilitate optimal instrument run efficiency. To date, this system has been demonstrated to run 60 samples in a week, each at 400x coverage. Data from the instrument can be processed by Bionano Access™ and VIA™ software to call SVs, copy number variants (CNVs), regions of loss of heterozygosity (LOH), on a genome wide basis, and to classify those variants automatically according to guidelines such as the National Comprehensive Cancer Network (NCCN), World Health Organization (WHO), and National Health Services (NHS, United Kingdom). This new high throughput system with advanced bioinformatic data analysis has the potential to further transform structural variation analysis in cancer research. Citation Format: Alex R. Hastie, Pat Lynch. Genome wide, high-throughput, high-resolution structural variation detection at low variant allele fraction for oncology samples [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 6257.

Tracking Karyotype Changes in Treatment-Induced Drug-Resistant Evolution.

Karyotype coding, which encompasses the complete chromosome sets and their topological genomic relationships within a given species, encodes system-level information that organizes and preserves genes' function, and determines the macroevolution of cancer. This new recognition emphasizes the crucial role of karyotype characterization in cancer research. To advance this cancer cytogenetic/cytogenomic concept and its platforms, this study outlines protocols for monitoring the karyotype landscape during treatment-induced rapid drug resistance in cancer. It emphasizes four key perspectives: combinational analyses of phenotype and karyotype, a focus on the entire evolutionary process through longitudinal analysis, a comparison of whole landscape dynamics by including various types of NCCAs (including genome chaos), and the use of the same process to prioritize different genomic scales. This protocol holds promise for studying numerous evolutionary aspects of cancers, and it further enhances the power of karyotype analysis in cancer research.

A narrative review with a step-by-step guide to R software for clinicians: Navigating medical data analysis in cancer research

Cancer causes immense suffering globally, and data constitute the cornerstone of cancer research. Analyzing data is pivotal, but manual analysis of vast datasets within constrained time frames is challenging and error-prone. Even minor inaccuracies can lead to false interpretations, affecting lives. This review explores the free, open-source, and widely acclaimed R software. Our goal was to facilitate data analysis and visualization in the scientific writing of clinical projects. R offers a wide range of features and packages for tasks like data manipulation, cleaning, analysis, and creating informative graphs, including traditional statistics, hypothesis testing, regression, time series, survival analysis, machine learning, and medical image analysis. These capabilities aid in accurate data analysis, facilitating a deeper understanding of cancer mechanisms and predicting outcomes. To prepare this review, we performed an online literature search in Scopus, PubMed, and Google for articles and books related to R software published between March 2012 and January 2024, using specific keywords such as “medical data analysis,” “RStudio,” “statistical software,” “clinical data management,” “R programming,” and “research tools.” Articles, books, and online sources lacking full-text options in English or complete information were excluded. A total of 66 articles and book chapters were retrieved, 22 were excluded, and 44 were included in this review. Through this article, our goal was to provide a user-friendly guide to employing R software for fundamental analysis with dummy data, making it accessible even to non-programmers. This will empower individuals to perform statistical analyses independently, contributing to cancer research with flexibility and accuracy.

Machine learning for survival analysis in cancer research: A comparative study

OverviewSurvival analysis is at the basis of every study in the field of cancer research. As every endeavor in this field aims primarily and eventually to improve patients’ survival time or reduce the potential for recurrence. This article presents a summary of some cancer survival analysis techniques and an up-to-date overview of different implementations of Machine Learning in this area of research. This paper also presents an empirical comparison of selected statistical and Machine Learning approaches on different types of cancer medical datasets. MethodsIn this paper we explore a selection of recent articles that: review the use of Machine Learning in cancer research and/or benchmark the different Machine Learning techniques used in cancer survival analysis. This search resulted in 12 papers that were selected following certain criteria. Our aim is to assess the importance of the use of Machine Learning for survival analysis in cancer research, compared to the statistical methods, and how different Machine Learning techniques may perform in different settings in the context of cancer survival analysis. The techniques were selected based on their popularity. Cox Proportional Hazards with Ridge penalty, Random Survival Forests, Gradient Boosting for Survival Analysis with a CoxPh loss function, linear and kernel Support Vector Machines were applied to 10 different cancer survival datasets. The mean Concordance Index and standard deviation were used to compare the performances of these techniques and the results of these implementations were summarized and analyzed for noticeable patterns or trends. Kaplan-Meier plots were used for the non-parametric survival analysis of the different datasets. ResultsCox Proportional Hazards delivers comparable results with Machine Learning techniques thanks to the Ridge penalty and the different methods for dealing with tied events but fails to produce results in higher dimensional datasets. All techniques benchmarked in the study had comparable performances. The use of prognostic tools when there is a mismatch between the patients and the populations used to train the models may not be advisable since each dataset provides a differently shaped survival curve even when presenting a similar cancer type.

True One Cell Chemical Analysis in Cancer Research: A Review

True One Cell (TOC) analysis Is becoming highly critical for functional studies of cancer cells. This is partially because it is the only form of analysis that provides an avenue for studying the heterogeneity and cell-to-cell variations of individual cancer cells, thus providing unique insight into complex regulatory processes that govern TOC functions within a tumor. Additionally, true one cell techniques are playing an increasingly important role in current attempts to implement TOC metabolomic and proteomic studies, as well as emerging attempts to spatially resolve TOC information. In this review we provide a brief overview of the basis of the field and discuss its applications in TOC metabolomics and proteomics.

Abstract 4284: Enhancing subclonal reconstruction algorithm for resolving complex tumor phylogenies from multi-sample tumor DNA sequencing

Abstract Cancer is characterized by the ongoing accumulation of somatic mutations, providing selective advantages that may lead to dysregulated cellular proliferation. While the cancer genome at diagnosis has been extensively studied, many cancer types still lack strong prognostic biomarkers. The continuous acquisition and selection for driver mutations in a population of cancer cells acts as a Darwinian process, resulting in clonal expansions of progressively more aberrant and fit phenotypes. Reconstructing tumor evolution allows us to understand key events that drive cancer progression and patterns of mutation co-occurrence within clones. These evolutionary features guide our understanding of fundamental mechanisms that lead to disease lethality. Inferring tumor evolution from DNA sequencing data is becoming part of routine analysis in cancer research. As sequencing costs drop, sequencing multiple tumor samples from a patient becomes routine. These multiple samples can represent different spatial regions of a tumor, longitudinal samples from a single region or a combination of both. This provides an opportunity to study tumor evolution in much greater detail and accuracy than was previously feasible through single-sample datasets. The most widely used methods to reconstruct the subclonal evolution of a tumor utilize stochastic-search algorithms. These approaches iterate through a parameter space to select phylogenetic solutions that maximize the likelihood of observed sequencing data. They are optimized for low complexity cases, where the size and number or subclones are relatively limited. As tumor subclonal structure increases in complexity, the parameter space grows exponentially, and stochastic-search algorithms become computationally intractable. For instance, recent benchmarking studies have revealed that many methods fail to reconstruct clone trees for data with as few as ten subclones. To circumvent current computational limitations, we developed a deterministic algorithm for subclonal reconstruction that leverages fundamental principles of cancer biology to encode heuristics that reduce the solution space to biologically plausible phylogenies. When applied to samples (4-36 tumors; median 16) from 12 patients with metastatic breast cancer, our method reduced the average runtime ten-fold. We were able to delineate the evolutionary history of up to 57 distinct subclones per patient, which is infeasible with most current methods. Benchmarking using methods developed for the SMCHet DREAM challenge on real and simulated datasets further quantifies the accuracy, resolution, and scalability. We have thus presented a novel method for rapid and optimized reconstruction of tumor evolutionary histories. Citation Format: Helena Winata, Daniel Knight, Juber A. Patel, Nicholas K. Wang, Pier Selenica, Stefan E. Eng, Caroline Kostrzewa, Jaron Arbet, Yingjie Zhu, Ronglai Shen, Jorge Reis-Filho, Pedram Razafi, Paul C. Boutros. Enhancing subclonal reconstruction algorithm for resolving complex tumor phylogenies from multi-sample tumor DNA sequencing. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 4284.

Abstract LB501: Making discoveries with Kids First Variant DataBase and WorkBench

Abstract The Gabriella Miller Kids First Pediatric Research Program (Kids First) aims at facilitating researchers to uncover new insights into the biology of childhood cancer (CC) and structural birth defects (SBD), including the discovery of shared genetic pathways between these disorders. Kids First has two initiatives, which are whole genome sequencing (WGS) of biospecimens from families with CC/SBD, and establishing Kids First Data Resource. Kids First Data Resource Center (KFDRC) developed Kids First Data Resource Portal (KFDRP; https://portal.kidsfirstdrc.org/), which is a centralized platform to search, view, analyze, and identify currently accessible data from both Kids First and collaborative cohorts, incorporating omics and phenotypic information of 30 studies and 26,300 participants. Recently KFDRC released two new KFDRP components named Variant DataBase (VDB) and Variant WorkBench (VWB), enabling users to query, mangle, analyze and visualize germline genomic variants. The current release includes ~309 million unique variants in a matrix of more than 61.3 billion individual-chromosomal position occurrences from over 11,500 participants in 17 studies. While VDB provides a quick variant summary, VWB supports scripting languages such as Python, Spark, SQL, R, and MarkDown as in-depth analysis tools enabled by Apache Zeppelin notebooks. In addition to variant calls and phenotypic information such as Human Phenotype Ontology (HPO) terms and Mondo IDs, VWB hosts rich external variant annotations in the public domain, such as Cancer Hotspots, ClinVar, COSMIC, dbNSFP, gnomAD, TOPMed, as well as gene-phenotype links provided by OMIM, HPO, Orphanet, and the Deciphering Developmental Disorders Project. Users can also load additional databases within a notebook (such as the subscription-based Human Gene Mutation Database [HGMD]), import custom datasets as temporary query tables, export analysis outputs to local drives, visualize analysis results in multiple chart styles, display local figures, and save notebooks for sharing, further use and Cavatica projects. In an effort to screen tier 1 genes (n=578) from the most recent Cancer Gene Census provided by COSMIC for rare deleterious variants in all six currently available CC cohorts in KFDRP, we identified over 1.2 million germline variants that are with minor allele frequency no more than 0.00001 among gnomAD/TOPMed datasets and with ratio of number of “damaging” predictions over number of all predictions in dbNSFP between 0.5 and 1, or that are already cataloged in the most recent version of ClinVar/HGMD. The whole process took less than two hours which is much faster than conventional methods. These variant tools in KFDRP enable efficient genomic variant analysis in cancer research with advanced big data technology. Citation Format: Yiran Guo, Jeremy Costanza, Christophe Botek, Miguel Brown, David Higgins, Yuankun Zhu, Bailey Farrow, Allison Heath, Adam Resnick, Vincent Ferretti, Kids First Data Resource Center. Making discoveries with Kids First Variant DataBase and WorkBench [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr LB501.

RPS13, a potential universal reference gene for normalisation of gene expression in multiple human normal and cancer tissue samples.

Reference genes are considered stable genes and are used for normalizing the gene expression profile across different cell types; as well as, in normal and diseased samples. However, these gene associates with different biological processes, and hence expression vary in different pathological conditions. Therefore, in the present study, eight different reference genes were used and compared to identify common reference gene usable for an array of different cell types and human cancers. The expression stability of the eight reference genes across eleven normal and cancerous tissues was confirmed through real time-qPCR. Ribosomal protein S13 (RPS13) was found to be a common and stable reference gene across intra- and inter-comparison between various normal and tumor tissue types. Further, TCGA data analysis across and between normal and tumor tissue types also showed minimum deviation in expression of RPS13 gene out of eight routinely used reference genes. RPS13 is the common stable reference gene in normalization for gene expression based analysis in cancer research.

Multi-Parameter Quantitative Imaging of Tumor Microenvironments Reveals Perivascular Immune Niches Associated With Anti-Tumor Immunity.

Tumors are populated by a multitude of immune cell types with varied phenotypic and functional properties, which can either promote or inhibit anti-tumor responses. Appropriate localization and function of these cells within tumors is critical for protective immunity, with CD8 T cell infiltration being a biomarker of disease outcome and therapeutic efficacy. Recent multiplexed imaging approaches have revealed highly complex patterns of localization for these immune cell subsets and the generation of distinct tumor microenvironments (TMEs), which can vary among cancer types, individuals, and within individual tumors. While it is recognized that TMEs play a pivotal role in disease progression, a better understanding of their composition, organization, and heterogeneity, as well as how distinct TMEs are reshaped with immunotherapy, is necessary. Here, we performed spatial analysis using multi-parameter confocal imaging, histocytometry, and CytoMAP to study the microanatomical organization of immune cells in two widely used preclinical cancer models, the MC38 colorectal and KPC pancreatic murine tumors engineered to express human carcinoembryonic antigen (CEA). Immune responses were examined in either unperturbed tumors or after immunotherapy with a CEA T cell bispecific (CEA-TCB) surrogate antibody and anti-PD-L1 treatment. CEA-TCB mono and combination immunotherapy markedly enhanced intra-tumoral cellularity of CD8 T cells, dominantly driven by the expansion of TCF1-PD1+ effector T cells and with more minor increases in TCF1+PD1+ resource CD8 T cells. The majority of infiltrating T cells, particularly resource CD8 T cells, were colocalized with dendritic cells (DCs) or activated MHCII+ macrophages, but largely avoided the deeper tumor nest regions composed of cancer cells and non-activated macrophages. These myeloid cell – T cell aggregates were found in close proximity to tumor blood vessels, generating perivascular immune niches. This perivascular TME was present in untreated samples and markedly increased after CEA-TCB therapy, with its relative abundance positively associated with response to therapy. Together, these studies demonstrate the utility of advanced spatial analysis in cancer research by revealing that blood vessels are key organizational hubs of innate and adaptive immune cells within tumors, and suggesting the likely relevance of the perivascular immune TME in disease outcome.

Methylation-eQTL analysis in cancer research.

DNA methylation is a key epigenetic factor regulating gene expression. While promoter methylation has been well studied, recent publications have revealed that functionally important methylation also occurs in intergenic and distal regions, and varies across genes and tissue types. Given the growing importance of inter-platform integrative genomic analyses, there is an urgent need to develop methods to discover and characterize gene-level relationships between methylation and expression. We introduce a novel sequential penalized regression approach to identify methylation-expression quantitative trait loci (methyl-eQTLs), a term that we have coined to represent, for each gene and tissue type, a sparse set of CpG loci best explaining gene expression and accompanying weights indicating direction and strength of association. Using TCGA and MD Anderson colorectal cohorts to build and validate our models, we demonstrate our strategy better explains expression variability than current commonly used gene-level methylation summaries. The methyl-eQTLs identified by our approach can be used to construct gene-level methylation summaries that are maximally correlated with gene expression for use in integrative models, and produce a tissue-specific summary of which genes appear to be strongly regulated by methylation. Our results introduce an important resource to the biomedical community for integrative genomics analyses involving DNA methylation. We produce an R Shiny app (https://rstudio-prd-c1.pmacs.upenn.edu/methyl-eQTL/) that interactively presents methyl-eQTL results for colorectal, breast and pancreatic cancer. The source R code for this work is provided in the Supplementary Material. Supplementary data are available at Bioinformatics online.

Gene expression and survival analysis in cancer research using online open access platforms: a comparative analysis

The development of genomics and transcriptomics and the potential associated with sharing data related with cancer, led to a growing understanding of cancer biology and to the identification of new biomarkers. Analysis of tumor gene expression and associated patient survival rate enables the dissection of the impact of certain genes in cancer patient ́s survival. For that purpose, it is essential to choose user-friendly platforms, where it is easy to analyze, compare and collect information for a certain set of genes. The goal of this article is to compare the content and utility of five open access online platforms for tumor gene expression and patient survival analysis from TCGA datasets – cBioPortal, USCS Xena, GEPIA, UALCAN and ONCOLNC. We explore these platforms from the point of view of a lay user, assessing their applicability to study differences in gene expression in tumor vs normal tissues, or according to cancer stage, and the impact of such expression patterns on patient survival. Although all five platforms are very intuitive and access to the data is easy, they vary in the information available, results visualization, and statistical tests performed. Therefore, the choice of a platform must take into account the study goals. For some purposes, a combination of platforms may be required.

Abstract 3383: DrOncoRight: A natural language-oriented analytics platform for cancer omics data

Abstract Over the past decade, high-throughput molecular profiling technologies have revolutionized cancer research. Petabytes of omics data (e.g., genomic, transcriptomic, proteomic, epigenomic, and metabolic data) have been generated from thousands of patients, animal models, and cell line samples, especially through some large consortium projects such as The Cancer Genome Atlas (TCGA) and International Cancer Genome Consortium (ICGC). These rich, high-throughput cancer omics data have provided an unprecedented opportunity to characterize cancer-related molecular mechanisms and identify biomarkers and therapeutic targets systematically. However, this tidal wave of molecular data also presents a major challenge for cancer researchers in analyzing the data and obtaining meaningful biological and clinical insights effectively. This is particularly true for a large proportion of researchers who have no or limited bioinformatics and statistical expertise. Many efforts have been made to overcome this challenge. First, many programming languages with specially designed modules or libraries to allow easy analysis and visualization of omics data have been developed. However, these tools still require users to acquire some programming skills, such as Python, R, and Perl, which is not feasible for most experimental researchers. Second, many web-based and stand-alone bioinformatics databases and applications have been developed to allow users to explore and analyze cancer omics data through a user-friendly, interactive interface. But these bioinformatics tools usually focus on one specific type of molecular data, provide only predefined analysis, and do not allow the customization of analytic and visualization tasks. Moreover, users still have to spend considerable time identifying appropriate tools, learning distinct user interfaces/procedures, in addition to keeping track of the status and updates for these quickly evolving tools. As a result, there is still a substantial barrier that prevents a large body of cancer researchers from performing cancer omics data analyses in an intuitive, efficient, and reproducible way. To address this challenge, our team has developed DrOncoRight (https://drbioright.org), an open-access, natural language-oriented, artificial intelligence (AI)-driven analytics platform for analyzing and visualizing cancer omics data. A major attractive feature of this platform is that it allows users to ask biological questions through natural language (text or voice). It automatically understands users' intentions, identifies related cancer genomic datasets, performs diverse analyses, and returns the results in a timely, visually attractive manner. Further, based on user feedback, the platform improves itself through active learning. Taken together, equipped with cutting-edge AI, informatics, and analytic technologies, DrOncoRight represents a revolutionary approach to next-generation cancer science which greatly increases the efficiency and reproducibility of data analysis in cancer research. Citation Format: Jun Li, Hu Chen, Yumeng Wang, Mei-Ju Chen, Han Liang. DrOncoRight: A natural language-oriented analytics platform for cancer omics data [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 3383.

Statistics and pitfalls of trend analysis in cancer research: a review focused on statistical packages.

Trend analysis is the analysis using statistical models to estimate and predict potential trends over time, space or any independent continuous-variable. It has been widely used in epidemiology and public health, but much less so in clinical oncology and basic cancer research. Methodological imitations of the chosen statistical package also appear to result in biased or less rigorous interrogation of cancer-related data. We thus review the basic statistics of trend analysis, commonly used commands of statistical packages and the common pitfalls of conducting trend analysis. Four free and 3 commercial statistical-packages were discussed in depth, including Joinpoint, Epi info, R package, Python, SAS, Stata and SPSS. We hope that this review could serve as a practical yet concise guide for using statistical packages for trend analysis in translational and clinical oncology, and help improve the scientific rigor of trend analyses in these fields. The guide, however, may also be applied to other research fields.

Cancer-Related Gene Signature Selection Based on Boosted Regression for Multilayer Perceptron

Gene expression profiling is a useful technique for analyzing cellular function, and gene expression profiles are widely studied in human cancer research. Gene expression data usually consist of a very large number of features and a relatively small number of samples, and extracting a small number of important features from these data is a major challenge of gene expression-based analysis in cancer research. In this paper, we propose an embedded feature selection algorithm using boosted linear regression-based feature selection. The boosting technique is applied to derive the ensemble feature selector and improve the performance of linear regression-based feature selection. The proposed feature selection algorithm, called boosted regression-based feature selection for the multilayer perceptron (BREG-MLP), repeats the boosted feature selection process to extract the smallest feature subset while maintaining good classification performance. We apply the proposed BREG-MLP to some human cancer-related gene expression data sets for the purpose of extracting important features, and we confirm that BREG-MLP offers improved performance compared to single regression-based feature selection methods.

Metformin Inhibits Cellular Proliferation and Bioenergetics in Colorectal Cancer Patient-Derived Xenografts.

There is increasing preclinical evidence suggesting that metformin, an antidiabetic drug, has anticancer properties against various malignancies, including colorectal cancer. However, the majority of evidence, which was derived from cancer cell lines and xenografts, was likely to overestimate the benefit of metformin because these models are inadequate and require supraphysiologic levels of metformin. Here, we generated patient-derived xenograft (PDX) lines from 2 colorectal cancer patients to assess the properties of metformin and 5-fluorouracil (5-FU), the first-line drug treatment for colorectal cancer. Metformin (150 mg/kg) as a single agent inhibits the growth of both PDX tumors by at least 50% (P < 0.05) when administered orally for 24 days. In one of the PDX models, metformin given concurrently with 5-FU (25 mg/kg) leads to an 85% (P = 0.054) growth inhibition. Ex vivo culture of organoids generated from PDX demonstrates that metformin inhibits growth by executing metabolic changes to decrease oxygen consumption and activating AMPK-mediated pathways. In addition, we also performed genetic characterizations of serial PDX samples with corresponding parental tissues from patients using next-generation sequencing (NGS). Our pilot NGS study demonstrates that PDX represents a useful platform for analysis in cancer research because it demonstrates high fidelity with parental tumor. Furthermore, NGS analysis of PDX may be useful to determine genetic identifiers of drug response. This is the first preclinical study using PDX and PDX-derived organoids to investigate the efficacy of metformin in colorectal cancer. Mol Cancer Ther; 16(9); 2035-44. ©2017 AACR.

Ovarian cancer study dropouts had worse health-related quality of life and psychosocial symptoms at baseline and over time.

Participant drop out is a major barrier to high-quality patient-reported outcome (PRO) data analysis in cancer research as patients with worsening health are more likely to dropout. To test the hypothesis that ovarian cancer patients with worse PROs would drop out earlier, we examined how patients differed by time of dropout on health-related quality of life (HRQOL), anxiety, depression, optimism and insomnia. This analysis included 619 participants, stratified by time of dropout, from the Australian Ovarian Cancer Study - Quality of Life substudy, in which participants completed PRO questionnaires at three-monthly intervals for 21 months. Trends in PROs over time were examined. Pearson correlations examined the relationship between time of dropout and baseline PROs. Multiple linear regression models including age, disease stage and time since diagnosis examined relationships between baseline and final PRO scores, and final PRO scores and dropout group. Participants who dropped out earlier had significantly worse baseline HRQOL (p<0.0001) and higher depression (p<0.0001). For all five PROs, final scores were significantly associated with baseline scores (p<0.0001). Time of dropout was significantly associated with final HRQOL (p=0.003), anxiety (p=0.05), depression (p=0.02) and optimism (p=0.02) scores. Depression, HRQOL and anxiety worsened at a faster rate overtime in dropouts than study completers. Poorer HRQOL and higher depression at baseline, and final HRQOL, anxiety, depression and optimism scores were predictive of time of dropout. These results highlight the importance of collecting auxiliary data to inform careful and considered handling of missing PRO data during analysis, interpretation and reporting.

Phospholipids as cancer biomarkers: Mass spectrometry-based analysis.

Lipids, particularly phospholipids (PLs), are key components of cellular membrane. PLs play important and diverse roles in cells such as chemical-energy storage, cellular signaling, cell membranes, and cell-cell interactions in tissues. All these cellular processes are pertinent to cells that undergo transformation, cancer progression, and metastasis. Thus, there is a strong possibility that some classes of PLs are expected to present in cancer cells and tissues in cellular physiology. The mass spectrometric soft-ionization techniques, electrospray ionization (ESI), and matrix-assisted laser desorption/ionization (MALDI) are well-established in the proteomics field, have been used for lipidomic analysis in cancer research. This review focused on the applications of mass spectrometry (MS) mainly on ESI-MS and MALDI-MS in the structural characterization, molecular composition and key roles of various PLs present in cancer cells, tissues, blood, and urine, and on their importance for cancer-related problems as well as challenges for development of novel PL-based biomarkers. The profiling of PLs helps to rationalize their functions in biological systems, and will also provide diagnostic information to elucidate mechanisms behind the control of cancer, diabetes, and neurodegenerative diseases. The investigation of cellular PLs with MS methods suggests new insights on various cancer diseases and clinical applications in the drug discovery and development of biomarkers for various PL-related different cancer diseases. PL profiling in tissues, cells and body fluids also reflect the general condition of the whole organism and can indicate the existence of cancer and other diseases. PL profiling with MS opens new prospects to assess alterations of PLs in cancer, screening specific biomarkers and provide a basis for the development of novel therapeutic strategies. © 2016 Wiley Periodicals, Inc. Mass Spec Rev 37:107-138, 2018.