Abstract

Preeclampsia (PE) is a multisystem disorder that contributes to maternal and fetal mortality and morbidity worldwide. It is characterized by de-novo hypertension and proteinuria or other maternal organ damage after 20 weeks of gestation. Evidence suggested that endogenous digitalis-like factor (EDLF) contributes to the pathogenesis of PE, and that the potential source of EDLF is the placenta. EDLF can inhibit the sodium pump (SP) specifically and may lead to hypertension, it has also been associated with hypoxia, oxidative stress and other abnormalitites present in PE.We studied whether normal human placenta responded to SP inhibition casued by EDLF with a change in abundance of lipids in the placental cytosol, and whether there was a characteristic set of lipid changes that could serve as a signature for EDLF exposure if there were such changes. Placenta tissues from 20 normal pregnancies were incubated for 48 hr in the presence and absence of ouabain, a widely studied EDLF, followed by tissue homogenization, lipid extraction, and the study of lipids using a mass spectrometery (MS) based lipidomics approach. 1207 lipidomic markers were surveyed by paired Student t-test, among which 26 markers had significantly different abundances between cases and control at the FDR=0.05 level. A set of 8 lipidomic markers were selected by a statistical model built with a sparse partial least squares discriminant analysis method (sPLS-DA) and a bootstrap procedure. All eight markers were then chemically characterized and partially identified using tandem MS. These markers might be used to identify placentas that have been previously exposed to EDLF in return.Endogenous peptides and small proteins might contribute to the pathophysiology of various diseases. Therefore, we investigated the potential peptidomic profile of placenta tissues in response to EDLF exposure as well. Placenta tissues from 20 normal pregnancies were incubated for 25 hr with and without the addition of ouabain, followed by homogenization, protein depletion, and the study of the peptides by a LC-MS based peptidomics approach. 275 peptidomic markers were evaluated by Student t-test. A set of 8 markers was chosen using a logistic regression model build with the Akaike information criterion (AIC). However, no peptidomics markers or set of markers showed specific, statististically significantly different changes in abundances between cases and controls after applying a false discovery rate (FDR) correction or using more conservative methods to overcome over-fitting. Using an optimal sPLS- DA, cross-validation studies and logistic regression models, we also found that the addition of any peptidomic marker to the previously selected lipidomic profile was unlikely to help identify placentas that had been exposed to EDLF.Alzheimer's disease (AD) is the most common form of dementia and the number of AD cases worldwide is currently estimated to be 36 million. The exact pathogenesis of AD remainsiielusive and available therapeutic strategies can only delay its progession temporarily. Several hypotheses have been proposed regarding the pathophysiology of AD and the beta-amyloid (AÎ²) hypothesis is considered the core mechanism. However, the majority of studies concerning AD, or AD biomarkers specifically, have ignored a potentially important variable that is gender, despite reported gender differences in the risk of developing AD, the risk factors, clinical symptoms and CSF biomarkers of the disease, among many other aspects.We analyzed data obtained from a previous study of diagnostic serum lipid biomarkers for AD with the consideration of potential gender difference. Firstly, we studied the interaction between gender and disease stage using analysis of variance (ANOVA) and analysis of covariance (ANCOVA). Lipid markers that showed statistically significant interaction were selected after applying a FDR correction. Secondly, using a lasso logistic regression model with binary classification (control vs. all AD stages), we identified gender-specific markers and found different coefficient estimates for different genders as well. Lastly, we build a new ordinal model with the addition of a gender-specific marker using a Bayesian lasso probit ordinal regression model. The predictive performance of the new model was found to be statistically significantly better than the previous model which was built without the consideration of gender.In conclusion, we successfully discovered, chemically characterized lipidomic markers indicative of EDLF exposure in placenta and detected gender-specific lipid markers for AD.

Degree

College and Department

Physical and Mathematical Sciences; Chemistry and Biochemistry

Rights

http://lib.byu.edu/about/copyright/