Many of the 2 million amputees in the United States suffer from chronic pain, either isolated to the residual limb itself or as phantom limb pain (PLP) perceived in the limb no longer present. Prevalence rates of residual limb pain vary widely, from 10% to 76%, while rates of PLP have been reported as high as 85%.1–4 Residual limb pain and PLP cause measurable decreases in prosthetic function and poor quality of life.5,6
Simplistically, residual limb pain is predominantly driven by cut nerve endings that form terminal-neuromas—disorganized axons encased in scar. Numerous treatments for neuromas have been described in the literature, though no single neuroma treatment has been shown to be consistently effective or superior. Previously reported pain management strategies have emphasized nerve ablation techniques with focused radiofrequency waves or injected neurotoxins. Alternative surgical approaches excise the neuroma and transpose the remaining nerve fascicles into a more favorable microenvironment such as bone, fat, vein, or even back onto itself.7,8 Of the surgical strategies for neuroma management, the most commonly considered for symptomatic neuromas is excision and burying the freshened nerve ending into a nearby healthy muscle.9 Common to all of these neuroma treatment procedures is the physiologic certitude that the freshly treated nerve will attempt to regenerate and subsequently will reform a new neuroma. Treatment success for these procedures requires that the newly created neuroma be less symptomatic than the neuroma that was removed.
Related to, but distinct from, residual limb pain is phantom limb pain. PLP is thought to be a complex interplay between the painful neuroma and multiple levels of the central nervous system resulting in cortical reorganization that has proven even more difficult than neuroma pain to prevent or reverse.10–14 While neuromodulators such as gabapentin may have some effect on phantom pain,15 2 recent meta-analyses failed to demonstrate a meaningful benefit to these or any other medical treatments.16–18
A conceptually different strategy for handling the terminal end of a divided nerve originated in a procedure first performed by Dumanian in 2002 called Targeted Muscle Reinnervation (TMR) for the brain control of advanced myoelectric prostheses for amputees.19 The terminal neuroma is removed and the newly freshened nerve is coapted to a newly divided nearby motor nerve. The fascicles, primed to regenerate, grow down the motor nerve to enter and re-innervate the newly denervated muscle.20,21 Some fascicles connect with motor end-plates, while others connect to the numerous sensory end organs such as proprioceptors that exist within the muscle. What distinguishes TMR from all other treatments of neuromas is that the fascicles of the mixed major and sensory nerves are channeled toward nerve receptor targets. Also important is the experience from surgeons performing muscle flap transfer surgery that the proximal aspect of a divided motor nerve never forms a symptomatic neuroma.
Contrary to early concerns that TMR could create or worsen pain, it was observed that TMR patients had less pain postprocedure. These observations were published in a multicenter retrospective study.22 A preclinical animal model confirmed histologic restoration of myelinated nerve morphology with TMR.23 TMR gives the regenerating fascicles “somewhere to go and something to do,” thus serving to heal rather than hide the amputated nerve ending. Functional motor units produced by TMR may reverse the pathologic central reorganization associated with PLP.24,25 In contrast, standard neuroma treatments do not provide a distal nerve receptor for potential reinnervation and do not attempt to heal the end of the nerve.
In this study, we performed a prospective, single-blinded, randomized controlled trial (RCT) of amputees with neuroma-related pain to compare the effectiveness of TMR to standard treatment of neuroma excision and burying the nerve ending in muscle.26 Outcomes included patient-reported residual limb pain and PLP measures, functional outcomes, and neuroma size by magnetic resonance imaging (MRI).
Twenty-eight major limb amputees over the age of 18 years with chronic pain were prospectively enrolled in an IRB-approved surgical trial at Northwestern and Walter Reed National Military Medical Center. Major limb amputees above the wrist or ankle, older than 18 years old, and who had not undergone prior neuroma treatments for pain after their initial amputation were randomized in the operating room to either undergo standard neuroma surgery or TMR with the opening of an envelope created using a random number generator dictating their method of treatment. The protocol for patients with multiple limb loss was to randomize the patient to a single procedure, but to obtain outcome data from each limb individually. Patients were blinded to their intervention for 1-year postsurgery. Patients in the standard treatment arm still suffering from significant neuroma-related pain at 1 year were offered TMR, if requested. Patient-reported outcome (PRO) data were obtained pre-and postoperatively at 3-month intervals for 1 year and at the conclusion of the study. There were 3 above elbow and 1 below elbow amputations, as well as 10 above knee and 16 below knee amputations for the 30 limbs treated.
Over the 3 years of patient recruitment, approximately 85 patients were screened to find the 28 patients who participated in this randomized clinical trial. As per protocol, a third cohort of 33 amputees was created who underwent TMR for residual limb pain or PLP but who were not randomized for reasons of prior surgical treatment for painful nerves, refusal to participate in the clinical trial, or the concomitant need for improved prosthetic control.
Standard Neuroma Treatment of Neuroma Excision and Muscle Burying
Standard neuroma treatment involves excising the neuroma back to visibly healthy appearing nerve fascicles (Fig. 1, left). The nerve is mobilized proximally and tunneled into the deep aspect of nearby muscle without tension. The muscle itself is chosen if it has limited excursion and is away from joint motion to avoid tugging on the nerve. The end of the nerve is held in place with fine sutures between the nerve and the entry point of the muscle.27 Selection of nerves to be treated for both groups was determined preoperatively by the location and distribution of pain found on physical examination including the presence or absence of Tinel's signs.
TMR Surgical Technique
Detailed descriptions of TMR have been previously published (Fig. 1, right).28–31 Neuromas are dissected and excised to healthy fascicles prior to transfer. Motor nerves innervating nearby muscles rendered functionless by the amputation are identified using a handheld nerve stimulator to serve as potential recipients of TMR nerve transfers. The major mixed nerve sectioned by the amputation (eg, tibial nerve) is then coapted to the surgically divided distal segment of the motor nerve (eg, motor nerve to the soleus muscle) using loupe magnification and 6-0 or 7-0 sutures. Pure sensory nerves including the sural or saphenous nerves were similarly treated when located. There were no differences in the postoperative recovery protocols between patient groups.
Pain data were captured via 2 different PRO scales. The 11-point (0–10) Numerical Rating Scale (NRS) was incorporated as the gold standard direct assessment and primary outcome for pain. Patients were asked to report their worst and best pain levels in the past 24 hours and their current pain levels. To supplement this NRS scale, secondary outcomes included 3 Patient-Reported Outcomes Measurement Information System (PROMIS) assessments: Pain Behavior – Short Form 7a, Pain Intensity—Short Form 3a, and Pain Interference—Short Form 8a.32–34 PROMIS is a validated toolbox for generalized pain, though the use of this tool to assess the localized pain and discomfort of an amputee had not previously been attempted. Participants in all cohorts were asked to complete the NRS scales and PROMIS measures distinguishing residual limb pain and PLP with the visual aid of an avatar (Fig. 2).
Enrolled patients underwent MRI neurogram of the affected limb preoperatively and 1-year postoperatively. Imaging studies were performed using 3 Tesla machines at both sites. Radiologists were asked to identify and measure neuromas in a blinded fashion. A neuroma was defined as a swelling on the end of a cylindrically shaped nerve. The boundaries of these swollen areas were measured and recorded in cubic millimeters.
Neuro-Quality of Life (Neuro-QOL) was used to assess functional outcomes in lower extremity amputees. The Orthotics Prosthetics Users Survey Upper Extremity form was used to assess functional outcomes in upper extremity amputees.
The 2 primary outcomes, change in NRS worst pain score from baseline to 1 year postsurgery for phantom and residual limb pain, were compared between treatments groups in an intention to treat analysis using 2-sample t tests with the Satterthwaite method for unequal variances. A Bonferroni adjustment was applied to the P values and confidence intervals of the effect estimate to account for the 2 comparisons and control the type I error for the primary outcomes at α=0.05 overall. The adjustment consisted of multiplying each P value by 2 and calculating 97.5% confidence intervals to achieve 95% simultaneous confidence intervals. The normality of the difference scores was evaluated before performing the t tests.
Four patients had missing data at the 1-year follow-up time and data were taken from the nearest available follow-up time. One patient in the standard care arm was in severe pain and converted to TMR after 6 months; NRS scores from the 6-month follow-up time were used for the primary analyses for this patient. One TMR patient (2 limbs) and 1 standard care patient were missing 1-year data but did have data at 18 months, which were used for the primary outcomes. Two patients (1 TMR and 1 standard care) had 2 amputations treated; pain scores for each limb were evaluated as independent observations.
As a sensitivity analysis for the primary outcomes, comparisons at the 1-year follow-up time were analyzed utilizing longitudinal linear mixed models incorporating all available data. The model outcomes were the NRS change from baseline scores for all follow-up times, calculated by subtracting the NRS worst pain value at each time point from the NRS worst pain value at baseline for each patient. A random subject effect was included, along with a spatial power correlated error structure modeling correlation in outcomes from the same patient over time, allowing stronger correlation in outcomes occurring closer together in time. Patients with missing data at the 1-year time point were accounted for through the restricted maximum likelihood estimation procedure utilizing all data available at other follow-up times. In each model, 1 for phantom pain and 1 for residual limb pain, a 1-degree of freedom parameter contrast comparing the group treatment effects at the 1-year follow-up outcomes was constructed to obtain the comparison of interest. Results for secondary outcomes are reported as point estimates for effect sizes and unadjusted confidence intervals. All statistical analyses were conducted using SAS/STAT software, version 9.4 (SAS Institute, Cary, NC).
Patient Enrollment and Study Termination
The trial intended to recruit 200 patients but was stopped early with recruitment of 28 patients, without a formal stopping rule. Study enrollment was slow in this surgical trial of an orphan patient population with limited access to a study site. The study was terminated after 3 years of study enrollment (2014–2017). The 28 patients (30 limbs) were allocated equally between the 2 intervention groups (Table 1).
At 1-year postsurgery there was evidence of greater reduction in NRS worst PLP for patients receiving TMR compared with those receiving standard of care, although the difference was not strictly significant after applying the Bonferroni adjustment [mean (adjusted CI) for difference in change scores = 3.4 (−0.1, 6.9), adjusted P = 0.06] (Table 2). The average decrease in PLP was 3.2 in the TMR arm compared with an average increase of 0.2 in the standard care arm. In longitudinal mixed model analysis, the difference in change scores at 1-year postsurgery was significantly greater in the TMR arm compared with standard care [mean (adjusted CI) = 3.5 (0.2, 6.7), adjusted P = 0.03].
Change in NRS worst residual limb pain did not differ significantly between randomized groups, although the effect was in direction of greater decrease for the TMR group [mean (adjusted CI) for difference in change scores = 1.9 (−0.5, 4.4), adjusted P = 0.15]. The average decrease in pain was 2.9 in the TMR arm and 0.9 in the standard care arm. The mixed model results for the group comparison at 1 year yielded similar results [mean (adjusted CI) = 2.1 (−0.3, 4.6), adjusted P = 0.10].
Group means and standard deviations are presented for NRS worst pain scores at baseline, last follow-up, and change from baseline to last follow-up. The point estimates for mean differences in change scores between groups are presented with unadjusted 95% confidence intervals. These results are presented both for the intention to treat randomized groups and additionally incorporating crossover results for patients randomized to and receiving standard care and then crossing over to TMR during the study (Table 3).
PROMIS scale outcomes are similarly reported as mean and standard deviation at baseline, 1-year, last follow-up, and change from baseline. The mean differences in change scores between groups are presented with unadjusted 95% confidence intervals (Tables 4 and 5).
Fully 72% of TMR patients had either no PLP or mild PLP at longest follow-up. This contrasts with the standard treatment arm, which had 40% of patients having no PLP to mild PLP after surgery. For residual limb pain, the percentage of the TMR patient cohort being pain free or suffering from mild pain increased from 0% preoperatively to 67% postoperatively. This compares to the standard arm with 27% having no residual limb pain to mild residual limb pain at longest follow-up.
As most upper extremity candidates desired improved terminal device control that precluded randomization, there were too few upper extremity patients (n = 4) for analysis of OPUS-UE data. Analysis of the lower extremity NEURO-QOL results, representing 24 patient responses, revealed little difference between groups at 1 year. When crossover data were included and at final follow-up, the mean NEURO-QOL t score increased from 39.9 to 45.2 in the TMR cohort showing functional improvement.
Twenty-five of the 28 enrolled patients underwent both preoperative and postoperative imaging of their affected limb(s). Postoperative nerve volumes were 378 mm3 for TMR and 552 mm3 for standard surgery. While 64 nerves were treated, only 25 nerves transferred were visualized by MRI by radiology for measurement.
There were no surgical complications to report.
This prospective, multicenter, randomized clinical trial provides evidence that TMR decreases phantom pain in major limb amputees, with an average decrease of 3.2 in the TMR arm compared with an average increase of 0.2 in the standard treatment arm at the defined 1-year end point. Changes in NRS of 2 points has been shown to be clinically important and correlated to a patient's need to take additional pain medication in studies of both chronic and acute pain.35,36 Residual limb pain showed a trend towards improvement in the TMR group over standard treatment, with average decreases in pain of 2.9 versus 0.9, though this did not reach statistical significance. A failure to reach statistical significance may be due to residual limb pain being caused not only by neuromas, but also due to bone spurs, ischemia, or other conditions37 that were not addressed by this surgical procedure. At final follow-up of just under 1 ½ years, including crossovers, TMR had a decrease of PLP of 3.6 versus an increase of 0.5 for standard treatment, and a decrease of residual limb pain of 3.7 versus 1.2 for muscle burying. To our knowledge, this is the first surgical RCT for the treatment of neuromas.
We postulate that TMR treats pathologic pain through a physiologic nerve healing mechanism that establishes a new afferent signal from muscular sensory receptors and thus, closes the efferent-afferent feedback loop. The process of TMR inherently creates a denervated muscle segment that in turn provides a neurotrophic signal for regeneration of fascicles down the distal motor nerve stump to empty motor endplates and proprioceptors. Studies regarding direct muscle neurotization have long hypothesized that the sensitivity of innervated muscles to acetylcholine is limited and denervated muscle fibers are more readily accepting of neurotization, and we postulate that it is the connection to the terminal receptor that is the cause for the lasting decrease in pain. A point to be repeated is that the newly cut motor nerves do not become symptomatic neuromas. No function is lost from the division of the recipient motor nerve, as the muscle does not maintain any motor function after limb amputation. A surgical concern that the size mismatch between donor and recipient nerves with TMR could create symptomatic neuromas-in-continuity did not materialize. In comparison, standard treatment of neuromas of excision and muscle burying simply places the nerve ending in a healthy, innervated, vascular bed without neurotrophic signals or a reinnervation target and was ineffective for the treatment of either phantoms or residual limb pain in these patients.
We believe that this data is the first to demonstrate in a randomized, blinded trial the persistent long-term improvement of phantom limb pain by any modality. Major limb amputation, and resulting peripheral nerve deafferentation, has been shown to affect multiple neural levels from the periphery to the sensorimotor cortex that contribute to phantom limb phenomena.9–13 Ectopic discharges from disorganized axonal sprouting at the cut nerve ending cause local residual limb pain, and there is evidence to suggest that neuroma pain is a driver of PLP. Harris hypothesized that incongruence between motor efferents and sensory feedback leads to pathologic pain,38 thus providing the basis for behavioral therapies such as mirror therapy and augmented reality training.39 These strategies, as well as those standard procedures that attempt to find the neuroma a more protected microenvironment, have been met with limited success likely because they do not address aberrant electric activity from the nerve ending.40 TMR in the established amputee may reverse maladaptive cortical reorganization, with functional MRI demonstrating motor and sensory cortical maps more similar to healthy controls than to non-TMR amputees.41 TMR for amputees demonstrates improvement but not complete elimination in PLP for most patients, implying that central changes may be only partly reversible.42 This also raises the question whether the performance of TMR concurrently at the time of amputation will be effective in the prevention of residual limb pain and PLP.43
Strengths of the study include the adherence to protocol, reasonable follow-up for all patients, and the crossover patients demonstrating that patients can be successfully treated for pain even after initial management failures. The concept that the “healing” of an injured nerve results in a long-term improvement in pain has already been established for the treatment of neuromas in patients with intact limbs.44,45 It is the special situation of the amputee that there is no distal nerve present for potential repair—hence the need to utilize a nearby expendable muscle filled with receptors for a TMR nerve transfer.
Functional outcomes did not show clear improvement with TMR, though the trend was toward improvement when crossover patients were included. Functional outcomes depend not only on the presence or absence of neuroma pain, but other issues including limb strength, prosthetic function, and patient motivation. MRI neurograms proved to inadequate and insensitive to locate the affected nerves found by physical examination in both treatment arms.
While the trial planned for 200 participants, many fewer procedures were performed than anticipated. At the beginning of the trial, it was hoped that multiple centers scattered geographically around the United States would be involved to perform these surgeries. Despite intensive work, only 2 of 7 planned centers both maintained the required surgeon complement and were able to obtain institutional review board clearance in a timely enough manner to participate. In addition, many more amputees than expected had undergone prior neuroma excision and burying, therefore disqualifying them from being randomized for this trial, as the participating surgeons felt it unethical to repeat an operation that had failed before. Third, the patients were communicating with each other through the internet, with several patients changing their minds and refusing to be randomized at the last minute after hearing more about standard surgery. We treated 33 patients outside of the trial and without randomization. Review of these patients showed outcomes remarkably similar to the TMR patients treated inside the trial. Looking at the overall conduct of this RCT, the lead authors feel that the overall minimum number of patients underwent an unsuccessful procedure (burying) while still being able to demonstrate the efficacy of TMR.
A weakness of this study is the requirement of patients to self-report pain and to distinguish residual limb pain from PLP. Patient-reported outcome instruments, while subject to patient bias, are the gold standard for evaluating pain. We acknowledge that the NRS data in this study represents a 1-time evaluation of pain that can often change over the course of hours, weeks, and months. We administered 3 supplemental PROMIS item banks, which have not yet been validated in people living with chronic postamputation pain. We observed some correlation between the 11-point NRS and PROMIS Pain Intensity, but poor correlation with PROMIS Pain Behavior and Pain Interference (data not shown). We postulate that longstanding behavioral adaptations from living with chronic pain may make these measures slow to change. Another weakness of the study is that only 1 standard neuroma surgery was used for the comparison to TMR. Neuroma excision and muscle burying is the most widely performed and accepted surgical treatment for neuromas. Muscle burying also has in common with all of the other treatments of neuromas the physiologic consequence that a new neuroma will form after treatment. Also not tested was the recently devised regenerative peripheral nerve interface or RPNI, that like TMR was created to achieve improved prosthetic control. Like TMR, RPNI's attempt to achieve a connection between nerve endings and nerve receptors that appear on newly revascularized free muscle grafts that are surgically wrapped around these newly divided nerve endings. Early reports show a decrease in both pain and phantoms.45 Considering the difficulty with patient recruitment in this randomized study, further comparisons of these various techniques may only be achieved using a unified patient outcome tool that will be shared between institutions.
TMR resulted in improved phantom limb pain and trended toward improved residual limb pain in major limb amputees compared with conventional surgical therapy. Future studies will focus on the surgical refinements of the procedure, as well as the application of TMR to specific indications for amputation such as vascular disease.
The authors thank Ms Carlee Culver and Ms Monalee Shah for their contributions to study execution.
1. Ephraim PL, Wegener ST, MacKenzie EJ, et al. Phantom pain, residual limb pain, and back pain in amputees: results of a national survey. Arch Phys Med Rehabil
2. Ehde DM, Czerniecki JM, Smith DG, et al. Chronic phantom sensations, phantom pain, residual limb pain, and other regional pain after lower limb amputation. Arch Phys Med Rehabil
3. Smith DG, Ehde DM, Legro MW, et al. Phantom limb, residual limb, and back pain after lower extremity amputations. Clin Orthop Relat Res
4. Hsu E, Cohen SP. Postamputation pain
: epidemiology, mechanisms, and treatment. J Pain Res
5. Pierce RO Jr, Kernek CB, Ambrose TA 2nd. The plight of the traumatic amputee. Orthopedics
6. Tintle SM, Baechler MF, Nanos GP, et al. Reoperations following combat-related upper-extremity amputations. J Bone Joint Surg Am
7. Gorkisch K, Boese-Landgraf J, Vaubel E. Treatment and prevention of amputation neuromas in hand surgery. Plast Reconstr Surg
8. Lu C, Sun X, Wang C, et al. Mechanisms and treatment of painful neuromas. Rev Neurosci
9. Dellon AL, Mackinnon SE, Pestronk A. Implantation of sensory nerve into muscle: preliminary clinical and experimental observations on neuroma
formation. Ann Plast Surg
10. Flor H, Elbert T, Knecht S, et al. Phantom-limb pain as a perceptual correlate of cortical reorganization following arm amputation. Nature
11. Flor H, Nikolajsen L, Staehelin Jensen T. Phantom limb pain
: a case of maladaptive CNS plasticity? Nat Rev Neurosci
12. Montoya P, Ritter K, Huse E, et al. The cortical somatotopic map and phantom phenomena in subjects with congenital limb atrophy and traumatic amputees with phantom limb pain
. Eur J Neurosci
13. Preissler S, Feiler J, Dietrich C, et al. Gray matter changes following limb amputation with high and low intensities of phantom limb pain
. Cereb Cortex
14. Vaso A, Adahan HM, Gjika A, et al. Peripheral nervous system origin of phantom limb pain
15. Wang X, Yi Y, Tang D, et al. Gabapentin as an adjuvant therapy for prevention of acute phantom-limb pain in pediatric patients undergoing amputation for malignant bone tumors: a prospective double-blind randomized controlled trial. J Pain Symptom Manage
16. Herrador Colmenero L, Perez Marmol JM, Martí-García C, et al. Effectiveness of mirror therapy, motor imagery, and virtual feedback on phantom limb pain
following amputation: a systematic review. Prosthet Orthot Int
17. Alviar MJ, Hale T, Dungca M. Pharmacologic interventions for treating phantom limb pain
. Cochrane Database Syst Rev
18. Bosanquet DC, Glasbey JC, Stimpson A, et al. Systematic review and meta-analysis of the efficacy of perineural local anaesthetic catheters after major lower limb amputation. Eur J Vasc Endovasc Surg
19. Hijjawi J, Kuiken TA, Lipschutz RD, et al. An improved brain-machine interface accomplished using multiple nerve transfers. Plast Reconstr Surg
20. Kuiken TA, Dumanian GA, Lipschutz RD, et al. The use of targeted muscle reinnervation
for improved myoelectric prosthesis control in a bilateral shoulder disarticulation amputee. Prosthet Orthot Int
21. Kuiken TA, Miller LA, Lipschutz RD, et al. Targeted reinnervation for enhanced prosthetic arm function in a woman with a proximal amputation: a case study. Lancet
22. Souza JM, Cheesborough JE, Ko JH, et al. Targeted muscle reinnervation
: a novel approach to postamputation neuroma
pain. Clin Orthop Relat Res
23. Kim PS, Ko JH, O'Shaughnessy KK, et al. Targeted muscle reinnervation
of a pedicled rabbit rectus abdominis flap: the demonstration of a cure for the common mixed nerve neuroma
? J Hand Surg Am
24. Lotze M, Grodd W, Birbaumer N, et al. Does use of a myoelectric prosthesis prevent cortical reorganization and phantom limb pain
? Nat Neurosci
25. Preissler S, Thielemann D, Dietrich C, et al. Preliminary evidence for training-induced changes of morphology and phantom limb pain
. Front Hum Neurosci
26. Watson J, Gonzalez M, Romero A, et al. Neuromas of the hand and upper extremity. J Hand Surg Am
27. Elliot D. Surgical management of painful peripheral nerves. Clin Plast Surg
28. Agnew SP, Schultz AE, Dumanian GA, et al. Targeted reinnervation in the transfemoral amputee: a preliminary study of surgical technique. Plast Reconstr Surg
29. Gart MS, Souza JM, Dumanian GA. Targeted muscle reinnervation
in the upper extremity amputee: a technical roadmap. J Hand Surg Am
30. Morgan EN, Kyle Potter B, Souza JM, et al. Targeted muscle reinnervation
for transradial amputation: description of operative technique. Tech Hand Up Extrem Surg
31. Fracol M, Janes L, Ko J, et al. Targeted muscle reinnervation
in the lower leg: an anatomic study. Plast Reconstr Surg
32. Revicki DA, Chen WH, Harnam N, et al. Development and psychometric analysis of the PROMIS pain behavior item bank. Pain
33. Chen W-H, Revicki D, Amtmann D, et al. Development and Analysis of PROMIS Pain Intensity Scale. Qual Life Res
2012; 20 (suppl 1):18.
34. Amtmann D, Cook KF, Jensen MP, et al. Development of a PROMIS item bank to measure pain interference. Pain
35. Farrar JT, Young JP Jr, LaMoreaux L, et al. Clinical importance of changes in chronic pain intensity measured on an 11-point numerical pain rating scale. Pain
36. Farrar JT, Portenoy RK, Berlin JA, et al. Defining the clinically important difference in pain outcome measures. Pain
37. Clarke C, Lindsay DR, Pyati S, et al. Residual limb pain is not a diagnosis: a proposed algorithm to classify postamputation pain
. Clin J Pain
38. Harris AJ. Cortical origin of pathological pain. Lancet
39. Ortiz-Catalan M, Guethmundsdottir RA, Kristoffersen MB, et al. Phantom motor execution facilitated by machine learning and augmented reality as treatment for phantom limb pain
: a single group, clinical trial in patients with chronic intractable phantom limb pain
40. Richardson C, Kulkarni J. A review of the management of phantom limb pain
: challenges and solutions. J Pain Res
41. Serino A, Akselrod M, Salomon R, et al. Upper limb cortical maps in amputees with targeted muscle and sensory reinnervation. Brain
42. Baliki MN, Petre B, Torbey S, et al. Corticostriatal functional connectivity predicts transition to chronic back pain. Nat Neurosci
43. Cheesborough JE, Souza JM, Dumanian GA, et al. Targeted muscle reinnervation
in the initial management of traumatic upper extremity amputation injury. Hand (N Y)
44. Guse DM, Moran SL. Outcomes of the surgical treatment of peripheral neuromas of the hand and forearm: a 25-year comparative outcome study. Ann Plast Surg
45. Souza JM, Purnell CA, Cheesborough JE, et al. Successful treatment of foot and ankle neuroma
pain with processed nerve allografts. Foot Ankle Int