In 1983, Sherman and Sherman²⁴ published a survey of 590 war veteran amputees in which 85% reported phantom pain. A study with 2,694 amputees showed that 51% experienced PLP severe enough to hinder lifestyle on more than 6 days per month, 21% reported daily pain over a 10- to 14-hour period, and 27% for more than 15 hours per day.²⁵ The incidence of PLP increases with more proximal amputation. RLP is reported in up to 50% of amputees.^{14,26,27,28,29,30,31}

PLP is also a major contributor to overall morbidity in veterans and service members after major limb amputation. Of 298 veterans who served in the Vietnam war and 283 service members and veterans who served in Operation Iraqi Freedom/Operation Enduring Freedom (OIF/OEF), 72.2% (Vietnam) and 76.0% (OIF/OEF) reported PLP, and 48.3% (Vietnam) and 62.9% (OIF/OEF) reported RLP.³² Investigation of pain phenotypes in 124 active military service members after traumatic amputation injury found a significant association between neuropathic RLP and posttraumatic stress disorder (PTSD) and depression.³²

Predisposing factors for developing PLP have been more challenging to discern. Most prospective studies report that increased pain before amputation is associated with increased PLP after amputation. However, other studies report no such association. Differences in these findings may be due, in part, to the fact that patients enrolled in prospective studies generally have a preexisting medical disease necessitating the amputation and often exclude patients who suffer traumatic amputation resulting in an overestimation of the association between preamputation pain and PLP.³³

A retrospective study by Dijkstra et al.³⁴ of 536 amputees further characterized risk factors based on a logistic regression analysis, finding bilateral amputation and amputation of a lower limb to be the strongest risk factors for PLP. Anxiety as well as passive coping and catastrophizing personality traits were found in one study to be independent contributors to persistent postamputation pain.³⁵ In addition, there may be a sympathetic component to PLP triggered by stress, anxiety, depression, and other emotional states.³⁶

PLP has been reported to occur as early as 1 week after amputation and as late as 40 years after amputation.^37,38 However, most studies report that the onset of PLP occurs within the first week for 75% of patients, with onset occurring within the initial 24 hours after amputation for nearly 50% of patients.³³ Phantom pain may diminish with time and eventually fade away. However, some prospective studies indicate that even 2 years after amputation, the incidence is not greatly diminished from that at onset.^31,39

Phantom pain may be modulated by multiple factors, both internal as well as external. Exacerbations of pain may be produced by trivial, physical, or emotional stimuli. Anxiety, depression, urination, cough, defecation, sexual activity, cold environment, or changes in the weather may worsen PLP.^{24,39,40,41,42,43,44,45} It also has been reported that general, spinal, or regional anesthesia in amputees may cause appearance of phantom pain in otherwise pain-free subjects.^{46,47,48,49,50,51}

The mechanisms underlying phantom pain are complex and multifactorial. Although incompletely understood, the pathophysiology likely involves peripheral, spinal, and supraspinal mechanisms as well as psychological factors.

After amputation, the distal end of axotomized afferent fibers initially undergoes retrograde degeneration with subsequent regenerative sprouting. These enlarged and disorganized endings of C fibers and demyelinated A fibers often form neuromas that show an increased rate of both spontaneous, ectopic activity, and abnormal, evoked activity in response to
mechanical (e.g., pressure, Tinel sign) and chemical (e.g., norepinephrine) stimulation.⁵ The proliferation of heterotopic sodium channels (Nav1.3, Nav1.7, Nav1.8) may also lower the stimulation threshold and provoke ectopic discharge.⁵²

Clinically, peripheral causes of PLP is supported by the observation that RLP and phantom pain can be temporarily reduced in some (but not all) patients after injection of local anesthetics into stump neuromas. There have also been reports of alleviation of PLP after the removal of neuromas.⁵³ In contrast, two independent studies found that repetitive touching of the residual limb results in increased PLP.³³ Studies reporting the reduction of phantom pain with drugs blocking the sodium channels lend further support to this theory.^54,55

Ectopic discharges can also occur at the level of the dorsal root ganglion (DRG), independently of stump neuromas. This can result in amplification of peripheral signals and recruitment of neighboring neurons. Electrophysiology studies by Nyström and Hagbarth⁵⁶ have shown that blocking stump neuromas eliminates the percussion-evoked Tinel sign and associated evoked activity, but does not block the ongoing spontaneous discharge recorded in the nerve. When compared to neuromas, the DRG has proved to be a more robust source of spontaneous firing than neuromas.⁵⁷

The sympathetic nervous system may play a role in potentiating phantom pain. Sympathetic nerve blocks can temporarily relieve phantom pain in some patients, whereas injection of norepinephrine can exacerbate pain. Catecholamines can promote firing of peripheral mechanoreceptors, which in turn may activate sensitized dorsal horn neurons. Increased sympathetic tone can also promote ephaptic neuronal transmission in the periphery. Sympathetic tone is inversely related to skin temperature at the amputated stump.⁶ Investigators have also shown an inverse relationship between phantom pain and skin temperature, suggesting that sympathetic tone promotes pain sensation.⁶

Peripheral nerve injury is also accompanied by reorganization of signal processing at the spinal cord level. Selective degeneration of unmyelinated C fibers results in functional denervation of lamina II neurons in the dorsal horn. A compensatory arborization of Aδ and Aβ fibers “sprouting” into lamina II can occur.^58,59 This change in innervation is accompanied by phenotypic switching, whereby Aβ-fiber terminals release substance P, a nociceptive peptide. These changes may form the anatomic and neurochemical substrate for the clinical phenomenon of allodynia, where a non-noxious mechanical stimulus is perceived as painful.

Increased excitatory input at the dorsal horn following nerve injury can cause apoptosis of inhibitory interneurons expressing γ-aminobutyric acid and glycine. Activation of microglia after neural injury can result in release of brain-derived neurotrophic factor, which promotes phenotypic switching of inhibitory interneurons, and may lead to release of excitatory neurotransmitters (e.g., glutamate). Opioid receptors are also downregulated along with upregulation of cholecystokinin, which is an endogenous opioid receptor antagonist.

Nikolajsen and Jensen⁶⁰ explained that the pharmacology of spinal sensitization involves increased activity in N-methyl-D-aspartate (NMDA) receptor complex,⁶¹ and many aspects of the central sensitization can be reduced by NMDA receptor antagonists. This was supported in human amputees where the evoked stump or phantom pain produced by repetitive stimulation of the stump by nonnoxious pinprick was reduced by the NMDA receptor antagonist, ketamine.⁶⁰

Waxman and Hains proposed that abnormal expression of Nav1.3 sodium channels in the second- and third-order neurons along nociceptive pathways after spinal cord injury may make these neurons hyperexcitable.⁶² These neurons may then function as pain amplifiers/generators and conceivably contribute to phantom phenomena.

Peripheral injury after amputation is also accompanied by remapping of supraspinal synaptic networks, including those in the primary somatosensory cortex. Some patients with PLP exhibit “topographical remapping,” where stimulation of an unaffected site (e.g., face) will result in a sensation perceived in the phantom limb. Functional imaging studies have shown activation of areas in the primary somatosensory cortex and are both adjacent and distant from the area normally subserving the affected limb. Topographical remapping appears to correlate with persistence of phantom pain, with data showing that upper extremity amputees with phantom pain have expansion of the mouth area into the hand area in the sensory homunculus (primary somatosensory cortex). These findings have also been described in the primary motor cortex of amputees, with good correlation between reorganization and presence of phantom pain symptoms.

Melzack⁶³ observed that a substantial number of children who are born without a limb feel a phantom of the missing part and suggested the existence of a neural network, or neuromatrix, that subserves body sensation and has a genetically determined substrate that is modified by sensory experience. Lotze et al.⁶⁴ revealed that functional magnetic resonance imaging (fMRI) data from amputees without pain and healthy volunteers during a lip pursing task were similar. In amputees with PLP, however, the cortical representation of the mouth extends into the region of the hand and arm. Giummarra et al.⁶⁵ proposed that phantom pain may reflect a maladaptive failure of the neuromatrix to maintain global bodily constructs. Cortical map reorganization may be facilitated via selective loss of C fibers, which occurs after amputations. C fibers appear to have an important role in the maintenance of cortical maps.

Psychological factors can also play a role in the pathogenesis of phantom pain. Although these factors may not play a causative role, they may certainly modulate the pain experience.

Longitudinal diary studies showed that there is a significant relation between stress and the onset and exacerbation of episodes of PLP, probably mediated by activity in the sympathetic nervous system and increases in muscle tension.⁶⁶ Patients who received less support before the amputation tend to report more PLP.²⁷

Animal work on stimulation-induced plasticity suggests that extensive behaviorally relevant (but not passive) stimulation of a body part leads to an expansion of its representation zone.⁶⁷ Intensive use of a myoelectric prosthesis is positively correlated with reduced cortical reorganization and analgesic effects.⁶⁴ These effects could not be achieved with standard medical treatment and general psychological counseling because it is felt that in order to achieve analgesia, input into the amputation zone of the cortex is needed in order to “undo” the reorganizational changes induced by amputation. Similar beneficial effects on phantom pain and cortical activation were reported for imagined movement of the phantom and may also occur to some degree with mirror treatment (where a mirror is used to trick the brain into perceiving movement of the phantom when the intact limb is moved).⁶⁸

PLP cannot currently be completely prevented; however, perioperative epidural techniques, placement of peripheral nerve catheters, and other analgesic strategies utilized preoperatively, intraoperatively, and postoperatively may at least address postoperative pain control better than not employing any specific perioperative analgesic techniques.^{69,70,71,72,73,74,75}

Madabhushi and colleagues⁷⁴ reported on a patient with a history of PLP from a below-knee amputation who then came for an above-knee amputation in the same extremity. Before transection, the sciatic nerve was infiltrated with 0.25% bupivacaine 5 mL and
clonidine 50 µg. After the nerve was severed, a 20-gauge epidural catheter was inserted into the nerve sheath and externalized laterally through a separate skin incision. Before closure, 0.25% bupivacaine 10 mL and clonidine 50 µg was injected and then 0.1% bupivacaine and clonidine 2 µg/mL was infused perineurally for the first 96 hours postoperatively. The mean postoperative pain score (from 0 to 10) for 96 hours was 1.2 ± 0.7.⁷⁴ The patient required a total of 10 mg of oxycodone postoperatively. Over a 1-year follow-up period, the patient never reported stump or phantom pain.⁷⁴

Treatment for phantom pain often requires a multimodal approach. Treatment options range from behavioral techniques and pharmacologic treatments to invasive electrical brain and spinal cord stimulation (SCS) and surgical interventions. There remains a paucity of data from large randomized controlled trials to guide treatment options. As a general rule, initial treatments should be low-risk, low-cost, and noninvasive, with more expensive and invasive treatments reserved for patients who fail conservative care.