Amygdala, Functional Imaging

Synonyms

Positron emission tomography (PET); functional magnetic resonance imaging (fMRI)

Definition

The amygdala is an essential key structure in the cerebral limbic network underlying emotion processing. As such, it is suggested to be part of the brain circuit involved in the processing of pain, which is known to include strong affective components. Neuroimaging studies pointing to amygdala involvement during pain processing are currently increasing. The amygdala is a small almond shape structure in the anterior temporal lobe with a variety of functions for emotion processing together with learning and memory. It is supposed to execute an evaluative associative function, combining external cues with internal responses, thereby assessing and defining the valence, relevance and significance of stimuli. It is its extensive connectivity with various cortical and subcortical areas that enables fast automatic, but also more conscious deliberate, responses. Its role in pain processing is however less clear.

Characteristics

Negative affect is typically evoked by acute pain. Key structures of the limbic system have been identified that play an important role in regulating affective behavior; among the most important are the subcortical and cortical areas, the anterior cingulate, the insula and the prefrontal cortex. Most notably, assessment of emotional valence of stimuli and the provocation of distinct emotional reactions are mediated by the amygdala. This central role in emotion processing can be executed due to a broad cortical and subcortical network in which the amygdala is located and which is able to provide it with raw information via the short thalamus route but also with highly processed polymodal input from sensory cortices. Finally, the amygdala is not a unitary structure, but consists of several nuclei exerting different functions. It is believed to have a major role in pain because of the strong association and interaction between pain and emotion, but also because of the specific nociceptive inputs to the latero-capsular part of the central nucleus, the major output system within the amygdala, indicating that, within this accumulation of nuclei, this part may represent the “nociceptive amygdala” (Neugebauer et al. 2004). For fMRI, mapping of activation within this region is, however, critical posing technical and methodological problems, which often call into question the validity and reliability of imaging results reporting amygdala activation. This may possibly be one of the reasons, why early neuroimaging findings mostly failed to demonstrate clear amygdala activation during pain perception. FMRI of this deep subcortical region is confronted with a set of difficulties, such as movement, respiratory, inflow and susceptibility artefacts (see inflow artefacts) and nonetheless the rapid habituation of amygdala responses to repeated stimulus presentations. This is of special relevance for experimental pain studies, which mostly rely on the application of block designs, which are especially prone to habituation. Recent methodological advances in neuroimaging may have partly overcome these inherent mapping difficulties, accounting for the increase in pain studies successfully demonstrating amygdala participation (Bingel et al. 2002; Bornhövd et al. 2002).

Alternatively, it is also conceivable that the majority of pain stimulation techniques failed to evoke pain that provoked strong emotional responses, hence falling short of observing amygdala involvement. The frequent failure of these early studies to report changes in autonomic arousal during painful stimulation corroborates this assumption. In an attempt to model acute traumatic nociceptive pain, a PET study used intracutaneous injection of ethanol (Hsieh et al. 1995). Affective and heart rate changes were described in subjects and cerebral activation was found in subcortical structures, specifically the hypothalamus and the periaqueductal gray. These regions are taken to constitute the brain defense system which functions as a modulator for aversive states. Although signal increases in the amygdala were detected by the authors, they failed to be significant.

Despite more recent neuroimaging findings reporting amygdala involvement in pain processing, a full characterization of its function during pain perception is still lacking and at first sight results seem to be equivocal, pointing to activations as well as deactivations of the amygdala in this context (Table 1).

Table 1 Overview of pain studies reporting amygdala activation

One fMRI investigation applied painful stimulation with a strong affective component to measure pain related changes in cerebral activity (Schneider et al. 2001). By inflating an indwelling balloon catheter, a dorsal foot vein of healthy volunteers was stretched to a noxious distress physical level, which induced vascular pain associated with a particularly strong negative affect. Since the sensory innervation of veins exclusively subserves nociception, non-painful co-sensations were excluded. Additionally, brief stimulations of only a few minutes produce vascular pain that escapes adaptation and is generally reported as particular aching in character. During noxious stimulation, the subjects continuously rated perceived pain intensity on a pneumatically coupled visual analogue scale, which was used as permanent feedback to adjust balloon expansion so that the pain intensity could be kept at intended values at all times. The analysis strategy that focused primarily on correlations of signal changes with these subjective ratings, rather than the generally applied signal variations to a stimulation based reference function (boxcar design), facilitated producing evidence for amygdala activation (Fig. 1). Hence, these results indicated a relevant role of the amygdala in the subjective component of painful experiences and suggested that in the widespread cerebral network of pain perception, the limbic system and especially the amygdala may be instrumental in the affective aspects of pain. Supporting evidence for these conclusions come from neuroimaging findings during air hunger (Evans et al. 2002) or fundus balloon distension (Lu et al. 2004). Dyspnea was induced in healthy subjects by mechanical ventilation until a sensation of “urge to breathe” and “starved for air” was reached and compared to mild hypocapnia. This pain is also very afflicted with strong negative affect. Correspondingly, a network of limbic and paralimbic nodes was activated, including anterior insula, anterior cingulate, operculum, thalamus, cerebellum, basal ganglia and also amygdala, that is the majority of regions forming part of the limbic network also involved in emotion processing. Similarly, moderate gastric pain was induced in 10 healthy subjects using fundus balloon distension (Lu et al. 2004) and resulted in a widespread activation pattern of subcortical as well as cortical regions, among them insula and amygdala. This may once again point especially to the strong affective component of visceral pain. Since visceral pain may be indicative of an urgent and marked system imbalance possible endangering survival, strong affective responses with the objective of initiating adequate adaptations and reactions seem to have an evolutionary purpose and be necessary.

Figure 1

Individual signal intensities in the amygdala following correlation with subjective ratings of the six individual participants (from Schneider et al. 2000).

Amygdala activation is however not restricted to visceral pain, but also visible during other kinds of painful stimulation in animals as well as humans (Bingel et al. 2002). Unilateral laser evoked painful stimuli of either side, which also avoided concomitant tactile stimulation and anticipation as well as habituation, successfully demonstrated bilateral amygdala activation, most probably representing the affective pain component (Fig. 2). In contrast, basal ganglia and cerebellum displayed corresponding unilateral activation and may probably be related to defensive and withdrawal behavior. RCBF (regional cerebral blood flow) changes were also found in limbic structures of rats during noxious formalin nociception (Morrow et al. 1998).

Figure 2

Amygdala activation emerged bilaterally in response to painful unilateral laser stimulation. Left: Fittet responses applied to the left (blue line) or right (red line) hand for the left (left graph) and right (right graph) hemispheres. The dotted lines show the standard error of the mean (SEM) (from Bingel et al. 2002).

Hence, the role of the amygdala as a “sensory gateway to the emotions” (Aggleton and Mishkin 1986) with an evaluative function seems to extend to pain perception as well. An increasing number of studies supported the notion of a common evaluative system with a central role of the amygdala in the processing of painful but also non-painful or novel stimuli. The amygdala not only demonstrated coding of the pain amount by showing a linearly increasing response to augmenting painfulness (Fig. 3) but also significant responses during uncertain trials in which the stimulus was not perceived and hence a judgment on the nature and valence is required (Bornhövd et al. 2002). Furthermore, the amygdala, here more specifically the sublenticular extended amygdala, seems to be characterized by early responses (to noxious thermal stimuli) in contrast to regions activated later and associated specifically to somatosensory processing, such as thalamus, somatosensory cortex and insula (Becerra et al. 2001) (Fig. 4). This is in accordance with the activation characteristic of the amygdala during classical conditioning (Büchel et al. 1998), in which a rapid adaptation to the conditioned stimulus has been observed, pointing to a major role of the amygdala during the early phase of learning, during the establishment of an association between the neutral stimulus and the (un)conditioned response. Hence, the early response during pain seems to reflect the association between the painful stimulus and an adequate internal response determining the negative valence of the stimulus.

Figure 3

Picture: Bilateral amygdala activation (p = 0.001) on a coronal slide. Graphs: Left side entails regression coefficients indicating amount of response for each trial (P0–P4). Right side depicts amount of signal change in the amygdala as a function of peristimulus time separately for all stimuli (P0–P4; from Bornhövd et al. 2002).

Figure 4

Coronal slices showing sublenticular extended amygdala (SLEA) activation in the early (left) and late phases (middle) in response to a noxious thermal stimulation (46°C). Overlap (white) of early (yellow/red) and late (blue) phases (right).

However, sometimes deactivation as opposed to activation has been observed in the amygdala during painful stimulation, for example with fMRI in response to thermal stimuli (45°C) (Becerra et al. 1999). In this study only 6 subjects were investigated and changes were low-level. Similar deactivations have also been reported using PET during mild or moderate pain due to CO₂ laser stimulation compared to non-painful warm sensations (Derbyshire et al. 1997). Hence, a possible moderating variable for activations and deactivations may be the specific thermal pain sensation, which was similar during both experiments. Alternatively, the deactivation may reflect another functional activation characteristic of the amygdala under certain circumstances. Hence, the deactivation may simply be the consequence of the nature of the experimental pain stimulus. An early activation in the amygdala for purposes of evaluation and affective judgment may be followed by a deactivation, possibly representing the attempt to regulate and cope with the affective aspects of the painful experience as well as the painful sensation itself that cannot be escaped in this special experimental setup. This interpretation is supported by recent PET findings. Petrovic et al. (2004) investigated the influence of context manipulations before the painful stimulation on the activation pattern during noxious (cold pressure) stimulation. Subjects were informed prior to stimulation if it was going to be painful or not and if it would last for 1 or 2 min. Anticipating that the pain was going to last longer was accompanied by a decrease in amygdala activation and changes in autonomic parameters, but also cognitive processes in the majority of subjects that consisted of strategies to cope with the stressful but unavoidable pain. This amygdala deactivation was paralleled by activation in the anterior cingulate, pointing to interactions within this limbic network constituting the brain’s pain matrix responsible for the development and modulation as well as coverage and termination of the affective noxious events.

This study also highlights some methodological problems of pain imaging studies in general and those with a special focus on the amygdala. Anticipation may alter amygdala response characteristics and may lead to deactivations instead of activations. Furthermore, the individual variability in pain responses and several methodological factors, such as imaging method, data analysis, control condition used for comparison with pain condition etc. influence results as well as their interpretation.

However, further indications that the amygdala serves coping functions during pain perception come from clinical trials. Here, visceral pain hypersensitivity is discussed as a possible relevant pathogenic factor in various chronic pain syndromes, such as irritable bowel syndrome (IBS). Reduced signals in the amygdala (as well as in further limbic network nodes such as insula and striatum) have also been observed in patients with irritable bowel syndrome during rectal pain stimulation (Bonaz et al. 2002) and are in accordance with the interpretation of deactivations found in healthy controls. It may be suggested that deactivations in patients may correspond to the effort to modulate and control the strong affective components of the painful experiences. Unfortunately this study failed to include healthy controls and hence, a conclusion on the dysfunctional or compensatory aspects of these activations in patients remains elusive. Interestingly, a recent fMRI study (Wilder-Smith et al. 2004) investigating rectal pain alone or accompanied by painful foot stimulation (ice water, activating endogenous pain inhibitory mechanisms) in patients with irritable bowel syndrome as well as healthy controls found differential activations between groups in the amygdala (activation in constipated patients) as well as further affective-limbic regions (hippocampus, insula, anterior cingulate, prefrontal cortex etc.) during heterotopic stimulation.

Hence, the amygdala is not only implicated in the affective aspects of pain processing, including both the appraisal of a painful stimulation with the initiation of adequate responses, and the experiential affective aspects, such as stress, fear or anxiety but also the modification, attenuation and coping of these affective experiential aspects. This multiple functionality is supported by behavioral findings demonstrating amygdala activation during enhancement as well as inhibition of pain (Neugebauer et al. 2004). First, it may be a protective mechanism to detect a possible harmful stimulus, hence amplifying the painful experience; however, in case of unavoidable harm or pain, it may be the most suitable response to reduce the painfulness by inhibition (for example via the periaqueductal gray). Finally, the central role in pain and emotion makes it highly likely that it may also be involved in the dysfunctional aspects of chronic (visceral) pain. For example, the involvement of the amygdala during memory and learning may be relevant facets for the development of chronic pain.

However, the diversity of functions exerted by the amygdala as indicated by the different imaging studies on experimental and chronic pain, such as affective painful experience but also modulation of this experience as an evolutionary sensible warning and evaluative survival system, including an effective adaptation mechanism in case of inescapable painful stimulation, suggests the involvement of other brain regions as well. Hence, the function of the amygdala cannot be determined alone but only within a greater cortical and subcortical network. Despite its relevance, it is only the continuous and intensive interconnections, interactions and feedback mechanisms with other brain regions that account for the complex and intact function of this structure in pain and emotion.