Sixteen healthy male students, mean age (± SD) 24.9 (± 4.8) years, were recruited from University College London (UCL) campus advertisements. Ten were Caucasian, 3 Indian-Asian, 2 Chinese-Asian, and 1 Latin-American. Volunteers were reviewed by a psychiatrist (NAH) and screened for a history of any relevant physical or psychiatric illness. One participant had a history of hay fever, and another had shellfish allergy. Four participants rated their general health as excellent, 7 very good, and 5 good. No participant rated their general health as poor or fair. All were medication free, with no nonsteroidal or steroidal inflammatory drug use in the preceding 2 weeks, and were nonsmokers. Volunteers who had received typhoid vaccine within 3 years or other vaccine within 6 months were excluded. Participants were advised to not consume caffeinated beverages or alcohol, avoid high-fat meals, and refrain from excessive exercise for 12 hours before testing. They were asked not to take aspirin, ibuprofen, or antibiotics for 14 days before testing. After complete description of the study to the subjects, written informed consent was obtained. Procedures were approved by the joint UCL/UCL Hospital Ethics Committee. Findings in the same group of participants are also reported elsewhere (19,16
We adopted a randomized, double-blind, cross-over repeated measures design in which all participants underwent imaging in two separate sessions, an average of 7 days apart as reported previously (16
). In the first session participants were randomly assigned to one of two experimental conditions (typhoid vaccine or placebo). Baseline blood sample was taken; then injections of .025 mg of Salmonella typhi
capsular polysaccharide vaccine (Typhim Vi, Aventis Pasteur MSD, Berkshire, United Kingdom) or .5 mL of normal saline placebo were administered intramuscularly into the deltoid muscle. An fMRI was performed 2 hours after injection in a 60-min session. During each session, participants performed three tasks. This article focuses on data acquired during an implicit facial-affect processing task. Immediately after scanning, a second blood sample was taken (3 hours after injection) for cytokine measurement. Body temperature was assessed at baseline and at 2 and 3 hours with a sublingual digital thermometer. See for study timeline. The second session was identical except that participants received the other injection (i.e., typhoid vaccination if they previously received saline and vice versa).
Figure 1 Study timeline. Participants completed mood rating questionnaires (Profile of Mood States [POMS]) and underwent venesection, then randomly received Salmonella typhi capsular polysaccharide vaccination (Typhim Vi, Aventis Pasteur MSD, Berkshire, United (more ...)
Mood and other psychological symptoms were assessed with a modified version of the Profile of Mood States (POMS) (21
). This consisted of 36 items, each of which was rated on a five-point scale (0 = not at all to 4 = extremely). Six items were taken from the vigor, tension-anxiety, depression-dejection, and confusion scales of the POMS and five from the fatigue scale. In addition, there were four symptom (feverish, aching joints, nauseated, and headache) and three filler items. Participants were asked to rate how they felt at that moment. Scores for the five POMS subscales were computed by summing ratings on individual items. Total mood scores were derived by the standard method detailed in the POMS rating manual of subtracting ratings on the negative scales (tension-anxiety, depression-dejection, confusion, and fatigue) from the vigor scores (21
). Of note, this method produces a composite total mood score that is sensitive to changes in cognitive-mood and neurovegetative contributions to mood.
We adopted a model of mild experimental inflammation with standard typhoid (Salmonella typhi
) vaccination that has previously been shown to induce both a low-grade toll-like-receptor-4 mediated inflammatory-cytokine response (associated with an approximate doubling of peripheral IL-6 levels peaking between 2 and 3 hours) (22
) and a transient negative total mood (peaking 1.5–3 hours after injection) (10,11
). Injection of .025 mg Salmonella typhi
vaccine (Typhim Vi, Aventis Pasteur MSD) or .9% sodium chloride placebo in identical 2-mL syringes was administered intramuscularly into the deltoid muscle by a qualified doctor (NAH). There were no complications of either injection.
Separate venepunctures were performed at baseline and 3 hours after injection for vaccine and placebo conditions. Blood (10 mL) was drawn into Vacutainer tubes (Becton Dickinson and Company, Franklin Lakes, New Jersey) containing ethylenediaminetetraacetic acid (EDTA) anticoagulant, centrifuged immediately at 1250 g for 10 min at room temperature. Plasma was removed, aliquoted, and frozen at −70°C before analysis. Plasma IL-6 and tumor necrosis factor α (TNF-α) were assessed with high-sensitivity, two-site enzyme-linked immunosorbent assays (ELISAs) (R&D Systems, Oxford, United Kingdom). The limit of detection of the IL-6 assay was .09 pg/mL, with intra- and interassay coefficients of variation (CVs) of 5.3% and 9.2%, respectively. The TNF-α assay had a detection limit of .10 pg/mL with intra- and interassay CVs of 6.9% and 8.4%, respectively. Plasma IL-1RA concentrations were determined by a commercial ELISA from R&D Systems. This assay had a limit of detection of 15 pg/mL and inter- and intra-assay CVs of <10%. Salivary cortisol was collected with cotton dental rolls at baseline and at 2 and 3 hours (Salivettes, Sarstedt, Leicester, United Kingdom) and analyzed with a time resolved immunoassay with fluorescence detection. Intra- and interassay variability were <8%.
Twenty faces (10 male) from a standardized series of facial emotional expressions (Karolinska-Directed-Emotional-Faces-Set [KDEF]) (23
) were selected displaying happy, sad, angry, and neutral expressions. Each face was presented for 500 msec in random order with an intertrial interval of 3400 msec. Each identity-expression combination was presented 4 times, along with 48 baseline trials (cross hair fixation). The same stimulation set was used after both vaccine and normal saline placebo injection. Participants performed an incidental age judgment task: indicating with a right-handed response pad if older or younger than 25 years of age. An orthogonal task and short stimulus presentation were chosen to elicit incidental affective processing that previous data suggest might precede explicit processing (24
Gradient-echo single-shot echo planar imaging was used to acquire T2*-weighted image volumes on a 1.5-T Siemens Sonata (Siemens AG Medical Solutions, Erlangen, Germany) scanner equipped with a standard head-coil. External restraint was used to minimize head movement. We acquired 284 volumes each with 44 slices (contiguous 2-mm slices with 1-mm inter-slice gap, echo-time 40 msec: spatial resolution 3 mm × 3 mm × 3 mm). Slices were tilted −30° from the intercommissural plane to reduce orbitofrontal dropout due to susceptibility artifact from frontal sinuses (25
). High-resolution inversion-recovery echo planar images were also obtained to aid image registration.
The fMRI data were analyzed with SPM5 (http://www.fil.ion.ucl.ac.uk/spm
). The first 5 volumes were discarded to allow for T1 equilibration. Individual scans were realigned, unwarped, normalized, and spatially smoothed with an 8-mm full-width-at-half-maximal Gaussian kernel with standard SPM methods. High-pass frequency filter (cut-off 120 sec) and corrections for auto-correlation between consecutive scans (auto-regressive [AR]1) were applied to the time series. Each event was modeled by a standard synthetic hemodynamic response function at each voxel across the whole brain. Presentations of neutral, happy, sad, and angry facial expressions were modeled as separate regressors. Null events (15% of presentations) were included to facilitate identification of hemodynamic responses to stochastically ordered stimuli.
First-level individualized design matrices were estimated in the following manner: effects of task (viewing happy, sad, angry, and neutral facial expressions) were computed on a voxel-wise basis for each participant for both vaccination and placebo conditions in the form of SPMs of discrete contrasts within the general linear model. Subsequent second-level analyses were performed on the SPM contrast images with a 4 (emotional expression) × 2 (inflammatory status) factorial design to permit formal inferences about population effects.
The main effect of viewing facial expressions (across both vaccine and placebo conditions) was calculated within a second-level analysis of variance encompassing individual contrasts for each of the emotional expressions. Functionally activated clusters in bilateral fusiform gyrus (fusiform face area [FFA]) and STS were identified at a family-wise error corrected threshold (FWE) of p <
.05 and extracted with the image analysis package MarsBaR (26
). Anatomical localizers from the same package were used for the amygdala bilaterally. Effects of inflammation on activity in each of these regions was determined by extracting the contrast estimates from the peak voxel in each region and analyzing it in a 4 (emotional expression) × 2 (inflammatory status) repeated measures analysis of variance in SPSS (SPSS, Chicago, Illinois).
We then performed a between-subject analysis to determine regions in which response to implicit observation of emotional facial expressions was modulated as a function of inflammation-associated total mood change. This whole brain regression analysis was performed with individual activation maps to facial emotional expressions (vs. implicit baseline) with inflammation-associated mood change as the between-subject dependent variable. Results are reported for the STS, FFA, and amygdala at uncorrected and stringent small volume FWE corrected thresholds.
We extended this approach to determine the effects of inflammation-associated mood change on brain regions previously implicated in the etiology of depression, targeting specific cortical (sACC, medial prefrontal, anterior cingulate) and subcortical (thalamus, nucleus accumbens) regions of interest, informed by prior research by Mayberg et al.
) and others. This whole brain between-subject regression analysis (with inflammation-associated total mood change as the dependent variable) identified bilateral sACC as the region showing the strongest positive correlation with inflammation-associated deterioration in total mood in keeping with previous analyses on the neurobiological basis of depression (27
) and anhedonia (28
). Of note, no region outside of this predefined region of interest showed a significant correlation with inflammation-associated mood change at an uncorrected threshold of p
< .001, 10 contiguous voxels.
We next performed an effective connectivity analysis to test for changes in interregional neural connectivity related to the psychophysiological interaction (PPI) between activity in sACC (physiological variable) and reported changes in total mood (psychological variable) (i.e., which brain regions increase or decrease their connectivity to sACC in a mood-dependent manner). Results are reported for uncorrected and stringent whole brain or region of interest FEW-corrected for multiple comparisons. Finally we used peripheral cytokine response (IL-6), as an index of peripheral inflammation, in correlational analyses to determine the influence of peripheral cytokines on emotional face processing, its interactions with inflammation-associated mood change, and mood-dependent connectivity of sACC.