Cancer Biomarker Discovery Research Articles

Investigation of the urinary metabolic variations and the application in bladder cancer biomarker discovery

Urine metabolomics have been used to identify biomarkers for clinical diseases. However, inter-individual variations and effect factors need to be further evaluated. In our study, we explored the urine metabolome in a cohort of 203 health adults, 6 patients with benign bladder lesions, and 53 patients with bladder cancer (BCa) using liquid chromatography coupled with high resolution mass spectrometry. Inter-individual analysis of both healthy controls and BCa patients showed that the urine metabolome was relatively stable. Further analysis indicated that sex and age affect inter-individual variations in urine metabolome. Metabolic pathways such as tryptophan metabolism, the citrate cycle, and pantothenate and CoA biosynthesis were found to be related to sex and age. To eliminate age and sex interference, additional BCa urine metabolomic biomarkers were explored using age and sex-matched urine samples (Test group: 44 health adults vs. 33 patients with BCa). Metabolic profiling of urine could significantly differentiate the cases with cancer from the controls and high-grade from low-grade BCa. A metabolite panel consisting of trans-2-dodecenoylcarnitine, serinyl-valine, feruloyl-2-hydroxyputrescine, and 3-hydroxynonanoyl carnitine were discovered to have good predictive ability for BCa with an area under the curve (AUC) of 0.956 (cross validation: AUC = 0.924). A panel of indolylacryloylglycine, N2 -galacturonyl-L-lysine, and aspartyl-glutamate was used to establish a robust model for high- and low-grade BCa distinction with AUC of 0.937 (cross validation: AUC = 0.891). External sample (26 control vs. 20 BCa) validation verified the acceptable accuracy of these models for BCa detection. Our study showed that urinary metabolomics is a useful strategy for differential analysis and biomarker discovery.

MassImager: A software for interactive and in-depth analysis of mass spectrometry imaging data

Mass spectrometry imaging (MSI) has become a powerful tool to probe molecule events in biological tissue. However, it is a widely held viewpoint that one of the biggest challenges is an easy-to-use data processing software for discovering the underlying biological information from complicated and huge MSI dataset. Here, a user-friendly and full-featured MSI software including three subsystems, Solution, Visualization and Intelligence, named MassImager, is developed focusing on interactive visualization, in-situ biomarker discovery and artificial intelligent pathological diagnosis. Simplified data preprocessing and high-throughput MSI data exchange, serialization jointly guarantee the quick reconstruction of ion image and rapid analysis of dozens of gigabytes datasets. It also offers diverse self-defined operations for visual processing, including multiple ion visualization, multiple channel superposition, image normalization, visual resolution enhancement and image filter. Regions-of-interest analysis can be performed precisely through the interactive visualization between the ion images and mass spectra, also the overlaid optical image guide, to directly find out the region-specific biomarkers. Moreover, automatic pattern recognition can be achieved immediately upon the supervised or unsupervised multivariate statistical modeling. Clear discrimination between cancer tissue and adjacent tissue within a MSI dataset can be seen in the generated pattern image, which shows great potential in visually in-situ biomarker discovery and artificial intelligent pathological diagnosis of cancer. All the features are integrated together in MassImager to provide a deep MSI processing solution at the in-situ metabolomics level for biomarker discovery and future clinical pathological diagnosis.

Bioapplications of Cell-SELEX-Generated Aptamers in Cancer Diagnostics, Therapeutics, Theranostics and Biomarker Discovery: A Comprehensive Review.

Currently, functional single-stranded oligonucleotide probes, termed aptamers, generated by an iterative technology, Systematic Evolution of Ligands by Exponential Enrichment (SELEX), are utilized to selectively target molecules or cells with high affinity. Aptamers hold considerable promise as multifunctional molecules or conjugates for challenging nanotechnologies or bioapplications now and in the future. In this review, we first describe recent endeavors to select aptamers towards live cancer cells via cell-SELEX. We then introduce several characteristic applications of selected aptamers, especially in imaging, drug delivery and therapy. In part, these advances have been made possible via synthesis of aptamer-based nanomaterials, which, by their sizes, shapes, and physicochemical properties, allow such aptamer-nanomaterial complexes to function as signal reporters or drug carriers. We also describe how these aptamer-based molecular tools contribute to cancer biomarker discovery through high-affinity recognition of membrane protein receptors.

Innovative methods for biomarker discovery in the evaluation and development of cancer precision therapies.

The discovery of biomarkers able to detect cancer at an early stage, to evaluate its aggressiveness, and to predict the response to therapy remains a major challenge in clinical oncology and precision medicine. In this review, we summarize recent achievements in the discovery and development of cancer biomarkers. We also highlight emerging innovative methods in biomarker discovery and provide insights into the challenges faced in their evaluation and validation.

Multiplex glycan bead array for high throughput and high content analyses of glycan binding proteins

Glycan-binding proteins (GBPs) play critical roles in diverse cellular functions such as cell adhesion, signal transduction and immune response. Studies of the interaction between GBPs and glycans have been hampered by the availability of high throughput and high-content technologies. Here we report multiplex glycan bead array (MGBA) that allows simultaneous analyses of 384 samples and up to 500 glycans in a single assay. The specificity, sensitivity and reproducibility of MGBA are evaluated using 39 plant lectins, 13 recombinant anti-glycan antibodies, and mammalian GBPs. We demonstrate the utility of this platform by the analyses of natural anti-glycan IgM and IgG antibodies in 961 human serum samples and the discovery of anti-glycan antibody biomarkers for ovarian cancer. Our data indicate that the MGBA platform is particularly suited for large population-based studies that require the analyses of large numbers of samples and glycans.

Integrated analysis of fine-needle-aspiration cystic fluid proteome, cancer cell secretome, and public transcriptome datasets for papillary thyroid cancer biomarker discovery.

Thyroid ultrasound and ultrasound-guided fine-needle aspiration (USG/FNA) biopsy are currently used for diagnosing papillary thyroid carcinoma (PTC), but their detection limit could be improved by combining other biomarkers. To discover novel PTC biomarkers, we herein applied a GeLC-MS/MS strategy to analyze the proteome profiles of serum-abundant-protein-depleted FNA cystic fluid from benign and PTC patients, as well as two PTC cell line secretomes. From them, we identified 346, 488, and 2105 proteins, respectively. Comparative analysis revealed that 191 proteins were detected in the PTC but not the benign cystic fluid samples, and thus may represent potential PTC biomarkers. Among these proteins, 101 were detected in the PTC cell line secretomes, and seven of them (NPC2, CTSC, AGRN, GPNMB, DPP4, ERAP2, and SH3BGRL3) were reported in public PTC transcriptome datasets as having 4681 elevated mRNA expression in PTC. Immunoblot analysis confirmed the elevated expression levels of five proteins (NPC2, CTSC, GPNMB, DPP4, and ERAP2) in PTC versus benign cystic fluids. Immunohistochemical studies from near 100 pairs of PTC tissue and their adjacent non-tumor counterparts further showed that AGRN (n = 98), CTSC (n = 99), ERAP2 (n = 98) and GPNMB (n = 100) were significantly (p < 0.05) overexpressed in PTC and higher expression levels of AGRN and CTSC were also significantly associated with metastasis and poor prognosis of PTC patients. Collectively, our results indicate that an integrated analysis of FNA cystic fluid proteome, cancer cell secretome and tissue transcriptome datasets represents a useful strategy for efficiently discovering novel PTC biomarker candidates.

Tissue and plasma proteomics for early stage cancer detection.

The pursuit of novel and effective biomarkers is essential in the struggle against cancer, which is a leading cause of mortality worldwide. Biomarkers can be used as specific diagnostic tools, prognostic predictors, markers of the development of therapeutic resistance or even as therapeutic targets themselves. Through the application of sensitive and specific proteomic techniques, oncoproteomics investigates the proteins associated with cancer processes, to better understand their biological function/s and their associated pathways during tumorigenesis. Such studies seek to identify both potential biomarkers and drug targets in order to improve patient survival and quality of life whilst reducing the global health budget. Tissue and plasma are the most commonly utilised biological samples for such studies as they are readily available, non-invasive and generally acceptable. Here, we outline the relative advantages and disadvantages of the most frequently used techniques for cancer diagnosis, prognosis, treatment and surveillance, concentrating on the latest advances and application of tissue and plasma proteomics for novel cancer biomarker discovery and disease surveillance.

Traps and trumps from adjacent-to-tumor samples in gastric cancer research.

The search for cancer biomarkers is frequently based on comparisons between tumors and adjacent-to-tumor samples. However, even after histological confirmation of been free of cancer cells, these adjacent-to-tumor samples might harbor molecular alterations which are not sufficient to cause them to look like cancer, but can differentiate these cells from normal cells. When comparing them, potential biomarkers are missed, and mainly the opportunity of finding initial aberrations presents in both tumors and adjacent samples, but not in true normal samples from non-cancer patients, resulting in misinterpretations about the carcinogenic process. Nevertheless, collecting adjacent-to-tumor samples brings trumps to be explored. The addition of samples from non-cancer patients opens an opportunity to increase the finds of the molecular cascade of events in the carcinogenic process. Differences between normal samples and adjacent samples might represent the first steps of the carcinogenic process. Adding samples of non-cancer patients to the analysis of molecular alterations relevant to the carcinogenic process opens a new window of opportunities to the discovery of cancer biomarkers and molecular targets.

Gelatin Nanoparticle-Coated Silicon Beads for Density-Selective Capture and Release of Heterogeneous Circulating Tumor Cells with High Purity.

Background: Circulating tumor cells (CTCs) are a burgeoning topic in cancer biomarker discovery research with minimal invasive blood draws. CTCs can be used as potential biomarkers for disease prognosis, early cancer diagnosis and pharmacodynamics. However, the extremely low abundance of CTCs limits their clinical utility because of technical challenges such as the isolation and subsequent detailed molecular and functional characterization of rare CTCs from patient blood samples.Methods: In this study, we present a novel density gradient centrifugation method employing biodegradable gelatin nanoparticles coated on silicon beads for the isolation, release, and downstream analysis of CTCs from colorectal and breast cancer patients.Results: Using clinical patient/spiked samples, we demonstrate that this method has significant CTC-capture efficiency (>80%) and purity (>85%), high CTC release efficiency (94%) and viability (92.5%). We also demonstrate the unparalleled robustness of our method in downstream CTC analyses such as the detection of PIK3CA mutations.Conclusion: The efficiency and versatility of the multifunctional density microbeads approach provides new opportunities for personalized cancer diagnostics and treatments.

NGS approach for new ovarian cancer biomarker discovery Ovarian cancer biomarker discovery

NGS approach for new ovarian cancer biomarker discovery Ovarian cancer biomarker discovery

Designation of fingerprint glycopeptides for targeted glycoproteomic analysis of serum haptoglobin: insights into gastric cancer biomarker discovery.

Gastric cancer (GC) is one of the leading causes of cancer-related death worldwide, largely because of difficulties in early diagnosis. Despite accumulating evidence indicating that aberrant glycosylation is associated with GC, site-specific localization of the glycosylation to increase specificity and sensitivity for clinical use is still an analytical challenge. Here, we created an analytical platform with a targeted glycoproteomic approach for GC biomarker discovery. Unlike the conventional glycomic approach with untargeted mass spectrometric profiling of released glycan, our platform is characterized by three key features: it is a target-protein-specific, glycosylation-site-specific, and structure-specific platform with a one-shot enzyme reaction. Serum haptoglobin enriched by immunoaffinity chromatography was subjected to multispecific proteolysis to generate site-specific glycopeptides and to investigate the macroheterogeneity and microheterogeneity. Glycopeptides were identified and quantified by nano liquid chromatography-mass spectrometry and nano liquid chromatography-tandem mass spectrometry. Ninety-six glycopeptides, each corresponding to a unique glycan/glycosite pairing, were tracked across all cancer and control samples. Differences in abundance between the two groups were marked by particularly high magnitudes. Three glycopeptides exhibited exceptionally high control-to-cancer fold changes along with receiver operating characteristic curve areas of 1.0, indicating perfect discrimination between the two groups. From the results taken together, our platform, which provides biological information as well as high sensitivity and reproducibility, may be useful for GC biomarker discovery. Graphical abstract ᅟ.

CancerDiscover: an integrative pipeline for cancer biomarker and cancer class prediction from high-throughput sequencing data.

Accurate identification of cancer biomarkers and classification of cancer type and subtype from High Throughput Sequencing (HTS) data is a challenging problem because it requires manual processing of raw HTS data from various sequencing platforms, quality control, and normalization, which are both tedious and time-consuming. Machine learning techniques for cancer class prediction and biomarker discovery can hasten cancer detection and significantly improve prognosis. To date, great research efforts have been taken for cancer biomarker identification and cancer class prediction. However, currently available tools and pipelines lack flexibility in data preprocessing, running multiple feature selection methods and learning algorithms, therefore, developing a freely available and easy-to-use program is strongly demanded by researchers. Here, we propose CancerDiscover, an integrative open-source software pipeline that allows users to automatically and efficiently process large high-throughput raw datasets, normalize, and selects best performing features from multiple feature selection algorithms. Additionally, the integrative pipeline lets users apply different feature thresholds to identify cancer biomarkers and build various training models to distinguish different types and subtypes of cancer. The open-source software is available at https://github.com/HelikarLab/CancerDiscover and is free for use under the GPL3 license.

Selected reaction monitoring approach for validating peptide biomarkers

We here describe a selected reaction monitoring (SRM)-based approach for the discovery and validation of peptide biomarkers for cancer. The first stage of this approach is the direct identification of candidate peptides through comparison of proteolytic peptides derived from the plasma of cancer patients or healthy individuals. Several hundred candidate peptides were identified through this method, providing challenges for choosing and validating the small number of peptides that might prove diagnostically useful. To accomplish this validation, we used 2D chromatography coupled with SRM of candidate peptides. We applied this approach, called sequential analysis of fractionated eluates by SRM (SAFE-SRM), to plasma from cancer patients and discovered two peptides encoded by the peptidyl-prolyl cis-trans isomerase A (PPIA) gene whose abundance was increased in the plasma of ovarian cancer patients. At optimal thresholds, elevated levels of at least one of these two peptides was detected in 43 (68.3%) of 63 women with ovarian cancer but in none of 50 healthy controls. In addition to providing a potential biomarker for ovarian cancer, this approach is generally applicable to the discovery of peptides characteristic of various disease states.

Current and Prospective Protein Biomarkers of Lung Cancer.

Lung cancer is a malignant lung tumor with various histological variants that arise from different cell types, such as bronchial epithelium, bronchioles, alveoli, or bronchial mucous glands. The clinical course and treatment efficacy of lung cancer depends on the histological variant of the tumor. Therefore, accurate identification of the histological type of cancer and respective protein biomarkers is crucial for adequate therapy. Due to the great diversity in the molecular-biological features of lung cancer histological types, detection is impossible without knowledge of the nature and origin of malignant cells, which release certain protein biomarkers into the bloodstream. To date, different panels of biomarkers are used for screening. Unfortunately, a uniform serum biomarker composition capable of distinguishing lung cancer types is yet to be discovered. As such, histological analyses of tumor biopsies and immunohistochemistry are the most frequently used methods for establishing correct diagnoses. Here, we discuss the recent advances in conventional and prospective aptamer based strategies for biomarker discovery. Aptamers like artificial antibodies can serve as molecular recognition elements for isolation detection and search of novel tumor-associated markers. Here we will describe how these small synthetic single stranded oligonucleotides can be used for lung cancer biomarker discovery and utilized for accurate diagnosis and targeted therapy. Furthermore, we describe the most frequently used in-clinic and novel lung cancer biomarkers, which suggest to have the ability of differentiating between histological types of lung cancer and defining metastasis rate.

P160 - Dithiothreitol-based protein equalization method: A powerful tool in bladder cancer biomarker discovery

P160 - Dithiothreitol-based protein equalization method: A powerful tool in bladder cancer biomarker discovery

Recent Advances in the Mass Spectrometry Methods for Glycomics and Cancer.

The review focuses on recent aspects (last three years) of glycosylation analyses that provide relevant information about cancer. It includes recent development in glycan and protein enrichment methods for discovery of cancer markers. It will however focus on the recent technological developments in mass spectrometry (MS), bioinformatics and separation methods as they apply toward identifying cancer markers. More specifically, it will cover advances in matrix-assisted laser desorption/ionization (MALDI), electrospray ionization (ESI), capillary electrophoresis (CE), and liquid chromatography (LC) coupled to mass spectrometry. The discussions will include glycans, recently identified as potential markers for cancer that have been discovered using the highlighted technologies. We will also discuss emerging glycoproteomic techniques and site-specific methods, and how these methods are being utilized for cancer biomarker discovery. The large amount of data and the complexity of glycoproteomic analysis have been the impetus for developing bioinformatic methods for assigning glycosylation sites and characterizing the potentially very large site- or microheterogeneity. This review will cover the most recent advancements in biomarker discovery of N- and O-glycosylation of proteins as well as the glycolipids. This group collectively constitutes glycosylation on the cell membrane or the glycocalyx. The review will also highlight methods that are highly reproducible, with low coefficient of variation (CV), and scalable for large sample sets. The reader is also referred to other notable earlier reviews on glycomic biomarkers for cancer. Mereiter et al. describe the recent glycomic effort in gastrointestinal cancer.1 A review focused on N-glycomic analysis of colorectal cancer has been published by Sethi and Fanayan.2 N-Glycan, O-glycan, and glycolipid characteristics of colorectal cancer were reviewed by Holst et al.3 Muchena et al. have provided a more general review of glycan biomarkers covering up to the current review period.4 The field of glycoscience also covers a broad area of structures and may include highly anionic (glycosaminoglycans) and monosaccharide (e.g. O-GlcNAc) modifications that require their specific and unique sets of analytical tools. The latter topics are not covered in this review. 1.1. Background of Glycosylation and Cancer There is nearly 50 years of research illustrating that changes in glycosylation accompany cancer.5 Glycosylation is a dynamic process intimately involved in key processes in cells, including cell-cell and cell-extracellular communication as well as cell-cell adhesion, and cellular metabolism. Glycans expressed in several types of glycoconjugates are known to change during cancer genesis and progression.6 These changes increase the structural heterogeneity and alter the functions of cells.7 Glycosylation has been found to enable tumor-induced immunomodulation and metastasis.8–10 The cell-surface structures allow the immune cells to differentiate self/normal cells from non-self/abnormal cells.11 For example, terminal residues on N-glycans, such as sialic acids, are involved in immunity and cell-cell communication.12 Changes in glycosylation of adhesion proteins can largely influence their binding properties, leading to altered cell-cell or cell-matrix contacts.13 Other types of glycans are also involved in cancer. Gangliosides and sphingolipids are involved in transmembrane communication vital in tumor cell growth and invasion.14 Glycosaminoglycans are involved in tumor cell migration15 and motility.16–18 The search for effective markers is aided by the understanding of how glycans are synthesized. The glycan biosynthetic process is a non-template process involving multiple enzymes, some performing competing activities. It is estimated that more than 300 metabolic enzymes, composed of glycosyltransferases and glycosidases, are involved in the biosynthesis and processing of glycans.19–20 The best-known series of pathways belongs to the production of N-glycans (Figure 1). They illustrate the large degree of structural heterogeneity in glycosylation. N-Glycans are produced in a step-wise process beginning with the production of high mannose structures on a lipid, which are transferred to the nascent polypeptide chain to guide protein folding. Once folded, the glycans are then trimmed back and extended to form complex and hybrid structures. The folded protein can be secreted with glycans that range from early in the process to yield high mannose structures to later in the process corresponding to complex or hybrid structures. The number of structures for one glycosylation site can vary by a large degree, from a handful for transferrin21 to over 70 structures for IgG, the most abundant serum glycoprotein.22–23 Open in a separate window Figure 1. Representation of the glycosylation pathway of proteins. The pathway illustrates the complexity and heterogeneity of structures. The proteins may exit the pathway with various levels of glycosylation.

Secreted Tumor Antigens - Immune Biomarkers for Diagnosis and Therapy.

With the advent of immunotherapies for cancer, there is growing interest in the identification of tumor antigens. Tumor antigens are self-molecules altered through e.g. genetic mutations (neoantigens), protein truncation, protein misfolding, or abnormal posttranslational modifications. To induce an immune response, tumor antigens need to be secreted into the tumor environment and presented to the immune system in the draining lymph nodes, resulting in the generation of tumor-specific effector cells and antibodies. Cytotoxic T cells are thought to be responsible for killing of tumor cells, and several recent studies have used MS, combined with exome/transcriptome sequencing and bioinformatics, to identify their cognate peptide ligands on tumor MHC class I molecules. Circulating (serum) antibodies have been more widely used to identify tumor antigens in a range of human cancers, using 2D Western blots, immunoaffinity, and microarray technologies. More specific antibody probes have been generated by harvesting antibodies directly from antibody-secreting cells through in vitro cultures of lymph node cells (antibody-secreting cells probes) or B-cell immortalization. Further identification and characterization of tumor antigens is likely to have important implications for cancer diagnostic and biomarker discovery, immune profiling, and the development of cancer vaccines and targeted immunotherapies.

Fifteen-Year Follow-Up of the NYU Lung Biomarker Screening Cohort: Indolent-Prevalent Screen-Detected Lung Cancers Demonstrate a Less Aggressive Form of Pulmonary Malignancy

SESSION TITLE: Advances in Lung Cancer SESSION TYPE: Original Investigation Slide PRESENTED ON: Wednesday, November 1, 2017 at 08:45 AM - 10:00 AM PURPOSE: Low-dose computed tomography (CT) for lung cancer screening is recommended in high-risk individuals to reduce lung cancer mortality. In our lung cancer screening cohort, we previously classified and characterized three groups of screen-detected lung cancers. “Incident (I)” cancers were defined as nodules not present on baseline CT-scan, but apparent on subsequent CT-scans. “Prevalent (P)” cancers were nodules documented at baseline CT and diagnosed as malignant within 6 months of the initial scan. “Indolent-prevalent (Id-P)” cancers were also present at baseline, but diagnosed after 6 months. In this update to our 2012 publication, we compared the baseline demographics, nodule characteristics, and clinical outcomes of these three sub-classifications of screen-detected cancers. METHODS: Since 2001, the NYU Lung Cancer Biomarker Center has conducted a longitudinal lung cancer biomarker discovery trial that included low-dose CT screening of high-risk individuals over the age of 50, with more than 20 pack-year smoking history, and living in an urban setting. Analysis included only patients with screen-detected lung cancer that have data to sub-classify them in this cohort study and report risk factors, prognosis, and subtypes of lung cancer. Mann-Whitney statistical analysis was used. RESULTS: A total of 1519 study subjects were enrolled from 2001 to 2016. We diagnosed 59 cancers in 56 patients. There were 42 adenocarcinomas, 4 small cell, 3 squamous-cell, 1 mesothelioma, and 9 “other” lung cancers. There were 29 Id-P cases, 20 P cases, and 10 I cases. While most Id-P and P cases were diagnosed at stage I-II (80% and 75%, respectively), most I cases were diagnosed at later stage (III-IV, 55%). The 1-year and 5-year survival rate at all stages for Id-P cases were 100% and 92%, respectively, while in P cases, 1-year and 5-year survival rates were both 85%. Recurrence rates were not statistically different between Id-P and P cases. There was more adenocarcinoma in-situ/lepidic predominant subtype (50%) in Id-P cases, more acinar predominant subtype (40%) in P cases, and no dominant subtype in I cases. Id-P cancers were more likely to be solid nodules compared with P cancers (p=0.05). On average, Id-P cancers were diagnosed 5.4 years after initial nodule detection and grew at a rate of 1.7mm per year. CONCLUSIONS: The three sub-classifications of screen-detected lung cancers have different histological subtypes and prognoses. Id-P screen detected lung cancers appear to have a less aggressive growth pattern despite being more likely to derive from solid nodules compared to P cancers. CLINICAL IMPLICATIONS: A better biological understanding of these three sub-classifications of screen-detected lung cancers may help us identify who may need further screening and improve screening intervals. DISCLOSURE: The following authors have nothing to disclose: Jiwoon Chang, Christopher Laboe, Myah Draine, Jamie Bessich, Gaetane Michaud, Daniel Sterman, William Rom, James Tsay No Product/Research Disclosure Information

Identification of potential tissue-specific cancer biomarkers and development of cancer versus normal genomic classifiers.

Machine learning techniques for cancer prediction and biomarker discovery can hasten cancer detection and significantly improve prognosis. Recent “OMICS” studies which include a variety of cancer and normal tissue samples along with machine learning approaches have the potential to further accelerate such discovery. To demonstrate this potential, 2,175 gene expression samples from nine tissue types were obtained to identify gene sets whose expression is characteristic of each cancer class. Using random forests classification and ten-fold cross-validation, we developed nine single-tissue classifiers, two multi-tissue cancer-versus-normal classifiers, and one multi-tissue normal classifier. Given a sample of a specified tissue type, the single-tissue models classified samples as cancer or normal with a testing accuracy between 85.29% and 100%. Given a sample of non-specific tissue type, the multi-tissue bi-class model classified the sample as cancer versus normal with a testing accuracy of 97.89%. Given a sample of non-specific tissue type, the multi-tissue multi-class model classified the sample as cancer versus normal and as a specific tissue type with a testing accuracy of 97.43%. Given a normal sample of any of the nine tissue types, the multi-tissue normal model classified the sample as a particular tissue type with a testing accuracy of 97.35%. The machine learning classifiers developed in this study identify potential cancer biomarkers with sensitivity and specificity that exceed those of existing biomarkers and pointed to pathways that are critical to tissue-specific tumor development. This study demonstrates the feasibility of predicting the tissue origin of carcinoma in the context of multiple cancer classes.

Lectins: an effective tool for screening of potential cancer biomarkers.

In recent years, the use of lectins for screening of potential biomarkers has gained increased importance in cancer research, given the development in glycobiology that highlights altered structural changes of glycans in cancer associated processes. Lectins, having the properties of recognizing specific carbohydrate moieties of glycoconjugates, have become an effective tool for detection of new cancer biomarkers in complex bodily fluids and tissues. The specificity of lectins provides an added advantage of selecting peptides that are differently glycosylated and aberrantly expressed in cancer patients, many of which are not possibly detected using conventional methods because of their low abundance in bodily fluids. When coupled with mass spectrometry, research utilizing lectins, which are mainly from plants and fungi, has led to identification of numerous potential cancer biomarkers that may be used in the future. This article reviews lectin-based methods that are commonly adopted in cancer biomarker discovery research.