Joint Pain : Basic Mechanisms

Pain Sensation in the Normal Joint and during Joint Disease

Sensory information from joints is involved in the sense of movement and position and in control of movement, but most of this information does not reach consciousness. The major conscious sensation from joints is pain. Deep tissue pain is often dull, aching, and poorly localized and is thus different from cutaneous pain ( ). In a normal joint, pain is typically elicited by twisting or hitting the joint. Experimental invasive sensory testing in conscious humans has revealed that pain in a normal joint can be elicited by the application of noxious mechanical and chemical stimuli to fibrous structures such as ligaments and the fibrous capsule ( , , , ). Pain is not evoked by stimulation of normal cartilage and rarely by stimulation of normal synovial tissue ( ). Innocuous mechanical stimulation of fibrous structures can evoke pressure sensations ( ). Joint pain can be referred from other deep tissue such as muscles, tendon, fascia, other joints, and ligaments ( ).

Typical symptoms of inflammatory joint diseases are hyperalgesia and/or persistent pain at rest, which is usually dull and poorly localized ( ; ; ). Noxious stimuli cause stronger pain than normal, and pain is even evoked by mechanical stimuli whose intensity does not normally elicit pain (i.e., movements in the working range and gentle pressure, such as palpation).

An even more frequent cause of joint pain is OA ( ). Although the cartilage is initially damaged, the whole joint is afflicted in the long term, and hence pain may originate from different structures. Some authors describe severe inflammatory processes such as infiltration of osteoarthritic joints with inflammatory cells ( ), and the profile of elevated cytokine production may be similar to that of inflammatory disease ( ). Thus it is conceivable that pain during OA has a strong inflammatory component. Initially, pain in osteoarthritic joints is elicited by movement and loading of the joint, but at later stages pain may occur at rest (Scott 2006, ).

Experimental Models

To investigate the neuronal mechanisms involved in clinically relevant joint pain, models of joint inflammation and OA are being used. Acute inflammation in the joint can be induced by the intra-articular injection of crystals such as urate and kaolin or by injection of carrageenan. Injection of kaolin and carrageenan (K/C) into the joint produces edema and granulocytic infiltration within 1–3 hours, with a plateau reached after 4–6 hours. In awake animals, limping develops in the injected joint, and the mechanical threshold for withdrawal responses at the injected knee is significantly lowered within the first 4 hours and remains stable for at least 12 hours when the inflammation is fully established. In the K/C model, identified neurons can be recorded throughout the development of inflammation to directly monitor the generation of hyperexcitability ( ). Injection of Freund’s complete adjuvant (FCA) into a single joint produces a monarthritis that is present for 2 to 4 weeks. Usually, the lesion is restricted to the injected joint. Hyperalgesia (limping or guarding of the leg, enhanced sensitivity to pressure on the joint) develops within a day, reaches a peak within 3 days, and is maintained to some degree for up to several weeks ( ). After injection of high-dose FCA into the tail base or lymph node, polyarthritis develops ( ). More recently, models of chronic arthritis that are used in basic rheumatology research are now being used in pain research, namely, collagen-induced polyarthritis (CIA) and antigen-induced monarthritis (AIA). CIA is elicited by immunization with type II collagen in FCA, which induces an autoimmune disease directed against the cartilage in the joint, and it exhibits prominent involvement of B cells and the innate immune system and activation of the complement cascade ( ). To evoke AIA in rats or mice, animals are immunized against the antigen methylated bovine serum albumin (m-BSA). Three weeks later, m-BSA is injected into the knee joint. The unilateral AIA elicited in this manner is mainly T cell dependent, has an acute phase (characterized by invasion of granulocytes and fibrin exudation), and spontaneously progresses to chronic inflammation (characterized by infiltration of mononuclear cells, synovial hyperplasia, fibrosis in periarticular structures, and some cartilage and bone destruction) ( , ). Both CIA and AIA are characterized by mechanical hyperalgesia at the inflamed joint, forms of secondary mechanical and thermal hyperalgesia ( , ), disturbances in gait ( , ), and other signs such as vocalization with pressure (Neugebauer et al 2007). K/BxN serum transfer arthritis, recently introduced into pain research, produces a time-dependent shift from inflammatory pain to a pain state that is only gabapentin sensitive (i.e., probably neuropathic; ).

Experimental osteoarthritic pain has been studied mainly in the mono-iodoacetate (MIA) model and in surgically induced OA. One day after injection of MIA into the joint (MIA inhibits glycolysis and is toxic to chondrocytes), chondrocytes are shrunken and show fragmented pyknotic nuclei, the synovial membrane is expanded by fibrin proteinaceous edema fluid, and the joint is mildly infiltrated by lymphocytes, macrophages, and plasma cells. Some days later, the inflammatory response in the synovium subsides, necrotic cartilage collapses, and chondrocytes are lost. Osteoclastic activity is increased, subchondral bone collapses, and fragmentation of bony trabeculae surrounded by osteoclasts with some replacement by fibrous tissue and newly laid trabecular bone occurs ( , ). The MIA model was established in pain research by the demonstration of long-lasting mechanical hyperalgesia, as is evident, for example, from assessment of weight bearing ( , , ). Because of its rapid time course, MIA is clearly different from slowly developing human OA, and it displays substantial differences from human OA in gene arrays ( , ). At least the initial mechanical hyperalgesia may represent inflammatory pain (e.g., , ). Neuropathic pain components were proposed by the demonstration of activated transcription factor 3 (ATF3) immunoreactivity in dorsal root ganglion (DRG) neurons (a marker of nerve injury) at 8 and 14 days ( ), up-regulation of galanin and neuropeptide Y, down-regulation of substance P and calcitonin gene–related peptide (CGRP) in DRG neurons (a pattern typical of neuropathy) ( ), and gabapentin sensitivity of the hyperalgesia ( ). Large doses of MIA (more than 1 mg in the rat) may lead to damage outside the joint and therefore make it difficult to interpret. OA induced by transection of the anterior cruciate ligament, lateral meniscectomy, partial meniscectomy, a meniscal tear, or any combination of these measures is often used in OA research in dogs, sheep, and guinea pigs ( , , ), as well as in pain research in rodents ( , ). The destabilization OA model shows similar changes in neuropeptides in DRGs as in the MIA model ( ). Recordings from joint afferents were also taken from guinea pigs in which OA develops spontaneously during aging ( ).

Innervation of Joints

Joints are innervated by nerves descending from main nerve trunks or by muscular, cutaneous, and periosteal branches. A typical joint nerve contains thick myelinated Aβ (group II), thin myelinated Aδ (group III), and a high proportion (≈80%) of unmyelinated C fibers that are either sensory (group IV) afferents or sympathetic efferents (each ≈50%) ( ). Articular Aβ fibers terminate as corpuscular endings of the Ruffini, Golgi, and Pacini type in the fibrous capsule, articular ligaments, menisci, and adjacent periosteum ( ). Aδ and C fibers terminate as free nerve endings in the fibrous capsule, the adipose tissue, the ligaments, the menisci, the periosteum, and the synovial layer, but the cartilage is not innervated ( ). Typical free nerve endings in the joint are ensheathed by Schwann cells, but some sites appearing as a string of beads are not covered, thus suggesting that these areas are receptive sites ( ).

In contrast to cutaneous afferents, the vast majority of articular sensory neurons are peptidergic and isolectin B4 (IB4) negative ( ). The major neuropeptides in joint nerves are substance P, CGRP, and somatostatin; others are neurokinin A, galanin, neuropeptide Y, and enkephalin. Neuropeptides influence the inflammatory process in the periphery and modify the peripheral and spinal nociceptive processes ( ).

Mechanosensitivity of Joint Afferents in Normal Joints

The mechanosensitivity of joint afferents was assessed mainly in articular nerves of the cat and rat knee and the rat ankle joint. Innocuous mechanical stimuli included the application of light to moderate pressure on the exposed joint and movements within the working range (usually not painful). Noxious and painful stimuli consisted of local pressure at high intensity and movements exceeding the working range of the joint, such as twisting against the resistance of the tissue.

Figure 44-1 shows types of joint afferents according to their sensitivity to movement ( ). The low-threshold Aδ fiber in Figure 44-1 A with two receptive fields in the fibrous capsule (dots) responded phasically to extension and tonically to inward rotation within the working range of the knee joint. The strongest responses were elicited by movements such as noxious inward rotation. Such neurons are also activated by light pressure on the receptive field. The Aδ fiber in Figure 44-1 B with a receptive field in the patellar ligament showed a weak response to outward rotation in the working range and a strong response to noxious outward rotation. The C fiber in Figure 44-1 C with a receptive field in the fibrous capsule responded only to noxious outward rotation. Such neurons also require high pressure to elicit a response when probing the receptive field. The Aδ fiber in Figure 44-1 D with a receptive field in the anterior capsule responded only to local noxious pressure, not to innocuous or noxious movements. A further group of sensory neurons are mechano-insensitive under normal conditions. These so-called silent nociceptors become mechanosensitive during inflammation (see later). Figure 44-2 displays the proportions of mechanosensitive A and C fibers in the categories defined in Figure 44-1 (excluding initially mechano-insensitive fibers). Most Aβ fibers were either strongly or weakly activated by innocuous stimuli, but more than 50% of the Aδ fibers and about 70% of the sensory C fibers were classified as high-threshold units ( ; ).

Many low-threshold Aβ and Aδ fibers in the fibrous capsule and in ligaments (e.g., the anterior cruciate ligament; ) fire in the innocuous range but have their strongest response in the noxious range. Responses in the innocuous range may control movements and prevent non-physiological movements. The discharge rate encodes the strength of a particular movement from the innocuous into the noxious range. However, the most adequate innocuous movement can evoke more impulses than a noxious movement in another direction.

Changes in Mechanosensitivity during Inflammation of the Joint

An inflamed joint hurts during movement in the working range and during palpation, and pain may occur under resting conditions. An important mechanism for the heightened pain sensitivity is the increase in mechanosensitivity in joint afferents. Unlike cutaneous nociceptors (e.g., ), joint nociceptors can be reliably sensitized for mechanical stimuli. During the development of inflammation, the responses of some low-threshold Aβ fibers to movement increase transiently in the initial hours of inflammation. These fibers do not develop resting discharges. Many low-threshold Aδ and C fibers show persistent increases in response to movements in the working range. Most strikingly, a large proportion of high-threshold afferents (see Fig. 44-1 C and D) are sensitized such that they respond to movements in the working range of the joint. Many units develop ongoing discharges in the resting position ( ; ; ). Increased mechanosensitivity has also been found during chronic forms of arthritis, thus suggesting that mechanical sensitization is an important neuronal basis for chronic persistent hyperalgesia of the inflamed joint ( , ).

Furthermore, initially mechano-insensitive afferents (silent nociceptors) are sensitized and become mechanosensitive ( , ). The C fiber shown in Figure 44-3 was identified by electrical stimulation of the joint nerve, but initially it did not respond to mechanical stimulation of the joint (control). However, during inflammation it began to respond to movements, and a receptive field became detectable (see dot in Fig. 44-3 C). Thus, silent nociceptors are recruited for the encoding of noxious events during an inflammatory process. A rough estimate is that about one-third of the sensory C fibers and a small percentage of the Aδ fibers are initially mechano-insensitive silent nociceptors.

There is some evidence that the mechanosensitivity of joint afferents is also increased in the course of OA. In the MIA model, the amount of sensitization of joint afferents was found to correlate with the MIA dose ( ). In a comparison of fiber samples, old guinea pigs with spontaneous OA exhibited increased firing of joint afferents on noxious movements, but overall there was no correlation between joint nociception and articular damage in this model ( ).

Chemosensitivity of Joint Afferents

Aδ and C fibers of the joint express receptors for inflammatory mediators that are produced and released in the joint during pathophysiological conditions. Such mediators can excite and/or sensitize joint afferents. Notably, joint afferents also express receptors with inhibitory function. Thus, attenuation of mechanosensitivity can be achieved either by reducing the effect of excitatory/sensitizing mediators or by the application of agonists with inhibitory function.

Table 44-1 shows the effects of mediators on the response properties of afferent fibers of the knee and ankle joint that have been assessed with in vivo recordings. The classic inflammatory mediators bradykinin, prostaglandin E ₂ (PGE ₂ ) and I ₂ (PGI ₂ ), and serotonin excite subpopulations of high- and low-threshold Aδ and C fibers, but not Aβ fibers, and/or sensitize them for mechanical stimuli. Even though each mediator has its own profile of effect (e.g., duration), the mediators do interact and produce synergistic actions. After bolus injection into the joint artery, bradykinin excites joint afferents briefly (typically for less than 1 minute) but sensitizes the fibers to mechanical stimuli for several minutes even if it does not excite the neuron ( ). Both PGE ₂ and PGI ₂ cause ongoing discharges and/or sensitization to mechanical stimulation of the joint. The effect of PGE ₂ has a slow onset and a duration of minutes, and the action of PGI ₂ begins quickly and has a short duration ( , ). In the rat ankle joint, PGI ₂ excites and sensitizes a much larger proportion of units than PGE ₂ does ( ). In addition, these prostaglandins sensitize joint afferents to the effects of bradykinin regardless of whether they have an excitatory effect by themselves ( , ). PGE ₂ and bradykinin together can cause stronger sensitization to mechanical stimulation than either bradykinin or PGE ₂ alone ( ). Conversely, prostaglandin synthesis inhibitors such as aspirin and indomethacin reduce spontaneous discharges from acutely and chronically inflamed joints and attenuate the responses to mechanical stimulation ( , ). Bradykinin receptors are also involved in mechanical sensitization evoked by proteinase-activated receptor 4 ( ).

Table 44-1 shows the effects of various mediators. Acid-sensing ion channel 3 may be involved in the increase in mechanosensitivity because ASIC3 gene knockout mice exhibited reduced arthritis-induced mechanical hyperalgesia ( ). Cannabinoid receptor agonists reduce mechanosensitivity, which may be of therapeutic relevance. Note, however, that the CB1 receptor agonist anandamide may also activate transient receptor potential vanilloid 1 (TRPV1) receptors. Substance P and vasoactive intestinal peptide increased and endomorphin-1 and somatostatin reduced mechanosensitivity in numerous afferents. Endogenous opioids seem to delay the occurrence of pain in experimental OA ( ), and intra-articular opioids produce profound analgesia ( , ). Activation of somatostatin receptors may be an interesting option for future pain therapy ( , , ). Galanin, neuropeptide Y, and nociceptin sensitized some neurons and reduced responses in others; whether the particular pattern is dependent on the functional state of the neuron is not known.

Pro-inflammatory cytokines such as tumor necrosis factor-α (TNF-α), interleukin-1β (IL-1β), and IL-6 play a major role in the pathogenesis of immune-mediated arthritis, and they are involved in human OA ( , ) and OA models ( ). A single injection of either TNF-α ( ) or IL-6 ( ) into a normal knee joint sensitized Aδ (TNF-α) and C fibers (TNF-α and IL-6) to innocuous and noxious rotation of the knee joint at a slow time course (taking about 1 hour to develop). The mechanical sensitization induced in this manner persisted for hours, which contrasts with the short-lasting increases in mechanosensitivity after the intra-arterial injection of “classic mediators” (see earlier). The recombinant TNF (p75) receptor–Fc fusion protein etanercept (which neutralizes TNF) prevented TNF-α–induced mechanical sensitization and partly reversed it once it was established ( ). In rat AIA ( ), mouse CIA ( ), and mouse K/BxN ( ) arthritis, neutralization of TNF reduced mechanical hyperalgesia. Injection of etanercept into the AIA joint decreased the responses of joint afferents to rotation within 1 hour, a rapid effect most likely caused by direct influences on the neurons themselves ( ).

Mechanical sensitization by IL-6 was enhanced by co-administration of the soluble receptor sIL-6R, which allows trans-signaling of IL-6. Soluble gp130 (sgp130), which prevents trans-signaling, prevented mechanical sensitization by intra-articular IL-6/sIL-6R but did not reverse established IL-6–induced mechanical sensitization ( ). Intra-articular sgp130 pretreatment attenuated the development of mechanical hyperalgesia in AIA, but systemic application of sgp130 after the onset of AIA only slightly reduced mechanical hyperalgesia ( ). Thus, sensitization by TNF-α is more reversible than that by IL-6.

In a recent study neutralization of nerve growth factor (NGF) with an antibody significantly inproved OA pain in humans ( ). It is likely, therefore, that NGF has a role in the generation of joint pain. Although the sensory endings are the target of mediators, DRGs may be another target. In the acute phase of inflammation, ED1-positive macrophages (rat ED1 is equivalent to human CD68; , ) invade the lumbar DRGs (which supply the knee joint) in the absence of neuronal damage, and this process depends on TNF-α ( , ). The invasion was bilateral, and although the inflammation was unilateral, it was associated with bilateral de novo expression of vascular cell adhesion molecule 1 (VCAM-1) in endothelial cells in lumbar DRGs, and it correlated with mechanical and thermal hyperalgesia on both sides ( ). Symmetrical pain with unilateral pathologic changes was observed experimentally and clinically ( ). Since TNF up-regulates the expression of TRPV1 in cultured DRG neurons ( ), the macrophage invasion may enhance the synthesis of molecules relevant for nociception.

Which receptors and receptor subtypes are expressed in joint afferents has not been fully explored. Furthermore, the proportion of neurons responding to a mediator can be different under control or inflammatory conditions. This could indicate regulation of receptor expression or mediator interactions. Given the complexity of the changes in mechanosensitivity, an alternative approach to pain therapy is being used clinically in patients with OA, namely, intra-articular injection of hyaluronan solutions. Although comprehensive meta-analyses either did not reveal a significant effect ( ) or stated that hyaluronan had a small effect on pain ( ), nerve fiber recordings did document a reduction in the discharge of fibers after such intra-articular injections ( ). The mechanisms are unclear. Gabapentin also reduced the responses of joint afferents in normal and inflamed joint ( ).