fMRI and Human Pain Perception
Abstract and Keywords
Almost 30 years ago, technology based on magnetic resonance imaging (MRI) made it possible to visualize the functional states of the human brain. This technology immediately spurred pain researchers to examine brain circuitry of human pain and relate brain activity patterns with verbal reports of subjective perception. There was a brief period prior to functional MRI (fMRI) when positron emission tomography (PET) and single-photon emission computed tomography (SPECT) technologies were used to identify brain states in humans reporting pain, but the noninvasiveness of fMRI and its higher spatial and temporal resolution quickly made the latter the preferred choice to study human brain physiology. Prior to the advent of such human brain imaging technologies, whether the neocortex was involved in pain perception was still an open question: In human brain injury studies, large cortical lesions seemed to have little effect on pain perception, and in animal electrophysiological studies (mostly done in anesthetized preparations) several years of single-unit electrophysiological explorations from large expanses of the cortex yielded a measly number of neurons responding to nociceptive stimuli and not a single neocortical column dedicated to nociception. What has been learned between the introduction of the technology and today? This chapter briefly reviews the subject, highlighting advances and novel insights and pointing to lingering gaps. It also outlines future directions from the viewpoint of understanding mechanisms for nociception, acute pain, and chronic pain. From a brain imaging viewpoint, the chapter tackles the last concepts regarding local neuronal representation and across neuronal integration of information.
fMRI and Neural Activity
The human brain comprises 86 billion neurons, with trillions of synaptic connections between them (Azevedo et al., 2009). Human cognitive states must underlie coordinated communication across these elements at space-time scales that still remain ambiguous with respect to the number of neurons and synapses involved. Although the blood oxygen level–dependent (BOLD) functional magnetic resonance imaging (fMRI) signal does not measure neural activity directly, but relies on a cascade of physiological events consequent to neural activity (Iannetti & Wise, 2007; Logothetis, Pauls, Augath, Trinath, & Oeltermann, 2001), fMRI provides a fairly accurate spatial localization of the information regarding neuronal activity in the human brain.
Standard fMRI protocols allow parceling the brain into as many as 200,000 voxels, each providing averaged information about the functioning of about 4,300,000 neurons per location (86 billion divided by 200,000 voxels). Moreover, the possibility of measuring the temporal variability of the fMRI signal for each of these voxels, and thereby calculating their pairwise interactions, also gives access to the human brain information integration across all pairs of voxels, giving rise to a network with the astronomic dimension of about 200,000,000 elements.
The time resolution of the fMRI signal is limited both technologically, by the scanner sampling rate, and physiologically, by the temporal profile of the hemodynamic response. Current state-of-the art multichannel MRI scanners routinely sample BOLD signal from the entire brain in approximately 500 ms, although the hemodynamic low-pass filtering of neuronal activity limits resolving neuronal information to, at best, a few seconds. This little exercise is meant to illustrate the richness of information afforded by whole-brain fMRI, which can be pushed even further with higher strength magnets. Thus, at least hypothetically, scientists have ready access to the human brain physiological states at a spatial and temporal resolution that must be ample to unravel the neurobiology of perceptions and behaviors (in practice, given the high-dimensional space, statistical limitations—especially regarding number of subjects studied per condition or state, we need to down-sample in both space and time the fMRI signal to reach to reproducible and generalizable conclusions). Although scientists who study activities at the level of a single neuron or single synapse object that the fMRI scale is too sluggish in time and far too imprecise in space (e.g., representations of individual odors are distributed across the piriform cortex with no apparent global or local spatial structure; Schaffer et al., 2018), the fMRI science is based on the conviction of the system being macroscopically organized into an interconnected network of homogeneous gray matter regions communicating functionally and anatomically through white matter connections (Bassett & Sporns, 2017).
For fMRI analysis, the gray matter is typically divided into brain regions composed of groups of adjacent voxels that have similar properties (the spatial scale at which specific experimental conditions are best represented remains unknown) and demonstrate specialized information processing and knowledge representation. The white matter provides gross structural connections between distant brain regions delimiting the structural wiring of the brain, which is supplemented with electrical couplings across groups of neurons and brain regions that seem to expand on the white matter connectivity (Zimmermann, Griffiths, & McIntosh, 2018).
Local knowledge shared across networks is hypothesized to support perception, cognition, and human behavior by communicating and efficiently integrating information across distant brain regions (Passingham, Stephan, & Kotter, 2002). Consequently, localized activation and coordinated communication, constrained by specialized processing of individual brain regions and reflected in functional connectivity across such brain regions, should identify the different brain states. Overall, this is a reasonable assumption, and on this basis fMRI has importantly advanced knowledge of human brain function. In fact, one can add the reverse argument that if to accurately understand human brain states we need to know the state of every neuron and every synapse, it follows that the endeavor of unraveling human cognitive neuroscience is guaranteed to be doomed. Still, despite the substantial progress that human cognitive neuroscience has achieved thanks to fMRI technology, and despite the that high dimensionality of the raw fMRI data points to a rich trove of information, only a small proportion of this information is currently used and is even less understood; novel approaches await further untangling the BOLD fMRI signal. Additional specific issues relate to the chain of physiological events linking neural activity to the BOLD signal when fMRI is used in pharmacological and clinical studies (Iannetti & Wise, 2007), as it is often the case in pain research.
Before expounding on the fMRI of human pain, we first visit definitions and constructs for pain and nociception.
A New Definition of Nociception
A series of recent articles revisited the distinction between nociception and pain (Apkarian, 2019) to reiterate and refine the original concept that the two must be differentiated from each other (Melzack & Wall, 1996). The driving observation is the fact that most of us, and as far as we can tell most animals as well, live a life most of the time devoid of pain. At the same time, we live this mostly pain-free life also in the absence of injuries, within an environment that however is potentially highly noxious. Thus, our tissues are routinely protected from injuries in the absence of painful signals. As nociceptors can be active in the range of positions, forces, or movements that approach levels where injury may be impending, one has to conclude that nociceptive activity, in the absence of conscious pain perception, continuously and subconsciously delimits and modulates our behavioral repertoires, thus constantly protecting tissues from injuries (even during sleep, where occasional body movements are necessary to protect tissue from inflammation, injury, bedsores, etc.). Therefore, we define nociception as a subset of those supramodal mechanisms (i.e., mechanisms also contributed by non-somatosensory modalities) that continuously and subconsciously protect the body from injury. Of course, nociception can, in certain circumstances, give rise to painful percepts.
A New Definition of Pain and Chronic Pain
In contrast to nociception, pain should be defined as a conscious state signaling the failure of protection of the body from injuries and thus necessitating a more urgent change in behavior. Pain also induces learning, thus shaping avoidance behaviors, as discussed in the previous paragraph. Formally, this revised definition of pain (Apkarian, 2019) is stated as a conscious somatically embodied negative emotional state, associated with strong urges to modify the current state of beingness signaling the failure, or the potential for failure, of nociceptive and non-nociceptive processes to protect the body from injury. A corollary is that negative emotions and moods by themselves are not pain, and injury and nociceptive signals per se also do not necessarily lead to pain. Instead, conscious subjective evaluation of the injury and related nociceptive inputs are necessary to pain. Such evaluations enable identification of body parts from which the input arises (even if sometimes with low precision, as in visceral pain) and simultaneously involve the valuation–motivational limbic circuits, generating an error/surprise signal (perhaps in proportion to the negative affective response) in comparison to past memories, which in turn provides a learning signal that conditions future responses to similar stimuli, and finally engages dorsal striatal–cortical motor circuitry for action selection.
Within this construct, chronic pain may be viewed as a spectrum of persistent conditions, where there is a gain of the nociceptive afferent system at peripheral and spinal cord levels, a gain of the affective cost of the nociceptive input through central limbic routes, or both. These mechanisms eventually create and maintain the chronic pain brain state. It should be emphasized that modulation of nociceptive inputs at spinal, mesolimbic, and prefrontal cortical levels must be considered fundamental to endowing pain its subjectivity (both acute and chronic) and its multidimensionality, as well as to defining its attentional–emotional–motivational attributes.
Perception and Pain Perception
From a sensory viewpoint, acute pain is a conscious somatically embodied state with clear subjective negative affective component—perhaps best exemplified by the saying, “If you are doubting reality, kick a rock with bare feet.” Thus, we commonly associate an acute painful experience as having a shared subjectivity or a “quale” of pain. In a sense, the experience would be equivalent to a color in visual experience. Thus, this subjective affective coloring of a sensory state would be represented by neocortical neural patterns and their interactions (connectivity).
We need to first emphasize that all subjective sensory states remain unknown regarding underlying physical mechanisms. Thus, any theory of perception or subjectivity for the time being is at best a neuronal correlate to verbal reports of subjectivity, no matter how many neurons or synapses one can record from. The same is true for pain. In the case of acute pain, there is even debate whether a common qualia is shared across humans. Unlike color vision where a veridical assessment of the wavelength bouncing off the object and hitting the retinal can be quantified (although its relationship to subjective color is not constant, similar to the relationship between skin temperature and pain), in the case of pain, the chair on which you are comfortably sitting in time “becomes” painful. There is no way to communicate whether your qualia of the painful chair somehow matches my qualia of the painful chair. Yet, as brain scientists with tools that give us access to the brain physiology, we have performed such experiments now thousands of times.
Nociception and Pain-Related Behaviors
All animals, and even single-cell organisms, possess nociceptive machinery, as mobility necessitates avoiding noxious surroundings. The mobile animal solves the problem of survival by continuously improving on the dual objective of seeking nutrition and avoiding injuries (improving on decreasing injuries gives rise to the habits of building protective nests, which one can argue eventually leads to creation of civilization). The most sophisticated expression in the advancements of avoiding injuries then becomes conscious perception of pain, which endows the organism with tools for propagating information about the event across all cognitive domains (consciousness) and enables immediate action (behavior) to escape the potential injury, above and beyond the subtle unconscious behavioral repertoires commonly employed by nociception that functions automatically without the need to attention (e.g., “fidgeting” while being seated on a chair to avoid “bedsores”).
We contend that the nociceptive automated repertoires remain essentially unexplored in pain research, mainly due to the confusion of nociception with pain. However, studies of modulation of nocifensive reflexes by stimulus position in egocentric coordinates, as well as by other factors such as stimulus movement (Wallwork, Talbot, Camfferman, Moseley, & Iannetti, 2016); environmental landscape (Sambo, Forster, Williams, & Iannetti, 2012); movement of environmental objects other than those eliciting the nocifensive reflex (Somervail et al., 2019), just to name a few, provide the best human evidence for the concept that the value of actions aiming to avoid contact between noxious stimuli and the body is dynamically adjusted to maximize survival and fitness (for an extensive discussion of the topic, see Bufacchi & Iannetti, 2018).
Pain and Learning
Part of the survival value of pain is its intimate association with learning. Pain induces learning even after single occurrences, the memory of which can last the rest of life. This property is taken advantage of in Pavlovian paradigms studying learning and memory, especially in fear conditioning, where the more painful the unconditioned stimulus is, the fewer trials it takes to establish an aversive emotional association to a conditioning stimulus that was originally affectively neutral (Schafe, Nader, Blair, & LeDoux, 2001). The ability to extinguish aversive associations of fearful or painful events with repeated exposure to the unconditioned stimulus is also important for normal behavior; impaired ability to extinguish is clinically relevant, and its mechanisms involve mesolimbic circuitry (Myers & Davis, 2002; Sotres-Bayon, Bush, & LeDoux, 2004).
Ten years ago, one of us advanced the hypothesis that chronic pain is a state of continuous learning in which aversive emotional associations are continuously made with incidental events simply due to the persistent presence of pain. Simultaneously, continued presence of pain does not provide an opportunity for extinction because whenever the subject is reexposed to the conditioned event, he or she is still in pain. Failing to extinguish, therefore, makes the event become a reinforcement of aversive association (Apkarian, 2008). It was proposed that living with this conundrum and the effort of disentangling its associations underlie at least some of the suffering of chronic pain.
If one regards chronic pain as the continuous presence of an unconditioned stimulus and as an inability to extinguish its associations with random events, then the brain circuitry underlying reward-/punishment-induced learning must be in a heightened state. Accumulating evidence now exists in both human studies and rodent experiments consistent with this notion, as reviewed in the material that follows.
Brain Responses to Acute Pain: What Do They Reflect?
One of the most reproducible results across laboratories and paradigms is that noxious experimental stimuli perceived as painful elicit fMRI responses in a large array of brain regions, including the primary (S1) and secondary (S2) somatosensory cortices, the insula, and the anterior cingulate cortex (ACC) (Apkarian, Bushnell, Treede, & Zubieta, 2005). The careless conflation of these fMRI responses with the neural substrates of pain has been a mistake that has caused tremendous confusion.1 This simplistic interpretation is not just a legacy of the past, limited to the early functional neuroimaging studies of pain claiming, for example, that these fMRI responses are “the neural substrates of pain” (Ploghaus et al., 1999). It remains surprisingly pervasive even in more recent literature; perhaps the best examples are the recent unjustified claims that the “the dorsal posterior insula subserves a fundamental role in human pain” (Segerdahl, Mezue, Okell, Farrar, & Tracey, 2015, p. 499), or that the dorsal anterior cingulate cortex is “selectively involved in pain-related processes” (Lieberman & Eisenberger, 2015, p. 15255).
But why was the rapid conclusion that these responses reflect the neural correlate of pain perception made? Besides the natural tendency of associating a perceptual process to statistical results visualized in fMRI scans (McCabe & Castel, 2008; but see also Farah & Hook, 2013), this conclusion was based on the observations that the fMRI responses elicited by nociceptive stimuli (a) are consistently observed in response to stimuli that individuals experience as acutely painful; (b) often (but not always, as discussed further in this chapter) their amplitude correlates with the amount of pain experienced by the subject; and (c) are modulated by the same factors that affect pain perception. On the basis of these facts, many studies either have claimed that they reflect the functioning of cortical areas devoted to pain elaboration or, even without making this claim explicitly, have used this assumption to interpret experimental results in a number of fields. These conclusions do not consider an important factor: the degree of exclusivity of the relationship between these responses and the state of experiencing pain (Iannetti & Mouraux, 2010; Poldrack, 2006).
As a matter of fact, fMRI responses largely similar to those elicited by a transient nociceptive stimulus can be elicited by nonpainful stimuli, provided that the eliciting stimuli are salient enough (Downar et al., 2003; Mouraux & Iannetti, 2009; Mouraux et al., 2011) (Figure 1). More recently, it was also shown that a virtually identical “pain matrix” response can be observed in patients with congenital insensitivity to pain (Salomons et al., 2016). Thus, while these fMRI responses are indeed observed in most situations where pain is perceived, they are also frequently observed in situations where pain is not present, providing strong evidence that the fMRI brain responses are largely nonspecific for pain.
But, what is it exactly meant by pain specificity of an fMRI response? An fMRI response is considered specific for pain when it is observed during pain, but never during all other nonpainful experiences. The issue of pain specificity of fMRI responses is particularly complicated given that pain is a multidimensional experience. Let us consider the International Association for the Study of Pain (IASP) definition of pain as “an unpleasant sensory and emotional experience associated with actual or potential tissue damage, or described in terms of such damage” (Merskey & Bogduk, 1994). Thus, pain is defined by the conjunction of several attributes: its intrinsic unpleasantness, its high saliency and ability to attract attention, and, because of its ability to signal threats for the body, its strong behavioral relevance. The key point is that stimuli causing sensations that are unpleasant, salient, and behaviorally relevant are not necessarily painful; a good example is the sound produced by scratching a blackboard with fingernails. Therefore, when assessing the specificity for pain of a given brain response, it is crucial to use nonpainful control stimuli that elicit sensations matched at least with respect to unpleasantness, salience, and behavioral relevance. These different aspects are strongly associated with pain, but they do not capture what distinguishes physical pain from other sensations, that is, the experience of actual or potential tissue damage (Merskey & Bogduk, 1994). If the responses to painful stimuli are not compared to nonpainful control stimuli that elicit sensations matched with respect to unpleasantness, salience, and relevance, it follows that it is impossible to determine whether the differences observed between the brain responses elicited by painful and nonpainful stimuli reflect neural activities that are selective for pain or neural activities that are selective for these other features.
Thus, the large similarity between the brain responses elicited to acute pain and the brain responses elicited by nonpainful tactile, auditory, or visual stimuli can be explained by the fact that these stimuli can overlap in one or more of the dimensions common to pain and these other sensory modalities. It follows that claims of pain selectivity cannot be made without considering these factors and testing the response selectivity using control stimuli matched with respect to these other dimensions. We have argued that the attempts to falsify the hypotheses that the brain responses elicited by nociceptive stimuli are selective for pain using appropriate control stimuli have been insufficient, and the liberal use of terms implying selectivity (or, worse, specificity) has biased the interpretation of several pain neuroimaging results (Mouraux & Iannetti, 2018).
For example, labeling as a “neurological pain signature” a spatial pattern of fMRI activity observed in different conditions entailing the experience of physical pain (Wager et al., 2013) because this pattern is not observed in a number of control conditions can be insufficient if these conditions are less salient, less behaviorally relevant, or not somatic. Examples are a low-salience, mild, warm stimulus versus a high-salience, burning, heat stimulus or a nonsomatic “social pain” stimulus versus a somatic painful stimulus. For this reason, it could well be that the spatial pattern of brain activity referred to as a neurological pain signature was, in fact, a spatial pattern of brain activity selective for the occurrence of a salient somatic stimulus, regardless of whether it elicits pain. Indeed, it has been observed that the same neurological pain signature fails to predict variations in pain induced by cognitive “self-regulation,” that is, by imagining that nociceptive stimuli are more painful or less painful than they are, thereby demonstrating that the Neurological Pain Signature (NPS) does not necessarily track the subjective pain experience (Woo, Roy, Buhle, & Wager, 2015).
A number of studies have been trying to carefully match painful and nonpainful stimuli with respect to their shared dimensions (e.g., stimulus saliency and stimulus intensity) in order to compare meaningfully the brain responses they elicit and tease out the differences more directly related to the painful quality of the perceptual experience. While the large part of the fMRI response identified at regional level using conventional mass-univariate analysis and general linear modeling is spatially similar (Liang, Su, Mouraux, & Iannetti, 2019; Mouraux, Diukova, Lee, Wise, & Iannetti, 2011; Su et al., 2019), a fine-grained analysis of the spatial distribution of the fMRI signal across the voxels composing each region multi-voxel pattern analysis (MVPA; e.g., Liang, Mouraux, Hu, & Iannetti, 2013) was able to successfully distinguish the responses elicited by the different stimuli, This indicated some differences in the fine-grained spatial distribution of the activity within the tested brain regions. Importantly, this was also the case when MVPA was performed using fMRI responses normalized across multiple voxels, demonstrating that the ability of MVPA to distinguish between the different types of stimuli was not due to mean differences in signal intensity. This demonstration of consistent differences in the spatial pattern of activation indicates that the fMRI responses elicited by transient nociceptive stimuli are functionally heterogeneous; they carry information related to both painfulness and stimulus intensity and saliency (Liang et al., 2019; Mouraux et al., 2011).
Finally, it is important to emphasize that testing whether a given spatial pattern of brain activity constitutes a “pain signature” requires exactly the same evidence that is needed to demonstrate the existence of “pain-specific” neurons or brain areas. In addition to showing that the identified spatial pattern is always present when one experiences pain, one must also show that the spatial pattern of activity is never present in the absence of pain.
Brain Responses to Chronic Pain: What Do They Reflect?
Early efforts to unravel the impact of chronic pain on brain activity explored activity patterns for acute, mostly thermal, stimuli in various pain patient groups in comparison to healthy controls (see Apkarian, 2005, for a review) (Figure 2). The underlying logic was the notion that the presence of pain would distort perception and representation for the added acute pain. Although many such studies were published, their interpretation was indirect and complicated and results often misinterpreted. The study by Baliki et al. (2006) demonstrated a double dissociation between brain activity to subjective ratings of chronic back pain (CBP) and acute thermal pain, applied in the same subjects, replicated in a new group of individuals with CBP on a different MRI scanner, as well as in healthy controls (Figure 2). Importantly, the study showed that although spontaneous back pain fluctuations are represented mainly in prefrontal and limbic brain regions, perceived magnitude for acute pain was localized in the insula, and the stimulus response in the insula was similar between healthy subjects and those with CBP. Subsequent studies demonstrated that distinct chronic pain conditions showed somewhat distinct brain activity patterns, engaging different components of prefrontal and limbic circuits (Apkarian, Baliki, & Geha, 2009).
Perhaps the most compelling evidence that the brain activity for pain shifts in brain representation is from a longitudinal study where subacute back pain (SBP) patients were followed in time to transition to either chronic pain or recovery (Hashmi et al., 2013). In these patients, brain activity for spontaneous fluctuations of back pain initially showed a pattern that corresponded to that seen for acute pain in healthy subjects. In the subjects for whom back pain diminished (over 1-year), related brain activity diminished to below noise and became undetectable. In contrast, for subjects whose pain persisted over the year, brain activity slowly shifted away from sensory regions to the prefrontal and limbic areas. Thus, the study directly showed reorganization of back pain representation, in the same subjects, with transition from acute to chronic pain engaging self-referential evaluation and emotion circuitry.
Simultaneously, accumulating evidence showed that brain gray matter regional density and volume, as well as across-region interactions, were distorted in chronic pain conditions and differently as a function of type of chronic pain (Baliki, Schnitzer, Bauer, & Apkarian, 2011). Moreover, brain white matter properties (anatomical connectivity across brain regions) were abnormal in chronic pain (Geha et al., 2008), and functional connectivity between resting state networks was also distorted (Baliki, Mansour, Baria, & Apkarian, 2014). Since these studies were cross-sectional in nature, they had an intrinsic problem: They could not disambiguate between causes and consequences of chronic pain.
Such disambiguation necessitated a longitudinal study (Baliki et al., 2012), where it was shown that brain gray matter properties and related functional connectivity changed in time in patients where pain persisted, thus identifying the consequence of the transition to chronic pain. The study also identified a brain functional connectivity that predicted who, in the future, would either recover from or persist with back pain, thus identifying a causal brain pathway (medial prefrontal cortex and nucleus accumbens) involved in the transition process. This result points to a role of a brain self-valuation and emotional motivation circuitry in maintaining chronic pain. A subsequent study (Vachon-Presseau et al., 2018) demonstrated a multiparameter limbic brain model that could explain about 60% of the variance of transition to chronic pain based on limbic brain properties: functional connectivity, white matter properties, and the size of the amygdala or the hippocampus.
With the advent of resting-state fMRI technology, it could be shown that functional network properties of the brain were distorted in chronic pain (Farmer, Baliki, & Apkarian, 2012). These studies culminated in showing that there was a global reorganization in functional information sharing that propagates throughout the brain and shifts the balance by decreasing information sharing for brain regions with high connectivity and increasing information sharing for regions with low connectivity (Mansour et al., 2016). The study could also demonstrate that this process was time dependent and emerged about 1 year after back pain persistence. Moreover, the same phenomenon could also be shown in rodents transitioning to chronic pain. Importantly, the effect size was proportional to the magnitude of chronic pain (in both chronic back pain and osteoarthritis) but was not affected by the presence of acute pain.
Biomarkers for Chronic Pain
Chronic pain and opiate addiction are intricately related, and their combination has had a huge impact on disability, healthcare costs, and overdose deaths. To combat these massive societal burdens, both the National Institutes of Health (NIH) and the Food and Drug Administration (FDA) recently launched large initiatives toward the development of novel tools and therapies for combating chronic pain and for diminishing reliance on opiates for managing chronic pain. This effort has emphasized the urgent need for developing validated biomarkers, with the assumption that such biomarkers facilitate the mechanistically driven discovery of novel therapies. Mechanisms and biomarkers are intricately interrelated, and advances in brain markers regarding causes and consequences of chronic pain lead to specific hypotheses regarding these relationships. It should be emphasized that given differences in brain anatomy and physiology between acute and chronic pain, biomarkers for acute pain do not generalize to chronic pain and in fact may be importantly misleading.
Figure 3 illustrates the general concept of different stages of chronic pain. The classic (peripheralist) viewpoint has been that all four stages are based on nociceptive afferent properties and their influence on spinal cord circuitry (central sensitization). Within this construct, quantitative sensory testing (QST) becomes the primary method with which one seeks biomarkers for chronic pain (Apkarian & Reckziegel, 2019). Additionally, skin biopsies, searching for anatomical evidence that would complement or confirm QST data, assumes that the anatomy of skin innervation defines and identifies chronic pain states. We presume that, although this approach may yield evidence of some correlates to chronic pain, these at best will be poor associations. If one is pursuing biomarkers for pain, be it acute or chronic, peripheral tissue properties are likely to be the wrong source of information. Peripheral tissue properties may provide information regarding abnormal or heightened nociceptive inputs to the nervous system, but cannot capture conscious subjective pain states that vary in time and context and are continuously modulated based on recent experiences and acquired memories.
Instead, the evidence regarding mechanisms of chronic pain suggests the following notions: (a) Pain is a subjective conscious perception that is distinct from nociception. It requires brain activity; thus, the main search for biomarkers must be an inquiry of underlying brain processes. (b) As similar injuries are commonly observed to result only in chronic pain in a small minority of subjects, and also in most of such conditions the actual injury seems to provide little explanatory value, we assume that the bulk of the risk for chronic pain is based on brain properties as a consequence of a lifetime accumulation of experiences (nurture) as well as genetic determinants (nature). (c) Even in chronic pain conditions where there may be a blatant nociceptive drive, persistence of pain may itself be sufficient to carve a new brain state with distinct interaction rules between nociceptive activity and the brain interpretation of this activity as pain. Therefore, over the last 20 years a large research effort in the laboratory of one of us has been directed toward searching for brain biomarkers contributing to the distinct stages of chronic pain.
This research can be briefly summarized regarding processes controlling the four stages in Figure 4: (1) Limbic–emotional circuitry defines predispositions; (2) Emotional learning mechanisms underlie and control the transitional stage (3), which is a consequence of the interaction between predispositions and injury-related nociceptive inputs to the nervous system. Moreover, maintenance or chronic pain (4) is a new brain state, with distinct anatomical and functional properties. Figure 4 establishes the types of biomarkers that one would associate with each of the four stages of chronic pain. Within this construct, properties of each kind of biomarker can be defined. For further details of expected properties of each type of biomarker and case examples, see Apkarian & Reckziegel (2019).
Overall, we propose a rationale and roadmap for biomarker development within a hypothesis-driven construct of mechanisms underlying chronic pain. Other domains that may in fact provide additional critical biomarker information would be genetic factors and electroencephalographic (EEG) features (e.g., Hu & Iannetti, 2019), as well as blood-borne cytokine and chemokines. The genetics of chronic pain remains a confusing topic, as many of the early gene associations have not been replicated in larger studies. Yet, undoubtedly genetic and epigenetic variations play a critical role, especially in prognostic biomarkers, and their influence needs to be uncovered. To our knowledge, no EEG biomarkers have been identified for chronic pain, although important efforts in this direction are underway (Schulz et al., 2015). Implementing EEG technology is becoming less costly and its analysis is becoming automated; thus, it could be readily used in a routine clinical setting if its utility can be demonstrated. Although not covered here, there is now evidence for a similar approach to developing prognostic biomarkers for placebo response in the clinical setting (Hashmi et al., 2012; Tetreault et al., 2016; Vachon-Presseau et al., 2018). It remains imperative to demonstrate that such markers in fact lead to more efficient, mechanism-driven, novel discoveries for treating the pain patient. Only such successes will change the clinical and research course of the field.
This overview charts novel concepts and critical issues that have arisen from fMRI studies of acute and chronic pain. The chapter emphasizes definitions and related brain mechanisms that have been recently observed, as well as important gaps, challenging the research community to critically address these issues. We conclude that the ongoing research has certainly expanded the realm of questions being tackled, but it has also demonstrated the limits within which current neuroscience can explore perception, especially pain perception. However, within these limits, there are also strong indications that understanding mechanisms—be they pathways, networks, synapses, or molecules that are engaged in chronic pain—are all critical on the road to better management and prevention of such conditions. In this direction, better understanding of the dimensions that make up the dimensions of biomarkers for chronic pain promises to impact the field from both basic science and clinical management viewpoints.
GDI acknowledges the support of The Wellcome Trust (COLL JLARAXR) and the European Research Council (Consolidator grant PAINSTRAT).
Apkarian, A. V. (2005). Retrieved from: https://www.ncbi.nlm.nih.gov/pubmed/15979027
Apkarian, A. V. (2008). Pain perception in relation to emotional learning. Current Opinion in Neurobiology, 18(4), 464–468. doi:10.1016/j.conb.2008.09.012Find this resource:
Apkarian, A. V. (2019). Definitions of nociception, pain, and chronic pain with implications regarding science and society. Neuroscience Letters, 702, 1–2. doi:10.1016/j.neulet.2018.11.039Find this resource:
Apkarian, A. V., Baliki, M. N., & Geha, P. Y. (2009). Towards a theory of chronic pain. Progress in Neurobiology, 87(2), 81–97. doi:10.1016/j.pneurobio.2008.09.018Find this resource:
Apkarian, A. V., Bushnell, M. C., Treede, R. D., & Zubieta, J. K. (2005). Human brain mechanisms of pain perception and regulation in health and disease. European Journal of Pain, 9(4), 463–484. doi:10.1016/j.ejpain.2004.11.001Find this resource:
Apkarian, A. V., & Reckziegel, D. (2019). Peripheral and central viewpoints of chronic pain, and translational implications. Neuroscience Letters, 702, 3–5. doi:10.1016/j.neulet.2018.11.040Find this resource:
Azevedo, F. A., Carvalho, L. R., Grinberg, L. T., Farfel, J. M., Ferretti, R. E., Leite, R. E., . . . Herculano-Houzel, S. (2009). Equal numbers of neuronal and nonneuronal cells make the human brain an isometrically scaled-up primate brain. Journal of Comparative Neurology, 513(5), 532–541. doi:10.1002/cne.21974Find this resource:
Baliki et al. (2006). Retrieved from https://www.ncbi.nlm.nih.gov/pubmed/17122041
Baliki, M. N., Mansour, A. R., Baria, A. T., & Apkarian, A. V. (2014). Functional reorganization of the default mode network across chronic pain conditions. PLoS One, 9(9), e106133. doi:10.1371/journal.pone.0106133Find this resource:
Baliki, M. N., Petre, B., Torbey, S., Herrmann, K. M., Huang, L., Schnitzer, T. J., . . . Apkarian, A. V. (2012). Corticostriatal functional connectivity predicts transition to chronic back pain. Nature Neuroscience, 15(8), 1117–1119. doi:10.1038/nn.3153Find this resource:
Baliki, M. N., Schnitzer, T. J., Bauer, W. R., & Apkarian, A. V. (2011). Brain morphological signatures for chronic pain. PLoS One, 6(10), e26010. doi:10.1371/journal.pone.0026010Find this resource:
Bassett, D. S., & Sporns, O. (2017). Network neuroscience. Nature Neuroscience, 20(3), 353–364. doi:10.1038/nn.4502Find this resource:
Bufacchi, R. J., & Iannetti, G. D. (2018). An action field theory of peripersonal space. Trends in Cognitive Science, 22(12), 1076–1090. doi:10.1016/j.tics.2018.09.004Find this resource:
Carmon, A., Mor, J., & Goldberg, J. (1976). Evoked cerebral responses to noxious thermal stimuli in humans. Experimental Brain Research, 25(1), 103–107.Find this resource:
Chapman, C. R., Chen, A. C., Colpitts, Y. M., & Martin, R. W. (1981). Sensory decision theory describes evoked potentials in pain discrimination. Psychophysiology, 18(2), 114–120.Find this resource:
Chapman, C. R., Colpitts, Y. H., Mayeno, J. K., & Gagliardi, G. J. (1981). Rate of stimulus repetition changes evoked potential amplitude: Dental and auditory modalities compared. Experimental Brain Research, 43(3–4), 246–252.Find this resource:
Downar et al., 2003. Retrieved from http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=14642466.
Farah, M. J., & Hook, C. J. (2013). The seductive allure of “seductive allure.” Perspectives on Psychological Science, 8(1), 88–90. doi:10.1177/1745691612469035Find this resource:
Farmer, M. A., Baliki, M. N., & Apkarian, A. V. (2012). A dynamic network perspective of chronic pain. Neuroscience Letters, 520(2), 197–203. doi:10.1016/j.neulet.2012.05.001Find this resource:
Geha, P. Y., Baliki, M. N., Harden, R. N., Bauer, W. R., Parrish, T. B., & Apkarian, A. V. (2008). The brain in chronic CRPS pain: Abnormal gray-white matter interactions in emotional and autonomic regions. Neuron, 60(4), 570–581. doi:10.1016/j.neuron.2008.08.022Find this resource:
Hashmi, J. A., Baliki, M. N., Huang, L., Baria, A. T., Torbey, S., Hermann, K. M., . . . Apkarian, A. V. (2013). Shape shifting pain: Chronification of back pain shifts brain representation from nociceptive to emotional circuits. Brain, 136(Pt. 9), 2751–2768. doi:10.1093/brain/awt211Find this resource:
Hashmi, J. A., Baria, A. T., Baliki, M. N., Huang, L., Schnitzer, T. J., & Apkarian, A. V. (2012). Brain networks predicting placebo analgesia in a clinical trial for chronic back pain. Pain, 153(12), 2393–2402. doi:10.1016/j.pain.2012.08.008Find this resource:
Hu, L., & Iannetti, G. D. (2019). Neural indicators of perceptual variability of pain across species. Proceedings of the National Academy of Sciences of the United States of America. 116(5), 201812499. doi:10.1073/pnas.1812499116Find this resource:
Iannetti, G. D., & Mouraux, A. (2010). From the neuromatrix to the pain matrix (and back). Experimental Brain Research, 205(1), 1–12. doi:10.1007/s00221-010-2340-1Find this resource:
Iannetti, G. D., & Wise, R. G. (2007). BOLD functional MRI in disease and pharmacological studies: Room for improvement? Magnetic Resonance Imaging, 25(6), 978–988. doi:10.1016/j.mri.2007.03.018Find this resource:
Liang, M., Mouraux, A., Hu, L., & Iannetti, G. D. (2013). Primary sensory cortices contain distinguishable spatial patterns of activity for each sense. Nature Communications, 4, 1979. doi:10.1038/ncomms2979Find this resource:
Liang, M., Su, Q., Mouraux, A., & Iannetti, G. D. (2019). Spatial patterns of brain activity preferentially reflecting transient pain and stimulus intensity. Cerebral Cortex, 29(5), 2211–2227. doi:10.1093/cercor/bhz026Find this resource:
Lieberman, M. D., & Eisenberger, N. I. (2015). The dorsal anterior cingulate cortex is selective for pain: Results from large-scale reverse inference. Proceedings of the National Academy of Sciences, 112(49), 15250–15255. doi:10.1073/pnas.1515083112Find this resource:
Logothetis, N. K., Pauls, J., Augath, M., Trinath, T., & Oeltermann, A. (2001). Neurophysiological investigation of the basis of the fMRI signal. Nature, 412(6843), 150–157. doi:10.1038/35084005Find this resource:
Mansour, A., Baria, A. T., Tetreault, P., Vachon-Presseau, E., Chang, P. C., Huang, L., . . . Baliki, M. N. (2016). Global disruption of degree rank order: A hallmark of chronic pain. Scientific Reports, 6, 34853. doi:10.1038/srep34853Find this resource:
McCabe, D. P., & Castel, A. D. (2008). Seeing is believing: The effect of brain images on judgments of scientific reasoning. Cognition, 107(1), 343–352. doi:10.1016/j.cognition.2007.07.017Find this resource:
Melzack, R., & Wall, P. D. (1996). The Challenge of Pain (2nd ed.). London, England: Penguin.Find this resource:
Merskey, H., & Bogduk, N. (1994). Pain terms, a current list with definitions and notes on usage. IASP Task Force on Taxonomy. In H. Merskey & N. Bobduk (Eds.), Classification of Chronic Pain (pp. 209–214). Seattle, WA: IASP Press.Find this resource:
Mouraux, A., Diukova, A., Lee, M. C., Wise, R. G., & Iannetti, G. D. (2011). A multisensory investigation of the functional significance of the “pain matrix.” Neuroimage, 54(3), 2237–2249. doi:10.1016/j.neuroimage.2010.09.084Find this resource:
Mouraux & Iannetti (2009). Retrieved from: http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=19339457
Mouraux & Iannetti (2018). Retrieved from: https://www.ncbi.nlm.nih.gov/pubmed/30462175
Myers, K. M., & Davis, M. (2002). Behavioral and neural analysis of extinction. Neuron, 36(4), 567–584.Find this resource:
Passingham, R. E., Stephan, K. E., & Kotter, R. (2002). The anatomical basis of functional localization in the cortex. Nature Reviews Neuroscience, 3(8), 606–616. doi:10.1038/nrn893Find this resource:
Ploghaus, A., Tracey, I., Gati, J. S., Clare, S., Menon, R. S., Matthews, P. M., & Rawlins, J. N. (1999). Dissociating pain from its anticipation in the human brain. Science, 284(5422), 1979–1981.Find this resource:
Poldrack, R. A. (2006). Can cognitive processes be inferred from neuroimaging data? Trends in Cognitive Science, 10(2), 59–63. doi:10.1016/j.tics.2005.12.004Find this resource:
Salomons, T. V., Iannetti, G. D., Liang, M., & Wood, J. N. (2016). The ‘Pain Matrix’ in Pain-Free Individuals. JAMA Neurology, 73(6):755–756. doi:10.1001/jamaneurol.2016.0653. https://www.ncbi.nlm.nih.gov/pubmed/27111250.Find this resource:
Sambo, C. F., Forster, B., Williams, S. C., & Iannetti, G. D. (2012). To blink or not to blink: Fine cognitive tuning of the defensive peripersonal space. Journal of Neuroscience, 32(37), 12921–12927. doi:10.1523/JNEUROSCI.0607-12.2012Find this resource:
Schafe, G. E., Nader, K., Blair, H. T., & LeDoux, J. E. (2001). Memory consolidation of Pavlovian fear conditioning: a cellular and molecular perspective. Trends in Neurosciences, 24(9), 540–546.Find this resource:
Schaffer, E. S., Stettler, D. D., Kato, D., Choi, G. B., Axel, R., & Abbott, L. F. (2018). Odor perception on the two sides of the brain: Consistency despite randomness. Neuron, 98(4), 736–742.e733. doi:10.1016/j.neuron.2018.04.004Find this resource:
Schulz, E., May, E. S., Postorino, M., Tiemann, L., Nickel, M. M., Witkovsky, V., . . . Ploner, M. (2015). Prefrontal gamma oscillations encode tonic pain in humans. Cerebral Cortex, 25(11), 4407–4414. doi:10.1093/cercor/bhv043Find this resource:
Segerdahl, A. R., Mezue, M., Okell, T. W., Farrar, J. T., & Tracey, I. (2015). The dorsal posterior insula subserves a fundamental role in human pain. Nature Neuroscience, 18(4), 499–500. doi:10.1038/nn.3969Find this resource:
Somervail, R., Bufacchi, R. J., Guo, Y., Kilintari, M., Novembre, G., Swapp, D., . . . Iannetti, G. D. (2019). Movement of environmental threats modifies the relevance of the defensive eye-blink in a spatially-tuned manner. Scientific Reports, 9(1), 3661. doi:10.1038/s41598-019-40075-xFind this resource:
Sotres-Bayon, F., Bush, D. E., & LeDoux, J. E. (2004). Emotional perseveration: An update on prefrontal-amygdala interactions in fear extinction. Learning and Memory, 11(5), 525–535. doi:10.1101/lm.79504Find this resource:
Su, Q., Qin, W., Yang, Q. Q., Yu, C. S., Qian, T. Y., Mouraux, A., . . . Liang, M. (2019). Brain regions preferentially responding to transient and iso-intense painful or tactile stimuli. Neuroimage, 192, 52–65. doi:10.1016/j.neuroimage.2019.01.039Find this resource:
Tetreault, P., Mansour, A., Vachon-Presseau, E., Schnitzer, T. J., Apkarian, A. V., & Baliki, M. N. (2016). Brain connectivity predicts placebo response across chronic pain clinical trials. PLoS Biology, 14(10), e1002570. doi:10.1371/journal.pbio.1002570Find this resource:
Vachon-Presseau, E., Berger, S. E., Abdullah, T. B., Huang, L., Cecchi, G. A., Griffith, J. W., . . . Apkarian, A. V. (2018). Brain and psychological determinants of placebo pill response in chronic pain patients. Nature Communications, 9(1), 3397. doi:10.1038/s41467-018-05859-1Find this resource:
Wager, T. D., Atlas, L. Y., Lindquist, M. A., Roy, M., Woo, C. W., & Kross, E. (2013). An fMRI-based neurologic signature of physical pain. New England Journal of Medicine, 368(15), 1388–1397. doi:10.1056/NEJMoa1204471Find this resource:
Wallwork, S. B., Talbot, K., Camfferman, D., Moseley, G. L., & Iannetti, G. D. (2016). The blink reflex magnitude is continuously adjusted according to both current and predicted stimulus position with respect to the face. Cortex, 81, 168–175. doi:10.1016/j.cortex.2016.04.009Find this resource:
Woo, C. W., Roy, M., Buhle, J. T., & Wager, T. D. (2015). Distinct brain systems mediate the effects of nociceptive input and self-regulation on pain. PLoS Biology, 13(1), e1002036. doi:10.1371/journal.pbio.1002036Find this resource:
Zimmermann, J., Griffiths, J. D., & McIntosh, A. R. (2018). Unique mapping of structural and functional connectivity on cognition. Journal of Neuroscience, 38(45), 9658–9667. doi:10.1523/jneurosci.0900-18.2018Find this resource:
(1.) It is instructive to remind the reader that when interpreting the functional significance of the EEG brain responses elicited by nociceptive stimuli, initial reports were extremely cautious, taking into consideration the notion that pain perception is only one of the possible outcomes of the cortical activation elicited by the incoming nociceptive input. For example, Carmon, Mor, and Goldberg (1976) stated “it is possible that only the arousing and alerting effect of pain is responsible” for the elicited brain responses, and Chapman, Chen, Colpitts, and Martin (1981), and Chapman, Colpitts, Mayeno, and Gagliardi (1981, 250) wrote that these responses “cannot be considered neurophysiological representations of pain sensations.”