Predicting patient response with models trained on cell lines and patient-derived xenografts by nonlinear transfer learning

2021 ◽  
Vol 118 (49) ◽  
pp. e2106682118
Author(s):  
Soufiane M. C. Mourragui ◽  
Marco Loog ◽  
Daniel J. Vis ◽  
Kat Moore ◽  
Anna G. Manjon ◽  
...  
Keyword(s):  
Drug Response ◽  
Clinical Importance ◽  
Chemotherapeutic Agents ◽  
Human Tumors ◽  
The Cancer Genome Atlas ◽  
Learning Approaches ◽  
Preclinical Models ◽  
Patient Response ◽  
Response Predictors ◽  
Cancer Genome Atlas

Preclinical models have been the workhorse of cancer research, producing massive amounts of drug response data. Unfortunately, translating response biomarkers derived from these datasets to human tumors has proven to be particularly challenging. To address this challenge, we developed TRANSACT, a computational framework that builds a consensus space to capture biological processes common to preclinical models and human tumors and exploits this space to construct drug response predictors that robustly transfer from preclinical models to human tumors. TRANSACT performs favorably compared to four competing approaches, including two deep learning approaches, on a set of 23 drug prediction challenges on The Cancer Genome Atlas and 226 metastatic tumors from the Hartwig Medical Foundation. We demonstrate that response predictions deliver a robust performance for a number of therapies of high clinical importance: platinum-based chemotherapies, gemcitabine, and paclitaxel. In contrast to other approaches, we demonstrate the interpretability of the TRANSACT predictors by correctly identifying known biomarkers of targeted therapies, and we propose potential mechanisms that mediate the resistance to two chemotherapeutic agents.

scholarly journals PRECISE: a domain adaptation approach to transfer predictors of drug response from pre-clinical models to tumors

2019 ◽  
Vol 35 (14) ◽  
pp. i510-i519 ◽  
Author(s):  
Soufiane Mourragui ◽  
Marco Loog ◽  
Mark A van de Wiel ◽  
Marcel J T Reinders ◽  
Lodewyk F A Wessels
Keyword(s):  
Cell Lines ◽  
Drug Response ◽  
Domain Adaptation ◽  
Predictive Performance ◽  
Human Tumors ◽  
Supplementary Information ◽  
Patient Response ◽  
Common Information ◽  
Response Predictors ◽  
Clinical Models

Abstract Motivation Cell lines and patient-derived xenografts (PDXs) have been used extensively to understand the molecular underpinnings of cancer. While core biological processes are typically conserved, these models also show important differences compared to human tumors, hampering the translation of findings from pre-clinical models to the human setting. In particular, employing drug response predictors generated on data derived from pre-clinical models to predict patient response remains a challenging task. As very large drug response datasets have been collected for pre-clinical models, and patient drug response data are often lacking, there is an urgent need for methods that efficiently transfer drug response predictors from pre-clinical models to the human setting. Results We show that cell lines and PDXs share common characteristics and processes with human tumors. We quantify this similarity and show that a regression model cannot simply be trained on cell lines or PDXs and then applied on tumors. We developed PRECISE, a novel methodology based on domain adaptation that captures the common information shared amongst pre-clinical models and human tumors in a consensus representation. Employing this representation, we train predictors of drug response on pre-clinical data and apply these predictors to stratify human tumors. We show that the resulting domain-invariant predictors show a small reduction in predictive performance in the pre-clinical domain but, importantly, reliably recover known associations between independent biomarkers and their companion drugs on human tumors. Availability and implementation PRECISE and the scripts for running our experiments are available on our GitHub page (https://github.com/NKI-CCB/PRECISE). Supplementary information Supplementary data are available at Bioinformatics online.

scholarly journals PRECISE: A domain adaptation approach to transfer predictors of drug response from pre-clinical models to tumors

2019 ◽  
Author(s):  
Soufiane Mourragui ◽  
Marco Loog ◽  
Marcel JT Reinders ◽  
Lodewyk FA Wessels
Keyword(s):  
Cell Lines ◽  
Drug Response ◽  
Domain Adaptation ◽  
Predictive Performance ◽  
Human Tumors ◽  
Supplementary Information ◽  
Patient Response ◽  
Common Information ◽  
Response Predictors ◽  
Clinical Models

AbstractMotivationCell lines and patient-derived xenografts (PDX) have been used extensively to understand the molecular underpinnings of cancer. While core biological processes are typically conserved, these models also show important differences compared to human tumors, hampering the translation of findings from pre-clinical models to the human setting. In particular, employing drug response predictors generated on data derived from pre-clinical models to predict patient response, remains a challenging task. As very large drug response datasets have been collected for pre-clinical models, and patient drug response data is often lacking, there is an urgent need for methods that efficiently transfer drug response predictors from pre-clinical models to the human setting.ResultsWe show that cell lines and PDXs share common characteristics and processes with human tumors. We quantify this similarity and show that a regression model cannot simply be trained on cell lines or PDXs and then applied on tumors. We developed PRECISE, a novel methodology based on domain adaptation that captures the common information shared amongst pre-clinical models and human tumors in a consensus representation. Employing this representation, we train predictors of drug response on pre-clinical data and apply these predictors to stratify human tumors. We show that the resulting domain-invariant predictors show a small reduction in predictive performance in the pre-clinical domain but, importantly, reliably recover known associations between independent biomarkers and their companion drugs on human tumors.AvailabilityPRECISE and the scripts for running our experiments are available on our GitHub page (https://github.com/NKI-CCB/PRECISE)[email protected] informationSupplementary data are available. online.

scholarly journals Predicting clinical drug response from model systems by non-linear subspace-based transfer learning

2020 ◽  
Author(s):  
Soufiane Mourragui ◽  
Marco Loog ◽  
Daniel J. Vis ◽  
Kat Moore ◽  
Anna G. Manjon ◽  
...  
Keyword(s):  
Deep Learning ◽  
Drug Response ◽  
Linear Manifold ◽  
Response Prediction ◽  
Human Tumors ◽  
Model Systems ◽  
Learning Approaches ◽  
Non Linear ◽  
Clinical Models ◽  
Clinical Drug

AbstractPre-clinical models have been the workhorse of cancer research for decades. While powerful, these models do not fully recapitulate the complexity of human tumors. Consequently, translating biomarkers of drug response from pre-clinical models to human tumors has been particularly challenging. To explicitly take these differences into account and enable an efficient exploitation of the vast pre-clinical drug response resources, we developed TRANSACT, a novel computational framework for clinical drug response prediction. First, TRANSACT employs non-linear manifold learning to capture biological processes active in pre-clinical models and human tumors. Then, TRANSACT builds predictors on cell line response only and transfers these to Patient-Derived Xenografts (PDXs) and human tumors. TRANSACT outperforms four competing approaches, including Deep Learning approaches, for a set of 15 drugs on PDXs, TCGA cohorts and 226 metastatic tumors from the Hartwig Medical Foundation data. For only four drugs Deep Learning outperforms TRANSACT. We further derived an algorithmic approach to interpret TRANSACT and used it to validate the approach by identifying known biomarkers to targeted therapies and we propose novel putative biomarkers of resistance to Paclitaxel and Gemcitabine.

scholarly journals Widespread expression of Sonic hedgehog (Shh) and Nrf2 in patients treated with cisplatin predicts outcome in resected tumors and are potential therapeutic targets for HPV-negative head and neck cancer

2020 ◽  
Vol 12 ◽  
pp. 175883592091122 ◽  
Author(s):  
Abu Shadat M. Noman ◽  
Rashed R. Parag ◽  
Muhammad I. Rashid ◽  
Mohammad Z. Rahman ◽  
Ali A. Chowdhury ◽  
...  
Keyword(s):  
Head And Neck ◽  
Sonic Hedgehog ◽  
Critical Role ◽  
Therapeutic Targets ◽  
Disease Free Survival ◽  
Chemotherapeutic Agents ◽  
The Cancer Genome Atlas ◽  
Chemotherapeutic Resistance ◽  
Cancer Genome Atlas ◽  
Relationship Of

Background: Sonic hedgehog (Shh) and Nrf2 play a critical role in chemotherapeutic resistance. These two genes have been found to be dysregulated in head and neck squamous cell carcinomas (HNSCC). The purpose of this study was to analyze the expression, function and clinical prognostic relationship of Shh and Nrf2 in HNSCC in the context of therapeutic resistance and cancer stem cells (CSCs). Methods: We analyzed a cohort of patients with HNSCC to identify potential therapeutic biomarkers correlating with overall survival (OS) as well as disease-free survival (DFS) from our own data and validated these results using The Cancer Genome Atlas dataset. Expression of Shh and Nrf2 was knocked down by siRNA and cell growth, sphere growth and chemotherapeutic resistance were evaluated. Results: Widespread abundant expression of Shh and Nrf2 proteins were associated with shorter OS and DFS. The combination of Shh and Nrf2 expression levels was found to be a significant predictor of patient DFS. The tumor stromal index was correlated with Shh expression and inversely associated with shorter OS and DFS. Inhibition of Shh by siRNA or cyclopamine resulted in the attenuation of resistant CSC self-renewal, invasion, clonogenic growth and re-sensitization to the chemotherapeutic agents. Concomitant upregulation of Shh and Nrf2 proved to be an independent predictor of poor OS and DFS in patients with HNSCC. Conclusions: These findings suggest that Shh and Nrf2 could serve as therapeutic targets as well as promising dual prognostic therapeutic biomarkers for HNSCC.

scholarly journals Clinical Importance of Epstein–Barr Virus-Associated Gastric Cancer

Cancers ◽  
2018 ◽  
Vol 10 (6) ◽  
pp. 167 ◽  
Author(s):  
Jun Nishikawa ◽  
Hisashi Iizasa ◽  
Hironori Yoshiyama ◽  
Kanami Shimokuri ◽  
Yuki Kobayashi ◽  
...  
Keyword(s):  
Gastric Cancer ◽  
Gastric Carcinoma ◽  
Epstein Barr Virus ◽  
Low Frequency ◽  
Clinical Importance ◽  
Molecular Classification ◽  
The Cancer Genome Atlas ◽  
Barr Virus ◽  
Cancer Genome Atlas ◽  
Epstein Barr

Epstein–Barr virus-associated gastric carcinoma (EBVaGC) is the most common malignancy caused by EBV infection. EBVaGC has definite histological characteristics similar to gastric carcinoma with lymphoid stroma. Clinically, EBVaGC has a significantly low frequency of lymph node metastasis compared with EBV-negative gastric cancer, resulting in a better prognosis. The Cancer Genome Atlas of gastric adenocarcinomas proposed a molecular classification divided into four molecular subtypes: (1) EBVaGC; (2) microsatellite instability; (3) chromosomal instability; and (4) genomically stable tumors. EBVaGC harbors a DNA methylation phenotype, PD-L1 and PD-L2 overexpression, and frequent alterations in the PIK3CA gene. We review clinical importance of EBVaGC and discuss novel therapeutic applications for EBVaGC.

scholarly journals Immune-Related Nine-MicroRNA Signature for Predicting the Prognosis of Gastric Cancer

2021 ◽  
Vol 12 ◽  
Author(s):  
Jingxuan Xu ◽  
Jian Wen ◽  
Shuangquan Li ◽  
Xian Shen ◽  
Tao You ◽  
...  
Keyword(s):  
Gastric Cancer ◽  
Cox Regression ◽  
Expression Profiles ◽  
Survival Rates ◽  
Clinical Importance ◽  
Risk Groups ◽  
Chemotherapeutic Agents ◽  
The Cancer Genome Atlas ◽  
Regression Analyses ◽  
New Biomarkers

Recent findings have demonstrated the superiority and utility of microRNAs (miRNAs) as new biomarkers for cancer diagnosis, therapy, and prognosis. In this study, to explore the prognostic value of immune-related miRNAs in gastric cancer (GC), we analyzed the miRNA-expression profiles of 389 patients with GC, using data deposited in The Cancer Genome Atlas database. Using a forward- and backward-variable selection and multivariate Cox regression analyses model, we identified a nine-miRNA signature (the “ImmiRSig,” consisting of miR-125b-5p, miR-99a-3p, miR-145-3p, miR-328-3p, miR-133a-5p, miR-1292-5p, miR-675-3p, miR-92b-5p, and miR-942-3p) in the training cohort that enabled the division of patients into high- and low-risk groups with significantly different survival rates. The ImmiRSig was successfully validated with an independent test cohort of 193 GC patients. Univariate and multivariate Cox regression analyses indicated that the ImmiRSig would serve as an independent prognostic factor after adjusting for other clinical covariates. Pending further prospective validation, the identified ImmiRSig appears to have significant clinical importance in terms of improving outcome predictions and guiding personalized treatment for patients with GC. Finally, significant associations between the ImmiRSig and the half-maximal inhibitory concentrations of chemotherapeutic agents were observed, suggesting that ImmiRSig may predict the clinical efficacy of chemotherapy.

scholarly journals Analytic approaches to clinical validation of results from preclinical models of glioblastoma: a systematic review

2021 ◽  
Author(s):  
Beth Fitt ◽  
Grace Loy ◽  
Edward Christopher ◽  
Paul M Brennan ◽  
Michael TC Poon
Keyword(s):  
Systematic Review ◽  
The Cancer Genome Atlas ◽  
Clinical Validation ◽  
Preclinical Models ◽  
Survival Analyses ◽  
Clinical Variables ◽  
Cohort Selection ◽  
Cancer Genome Atlas ◽  
Validation Of Results ◽  
Genome Atlas

AbstractBackgroundAnalytic approaches to clinical validation of results from preclinical models are important in assessment of their relevance to human disease. This systematic review examined consistency in reporting of glioblastoma cohorts from The Cancer Genome Atlas (TCGA) and assessed whether studies included patient characteristics in their survival analyses.MethodsWe searched Embase and Medline on 02Feb21 for studies using preclinical models of glioblastoma published after Jan2008 that used data from TCGA to validate the association between at least one molecular marker and overall survival in adult patients with glioblastoma. Main data items included cohort characteristics, statistical significance of the survival analysis, and model covariates.ResultsThere were 58 eligible studies from 1,751 non-duplicate records investigating 126 individual molecular markers. In 14 studies published between 2017 and 2020 using TCGA RNA microarray data that should have the same cohort, the median number of patients was 464.5 (interquartile range 220.5-525). Of the 15 molecular markers that underwent more than one univariable or multivariable survival analyses, five had discrepancies between studies. Covariates used in the 17 studies that used multivariable survival analyses were age (76.5%), pre-operative functional status (35.3%), sex (29.4%) MGMT promoter methylation (29.4%), radiotherapy (23.5%), chemotherapy (17.6%), IDH mutation (17.6%) and extent of resection (5.9%).ConclusionsPreclinical glioblastoma studies that used TCGA for validation did not provide sufficient information about their cohort selection and there were inconsistent results. Transparency in reporting and the use of analytic approaches that adjust for clinical variables can improve the reproducibility between studies.Importance of the StudyDespite using the same data from The Cancer Genome Atlas, translational preclinical studies in glioblastoma research included different numbers of patients into their analyses and their results were inconsistent.Fewer than a third of the studies used multivariable survival analysis to adjust for clinical variables but most did not take treatment factors into account.Greater transparency in cohort selection from open access data and integration of clinical variables into analyses will help improve reproducibility in glioblastoma research.

scholarly journals Deep generative neural network for accurate drug response imputation

2021 ◽  
Vol 12 (1) ◽  
Author(s):  
Peilin Jia ◽  
Ruifeng Hu ◽  
Guangsheng Pei ◽  
Yulin Dai ◽  
Yin-Ying Wang ◽  
...  
Keyword(s):  
Tumor Microenvironment ◽  
Cell Line ◽  
Drug Response ◽  
Dimensional Space ◽  
The Cancer Genome Atlas ◽  
Data Sets ◽  
Variational Autoencoder ◽  
Cancer Genome Atlas ◽  
Low Dimensional ◽  
The Impact

AbstractDrug response differs substantially in cancer patients due to inter- and intra-tumor heterogeneity. Particularly, transcriptome context, especially tumor microenvironment, has been shown playing a significant role in shaping the actual treatment outcome. In this study, we develop a deep variational autoencoder (VAE) model to compress thousands of genes into latent vectors in a low-dimensional space. We then demonstrate that these encoded vectors could accurately impute drug response, outperform standard signature-gene based approaches, and appropriately control the overfitting problem. We apply rigorous quality assessment and validation, including assessing the impact of cell line lineage, cross-validation, cross-panel evaluation, and application in independent clinical data sets, to warrant the accuracy of the imputed drug response in both cell lines and cancer samples. Specifically, the expression-regulated component (EReX) of the observed drug response achieves high correlation across panels. Using the well-trained models, we impute drug response of The Cancer Genome Atlas data and investigate the features and signatures associated with the imputed drug response, including cell line origins, somatic mutations and tumor mutation burdens, tumor microenvironment, and confounding factors. In summary, our deep learning method and the results are useful for the study of signatures and markers of drug response.

Sign in / Sign up

Export Citation Format

Share Document