Jon Y. Zhou, MD
Assistant Clinical Professor
Department of Anesthesiology and Pain Medicine
U.C. Davis Medical Center
Phantom limb pain (PLP), a condition defined as chronic, severe pain in parts of the body that no longer exist secondary to amputation was described by a French surgeon, Ambroise Pare, in 1551. He noted that soldiers who had amputated limbs would develop chronic pain referred to the absent limb after surgery. Although PLP generally follows amputations in the limbs, it may develop in any body part including the breast, tongue, or even parts of the GI tract. PLP may develop anytime from days to years after the initial surgery. However, in the majority of patients, PLP will develop within 1 week after amputation with a high percentage developing sensations within 24 hours. PLP is a chronic pain syndrome with significant long-term ramifications. It is estimated that approximately 185,000 limb amputations occur in the United States each year, with a total of more than 1.5 million patients total with limb loss in the US. Treating patients with PLP, although challenging, may have a significant impact on our patients’ quality of life and decrease the cost of health care in the future. The most successful treatments include multidisciplinary measures with a combination of physical therapy, medication management, and, if indicated, interventional procedures.
Patients who have undergone limb amputation often have a combination of stump pain, phantom limb sensations, and PLP postoperatively. PLP is often described as a neuropathic pain state, with sensations of sharp, stabbing, burning, twisting, or cramping pain that is worse in the distal part of the amputated limb. PLP is distinguished from phantom limb sensation, in which the patient’s brain processes the sensations of the missing limb but does not have pain. Not all phantom limb sensations are unpleasant; some are feelings of warmth, cold, itching, etc. Stump pain is the somatic pain that occurs from the surgical incision and is easily localized and different from PLP. Stump pain may occur from neuroma formation at the scar site, pain from chronic prosthesis use, or pain from the surgical incision. Patients after amputation may suffer from a combination of stump pain, phantom limb sensation, and PLP, which may lead to a confusing picture when patients present for treatment. Despite multiple treatment modalities, the incidence of long-term PLP ranges from 43% to 79%. Although some studies show that PLP decreases over time, there is no consensus about the long-term outcome of this painful condition.
The etiology of the PLP is multifactorial and includes a combination of peripheral, central, and psychological factors. Some risk factors for PLP include pre-amputation pain and uncontrolled acute pain postoperatively. Females and patients having amputations of the upper extremity have higher incidence of PLP.
The mechanisms of PLP are multifactorial in nature with changes both peripherally and centrally. In addition, there is a psychosocial component to the progression of PLP that may exacerbate preexisting pain states. The progress from amputation to PLP is likely not attributed to one factor but a combination of central and peripheral neural mechanisms.
Peripherally, the surgical incision will acutely sever the arteries, tissues, and nerves, which leads to an immediate inflammatory response at the level of the spinal cord. Deafferentation is described as a loss of sensory input from a portion of the body from severing of the nerves. Deafferentation pain usually displays varying degrees of sensory loss characterized by disturbances in pain and temperature sensation. In addition to sensory loss, patients may have allodynia, hyperalgesia, hyperpathia, and dysesthesias. The mechanism may be due to the formation of neuromas, which form at the nerve sprouts of the severed limb. There is enhanced expression of sodium channels at these neuromas that are hyperexcitable and may lead to progression of chronic pain.
Centrally, patients with PLP may have cortical reorganization, which leads to progression of this disease. This may be preceded by spinal cord sensitization at the level of the dorsal root ganglion (DRG). In addition, sprouting of new neurons not initially injured at the peripheral level at the level of the spinal cord may further the pain cycle. There will also be increased levels of pro-inflammatory enzymes, such as substance P and neurokinins at the dorsal horn, and decreased levels of inhibitory enzymes, such as GABA. These changes at the spinal cord are separate from the changes in the level of the cerebral cortex. Cortical reorganization is the adaptation of the cortex for the parts of the brain that stop receiving information due to the amputation of the limb. This somatosensory cortex evolution may describe the sensations of PLP in certain patients.
Lastly, patients who are poor adaptors and have increased levels of anxiety and depression may have a worse prognosis with PLP compared to patients with better coping skills.
Based on information from Kern, Busch
Treatment of PLP is challenging, and a multimodal approach that targets the psychosocial, peripheral, and central mechanisms of this pain state may best benefit patients.
Pharmacologic agents for PLP include traditional opioids, nonsteroidal anti-inflammatory drugs (NSAIDs), calcium channel blockers (pregabalin, gabapentin), local anesthetics (lidocaine), tricylic antidepressants (amitriptyline), and NMDA receptor antagonists (ketamine, memantine). Opioids are effective for producing analgesia postoperatively but have not been proven to alleviate PLP or stop the progression of this disorder long term. Higher opioid requirements preoperatively have been associated with higher rates of PLP progression and, as expected, higher postoperative opioid requirements. Despite this, opioids were shown to provide superior postoperative analgesia compared to placebo in several case series. NSAIDS such as ibuprofen work by inhibition of prostaglandin synthesis and decrease the perception of nociception at the peripheral and central levels. Although NSAIDS are used widely in other pain states, there is little evidence that this class of medications is effective in the treatment and prevention of PLP.
Typically tricyclic antidepressants are the first-line pharmacologic agents used to treat neuropathic pain and have been hypothesized to also help treat PLP. The mechanism of action involves inhibition of serotonin-norepinephrine reuptake, NMDA receptor antagonism, and sodium channel blockade. A study by Robinson et al studied 39 patients receiving a six week trial of amitriptyline, up to 125 mg/day, but failed to show any improvement in their pain compared to placebo. All patients in that study had chronic PLP greater than six months at presentation. However, other studies have showed dramatic improvement in PLP VAS scores after 1 month of TCA use. The role of TCA is clear in neuropathic states, but its role for PLP is less clear. Desipramine and nortriptyline are also TCAs similar to amitriptyline but with fewer centrally acting side effects, and they may help with neuropathic states. In summary, TCAs are helpful for neuropathic pain states but further research needs to be done for their role in the treatment of PLP.
Anticonvulsants such as gabapentin and pregabalin have been shown to be effective for neuropathic states, but the evidence for long-term treatment of PLP is equivocal. Studies that show efficacy have small sample sizes, with one randomized crossover study of 19 patients with PLP greater than six months who received either monotherapy with gabapentin up to 2,400 mg a day or placebo. Although the VAS was significantly lower in the gabapentin group, there were no differences in mood, sleep quality, and ability to perform activities of daily living. However, another RCT of 24 patients that compared gabapentin up to 3,600 mg a day versus placebo showed no difference. Although the studies are equivocal, perhaps the monotherapy treatment of a multifactorial pain state may explain the ambiguous results. Other anticonvulsants have been studied, including carbamazepine and topiramate, with positive treatment results for PLP but quality of evidence was low.
Calcitonin as an analgesic was first reported in 1975 when it was injected into the CSF of rabbits to treat vertebral fracture pain. The exact mechanism by which calcitonin inhibits pain is still unknown. Calcitonin has been theorized to inhibit prostaglandin release and cytokine release for an anti-inflammatory effect. Like many monotherapy agents, the research for calcitonin has both positive and negative studies, all of which are small. Jaer et al., in 1991, studied 21 patients who underwent surgical amputations and had long-term PLP. These patients received IV calcitonin (200 units over 20 minutes), up to three doses. One week after treatment, 19 of the 21 patients reported pain relief by at least 50%, and 16 (76%) were pain free. Adverse side effects were mild and included nausea and vomiting, which resolved with anti-emetics. Despite the positive outcome of this study, the study is underpowered, the authors did not state p values for observations, and the authors did not list the other medications that the patients were taking.
NMDA antagonists include ketamine, memantine, and dextromethorphan. IV ketamine infusions have shown clinical efficacy in multiple clinical trials. One RCT of 45 patients undergoing below-the-knee and above-the-knee amputations compared IV ketamine perioperative infusions to IV saline infusions and showed that, at 6 months, 47% of the ketamine infusion groups had PLP while 71% of the control group had persistent PLP. Patients started the ketamine at 0.5mg/kg/h intraoperatively and continued the infusion for 72 hours postoperatively at 0.5mg/kg/h. Another study of 31 patients by Galova et al, compared three groups: high-dose ketamine (0.1 mg/kg/h), low-dose ketamine (0.05mg/kg/h), or magnesium infusions. They showed improved efficacy with higher doses of ketamine.  The study demonstrated a 50% reduction in PLP with low-dose ketamine infusions and complete resolution with high-dose ketamine infusions. Katz et al compared the use of epidural ketamine and bupivacaine with bupivacaine alone and demonstrated a decrease in PLP immediately postoperative but no significant differences in pain scores at eight days to one year after the procedure. Memantine is an oral NMDA antagonist, typically used as a treatment for Alzheimer’s disease, that has been purported to be effective for PLP. However, there is not good evidence for the long-term benefits of this medication for PLP. Maier et al compared 36 patients receiving memantine vs placebo, and, although there was improved pain relief at 4 weeks compared to placebo, there was no difference at 18 months. The treatment group received oral memantine 30 mg daily for four weeks. Another study by Wiech et al studied 8 patients with traumatic upper-extremity amputations and showed no effect on the intensity of chronic PLP after 4 weeks of treatment with memantine. In summary, IV ketamine has been shown to be effective for neuropathic pain states and may play a role in the PLP, but further studies are needed with increased power. Memantine used orally may have short-term benefits but has not been shown to be effective for long-term treatment of PLP after a four-week regimen of 30 mg orally daily.
There are a wide variety of nonpharmacologic mechanisms for treating PLP, including invasive surgical procedures and noninvasive therapies such as physical therapy and transcutaneous electric nerve stimulation (TENS) units. TENS has been found to be effective in treating some of the symptoms of PLP based on case series. TENS works via the gate theory of pain control by causing non-noxious stimulation in order to prevent the brain from processing cutaneous pain sensations. Because TENS units are noninvasive and have minimal side effects, they are an appropriate nonpharmacologic method of treating pain. The TENS stimulates large diameter A-beta afferents in order to produce analgesia locally, inhibits nociceptive receptors, increases blood flow to the region, and reduces muscle spasms. TENS has been shown to be effective for other chronic pain states including chronic low-back pain, recurrent headache, and musculoskeletal pain. However, there are not many large, randomized controlled trials, with most of the research based on case series. Interestingly, Giuffrida et al. evaluated the use of a TENS unit on the contralateral limb that developed PLP. In a small case series of two patients, they showed that both patients had significant improvement in PLP after regular use of the TENS unit on the opposite limb. PLP has both spinal and supraspinal mechanisms, and it is believed that cortical remapping may play a role. Contralateral responses to unilateral neurological lesions are common and may be due to neurological mechanisms that cross in the spinal cord. This was a small trial that warrants further larger studies to evaluate the use of TENS units on the contralateral limb for PLP. In conclusion, TENS is often recommended as part of a multidisciplinary treatment plan for PLP due to its favorable side-effect profile and patient tolerance.
Mirror therapy is a specialized form of physical therapy in which a mirror is placed near the midline so that the patient moves the contralateral limb but visualizes the amputated limb moving due to the mirror reflection. It has been shown to be helpful in remapping the patient’s cortex, and multiple clinical trials have shown its efficacy in decreasing pain and increasing activities of daily living. Chan et al. evaluated 22 patients into three groups: one that viewed the reflected image on an intact mirror, one that viewed a covered mirror, and one group that was trained in mental visualization without a mirror. After four weeks of regular therapy, only the intact mirror imaging group had significant decreases in overall pain score. After four weeks, all groups received the intact mirror therapy, and, after eight weeks, all three groups showed similar decreases in VAS pain score.
More invasive treatments of pain are the last resort for patients who fail medical management and other conservative pain management techniques. Surgical procedures such as selective ablation of the dorsal spinal nerve root, anterolatral cordotomy, thalamotomy (resection of thalamus to prevent pain transmission), and sympathectomy have short-term benefit but no difference in long-term relief.[25-28] These procedures may have long-term side effects including worsening pain, nerve injury, and a high rate of relapse. These surgical procedures are typically reserved for terminally ill patients.
There has been more interest recently in the use of neuromodulation interventions to alleviate PLP including spinal cord stimulation (SCS), selective DRG stimulation, and peripheral nerve stimulation to decrease the perception of the PLP. SCS has be hypothesized to alleviate PLP. It works by the gate theory of control, first postulated by Melzack and Wall, which showed that activation of large myelinated fibers can inhibit the transmission of pain signals from the level of the spinal cord to the cerebral cortex. SCS has been used as a modality for treatment of PLP since 1969, with small case series presented in 1975. A recent case series in 2010 from MD Anderson evaluated the use of SCS for four patients with lower extremity amputations and persistent PLP and showed over 80% efficacy. Older studies with larger sample sizes include one by Krainick et al. who reported long-term outcomes for 61 patients who received SCS implants in the subdural space. Although 26 patients received 25% relief, 28% of the patients received no relief. Another study by Katayama et al. in 2001 evaluated 19 patients with PLP with six of the patients with long-term pain control with the SCS implants. A recent case report by McAuley showed that 11 of 12 patients had good short-term pain relief but only five had long-term relief. There are no large trials of the use of SCS for PLP.
DRG SCS is a more selective form of SCS in which epidural leads are placed near the sensory neurons of the DRG rather than the midline posterior epidural space where typical SCS leads are placed. The main advantage of this process is the ability to have more selective non-painful paresthesia coverage versus complete stimulation of the dorsal column with traditional SCS. Eight patients were selected to receive DRG stimulation, with all receiving adequate pain relief during the trial process and decreased pain scores. Leads were placed at L3-S1 DRGs for lower-extremity pain and at C6-C7 DRGs for upper-extremity pain. The average VAS score was 83 mm prior to implantation and decreased to 38.9 mm after the trial.
A newer technique of peripheral nerve stimulation involves surgically placing electrodes close to a peripheral nerve and stimulating electrical current through the leads, leading to a comfortable and non-painful paresthesia. Sixteen patients were evaluated, and 14 of the 16 had decreased pain with the peripheral nerve stimulation. A single 24-gauge electrode was inserted via a peripheral nerve block needle and connected to a battery-powered electrical stimulator. The current was adjusted so that the patients felt no motor contraction or uncomfortable sensations near the femoral and sciatic nerve. This new novel technique may provide long-term relief from PLP, but further larger, randomized, controlled trials need to be conducted.
Unfortunately, PLP can be a debilitating disease in which patients may have little relief from their constant pain with standard opioid pharmacological therapy. Newer neuromodulation techniques, such as DRG stimulation, peripheral nerve stimulation, and dorsal column SCS, may be indicated for patients who have intolerable side effects from medications or have inadequate pain control despite multimodal pharmacologic treatments. In these patients, the benefits of a SCS trial may outweigh the inherent risks of the procedure itself.
PLP occurs in a high percentage of untreated patients who have undergone limb amputation. Risk factors for progression of PLP include poorly pain prior to amputation, intensity of postoperative stump pain, and traumatic amputations. Women and patients receiving upper-extremity surgeries tend to produce a higher incidence of PLP as well. Both central and peripheral mechanisms may partially explain the etiology of PLP, but the exact mechanism is still unclear. This may explain the difficulty in treating this pain state due to its multifactorial nature. Two patients with similar demographics and risk factors may have different degrees of PLP. Currently, a multimodal approach with conservative treatments of TENS, mirror therapy, and pharmacologic treatments is a reasonable start. In patients who fail conservative treatments, neuromodulation with SCS, DRG stimulation, or peripheral nerve stimulation maybe options. Surgical treatments such as cordotomy  are usually the last option for patients who terminally ill. With new technologies emerging continually emerging, treatment of this debilitating condition may improve.