Predicting brain age with deep learning from raw imaging data results in a reliable and heritable biomarker.

Machine learning analysis of neuroimaging data can accurately predict chronological age in healthy people. Deviations from healthy brain ageing have been associated with cognitive impairment and disease. Here we sought to further establish the credentials of 'brain-predicted age' as a biomarker of individual differences in the brain ageing process, using a predictive modelling approach based on deep learning, and specifically convolutional neural networks (CNN), and applied to both pre-processed and raw T1-weighted MRI data. Firstly, we aimed to demonstrate the accuracy of CNN brain-predicted age using a large dataset of healthy adults (N = 2001). Next, we sought to establish the heritability of brain-predicted age using a sample of monozygotic and dizygotic female twins (N = 62). Thirdly, we examined the test-retest and multi-centre reliability of brain-predicted age using two samples (within-scanner N = 20; between-scanner N = 11). CNN brain-predicted ages were generated and compared to a Gaussian Process Regression (GPR) approach, on all datasets. Input data were grey matter (GM) or white matter (WM) volumetric maps generated by Statistical Parametric Mapping (SPM) or raw data. CNN accurately predicted chronological age using GM (correlation between brain-predicted age and chronological age r = 0.96, mean absolute error [MAE] = 4.16 years) and raw (r = 0.94, MAE = 4.65 years) data. This was comparable to GPR brain-predicted age using GM data (r = 0.95, MAE = 4.66 years). Brain-predicted age was a heritable phenotype for all models and input data (h2 ≥ 0.5). Brain-predicted age showed high test-retest reliability (intraclass correlation coefficient [ICC] = 0.90-0.99). Multi-centre reliability was more variable within high ICCs for GM (0.83-0.96) and poor-moderate levels for WM and raw data (0.51-0.77). Brain-predicted age represents an accurate, highly reliable and genetically-influenced phenotype, that has potential to be used as a biomarker of brain ageing. Moreover, age predictions can be accurately generated on raw T1-MRI data, substantially reducing computation time for novel data, bringing the process closer to giving real-time information on brain health in clinical settings.

Cell-specific enhancement of doxorubicin toxicity in human tumour cells by docosahexaenoic acid.

We examined the cytotoxicity of doxorubicin alone, or in combination with docosahexaenoic acid (22:6 n-3), in glioblastoma cell lines A-172 and U-87 MG and bronchial carcinoma cell lines A-427 and SK-LU-1. For both glioblastoma cell lines we found an enhanced cytotoxicity of doxorubicin when given with concentrations of docosahexaenoic acid that alone are non-toxic. In SK-LU-1 cells no such enhancement was observed, whereas a small increase was observed for A-427 cells. The enhanced cytotoxicity in glioblastoma cells was not caused by lipid peroxidation products. In A-427 cells, however, the modest potentiation could be explained by the formation of cytotoxic lipid peroxidation products. Se-glutathione peroxidase activity increased after doxorubicin exposure and even more after addition of Na-selenite, but this did not reduce the cytotoxicity of doxorubicin. These results demonstrated that the mechanisms of enhancement of cytotoxicity by docosahexaenoic acid are complex and cell-specific and do not require increased lipid peroxidation.

Acrolein cytotoxicity and glutathione depletion in n-3 fatty acid sensitive- and resistant human tumor cells.

Acrolein is a highly reactive unsaturated aldehyde formed endogenously and present in the environment. Acrolein efficiently reduces glutathione-contents and is highly cytotoxic in two lung carcinoma cell lines (A-427 and SK-LU-1) and the glioblastoma cell line A-172. A-427, which has the lowest GSH content of the cell lines, is also more sensitive to growth inhibition and more depleted in GSH after acrolein exposure. A-427 is also highly sensitive to docosahexaenoic acid (22:6 n-3, DHA) and acrolein potentiates the cytotoxic effect of DHA in this cell line, but not in the DHA-resistant cell lines SK-LU-1 and A-172. Surprisingly, the cytotoxic effect of acrolein was partially reversed by vitamin E, selenite and 2-phenyl-1,2-benzisoselenazol-3(2H)-one (ebselen, a Se-glutathione peroxidase mimic) in A-427 cells, but not in SK-LU-1 and A-172 cells. Using the TUNEL assay a strong nuclear fluorescence was observed in DHA-treated A-427 cells, indicating death by apoptosis, whereas acrolein apparently did not induce apoptosis.

Evidence that changes in Se-glutathione peroxidase levels affect the sensitivity of human tumour cell lines to n-3 fatty acids.

The human lung adenocarcinoma cell line A-427 is significantly more sensitive to cytotoxic lipid peroxidation products of eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) than the human lung adenocarcinoma cell line SK-LU-1, and the glioblastoma cell lines A-172 and U-87 MG. The cytotoxic effect as well as lipid peroxidation were abolished by vitamin E. The differential sensitivities of the cell lines were not correlated to the levels of lipid peroxidation products (measured as the end product malondialdehyde), indicating differences in sensitivities to products of lipid peroxidation. The high sensitivity of A-427 is apparently due to a low level of selenium-dependent glutathione peroxidase (GSH-Px), because pretreatment with sodium selenite (250 nM) increased the GSH-Px activity 3- to 4-fold and protected the cells almost completely against the growth inhibitory effect of DHA. Furthermore, 2-phenyl-1,2-benzisoselenazol-3(2H)-one (ebselen) a seleno-organic GSH-Px mimic, suppressed the cytotoxic action of DHA to A-427 in a dose dependent manner. Northern analysis demonstrated that pretreatment with sodium selenite (250 nM) was accompanied by an increased level of GSH-Px mRNA (1.8-fold) in A-427 cells, while the level remained unchanged under the same conditions in DHA/EPA-resistant A-172 cells. In addition, the level of selenophosphate synthetase mRNA (SelD), a key intermediate in tRNA(Sec) formation, increased 1.2- to 1.7-fold in A-427 and A-172 cells after pretreatment with sodium selenite. These results indicate that upregulation of GSH-Px activity by sodium selenite in the EPA/DHA sensitive cell line A-427 may be due to an increase in mRNAs for GSH-Px and a precursor important for formation of tRNA(Sec) which is required for incorporation of selenocysteine in GSH-Px during translation. These results demonstrate an important role for GSH-Px in the cellular defence against cytotoxic lipid peroxidation products. Furthermore, measurement of GSH-Px activities in tumour cells may be one useful biochemical predictor for their sensitivities to polyunsaturated fatty acids.