Fourteen regression models (i.e., seven DVs each with two tests run, one model with HCV status entered as a predictor on its own, and a second model with HCV status and the immune factors entered as predictors together) were calculated. A Bonferroni correction with a cutoff

of P = 0.0035 (i.e., 0.05/14 tests) determined if the models were significant after a correction. Thus, the Inhibitors,research,lifescience,medical P-value for the omnibus model must be below this cutoff for models to be significant; individual variable P-values are considered significant if <0.05 as long as the model is significant. For the models with only HCV Status entered, only the FSS was significant. For the models with multiple analytes Inhibitors,research,lifescience,medical entered, the significant models were for Depression Total, Depression-Cognitive Affective, Depression-Somatic, Anxiety, and Fatigue. The two pain scales had P-values above this threshold. Results from these models are interpreted cautiously. Ignoring the correction briefly, as summarized in Table 4, and consistent with the group comparisons in Table 2, HCV status was a significant predictor of increased Depression-Total, Depression-Somatic Factor, Anxiety, Fatigue, and Pain Interference Inhibitors,research,lifescience,medical in regression

analyses with HCV status entered as the only independent variable. Depression-Cognitive Affective and Pain Severity were not significant in either check details Tables 2012 or 2000. Therefore, HCV status was entered as an independent variable along with the 33 detectable immune factors Inhibitors,research,lifescience,medical in subsequent regression analyses. In the final regression models (Table 4), HCV status was a significant predictor of the severity of Depression-Total, Depression-Cognitive Affective Factor, Depression-Somatic Factor, and Anxiety. All of the final regression models accounted for a larger percentage of the variance in each neuropsychiatric variable than HCV status alone. The final regression models yielded protein signatures of 4–10 plasma immune

factors that significantly predicted Inhibitors,research,lifescience,medical the severity of each neuropsychiatric variable. The protein signatures accounted for 36% of the variance in Depression-Total (seven factors), 40% in Depression-Cognitive Affective Factor (10 factors), 31% in Depression-Somatic Factor (eight factors), 25% in Anxiety (four factors), and 34% in Fatigue (seven factors). Results were interpreted cautiously because models were significant, but not at the corrected level for Pain Severity (19%; four factors) and Pain Interference others (20%; four factors). Because of extant group differences in rates of hypertension, asthma, and current tobacco use (Table 2), we evaluated whether HCV was a proxy that would account for differences in our models (Sluzewska et al. 1995; Maes et al. 1999, 2009; Kubera et al. 2001; O’Brien et al. 2007). To do this, we conducted linear regression analyses to evaluate these variables as potential covariates in our models.