Feature-specific quantile normalization and feature-specific mean-variance normalization deliver robust bi-directional classification and feature selection performance between microarray and RNAseq data.

BACKGROUND: Cross-platform normalization seeks to minimize technological bias between microarray and RNAseq whole-transcriptome data. Incorporating multiple gene expression platforms permits external validation of experimental findings, and augments training sets for machine learning models. Here, we compare the performance of Feature Specific Quantile Normalization (FSQN) to a previously used but unvalidated and uncharacterized method we label as Feature Specific Mean Variance Normalization (FSMVN). We evaluate the performance of these methods for bidirectional normalization in the context of nested feature selection. RESULTS: FSQN and FSMVN provided clinically equivalent bidirectional model performance with and without feature selection for colon CMS and breast PAM50 classification. Using principal component analysis, we determine that these methods eliminate batch effects related to technological platforms. Without feature selection, no statistical difference was identified between the performance of FSQN and FSMVN of cross-platform data compared to within-platform distributions. Under optimal feature selection conditions, balanced accuracy was FSQN and FSMVN were statistically equivalent to the within-platform distribution performance in multivariable linear regression analysis. FSQN and FSMVN also provided similar performance to within-platform distributions as the number of selected genes used to create models decreases. CONCLUSIONS: In the context of generating supervised machine learning classifiers for molecular subtypes, FSQN and FSMVN are equally effective. Under optimal modeling conditions, FSQN and FSMVN provide equivalent model accuracy performance on cross-platform normalization data compared to within-platform data. Using cross-platform data should still be approached with caution as subtle performance differences may exist depending on the classification problem, training, and testing distributions.

Individual Survival Distributions Generated by Multi-Task Logistic Regression Yield a New Perspective on Molecular and Clinical Prognostic Factors in Gastric Adenocarcinoma.

Recent advances in our understanding of gastric cancer biology have prompted a shift towards more personalized therapy. However, results are based on population-based survival analyses, which evaluate the average survival effects of entire treatment groups or single prognostic variables. This study uses a personalized survival modelling approach called individual survival distributions (ISDs) with the multi-task logistic regression (MTLR) model to provide novel insight into personalized survival in gastric adenocarcinoma. We performed a pooled analysis using 1043 patients from a previously characterized database annotated with molecular subtypes from the Cancer Genome Atlas, Asian Cancer Research Group, and tumour microenvironment (TME) score. The MTLR model achieved a 5-fold cross-validated concordance index of 72.1 ± 3.3%. This model found that the TME score and chemotherapy had similar survival effects over the entire study time. The TME score provided the greatest survival benefit beyond a 5-year follow-up. Stage III and Stage IV disease contributed the greatest negative effect on survival. The MTLR model weights were significantly correlated with the Cox model coefficients (Pearson coefficient = 0.86, p < 0.0001). We illustrate how ISDs can accurately predict the survival time for each patient, which is especially relevant in cases of molecular subtype heterogeneity. This study provides evidence that the TME score is principally associated with long-term survival in gastric adenocarcinoma. Additional external validation and investigation into the clinical utility of this ISD model in gastric cancer is an area of future research.

Pancreatic ductal adenocarcinoma is associated with a distinct urinary metabolomic signature.

BACKGROUND: Pancreatic ductal adenocarcinoma (PDAC) is an aggressive malignancy with poor prognosis in part due to the lack of early detection and screening methods. Metabolomics provides a means for noninvasive screening of tumor-associated perturbations in cellular metabolism. METHODS: Urine samples of PDAC patients (n = 32), healthy age and gender-matched controls (n = 32), and patients with benign pancreatic conditions (n = 25) were examined using (1)H-NMR spectroscopy. Targeted profiling of spectra permitted quantification of 66 metabolites. Unsupervised (principal component analysis, PCA) and supervised (orthogonal partial-least squares discriminant analysis, OPLS-DA) multivariate pattern recognition techniques were applied to discriminate between sample spectra using SIMCA-P(+) (version 12, Umetrics, Sweden). RESULTS: Clear distinction between PDAC and controls was noted when using OPLS-DA. Significant differences in metabolite concentrations between cancers and controls (p < 0.001) were noted. Model parameters for both goodness of fit, and predictive capability were high (R (2) = 0.85; Q (2) = 0.59, respectively). Internal validation methods were used to confirm model validity. Sensitivity and specificity of the multivariate OPLS-DA model were summarized using a receiver operating characteristics (ROC) curve, with an area under the curve (AUROC) = 0.988, indicating strong predictive power. Preliminary analysis revealed an AUROC = 0.958 for the model of benign pancreatic disease compared with PDAC, and suggest that the cancer-associated metabolomic signature dissipates following RO resection. CONCLUSIONS: Urinary metabolomics detected distinct differences in the metabolic profiles of pancreatic cancer compared with healthy controls and benign pancreatic disease. These preliminary results suggest that metabolomic approaches may facilitate discovery of novel pancreatic cancer biomarkers.

Urinary metabolomic signature of esophageal cancer and Barrett's esophagus.

BACKGROUND: Esophageal adenocarcinoma (EAC) often presents at a late, incurable stage, and mortality has increased substantially, due to an increase in incidence of EAC arising out of Barrett's esophagus. When diagnosed early, however, the combination of surgery and adjuvant therapies is associated with high cure rates. Metabolomics provides a means for non- invasive screening of early tumor-associated perturbations in cellular metabolism. METHODS: Urine samples from patients with esophageal carcinoma (n = 44), Barrett's esophagus (n = 31), and healthy controls (n = 75) were examined using (1)H-NMR spectroscopy. Targeted profiling of spectra using Chenomx software permitted quantification of 66 distinct metabolites. Unsupervised (principal component analysis) and supervised (orthogonal partial least-squares discriminant analysis OPLS-DA) multivariate pattern recognition techniques were applied to discriminate between samples using SIMCA-P(+) software. Model specificity was also confirmed through comparison with a pancreatic cancer cohort (n = 32). RESULTS: Clear distinctions between esophageal cancer, Barrett's esophagus and healthy controls were noted when OPLS-DA was applied. Model validity was confirmed using two established methods of internal validation, cross-validation and response permutation. Sensitivity and specificity of the multivariate OPLS-DA models were summarized using a receiver operating characteristic curve analysis and revealed excellent predictive power (area under the curve = 0.9810 and 0.9627 for esophageal cancer and Barrett's esophagus, respectively). The metabolite expression profiles of esophageal cancer and pancreatic cancer were also clearly distinguishable with an area under the receiver operating characteristics curve (AUROC) = 0.8954. CONCLUSIONS: Urinary metabolomics identified discrete metabolic signatures that clearly distinguished both Barrett's esophagus and esophageal cancer from controls. The metabolite expression profile of esophageal cancer was also discrete from its precursor lesion, Barrett's esophagus. The cancer-specific nature of this profile was confirmed through comparison with pancreatic cancer. These preliminary results suggest that urinary metabolomics may have a future potential role in non-invasive screening in these conditions.

Metabolomics and surgical oncology: Potential role for small molecule biomarkers.

Metabolomics, the newest of the "omics" sciences, has brought much excitement to the field of oncology as a potential new translational tool capable of bringing the molecular world of cancer care to the bedside. While still early in its development, metabolomics could alter the scope and role of surgery in the multidisciplinary treatment of cancer. This review examines potential roles of metabolomics in areas of early cancer detection, personalized therapeutics and tumorigenesis.