Neuroanatomy of the nociceptive system: From nociceptors to brain networks

Abstract

This chapter reviews the neuroanatomy of the nociceptive system and its functional organization. We describe three main compartments of the nervous system that underlie normal nociception and the resulting pain percept: Peripheral, Spinal Cord, and Brain. We focus on how ascending nociceptive processing streams traverse these anatomical compartments, culminating in the multidimensional experience of pain. We also describe neuropathic pain conditions, in which nociceptive processing is abnormal, not only because of the primary effects of a lesion or disease affecting peripheral nerves or the central nervous system (CNS), but also due to secondary effects on ascending pathways and brain networks. We discuss how the anatomical components (circuits/networks) reorganize under various etiologies of neuropathic pain and how these changes can give rise to pathological pain states.

1. Introduction

The International Association for the Study of Pain (IASP) defines pain as “an unpleasant sensory and emotional experience associated with, or resembling that associated with, actual or potential tissue damage” (Raja et al., 2020). Under normal conditions, noxious stimulation applied to the periphery evokes a cascade of nociceptive signaling in ascending somatosensory pathways, which culminates in a distributed multidimensional brain representation that is required to localize, interpret, and appropriately respond to the stimulus. From the periphery to the brain, the nociceptive somatosensory system exhibits a remarkably complex hierarchical organizational structure that allows each tissue compartment to influence processing at other levels. This complex organization contributes to significant interindividual and contextual variability in the relationship between the magnitude of the provoking stimulus and the pain reported.

Neuropathic pain is defined as “pain arising as a direct consequence of a lesion or disease affecting the somatosensory nervous system” (Treede et al., 2008). In response to nerve injury or disease, extensive plastic changes throughout the somatosensory system emerge that can profoundly disrupt the resulting pain experience and contribute to the instantiation of chronic pain (Kuner & Flor, 2016). In some cases, the absence of brain EEG responses to noxious stimulation of the periphery is confirmed in patients with neuropathic pain, indicating direct interruption of the nociceptive circuitry (Rosner et al., 2023). However, based on neuroanatomy, uncertainty is common in diagnostics (Finnerup et al., 2016). For example, the painful area reported by individuals can spatially grow or shrink over time, reflecting the extent to which the nociceptive system changes functionally and structurally after injury.

Of particular importance is the demonstration that various etiologies of chronic pain, including tissue injury-induced inflammatory and nerve injury-induced neuropathic pain, involve long-lasting sensitization of circuits at the level of the spinal cord and the brain, which contributes to the ongoing pain and hypersensitivity that characterize most clinical pain conditions (Basbaum et al., 2009; Woolf, 2011). Importantly, although the location and type of damage significantly affect the symptoms and clinical trajectory of chronic neuropathic pain, there are shared features across conditions involving convergent dysregulation of brain networks that may offer viable new targets for therapeutic development.

In this chapter, we review the neuroanatomy and function of three main anatomical compartments (Peripheral, Spinal Cord, and Brain) that underlie normal nociceptive processing (Fig. 1). We describe how the cascade of nociceptive processing through these regions culminates in a pain percept. We also underscore how the connections between compartments dynamically influence nociceptive processing. Lastly, we review alterations in these compartments that are often associated with, and may give rise to, chronic neuropathic pain.

Fig. 1. Three main anatomical compartments (Peripheral, Spinal Cord, and Brain) underlie normal nociceptive processing.

Nociceptors detect noxious stimuli through their free nerve endings and transmit these signals to the CNS, specifically the dorsal horn of the spinal cord. Dorsal horn neuronal circuits process the nociceptive input and transmit ascending nociceptive signals to the brain. Ascending nociceptive information engages complex circuits for top-down neuromodulation and is distributed across brain networks, resulting in the multidimensional experience of pain. Created with BioRender.com.

2. Peripheral: nociception

Nociception, the sensory process that detects harmful stimuli, alerts and protects the organism from danger through immediate behavioral responses. Specific classes of primary sensory neurons, the “nociceptors”, are specialized to receive the noxious environmental information capable of producing injury and generating pain. Nociceptor cell bodies reside in the periphery, and their extensive processes cover the body of mammals, including the skin and viscera. Interestingly, invertebrates have a similar bodily coverage system, indicating an early phylogenetic start to this noxious-sensing sensory system (Burrell, 2017). Through the constant survey of non-neural tissue, peripheral nociceptors are poised to detect changes induced by the environment or internal aberrations, transforming the environmental signals into neural activity.

At a low level, the organization of the somatosensory system is evident in the dermatomal patterns of the body, which reflect skin or visceral areas innervated by sensory neurons from a specific ganglion and which target specific spinal segments. In somatic areas besides the head, the ganglia containing sensory neurons are in the dorsal root ganglia (DRG), and each ganglion forms a corresponding dorsal root, which heads toward the dorsal horn of the spinal cord. The pseudo-unipolar morphology of DRG neurons effectively creates a straight “afferent” wire between the periphery and central neural circuits of the spinal cord. An analogous circuit exists for the transmission of noxious stimuli from the head and face. In this case, primary afferent cell bodies are in the trigeminal ganglion, which targets the brainstem trigeminal nuclei. The nucleus caudalis of the spinal trigeminal nucleus is the homolog of the spinal cord dorsal horn and processes nociceptive information from the head and face.

Of course, thermal, mechanical, and chemical qualities of stimuli encountered in the environment are psychophysically dissociable. The detection of modality starts at the primary sensory neurons through receptor proteins (Julius & Basbaum, 2001). For example, dissociated sensory neurons in a dish maintain responsiveness by generating action potentials to specific physical entities (temperatures, chemicals, etc.). And TRPV1 afferents, but not those neurons without this receptor, respond to the spicy chemical capsaicin.

The most remarkable structural differences in peripheral sensory neurons clustered together in the DRG are cell body size and myelination status. C-fibers are not myelinated, which is quite a distinction as the information from the A-delta myelinated class of nociceptors gets to the spinal cord faster. This temporal difference results in a first and second pain experience (Price & Dubner, 1977). This dissociation is further evident when C-fiber nociceptors are selectively stimulated in an awake human and evoke distinct verbal descriptions of the experience, such as burning pain (Ochoa & Torebjörk, 1989). Indeed, most nociceptors are C-fibers with free nerve endings in the skin. Yet, free nerve endings per se can serve different functions or evoke different behaviors (i.e., pruriceptors that drive scratching). From an evolutionary perspective, asking why “pain” processing afferents have the slowest conduction velocities is interesting.

Generally, the composition of DRG cell types is uniform along the body axis, including nociceptors, and though there are notable differences, mice and humans have similar cell types (Ray et al., 2018; Tavares-Ferreira et al., 2022; Yang et al., 2022). Elaboration of cell-type specific morphology, molecular composition, and function have recently benefited from single-cell transcriptomics and genetic manipulation of specific cell types (Sharma et al., 2020). Under these evolving “omic” classifications, a nomenclature of all DRG cell types by function is anticipated. However, these transcriptomic-based classifications must reconcile with the specific classes of nociceptors and their function that have been studied for quite some time based on physiology and protein-based neurochemistry (Emery & Ernfors, 2018; Kupari & Ernfors, 2023). A detailed proteomic analysis that correlates with existing transcriptomic studies is also essential (Rouwette et al., 2016).

Lastly, a single primary sensory neuron can activate many spinal neurons, depending on their spinal innervation patterns (Qi et al., 2024). Of course, when spinal neural circuits become involved, the complexity expands, and the quality of nociception can dramatically change.

3. Spinal: nociceptive processing

Dynamic, polysynaptic processing of nociception arises in the spinal cord dorsal horn, where peripheral information immediately gets entangled in complex neuroanatomy. Amazingly, the dorsal horn’s laminated anatomy provides defined boundaries between high-threshold nociceptive stimuli and low-threshold, innocuous tactile stimuli (Abraira et al., 2017). However, the neurons populating these laminae interact extensively, and numerous spinal microcircuits support acute nociception (Peirs & Seal, 2016). Structurally, different dendritic and axonal morphologies of spinal neurons subserve all the connections of microcircuits (Todd, 2010). The selective engagement of specific neuronal types is critical, of course. Or, at least, there are examples where more complex behaviors can be controlled, such as the cellular/circuit discrimination of itch from pain (Braz et al., 2014; Dong & Dong, 2018).

Similar to the classification of sensory neurons, developmental and genetic characterization of spinal cord cell types has recently benefited from unbiased transcriptomic profiling in the mouse and in humans (Russ et al., 2021; Yadav et al., 2023). Besides the excitatory vs. inhibitory classes, the most prominent characteristic is whether a spinal neuron is an interneuron or a brain-projecting neuron (Osseward et al., 2021). The well-known gate control theory of pain explains how primary sensory neurons and dorsal horn circuits in the spinal cord interact to either block or transmit pain signals. This modulation occurs through inhibition by myelinated afferent connections with inhibitory interneurons or by enhancing nociceptive processing via disinhibitory circuits (Melzack and Wall, 1965). Indeed, inhibitory interneurons are critical to all microcircuit logic in the dorsal horn. However, by way of being the output, the extent to which a particular stimulus is perceived as painful depends on the brain’s interpretation of activity generated in the spinal dorsal horn projection neurons (Braz et al., 2014). Thus, as the gate control theory emphasizes, the output of the dorsal horn projection represents dorsal horn processing of nociceptive signals, either inhibited (gate closed) or amplified (gate open).

Multiple ascending pathways (spinothalamic, spinoreticular, spinoparabrachial, and spinomesencephalic) engage their target supraspinal structures, further expanding nociceptive signaling. In fact, despite being <5 % of all dorsal horn neurons in the mouse (Polgár et al., 2010), anterolateral tract (spinothalamic) activity is sufficient for the brain to generate multidimensional percepts of pain, temperature, and itch (Mayer et al., 1975). However, the concept of a labeled line or a specific threshold causing pain continues to be questioned, and a polymodal encoding problem, in the awake state, is a frontier in research. Recent ribosomal profiling (Wercberger et al., 2021) and calcium imaging studies (Ahanonu et al., 2023) from our laboratory favor the view that polymodality of superficial dorsal horn projection neurons predominates.

Molecular subgroupings of spinal projection neurons provide evidence for functional divergence within these ascending pathways (Barik et al., 2021; Wercberger et al., 2021), which can partly be explained based on different supraspinal targets (Chen et al., 2024). Notably, regarding laminar location, ‘pain’ relevant projection neurons are distributed throughout the superficial and deep dorsal horn and even in adjacent white matter (lateral spinal nucleus, LSN). Unlike Lamina I, which is mainly innervated by nociceptors, the deep dorsal horn is also innervated by Aβ-LTMR, low threshold mechanoreceptive afferents that directly connect with deep Lamina V projection neurons (Kókai et al., 2022). Direct LTMR connections contribute to the relatively homogenous ‘wide dynamic range’ (WDR) response profile of deep projection neurons, responding to innocuous mechanical stimuli but also more strongly to noxious heat or pinch (Palecek et al., 1992). Disinhibition at the spinal level of either of these projection neuron populations could underpin conditions like allodynia; in the dynamic mechanical form, patients perceive a light brush stroke stimulus as painful (Jensen & Finnerup, 2014; Samuelsson et al., 2011).

Lastly, spinal circuits are critical for processing all somatosensory input from the periphery. These circuits support the organism’s ability to distinguish, manage, or override inputs from nociceptors, also known as antinociception. But these circuits are subject to powerful descending control that can alter nociceptive processing in states as varied as playing sports and being hungry (Alhadeff et al., 2018). Beyond an immediate spinal cord-mediated reflex (i.e., accidentally touching a hot stovetop), the feeling of pain persists. And without question, nociception can be dramatically different if one is already in pain (Yarnitsky, 2010). Profound modulation of all dorsal horn circuits is mediated by neurotransmitters, namely serotonin (5-HT), norepinephrine (NE), and endorphins, originating in the brainstem and, as yet, less well-characterized pathways from the cerebral cortex (Sirucek et al., 2023).

4. Brain: nociceptive pathway expansion and generation of the pain percept

As ascending peripheral and spinal nociceptive pathways are engaged, wherein does nociception (i.e., the encoding and transmission of noxious peripheral stimuli) take on the characteristic unpleasant emotional valence and motivational drive that transforms it to pain?

Selective pathways for sensory and emotional, or affective, pain processing begin to diverge in the spinal cord (e.g., into spinothalamic vs. spinoparabrachial, spinoreticular, and spinomesencephalic tracts) and are further established in the brainstem (Fig. 2). Whereas information about pain location, quality, and intensity is transmitted via the ventral thalamus to the primary and secondary somatosensory cortex and posterior insular cortex, information about pain’s inherent unpleasantness is transmitted through multiple pathways, primarily via the parabrachial nucleus and medial thalamus to limbic structures, including the amygdala, ventral striatum, anterior insular cortex, and cingulate cortex. These limbic structures together are thought to contribute to immediate, or early, pain unpleasantness. Collateral fibers from these ascending pathways impinge on several brainstem nuclei involved in arousal and autonomic functions, further influencing the aversiveness of pain and preparing the organism to mount an appropriate behavioral response (typically escape/avoidance).

Fig. 2. Sensory and affective pain pathways, cognitive-evaluative circuits, modulatory circuits and cortical networks.

This schematic provides an overarching summary of pain pathways and circuits. Nociceptive information from superficial and deep dorsal horn projection neurons (“Cord”) ascends through lateral “sensory” and medial “affective” pathways en route to brainstem and thalamic relays to specific cortical and subcortical targets. Sensory-discriminative circuits (blue-purple hues, top left) largely originate from deep laminae and ascend through ventral thalamic nuclei (VPM, VPL) to SI. Projections from superficial dorsal horn laminae to VPI thalamus provide a direct route to SII. Both SI and SII project to PIC. Affective-motivational circuits (yellow-orange hues, top right) originate primarily from superficial laminae and traverse the parabrachial nucleus and medial thalamic nuclei (MD, CMpf, VMpo) en route to AIC, ACC, amygdala, and ventral striatum/nucleus accumbens. AIC receives inputs primarily from VMpo, whereas ACC receives inputs from MD and CMpf. Parabrachial nucleus sends overlapping projections to MD, AIC, and ACC, among other regions not shown. Connections between regions and between sensory and affective circuits are extensive, but omitted here for clarity. Cognitive-evaluative circuits (green hues, middle left) are not pain-specific but include regions like the medial and lateral prefrontal cortex, orbitofrontal cortex, and lateral parietal cortex, as well as affective-motivational regions like ACC and AIC. Modulatory circuits (red, middle right) include most affective-motivational and cognitive-evaluative regions and other areas, like M1, with documented projections to opioidergic and monoaminergic brainstem nuclei that directly modulate incoming nociceptive signals in the dorsal horn. Key brainstem nuclei from the text are shown, but connections between regions, which are extensive and reciprocal, are omitted for clarity. Of note, the RVM includes the nucleus raphe magnus (NRM) and adjacent reticular formation. A subset of four canonical resting state networks (RSNs) linked with pain processing are shown in the gray box (bottom). Other RSNs that may be pain relevant, like the limbic network and ventral attention network, are omitted. One can appreciate that these RSNs consist of different combinations of regions from pain relevant circuits. See text for a more detailed description of region roles and connections. Created with BioRender.com. Abbreviations: DMN, default mode network; LC, locus coeruleus; PAG, periaqueductal gray; RVM, rostral ventral medulla; all others in text.

As these parallel ascending projections reach the brain, they form extensive reciprocal connections with a distributed network of interconnected brain regions involved in pain processing; the so-called Pain Matrix (Garcia-Larrea & Peyron, 2013; Iannetti & Mouraux, 2010; Melzack, 1999). These regions, in turn, project to higher order association cortex regions involved in appraising meanings and future implications of pain, which contribute to a ‘secondary’, or delayed, pain unpleasantness (Fields, 1999; Price, 2000). This cascade of processing ultimately culminates in the holistic experience of ‘pain’, distinct from nociception, and in the selection of a contextually appropriate response, which can range from evolving internal ‘feeling’ states to overt escape behaviors.

The specific organization of these pathways and the nature of their connections are a subject of ongoing research and debate. Preliminary circuit models of pain processing conceptualized central sensory and affective processing functions as occurring in series (Melzack & Casey, 1968). In other words, initial somatosensory computations about pain location and intensity are fed ‘forward’ to circuits that generate affective responses and behavioral reactions to pain. Although this simplistic feedforward hierarchical model of processing is at least partially true, the divergence of intensity and unpleasantness ratings under certain experimental conditions (e.g., during prolonged or tonic pain), and their independent modulation by certain experimental manipulations or targeted brain lesions, argues for a more complex organizational framework, in which extensive reciprocal connections influence local processing at each level of the hierarchy (Garcia-Larrea & Peyron, 2013; Motzkin, Kanungo, et al., 2023). Before turning to emerging models of pain processing that highlight the contributions of distributed networks to nociceptive processing and pain perception, we first review central pain circuits involved in somatosensory, affective, and evaluative processing.

4.1. Somatosensory pain circuits

A critical function of central pain processing is to accurately localize and identify noxious stimuli, which may be considered “cortical nociception” (Garcia-Larrea & Peyron, 2013). Such a ‘sensory-discriminative’ nociceptive processing circuit receives the bulk of its input from ascending spinothalamic tract projections, via the so-called ‘lateral pain pathway’ through the ventral thalamus to defined cortical targets. This pathway maintains a conserved somatotopic organization at each level of the hierarchy. In other words, neurons sampling distinct dermatomes (e.g., sacral, lumbar, thoracic, cervical, and facial) are organized with respect to their body site of origin. Fibers from the ventroposterolateral (VPL) thalamus, which receive spinothalamic tract inputs from the body, and ventroposteromedial (VPM) thalamus, which receive inputs from the face and head, originate primarily, though not exclusively, in deeper laminae IV and V of the dorsal horn and nucleus caudalis, respectively. VPL and VPM thalamic neurons in turn project through the posterior limb of the internal capsule toward the postcentral gyrus of the parietal lobe, the location of the primary somatosensory cortex (SI), where specific computations regarding stimulus location and intensity occur.

In general, VPL/VPM and SI neurons have narrow somatotopic receptive fields that reflect the anatomical origin of corresponding inputs, although some neurons with broad, bilateral receptive fields have been found in SI (Kenshalo & Isensee, 1983; Lamour et al., 1983). Activity in individual neurons throughout this circuit, and region-level activations in VPL/VPM and SI in human brain imaging studies, increase as a function of stimulus intensity (Coghill et al., 1999, 2003; Kenshalo & Isensee, 1983; Kenshalo et al., 1980; Peyron et al., 1999; Vierck et al., 2013). These computational features (i.e., somatotopic spatial organization and intensity-related activity) are well-suited to encode the “where?” and “how much?” of a stimulus. There is also evidence for modality-specificity (i.e., “what?”) in SI, such that qualitatively different sensations evoked from the same body region (e.g., vibration, pinprick, pressure) exhibit distinct cortical representations, although this is less well understood (Friedman et al., 2004). Of note, lesions of SI in humans can impair the ability to detect and discriminate noxious stimuli, but leave intact affective judgments regarding their unpleasantness (Ploner et al., 1999). On the other hand, electrical stimulation of VPL/VPM and SI in humans typically induces non-noxious and, at times, pleasant sensory experiences (Penfield & Boldrey, 1937); indeed, pain has never been reported with stimulation in either area. These observations highlight the intermingling of innocuous somatosensory inputs from the dorsal column medial lemniscus pathway with ascending nociceptive inputs, at least at lower levels of the sensory processing hierarchy.

The secondary somatosensory cortex (SII), located between SI and the posterior insular cortex (PIC) in the parietal operculum, is a higher order somatosensory processing region that shares several notable features with VPL/VPM and SI, including somatotopic responses that correlate with stimulus intensity. SII receives processed somatosensory information from SI, as well as separate direct inputs via the ventroposterior inferior (VPI) thalamus subjacent to VPM/VPL, which itself receives inputs primarily from Lamina I spinothalamic neurons of the dorsal horn (Apkarian & Hodge, 1989; Ralston & Ralston, 1992). SII is an association cortex with multiple distinct functional subregions that receive additional inputs from the hippocampus and parietal lobe, contributing to slightly larger and more complex receptive fields than neurons in SI (Eickhoff et al., 2006). Most notably, SII and adjacent PIC are the only brain regions where stimulation has been found to elicit painful sensations in humans (Mazzola et al., 2006, 2009; Ostrowsky et al., 2002). Activity in PIC is most closely linked to the intensity of the provoking stimulus, and lesions involving both areas can significantly alter pain processing (Garcia-Larrea et al., 2010; Greenspan et al., 1999; Segerdahl et al., 2015). Taken together, SII and PIC are the only brain regions with modality, intensity, and location specific cortical representations of nociceptive sensory stimuli, supporting their key contribution to somatosensory pain processing.

4.2. Affective pain circuits

The inherent unpleasantness of pain argues for the rapid recruitment of brain regions involved in affective processing. Parallel processing of pain unpleasantness is likely mediated in part by specific populations of spinal projection neurons, primarily originating in Lamina I, via the so-called ‘medial pain pathway’ to the brainstem, subcortical, and neocortical ‘affective-motivational’ circuits involved in autonomic control, arousal, and emotional expression. Most dorsal horn projections in this pathway traverse spinal parabrachial and medial thalamic nuclei en route to subcortical and cortical targets (Fig. 2). However, additional parallel pathways also contribute (e.g., the spinoreticulothalamic tract carrying fibers from deep lamina VII to medial thalamus), and direct projections from dorsal horn to the amygdala, basal forebrain, and various sites in the neocortex are also described (Cliffer et al., 1991; Fields, 1999; Huang et al., 2019; Price, 2000). Ascending medial pain pathway projections additionally impinge on several brainstem and diencephalic nuclei, including medullary and midbrain reticular formation, raphe nuclei, locus coeruleus, periaqueductal gray (PAG), and hypothalamus, which contribute to early autonomic and arousal changes that promote readiness for escape. Together, this organization establishes early parallel processing of ‘affective-motivational’ dimensions of pain alongside ‘sensory-discriminative’ processing.

Within the brain, ‘affective-motivational’ regions have characteristic features that make them well-suited to encode pain unpleasantness. Activity in this circuit tracks the intensity of noxious stimuli like ‘sensory-discriminative’ regions, but in contrast, neurons in affective regions typically have broad, bilateral receptive fields with little somatotopic organization, supporting a more general role in encoding aversiveness (Fields, 1999). Importantly, functions attributed to affective regions are not specific to somatosensory processing or pain. Key affective regions like the amygdala, ventral striatum, anterior insular cortex, and anterior cingulate cortex are broadly implicated in several core processing domains, including learning, reward processing, error monitoring, self-referential processing, and emotional expression, among many, many more specific functional attributions. This is perhaps unsurprising, as unpleasantness is certainly not unique to painful somatic sensations, but rather, is a more general characteristic of noxious and generally aversive stimuli ranging from non-painful tactile stimulation (e.g., itch) to other salient aversive stimuli across sensory modalities (e.g., visual, auditory, gustatory). Thus, although pain-provoking stimuli clearly activate these regions, their involvement is unlikely specific to pain unpleasantness and may instead reflect general coding of salience or aversiveness across modalities (Iannetti & Mouraux, 2010; Mouraux et al., 2011). To fully appreciate the contribution of each region to the experience of pain, it is helpful to develop a more general understanding of the component functions of each structure, which are reviewed elsewhere (Craig, 2009; De Olmos & Heimer, 1999; Haber, 2011; Shackman et al., 2011; Vogt, 2005). Below, we review the structure and function of a subset of affective regions in the context of pain.

The best studied cortical regions in the affective circuit are the anterior insular cortex (AIC) and anterior cingulate cortex (ACC), which were convincingly revealed in the first in vivo studies of acute pain in humans using Positron Emission Tomography (PET) and functional magnetic resonance imaging (fMRI) scans (Casey et al., 1994; Coghill et al., 1994; Talbot et al., 1991). The frequency with which these regions have appeared in the subsequent brain imaging literature has led some to suggest that both regions, particularly ACC, subserve a specific function in the context of pain (Lieberman & Eisenberger, 2015), though this view is controversial (Wager et al., 2016). Indeed, these two regions are among the most commonly (co)activated brain areas across the neuroscience literature, and their dysfunction is implicated in most neuropsychiatric disorders, including chronic pain (Craig, 2009; Motzkin, Kanungo, et al., 2023). In support of a key contribution of both regions to pain unpleasantness, however, is the thermal grill illusion, in which an illusory painful burning sensation generated by interlaced non-noxious warm and cool stimuli potently activates both regions, but not somatosensory cortex (Craig et al., 1996). Further, lesions to both regions, as well as to the cingulum bundle, the major white matter tract linking ACC with other pain processing regions, significantly disrupt the unpleasantness and aversiveness of pain, and, in certain cases, contribute to an “asymbolia” for pain, in which judgments about stimulus intensity are maintained but subjective aversiveness and the corresponding drive to escape are significantly attenuated (Berthier et al., 1988; Foltz & White, 1962; Freeman & Watts, 1946; Motzkin, Hiser, et al., 2023). Taken together, these observations suggest that AIC and ACC are likely required for the experience of pain unpleasantness.

The anterior insular cortex (AIC), a primate homologue of the rodent dysgranular insular cortex, is implicated in various specific functions, including gustatory processing, visceral sensation, and autonomic control. AIC receives projections from the thalamic ventromedial nucleus, posterior part (VMpo), a homologue of posterior thalamus in the rodent, and is reciprocally connected to the parabrachial nucleus and PAG (Dostrovsky & Craig, 1996). AIC also exhibits extensive connections with broad regions of the cortex, especially SII and posterior insula. AIC is potently activated by both positively and negatively valenced interoceptive stimuli, ranging from sensual touch to pain, itch, and bladder distension (Cechetto & Saper, 1987). Pain-responsive AIC neurons also respond to visceral, baroreceptor, and osmotic stimuli (Hanamori et al., 1998). In general, the AIC is thought to integrate sensory and interoceptive information to generate subjective “feelings” and promote awareness of internal body states. Whereas the magnitude of nociceptive stimuli is tracked by activity in SII and posterior insula, anterior insula activity reliably correlates with subjective judgments of pain intensity, consistent with its proposed role in body awareness (Craig, 2009; Segerdahl et al., 2015).

The ACC is a rostral subdivision of the cingulate cortex, a large and functionally heterogeneous midline frontal association cortex similarly implicated in a broad range of functions, including attention, cognitive control, motor planning, memory, decision making, reward processing, and emotional expression. The ACC receives projections from the mediodorsal (MD) thalamus and adjacent centromedian parafascicular (CMpf) thalamus (also known as the caudoventral mediodorsal thalamus), which transmits inputs from lamina I of the dorsal horn, medullary reticular formation, and pontine parabrachial nucleus (Craig & Dostrovsky, 1997). The cingulate cortex exhibits a gradient of connectivity along its rostrocaudal axis, with strong connections to limbic and prefrontal structures anteriorly and premotor structures posteriorly (Morecraft & Tanji, 2009; Shackman et al., 2011; Vogt, 2005). Consistent with a proposed role in pain unpleasantness, ACC activity in rodents tracks pain aversiveness, and ACC lesions attenuate pain avoidance behaviors (Johansen et al., 2001; LaBuda & Fuchs, 2005; Qu et al., 2011; Uhelski et al., 2012). In humans, the ACC is activated by acutely painful stimuli (Casey, 1999) and tonically hyperactivated in chronic pain (Apkarian et al., 2001; Moisset & Bouhassira, 2007; Price, 2000). Activity in ACC is selectively enhanced with hypnotic suggestion to increase the unpleasantness of a painful stimulus (Rainville et al., 1997). Further, individual pain responsive ACC neurons have been identified in humans alongside neurons that respond to observed pain in others (Hutchison et al., 1999). Perhaps most notably, ACC is a major source of identified projections to pain-modulatory brainstem nuclei (see below), suggesting that it may serve as a hub between ascending nociceptive transmission and descending modulation circuits. However, although ACC is broadly implicated in several pain-relevant functions, there are important cytoarchitectonic and functional distinctions among its subregions. “ACC” activations in human imaging data often involve more rostral divisions of the midcingulate cortex, which overlap with premotor regions that generate adaptive escape/avoidance responses to threatening stimuli (Morecraft & Tanji, 2009; Vogt, 2005). Regions involved in modulation and stimulus valuation tend to reside in more rostral subdivisions of ACC, generally corresponding with Brodmann areas 24a,b.

The amygdala and ventral striatum, often cited as key mediators of pain unpleasantness, are subcortical nuclei intimately involved in learning, reward, motivation, and emotional expression, among other key functions. The amygdala receives substantial projections from the parabrachial nucleus and projects to descending modulation circuits, including the PAG (see Pain Modulation Circuits). Activity in a subpopulation of basolateral amygdala neurons has been shown to specifically encode pain unpleasantness (Corder et al., 2019). The ventral striatum, although a recipient of fewer identified ascending spinal projections, is a central locus for reward and motivation and likely plays a key role in the transition from acute to chronic pain (Apkarian et al., 2013; Baliki et al., 2008, 2010, 2012). The ventral striatum, especially the nucleus accumbens (NAc), receives extensive dopaminergic projections from the ventral tegmental area, which are essential for reward processing and reinforcement learning (Haber, 2011; Ikemoto, 2010). Activity in the amygdala and ventral striatum are seen early in response to noxious stimulation and may be involved in the early recruitment of descending antinociceptive circuits (see Pain Modulation Circuits), or in encoding the ‘decision value’ of the stimulus to guide future behavior (Baliki et al., 2010; Becerra et al., 2001; Schwartz et al., 2017). Together, these regions are thought to generate potent salience and reinforcement signals to promote learning and the selection of appropriate behavioral responses to pain.

4.3. Evaluative pain circuits

Secondary emotional reactions to pain depend on the integration of somatosensory and affective processing streams with memories, goals, and contextual cues, which together generate subjective meanings from the pain experience. Components of the ‘affective-motivational’ circuit, like ACC and AIC, with extensive reciprocal connections to higher order orbitofrontal, lateral prefrontal, and lateral parietal structures, contribute to a ‘cognitive-evaluative’ circuit involved in self-referential and contextual processing functions, which relate to long-term implications of pain and the experience of suffering. Unlike somatosensory and affective regions, orbitofrontal, lateral prefrontal, and parietal regions involved in evaluative processes do not track the intensity of nociceptive stimuli, though orbitofrontal neurons have been shown to encode the ‘decision value’ of pain in the context of learning (Murray et al., 2007; Roy et al., 2012, 2014; Schoenbaum et al., 2007). Evaluative circuits instead appear to integrate somatosensory and affective information with context and expectations, and thus inform interpretations of ongoing body states (Garcia-Larrea & Peyron, 2013). Some evaluative regions, especially the orbitofrontal cortex and medial prefrontal cortex, are also directly implicated in emotional reappraisal and may therefore contribute to expectation and context-dependent modulation of activity in somatosensory and affective regions that ‘reinterpret’ or ‘reappraise’ the meaning of a painful stimulus (Bushnell et al., 2013; Hiser & Koenigs, 2018; Rudebeck et al., 2013). Indeed, the modulation of pain by placebo, expectations, and perceived controllability are all associated with activity in orbitofrontal and medial prefrontal evaluative regions (Atlas et al., 2010; Buchel et al., 2014; Salomons et al., 2004; Wiech et al., 2008; Wiech, 2016).

Taken together, ‘affective motivational’ and ‘cognitive evaluative’ regions form complex, highly interconnected circuits involved in a range of emotional and cognitive functions. These processes depend on extensive computations across distributed regions throughout the brain and, likely, resonant activity in corticostriatal and corticolimbic circuits, which markedly increases the complexity of their localization. Although these functions are clearly involved in the experience of pain, much remains to be learned about their organization and specific contributions to acute and chronic pain states.

4.4. Pain modulation circuits

The profound variability in neural and behavioral responses to identical nociceptive stimuli highlights the susceptibility of pain to modulation by a variety of factors. This variability results in part from descending projections that directly modulate spinal nociceptive processing (Fig. 2), and in part from network-level interactions among brain regions that influence central pain processing, some of which involve the ‘cognitive evaluative’ regions reviewed above. Fundamentally, a ‘pain modulation circuit’ consists of regions where activity modulates pain-related behaviors; for example, where activity correlates with pain relief, or where excitatory stimulation elicits analgesia. Regions with such characteristics often share a key feature: the involvement of endogenous opioidergic signaling. Opioidergic signaling is essential for normal function in modulation circuits and is reliably altered in chronic pain states, underscoring the relevance of these circuits to pathological pain (Fields, 2004).

Much of our present understanding of the organization of pain modulatory circuits is derived from physiological and behavioral observations following targeted neuraxial stimulation. The influence of supraspinal structures on nociceptive transmission in the dorsal horn was first revealed in stimulation experiments by Patrick Wall (Wall, 1967). Shortly thereafter, periaqueductal gray (PAG) was identified as a key locus of top down control with direct influence on dorsal horn nociceptive processing (Basbaum & Fields, 1978). The PAG is an evolutionarily conserved midbrain structure that plays an essential role in coordinating survival responses to threat, mediated through its top down influence on arousal, motor, and sensory processes (Koutsikou et al., 2017; Satpute et al., 2013). The PAG receives convergent inputs from the hypothalamus, central nucleus of the amygdala (CeA), and multiple locations in the frontal lobes, especially the ACC and AIC (An et al., 1998; Beitz, 1982; Shipley et al., 1991). The PAG projects to rostroventral medulla (RVM) and dorsolateral pontine tegmentum (DLPT), which in turn project through dorsolateral funiculus and selectively target nociceptive microcircuits in the dorsal horn. The PAG coordinates an opioidergic brainstem antinociceptive circuit largely through projections to RVM, which contains opposing populations of “ON” and “OFF” cells that directly facilitate or inhibit, respectively, nociceptive processing at the level of the dorsal horn. Activity in “ON” and “OFF” cells is related to a variety of state-dependent modifications, including arousal and sleep (De Preter & Heinricher, 2024). The RVM also contains spinally-projecting serotonergic cells that do not behave like ON or OFF cells but are also involved in descending pain modulation. Although serotonergic spinal projections from the brainstem were initially thought to drive descending inhibition, more recent work is beginning to characterize separable populations of pro- and anti-nociceptive dorsal horn-projecting serotonergic neurons in the brainstem, highlighting how similar monoaminergic projections can have divergent effects on spinal nociception (Ganley et al. 2023).

Opioids applied to each locus along the PAG-RVM-Spinal descending pathway inhibit behavioral responses to pain, and local or systemic application of opioid antagonists (e.g., naloxone, naltrexone) block the effects of stimulation (Fields, 2004). Importantly, PAG stimulation to activate descending antinociceptive pathways was rapidly translated to humans with neurosurgical deep brain stimulation of the PAG and neighboring rostral periventricular gray. In humans, PAG stimulation increases endogenous opioid release and is reversible by the opioid antagonist, naloxone (Akil et al., 1978; Hosobuchi et al., 1977; Hosobuchi, 1978; Richardson & Akil, 1977a, 1977b; Sims-Williams et al., 2017). Although variability in clinical response and off-target stimulation effects on adjacent oculomotor and escape circuits have limited its widespread use and clinical utility, PAG stimulation remains a viable neurosurgical intervention for pain relief in super refractory pain conditions.

Many brain regions comprising the ‘affective-motivational’ and ‘cognitive-evaluative’ circuits, and several additional cortical regions, most notably the primary motor cortex (M1), where stimulation is known to be analgesic, have extensive descending projections to the PAG and are thus likely important coordinators of brainstem antinociceptive circuits (Gan et al., 2022; Pagano et al., 2012; Shipley et al., 1991). Among these regions, the ACC is best studied. In rodents, ACC stimulation attenuates pain escape/avoidance behaviors and PAG lesions abolish this effect, indicating that “pain relieving” effects of ACC stimulation are mediated by the PAG (Gu et al., 2015; LaBuda & Fuchs, 2005; Senapati et al., 2005). In humans, functional connectivity between ACC and PAG is associated with pain relief and is typically enhanced during successful placebo and opioid analgesia (Bingel et al., 2006; Eippert et al., 2009; Krummenacher et al., 2010; Schmidt-Wilcke et al., 2014; Wager et al., 2007). Notably, placebo-related activations in these regions precede noxious stimulation, highlighting how expectations induce state-dependent modifications of neural activity with powerful downstream effects on pain processing (Atlas et al., 2010; Petrovic et al., 2002; Ploner et al., 2010; Wager et al., 2004).

On the other hand, as reviewed in detail above, the ACC is also directly implicated in encoding pain aversiveness; for example, ACC neurons in humans track pain intensity, bulk stimulation of ACC in rodents can provoke pain avoidance behaviors, and lesions of ACC across species attenuate pain behaviors. Thus, ACC activation can drive both pain aversiveness and pain relief. Although the specific mechanisms responsible for the divergence of ACC function in the context of pain or relief are not known, evidence from our laboratory suggests that projections from MD thalamus specifically activate subcortically-projecting ACC neurons involved in antinociception. In contrast, projections from the basolateral amygdala to ACC have the opposite effect (Meda et al., 2019). Another recent study found that ACC projections via PIC to the nucleus raphe magnus can establish peripheral hypersensitivity even in the absence of nociceptive input, highlighting another viable pro-nociceptive pathway originating from the ACC (Tan et al., 2017). Rostral subregions of ACC project to the nucleus accumbens (NAc), a ventral striatal nucleus involved in reward processing (Haber, 2011). Activation of these “reward circuits” precedes activation of classic pain circuitry (Becerra et al., 2001) and may influence the decision value and, therefore, behavioral responses to nociceptive stimulation. Also, ACC-NAc projections may promote a hedonic pain-relieving state through primarily supraspinal mechanisms (Navratilova et al., 2012). Studies of placebo and opioid analgesia suggest that other frontal regions implicated in evaluative processing, including the dorsolateral prefrontal cortex, medial prefrontal cortex, and orbitofrontal cortex, are also well suited to modulate activity across the network, potentially through similar top-down recruitment of antinociceptive circuits in the PAG (Wiech, 2016).

Several additional cortical, subcortical, and brainstem regions are implicated in pain modulation, including the AIC and a variety of brainstem nuclei, including the locus coeruleus and the reticular formation, among others. These parallel descending pathways involve mixed opioidergic and monoaminergic signaling and are likely coordinated by overlapping inputs from higher cortical and subcortical regions. For example, the locus coeruleus (LC) in the dorsal pons is the primary source of noradrenergic projections to the brain and spinal cord, especially the dorsal horn (Bruinstroop et al., 2012). Selective activation of descending spinal projections from LC produces robust ipsilateral antinociceptive effects, consistent with therapeutic effects of norepinephrine reuptake inhibitors like duloxetine and venlafaxine on neuropathic pain, whereas activation of ascending prefrontal cortex noradrenergic projections increase pain and anxiety behaviors (Hirschberg et al. 2017). Recently, converging projections from the PAG and RVM to the LC have been shown to directly interact to influence LC projections to the dorsal horn in the context of opioidergic antinociception (Lubejko et al., 2024) and descending A5 noradrenergic projections have been shown to be required for DNIC (Kucharczyk et al., 2023). Clearly, complex interactions among brainstem nuclei can enhance and suppress nociceptive processing, providing a mechanism for its bidirectional control.

4.5. Network-level pain processing

As we noted above, the full complement of brain regions implicated in pain processing is commonly referred to as the Pain Matrix (Melzack, 1999). Component regions occupy nearly 15 % of the cortical mantle, which reflects the richness and depth of the pain experience, the complexity of which cannot be fully explained by activity in any single brain region (Baliki & Apkarian, 2015). However, as was particularly apparent for brain regions comprising ‘affective motivational’ and ‘cognitive evaluative’ circuits, functional attributions to most Pain Matrix regions are not specific to pain. How, then, does the holistic experience of pain emerge from the component functions of pain matrix regions? Although this question remains largely unanswered, it is helpful to consider how component regions assemble into functional networks and how regions within these networks may work together to instantiate and modulate pain. This emerging network-based approach promises to be the next frontier in pain research.

Prior theoretical accounts of the neural bases of language, memory, and emotion provide a helpful framework to understand how complex experiences emerge from distributed networks of functionally related brain regions (Mesulam, 1990). For example, whereas lesions to several specific areas may contribute to convergent impairments in memory or language, no single “memory” or “language” region is responsible for these processes. Thus, if each of the “pain relevant” brain regions summarized in this chapter were critical to the experience of pain, we might expect that activating or inhibiting any one of them would disrupt or enhance pain. However, with a few notable exceptions, stimulation of individual brain regions rarely elicits pain and lesions to single regions seldom abolish pain. In support of the primacy of distributed network representations to the subjective experience of pain, the neural pain signature (NPS), a weighted combination of fMRI activity across several Pain Matrix regions, can reliably discriminate acute pain from related sensations and even predict the magnitude of subjective pain in individual subjects, a feat untenable with estimates from single brain regions (Wager et al., 2013). Indeed, it has been postulated that conscious awareness of pain requires a transition from subconscious processing by single regions or subnetworks to coherent brain-wide representations across large-scale distributed brain networks (Dehaene & Naccache, 2001; Garcia-Larrea & Peyron, 2013). Together, these observations indicate that large-scale interactions across brain networks may be more closely related to the holistic experience of pain than activity in individual regions.

Recently, advances in functional brain imaging techniques have enabled the exploration of large-scale distributed brain networks involved in pain processing. This approach leverages the discovery that spontaneous fluctuations in fMRI signals during rest exhibit correlated low-frequency oscillations (<0.1 Hz), which can effectively map connections between different brain regions (Buckner et al., 2013). Using this method, researchers can investigate how various brain areas collaborate to support complex functions and how alterations in these connections might influence different experiences, such as pain. Studies utilizing these approaches have identified and characterized several intrinsic resting state networks (RSNs), which are defined by synchronized temporal fluctuations in resting brain activity.

Several identified RSNs correspond directly with established sensory-discriminative, affective-motivational, and cognitive-evaluative pain circuits (Davis & Moayedi, 2013; Kucyi & Davis, 2015; Motzkin, Kanungo, et al., 2023). Of particular relevance to pain are the sensorimotor network, consisting of primary motor cortex and primary and secondary somatosensory cortex; the default mode network (DMN), consisting of evaluative regions involved in self-referential processing including the medial prefrontal cortex, posterior cingulate cortex, and angular gyrus; the frontoparietal network (FPN, also known as the central executive network), consisting of additional cognitive-evaluative regions including lateral prefrontal and lateral parietal cortex involved in sustained attention, problem solving, and working memory; and the salience network (SN, also known as the ventral attention network and cingulo-opercular network), consisting of key affective regions, including the ACC and AIC, which are thought to mediate switching between ruminative functions of the DMN and the externally directed attentional functions mediated by the FPN (Fig. 2).

In the context of pain, intrinsic RSNs may help to explain how subconscious nociceptive processes gain access to conscious awareness, resulting in sustained attention toward pain and the establishment of “remembered pain” (Garcia-Larrea & Peyron, 2013). The RSNs may also shed light on the mechanisms through which between-region connections could contribute to effective or deficient pain modulation. For example, attention to pain is associated with activation of the salience network (containing ACC and AIC) and frontoparietal networks (containing lateral cortical evaluative regions involved in sustained attention), whereas mind wandering away from pain is associated with activation of the default mode network (containing a subset of medial prefrontal and lateral parietal evaluative regions involved in ruminative and introspective processing). In fact, functional coupling between frontoparietal and somatosensory and affective networks is thought to be required to permit nociceptive processes access to conscious awareness (Garcia-Larrea & Peyron, 2013). Further, connectivity between DMN subregions and the PAG increases during periods of distraction away from pain, providing a viable mechanism through which distraction or mind wandering may directly contribute to spinal nociceptive modulation (Kucyi & Davis, 2015; Wiech, 2016).

Although much remains to be learned about the contribution of brain networks to pain, a network-centered approach is particularly helpful in understanding the extensive cortical reorganization that accompanies the transition to chronic pain. In the remaining sections, we describe how circuit representations shift and change in the context of injury and/or longstanding pain.

5. Reorganization after injury: circuits/networks

5.1. Nociceptors and spinal circuits

Peripheral neuropathies, including those caused by direct injury, neurotoxicity, and gain-of-function mutations, profoundly affect the physiology of sensory neurons (Dib-Hajj et al., 2017). Nociceptor hyperexcitability is theorized as a central mechanism of hyperalgesia in peripheral neuropathic pain (Devor, 2006). In an injured state, dorsal root ganglion (DRG) neurons can exhibit spontaneous ectopic activity (Devor, 1991), transcriptomic signs of dedifferentiation (Renthal et al., 2020), abnormal peripheral terminal regeneration (Gangadharan et al., 2022), and even ectopic gap junction communication with intact DRG neurons (Bráz et al., 2011; Kim et al., 2016). While lidocaine can, in most cases, provide immediate relief from peripheral neuropathic pain, the effects of dorsal rhizotomy (which removes sensation from the affected nerve) are complex and may either alleviate or exacerbate pain. Notably, lesser-known mechanisms, such as age-related factors, can suppress pain expression even when nociceptor hyperexcitability is present (Wimalasena et al., 2023).

Injured nociceptors also play a significant role in central sensitization in dorsal horn circuits. Although the term “central sensitization” is sometimes misapplied clinically to explain symptomatology, it refers specifically to a spinal circuit mechanism capable of producing a persistently increased responsiveness to normally subthreshold sensory input (Latremoliere & Woolf, 2009). A classic example is secondary hyperalgesia, in which areas surrounding an injury become hypersensitive to mechanical stimuli but not heat (Treede et al., 1992). Similar mechanisms likely support other characteristic features of neuropathic pain states, like dynamic tactile allodynia and the spreading of tenderness. Some researchers suggest that central sensitization may sustain allodynia and hyperalgesia without ongoing peripheral abnormalities (Latremoliere & Woolf, 2009), although this view is somewhat controversial (Meacham et al., 2017).

Within the spinal cord, neuropathic pain is associated with a variety of neuroplastic changes, such as synapse growth or loss, altered connectivity, reduced inhibition, and increased excitability in dorsal horn neurons (Basbaum et al., 2009; Kuner & Flor, 2016b). Historically, anterolateral tractotomy—the surgical severing of ascending axons from Laminae I and V projection neurons—was used to treat intractable pain (Gowers, 1886; Spiller, 1912). While this procedure often provided immediate relief, the pain frequently returned, likely due to spinal circuit reorganization involving multisynaptic commissural pathways (Basbaum, 2022). However, spinal neuroplasticity can also promote repair, as demonstrated by GABAergic cell transplants in chronic pain models, which have reversed hypersensitivity by silencing hyperexcitable spinal circuits (Bráz et al., 2012).

Importantly, significant interindividual variability can be seen in populations of patients with ostensibly the same peripheral injury or ascending pathway disruption, which indicates that secondary brain changes are likely required for the development of a chronic pain condition. Specific brain changes in descending pain modulation pathways are proposed to play a critical role in the “chronification” of pain (Drake et al., 2021; Ossipov et al., 2014). Descending modulatory pathways are highly plastic and responsive to injury-induced changes in spinal circuits (Bannister & Dickenson, 2017; Suzuki et al., 2002).

5.2. Brain circuits and networks

In the preceding section, we summarized how injury-related modifications in spinal circuits impact nociceptive processing and pain-related behaviors. Although these changes can account for a range of somatosensory characteristics typical of chronic neuropathic pain (e.g., hyperalgesia, allodynia), most chronic pain syndromes involve additional alterations in emotion processing, motivational behavior, cognitive and attentional processes, and memory, which together drive complex experiences of suffering in pathological chronic pain states. Below, we review these changes.

Injuries to the CNS provoke extensive plastic reorganization of central brain circuits that can significantly alter central sensorimotor representations and contribute to pathological pain (Baliki & Apkarian, 2015; Bushnell et al., 2013; Garcia-Larrea & Peyron, 2013; Kuner & Flor, 2016; Rosner et al., 2023). These changes are best understood in the context of limb amputation and spinal cord injury, which are typically accompanied by plastic expansions or shifts in cortical representations of the affected body part (Flor, 2003; Flor, Nikolajsen, & Staehelin Jensen, 2006; Kuner & Flor, 2016). Lesions anywhere along the spinothalamic tract (including damage affecting terminal projections to cortex) can provoke a central pain syndrome characterized by coexisting sensory impairments and unremitting spontaneous neuropathic pain, the latter of which is ascribed to plastic reorganization above and below the level of the lesion (Boivie et al., 1989; Leijon et al., 1989; Rosner et al., 2023). Although some of these plastic changes are directly related to the emergence of painful sensations, most of our understanding of the central changes that accompany (or perhaps give rise to) chronic pain are from cross-sectional studies of patients with non-neuropathic pain, such as chronic low back pain, osteoarthritis, and fibromyalgia, which may involve distinct central mechanisms. As such, it remains unclear whether many of the changes described below are a cause or consequence of the painful condition.

The most commonly identified anatomical changes that occur in a chronic pain setting are found in brain regions containing affective-motivational and cognitive-evaluative circuits. In fact, the three cortical regions that most consistently show decreases in gray matter volume across pain conditions are the ACC, AIC, and prefrontal cortex (Bushnell et al., 2013), though changes have also been found in SI, hippocampus, basal ganglia, and primary motor cortex (Kuner & Flor, 2016). Notably, in all three regions, molecular imaging studies show decreases in opioid receptor binding in patients with chronic pain (Bushnell et al., 2013). Rodent models of chronic neuropathic pain have also identified changes in dendritic size and spine density in similar frontal regions (Metz et al., 2009; Seminowicz et al., 2009).

It is less clear how observed structural changes influence the function of affected brain regions. Recent large-scale meta-analyses comparing responses to acute experimental pain between patients with chronic pain and healthy adults have identified abnormal activity in left AIC as a specific functional marker of chronic pain (Ferraro et al. 2022; but see: Kim et al. 2021; Xu et al. 2021). However, both increases and decreases in activity have been observed in left AIC. These divergent findings highlight the possibility of profound alterations in structure-function relationships in chronic pain, or perhaps simply reflect important differences between evoked experimental pain used in research and spontaneous chronic clinical pain. Other activity patterns seen in patients with neuropathic pain include relative hypoactivity of the ventromedial and orbitofrontal cortex, which are thought to contribute to deficient appraisals of pain, as well as the emergence of thalamic hypoactivity contralateral to pain and a shift in pain-related activations from contralateral to ipsilateral PIC, which together suggest that additional pathology-specific reorganizations may contribute to persistent pain, allodynia, and hyperpathia, although much more research is needed in clinical populations (Derbyshire, 1999; Garcia-Larrea & Peyron, 2013).

At the level of functional networks, the most consistent difference identified between patients with chronic pain and healthy comparison adults is a “disconnection” of the frontal component of the default mode network (DMN). Similar alterations in gray matter volume and functional connectivity in the DMN have been described in multiple pain conditions (Baliki et al., 2008, 2014; Brandl et al., 2022; Cauda et al., 2014; Kucyi & Davis, 2015). Across pain etiologies, consistent changes are also seen in salience and frontoparietal attention networks, including ACC, AIC, and medial prefrontal regions. Indeed, a recent study comparing chronic pain with other neuropsychiatric disorders found that reduced connectivity between DMN and salience networks is likely specific to pain (Brandl et al., 2022). In contrast, condition-specific network reorganization is more commonly seen in somatosensory regions (Cauda et al., 2014). Importantly, the neural pain signature, which can accurately predict acute pain from distributed brain activity, performs poorly when applied to clinical chronic pain states (Lee et al., 2021). These observations suggest that there is likely a fundamental reorganization of brain function in chronic pain. Indeed, more recent research applying new mathematical methods from graph theory to brain networks suggests that, across pain conditions, chronic pain is associated with large-scale disruptions in the overall pattern of brain-wide connections (i.e., network topology) (Lenoir et al., 2021; Mansour et al., 2016). We and others have proposed that such large-scale disruptions may reflect dysfunction in highly connected network hub regions, leading to a disorganized and hyperconnected brain (De Pauw et al., 2020; Motzkin, Kanungo, et al., 2023).

Finally, affective and evaluative circuits seem particularly relevant to transitioning from acute to chronic pain. Early structural changes in the nucleus accumbens and insula predict the transition from subacute to chronic low back pain, indicating that volumetric changes in these regions may be particularly relevant to the development of chronic pain (Apkarian et al., 2013; Geha et al., 2008; Hashmi et al., 2013; Vachon-Presseau et al., 2016). These structural changes are also associated with enhanced functional connectivity between the nucleus accumbens and prefrontal cortex, a finding that can predict the likelihood of developing chronic pain with up to 80 % accuracy (Baliki & Apkarian, 2015; Baliki et al., 2012). In preclinical models, changes in similar regions are seen, including comparable changes in connectivity, reductions in dopamine receptor expression, and altered long term depression (LTD), which together impair motivated behavior (Chang et al., 2014; Schwartz et al., 2014). Overall, the preponderance of changes in frontal and limbic circuits, especially the prefrontal cortex and ventral striatum, has led some to suggest that plastic reorganization in these regions during chronic pain renders incoming nociceptive inputs more aversive and emotional (Mansour et al., 2014). Alterations in these circuits in chronic pain mirror changes reported in mood and anxiety disorders, and, indeed, pathology in similar regions is implicated in most neuropsychiatric disorders. Shared changes, to the extent that they reflect similar symptomatic manifestations across disorders, may shed light on fundamental mechanisms of pain chronification, which could improve brain-based treatments.

6. Conclusion

When we encounter noxious environmental signals, the nociceptive somatosensory system encodes this information, extracting its qualities to form specific, interpretable, actionable perceptions. Under normal conditions, peripheral noxious information ascends to the brain within nociceptive pathways, provoking the experience of pain, which involves activation in brain regions involved in sensory-discriminative, affective-motivational, and cognitive-evaluative processing. However, the link between peripheral nociceptive activity and pain is not immutable, and a variety of factors can influence the transmission of injury-induced signals en route to the brain, including descending modulatory projections that strongly influence the first "pain" messages reaching the spinal cord from the periphery. Under pathological conditions, alterations in anatomy and physiology at each level of the system can provoke pain and, in some cases, contribute to ongoing or sustained pain that outlasts abnormal peripheral nociceptor activity. Understanding relationships between anatomical compartments is essential to guide the development of new treatment strategies and provide tools for analyzing and predicting treatment effectiveness.