The authors of this paper do not have any commercial associations that might pose a conflict of interest in connection with this manuscript.
Dr. Chantal Henry, Pôle de Psychiatrie, Hôpital Albert Chenevier, 40 rue Mesly, 94000 Créteil, FRANCE. Fax: + 33-1-49-81-30-99; e-mail: firstname.lastname@example.org
Background: Normothymic states in bipolar disorders are generally considered to be devoid of severe symptoms. However, bipolar patients present subsyndromal symptoms for half of their lives, and often have comorbid psychiatric disorders. If we go beyond the concept of temperamental features, can we identify certain emotional characteristics distinguishing normothymic bipolar patients from normal controls? We previously showed, using self-completed questionnaires, that normothymic bipolar patients display higher levels of emotional lability and intensity than controls.
Objectives: The aim of this study was to assess the emotional reactivity of normothymic bipolar patients, comparing such patients with a normal control group during an experimental mood induction procedure.
Method: We evaluated the subjective emotional reactivity of 145 subjects (90 control subjects and 55 normothymic bipolar patients), using an emotional induction method based on the viewing of a set of 18 pictures (6 positive, 6 negative, 6 neutral) extracted from the International Affective Picture System. Subjective valence and arousal were recorded with the Self-Assessment Manikin. We also recorded startle reflexes, triggered by a tone occurring during the viewing of two-thirds of the pictures. We controlled for confounding factors, such as concurrent treatments, in all analyses.
Results: Normothymic bipolar patients and normal controls assessed valence and arousal similarly for positive and negative images. However, neutral images were considered more pleasant [F(1,143) = 8.4; p = 0.004] and induced a higher level of arousal [F(1,143) = 12.3; p = 0.001] in normothymic bipolar patients than in control subjects. Neutral pictures also triggered a stronger startle reflex in normothymic bipolar patients compared to controls [F(3,123) = 3.1; p = 0.03].
Conclusion: Normothymic bipolar patients displayed emotional hyper-reactivity, mostly evidenced in neutral situations. This feature may be linked to emotional dysregulation and is a potential endophenotype and/or a risk factor for bipolar disorders. This trait may be responsible for vulnerability to minor stressful events in everyday life. These findings have potential implications for the daily management of bipolar disorder between crises.
Current classifications of mental disorders do not take into account the normothymic phase in bipolar disorders. However, it has now been established that many patients display comorbid conditions and/or residual symptoms during such phases (1, 2). Personality traits or temperaments specific to patients with bipolar disorder have also been identified (3–8), but few studies have focused on dimension evaluation (9). For example, the studies on impulsivity in bipolar patients have demonstrated the value of this type of approach for improving our understanding of the modifications occurring during episodes and for a better characterization of the normothymic period (10, 11). Dimension-based studies have the advantage of making it easier to establish links with neurobiological modifications.
Surprisingly, few studies have addressed the emotional reactivity of bipolar patients. The thymic episodes characterizing bipolar disorders are mostly defined on the basis of emotional tone (sadness/euphoria). However, these emotions are characterized not only by their tone, but also by the intensity with which they are felt. Most bipolar patients spontaneously report having excessively strong emotional reactions, even between episodes. They often say that they are more sensitive than other people. Actually, several lines of evidence suggest that emotional reactivity is disturbed in bipolar patients during normothymic periods. First, emotional reactivity is related to stress management in bipolar patients during the normothymic phase. Several studies have shown that stressful life events have deleterious effects, increasing the frequency of relapse in patients with bipolar disorders (12, 13). Second, bipolar patients report considerable daily variation in affect, even when normothymic (14). Personality traits such as neuroticism are associated with a higher level of emotional instability. Similarly, Benazzi (15) showed that having ‘ups and downs’ during normothymic periods is predictive of bipolarity in patients presenting recurrent depression. Finally, we have previously shown that during normothymic periods, bipolar patients experience more intense emotions than patients with personality disorders or control subjects, resulting in a higher level of emotional variability (16, 17); these results were obtained in studies involving self-completed questionnaires, requiring the patients to have good insight into subtle traits.
The emotional reactivity of subjects can be studied more objectively with the emotional induction method. A classical technique involves showing the subject positive, neutral, or negative images and evaluating subjective and objective factors linked to the emotions triggered by the images.
The aim of this study was to compare emotional reactivity in normothymic bipolar patients and control subjects using an emotional induction method based on viewing images with positive, neutral, or negative connotations. This approach made it possible to evaluate: (i) subjective criteria (evaluation of the valence of images and of subjective emotional reactivity), and (ii) objective criteria of emotional reactivity, by recording startle reflexes. We tested the hypothesis that bipolar patients present high levels of emotional reactivity even during normothymic periods.
The sample studied consisted of two groups of subjects. The first group included control subjects, without bipolar disorder, recruited by means of an advertisement (for example, in shopping areas and sporting events). The second group consisted of normothymic patients with bipolar disorder diagnosed according to DSM-IV criteria and recruited during specialist consultations in an adult psychiatry department (Bordeaux Hospital, Bordeaux, France). Patients were included after giving informed consent, and neither controls nor bipolar patients received any financial compensation.
For inclusion in the group of normothymic bipolar patients, subjects did not, at the time of the evaluation, fulfill the criteria for a major depressive episode or a manic, mixed, or hypomanic episode, according to DSM-IV criteria. Normothymia was confirmed by a general clinical evaluation, carried out by the treating psychiatrist, and by low scores on depressive and manic scales [Montgomery-Åsberg Depression Rating Scale (MADRS) (18) ≤12 and Bech-Rafaelsen Mania Scale (BRMS) (19) ≤4]. Patients with current alcohol or substance misuse were excluded according to the DSM-IV criteria. Moreover, in the control group we excluded subjects with a familial history of bipolar disorders (we did a family tree for all participants). The final sample consisted of 145 subjects: 90 controls and 55 normothymic bipolar patients.
All the subjects (patients and controls) were evaluated with the mood section of the Diagnostic Interview for Genetic Studies (DIGS) (20, 21), a semistructured interview providing DSM-IV criteria. Normothymia was also checked with the MADRS to evaluate the intensity of depression and with the BRMS to evaluate mania. The current symptoms were assessed the same day as the emotional induction.
Emotional induction procedure
Emotional reactivity was evaluated by inducing emotions through emotional visual stimuli and measuring their subjective (valence, arousal) and objective (startle reflex) effects. The protocol comprised three steps.
(i) The first phase involved recording the intensity of the startle reflex when an audible signal was emitted in the absence of any emotional induction by images. The reference value used was the mean of all these recordings. An electromyogram was recorded for each eye. The filters were regulated as follows: one filter was set at 30 Hz (high band-pass filter) and another was set at 200 Hz (low band-pass filter); the frequency selective damping scale was set at 100 μV to limit interference. The measurement of startle reflexes required five electrodes: two under each eye, close to the lashes, and the fifth between the eyes, as described by Grillon and Baas (22).
(ii) In the second phase, the startle reflex was measured in a situation of emotional induction, as described by Herpertz et al. (23). The subject was placed about one meter in front of a screen. A series of 18 photographs from the International Affective Picture System (IAPS) (24), six positive, six neutral, and six negative, were projected onto the screen. The photographs selected were numbers 1300, 2800, 3030, 6550, 9300, 9920, 1670, 5530, 5720, 7002, 7009, 7140, 2352, 4659, 4660, 4680, 7230, and 8031. These photographs were displayed as a succession of six trios, each trio including one positive image, one neutral image, and one negative image. The effect of presentation order was limited by ensuring that positive, neutral, and negative images were a balanced presentation. Each image was presented to the subject for 10 seconds and images were separated by a 10-second period in which the screen was dark gray. Habituation was limited by emitting an audible signal during the viewing of 12 photos only. This signal was emitted 5.5 or 7.5 seconds into the presentation of the image. As for the order of presentation of the images, we also ensured a balanced presentation of sounds.
(iii) During the last phase, the subjects were asked to evaluate, for every image, a valence (the extent to which the image was pleasant or unpleasant) and arousal (the extent to which the image was emotionally moving). This step was carried out with the Self-Assessment Manikin, evaluating each dimension with a nine-point Likert scale (scores of 1 to 9). For the evaluation of valence, low scores were indicative of unpleasant images and high scores were indicative of pleasant images. Photographs with scores close to the middle of the range were considered neutral. For the evaluation of arousal, low scores were indicative of a low intensity of emotion and high scores were indicative of strong emotion.
Sociodemographic variables (age, sex, marital status, profession, educational level) and MADRS and BRMS scores were compared using Chi-squared or Student’s t-tests, in the SPSS-12 package (SPSS, Inc., Chicago, IL, USA).
For the measurement of startle reflexes, we analyzed for each subject the difference in blinking amplitude between the emotional induction phase and the reference phase. Blinking amplitude was transformed into a logarithmic value, as the results were found to be non-normally distributed. We used a repeated-measures ANOVA to compare differences between groups, the effects of repeated measures, and the effects of interactions. Because anxiolytics and neuroleptics were found to have an effect on the variables studied, we controlled for the impact of these drugs on our analyses. A post hoc analysis of statistical power, based on 110 subjects, indicated a minimum power of 0.74 (α = 0.05) for the detection of a mean effect of F = 0.25.
Description of the study sample
The total sample consisted of 145 subjects, including 90 controls and 55 normothymic bipolar patients. Controls and patients did not differ significantly in terms of mean age: 37.8 years (±14.2; range: 20–67) for controls and 39.0 years (±12.61; range: 17–78) for normothymic bipolar patients. Control subjects and normothymic bipolar patients did not differ in terms of MADRS [0.31 (±1.46) and 1.49 (±2.75); range: 0–11, respectively] or BRMS [0.19 (±0.56) and 0.96 (±1.35); range: 0–4, respectively] scores. Among bipolar patients, 58% received a mood stabilizer, 9% antidepressants, 20% antipsychotics, and 20% benzodiazepines. In our statistical analyses we took into account the possible effect of the treatment.
Subjective evaluation of valence
As expected, we found that valence differed significantly according to the type of image [positive, neutral, or negative; F(2,286) = 1000.64, p < 0.0001]. There was a significant group effect [F(3,141) = 2.79, p = 0.04]. This effect resulted from differences in valence evaluation for neutral images. Normothymic bipolar patients and controls attributed similar valences to positive [F(1, 143) = 0.18, p = 0.68] and negative [F(1, 143) = 0.53, p = 0.47] images. However, the bipolar patients assigned a higher valence to neutral images than did the controls [F(1, 143) = 8.40, p = 0.004; see Fig. 1].
Subjective evaluation of arousal
As expected, the level of arousal differed significantly according to the type of image viewed [positive, neutral, or negative; F(2,286) = 253.14, p < 0.0001]. There was a significant difference between control subjects and normothymic bipolar patients [F(3,141) = 4.58, p = 0.004]. The level of arousal reported did not differ between the two groups for positive or negative images [F(1, 143) = 0.90, p = 0.34 for positive images and F(1,143) = 0.19, p = 0.66 for negative images]. However, the normothymic bipolar patients were more moved than the controls by neutral images [F(1,143) = 12.33, p = 0.001; see Fig. 2).
In a first analysis, we showed that the current treatment can have some impact on the blink amplitude. Among all bipolar patients, 31.5% (n = 17) had had no treatment at the moment of the evaluation. The blink amplitude was higher in this group than in the group of patients receiving psychotropic agents, whatever the type [t(91) = 1.79; p = 0.04]. We then explored the influence of each type of treatment. While antidepressants and mood stabilizers did not have a significant effect on the blink amplitude (p > 0.05), antipsychotics and anxiolytics led to a decrease in the blink amplitude (respectively, p = 0.06 and p < 0.001). Then, in the ANOVA analysis, we took into account the presence or absence of antipsychotics and anxiolytics as a confounding factor.
As expected, variations in blinking amplitude differed significantly according to the type of image viewed [positive, negative, or neutral; F(2,274) = 13.36; p < 0.0001]. After controlling for the effect of treatment with anxiolytic or neuroleptic drugs, the normothymic bipolar patients displayed a significantly larger change in blinking amplitude than the control subjects when viewing neutral images [F(3,123) = 3.14; p = 0.03]. This difference in blinking amplitude tended to be larger during the viewing of positive images [F(3,123) = 2.31; p = 0.08] but did not differ significantly during the viewing of negative images [F(3,123) = 1.70; p = 0.17; see Fig. 3].
Potential impact of the most recent episode on emotional reactivity
In order to assess the potential impact of the most recent episode on emotional reactivity, we divided the group of bipolar patients into two groups: the subjects in one group had had an episode during the last 12 months (n = 29) and those in the other had had their last episode after at least one year of remission (n = 22). There was no significant difference between these two groups in emotional reactivity (arousal). Moreover, the last episode (manic, hypomanic, or mixed versus depression) had no impact on emotional reactivity (arousal).
This study of emotional induction shows that the subjective evaluation of valence and arousal and the startle reflex during the viewing of neutral images are greater in normothymic bipolar patients than in control subjects. This demonstrates a higher level of emotional reactivity in normothymic bipolar patients than in control subjects in response to stimuli considered neutral by the control subjects.
These results, obtained with a new methodology, are consistent with those obtained with self-completed questionnaires, showing that normothymic bipolar patients experience stronger emotions and have greater affective reactivity than control subjects (16, 17). Our findings are also consistent with those of Solomon et al. (25) linking the high neuroticism score of bipolar patients with probable emotional hypersensitivity. These previous studies evaluated the emotional dimension of bipolar disease through a questionnaire or scale. Our study evaluated this dimension through an emotional induction protocol, with the collection of data linking subjective and physiological measurements.
The results of previous studies can be reinterpreted in light of these findings of fundamental emotional hyper-reactivity in bipolar patients. According to Judd et al. (26), bipolar patients display episodes of hypomania or depression interspersed with intercrisis subsyndromic elements. These minimal thymic elements may result from fundamental emotional hyper-reactivity, leading to an emotional reactivity that becomes chronic and corresponds to subsyndromal symptoms. The tone of these symptoms in a subdepressive or euphoric and irritable mode may depend on the environmental situation, and also on the temperamental traits of the subjects or even comorbid conditions, including, in particular, anxiety.
This emotional hyper-reactivity seems to increase considerably during manic states (16, 27), being associated with all types of stimuli (positive, neutral, and negative). In normothymic periods, this emotional hyper-reactivity seems to be limited to neutral situations. This can be explained by the fact that positive and negative pictures induced a stronger reactivity that does not allow discriminating differences by reaching a saturation point.
Is emotional reactivity hereditable? It would be valuable to carry out similar studies in subjects’ relatives at risk of developing bipolar disorder. The presence of the same characteristics in the patients and their relatives at risk would identify this emotional hyper-reactivity as an endophenotype associated with susceptibility to bipolar disorder. Emotional hyper-reactivity may also be a risk factor for developing acute episodes in bipolar patients.
Emotional hyper-reactivity may provide an explanation for the particularly acute sensitivity of bipolar patients to life events. Many studies have considered the impact of stress on bipolar disorders, and on recurrence in particular (28). Bipolar subjects not only have a high risk of recurrence when faced with major life events, they also seem to be particularly sensitive to minor life events. Malkoff-Schwartz et al. (29) found that events that seemed minor and could be considered almost neutral were nonetheless likely to disturb social routine and to trigger bouts of mania. Our results are consistent with this reported vulnerability to events that seem to be minor but have major repercussions in bipolar patients. However, they conflict with the conclusions of Johnson (30), who reported that the emotional hyper-reactivity of bipolar patients was limited to positive and encouraging stimuli.
Our results for the startle reflex are consistent with those of several other studies showing that the blinking amplitude of the eyes varies with the valence of the stimulus (31–35). Activation seems to be greater for negative than for neutral stimuli, and greater for neutral than for positive stimuli.
Our study is subject to several limitations concerning the method of induction by images and the selection of the images used. We selected images from the IAPS (24) based on ethical criteria (images unlikely to produce too strong sensations of discomfort or anguish in patients or representing common phobia-related objects). Another possible limitation concerns the treatments. However, in order to limit this potential bias, we controlled this variable in the statistical analyses. Furthermore, there is important psychiatric comorbidity in bipolar disorders and a future study should evaluate other psychiatric diagnoses. In our study, we only excluded patients with current substance or alcohol misuse.
This study raises several questions. Is emotional reactivity an endophenotype of bipolar disorder? If so, it could be used to identify subjects at risk of developing bipolar disorder. What effect does treatment have on emotional reactivity? Is a good response to thymoregulatory treatment associated with a decrease in emotional reactivity? Does an increase in the number of cycles on antidepressants lead to an increase in emotional reactivity? El-Mallakh and Karippot (36) recently showed that bipolar patients treated for long periods (several years) with antidepressants tended to progress to a chronic state of emotional reactivity, mainly dysphoric in tone, associated with insomnia. This raises the question of possible aggravation of the basal emotional reactivity of bipolar patients by the long-term prescription of antidepressants. Studies of emotional reactivity may thus help to improve our understanding of the mechanisms underlying mood changes. Phillips et al. (37), studying the neurobiological disturbances implicated in bipolar disorders, suggest that bipolar disorders are characterized by an increased sensitivity to emotionally salient environmental information and an inability to regulate mood. Moreover, it seems that bipolar patients have an inability to differentiate between relevant and irrelevant emotional stimuli.
Neuroimaging studies in adults and children support the involvement of prefrontal brain regions, and more specifically the dorsolateral prefrontal cortex and the anterior cingulate cortex (38, 39). Thus, the definition of bipolar disorder could be refined considering the sustainable dyscontrol of emotional reactivity, rather than just as the recurrence of episodes. A mild emotional hyper-reactivity characterizes intercrisis periods and partly accounts for the high prevalence of subsyndromal symptoms. Moreover, we have previously shown that bipolar mood episodes can also be defined through emotional reactivity, not only through the tone of affects (euphoria/depressed mood). Manic and mixed states are characterized by a high emotional hyper-reactivity (17) and this concept, when extended to depressive episodes, allows distinguishing two types of bipolar depression. One group is characterized by emotional hyporeactivity, and the other by an increase in emotional reactivity, clinically closer to mixed states (40, 41). This concept helps to understand why mixed patients can feel, at the same time or in a very short interval, emotions as different as euphoria, sadness, irritability, or anxiety. Thus, bipolar disorders should be viewed as a global emotional dysregulation, fluctuating from inhibition to excitation.
This work was funded by the Programme Hospitalier de la Recherche Clinique, 2003. KM was supported by a grant from the Ministère de la Recherche.