Research paperConsidering sex differences clarifies the effects of depression on facial emotion processing during fMRI
Introduction
Facial emotion perception is a fundamental element of social functioning that facilitates individuals’ abilities to evaluate and respond to the reactions of others, build and strengthen interpersonal relationships and navigate social networks (Russell et al., 2003). As such, difficulties in facial emotion perception might limit the extent and benefits of social support systems. Individuals with Major Depressive Disorder (MDD) have exhibited processing biases of emotional material, including facial expressions (see Mathews and MacLeod, 2005). These biases are generally away from positive and/or toward negative facial expressions (e.g., Hale, 1998; Joormann and Gotlib, 2007; Leyman et al., 2006; Surguladze et al., 2004). Findings of biases in emotion perception in MDD support interpersonal models of MDD, which highlight the importance of social skills deficits and biases in the etiology and maintenance of MDD. It has been suggested, for example, that the tendency of depressed individuals to be biased toward negative social cues contributes to their feelings of decreased social support and thereby exacerbates depressive symptoms (Gotlib and Hammen, 1992). These findings also converge with Beck's classic (1967) cognitive model of depression, which places biased processing of emotional information at the core of MDD. Notably, Beck holds that such biases are not only state markers of active MDD (aMDD), but also underlying trait vulnerabilities for MDD onset and recurrence.
MDD is highly prevalent (Kessler et al., 2005) and recurrent; an estimated half to two-thirds of individuals who experience a major depressive episode will have at least one more over the course of their lives (Kessler, 2002). Elucidating state and trait risk factors for MDD, such as those associated with impaired emotion perception, could aid in identifying individuals at greater risk for onset and relapse, and potentially lead to earlier interventions. The present study therefore sought to identify performance and neurobiological intermediate phenotypes of MDD whilst minimizing the potential confounding effects of active symptoms by studying individuals with remitted MDD (rMDD) in late adolescence and early adulthood. This approach offers a unique opportunity for determining whether behaviours such as biased facial emotion processing are a trait risk marker for MDD, instead of potentially reflecting state experiences of current depression. Although we cannot completely rule out ‘scar’ effects of chronic illness burden, we can limit their influence by restricting the sample to those who have experienced few depressive episodes. Further, by restricting the age range of participants to young adults in the study we can also minimize developmental, disease progression, and age/degenerative effects.
Most studies on facial emotion perception have been conducted in aMDD. However, there is some evidence that response to positive facial cues may be especially important in distinguishing individuals with rMDD from those who have never been clinically depressed. For example, participants with rMDD have been shown to require greater intensity of emotional expression to identify happy faces compared to never-depressed controls (LeMoult et al., 2009). Healthy controls (HCs) have also demonstrated a bias towards positive facial expressions and away from negative facial expressions in an attention bias (i.e. dot probe) task, whereas currently and formerly depressed participants showed no such bias (Joormann and Gotlib, 2007). Still other studies have shown that individuals with rMDD have similar accuracy as HCs (Jenkins et al., 2016), unlike those with aMDD (e.g., Kohler et al., 2011; Mikhailova et al., 1996).
Studies using functional magnetic resonance imaging (fMRI) have revealed neural differences during facial emotion processing in rMDD. Two general, and sometimes contradictory, patterns of findings have emerged. First, there is evidence of increased neural activity in response to emotional faces among individuals with rMDD compared with HCs. This has included increased activity in the bilateral middle temporal gyrus and left superior frontal gyrus for blocks of emotional faces compared to animals (Jenkins et al., 2016), as well as increased activity in the bilateral dorsolateral prefrontal cortex (DLPFC) and right caudate for blocks of fearful faces (Norbury et al., 2010). Increased activity in the remitted state has also been reported in the amygdala in response to masked sad minus neutral and masked happy minus neutral faces (Victor et al., 2010) and unmasked sad faces (Neumeister et al., 2006) during implicit emotion tasks as well as during explicit classification of emotional faces compared to animals (Jenkins et al., 2016). Notably, other studies have shown that amygdala hyperactivation normalizes with antidepressant treatment (Arnone et al., 2012, Sheline et al., 2001). It remains unclear whether this normalization is a byproduct of antidepressant use or a remission effect.
The second pattern of imaging findings is of decreased activation in response to specific emotional faces among individuals with rMDD versus HCs, typically after subtracting BOLD signal in response to neutral faces. For example, studies have reported reduced activation in the left ventral striatum in response to sad faces (Neumeister et al., 2006), right insula, fusiform gyrus, putamen and bilateral hippocampus in response to sad minus neutral faces and in the left hippocampus and BA44 in response to fearful minus neutral faces (Thomas et al., 2011) and the left orbitofrontal cortex (OFC) and bilateral DLPFC in response to fearful minus neutral faces (Kerestes et al., 2012). A recent meta-analysis of MDD studies highlighted the severe lack of convergence across neuroimaging experiments, even across meta-analytic findings (Müller et al., 2017). The authors noted that this ‘perplexing’ divergence could be due to analytic factors such as use of uncorrected inference procedures, methodological differences, and heterogeneous clinical populations. For example, methodological factors could include whether neutral or happy faces are used for the contrast (Kerestes et al., 2012, Norbury et al., 2010), and the type and duration of treatment participants have experienced contributes to clinical heterogeneity This in turn could impact activation in the DLPFC during implicit negative facial emotion processing (Fales et al., 2009, Fu et al., 2004). Another crucial variable to consider that could potentially be responsible for inconsistencies in results is sex. While many studies attempt to collect both male and female participants, too often Sex is neglected as a variable of interest, and studies of its interaction with history of depression are completely lacking. It is particularly important to consider the interaction of sex and history of depression given that females are twice as likely as males to experience a depressive disorder (e.g., Wade et al., 2002). Sex could therefore be another key factor contributing to and or obscuring the literature on facial emotion perception.
Biases in facial emotion processing could be a potential mechanism leading to sex differences in MDD, given that research consistently shows between-sex behavioural and neural differences in this domain (Stevens and Hamann, 2012; Weisenbach et al., 2014; Whittle et al., 2011; Wright et al., 2009). In general, females are more accurate than males at identifying facial emotions, particularly negative ones (Montagne et al., 2005, Schienle et al., 2005). Neuroimaging studies have also shown that females recruit different brain regions compared to males during facial emotion processing (Weisenbach et al., 2014; Whittle et al., 2011). In a meta-analysis of neuroimaging studies, Stevens and Hamann (2012) found that the left amygdala showed greater activation for negative emotions in females and for positive emotions in males. A recent study by our group with a separate sample reported an interaction between aMDD (vs HC), age (younger vs. older) and sex in several regions, using a block design (Briceño et al., 2015). Notably, these included the DLPFC, inferior and superior frontal gyri, and cingulate, with hyperactivation in these regions for young females with aMDD and hypoactivation in young males with aMDD compared to their same sex HC counterparts.
What is intriguing in light of prior research is the seeming discrepancy between the increased risk for MDD in females, who are typically more accurate at emotion processing than males, while those with MDD typically exhibit worse emotion processing abilities than HCs. To date, research has been inconclusive on why emotion processing strengths and weaknesses may emerge/present differentially in males and females. Primarily, based on a recent finding from a study of aMDD (Briceño et al., 2015), we hypothesized an interaction between sex and diagnosis in a young rMDD sample (a trait hypothesis), as to our knowledge no previous study has sought to examine sex differences in neural activity during facial emotion processing in individuals with rMDD compared to HCs. Specifically, we predicted that males with rMDD would show hypoactivity compared to females with rMDD in emotion processing regions, and that male HCs would show hyperactivity compared to female HCs in these regions.
To our knowledge, no fMRI studies of rMDD have examined explicit facial emotion perception in an event-related design. The present study is not only the first to do so, but is also the first to examine the interaction of Sex with diagnosis in this population.
Section snippets
Participants
The study included 104 individuals aged 18–23 years. The rMDD group comprised 58 (40 female) individuals with a history of MDD (1–3 prior episodes1) who were in remission at the time of the study, as defined by DSM-IV-TR criteria. The mean number of years well was 2.98 (SD = 1.73). The HC
Performance results
For behaviour accuracy, there was no significant main effect of Diagnosis, F(5, 66) = 1.36, p = .251, η2p = .09, no significant main effect of Sex, F(5, 66) = 1.36, p = .250, η2p = .09, and no significant Diagnosis by Sex interaction, F(5, 66) = 1.52, p = .197, η2p = .10. However, there was a significant effect of the covariate Site, F(5, 66) = 2.64, p = .031, η2p = .17, with participants from UM performing better than those from UIC. Due to site differences in Accuracy, we included Site as a
Discussion
The present study illustrates the complexity of studying the neural systems underlying facial emotion perception in mood disorders, as there were significant interactions between Diagnostic group and Sex. This was demonstrated in blood oxygen level-dependent (BOLD) fMRI activity during perception of Sad and of Neutral expressions. As hypothesized, the general pattern was of Females with rMDD demonstrating hyperactivity compared to Males with rMDD, whereas Males HCs showed hyperactivity compared
Funding
This work was supported by funding provided by NIMH grant RO1 091811 (SAL).
Acknowledgements
We would like to thank the individuals that participated in this study. We thank the Multifaceted Explorations of the Neurobiology of Depressive Disorders laboratory (MEND2, Kelly A Ryan, Laura B. Gabriel, Anne L. Weldon, Kortni K. Meyers, Erica Hymen, Bethany Pester, and Kristy A. Skerrett,) for assistance in data collection and diagnostic interviews.
References (57)
- et al.
Age and gender modulate the neural circuitry supporting facial emotion processing in adults with Major Depressive Disorder
Am. J. Geriatr. Psychiatry
(2015) - et al.
The reorienting system of the human brain: from environment to theory of mind
Neuron
(2008) - et al.
Antidepressant treatment normalizes hypoactivity in dorsolateral prefrontal cortex during emotional interference processing in major depression
J. Affect. Disord.
(2009) Judgment of facial expressions and depression persistence
Psychiatry Res.
(1998)- et al.
Improved optimization for the robust and accurate linear registration and motion correction of brain images
Neuroimage
(2002) The functional neuroanatomy of reappraisal: time matters
Neurosci. Biobehav. Rev.
(2009)- et al.
Neural correlates of emotion regulation deficits in remitted depression: the influence of regulation strategy, habitual regulation use, and emotional valence
Neuroimage
(2012) - et al.
Prefrontal cortical response to emotional faces in individuals with major depressive disorder in remission
Psychiatry Res.: Neuroimaging
(2012) - et al.
Insula reactivity and connectivity to anterior cingulate cortex when processing threat in generalized social anxiety disorder
Biol. Psychol.
(2012) - et al.
Facial emotion perception in depression and bipolar disorder: a quantitative review
Psychiatry Res.
(2011)