Finally we searched the Allele Frequencies in Worldwide

Finally, we searched the “Allele*Frequencies in Worldwide Populations” (http://www.allelefrequencies.net/) website for known haplotypes related to the 6 discriminating alleles, within the 3 databases that we used. Haplotype information was available only for the SF database, as shown in Table 5. It can be seen that no identified haplotype contains more than one of the 6 discriminating alleles, indicating their unique contribution.
Discussion These results document for the first time significant differences in the HLA makeup between veterans with GWI and Gulf War era veterans without it. All the evidence from this study points to an enhanced vulnerability (or lack of protection) of the GWI veterans and, conversely, a reduced vulnerability (or additional protection) of the healthy veterans who served in the Gulf War but did not suffer from GWI. Collectively, our findings are in keeping with other evidence for an immune dysfunction in GWI (Broderick et al., 2012, 2013; Craddock et al., 2015; Hotopf et al., 2000; Israeli, 2012; Moss, 2013; Toubi, 2012; Whistler et al., 2009), in addition to other factors, including inflammatory components (Broderick et al., 2012, 2013; Johnson et al., 2013), mitochondria dysfunction (Koslik et al., 2014) and genetic variants regarding butyrylcholinesterase enzyme activity (Steele et al., 2015). In fact, our findings provide a genetic susceptibility framework within which environmental triggers (e.g. vaccines, exposure to chemicals, stress, etc.) can be discussed, investigated and evaluated. All six HLA hexokinase inhibitor that were singled out by the discriminant analysis to classify successfully (Fig. 1) control and GWI participants belonged in Class II and came from all three major genes (DQB1, DPB1, DRB1). Two alleles (DQB1*02:02 and DRB1*08:11) were absent from the GWI population, and the frequencies of the remaining four were all lower in the GWI group than in the control group (Table 4A). In addition, the percentage of participants with any of the six alleles was lower than in the controls (Table 4B). These results, together, document the lower frequency of occurrence of these alleles in the GWI group, as compared to the control group. The results of the regression analysis further documented the protective association of those alleles, given the significant negative slope in Fig. 2. In addition, this analysis further differentiated this effect among specific symptom domains, since it showed highly significant overall protective effects for Pain, Fatigue, Neurological-Cognitive-Mood domains. Remarkably, the effect of each allele on symptom severity in each one of these domains was protective, as evidenced by the universally negative partial regression coefficients in those three separate regression analyses. In contrast, the effect was not significant for the Skin, Gastrointestinal and Respiratory domains. The comparison of the observed allele frequencies to those reported in published databases (Ovsyannikova et al., 2005; Rossman et al., 2003; Skibola et al., 2012) (Figs. 3–5) documented an additional finding, namely that both GWI and controls differed systematically and in opposite directions (higher for controls, lower for GWI, Figs. 4 and 5) with respect to the published allele frequencies. This suggests the following hypothesis regarding the role of the six discriminating alleles we identified. First, we assume that the population of GW-era veterans came from a larger population whose allele frequencies would be very similar to those reported in the three databases we used in this study. This is a reasonable assumption. Second, it is known that GW veterans were administered various vaccines, possibly together and/or multiply (Institute of Medicine, 2000; Petersdorf et al., 1996; Schwartz et al., 1997; Steele, 2000) and they were exposed to various chemical agents (Institute of Medicine, 2000; Steele et al., 2015). Third, a proportion of GW veterans developed GWI consisting of chronic symptoms affecting multiple organ systems (Fukuda et al., 1998; Institute of Medicine, 2006, 2010; Kang et al., 2009; Kelsall et al., 2009; Steele, 2000), mostly affecting veterans who were deployed but also present in non-deployed or minimally exposed veterans (Steele, 2000). Based on those facts, we hypothesize that vaccinations and/or chemical exposures of GW era veterans served as environmental triggers (“hits”) that contributed to developing GWI in genetically (HLA) “vulnerable” veterans. Specifically, we propose that the presence of certain HLA alleles in higher frequencies conferred protection, whereas their relative paucity conferred vulnerability. This, in turn, resulted in the two distinct subpopulations of our study, namely control, deployed GW veterans with higher allele frequencies and absence of GWI on the one hand, and deployed GW veterans with lower frequencies and presence of GWI (Fig. 3). It should be noted that both vaccinations (Hotopf et al., 2000; Israeli, 2012; Toubi, 2012; Rook and Zumla, 1997) and chemical exposures (Moss, 2013) have been implicated previously as contributing factors for the development of GWI. The results of our study simply add a genetic susceptibility framework within which the effects of the factors above could be interpreted and investigated in future work. The nature of this postulated protection is likely to relate to autoimmune as well as inflammatory processes, since HLA has been implicated in both (Trowsdale and Knight, 2013; Johnson et al., 2013; Blackwell et al., 2009). In addition, HLA genetic underpinnings in immune responses to vaccines have been well established (Ovsyannikova et al., 2006; Ovsyannikova and Poland, 2011; Poland et al., 2007, 2008a, 2008b). Specific HLA makeup could thus manifest as autoimmune reactions, aberrant immune response to vaccinations, and/or increased susceptibility to infections (Nicolson, 1998); all three of them have been implicated in GWI.