Frontal Brain Asymmetry as a Biological Substrate of Emotions in Patients With Panic Disorders FREE

FRONTAL BRAIN asymmetry (FBA) as measured by the electroencephalograph is associated with differences in basic dimensions of emotion (see references 1 and 2 for reviews).^3- 8 Increased activation in right anterior regions is linked to negative affect, whereas higher activation in the left frontal hemisphere is associated with positive affect. Frontal brain asymmetry seems to reflect the activation of specialized systems for avoidance-withdrawal behavior in the right frontal hemisphere and for approach behavior in the left frontal region,² and the alpha power difference in electroencephalography between left and right frontal electrodes presumably is the best measure for activation of these systems.^9- 10 However, alpha asymmetry represents a relative index, ie, the same frontal asymmetry in favor of the right hemisphere could be due to either increased right frontal hemisphere activation or decreased left frontal hemisphere activation.

Frontal brain asymmetry represents a trait marker related to psychopathology^{2,4- 5,10- 11} as well as a state marker associated with acute emotional reactions.⁹

Depression consistently has been found to be associated with FBA with decreased left frontal hemisphere activation.^4- 6,12 Regarding anxiety, recent studies revealed FBA with a right frontal hyperactivation during rest in anxious, depressed patients,¹³ while an inverse FBA was found in anxious undergraduate students.¹⁴ Both studies^13- 14 also found relatively greater right than left parietal activation in anxious, depressed patients and anxious undergraduates during phases of anxious arousal. These posterior asymmetries are in line with the proposal of Heller et al^3,14 that anxious arousal including panic should be associated with a relatively increased right parietotemporal activation, while anxious apprehension should be associated with a relatively increased left parietal activation.

Several observations suggest that panic disorder should be characterized by right frontal hyperactivation: (1) The right hemisphere predominantly controls and processes autonomic changes and interoceptive perceptions,¹⁵ both important factors in the development and maintenance of panic disorder.^16- 18 (2) Most patients with panic disorder are characterized by avoidance-withdrawal behavior,¹⁹ which seems to be controlled by right frontal brain regions.²⁰

This study is the first to examine FBA in patients with panic disorder. In contrast to prior studies, patients with anxiety disorder without comorbidity of depression were examined to separate effects of anxiety and depression on FBA. Stimulus-specific FBA effects were examined by the comparison of panic-relevant and panic-irrelevant stimuli conditions. The main hypothesis of this study was that patients with panic disorder exhibit a dispositional FBA with relatively lower right than left frontal alpha power during rest. Also, FBA of patients with panic disorder should vary depending on the circumstances. First, we predicted an FBA in these patients when confronted with panic-relevant stimuli (emergency situations), but not when confronted with emotionally neutral stimuli (mushroom picture). Second, no FBA should be elicited in patients with panic disorder by anxiety-relevant but panic-irrelevant stimuli (spider picture), by emotionally relevant but anxiety-irrelevant stimuli (erotic picture), or by a concentration-consuming but emotionally neutral task (motor task).

Twenty-three patients with panic disorder were compared with 25 healthy control participants. Diagnoses were made by 2 independent, experienced psychiatrists according to theDSM-III-R²¹ on the basis of the Diagnostisches Interview bei psychischen Störungen.²² The Diagnostisches Interview bei psychischen Störungen represents an extended modification of the American Anxiety Disorder Interview Schedule-Revised.²³

All patients met DSM-III-R criteria for panic disorder with or without agoraphobia. Fourteen reported a history of depressive symptoms, but none fulfilled diagnostic criteria for an affective disorder. Patients with and without a history of depressive symptoms did not differ in subjective or electrophysiological measures. Two patients had histories of alcohol abuse and 2 of benzodiazepine abuse. Additional diagnoses were generalized anxiety disorder (2 patients), past generalized anxiety disorder (1 patient), mild social phobia (1 patient), and past social phobia (1 patient). No patient had a specific phobia according toDSM-III-R criteria. No other current or past Axis I or II disorder (including dysthymia) was detected. With the exception of 3 patients who each received a very low dose of medication, all others had not received any psychoactive medication for at least 4 weeks and no neuroleptic medication for at least 6 months. All participants were physically healthy as confirmed by history, physical examination, and screening laboratory tests. The control participants had no history of Axis I disorder and reported no such disorder in first-degree relatives. All participants were right-handed as assessed by the Edinburgh Inventory.²⁴

The erotic, spider, and mushroom pictures were taken from the International Affective Picture System,²⁵ the emergency picture from Pauli et al.²⁶ During the standardized motor performance task, participants had to track a line with a pencil on a sheet of paper along a labyrinth as quickly as possible and without touching the margins.

All participants completed a psychometric test battery (Table 1) including the Agoraphobic Cognition and Body Sensation Questionnaires,^27- 28 the State (STAI-S) and Trait (STAI-T) versions of the Spielberger Anxiety Inventory,^29- 30 the Beck Depression Inventory (BDI^31- 32), and the Symptom Checklist 90-Revised (SCL-90-R^33- 34).

The participants' acute affective states in the experimental situations were measured by the Self-Assessment Manikin (SAM).^35- 36 The SAM assesses arousal, valence, and dominance on 9-point rating scales. Each dimension is depicted by a graphic figure on a continuously varying scale. When representing the valence dimension, the SAM ranges from a smiling, happy figure to a frowning, unhappy figure. For the arousal dimension, the SAM ranges from an excited, wide-eyed figure to a relaxed, sleepy figure. For the dominance dimension, the size of the manikin changes from a relatively small to a relatively big figure. High values represent negative valence, high arousal, or high dominance.

Brain electrical activity was recorded continuously during the whole session with an electroencephalography amplifier (bandwidth 0.016-35 Hz, sampling rate 128 Hz) from 19 sites using an electrode cap according to the International 10-20 System. All electrodes were recorded with reference to the vertex. Electrical impedances were kept below 5 kΩ.

Vertical eye movements were measured at the right eye using the frontal pole electrode (Fp1) and an electrode below the right eye. Horizontal electro-oculography was recorded between the outer canthi of both eyes. A marker channel was used to identify experimental phases.

Electroencephalographs were off-line filtered with a bandwidth from 0.3 to 35 Hz. First, electroencephalography records were visually examined for electro-oculographic and electromyographic artifact, and epochs with obvious artifacts (eg, deflections greater than 50 µV in any channel) were excluded. In a second step, ocular artifacts were corrected with a regression method based on the vertical and horizontal electro-oculogram.³⁷

Participants had to avoid caffeine and alcohol starting the evening before testing. After informed consent was obtained, participants were seated in a relaxation chair. The experiment comprised an adaptation phase of 2 minutes, 4 resting phases, and 5 stimulus phases. Participants were asked to relax as much as possible during the resting phases of 1-minute duration. Eyes were alternately kept open and closed, with half of the participants starting with the eyes open. During the stimulus phases, 1 of the 4 pictures or the motor task were presented to the participants for 1 minute each. The sequence of these phases was counterbalanced between participants. The SAM questions were assessed immediately after each phase. Phases were separated by intervals of 2 minutes.

A fast Fourier transformation with overlapping Parzen windows (50% overlap) of 2 seconds' duration was applied to the 1-minute segments of relevant phases for the channels of the electroencephalograph (ie, superior frontal electrodes F3 and F4, and parietal electrodes P3 and P4). Resulting power spectra were averaged.

Only the alpha frequency band was further analyzed, because this frequency band has been most consistently linked to electroencephalograph asymmetry for emotion and cognition.^11,38- 39 These data were log transformed to normalize their distribution. To get a single measure for alpha power during rest, alpha power was averaged across the 4 one-minute resting phases.

Following Davidson³⁸ and our main a priori hypotheses, analyses were first conducted for frontal electrodes. The respective analyses of variance (ANOVAs) contained the factors Group (patients with panic disorders vs control participants) and Hemisphere (left vs right). To test the main hypothesis about stimulus specificity, the ANOVA included an additional factor, Picture Category (panic-relevant vs emotionally neutral).

Regional specificity of FBA was analyzed with 2 approaches. First, as in previous studies,^{5,9,12,40- 41} separate ANOVAs for parietal electrodes were performed to analyze parietal brain asymmetries in an explorative manner. This rather conservative approach was taken because we hypothesized no group differences; separate ANOVAs maximized our chances of finding evidence contradicting our hypotheses. Second, for cases where a statistically significant FBA was found, an additional ANOVA including frontal and parietal electrodes was conducted. A significant Group × Hemisphere × Region interaction would indicate specificity of asymmetry to frontal regions. Alpha level of significance was set atP<.05.

Significant effects were further evaluated by post hoc mean comparisons. Group differences regarding subjective measures were tested with t tests.

Patients with panic disorders reported a significantly more negative valence than control participants (t₄₆=3.0,P=.005) (Table 2). No other significant differences were found.

For frontal electrodes (F3 and F4), there was a significant Group × Hemisphere interaction (F_1,46=5.8,P=.02). Patients with panic disorders showed significantly less right than left frontal alpha power (t₂₂=2.3, P=.03), while control participants did not show any FBA. Neither Group nor Hemisphere main effects reached statistical significance.

For parietal electrodes (P3 and P4), a significant Group × Hemisphere interaction (F_1,46=5.7,P=.02) was found. Post hoc tests revealed no posterior brain asymmetry for patients with panic disorders. However, for control participants, the right parietal alpha power was enhanced compared with the left (t₂₄=2.7, P=.01). No significant main effects were obtained.

An overall ANOVA including frontal and parietal electrodes revealed no Group × Hemisphere × Region interaction (F_1,46=0.81,P=.37).

Similar to Schaffer et al⁶ and Jacobs and Snyder,¹² Pearson correlations were computed between the frontal or parietal right minus left alpha power difference [(F4−F3), or (P4−P3)] and the BDI, the SCL-90-R depression and anxiety score, the STAI-S and STAI-T. For all participants, significant negative correlations were found between the FBA and BDI (r=−0.39,df=48, P=.007), SCL-90-R depression (r=−0.34, df=48, P=.02), SCL-90-R anxiety (r=−0.52,df=48, P=.001), STAI-S (r=−0.41, df=−47, P=.004), and STAI-T (r=−0.27,df=48, P=.07, trend only), whereby negative correlations indicated that high questionnaire scores were associated with a relatively reduced right frontal alpha power. For parietal electrodes, no significant correlations were found.

However, correlational analysis within control participants only revealed a trend for a negative correlation between FBA and BDI (r=−0.39, df=25,P=.06), but not for the SCL-90-R or the STAI-S or STAI-T. By controlling for the BDI, the partial correlation coefficient of all anxiety measures remained nonsignificant.

Correlational analysis within the patients with panic disorders indicated for FBA a significant negative relationship with SCL-90-R anxiety (r=−0.50,df=23, P=.02) and a trend for a negative relationship with STAI-S (r=−0.40, df=23, P=.06), but no significant correlation with the BDI (r=−0.17,df=23, P=.43). No correlations were found for parietal asymmetry (SCL-90-R anxiety, r=−0.01, STAI-Sr=0.01).

By controlling for the BDI, the partial correlation coefficient between FBA and SCL-90-R (r=−0.50,df=20, P=.01) or STAI-S (r=−0.36, df=20, P=.05) remained significant. Comparable results were found by using the SCL-90-R depression score as a covariate.

The ANOVA for valence ratings revealed significant main effects for the factors Group (F_1,46=13.9,P<.001) and Picture Category (F_1,46=64.1, P<.001). Furthermore, there was a significant Group × Picture Category interaction (F_1,46=3.9, P<.05). Patients with panic disorders and control participants rated the valence of the mushroom picture similarly (t₄₆=1.3, P =.19), while the emergency picture was rated significantly more negative by patients than by controls (t₄₆=4.2, P<.001). Similar ANOVAs for arousal and dominance ratings revealed only significant main Picture Category effects (both F_1,46>30.0, P<.001). The emergency picture elicited more arousal and less dominance than the mushroom picture in both groups (Table 3).

For frontal electrodes, the ANOVA revealed a significant Picture Category main effect (F_1,46=7.6,P=.008), indicating that the emergency picture elicited a generally lowered alpha power compared with the mushroom picture. Likewise, there were significant Hemisphere × Group (F_1,46=4.6, P=.04) and Hemisphere × Group × Picture Category (F_1,46=4.8, P=.03) interactions.

Post hoc analyses in the control group revealed neither significant main effects of Picture Category or Hemisphere nor a significant Hemisphere × Picture Category interaction. Hence, there is no indication of any asymmetry in frontal alpha power in control participants during viewing of mushroom or emergency pictures.

In patients with panic disorders, however, the ANOVA revealed a significant main effect of Picture Category (F_1,22=4.4, P=.05) and a significant Hemisphere × Picture Category interaction (F_1,22=9.7,P=.005). While looking at the emergency picture, patients with panic disorders showed a significantly reduced right frontal alpha power compared with left frontal alpha power (t₂₂=2.7, P=.01). For the mushroom picture, this FBA did not emerge. These findings indicate that a panic-related stimulus induced a reduced right frontal alpha power in patients with panic disorder only.

For parietal electrodes, the ANOVA revealed a significant Picture Category main effect (F_1,46=5.8,P=.02) and a significant Picture Category × Group interaction (F_1,46=4.2, P=.05). A post hoc analysis for patients with panic disorders showed a significant main effect of Picture Category (F_1,46=8.2,P=.009), indicating that their parietal alpha power was generally lower while looking at the emergency than the mushroom picture. A corresponding post hoc analysis for control participants revealed no significant effects. No parietal asymmetry effects were found for either patients or controls.

An overall ANOVA including frontal and parietal electrodes revealed no significant Group × Hemisphere × Region interaction (F_1,46=0.52,P=.48).

Patients with panic disorders reported a significantly more negative valence than control participants during inspection of the spider picture (t₄₆=3.3,P=.002) (Table 4). No other group differences reached statistical significance.

Frontal power during the execution of the motor task was virtually identical for left and right electrodes and for patients with panic disorders and control participants (Table 4).

The ANOVAs on frontal alpha power for the spider and erotic picture revealed no significant main effects of Group or Hemisphere, but significant Hemisphere × Group interactions: spider picture (F_1,46=6.4,P=.02); erotic picture (F_1,46=4.6, P=.04). Post hoc analyses revealed that patients with panic disorders showed significantly lower alpha power at the right compared with the left frontal electrode while looking at the spider (t₂₂=2.1,P=.05) or erotic picture (t₂₂=2.3, P=.03). The same analysis for control participants revealed no FBA.

For parietal electrodes, the ANOVAs for the motor task, the spider picture or the erotic picture revealed no significant main or interaction effects. Overall, ANOVAs, including frontal and parietal electrodes, revealed no Group × Hemisphere × Region interaction for the spider (F_1,46=1.88,P=.18) or the erotic (F_1,46=1.33, P=.25) picture. As sex is suspected to be correlated with hemispheric differences, especially concerning the response to erotic pictures, we additionally ran the analyses for the erotic pictures for women only. The reported difference was still present.

These data are the first to provide evidence that patients with panic disorders are characterized by an FBA with a relatively reduced right alpha power during rest and when confronted with anxiety-relevant stimuli. No FBA was found in the control participants. According to Davidson et al,⁹ these findings indicate an enhanced activation of the right frontal avoidance-withdrawal system in patients with panic disorders.

The FBA found in patients with panic disorders was not caused by symptoms of depression. First, patients with panic disorders were selected to have no depression comorbidity. Second, as in previous studies, FBA was correlated with a psychometric measure of depression (ie, BDI)^6,12 in the healthy controls, while in patients with panic disorders FBA was not correlated with the BDI, but with anxiety measures. Third, the data of Davidson et al⁹ showed that FBA in depression is more supportive of decreased left frontal hemisphere activation, while the FBA of patients with panic disorders seems to be caused by increased right frontal hemisphere activation.

INCREASED RIGHT frontal hemisphere activation may reflect an acute emotional reaction,⁹ indicating the prevalence of negative emotions and an acute activation of the avoidance-withdrawal system. The correlation between the FBA of patients with panic disorders during rest and STAI-S, but not the STAI-T indicates that FBA is a state rather than a trait marker. The affective judgments during the rest phase also indicated that the patients with panic disorders experienced this situation as more aversive than the healthy controls. Ruminative thoughts related to the illness during the rest phase might be among the reasons for the more negative valence ratings. Previous studies showed that patients with panic disorders experience baseline or rest phases as more aversive^42- 44 and periods of relative relaxation as anxiety-inducing^45- 46

The interpretation of FBA in patients with panic disorders as a state marker is also in line with our additional findings that patients with panic disorders show FBA while confronted with anxiety-relevant stimuli (emergency situation, spider). In both situations, more negative valence ratings were observed in patients with panic disorders than in healthy controls, and patients with panic disorders exhibited FBA. On the other hand, both groups showed comparable valence ratings and no FBA for the mushroom picture or for the motor task. Therefore, the FBA of patients with panic disorders may reflect negative emotional reactions elicited by these situations.

However, the finding that an anxiety-irrelevant erotic picture also induced an FBA in patients with panic disorders seems to be inconsistent with this interpretation. Erotic stimuli are arousing, and patients with panic disorders may exhibit FBA in all arousing situations, irrespective of their positive or negative valence. Arousal may induce bodily reactions (ie, heart rate acceleration) that are known to be important triggers of panic attacks.^16,42,47 This explanation is supported by the fact that patients with panic disorders showed a tendency to more negative valence ratings for the erotic picture than control participants.

However, a more detailed analysis of the data suggests another interpretation. While looking at the erotic picture, the FBA of patients with panic disorders did not derive from a reduction of right frontal alpha power, but instead from an increase in left frontal alpha power. This distinguished the reactions of patients with panic disorders to anxiety-irrelevant erotic pictures from those to anxiety-relevant emergency or spider pictures. To what extent frontal asymmetry from either a decrease of right frontal alpha power or an increase of left frontal alpha power reflects functional differences cannot be solved with our data. It might be assumed, however, that an increase in left frontal alpha power is associated with the inhibition of the approach system and a decrease of right frontal alpha power with the selective activation of the avoidance-withdrawal system.

Another way of interpreting our findings is to assume that patients with panic disorders are characterized by generally negative emotions and a tonically overactivated withdrawal-avoidance system. This disposition causes FBA in most situations, except when patients with panic disorders are occupied with explicitly neutral stimuli. Neutrally valenced and unarousing stimuli may ameliorate the tonic right frontal overactivation. Consequently, no FBA is observable. It is well known by clinicians that many patients with panic disorders try (and often succeed) to cope with their symptoms with distractions. Findings in depression showing that negatively valenced thoughts are significantly reduced if patients engage in some form of distraction also support this notion. Finally, the frequent observation that patients with panic disorders feel safe at home might be explained. The home presumably is an explicitly neutral stimulus that reduces right frontal overactivation.

Although recorded frontal alpha levels consistently have been shown to be associated with phenomena interpreted as right or left frontal hemisphere activation, the possible neurobiological mechanism by which such measures are associated with frontal activation are unclear. To date, no data are available that suggest frontal cortex alpha generators. But recent neuroimaging studies show a reverse relationship between the activation of prefrontal cortex and amygdala.^48- 49 These data suggest that the frontal cortex regulates and restrains subcortical structures related to affect (see also references 50-53). Weakening of these influences can facilitate activity of the amygdala, which, in turn, can release uncontrolled affect such as anxiety and panic. Nevertheless, further research is warranted to elucidate what brain processes underlie changes in recorded frontal alpha levels.

While our data clearly support the hypothesis that patients with panic disorders are characterized by FBA, our findings on parietal brain asymmetries (PBA) are equivocal. All specific tests for PBA in patients with panic disorders failed to reach significance, and significant correlations between asymmetry measures and psychopathology were restricted to frontal areas exclusively. However, the overall tests, which included frontal and parietal electrodes, failed to confirm that the FBA in patients with panic disorders was restricted to frontal brain areas. Although we cannot draw clear-cut conclusions about PBA, these findings can be viewed as consistent with an overall favoring of right over left frontal hemisphere activation in patients with panic disorders. Also, the finding during the resting phase that the control participants showed a greater parietal asymmetry favoring the right over the left side in alpha power could be viewed as consistent with this overall favoring of right over left. Other recent studies^13- 14 also found a relatively enhanced right parietal brain activation in anxious participants during rest. Together, these findings support the model of Heller et al³ that anxious arousal is associated with a relatively enhanced right parietal activation.

In summary, patients with panic disorders are characterized by an asymmetry in frontal activation, which can be taken as evidence for an overactive avoidance-withdrawal system. The asymmetry was independent of symptoms of depression but was influenced by situational information.

Accepted for publication October 15, 1998.

Research was supported by the German Research Society, Bonn (Deutsche Forschungsgemeinschaft, DFG).

We acknowledge the help of Petra Hinderer for running the experimental sessions.

Corresponding author: Georg Wiedemann, MD, University Hospital of Psychiatry and Psychotherapy, University of Tübingen, Osianderstrasse 24, D-72076 Tübingen, Germany (e-mail: georg.wiedemann@med.uni-tuebingen.de).