Participants and assessments

We recruited 18 young people (9 women, 9 men), mean age 19.8 years (range 19–21) who had never personally had depression but who had a biological parent with a history of major depression. Potential participants were assessed with the Structured Clinical Interview for DSM–IV Axis I Disorders Schedule (SCID–I) ^{Reference Spitzer, Williams, Gibbon and First5} to exclude a personal current or previous major depression or other Axis I disorder. The presence of major depression in a parent was assessed by the family history method using the participant as an informant. ^{Reference Andreason, Rice, Endicott, Reich and Coryell6} The criteria used included description of the symptoms of major depression together with the prescription of specific antidepressant treatment, either psychotherapy or medication. A history of bipolar disorder or schizophrenia in a parent was an exclusion criterion. We also recruited 16 controls (10 women, 6 men), mean age 19.9 years (range 18–21) who were determined by the same instruments to have no current or previous major depression and no history of depression in a biological parent or other first-degree relative. All participants were right-handed, according to the Edinburgh Handedness Inventory, ^{Reference Oldfield7} and had normal or corrected to normal vision.

Participants were assessed on a number of measures of current emotional state, including the Mood and Feeling Questionnaire (MFQ), ^{Reference Wood, Kroll, Moore and Harrington8} the Hospital Anxiety and Depression Scale (HADS) ^{Reference Zigmond and Snaith9} and the Perceived Stress Scale (PSS). ^{Reference Cohen, Kamarck and Mermelstein10} We also administered the Parental Bonding Instrument (PBI) ^{Reference Parker11} and the Life Events Rating Scale (LERS), ^{Reference Goodyer, Herbert, Tamplin, Secher and Pearson12} which assesses threat and loss events in the past year and over the lifespan. All participants gave full informed consent to the study, which was approved by the local ethics committee.

Image acquisition

Imaging data were collected using a 1.5 T Siemens Sonata scanner located at the Oxford Centre for Clinical Magnetic Resonance Research. Functional imaging consisted of 35 T₂*-weighted echo-planar image slices (repetition time (TR) 3000 ms, echo time (TE) 50 ms, flip angle 90°, matrix 64×64, 3 mm isotropic voxels). To facilitate later co-registration of the fMRI data into standard space, we also acquired a Turbo FLASH sequence (TR 12 ms, TE 5.65 ms, voxel size 1 mm³). The first two echo-planar image volumes in each session were discarded to avoid T₁ equilibration effects.

Emotional counting Stroop task

Participants were scanned while performing a modified version of the emotional counting Stroop called the ‘name the number of words’ task. ^{Reference Whalen, Bush, McNally, Wilhelm, McInerney, Jenike and Rausch4} Word stimuli were a subset drawn from a larger pool used in previous research ^{Reference Mathews, Mogg, May and Eysenck13} examining depression and anxiety, and selected to be either neutral (e.g. mileage, molecule), physically threatening (e.g. fatal, accident), socially threatening (e.g. worthless, inferior) or positive (e.g. generous, achievement). Physically threatening and socially threatening words were combined to generate a negative word category. Words were matched for word length, frequency and imageability. (For further information see online Table DS1 and www.psy.uwa.edu.au/mrcdatabase/uwa_mrc.htm.)

Participants completed one run of the task with a total of 160 words being presented across 16 blocks. Four 20-word blocks of each stimulus type were presented in a pseudo-randomised order and interspersed with 20-s blocks of fixation, free of stimulus (no motor response) as baseline. Presentation of the four conditions was counter-balanced across participants and between the two groups. Participants completed 10 trials during each presentation block (stimulus presentation 1500 ms, intertrial interval 500 ms). For each trial, participants viewed between one and four identical words and were instructed to report (via keypad response) the number of words presented in each trial. Stimuli were presented on a personal computer using E-Prime (version 1.0; Psychology Software Tools Inc., Pittsburgh, Philadelphia, USA) and projected onto an opaque screen at the foot of the scanner bore, which participants viewed using angled mirrors. Both accuracy and reaction times were recorded by E-Prime.

Functional MRI data analysis

Functional MRI data were preprocessed and analysed using the functional magnetic imaging of the Brain Software Library (FSL version 3.3; www.fmrib.ox.ac.uk/fsl), implemented in Linux SUSE, version 9.1. ^{Reference Smith, Jenkinson, Woolrich, Beckmann, Behrens, Johansen-Berg, Bannister, DE Luca, Drobnjak, Flitney, Niazy, Saunders, Vickers, Zhang, De Stefano, Brady and Matthews14} Preprocessing included within-participant image realignment, ^{Reference Jenkinson, Bannister, Brady and Smith15} non-brain removal, ^{Reference Smith16} spatial normalisation to a standard template (Montreal Neurological Institute (MNI) 152 stereotactic template) ^{Reference Jenkinson and Smith17} using an affine procedure and spatial smoothing using a Gaussian kernel (5 mm full-width, half-maximum). The time series in each session was high pass filtered (to a maximum of 0.007 Hz).

Analyses of data from individual participants were computed using the general linear model with local autocorrelation correction. ^{Reference Woolrich, Ripley, Brady and Smith18} Three explanatory variables were modelled: ‘neutral’, ‘positive’ and ‘negative’ words. In addition, temporal derivatives were included in the model as covariates of no interest to increase statistical sensitivity. All variables were modelled by convolving each block with a haemodynamic response function, using a variant of a gamma function (i.e. a normalisation of the probability density function of the gamma function) with a standard deviation of 3 s and a mean lag of 6 s.

Individual participant data were combined at the group level using full mixed effects analyses. ^{Reference Woolrich, Behrens, Beckmann, Jenkinson and Smith19} Significant activations were identified using cluster-based thresholding of statistical images with a height threshold of Z=2.0 and a (corrected) spatial extent threshold of P<0.05. ^{Reference Friston, Worsley, Frackowiak, Mazziotti and Evans20} Approximate Brodmann areas (BA) were identified by transformation of MNI coordinates into Talairach space (additional information available at www.mrc-cbu.cam.ac.uk/Imaging).

Participant characteristics and emotional counting Stroop performance

There were no significant differences between FH+ participants and controls in age, gender, current mood state, level of perceived stress and experience of life events. Controls rated their mothers as being more overprotective (Table 1). Owing to technical difficulties, accuracy and reaction time data for 8 participants were not available and the analyses of behavioural data were therefore carried out on 26 individuals (14 FH+ and 12 controls). There were no between-group differences in accuracy and reaction time (all P>0.40).

Functional MRI data

Because of our interest in activation differences between FH+ participants and controls, we report significant between-group comparisons (thresholded at Z=2.0, P=0.05, whole brain corrected) rather than effects of task performance within the two groups. For the orthogonal contrast negative words v. neutral words, we observed significantly greater activation in controls in the anterior cingulate cortex, medial frontal and right superior frontal gyrus (BA 24/10 and 9 respectively), left middle frontal gyrus (BA 8/6) and left caudate nucleus and bilateral activation in the inferior parietal lobe (BA 7) (Table 2). Comparing positive words with neutral words, controls had significantly greater blood-oxygen-level dependent (BOLD) response in the right anterior cingulate (BA 24), left thalamus, left superior frontal gyrus (BA 10) and left precuneus (BA 19) (Table 3).

Given our a priori hypothesis regarding the role of the affective subdivision of the anterior cingulate cortex in the emotional counting Stroop task, we focused subsequent analyses on this brain region. To examine group×emotion interactions we first extracted percentage signal change from the significant clusters of this subdivision identified above in the whole brain analysis (negative v. neutral words and positive v. neutral words). Further statistical analysis was implemented using a repeated measures analysis of variance (ANOVA) model with ‘group’ (FH+ v. controls) as the between-participant factors (FH+ v. control) and ‘word type’ (positive/negative/neutral words) as the within-participant factor for all participants. Significant interactions were followed up using simple main effects (independent and repeated-samples t-tests) to elicit the degree of this differential activation.

Limitations

An important limitation of the study is that we did not systematically conduct personal psychiatric interviews with relatives in either FH+ or control groups and it is therefore possible that some of the parents in the FH+ group did not have depression or that some parents in the control group did. Presumably, however, misclassifications of this kind would tend to decrease rather than increase biological differences between the two groups. In addition, we have previously shown in a larger study that FH+ participants identified in this way have increased waking salivary cortisol secretion relative to controls. ^{Reference Mannie, Harmer and Cowen21} It has been estimated that by young adulthood up to 40% of children of parents with a clinical mood disorder will have suffered a personal episode of depression; ^{Reference Beardslee, Versage and Gladstone22} however, the FH+ participants in the current study did not differ from controls in terms of current affective symptomatology and levels of perceived stress. In addition, albeit on limited data, it does not appear that their experience of parental depression is reflected in problems with parental attachment or in increased life events either recently or over the life span.

Anterior cingulate cortex activation in controls and FH+ participants

The anterior cingulate cortex has cognitive and affective divisions that are separable both anatomically and functionally. ^{Reference Bush, Luu and Osner2} Previous studies in healthy individuals have shown that the affective division of the anterior cingulate cortex is activated by a number of emotional manipulations ^{Reference Bush, Luu and Osner2,Reference Elliott, Rubinsztein, Sahakian and Dolan23} and our data in healthy participants confirm the findings of Whalen et al, ^{Reference Whalen, Bush, McNally, Wilhelm, McInerney, Jenike and Rausch4} who used an emotional counting Stroop to demonstrate increased activation in the pregenual region of the affective subdivision of the anterior cingulate cortex in response to emotional relative to neutral words. Also, in agreement with Whalen et al, we found that the emotional counting Stroop task was associated with an overall deactivation in this subdivision compared with fixation but that the deactivation was relatively less during presentation of emotional words. It has been suggested that the overall deactivation of the affective subdivision of the anterior cingulate cortex in response to the emotional counting Stroop reflects reciprocal inhibition from the cognitive subdivision with the purpose of maintaining cognitive performance where there is increased competition for attentional resources. ^{Reference Bush, Luu and Osner2,Reference Whalen, Bush, McNally, Wilhelm, McInerney, Jenike and Rausch4} Despite this deactivation, the relative increase in activity of the affective subdivision following presentation of emotional v. neutral words demonstrates the continuing ability of the anterior cingulate cortex to monitor emotional information during the emotional counting Stroop task.

In contrast to these findings in healthy participants, the affective subdivision of the anterior cingulate cortex in FH+ participants showed no difference in activation pattern to emotional v. neutral words. This suggests that in people at increased familial risk of depression the affective subdivision responds less efficiently to the changing emotional valence of incoming stimuli. This difference in activation pattern appeared to be driven partly by the lessened deactivation to neutral words shown by the FH+ participants. This might imply that in people at increased familial risk of depression the affective subdivision of the anterior cingulate cortex reacts to neutral stimuli as if they had an emotional valence. Whether or not this is the case, our findings suggest that in FH+ people the affective subdivision of the anterior cingulate cortex is less efficient in detecting differences in the emotional quality of sensory input.

Changes in anterior cingulate cortex activity in acute depression

Changes in activity of the anterior cingulate cortex have been reported frequently in imaging studies of patients who are acutely depressed, particularly hypoactivity in the cognitive subdivision which may correlate with impaired performance on cognitive tasks. ^{Reference Elliott, Baker, Rogers, O'Leary, Paykel, Frith, Dolan and Sahakian24,Reference Davidson, Pizzagalli, Nitscke and Putnam25} Findings in relation to the affective subdivision are more complex. Wagner et al Reference Wagner, Sinsel, Sobanski, Köhler, Marinou, Mentzel, Sauer and Schlösser ²⁶ used a counting (non-emotional) Stroop, in conjunction with fMRI, to study anterior cingulate cortex activity in patients with depression who were not receiving medication. They found no difference in either task performance or activation in the cognitive subdivision relative to controls; however, patients demonstrated less deactivation in the affective subdivision, a finding rather similar to our own. Other imaging studies in patients with acute depression have measured activation patterns in the anterior cingulate cortex in response to tasks involving the processing of emotional information. Findings have been variable, with some investigations demonstrating increased neural responses to negative emotional stimuli, consistent with the emotional biases associated with acute depression. ^{Reference Fu, Williams, Cleare, Brammer, Walsh, Kim, Andrew, Pich, Williams, Reed, Mitterschiffthaler, Suckling and Bullmore27} However, Elliott et al, ^{Reference Elliott, Rubinsztein, Sahakian and Dolan28} using an affective go/no-go task, noted findings similar to our own, namely blunted neural activation to both positive and negative emotional stimuli in the affective subdivision of the anterior cingulate cortex.

Implications

Our data indicate that abnormalities in the neural response to emotional stimuli in the anterior cingulate cortex can exist independently of the presence of acute depression and appear to be present in those at increased familial risk of illness. The pregenual region of the affective subdivision of the anterior cingulate cortex, implicated in our study, has connections to other brain regions known to be involved in emotional experience and expression, including the orbitofrontal cortex, amygdala, hippocampus and ventral striatum. ^{Reference Davidson, Pizzagalli, Nitscke and Putnam25,Reference Drevets29} In this respect the pregenual affective subdivision of the anterior cingulate cortex is well placed to play a key role in the integration of emotional and cognitive information. ^{Reference Bush, Luu and Osner2} Hence, abnormalities in this area could be associated with impaired ability to use emotional information to influence decision-making, as we observed in an emotional categorisation task in FH+ participants. ^{Reference Mannie, Bristow, Harmer and Cowen1} It is possible that deficits of this sort could result in difficulties in making complex social decisions; this may be one mechanism through which increased familial risk of depression could be expressed. ^{Reference Goodyer30}

It is important to note that we also observed differential neural activations to the emotional counting Stroop between FH+ and controls in brain regions other than the anterior cingulate cortex, including the thalamus and areas of prefrontal cortex, some of which are known to be associated with emotional processing. Although many of these brain regions have connections to the anterior cingulate cortex, these more widespread changes support the idea of a distributed circuitry underpinning both emotional processing and the risk of clinical mood disorders. ^{Reference Phillips, Drevets, Rausch and Lane31,Reference Phillips, Drevets, Rausch and Lane32} Hence, vulnerability to depression is likely to be associated with changes across a network of areas rather than dysfunction solely in the anterior cingulate cortex. For example, reduced connectivity between thalamus and anterior cingulate cortex has been demonstrated in patients with depression compared with healthy controls. ^{Reference Anand, Li, Wang, Wu, Gao, Bukhari, Mathews, Kalnin and Lowe33} Future studies investigating temporal correlations between BOLD response in the anterior cingulate cortex and the prefrontal/limbic regions reported above are warranted to examine whether altered functional integration and/or aberrant connectivity pre-date the onset of depression in at-risk individuals.