Dataset Information

Prediction of in vivo radiation dose in radiotherapy patients using ex vivo and in vivo gene expression signatures

ABSTRACT: After defining a gene expression signature that predicted radiation exposure dose with high accuracy in human peripheral white blood cells irradiated ex vivo, we now demonstrate the predictive power of gene expression signatures in blood from patients undergoing total body irradiation. Using whole genome microarray analysis, we have identified genes that respond to radiation exposure in cancer patients in vivo. A 3-nearest neighbor classifier built from these genes correctly predicted samples as exposed to 0, 1.25 or 3.75 Gy with 94% accuracy even when samples from healthy donor controls were included. The same samples were classified with 98% accuracy using a signature previously defined from ex vivo irradiation data. The samples could also be classified as exposed or not exposed with 100% accuracy using multiple methods. The demonstration that ex vivo irradiation is an appropriate model that can provide meaningful prediction of in vivo exposure, and that the signatures are robust across diverse disease states, is an important advance in the application of gene expression for biodosimetry. Translation of these signatures to a fully automated “lab-on-a-chip” device will enable high-throughput screening for large-scale radiological emergencies, as well as making such tests practical for clinical uses.

ORGANISM(S): Homo sapiens

PROVIDER: GSE20162 | GEO | 2011/01/31

SECONDARY ACCESSION(S): PRJNA125625

REPOSITORIES: GEO

ACCESS DATA

Similar Datasets

Project description:Background: Understanding individual patient host-response to viruses is key to designing optimal personalized therapy. Indeed, a subjectM-bM-^@M-^Ys epigenetic, transcriptomic and proteomic profiles dynamically respond to environmental challenges according to their intrinsic specific genetics. Unsurprisingly, in vivo human experimentation to understand individualized dynamic response of the transcriptome to viruses are rarely studied because of the obviously limitations stemming from ethical considerations of the clinical risk. In this rhinovirus study, we first establish that ex vivo human cells response to virus can serve as proxy for otherwise controversial in vivo human experimentation of viral infection. We further demonstrate that the N-of-1-pathways framework, that we previously validated for single patient M-bM-^@M-^Xomics analyses in cancer, can be utilized to understand individualized ex vivo response to viral stress and inform clinicians involved in personalized therapeutics. Method: N-of-1-pathways, validated retrospectively in whole tumors and experimentally in cancer cell lines, is a framework designed for identifying deregulated pathways at the single patient level between two M-bM-^@M-^Xomics profiles. It computes a significance score for a list of given genesets, using the M-bM-^@M-^Xomics profiles of a mere two samples as input (e.g. normal/tumoral, pre/post-treatment). We hypothesized that its usage could be extended to virus exposed cells. Additionally, we developed the Similarity Venn Diagram, an efficient and deceptively simple method for comparing results expressed in an ontology organized as a directed acyclic graph. We extracted the peripheral blood of four human subjects, with and without an ex vivo infection to rhinovirus. Their deregulated pathways were compared to those of 9 human subjects prior and after intranasal inoculation M-bM-^@M-^\in vivoM-bM-^@M-^] with rhinovirus. Results: We compared the N-of-1-pathways results using two established cohort-level methodologies: GSEA and enrichment of differentially expressed genes. Results are significantly and biologically similar between in vivo and ex vivo studies, both at the genes and enriched pathways levels. ROC curves demonstrate that deregulated pathways identified by N-of-1-pathways in cells each single subject infected ex vivo recapitulate the biologically relevant pathways observed in vivo in a whole cohort. Conclusion: In the context of less than five published transcriptomes of human viral infections in vivo and one ex vivo, the latter can be supplemented by 2-sample analyses yielding increased insight in individualized response without clinical risks. PBMCs incubated with viruses: The live PBMCs had been isolated from blood samples collected from four human subjects under a protocol approved by The University of Arizona Internal Review Board. Whole blood was obtained from donors and placed in heparin tubes that were centrifuged according to standard protocols to obtain PBMCs, then each aliquoted in two paired samples. Each sample of the pair was subsequently exposed to and incubated with either (i) Human Rhinovirus serotype 16 obtained from the American Type Culture Collection (RV; ATCCM-BM-. VR-283; ex vivo infected sample), or to (ii) sterile medium (control ex vivo non-infected sample) and incubated at 35M-BM-0C. This protocol resulted in 4 ex vivo infected + 4 ex vivo controls = 8 paired samples. RNA was extracted from these samples, amplified, tagged and hybridized on Affymetrix Human Gene 1.0 ST microarrays according to standard operating procedures. Dataset and preprocessing: Robust Multiple-array Average (RMA) normalization was applied on each patient data independently (2 paired samples at a time, to avoid bias in the single-patient experiments) using Affymetrix Power Tools (APT)

Dataset Information

Prediction of in vivo radiation dose in radiotherapy patients using ex vivo and in vivo gene expression signatures

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure