Browsing by Subject "Biomarker"

The accumulation of high-throughput data from vast sources has drawn a lot attentions to develop methods for extracting meaningful information out of the massive data. More interesting questions arise from how to combine the disparate information, which goes beyond modeling sparsity and dimension reduction. This dissertation focuses on the innovations in the area of heterogeneous data integration.

Chapter 1 contextualizes this dissertation by introducing different aspects of meta-analysis and model frameworks for high-dimensional genomic data.

Chapter 2 introduces a novel technique, joint Bayesian sparse factor analysis model, to vertically integrate multi-dimensional genomic data from different platforms.

Chapter 3 extends the above model to a nonparametric Bayes formula. It directly infers number of factors from a model-based approach.

On the other hand, chapter 4 deals with horizontal integration of diverse gene expression data; the model infers pathway activities across various experimental conditions.

All the methods mentioned above are demonstrated in both simulation studies and real data applications in chapters 2-4.

Finally, chapter 5 summarizes the dissertation and discusses future directions.

Introduction

Methods

Results

Discussions

Take-home message

Deep brain stimulation (DBS) is a highly successful neuromodulation therapy for treating the motor symptoms of Parkinson’s disease (PD). However, DBS has been used for over 30 years with little change in clinically used stimulation parameters and technology, and consequently, there have been few improvements in therapeutic efficacy during that period. Fortunately, recent advances in DBS devices and techniques, including automated stimulation parameter selection, directional leads, closed-loop stimulation, and model-optimized temporal patterns of stimulation, have the potential to improve symptom reduction, decrease side effects, and extend device battery life. However, making use of many of these techniques requires a recordable electrophysiological signal, or biomarker, that correlates strongly with clinical outcomes. The goals of this dissertation were to develop tools that assist in the recording and application of biomarkers, to characterize a new potential biomarker, the cortical evoked potential (cEP), and correlate it with symptom reduction, and to understand mechanistically how the cEP relates to symptom reduction during DBS. First, we quantified the effects of a novel electrodeposited platinum-iridium coating (EPIC) on single unit recording performance. We implanted electrodes in rats and used electrophysiological and histological measurements to compare quantitatively the single unit recording performance of coated vs. uncoated electrodes over a 12-week period. The coated electrodes had lower impedance, reduced noise, increased signal-to-noise ratio, and an increased number of discernible units per electrode as compared to the uncoated electrodes. These results demonstrated that EPIC electrodes provided recording performance and longevity superior to uncoated electrodes, thus improving our ability to quantify potential biomarkers from single unit recordings. Second, we developed a modified genetic algorithm (GA) designed to optimize temporal patterns of stimulation. We developed five modifications to the standard GA repopulation step that adapted the GA to design patterns for neuromodulation applications. We evaluated each modification individually and all modifications collectively by comparing performance to a standard GA across three test functions and two biophysically-based models of neural stimulation. The modifications improved performance across the test functions and performed best when all were used collectively. Thus, we developed a powerful tool for optimizing temporal patterns of stimulation using model-based proxies of DBS biomarkers. Third, we characterized a new candidate biomarker for DBS, the cEP, and quantified its correlation with symptom reduction during DBS. We used the unilateral 6-hydroxydopamine (6-OHDA) lesioned rat model or parkinsonism, with stimulating electrodes implanted in the subthalamic nucleus (STN) and the electrocorticography (ECoG) recorded above motor cortex (M1). We recorded the cEP during a range of stimulation conditions and while performing behavioral assessments of hypokinetic symptoms. The cEP was strongly affected by stimulation condition, and the cEP magnitude declined and the cEP latency increased with higher stimulation frequencies. These effects occurred over multiple minutes and with multiple time-scales. Additionally, the cEP magnitude and latency were each strongly correlated with symptom reduction during DBS, with correlations that were stronger and more consistent than those of conventional spectral-based biomarkers. This study demonstrated the potential utility of the cEP as a biomarker for symptom reduction from DBS. Fourth, to understand better how the cEP may relate mechanistically to symptom reduction from DBS, we developed a computational model of antidromic cortical activation during STN DBS and assessed the ability of DBS to desynchronize pathological cortical beta band oscillations. We tuned and validated the model using experimental data from the 6-OHDA lesioned rat model, and we implemented a stochastic model of antidromic spike failure, which is the presumed cause of the observed changes in the cEP magnitude, to determine how changes in the cEP relate to cortical desynchronization. STN DBS desynchronized pathological oscillations at high stimulation frequencies via a mechanism analogous to the informational lesion theory. Specifically, the DBS-evoked spikes masked the intrinsic pathological spiking via a combination of refractoriness, spike collision, and synaptic depletion. Further, the model revealed that antidromic spike failure played a critical role in shaping the therapeutic frequency profile of this masking effect, enforcing a parabolic shape with maximum desynchronization at ~130 Hz. The results in this dissertation advance our understanding of the therapeutic mechanism of STN DBS, provide important tools for application of electrophysiological biomarkers in DBS, and characterize the utility of the cEP as a potential biomarker to improve therapeutic outcomes.

The translational potential of diagnostic and prognostic platforms developed using expression microarray technology is evident. However, the majority of array-based diagnostics have yet to make their way into the clinical laboratory. Current approaches tend to focus on development of multi-gene classifiers of disease subtypes, but very few studies evaluate the translational potential of these assays. Likewise, only a handful of studies focus on development of approaches to optimize array-based tests for the ultimate goal of clinical utility. Prior to translation into the clinical setting, molecular diagnostic platforms should demonstrate a number of characteristics to ensure optimal and efficient testing and patient care. Assays should be accurate and precise, technically and biologically robust, and should take into account normal sources of biological variance that could ultimately affect test results. The overarching goal of the research presented in this dissertation is to develop methods for evaluating and optimizing the translational potential of molecular diagnostics developed using expression microarray technology.

The first research section of this dissertation is focused on our evaluation of the impact of intratumor heterogeneity on precision in microarray-based assays in breast cancer. We conducted genome-wide expression profiling on 50 needle core biopsies from 18 breast cancer patients. Global profiles of expression were characterized using unsupervised clustering methods and variance components models. Array-based measures of estrogen (ER) and progesterone receptor (PR) status were compared to immunohistochemistry. The precision of genomic predictors of ER pathway status, recurrence risk, and sensitivity to chemotherapeutics were evaluated by interclass correlation. Results demonstrated that intratumor variation was substantially less than the total variation observed across the patient population. Nevertheless, a fraction of genes exhibited significant intratumor heterogeneity in expression. A high degree of reproducibility was observed in single gene predictors of ER (intraclass correlation coefficient (ICC)=0.94) and PR expression (ICC=0.90), and in a multi-gene predictor of ER pathway activation (ICC=0.98) with high concordance with immunohistochemistry. Substantial agreement was also observed for multi-gene signatures of cancer recurrence (ICC=0.71), and chemotherapeutic sensitivity (ICC=0.72 and 0.64). Together, these results demonstrated that intratumor heterogeneity, although present at the level of individual gene expression, does not preclude precise micro-array based predictions of tumor behavior or clinical outcome in breast cancer patients.

Leading into the second research section, we observed that in some cancer types, certain genes behave as molecular switches and have either an "on" or "off" expression state. Specifically, we observed these molecular switch genes exist in breast cancer as robust diagnostic and prognostic markers, including ER, PR, and HER2, and define tumor subtypes associated with different treatment and patient survival. We hypothesized that clinically relevant molecular switch (bimodal) genes exist in epithelial ovarian cancer, a type of cancer with no established molecular subgroups. To test this hypothesis, we applied a bimodal discovery algorithm to a publically available ovarian cancer expression microarray dataset (GSE9891:285 tumors; 246 malignant serous (MS), 20 endometrioid (EM), 18 low malignant potential (LMP) ovarian carcinomas). Genes with robust bimodal expression were identified across all ovarian tumor types and within selected subtypes. Of these bimodal genes, 73 demonstrated differential expression between LMP vs. MS and EM, and 22 genes distinguished MS from EM. Fourteen bimodal genes had significant association with survival among MS tumors. When these genes were combined into a single survival score, the median survival for patients with a favorable versus unfavorable score was 65 versus 29 months (p<0.0001, HR=0.4221). Two independent datasets (n=53 high grade, advanced stage serous and n=119 advanced stage ovarian tumors) validated the survival score performance. Taken together, the results of this study revealed that genes with bimodal expression patterns not only define clinically relevant molecular subtypes of ovarian carcinoma, but also provide ideal targets for translation into the clinical laboratory.

Finally, the third research section of this dissertation focuses on development of robust blood-based molecular markers of decompression stress (DS). DS is defined as the pathophysiological response to inert gas coming out of solution in the blood and tissues when a body experiences a reduction in ambient pressure. To date, there are no established molecular markers of DS. We hypothesized that comparing gene expression before and after human decompression exposures by genome-wide expression profiling would identify candidate molecular markers of DS. Peripheral blood was collected 1hr before and 2hr after 93 hyperoxic, heliox experimental dives (n=54). Control arms included samples collected 1 hour before and 2 hours after high pressure oxygen breathing (n= 9) and surface exercise (n=9), and samples collected at 7am and 5pm for time of day (n=11). Pre and post-dive expression data collected from normoxic nitrox experimental dives were utilized for independent validation. Blood samples were collected into PaxGene RNA tubes. RNA was extracted and processed for globin reduction prior to cDNA synthesis and Affymetrix U133A GeneChip hybridization. 746 genes were differentially expressed following hyperoxic, heliox decompression exposures (permutation adjusted p-value cutoff 1.0E-4). After filtering control significant genes, 726 genes remained. Pathway analysis demonstrated a significant portion of genes were associated with innate immune response (p<0.0001). A 362 multi-gene signature of significant, covariant genes was then applied to the independent dataset and demonstrated differentiation between pre and post-dive samples (p=0.0058). There was no significant correlation between signature and venous bubble grade or bottom time in the validation study. Our results showed that expression profiling of peripheral blood following decompression exposures, while controlling for experimental and normal sources of biological variance, identifies a reproducible multi-gene signature of differentially expressed genes, primarily comprising genes associated with innate immune response and independent of venous bubble grade or dive profile.

Taken together, the research and results presented in this dissertation represent considerable advances in the development of approaches to guide microarray-based diagnostics towards the ultimate goal of clinical translation.

Noble metal nanoparticles exhibit unique optical properties that are beneficial to a variety of applications, including molecular imaging. The large scattering cross sections of nanoparticles provide high contrast necessary for biomarkers. Unlike alternative contrast agents, nanoparticles provide refractive index sensitivity revealing information regarding the local cellular environment. Altering the shape and composition of the nanoparticle shifts the peak resonant wavelength of scattered light, allowing for implementation of multiple spectrally distinct tags. In this project, nanoparticles that scatter in different spectral windows are functionalized with various antibodies recognizing extra-cellular receptors integral to cancer progression. A hyperspectral imaging system is developed, allowing for visualization and spectral characterization of cells labeled with these conjugates. Various molecular imaging and microspectroscopy applications of plasmonic nanoparticles are then investigated. First, anti-EGFR gold nanospheres are shown to quantitatively measure receptor expression with similar performance to fluorescence assays. Second, anti-EGFR gold nanorods and novel anti-IGF-1R silver nanospheres are implemented to indicate local cellular refractive indices. Third, because biosensing capabilities of nanoparticle tags may be limited by plasmonic coupling, polarization mapping is investigated as a method to discern these effects. Fourth, plasmonic coupling is tested to monitor HER-2 dimerization. Experiments reveal the interparticle conformation of proximal HER-2 bound labels, required for plasmonic coupling-enhanced dielectric sensing. Fifth, all three functionalized plasmonic tags are implemented simultaneously to indicate clinically relevant cell immunophenotype information and changes in the cellular dielectric environment. Finally, flow cytometry experiments are conducted utilizing the anti-EGFR nanorod tag to demonstrate profiling of receptor expression distribution and potential increased multiplexing capability.

Computed Tomography (CT) is a prevalent imaging technique in modern medicine that provides physicians a non-invasive method to evaluate and diagnose various clinical conditions. To aid in diagnosis, it is important to have a high accuracy and reliability in these images. In the first phase of this study, the variability of clinically-relevant imaging biomarkers was analyzed across different scanners and imaging parameters through usage of a customized anthropomorphic chest phantom with several experimental sample inserts. This phantom was scanned across 10 different scanners. Imaging biomarkers were computed for each scan. Intra and inter-scan variability was assessed by computing coefficients of variation and standard deviations of the measurements. It was found that LAA -950 and LAA -856 were the biomarkers with the highest levels of variability, while the majority of other biomarkers had variability less than 10 HU or 10% CV in both inter and intra-scan measurements. No clear trend was found between the variability of the biomarkers and radiation dose (i.e., CTDI).

Traditional assessments of CT technologies are limited in the sense that they work with real patient data and are not efficient. Alternatively, Virtual Imaging Trials (VITs), which use virtual scanners and patients, are more efficient and avoid unnecessary radiation exposure. DukeSim is a CT simulator that has been validated with simple cylindrical phantoms in the past, but not with more clinically-relevant phantoms and conditions. Biomarkers computed from real CT image data were compared to those from simulated CT scans of a computational version of an anthropomorphic chest phantom. Overall, relative percent errors ranged from 0.187% to 18.269%.

Having validated DukeSim in a clinically relevant context, the utility of DukeSim as a VIT tool was shown by investigating the effects of imaging and reconstruction parameters on the clinically relevant biomarkers. It was found that sharper reconstruction kernels and lower tube currents tended to reduce the accuracy of measured biomarkers. These findings will help to spark further studies in virtual imaging, which can help to yield further clinical insights to improve patient health outcomes.