Background: Gulf War veterans (GWV) experienced an unprecedented array of environmental exposures to chemicals, vaccines, oil fires, heat, and other factors that each contribute to a specific type of toxic effects to cells called "oxidative stress" (OS). It is this against which antioxidants seek to protect. There is evidence that some of these exposures are linked to Gulf War illness (GWI); and in fact mediate their toxicity via OS. Additionally, there is reason to suspect that GWI may be driven by cumulative exposure to OS. Large short-term exposures to external (environmental) OS can lead to long-term increases in internally produced OS. This is because OS can damage cell elements in a fashion that leads them to produce, on an ongoing basis, more oxygen free radicals (i.e. OS). (They may arise further if there are any depots of toxic substances in the tissue; but this is not necessary for long-term increases in OS.) This ongoing OS can produce further damage and may lead to symptoms that fit the pattern that ill GWV report. If this is shown to be true, it would be a breakthrough because it would lead to means to protect against similar problems in future personnel in military and also civilian settings. In addition, it may lead to approaches to treat ill GWV; and approaches to track the benefits of treatments being tested. Thus, it is important to determine if markers of OS can be identified that individually or in their pattern are linked to illness in GWV. There are a range of markers that are related to OS, and each has been found useful in a different complement of settings. Some protect against OS, or are driven up or down with OS, or both.

Goal: This study seeks to examine a spectrum of such markers related to OS, to see which markers, individually or as part of a pattern, discriminate GWI from controls; and relate to severity of illness within affected GWV.

Approach: To accomplish this we will recruit 40 affected GWV, and 40 healthy controls. In a first phase, we will look at half -- assessing a spectrum of OS markers in 20 with GWI and 20 controls to see which OS markers or patterns discriminate GWI, and are most strongly linked to illness severity within GWI. The most promising markers will be reappraised in the remaining 20 affected GWV and 20 controls to see whether patterns of markers retain their discriminating power. We capitalize on two innovative approaches that learn discrimination patterns by being told the answer ("supervised" learning via backpropagation neural networks); or by picking up on natural groupings in the data ("unsupervised" approach using a technique called Independent Component Analysis or ICA). A pioneer in both techniques will serve as a co-investigator and direct this portion of the study.

Benefits: This study will do several important things. It will determine the relation of GWI to OS -- a pivotally important factor that may be absolutely fundamental to advancing knowledge regarding GWV. It will enhance understanding of mechanism of GWI, which in turn will suggest approaches for promising treatments. It will yield objective markers tied to illness that can be tracked during testing of candidate treatments. By assessing which are the very best OS markers in relation to GWI, it will enhance quality and cost effectiveness of future studies that seek to include assessments of OS, enabling them to focus on only the best markers of OS and avoid unproductive markers. It will advance understanding the cause of GWI. Finally, it will directly suggest approaches to protect future servicepersons and also civilians from vulnerability to problems in similar settings. It may afford patterns of OS that enable objective diagnosis of GWI, overcoming the longstanding concerns and complications to GWV associated with GWI being defined purely by symptoms. Use of sophisticated unsupervised and supervised approaches that uncover higher order statistics in the data, undertaken by a premier authority in their use, will maximize prospects for that objective.