Premium
A novel method for integrating transcriptomics and neuroimaging
Author(s) -
Povala Guilherme,
Bastiani Marco Antônio,
Bellaver Bruna,
Ferreira Pamela C.L.,
Souza Débora Guerini,
Brum Wagner Scheeren,
Zatt Bruno,
Zimmer Eduardo R.
Publication year - 2020
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1002/alz.039779
Subject(s) - neuroimaging , voxel , positron emission tomography , transcriptome , context (archaeology) , standardized uptake value , pet imaging , nuclear medicine , medicine , computational biology , gene expression , neuroscience , biology , gene , radiology , genetics , paleontology
Abstract Background Positron emission tomography (PET) imaging has been playing a fundamental role in diagnosis of Alzheimer’s disease (AD). Also, in the context of AD, blood‐based biomarkers that are capable of predicting PET brain imaging findings are of high interest. Both PET imaging and transcriptomics are rich sources of different biological information. Hence, an interesting strategy to find potential novel blood biomarkers is by integrating PET imaging data with blood transcriptomics. We aim to develop a method for combining blood transcriptomics profile and PET imaging data, which may highlight novel AD biomarkers. Here, we hypothesize that by integrating blood transcriptomics and PET imaging data we will be able to identify clinically relevant novel peripheral biomarkers. Method Imaging and transcriptomics data were acquired from Alzheimer’s Disease Neuroimaging Initiative (ADNI). Microarray gene expression profiling from blood samples of 69 cognitively unimpaired (CU) individuals and 158 mild cognitively impaired (MCI) were submitted to differential expression (DE) analysis using the limma R package. The [F 18 ]FDG‐PET standardized uptake ratio (SUVr), using cerebellum as the reference region, were calculated. Genes obtained from DE analysis were selected to undergo integration with [F 18 ]FDG‐PET images using voxel‐wise generalized linear regressions (GLR) (RMINC package). Results The DE analysis resulted in 1232 differentially expressed genes (DEGs) (p‐value < 0.05). The GLR computed the associations between gene expression and [F 18 ]FDG for each voxel, resulting in t‐value maps. Afterwards, only gray matter voxels presenting absolute t‐values higher than 2.3 were retained. Then, we transformed t‐value maps into proportion maps. In brief, each volume of interest (VOI) shows the percentage of voxels statistically correlated with gene expression (see Figure 1). Finally, we ranked DEGs according to the amount of VOIs with a proportion higher than 30%. Conclusion With the use of the proposed method, we were able to integrate blood transcriptomics with PET neuroimaging. The implementation of this method shows great potential in the search for new blood biomarkers, which can accelerate early diagnosis and provides a template for future research in the field.