Framework For Biomarker Discovery Research Articles

Abstract 6446: ResponderID™ KRAS: Biology-driven machine learning to personalize KRAS inhibitor therapeutics

Abstract Breakthroughs in targeted KRAS therapeutics (KRASi) have the potential to transform the treatment landscape for several of the most common cancers including lung, colorectal, and pancreatic. Despite the recent approvals of KRASi and the anticipation of more to come, both the rate of patient response and the durability of these responses remain significant areas requiring improvement. Biomarkers that can predict response to KRASi and guide effective patient selection and drug combination strategies will be key to realizing the full potential of this emerging therapeutic field. While most biomarkers predominantly rely on a single analyte (e.g. KRAS mutation status), Genialis’ biomarkers are constructed using high-dimensional and/or multimodal data that capture the underlying biological complexity unique to each individual patient. Genialis' ResponderID™ is a machine learning-based biomarker discovery framework that models fundamental aspects of cancer biology to predict the clinical benefit based on the patient’s own biology. Here we report progress towards the development of a first-in-class, RNA-based biomarker, ResponderID™ KRAS, capable of stratifying KRAS G12C inhibitor response in lung cancer patients using RNA sequencing data. Trained on thousands of lung cancer samples, our biomarker models therapeutic response by unifying two core KRAS biologic axes, dependency and activation, to identify those patients most likely to respond. The performance characteristics of ResponderID™ KRAS thus far has been evaluated on a real world dataset of lung cancer patients treated with Sotorasib. ResponderID™ KRAS serves as an independent biomarker designed to inform clinical trial design, select for therapeutic efficacy, identify rational combination strategies, and expedite approvals across various therapeutic contexts. Citation Format: Josh Wheeler, Anže Lovše, Klemen Žiberna, Miha Štajdohar, Luka Ausec, Janez Kokošar, Daniel Pointing, Aditya Pai, Rafael Rosengarten, Mark Uhlik. ResponderID™ KRAS: Biology-driven machine learning to personalize KRAS inhibitor therapeutics [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 6446.

The Oncology Biomarker Discovery framework reveals cetuximab and bevacizumab response patterns in metastatic colorectal cancer

Precision medicine has revolutionised cancer treatments; however, actionable biomarkers remain scarce. To address this, we develop the Oncology Biomarker Discovery (OncoBird) framework for analysing the molecular and biomarker landscape of randomised controlled clinical trials. OncoBird identifies biomarkers based on single genes or mutually exclusive genetic alterations in isolation or in the context of tumour subtypes, and finally, assesses predictive components by their treatment interactions. Here, we utilise the open-label, randomised phase III trial (FIRE-3, AIO KRK-0306) in metastatic colorectal carcinoma patients, who received either cetuximab or bevacizumab in combination with 5-fluorouracil, folinic acid and irinotecan (FOLFIRI). We systematically identify five biomarkers with predictive components, e.g., patients with tumours that carry chr20q amplifications or lack mutually exclusive ERK signalling mutations benefited from cetuximab compared to bevacizumab. In summary, OncoBird characterises the molecular landscape and outlines actionable biomarkers, which generalises to any molecularly characterised randomised controlled trial.

A biomarker discovery framework for childhood anxiety.

Anxiety is the most common manifestation of psychopathology in youth, negatively affecting academic, social, and adaptive functioning and increasing risk for mental health problems into adulthood. Anxiety disorders are diagnosed only after clinical symptoms emerge, potentially missing opportunities to intervene during critical early prodromal periods. In this study, we used a new empirical approach to extracting nonlinear features of the electroencephalogram (EEG), with the goal of discovering differences in brain electrodynamics that distinguish children with anxiety disorders from healthy children. Additionally, we examined whether this approach could distinguish children with externalizing disorders from healthy children and children with anxiety. We used a novel supervised tensor factorization method to extract latent factors from repeated multifrequency nonlinear EEG measures in a longitudinal sample of children assessed in infancy and at ages 3, 5, and 7 years of age. We first examined the validity of this method by showing that calendar age is highly correlated with latent EEG complexity factors (r = 0.77). We then computed latent factors separately for distinguishing children with anxiety disorders from healthy controls using a 5-fold cross validation scheme and similarly for distinguishing children with externalizing disorders from healthy controls. We found that latent factors derived from EEG recordings at age 7 years were required to distinguish children with an anxiety disorder from healthy controls; recordings from infancy, 3 years, or 5 years alone were insufficient. However, recordings from two (5, 7 years) or three (3, 5, 7 years) recordings gave much better results than 7 year recordings alone. Externalizing disorders could be detected using 3- and 5 years EEG data, also giving better results with two or three recordings than any single snapshot. Further, sex assigned at birth was an important covariate that improved accuracy for both disorder groups, and birthweight as a covariate modestly improved accuracy for externalizing disorders. Recordings from infant EEG did not contribute to the classification accuracy for either anxiety or externalizing disorders. This study suggests that latent factors extracted from EEG recordings in childhood are promising candidate biomarkers for anxiety and for externalizing disorders if chosen at appropriate ages.

XAI-CNVMarker: Explainable AI-based copy number variant biomarker discovery for breast cancer subtypes

Breast cancer is a leading cause of cancer-related deaths among women. The multi-omic data has revolutionized the methodology to unravel molecular heterogeneity in breast cancer. As genetic variations captured from Copy Number Variation (CNV) data are considered the most stable amongst the multi-omic data, it leads to robust biomarkers. Thus, this paper targets the discovery of a set of CNV biomarkers for dissecting this heterogeneity. The existing algorithms yield biomarkers, too huge to be interpreted clinically. So, in this paper, we have proposed XAI-CNVMarker—an explainable AI-based post-hoc biomarker discovery framework to discover a small set of interpretable biomarkers. We exploit the power of deep learning to build DLmodel—a deep learning model for breast cancer classification. Subsequently, the trained model is analyzed using different explainable AI methods to arrive at a set of 44 CNV biomarkers. Using 5-fold cross-validation, we obtained a classification accuracy of 0.712 (± 0.048) at a 95% confidence interval. Gene set analysis revealed 37 subtype-specific enriched Reactome and Kegg pathways, 21 druggable genes, and 13 biomarkers linked with the prognostic outcome. Finally, we validated the efficacy of the identified biomarkers on METABRIC. Thus, the proposed framework demonstrates the role of explainable AI in discovering clinically reliable biomarkers.

An explainable AI-driven biomarker discovery framework for Non-Small Cell Lung Cancer classification

Non-Small Cell Lung Cancer (NSCLC) exhibits intrinsic heterogeneity at the molecular level that aids in distinguishing between its two prominent subtypes — Lung Adenocarcinoma (LUAD) and Lung Squamous Cell Carcinoma (LUSC). This paper proposes a novel explainable AI (XAI)-based deep learning framework to discover a small set of NSCLC biomarkers. The proposed framework comprises three modules — an autoencoder to shrink the input feature space, a feed-forward neural network to classify NSCLC instances into LUAD and LUSC, and a biomarker discovery module that leverages the combined network comprising the autoencoder and the feed-forward neural network. In the biomarker discovery module, XAI methods uncovered a set of 52 relevant biomarkers for NSCLC subtype classification. To evaluate the classification performance of the discovered biomarkers, multiple machine-learning models are constructed using these biomarkers. Using 10-Fold cross-validation, Multilayer Perceptron achieved an accuracy of 95.74% (±1.27) at 95% confidence interval. Further, using Drug-Gene Interaction Database, we observe that 14 of the discovered biomarkers are druggable. In addition, 28 biomarkers aid the prediction of the survivability of the patients. Out of 52 discovered biomarkers, we find that 45 biomarkers have been reported in previous studies on distinguishing between the two NSCLC subtypes. To the best of our knowledge, the remaining seven biomarkers have not yet been reported for NSCLC subtyping and could be further explored for their contribution to targeted therapy of lung cancer.

Biomarker Development in Cardiology: Reviewing the Past to Inform the Future.

Cardiac biomarkers have become pivotal to the clinical practice of cardiology, but there remains much to discover that could benefit cardiology patients. We review the discovery of key protein biomarkers in the fields of acute coronary syndrome, heart failure, and atherosclerosis, giving an overview of the populations they were studied in and the statistics that were used to validate them. We review statistical approaches that are currently in use to assess new biomarkers and overview a framework for biomarker discovery and evaluation that could be incorporated into clinical trials to evaluate cardiovascular outcomes in the future.

Data-Driven Kidney Transplant Phenotyping as a Histology-Independent Framework for Biomarker Discovery.

In transplant medicine, clinical decision making largely relies on histology of biopsy specimens. However, histology suffers from low specificity, sensitivity, and reproducibility, leading to suboptimal stratification of patients. We developed a histology-independent immune framework of kidney graft homeostasis and rejection. We applied tailored RNA deconvolution for leukocyte enumeration and coregulated gene network analysis to published bulk human kidney transplant RNA transcriptomes as input for unsupervised, high-dimensional phenotype clustering. We used framework-based graft survival analysis to identify a biomarker that was subsequently characterized in independent transplant biopsy specimens. We found seven immune phenotypes that confirm known rejection types and uncovered novel signatures. The molecular phenotypes allow for improved graft survival analysis compared with histology, and identify a high-risk group in nonrejecting transplants. Two fibrosis-related phenotypes with distinct immune features emerged with reduced graft survival. We identified lysyl oxidase-like 2 (LOXL2)-expressing peritubular CD68+ macrophages as a framework-derived biomarker of impaired allograft function. These cells precede graft fibrosis, as demonstrated in longitudinal biopsy specimens, and may be clinically useful as a biomarker for early fibrogenesis. This study provides a comprehensive, data-driven atlas of human kidney transplant phenotypes and demonstrates its utility to identify novel clinical biomarkers.

An integrated multi-omics analysis identifies prognostic molecular subtypes of non-muscle-invasive bladder cancer

The molecular landscape in non-muscle-invasive bladder cancer (NMIBC) is characterized by large biological heterogeneity with variable clinical outcomes. Here, we perform an integrative multi-omics analysis of patients diagnosed with NMIBC (n = 834). Transcriptomic analysis identifies four classes (1, 2a, 2b and 3) reflecting tumor biology and disease aggressiveness. Both transcriptome-based subtyping and the level of chromosomal instability provide independent prognostic value beyond established prognostic clinicopathological parameters. High chromosomal instability, p53-pathway disruption and APOBEC-related mutations are significantly associated with transcriptomic class 2a and poor outcome. RNA-derived immune cell infiltration is associated with chromosomally unstable tumors and enriched in class 2b. Spatial proteomics analysis confirms the higher infiltration of class 2b tumors and demonstrates an association between higher immune cell infiltration and lower recurrence rates. Finally, the independent prognostic value of the transcriptomic classes is documented in 1228 validation samples using a single sample classification tool. The classifier provides a framework for biomarker discovery and for optimizing treatment and surveillance in next-generation clinical trials.

A Precision Medicine Approach to Biomarker Utilization in Pediatric Sepsis-Associated Acute Kidney Injury.

Sepsis is a leading cause of morbidity and mortality in critically ill children, and acute kidney injury (AKI) is a frequent complication that confers an increased risk for poor outcomes. Despite the documented consequences of sepsis-associated AKI (SA-AKI), no effective disease-modifying therapies have been identified to date. As such, the only treatment options for these patients remain prevention and supportive care, both of which rely on the ability to promptly and accurately identify at risk and affected individuals. To achieve these goals, a variety of biomarkers have been investigated to help augment our currently limited predictive and diagnostic strategies for SA-AKI, however, these have had variable success in pediatric sepsis. In this mini-review, we will briefly outline the current use of biomarkers for SA-AKI, and propose a new framework for biomarker discovery and utilization that considers the individual patient's sepsis inflammatory response. Now recognized to be a key driver in the complex pathophysiology of SA-AKI, understanding the dysregulated host immune response to sepsis is a growing area of research that can and should be leveraged to improve the prediction and diagnosis of SA-AKI, while also potentially identifying novel therapeutic targets. Reframing SA-AKI in this manner – as a direct consequence of the individual patient's sepsis inflammatory response – will facilitate a precision medicine approach to its management, something that is required to move the care of this consequential disorder forward.

A random forest based biomarker discovery and power analysis framework for diagnostics research

BackgroundBiomarker identification is one of the major and important goal of functional genomics and translational medicine studies. Large scale –omics data are increasingly being accumulated and can provide vital means for the identification of biomarkers for the early diagnosis of complex disease and/or for advanced patient/diseases stratification. These tasks are clearly interlinked, and it is essential that an unbiased and stable methodology is applied in order to address them. Although, recently, many, primarily machine learning based, biomarker identification approaches have been developed, the exploration of potential associations between biomarker identification and the design of future experiments remains a challenge.MethodsIn this study, using both simulated and published experimentally derived datasets, we assessed the performance of several state-of-the-art Random Forest (RF) based decision approaches, namely the Boruta method, the permutation based feature selection without correction method, the permutation based feature selection with correction method, and the backward elimination based feature selection method. Moreover, we conducted a power analysis to estimate the number of samples required for potential future studies.ResultsWe present a number of different RF based stable feature selection methods and compare their performances using simulated, as well as published, experimentally derived, datasets. Across all of the scenarios considered, we found the Boruta method to be the most stable methodology, whilst the Permutation (Raw) approach offered the largest number of relevant features, when allowed to stabilise over a number of iterations. Finally, we developed and made available a web interface (https://joelarkman.shinyapps.io/PowerTools/) to streamline power calculations thereby aiding the design of potential future studies within a translational medicine context.ConclusionsWe developed a RF-based biomarker discovery framework and provide a web interface for our framework, termed PowerTools, that caters the design of appropriate and cost-effective subsequent future omics study.

Animal models in translational studies of PTSD

Understanding the neurobiological mechanisms of post-traumatic stress disorder (PTSD) is of vital importance for developing biomarkers and more effective pharmacotherapy for this disorder. The design of bidirectional translational studies addressing all facets of PTSD is needed. Animal models of PTSD are needed not only to capture the complexity of PTSD behavioral characteristics, but also to address experimentally the influence of variety of factors which might determine an individual's vulnerability or resilience to trauma, e.g., genetic predisposition, early-life experience and social support. The current review covers recent translational approaches to bridge the gap between human and animal PTSD research and to create a framework for discovery of biomarkers and novel therapeutics.

Metabolomics of Psychotic Disorders

Metabolomics, the global study of metabolites, has recently emerged as a promising approach for identification of potential diagnostic and treatment response biomarkers for psychotic disorders. To date, numerous studies have utilised metabolomics to better understand psychotic disorders and findings from these studies have begun to converge. In this review, we briefly describe the metabolomics approach including the different platforms used to analyse metabolites in biological samples from patients. We also summarise promising metabolic and pharmaco-metabolic biomarkers reported in the current psychotic disorder literature, which point to the dysregulation of fatty acid metabolism and the imbalance in oxidants/antioxidants that is present at illness onset. Finally, we conclude with a commentary on the challenges and future contribution of the metabolomics approach within the larger biomarker discovery framework currently being utilised in the field of psychiatry.

A statistical framework for biomarker discovery in metabolomic time course data.

Metabolomics is the study of the complement of small molecule metabolites in cells, biofluids and tissues. Many metabolomic experiments are designed to compare changes observed over time under two experimental conditions or groups (e.g. a control and drug-treated group) with the goal of identifying discriminatory metabolites or biomarkers that characterize each condition. A common study design consists of repeated measurements taken on each experimental unit thus producing time courses of all metabolites. We describe a statistical framework for estimating time-varying metabolic profiles and their within-group variability and for detecting between-group differences. Specifically, we propose (i) a smoothing splines mixed effects (SME) model that treats each longitudinal measurement as a smooth function of time and (ii) an associated functional test statistic. Statistical significance is assessed by a non-parametric bootstrap procedure. The methodology has been extensively evaluated using simulated data and has been applied to real nuclear magnetic resonance spectroscopy data collected in a preclinical toxicology study as part of a larger project lead by the COMET (Consortium for Metabonomic Toxicology). Our findings are compatible with the previously published studies. An R script is freely available for download at http://www2.imperial.ac.uk/~gmontana/sme.htm.

A Proteome Resource of Ovarian Cancer Ascites: Integrated Proteomic and Bioinformatic Analyses To Identify Putative Biomarkers

Epithelial ovarian cancer is the most lethal gynecological malignancy, and disease-specific biomarkers are urgently needed to improve diagnosis, prognosis, and to predict and monitor treatment efficiency. We present an in-depth proteomic analysis of selected biochemical fractions of human ovarian cancer ascites, resulting in the stringent and confident identification of over 2500 proteins. Rigorous filter schemes were applied to objectively minimize the number of false-positive identifications, and we only report proteins with substantial peptide evidence. Integrated computational analysis of the ascites proteome combined with several recently published proteomic data sets of human plasma, urine, 59 ovarian cancer related microarray data sets, and protein-protein interactions from the Interologous Interaction Database I (2)D ( http://ophid.utoronto.ca/i2d) resulted in a short-list of 80 putative biomarkers. The presented proteomics analysis provides a significant resource for ovarian cancer research, and a framework for biomarker discovery.