Analysis of Longitudinal High Dimensional Data for Burn and Trauma Studies

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): Severe burns are characterized by a hypermetabolic stress response which causes a loss of body mass and muscle wasting, immunologic compromise, slowed wound healing, and related bone loss, all of which contribute to increased morbidity, mortality, and prolonged recovery from injury. In fact, pediatric burn victims, after their acute hospitalization, are severely weakened, and often do not grow for a year post injury. The NIH is funding an Inflammation and the Host Response to Injury study (also known as the Glue grant because it is a consortium of clinical and basic scientists) to more deeply understand the complex set of events that culminate in the body's immune response to the injury of burn and trauma. One aim of the Glue grant in the burn population is to correlate the well defined phenotypic changes in metabolism in patients with burns with genetic and proteomic expression profiles analyzed from blood, fat, muscle and skin with the goal of identifying families of genes that are modulated in a time course prior to a physiologic response of tissues after a burn. The primary objective of this proposal is to develop computationally tractable methods of statistical analysis for identifying gene arrays that follow a distinct pattern over the time prior to a phenotypic change in burn patients, as indicated by the return of metabolic markers and growth to normal levels. The statistical analysis of these data cannot be performed using standard available methods: in fact, the analysis of gene microarrays presents unique statistical challenges because of the high- dimensionality of the data. This issue is further complicated when microarray data are collected serially over time. Additionally, when the phenotype is an event time, the methodology is further complicated by the fact that for some patients the event may be censored by the end of follow-up or death. We propose to adapt methods for analyzing time to event data to the setting of longitudinally collected microarray data. We will develop a score test that is sensitive to which genes are expressed differentially in patients who recover. We will extend these methods to handle mortality and to develop models that can utilize longitudinal microarrays to predict which patient will recover from severe trauma or burn. We will apply the methods developed in this grant to study the longitudinal genomic correlates to various clinical outcomes in data from pediatric burn patients in the Glue study. The objective of this proposal is to develop statistical methods for identifying gene expression patterns over time that would predict changes in burn patients that are associated with the return of metabolic markers and growth to normal levels. These methods will be applied to data from pediatric burn patients in the NIH-funded Inflammation and the Host Response to Injury (Glue) project. This will provide a better understanding of the complex set of events that culminate in the body's immune response to burn injuries that contributes to increased morbidity and mortality in these patients. This knowledge can be used to improve the treatment of severely burned patients resulting in more rapid and complete recovery.