Modulation of Spinal Nociceptive Processing

Timothy J. Ness

Alan Randich

Jennifer J. Deberry

The preceding chapter addressed the neuroanatomy and neurochemistry of neurons located within the spinal cord that process information related to pain-related sensation. These neurons are highly regulated components of the central nervous system (CNS) with inhibitory and excitatory feedback mechanisms. Pain may be due to increased primary afferent input or may be the result of a failure of feedback regulation or a combination of the two. Too much “gain” or inadequate “braking” can both result in an excess of sensory transmission. In the normally functioning state, there are tonic influences present as well as evocable mechanisms whereby the responses of second-order neurons can be suppressed or facilitated dependent on other events important to the organism. Some modulatory effects are relatively “hardwired” occurring in a reliable and predictable fashion. Other modulators are less predictable and may be dependent on psychic/cognitive processes that vary from organism to organism and may involve learning, motivation, or emotional factors. In most cases, modulatory systems are adaptive in that they help an organism to function optimally. Unfortunately, with disease, some of these same modulatory systems have become maladaptive and serve to impede both physiologic and social processes of healing. The complex nature of these modulatory influences and the neurotransmitters involved are discussed in three parts beginning with a discussion of mechanisms based at spinal levels, followed by a discussion of mechanisms related to descending influences, and finally by a brief discussion of some “triggers” of spinal neuron hyperexcitability that may occur pathologically and/or iatrogenically which lead to hypersensitivity. These triggers are used as examples of the interactive nature of these modulatory forces. There is value in understanding endogenous modulatory systems because they are the systems which the clinician activates or suppresses by using exogenous modulators, such as electrical stimulation or pharmacologic agents. Modulation occurs at each step of processing within the CNS, but the focus of the present discussion is the modulation of the spinal second-order nociceptive neuron.

Spinal Cord-Based Modulatory Mechanisms

ACUTE SEGMENTAL MODULATORY EFFECTS

Sensory inputs to the spinal cord and trigeminal nucleus begin to interact at the very first steps of transmission. Activation of large diameter afferents (Aδ) produces an inhibitory effect on the processing of signals from small diameter (Aδ and C-fiber) afferents. This effect has been documented since ancient times and is relearned by every child who rubs or massages injured parts of their bodies in order to achieve pain relief. The mechanism of this manipulation has been more difficult to explain than the time-honored efficacy of the effect.¹ What is clear is that second-order neurons of the spinal cord have both excitatory and inhibitory “receptive fields.” Namely, stimulation of different parts of the body using one or more types of stimuli (e.g., noxious heat, low threshold mechanical, high threshold mechanical) results in the depolarization or hyperpolarization of individual second-order neurons. Inhibition of second-order neurons which is produced by noxious stimuli can be evoked from heterosegmental sites and so is discussed separately later. Inhibition of second-order neurons which is produced by nonnoxious stimulation appears to be predominantly segmentally organized. Theoretically formulated as the initial gate control theory of Melzack and Wall,² this effect was hypothesized to occur because of a combination of presynaptic inhibition and the actions of inhibitory interneurons located in lamina II (substantia gelatinosa) of the spinal cord which are activated by large diameter afferents (Fig. 5.1). Although the specifics of this theory have evolved further to include nonsegmental effects, its general description has served as the theoretical underpinnings for the clinical effects of neuromodulatory (electrostimulatory analgesic) techniques ranging from transcutaneous electrical nerve stimulation to spinal cord stimulation and aspects of acupuncture.³

Numerous electrophysiologic studies of second-order neurons have observed that low-intensity, high-frequency electrical stimulation of nerves or somatic tissues located at the same segmental level as the neuron produces an inhibitory effect which is not reduced by naloxone (nonopioidergic) but which may involve GABAergic or glycinergic mechanisms. This phenomenon is present in both spinally transected and intact animals and so does not necessarily involve a brainstem mechanism. Dorsal column stimulation, which produces retrograde activation of Aδ fiber inputs to the spinal cord (but may also activate descending modulatory pathways), produces similar nonopioid, GABAergic, and/or glycinergic inhibition of spinal nociceptive processing.

FIGURE 5.1 Gate control theory as originally schematically described by Melzack and Wall² where large-diameter (L) and small-diameter (S) primary afferent fibers project to substantia gelatinosa (SG) and second-order transmission (T) neurons in the spinal dorsal horn. The inhibitory effect of SG neuronal activity is increased by L and decreased by S fiber activity. T neurons transmit information to the brain and other action sites. Activation of peripheral or central projections of L fibers using transcutaneous nerve stimulation, peripheral nerve stimulators, or dorsal column stimulators would all be expected to produce inhibition of S fiber input to the T cells.

HETEROSEGMENTAL MODULATORY SYSTEMS

Both excitatory and inhibitory effects can occur at one spinal level when stimuli are presented to distant portions of the body. Excitatory effects have generally been described in imprecise terms as extended connectivity or as propriospinal pathways that are just one component of the systems that alter the “gain control” mechanisms of nociceptive systems.⁴ Some of these intraspinal networks serve to integrate both sensory and motor functions involving the upper and lower extremities with an example being crossed flexion-extension reflex responses to noxious stimuli. A coordination of pelvic organ function also relies on intraspinal excitatory and inhibitory connections that link processing of sensory information from afferents traveling in the pelvic nerve to the lumbosacral cord with that of afferents traveling in sympathetic nerves to the thoracolumbar spinal cord. Intraspinal networks of neurons, which form a reticular network in the deeper parts of the spinal dorsal horn, have been described in the context of the multisynaptic ascending system of Noordenbos.⁵ This same network could just as easily serve as the substrates for multisynaptic descending modulatory influences.

The most formally studied heterosegmental interaction related to nociception is the phenomenon known as diffuse noxious inhibitory controls (DNIC). This endogenous inhibitory system is activated by noxious stimuli presented to a distant nonsegmental site and results in the inhibition of ongoing or evoked dorsal horn neuronal activity. The mechanisms of DNIC are postulated to involve the activation of brainstem nuclei that subsequently produce inhibition of spinal dorsal horn neurons through a descending modulatory mechanism, but it is notable that the neurophysiologic phenomena associated with DNIC have been demonstrated in spinally transected preparations. These “propriospinal” phenomena represent a general inhibitory system activated by heterosegmental noxious conditioning stimuli which is either synonymous with or highly augmented by the presence of a brainstem and mechanisms of DNIC. According to its original description by Le Bars et al.,⁶^,⁷ DNIC results in the inhibition of class II (wide dynamic range [WDR]; convergent) neurons and has no effect on class III (nociceptive specific) neurons. A consistency of many studies related to DNIC (and propriospinal heterosegmental inhibition) is that they have identified that most spinal neurons responsive to noxious stimuli effectively have “total body” receptive fields in that noxious stimuli will produce excitation or inhibition that is dependent on precise body site. DNIC-sensitive neurons have a resultant “inhibitory surround” effect where all noxious inputs outside of a defined area are inhibitory to the individual neurons. Human studies of the similar, but more general, phenomenon of conditioned pain modulation have described the potential for complex interactions of heterosegmental stimuli in which one pain typically inhibits other pains (as in DNIC) except in case of pathologic pain disorders where the expected inhibition may be absent or facilitation may be noted instead.⁸

C-FIBER WIND-UP AND CENTRAL SENSITIZATION

Changes in excitability occur in second-order neurons when repetitive or prolonged high-intensity input is received from primary afferent C-fibers. One of these changes in excitability is termed C-fiber wind-up. Noted by Mendell⁹ when recording from ascending axons of spinal dorsal horn neurons, wind-up is the phenomenon whereby repeated electrical C-fiber activation at certain rates (i.e., ≥1 Hz) leads to a sequential increase in the number of action potentials evoked by each stimulus (Fig. 5.2). Slower stimulus rates do not produce progressive increases in activation. Mechanical and thermal stimuli at intensities sufficient to activate C-fibers also produce similar wind-up. This sequential increase in response can be blunted through use of N-methyl-D-aspartate (NMDA) receptor antagonists and the effect disappears after a few seconds of nonstimulation.

FIGURE 5.2 Wind-up responses of single dorsolateral column axon to repeated stimulation of the sural nerve at sufficient intensity to activate A and C fibers (no wind-up seen with A-fiber stimulation by itself). The vertical time markers on the far right represent 100 milliseconds. Each mark at the bottom of the time line represents the stimulation artifact and the burst of activity immediately above each of these stimulations is the response to A-fiber stimulation (each dot represents an action potential). The more delayed responses are to the more slowly conducting C-fiber inputs. Response to stimulation shows increasing C-fiber wind-up responses on to 1 per second stimulation (not to 1 every 2 or 1 every 4 second stimulation rates at right). Wind-up lasts for only several seconds following the stimulation as seen by transient increase in spontaneous activity. (Redrawn from Mendell LM. Physiological properties of unmyelinated fiber projections to the spinal cord. Exp Neurol 1966;16:316-332.)

Another general category of increased neuronal excitability is termed central sensitization. This term has been used in a focused manner to describe acute changes in the responsiveness of second-order neurons following high-intensity or prolonged stimuli such as those that occur with nonneuronal tissue injury and subsequent inflammation. The term has also been used to describe phenomena such as delayed-onset nerve injury-related hypersensitivity and, in that case, is more subacute or chronic in nature with a potential for morphologic as well as biochemical alteration of second-order neurons. For purposes of the present discussion, injury-induced central sensitization as described by Woolf¹⁰ is used as the archetype model of central sensitization (Fig. 5.3). Multiple studies have demonstrated that tissue injury produces an augmentation of nociceptive reflexes that is NMDA receptor-dependent. In preclinical models, pharmacologic treatment has the greatest effect if given prior to injury and a blocking of afferent input serves to delay the onset of development of hypersensitivity. Extrapolating from this data and coupling it with evidence of long-term potentiation (LTP) of synaptic efficacy in the spinal cord after even brief bouts of NMDA receptor activation and interaction with glia,¹¹ some have further extrapolated these laboratory data to the clinical concept of “preemptive analgesia.” Treating pain before (and after) it begins has a clear potential for clinical benefit, although the true clinical significance of early intervention has proven difficult to define. On a neurophysiologic basis, an expansion of cutaneous excitatory receptive fields has been noted following tissue injury which follows a similar pharmacology, but specific results have been model- and species-dependent.

Supraspinal Modulatory Systems

TONIC DESCENDING INFLUENCES

A characteristic of spinal nociceptive systems is that they are under tonic descending inhibition such that a common effect of injury to spinal pathways is a release from this inhibition. Hyperreflexive states with secondary spasticity and autonomic
lability can occur. The precise neurophysiologic circuits associated with this descending inhibition is of significant debate, but known inhibitory neurotransmitters such as norepinephrine (NE) and serotonin (5-hydroxy-tryptophan; 5-HT) are synthesized in the brainstem and transported to the spinal cord from multiple supraspinal sites. This role for supraspinal structures in providing descending influences on spinal reflexes has long been recognized. In 1915, Sherrington and Sowton¹² demonstrated enhanced flexion reflexes following spinal transection. Later in 1926, Fulton¹³ suggested that this effect reflected removal of tonic descending inhibitory modulation of spinal interneurons mediating those reflexes. Descending control of flexion reflexes was extensively studied in ensuing years,¹⁴ but these studies did not target the issue of how the brain might specifically modulate incoming nociceptive signals from peripheral tissue.

FIGURE 5.3 Raster dot displays of a single biceps femoris unit activated by stimulation of the sural nerve once every 2 seconds before an ipsilateral thermal injury (Control), 30 and 60 minutes postinjury, and 10 minutes after the injured foot has been completely anesthetized with local anesthetic (LA). Each dot represents a unit discharge. The vertical scale is the latency of the responses after sural nerve stimulation, and the stimulus artifact can be seen at time 0. Stimulation strengths were sufficient to activate Aδ, Aδ, and C fibers. Note the different time scales used in the three panels to record the activity evoked by the three different fiber populations. In the preinjury state, only Aδ input was evoked. Thirty minutes after injury, a C-fiber response begins to occur, whereas at 60 minutes, both Aδ and C-fiber evoked responses are present (the C-fiber responses with wind-up). Ten minutes after LA, the C-fiber evoked responses remain higher than before the injury suggesting a central component of the sensitization. (Redrawn from Woolf CJ. Evidence for a central component of post-injury pain hypersensitivity. Nature 1983;306:686-688.)

FIGURE 5.4 A modification of the gate control theory schematic models includes excitatory (white circle) and inhibitory (black circle) links from the substantia gelatinosa (SG) to the transmission (T) cells as well as descending inhibitory control from brainstem systems. The round knob at the end of the inhibitory link indicates that its actions may be presynaptic, postsynaptic, or both. All connections are excitatory except the inhibitory link from SG to T cells. (Redrawn after Melzack R, Wall PD. The Challenge of Pain. New York: Basic Books; 1983.)

A series of seminal events in the late 1960s and early 1970s led to a full-fledged appreciation and analysis of descending modulation of spinal nociceptive processing. These included a modification of the original gate control theory to include supraspinal systems (Fig. 5.4). This change was prompted by studies which showed that spinal dorsal horn neurons were subject to tonic descending inhibitory influences¹⁵ and Reynolds’s¹⁶ demonstration that electrical stimulation of the midbrain periaqueductal grey (PAG) produced analgesia sufficient to perform abdominal surgery in a rat. This last phenomenon was referred to as “stimulation-produced analgesia” (SPA).¹⁷^,¹⁸ SPA can also be produced in humans¹⁹ in the form of deep brain stimulation, and it suggests the existence of endogenous systems that can selectively modulate pain. This served as the impetus for the extensive, formal analyses of supraspinal structures involved in descending modulation of spinal nociceptive processing that continues today. Later, a number of investigators found that electrical or chemical stimulation of other brain regions could also promote facilitation of nociceptive processing,²⁰^,²¹ suggesting the existence of similar descending facilitatory systems.

SUPRASPINAL SUBSTRATES MEDIATING THE DESCENDING MODULATION OF PAIN

Periaqueductal Grey of the Mesencephalon and the Rostral Ventral Medulla

The midbrain PAG and the rostral ventral medulla (RVM), in particular the medullary nucleus raphe magnus (NRM) figured prominently in the original analyses of descending modulation of pain. Indeed, they are often viewed as the “backbone” of the pain modulatory system²² and have been more extensively studied than any other brain regions. Yet, other brainstem nuclei/cell groups also serve in this role and include the nucleus gigantocellularis (NGC), nucleus reticularis gigantocellularis pars alpha (NGCα), nucleus paragigantocellularis (NpGC),
midbrain reticular formation, locus coeruleus/A6 cell group (LC/A6), the A5 cell group, the lateral reticular nucleus (LRN) and nearby A1 and C1 cell groups, the parabrachial nucleus/A7 cell group (PBN/A7) and the nucleus tractus solitarius (NTS). A limited amount of information is also available on cortical and limbic systems such as the anterior cingulate cortex (ACC), the amygdala, and hypothalamus that contribute to descending modulation. It is the investigation of these structures that has led to our current understanding of how descending pain modulatory systems affect pain perception. This topic has been well reviewed.¹^,²³^,²⁴^,²⁵^,²⁶^,²⁷^,²⁸^,²⁹ Table 5.1 summarizes results related to many of these areas separately along with the neurotransmitters associated with their putative inhibitory versus facilitatory effects on nociceptive transmission. A summary of the most important components is described in Figure 5.5.

The PAG was the initial site of investigation for endogenous pain control systems and is still viewed as an integral component of these systems. Antinociception produced by electrical stimulation of the PAG (SPA) is profound and comparable to that produced by a high dose of morphine. It eliminates behavioral and spinal dorsal horn neuronal responses to noxious stimuli including electric shock applied to the tooth pulp or limbs, noxious heating of the tail and hind paws, noxious pinching of the limbs, and injection of irritants into the viscera. The effects of SPA are produced almost immediately after the onset of stimulation and may last from a few seconds to hours after termination of stimulation. Microinjection of opiates into the PAG also produces behavioral antinociception and inhibition of spinal nociceptive transmission via disinhibition of inhibitory interneurons in the PAG. The subsequent discovery of endogenous opioid receptors and peptides⁵⁶^,⁵⁷^,⁵⁸^,⁵⁹ and demonstration of the presence of opioid receptors throughout the brainstem⁶⁰ further established a role for the PAG in pain modulation. Similarities were also observed between phenomena associated with PAG-derived inhibitory effects and opiate-induced analgesia, including tolerance and cross tolerance.⁶¹ SPA- and morphine-induced antinociception from the PAG also involve a spinal release of 5-HT and NE and mediation by both spinal 5-HT receptors and α₂ adrenoreceptors suggesting the need for a relay through serotonergic and noradrenergic brainstem sites. Anatomical studies reveal relatively few fibers that descend from the PAG directly to the spinal cord.²⁸ However, the PAG does have strong projections to the NRM and adjacent areas of the RVM. Attention has therefore been focused on the NRM as the primary relay in mediating the antinociceptive effects of activation of PAG neurons. Studies of the NRM were performed in a manner analogous to those performed in the PAG and often with comparable results.²⁸^,²⁹ Electrical stimulation of sites within the PAG or NRM produces inhibitory postsynaptic potentials (IPSPs) in dorsal horn neurons including those with ascending projections.⁶² Lesions of the dorsolateral funiculi (DLFs) of the spinal cord eliminate this inhibition and so this white matter pathway has been viewed as the primary spinal locus for descending fibers from the RVM.⁶³ Ventrolateral funiculi (VLFs) have also been implicated as the spinal pathways by which descending systems access the spinal dorsal horn (e.g., from the LRN), but there is greater evidence for these descending paths to promote facilitatory influences as opposed to inhibitory influences.

TABLE 5.1 Central Nervous System (CNS) Sites Modulating Nociceptive Transmission (NT)

CNS Site	Direct Projection to Spinal Segments	Possible Relay Sites	Facilitation NTs	Inhibition NTs
*Spinal Cord*
Segmental	NA	Multisegmental	GLUT	GABA, glycine
Propriospinal	Many	Multisegmental	GLUT	GABA, glycine, opioid, ACh
Heterosegmental	Many	Multisegmental-RVM-DRt	GLUT	Opioid, 5-HT, NE, GABA
*Medulla*
NTS	Few	RVM, LC, PAG, A5, cortex	GLUT	NE and 5-HT together
RVM (region)	Many		5-HT2, 5-HT3, NE (α1), GLUT, ACh	5-HT1A, 5-HT1B, 5-HT1D, 5-HT7, opioid, NE (α2), ACh, GABA, glycine
NRM	Many		5-HT2, 5-HT3	5-HT, opioid, ACh, GABA
NGC, NGCα, NpGC	Many	LC	GLUT, 5-HT, CCK-B, NE (α1)	5-HT, NE(α2)
LRN	Some	LC, RVM, A5		NE (α2)
A1	Few		NE (α1)	NE (α2)
DRt	Many		GLUT
*Pons*
LC/A6/A5	Many		NE (α1)	NE (α2)
PBN/A7	Many/few	PAG, RVM	NE (α1)	NE (α2), oxytocin?
*Mesencephalon*
PAG	Few	RVM, LC	NE (α1), 5-HT	5-HT, NE (α2), opioid, ACh
*Diencephalon Cortex*
Hypothalamus	Some	RVM, PAG	GLUT	5-HT, NE(α2), DA
Amygdala	Some	PAG, PBN	GLUT, oxytocin? CRF-related?	NE(α2), oxytocin? CRF-related?
ACC	Few	RVM, PAG	GLUT, 5-HT
Sensory cortex	Few	RVM	GLUT	5-HT, NE(α2), opioid
Sensory cortex	Few	?	GLUT	5-HT, NE(α2), opioid
VLO	Few	RVM, PAG	GLUT	5-HT, NE(α2), opioid
NOTE: A1 to A7 designation are noradrenergic nuclei as defined by Dahlström and Fuxe.⁶⁴
5-HT, serotonin; ACC, anterior cingulate cortex; ACh, acetylcholine; CCK, cholecystokinin; CRF, corticotropin-releasing factor; DA, dopamine; DRt, dorsal reticular nucleus; GABA, γ-aminobutyric acid; GLUT, glutamate; LC, locus coeruleus; LRN, lateral reticular nucleus; NA, not applicable; NE, norepinephrine; NGC, nucleus gigantocellularis; NGCα, nucleus reticularis gigantocellularis pars alpha; NpGC, nucleus paragigantocellularis; NRM, nucleus raphe magnus; NTS, nucleus tractus solitarius; PAG, periaqueductal gray; PBN, parabrachial nucleus; RVM, rostral ventrolateral medulla; VLO, ventrolateral orbital cortex.
From Randich and Ness¹; Mendell³; Le Bars et al.⁷; Zhuo²¹; Mason²²; Boadas-Vaelllo et al.²³; Lau and Vaughan²⁴; Ossipov et al.²⁵; Kwon et al.²⁶; Peirs and Seal²⁷; Basbaum and Fields²⁸,²⁹; Taylor and Westlund³⁰; Tsuruoka et al.³¹; Stevens et al.³²; Janss and Gebhart³³,³⁴; Randich and Aicher³⁵; Ren et al.³⁶; Aimone and Gebhart³⁷; Neugebauer³⁸; Senapti et al.³⁹; Kuroda et al.⁴⁰; Zhang et al.⁴¹; Zhang et al.⁴²; Hutchinson et al.⁴³; Millan⁴⁴; Gebhart and Randich⁴⁵; Thurston and Randich⁴⁶; Urban et al.⁴⁷; Schaible et al.⁴⁸; Ren and Dubner⁴⁹; Butler and Finn⁵⁰; Jennings et al.⁵¹; Robbins and Ness⁵²; Suzuki et al.⁵³; Gao and Mason⁵⁴; Kaplan and Fields.⁵⁵