Phantom Limb Pain: Pathophysiology and Treatment

Background
There were an estimated 1.6 million people in the US living with limb loss in 2005. This number is projected to increase to 3.6 million by 2050. Of these patients, the incidence of phantom limb pain (PLP) is estimated to be 42.2 – 78.8% (Houghton, 1999). The phenomenon was first described by Ambroise Pare, a French military surgeon, in 1552. The term “phantom limb pain” was coined later in 1872 by Silas Weir Mitchell, a Civil War surgeon (Flor H, 2002). PLP is commonly classified as neuropathic pain that is sensed in the amputated limb. It is often described by patients as a stabbing, throbbing, burning, and/or cramping sensation that may be positional or related to movement, and that may be elicited or exacerbated by physical or psychological factors. Recurring evidence has found that the sensations are often more intense in distal portions of the phantom limb. During evaluation the sensations should be characterized, scaled, and differentiated from other sensations such as non-painful phantom phenomena, residual limb pain, and non-painful residual limb phenomena (Flor H, 2002).

Risk Factors
There are a number of risk factors that increase the likelihood of developing PLP. For instance, it appears that females have a higher rate of developing PLP than males. The reason for this is not clearly understood, but theories include (1) confounding due to psychological factors such as catastrophizing (catastrophizing is associated with increased PLP) and (2) a higher rate of reporting PLP by females. In addition, PLP appears to occur more frequently in upper extremity than lower extremity amputees and more often co-exists with residual pain the in remaining limb (Subedi B, 2011). The rate is also increased in amputees who experience pain prior to amputation, especially in amputees with chronic pain. The good news is that PLP appears to dissipate over time. The most common onset is within 1 month of amputation with a second peak rate of onset occurring 1 year after amputation (Flor H, 2006).

Peripheral mechanisms are postulated to occur at the level of (1) the individual peripheral nerves that are severed during amputation and (2) at the level of the dorsal root ganglia.

Injury to the nerve is associated with changes in the electrical properties of the cell membrane including upregulation or novel expression of Na channels, altered trafficking at the Na channels, decreased expression of K channels, and altered transduction at the membrane level. The changes in transduction are suggested by evidence that demonstrates ectopic discharges that occur in relation to stimulation of the stump by pressure and temperature. The combination of these changes results in ectopic or spontaneous discharges and hyper-excitability (Flor H, 2006).The changes occurring within a peripheral nerve that is damaged or nearby damaged nerves have been supported by a number of clinical and animal studies investigating PLP and neuropathic pain. One study conducted by Nystrom and Hagbarth demonstrated that tapping on a neuroma increased activity in afferent C-fibers and increased pain sensation within the phantom limb (Nystrom B, 1981).

Another study conducted by Zimmerman demonstrated that in regenerating C-fibers, there is an early development of chemosensitivity to various substances including bradykinin, histamine, serotonin, capsaicin, and many other chemicals that also excite normal nociceptors in skin or muscle. However, when exposed to adrenaline prior to bradykinin there was a greatly enhanced response, which is not seen in unaffected nerves, suggesting that changes occur as the nerve regenerates (Zimmerman M, 2001).

Lastly, the theory of peripheral nerve involvement in PLP is supported by studies demonstrating that drugs that block Na channels reduce phantom pain. However, clinical studies have also demonstrated that although anesthetic blockade of a neuroma eliminated stump pain, it did not eliminate PLP. In addition, PLP often presents before a palpable, swollen neuroma is formed, which has prompted a search for additional mechanisms that are more proximal to the residual limb (Flor H, 2006).

The DRG is an additional site of ectopic discharges that is thought to be involved in various aspects of PLP. Firstly, it appears that ectopic discharges from the DRG can amplify ectopic discharges from the residual limb. It also appears that activity at the DRG can cause cross-excitation resulting in depolarization of neighboring neurons. The result is increased intensity of painful stimuli in the stump and phantom limb (Flor H, 2002).The sympathetic nervous system might also contribute to the model and is related to sympathetic-sensory coupling at the level of the neuroma and at level of the DRGs. Current mechanisms include sympathetically triggered ephaptic transmission, sympathetic activation of nociceptors, and activation of low threshold mechanoreceptors that trigger sensitized spinal cord neuromas.  In addition, sympathetic discharge can elicit and exacerbate ectopic neuronal activity, accounting for exacerbation during emotional distress. This model is supported by evidence that systemic adrenergic blocking agents sometimes reduce PLP and injections of adrenaline into stump neuromas increase PLP in some patients (Costigan M, 2009).

Central mechanisms are also thought to play a role in PLP since studies show that local anesthesia of the stump, and epidural anesthesia do not eliminate ongoing PLP in all amputees. For instance, ne study showed that only 50% of amputees with PLP were pain-free after a brachial plexus block (Flor H, 2006).

One category of central mechanisms involves changes at the level of the spinal cord. Most of the evidence for changes at the level of the spinal cord is from experimental data from animal studies. One of the primary changes thought to occur in PLP and neuropathic pain is central sensitization, which refers to an increase in the excitability of neurons within the CNS. The change is triggered by a burst of activity in nociceptors, which alters the strength of synaptic connections between the nociceptor and the neurons of the spinal cord. This principle is called activity-dependent synaptic plasticity. So first, because of increases in synaptic efficacy and reductions in inhibition, a central amplification occurs enhancing the pain response to noxious stimuli in amplitude, duration, and spatial extent resulting in hyperalgesia (Figure 5). In addition, changes occur within low-threshold mechanoreceptors that are activated by light touch (such as when touching the skin lightly with a feather). Normally these receptors activate a pathway that senses the feeling of touch. However, once these receptors undergo central sensitization, the pathway crosses with nociceptor pathways and activate neurons in the spinal cord or brainstem that normally only respond to noxious stimuli (Figure 6). Therefore, an input that would normally evoke an innocuous sensation now produces pain (Woolf C, 2011).

In addition, nerve injury may cause central hyper-excitability, which results from increased firing of dorsal horn neurons, structural changes at the central endings of primary sensory neurons, reduced spinal cord inhibitory processes because of destruction of inhibitory GABA and glycinergic interneurons in SC from rapid ectopic discharges, and downregulation of opioid receptors on primary afferent and intrinsic spinal neurons which also results in decreased action of inhibitory GABA and glycine activity (Flor H, 2006).Lastly, the “windup phenomenon” contributes to the changes that occur at the spinal cord. Essentially, this results in a loss of target neurons at the level of the spinal cord which transmit descending inhibitory signals from the supraspinal level, and reduces local intersegmental inhibitory mechanisms. Ultimately this results in spinal disinibition and nociceptive inputs reaching the supra spinal centers (Flor H, 2006).

There are thought to be a number of changes that occur in the brain after an amputation that contribute to the sensation of PLP. One of the best studied changes is called topographical remapping in which neurons from somatotopically adjacent areas invade the deafferented region. Studies using fMRI have demonstrated this concept by showing invasion of adjacent areas into the representation zone of the deafferentated limb. For instance, one study demonstrated the reorganization of the primary somatosensory and motor cortex in patients with unilateral arm amputation. Subjects were asked to pucker their lips at a metronome-paced speed while fMRI images were taken.  A shift of the mouth representation into the hand representation occurred only in amputees with PLP; not in amputees without PLP or in healthy controls (Flor H, 2002).

Another principle that influences PLP is called point-to-point correspondence. Ramachandran et al suggested that referred sensations in the phantom limb are perceptual correlates of reorganizational processes in the cortex by showing a point-to-point correspondence between stimulation sites and areas of sensation from the mouth to the phantom limb in arm amputees. The larger the shift of the mouth representation into the zone that formerly represented the arm, the more pronounced the PLP. Thus, there appears to be a close association between cortical reorganization and the magnitude of PLP (Ramachandran VS, 1998).In addition, Melzack’s neuromatrix theory proposes that an extensive network of neurons connecting the thalamus, somatosensory cortex, reticular formation, limbic system, and posterior parietal cortex, creates an anatomical representation of the self that provides information about the body and its sensation. The sensory input from the neuromatrix is used to create a neurosignature for each region of the body. The neurosignature is specific for an individual and is thought to be genetically determined, but can be modified by experience. Amputation creates an abnormal input into the neuromatrix because of the abnormal firing pattern of damaged nerves, while the neuromatrix continues to send an outgoing signal. The discrepancy and abnormal sensory input may contribute to the perception of pain (Bittar R, 2005).

Still, another theory thought to contribute to PLP is that of illusory perception, in which cortical reorganization is thought to be affected by perceiving a foreign object as part of the body. In one instance, illusory perception was studied by examining fMRI changes when amputees perceived a rubber hand as part of their own body. Evidence supporting this theory suggests that frontal and parietal areas are involved in perception of abnormal somatosensory phenomena and that abnormal painful sensations are related to incongruence of motor intention and sensory feedback. Another study used a mirror to create a discrepancy between actual and seen movement and found that the presence of painful and non-painful parasthesias resulted as a consequence of the incongruent movement (Giummarra MJ, 2010).

Finally, the theory of telescoping, which refers to shrinking and retraction of the phantom limb towards the residual limb, is thought to contribute to the perception of PLP. It was originally thought to be sign of adaptive plasticity, however, studies now suggest that it is associated with increased intensity of PLP. The image below (Figure 8) shows brain activation related to imagine movement (represented by IM) of telescoped and non-telescoped phantom limbs and cortical activation related to executed movements of the intact hand represented by EM. The red shading (primary somatosensory cortex) follows the perceived location of the movement (which was opening and closing the hand) and not the actual anatomical location of the hand. Completely telescoped phantom limbs create activity in the cortical region of the arm while non-telescoped phantom limbs create activity in the cortical region of the hand. fMRI data shows that the cortical representation of phantom limb movement actually follows the location of perceived movement and is not statically fixed in the cortical representation zone of the hand. Thus, the study supports the theory that visual, sensory, and motor feedback to the cortex might be an important determinant of the phantom limb phenomena.

In addition to peripheral and central mechanisms of PLP, clinical studies have also suggested a strong psychogenic component. One of the primary psychogenic mechanisms that has been explored is called the pain memory hypothesis, which suggests that pain memories established prior to amputation are powerful elicitors of PLP. Evidence suggests that long-lasting noxious input might lead to long-term changes at the cortical level affecting sensory-discriminative aspects of pain. In addition, there is also thought to be reorganization that occurs in regions of the brain associated with affective-motivational aspects of pain including the insula, anterior cingulate and frontal cortices. These mechanisms are thought to contribute to causation and may also influence the course and severity of PLP (Flor H, 2002).

Currently, the standard of care is to provide a multi-modal approach to treat PLP. The mainstay of treatment is still pharmacotherapy and includes analgesics (acetaminophen, NSAIDs, opioids), antidepressants (tricyclic antidepressants, selective serotonin reuptake inhibitors, serotonin/norepinephrine reuptake inhibitors), anticonvulsants (gabapentin, carbamezapine), and NMDA receptor antagonists (Memantine). Additional approaches and more novel therapies include transcutaneous electrical nerve stimulation (TENS) and biofeedback, integrative, and behavioral methods such as mirror therapy and memory formation therapy (Flor H, 2006).