Preserved emotional awareness of pain in a patient with extensive bilateral damage to the insula, anterior cingulate, and amygdala

Roger is a 55-year-old fully right-handed male with 16 years of education. At the age of 28, he survived a life-threatening episode of herpes simplex encephalitis (HSE), a viral attack that triggers a process of necrosis within the brain causing the “total disintegration of the affected tissue” (Hierons et al. 1978). In this case, HSE destroyed most of his limbic system, bilaterally, including his insula, ACC, and amygdala (Figs. 1, 2). Of note, the term limbic system has been used to describe Roger’s brain damage since virtually all of the key structures typically defined as comprising the limbic system have been bilaterally damaged in Roger (Feinstein et al. 2010). Detailed descriptions of his medical history and analyses of his lesion and neuropsychological profile have been reported elsewhere (Feinstein et al. 2010; Philippi et al. 2012). Remarkably, much of Roger’s cognitive abilities are within the normal range, including his intellectual functioning, speech, language, attention, working memory, and metacognition. His main presenting deficits include a dense global amnesia, along with anosmia and ageusia.

Roger’s subjective pain ratings were compared to a group of 29 healthy non-brain-damaged male subjects (average age: 29 years; range 18–55) using an identical cold pressor task and similar procedures from a previously published dataset (Lee et al. 2010).

Three T1-weighted magnetic resonance imaging (MRI) scans were acquired on a 3.0-Tesla Siemens Trio MRI scanner (MPRAGE, AC-PC aligned coronal acquisition; TR = 2,530 ms; TE = 3 ms; TI = 800 ms; flip angle = 10°; FOV = 256 × 256 mm²; slice thickness = 1 mm). The images were bias field corrected and registered together with a rigid body transformation and a sinc interpolation (AIR 3.08). The three scans were then averaged together in order to reduce motion artifacts, increase the signal-to-noise ratio, and enhance the contrast-to-noise ratio between gray and white matter (see Fig. 1). The images were then converted into MNI space (see Fig. 2).

To assess the experience of pain, we utilized the cold pressor paradigm, a gold standard that has been repeatedly used over the past century to safely induce transient states of intense pain (Edes and Dallenbach 1936; Lovallo 1975; Rainville et al. 1992). Roger underwent four cold pressor immersions (two left-hand trials and two right-hand trials), each on a separate day to avoid the effects of pain summation. During each cold pressor trial, Roger’s hand was immersed (up to the wrist) in circulating water maintained at a temperature of 0 °C (32 °F). Water was circulated continuously to maintain target water temperature and avoid localized warming around the hand. The forearm was supported using a soft armrest and the hand was maintained at a constant immersion depth. Prior to each cold pressor trial, Roger underwent a baseline warm water immersion, placing the same hand in a different tank of water maintained at 33 °C (91.4 °F). The warm water trials served as a control condition and normalized limb temperature prior to each cold pressor test. For all trials, a white curtain was hung between Roger and the immersion tank, blocking his ability to see his immersed hand (in either the warm or cold water). Trials were conducted in a single-blinded fashion such that Roger was not informed beforehand whether his hand would be immersed in warm or cold water.

Throughout each immersion, Roger continuously rated his moment-to-moment level of pain (instructions detailed below) using a 10-cm electronic visual analog scale (eVAS) along both sensory and affective dimensions. Pain ratings (recorded to the nearest millimeter) were transmitted from the digital linear potentiometer to a laptop computer at a sampling rate of 50 Hz. A stopwatch was used to calculate the total immersion time. During each immersion, Roger’s facial and vocal responses were recorded using a digital video camera and we also recorded continuous measures of heart rate, skin conductance, and facial electromyography (EMG) of the corrugator.

For each hand, there were two different types of trials: “sensory” and “affective”. During the sensory trials, Roger continuously rated the intensity of his pain, ranging from “No Pain” to “Worst Pain Imaginable”. During the affective trials, Roger continuously rated how “unpleasant” or “bothersome” the pain felt to him, ranging from “Not at all unpleasant” to “Extremely unpleasant”. For all immersions, rating instructions were repeated every 30 s because of his amnesia. Prior to each immersion, the difference between pain affect and pain sensation was described to Roger using a standardized set of instructions (Price et al. 1983): “There are two aspects of pain which we are interested in measuring: the intensity, how strong the pain feels, and the unpleasantness or how disturbing the pain is for you. The distinction between these two aspects of pain is like listening to a radio. As the volume increases, I can ask you how loud it sounds or how unpleasant or bothersome it is to hear. The intensity of pain is like loudness; the unpleasantness of pain is like how disturbing or bothersome the sound is. Please indicate how intense or unpleasant/bothersome this task is when we ask you.”

All facial coding was performed by an expert coder (KM Prkachin) using the Facial Action Coding System (Ekman et al. 2002). Previous work (Prkachin 1992; Prkachin and Solomon 2008) suggests that four facial actions comprise the bulk of a prototypical facial expression of pain: (1) brow-lowering; (2) tightening the eyelids or raising the cheeks (orbit tightening); (3) nose wrinkling or upper-lip raising (levator contraction); and (4) eye closure. Each of the first 3 facial actions was coded on a 6-point intensity scale (from 0 = no action, 1 = minimal action to 5 = maximal action). Eye closure was coded on a binary scale (0 = eyes open, 1 = eyes closed). A composite score was computed by summing the 4 facial action scores together (0 = no trace of a pain face to 16 = maximum expression of a pain face). Video clips of equal durations (corresponding to the first 50 s of immersion) were created for all four cold pressor and four warm water immersions. The eight video segments were randomized prior to coding and all facial coding was performed without sound so that the rater was blinded to the condition being viewed. Actions were coded on a frame-by-frame basis at a time resolution of 67 ms per frame.

Physiological data (including heart rate, skin conductance, and corrugator facial EMG) were recorded continuously during all trials with an MP100 acquisition unit (Biopac Systems, Inc). Heart rate was collected via lead II configuration, at a sampling rate of 250 Hz. EMG responses were recorded from the corrugator muscle, at a sampling rate of 1,000 Hz. Skin conductance level was recorded using two electrodes placed on the thenar and hypothenar eminences on the non-immersed hand, at a sampling rate of 1,000 Hz. For all physiological measures, change scores were computed moment-to-moment based upon the averaged 10-s period of time immediately preceding each immersion.

These findings have important theoretical implications for the notion of a “pain matrix”. Previous work has already raised the question about whether the pain matrix is specific to nociception, and the evidence shows that activation within the matrix is not sufficient for the experience of pain and can be induced by a variety of non-nociceptive stimuli (Iannetti and Mouraux 2010; Mouraux et al. 2011). Here, we extend this body of work by showing that key structures of the pain matrix are also not necessary for the experience of pain, including pain’s affective dimension (Price 2000; Rainville et al. 1997). In this context, it is important to note that Roger’s damage also includes structures posited to play a more basic role in pain sensation, including the posterior insula (bilaterally) and the adjacent parietal operculum and secondary somatosensory cortex (in the right hemisphere). Based on this additional damage and the noted presence of pain following stimulation of both sides of the body, it can be further deduced that these lesioned sensory structures are also not necessary for the experience of pain, calling into question recent claims that the dorsal posterior insula is the brain’s “primary cortex for pain” (Garcia-Larrea 2012). Thus, the core limbic structures of the pain matrix, including insula and ACC, are neither necessary nor sufficient for the experience of pain.

Such a conclusion refutes the reverse inferences being made in functional neuroimaging investigations of social pain. For example, activation of the ACC during studies of social rejection may not be the causal factor explaining why “rejection hurts” (Eisenberger et al. 2003; Eisenberger and Lieberman 2004; Eisenberger 2012). Likewise, activation of the anterior insula during studies of empathy may not represent the vicarious experience of another’s pain (Singer et al. 2004). The results in Roger raise the possibility that the overlapping activation found in the insula and ACC during states of physical and social pain does not necessarily reflect the experience of pain itself. Indeed, patients with congenital insensitivity to pain showed normal levels of insula and ACC activity when observing pain in others, even though they themselves are unable to feel pain (Danziger et al. 2009). On the other hand, psychopaths showed abnormally high levels of insula activity when observing pain in others, leading the authors of this study to conclude, “the role of the insula in emotion and empathy is complex and far from being understood” (Decety et al. 2013). Collectively, these data challenge the core assumptions underlying the neural connection between physical and social pain (cf. Iannetti et al. 2013), while underscoring the importance of resisting causal attributions based purely on functional neuroimaging data (Feinstein 2013).

Beyond pain, the insula and ACC are the two most commonly activated structures in any functional neuroimaging investigation of emotion and feeling (Craig 2009; Phan et al. 2002). While it would be tempting to conclude that the insula and ACC are the brain’s primary substrate for emotional experience, the case of Roger reveals that neither region is actually necessary for such experience to occur. This is an important point since several investigators have recently claimed that the anterior insula is the necessary substrate underlying all forms of emotional awareness (Craig 2009; Gu et al. 2013). We have previously shown that Roger’s self-awareness is remarkably preserved across a large battery of tests (Philippi et al. 2012), including a measure of interoceptive awareness (Khalsa et al. 2009). In this study, we demonstrate that many aspects of Roger’s emotional awareness are also preserved, a finding which casts grave doubt on the assertion that the insula is the brain’s most important substrate for feeling (Craig 2015; Damasio 2003).

Quite unexpectedly, not only did Roger feel pain, but his pain experience was at times excessive in nature and potentially hyperpathic. The International Association for the Study of Pain characterizes hyperpathic pain as “an abnormally painful reaction to a stimulus” that “is often explosive in character” (cf. http://​www.​iasp-pain.​org). Consistent with this definition, Roger’s subjective rating of pain during cold pressor trials was often at the maximum level, well above the 75th percentile of the corresponding ratings from the comparison group (Fig. 3a). Likewise, he reached his pain tolerance threshold and withdrew his hand much earlier than the majority of subjects in the comparison group. Perhaps the clearest evidence of hyperpathic pain was in his vocalizations, which were explosive and disinhibited (Table 1). These indications of hyperpathic pain in Roger will need to be further investigated using more precise thermal and mechanical pain-induction techniques, and compared to an age- and race-matched sample. Despite these limitations, Roger’s data suggest that the limbic structures commonly associated with pain may play a fundamental role in pain regulation. Under this view, the missing regions in Roger’s brain would impair his ability to control and downregulate his pain responses. This would be in line with a large body of literature on the role of the ACC in adaptive control, not only for pain, but also cognition and autonomic arousal (Shackman et al. 2011; Botvinick et al. 2004; Critchley et al. 2003). Such an interpretation is also consistent with the significant disturbances found in circuitry involving the insula, ACC, and amygdala in patients and animals under conditions of chronic pain (Borsook and Becerra 2006; Bushnell et al. 2013; Veinante et al. 2013). However, Roger is not a chronic pain patient and it is unclear how his brain damage might reflect alterations in the chronic pain state. Together, these findings suggest that the functional role of the limbic structures comprising the pain matrix may be more aligned with the adaptive regulation and response to pain rather than providing the decisive substrate for pain’s conscious experience, affective or otherwise.

The question remains as to which brain regions might be supporting Roger’s preserved experience of pain. We attempted to investigate this question with Roger using fMRI. Over the course of many different runs (collected over multiple days), Roger was unable to refrain from pain-related movements, an unfortunate byproduct of his aforementioned hyperpathic responses. In spite of all our efforts at correcting his movement artifacts, the pain fMRI data were not exploitable. We hope that methodological solutions will be found for rendering future pain-related fMRI studies with Roger possible. Until then, we can only speculate as to which brain regions might be supporting his pain experience. From a neuroanatomical perspective, nearly all pain signals traverse through nuclei within the brainstem and thalamus, both of which are largely intact in Roger and could be playing a critical role in his pain experience. While Roger’s ACC is destroyed bilaterally, there is some remaining tissue in the left hemisphere that corresponds to a dorsal region of Brodmann area 32 within the paracingulate gyrus. This region partially overlaps with a recent functional neuroimaging meta-analysis of pain-related activations (cf. Box 1 and Fig. 2 in Shackman et al. 2011) suggesting that adjacent territories lying dorsal to the anterior midcingulate cortex (including the paracingulate gyrus and supplementary motor area) may be just as important for pain as the ACC itself. Such an interpretation is consistent with the observation that most patients with akinetic mutism—a state that is often accompanied by a complete indifference to pain—have large bilateral lesions that typically impact both the ACC and the supplementary motor area (Damasio and Van Hoesen 1983). Roger’s secondary somatosensory cortex in the left hemisphere is also intact and might provide a viable compensatory route for the damage to this region in his right hemisphere. Of note, activation in the secondary somatosensory cortices appears to have one of the strongest relationships with the “subjective reality of pain” (Raij et al. 2005). Another possibility to consider is the primary somatosensory cortices, which are intact in both hemispheres, and have previously been shown to play a vital role in Roger’s preserved interoceptive awareness for cardiovascular sensations (Khalsa et al. 2009). Interestingly, a recent fMRI study tested two lesion patients with unilateral left insula damage and found “dramatically elevated” levels of activation in the primary somatosensory cortex related to the patients’ higher pain ratings during noxious heat stimulation (Starr et al. 2009). Another noteworthy point of the Starr et al. (2009) study was the surprising absence of activation in both patients’ right insular cortex during painful stimulation of the right leg. This provides further evidence that the insula is not essential for the experience of pain. Nevertheless, it is worth considering the possibility that the small island of tissue remaining in Roger’s left dorsal anterior insula (accounting for less than 10 % of total insular volume) could be contributing to his pain experience. Several points argue against this possibility: (1) Patient 2 in the Starr et al. (2009) study had damage that completely subsumed the left anterior insula, yet he continued to experience pain; (2) there are anecdotal reports of pain being experienced by another encephalitic patient with 100 % complete bilateral insula destruction (Damasio et al. 2013); and (3) in a previous study, we found that the tissue in Roger’s left anterior insula was both structurally and functionally disconnected from the rest of his brain (Philippi et al. 2012). Based on this evidence, it is highly unlikely that this small island of tissue would be playing a prominent role in Roger’s preserved pain experience.

The case of Roger establishes that the emotional experience of pain can be instantiated by brain structures outside of those traditionally presumed to be critical for pain affect, thus highlighting the widely distributed nature of pain processing in the brain (Coghill et al. 1999). Due to the chronicity of Roger’s damage, substantial recovery of function is plausible via reorganization and transfer to other brain systems (Rudrauf 2014). It is important to emphasize that in such a scenario, the very possibility of recovery would imply that the damaged regions are not necessary for such experience to occur (in contrast, for instance, to early visual cortices which are necessary for visual awareness). This brings forth an educated guess about Roger’s case, namely that his intact affective experience of pain is due to plasticity, which helped preserve a vital function for survival by maintaining his affective response to pain despite damage to regions that might normally serve this function. In other words, the adaptive role of pain affect is so essential that the brain may automatically rewire in service of self-preservation. Consequently, the neural circuitry underlying pain and the associated feelings of suffering and distress is more complicated than previously thought, with multiple pathways and built-in redundancy allowing for maximal adaptation and resilience in the face of brain injury.