14.1.1 Event-Related Potentials

14.1.2 P300

The P300 component is a positive deflection in the EEG signal associated with the process of item discrimination in attention and working memory. P300 is commonly measured by the auditory “oddball paradigm” during which participants are presented with a common tone that is infrequently replaced with a deviant tone stimulus. P300 research has revealed it is, in fact, multidimensional with separate subcomponents reflecting the detection of novel stimuli (P3a) and discrimination of task-relevant versus irrelevant information (P3b). This critical distinction between components of the P300 complex has been brought to bear on elucidating the electrophysiological correlates of the cognitive profile seen in depressive disorders.

Figure 14.1 Schematic examples of event-related potentials (ERPs) commonly described in studies of major depressive disorder (MDD). (a) Example of the grand average ERPs during an auditory oddball experiment. (b) An example of the response difference between correct and incorrect task performance. (c) The N1 and P2 components of the auditory evoked potential are increased in response to increased stimulus amplitude (in decibels). The slope of the peak to peak amplitudes over the stimulus conditions is used as a metric for the excitability of central serotonergic pathways

14.1.3 Diagnosis

In patients with MDD, there is a tendency for a reduction in the amplitude and increase in the latency of the P3a in response to auditory and visual oddball stimuli (2–6). In contrast, the P3b in MDD does not typically demonstrate any difference from healthy controls (7, 8). These findings point toward the disturbance of a perceptual orienting response but emphasize the retention of higher-level cognition (9, 10). However, when studied in patients with BD, the P300 measurements tend to vary depending on the phase of the illness. During the depressive episodes, the P300 profile behaves similarly to MDD patients. As individuals recover toward a euthymic state, there is a gradual improvement in the measurement of the P300 (11, 12), highlighting the state dependency of the P300 profile of BD patients. In a review by Bruder and colleagues (13), the authors describe the consistency of P300 findings across thirty years of research, however, emphasizing that there are a number of negative findings, and that positive findings typically had weak effect sizes. The response to this has been to break down patient groupings into subgroups that separate the individuals along important cognitive and neurophysiological distinctions (7). Cognitive deterioration has been shown to increase P300 latency in auditory oddball paradigms (2), and to decrease the N2b–P3a complex amplitude during choice-reaction time tasks (7). In MDD patients with psychotic features, the P300 amplitude in response to auditory oddball stimuli was lower than controls and MDD patients without psychotic features (14), whereas the opposite is true in patients with comorbid anxiety (7, 15, 16).

14.1.4 Treatment Prediction

Treatment with selective serotonin reuptake inhibitor (SSRI) antidepressants is typically associated with a normalizing of the P300 response toward that of controls (14, 17, 18). In tandem, the response to treatment with serotonergic agents has been shown to be greatest in patients with more impaired P3a and P3b characteristics (both amplitude and latency)(2, 19). This finding has also been mirrored in the study of predictors of treatment response to repetitive transcranial magnetic current stimulation (rTMS) for MDD. The multisite study found that P300 amplitude at baseline had significantly predicted the likelihood of response (20). In each of these studies, the patients who typically experienced the greatest response to treatment had the most deteriorated P300 profile, which suggests a greater severity of depression (21). From these findings it would appear that P300 might represent a candidate for identifying patients with poor frontal regulation of the serotonergic system. However, future research would benefit from more rigorous study design including a focus on the interaction between P300 and other neural correlates of depression to understand their unique contributions and causal relationships to pathophysiologic mechanisms and therapeutic responses.

14.2.1 Diagnosis

Based on the cognitive and physiological profiles of patients with MDD, their ERN should theoretically be distinct from healthy controls. However, the literature presents mixed findings, which likely reflect differences in the experimental design between studies. Varying task conditions will alter the network engaged during task performance and contribute to the efficiency of error monitoring. For example, in depressive disorders, where the processing of affective information is altered relative to controls, the ERN amplitude in response to errors committed during affective go/no-go trials was reduced in MDD patients relative to controls but remained equivalent during cognitive trials (24). Likewise, there is evidence of a blunted ERN in response to flanker tasks with a monetary reward component (25). Both tasks engage dopaminergic pathways that are known to be dysfunctional in patients with MDD and typically scale with the severity of the disease (26–29). Conversely during neutral and punish trials of a flanker task, the ERN amplitude has been shown to increase relative to controls (30), possibly emphasizing a biasing toward a heightened focus on negative information (31).

14.2.2 Treatment Prediction

In a study of geriatric depression patients, one of the few investigations of ERN found that those who achieved remission from depression following citalopram treatment demonstrated a reduction in ERN amplitude, a finding not present in those who failed to achieve remission (32). In contrast, Schrijvers and colleagues (28) found similarities between controls and MDD patients in ERN amplitude at the start and after seven weeks of antidepressant intervention. However, they observed a correlation between the ERN amplitude and absolute change in symptom severity score. A valuable insight to this was provided by Weinberg and colleagues (29) through their investigation of the ERN in patients with melancholic features of depression. They found that even compared with patients with an otherwise similar profile of disease severity those with melancholia exhibited a blunted ERN that also carried over into remission. Although not being improved through pharmacological intervention, patients showed increased ERN amplitude after a course of mindfulness therapy, albeit in the absence of symptom changes (33). The lack of consistent ERN findings in depression may question the use of the ERN in monitoring depression and treatment response. Future investigations could investigate whether this may reflect differential impact due to severity or subtype of depression.

14.3.1 Diagnosis

In patients with depression, the N1/P2 LDAEP has consistently been found to be increased relative to control subjects reflecting the findings of serotonergic suppression on LDAEP in animals (41, 42). However, this concept is contested by some studies that either have found no differences between patients and controls (43, 44) or have observed a dimensional effect of disease subtype on LDAEP properties. For example, Gopal and colleagues (42) noted that the amplitude of the N1/P2 complex was higher with increasing disease severity, and similarly, patients who had reported a history of suicidality had steeper LDAEP slopes than MDD patients without any suicidality (45). Conversely, MDD patients with melancholic symptoms have been shown to demonstrate a shallower LDAEP slope than non-melancholic patients (46). A similar counterintuitive finding was made by Jaworska and colleagues (47), who identified a negative correlation between the N1/P2 LDAEP slope and Montgomery-Asberg Depression Rating Scale (MADRS) score.  Both studies highlight the importance of considering treatment history and the potential effects of interventions on the monoaminergic systems being targeted. For example, the use of high doses of drugs that affect levels of serotonergic neurotransmitters will alter what is considered the baseline state of that individual.

14.3.2 Treatment Prediction

Despite some conflicting findings regarding the LDAEP as a diagnostic marker of MDD, there has been a strikingly consistent pattern of reports suggesting that the LDAEP is highly predictive of treatment response. Patients with a greater baseline LDAEP tend to show a more significant improvement in depression symptom scores following both short- and long-term treatment with a variety of serotonergic agents (48–50). The pretreatment LDAEP slope has also been shown to predict treatment response, with high baseline LDAEP being complicit with a positive response to pharmacological intervention (51).

There have also been a series of recent LDAEP investigations with rTMS as a treatment for MDD. The mechanism of action for rTMS at the left dorsolateral prefrontal cortex (DLPFC) for MDD appears to include downstream desensitization of 5HT1a receptors in the raphe nucleus and the hypothalamus of rats (52, 53). rTMS has also been shown to improve the metabolism of 5-HT in the human limbic system (54). The LDAEP has demonstrated a similar predictive relationship with rTMS, with higher baseline LDAEP slope correlated with greater improvements in HAMD scores post-intervention (55). Unlike pharmacological interventions, patients did not exhibit a posttreatment effect on the LDAEP slope implying that the resulting changes to 5-HT1a activity are more acute following rTMS. This potentially implies that symptom improvement is due in part to regulation of serotonergic activity in tandem with improved emotional regulation arising from the alteration of the DLPFC neuron firing. Through functional connectivity studies, we understand that there is a pattern of communication between the DLPFC and the subgenual ACC (SGACC), which is related to the control of affective cognition (56). Therefore, symptom improvement is likely stemming from a more complicated pattern of system-level changes.

Overall, the LDAEP may be a robust marker for treatment response due to it reflecting gross central serotonergic activity, while inadvertently highlighting the dimensionality of the neurophysiological profile associated with depression through its variations across subtypes of dementia.

14.4.1 Diagnosis

One of the most well-replicated EEG findings in the diagnosis of depressed patients is elevated alpha activity during rest where maximum amplitudes are observed at parieto-occipital locations in the eyes-closed condition (70). Several studies have reported elevated alpha power in depressed patients compared to healthy controls (71–73). A number of studies have localized elevated alpha activity to parietal, occipital, and frontal regions (73–75). Another well-replicated finding derived from alpha power as a biomarker for MDD is alpha asymmetry with decreased alpha power in the frontal right hemisphere regions compared to the left frontal regions, also known as frontal alpha asymmetry (FAA)(73, 76–86). FAA is thought to relate to the finding that MDD is characterized by a hyperactive right prefrontal cortex and hypoactive left prefrontal cortex (82, 87). The essential features of emotional affect can be described using the diathesis model developed by Davidson and Tomarken (82), where two fundamental motivational systems in response to external stimuli exist, namely appetitive (approach) and aversive (withdrawal)(88).  The differential hemispheric activity in EEG can be attributed to the balance in the activations of these two systems (88). Left frontal activation is thought to index appetitive behavior (approach), and right frontal activation is thought to index aversive behavior (withdrawal) (89, 90). Elevated frontal left versus right alpha activity (73, 78, 84–86, 91–93) (Inferred as reduced frontal left versus right activation) thus indexes reduced approach motivation and sensitivity to reward.  Stewart and colleagues (94) found that depressed individuals exhibited a similar pattern of reduced relative left frontal activity during all facial expressions, regardless of valence or approach/withdrawal related. This suggests a trait-like mechanism of emotional responding that is similar to most of the resting EEG asymmetry literature on depression (94, 95). However, many studies, including large sample studies failed to replicate the findings of the FAA in MDD (96–108). The uniformity and generalizability of most studies measuring frontal alpha asymmetry are lacking, owing to differences in technical aspects of EEG data collection and analysis as well as subject profiles. This necessitates meta-analysis and larger-scale studies (88).

14.4.2 Treatment Prediction

FAA and alpha power have shown some promise in differentiating responders and nonresponders to tricyclic antidepressants (TCA) such as clomipramine and imipramine, and with SSRIs(96, 109–113). In the International Study to Predict Optimized Treatment in Depression (iSPOT-D), a multicenter, randomized, prospective open-label trial, 1,008 MDD participants were randomized to eight-week treatment with escitalopram, sertraline, or venlafaxine-extended release (96). The patients were compared to 336 healthy controls (96). From the baseline EEG measures, a gender and drug-class interaction effect was found for FAA (96), with FAA associated with response to the SSRI escitalopram and sertraline, but in females only (96). In a study by Bruder and colleagues (109), MDD patients were treated with fluoxetine for twelve weeks. They found that nonresponders showed greater activation (less alpha) over the right hemisphere, but responders did not. Again, the difference was significant in females but not males. This study was replicated by the same group (75) where they found that occipital alpha asymmetry could be used to differentiate responders from nonresponders to SSRIs where responders showed greater alpha (less activity) over right than left hemisphere and nonresponders showed the opposite. They also found greater alpha power in treatment of responders compared with nonresponders and healthy controls at occipital sites (75), and hypothesized that increased pretreatment alpha activity might be indicative of the relationship between low serotonergic activity and low arousal. This is because serotonergic activity mediates behavioral arousal; thus, low serotonergic activity could reflect the reduced activity of the mesencephalic raphe nuclei and cortical afferents. Also, it is known that depression may be related to dysfunction of temporoparietal mechanisms, which may play a role in mediating emotional arousal (114, 115). Increased alpha power and alpha asymmetry found in SSRI responders could be due to this biological mechanism (114). In rTMS studies, a slow alpha peak frequency has been consistently found to be a predictor for nonresponse (20, 116).

14.5.1 Diagnosis

A number of earlier studies reported increased theta band activity in MDD (71, 129–131), with more recently, localized to frontal regions including ACC (57, 73, 132). Recently, Grin-Yatsenko and colleagues (133) found increased alpha (and theta power) in a large cohort of patients in the early stages of depression. A recent review of BD literature has found increased theta as one of the most robust findings for resting EEG (134). However, there are many mixed findings where some studies have either reported no differences between MDD and controls or decreased ACC activity in MDD (128, 135, 136). Thus, there are no promising findings to conclude elevated theta activity as a diagnostic biomarker and future studies need to follow stringent methodological considerations such as different mood states and medication status of patients.

14.5.2 Treatment Prediction

Changes in theta activity have shown to correspond with treatment with various antidepressants (110, 113) and with electroconvulsive therapy (137). However, there are mixed findings in these studies investigating pretreatment and early changes in the theta band. Reduced pretreatment theta band activity has been observed for TCA, imipramine, and open-label SSRIs at eight weeks with 63% accuracy (113, 114, 138), and high frontal theta activity is associated with nonresponse to antidepressant treatments (139). With a 60% accuracy, Iosifescu and colleagues (138) showed that reduced frontal theta relative power at one-week posttreatment was predictive of treatment response at eight weeks. However, the same finding, but in the opposite direction was found by Spronk and colleagues (140), who reported higher pretreatment theta power as predictive of a higher decrease in depressive symptoms after antidepressant treatment. It may be useful to note that the widespread report on frontal (not midline) theta activity could be due to “drowsiness” theta power. These findings are in line with PET and fMRI studies, where low metabolic activity in the ACC is demonstrated to be associated with worse treatment outcome. Another study reported increased rostral ACC and frontal theta activity to be associated with nonresponse (132), but not with non-remission. Noteworthy is that these results are mostly driven by antidepressant treatment resistance. This suggests that future studies should also investigate the role of treatment resistance for the association of rostral ACC and treatment outcome (132). Increased pretreatment theta current density localized to the rostral ACC has been associated with responses to nortriptyline, citalopram, reboxetine, fluoxetine, or venlafaxine, in depressed patients (141–143).

14.6.1 Diagnosis

Gamma rhythms could be a novel biomarker for major depression with more recent results that provide some objective information on major depressive disease status (148). In a review of gamma disturbances in MDD, Fitzgerald and Watson (148) concluded that depending on the task, gamma rhythms could be either elevated or reduced in depressed patients compared to healthy controls and can also be used to distinguish unipolar depression from bipolar depression. One study found that gamma activity increased in frontal and temporal regions during spatial and arithmetic tasks (153) while other studies found decreased gamma in the frontal region during emotional tasks (154, 155). Another study found reduced gamma in the ACC in MDD patients compared to healthy controls at baseline (156). However, these results need to be replicated with larger-scale controlled studies to identify the underlying pathophysiology that is probed by each task and how the changes in gamma activity can inform the dysfunction. Future studies should also combine profiles of gamma band power across the brain to assess ratios of activity across regions (148). Differentiating MDD and bipolar disorder (BD) can still be a clinical challenge (157). Numerous studies have identified neuroimaging biomarkers that can differentiate between MDD and BD (157), but neurophysiological studies benefit from the ability to measure direct consequences of the electrical activity of neurons at a high temporal resolution (157) and thus can be used to probe gamma activity. One of the most widely investigated tasks with respect to gamma oscillations is the auditory steady-state response (ASSR) task. A recent MEG study found that subjects with unipolar depression had greater gamma power than those with bipolar disorder (157) globally across the whole brain. Another study found that BD patients showed decreased auditory-evoked gamma in a variety of brain regions compared to healthy controls (158). A few studies found that emotional tasks can be used to differentiate unipolar and bipolar depression. Compared to bipolar disorder, patients with unipolar depression showed increased gamma power in temporal regions and decreased frontal gamma power (148, 154, 155, 159).

14.6.2 Treatment Prediction

Serotonergic and noradrenergic drugs have opposing effects on gamma power in different brain regions (148), with serotonergic antidepressants suppressing gamma and noradrenergic antidepressants increasing gamma in animal models. This finding serves as a potentially important guide to pinpoint how the pathophysiology of depression involves altered signaling in different circuits, which could be used to better understand the individual’s biotype of depression. Ketamine is another therapeutic agent that has recently gained much interest concerning understanding its role in depressive circuitry and electrophysiological effect due to its reported rapid and potentially durable effects (148, 160). A significant increase of gamma power as soon as ketamine is administered, preceding most mood effects implies that gamma power could play an important role in the mechanism of action of ketamine (161–163). A recent MEG study found that ketamine had an antidepressant effect in depressed patients and induced a mild depression in healthy controls (164). This study also noted that in patients who had lower baseline gamma, higher drug-induced gamma power across various brain regions was associated with the improved response. This was the opposite for patients with a higher baseline gamma. This study posited an “inverted U” relationship with optimal gamma power being associated with euthymia (148). However, it is still uncertain if gamma activity is causally related to the therapeutic actions of ketamine and monoaminergic antidepressants (148). Noninvasive, non-pharmaceutical treatments like TMS show increased gamma signaling after recovery from depression, and this is especially the case with baseline gamma (165–167). Specifically, a study (165) found that treatment response can be associated with increases in prefrontal gamma power as well as measures of theta gamma coupling.