How are you taking part in this consultation?

You will not be able to change how you comment later.

You must be signed in to answer questions

    The content on this page is not current guidance and is only for the purposes of the consultation process.

    Interventional procedure overview of targeted muscle reinnervation for managing limb amputation pain

    Closed for comments This consultation ended on at   Request commenting lead permission

    Evidence summary

    Population and studies description

    This interventional procedures overview is based on about 730 TMR procedures from 1 systematic review (Tham 2023), 1 randomised controlled trial (Dumanian 2019), 1 retrospective propensity score-matched study (Shammas 2022), 1 prospective case series (O'Brien 2022) and 6 retrospective case series or cohort studies (Kang 2022; Goodyear 2024; Chang 2021; Li 2024; Chang 2024 and Smith 2024). Of the 9 primary studies, 3 were also included in the systematic review (Dumanian 2019; O'Brien 2022 and Chang 2021). This is a rapid review of the literature, and a flow chart of the complete selection process is shown in figure 1. This overview presents 10 studies as the key evidence in table 2 and table 3, and lists 26 other relevant studies in appendix B, table 5.

    Most of the studies were based in the US and there was some overlap in authorship. The systematic review by Tham et al. (2023) included 10 studies, with 1,099 upper and lower limbs and 448 TMR procedures. Of the 10 studies, 1 was a randomised controlled trial, 6 were cohort studies and 3 were case series. The randomised controlled trial was graded as high quality and observational studies were moderate to very low quality. The TMR procedure was done either at the same time as the amputation or as a later, secondary procedure. The comparators were other interventions for chronic, postamputation pain, standard care, or no treatment. Most of the studies included amputations from any cause, but 1 only included people who had amputations because of cancer. The mean age of people who had TMR ranged from 35 to 59 years and the proportion of males ranged from 56 to 86%. The mean follow-up was 17.9 months (range 9.6 to 24.0).

    The randomised controlled trial reported by Dumanian et al. (2019), which was also included in the systematic review, compared delayed TMR (n=15 limbs) with neuroma excision and muscle burying (n=15 limbs). Treatment allocation was single blinded for the first year, after which people in the standard care group could choose to have TMR. The trial was intended to recruit 200 patients, but it was stopped early with recruitment of 28 patients, without a formal stopping rule. Only 2 of 7 planned centres in the US had the necessary surgeon complement and institutional review board clearance in time to participate. In addition, many more amputees than expected had already had neuroma excision and burying, which excluded them from the trial. The authors also noted that patients were communicating with each other through the internet, and some refused to be randomised after hearing more about standard surgery. Most people in the trial had lower limb amputations and the main reason for amputation was trauma. The mean age was 39 years in the TMR group and 45 years in the control group and the proportion of males was 86% and 57%, respectively. The mean follow-up was 17.7 months in the TMR group and 19.3 months in the control group.

    The matched sample in the propensity score-matched study by Shammas et al. (2022) included 96 people who had below-knee amputation with or without TMR. The main aim of the study was to assess the risk of postoperative complications when TMR is done at the time of below-knee amputation. The mean age was 58 years (range 31 to 90) and 65% were male. The main indications for amputation were infection (67%) and ischaemia (27%). Follow-up was 60 days.

    The prospective case series by O'Brien et al. (2022), which was also in the systematic review by Tham et al. (2023), included 81 people with major upper (19%) or lower limb (81%) amputations with concurrent TMR. The mean age was 52 years (range 18 to 85) and 42% were male. The main reasons for amputation were cancer (52%), trauma (20%) and infection (14%). Follow-up ranged from 3 months to 4.6 years.

    The retrospective case series by Kang et al. (2022) was based in the UK and included 36 people with upper (27.5%) or lower (72.5%) limb amputation and intractable neuroma pain or phantom limb pain. The reasons for amputation included trauma (64%), peripheral vascular disease (8%), infection (8%) and tumour (6%). The TMR was delayed and the mean duration from amputation was 11 years. The mean age at the time of TMR surgery was 49 years (range 23 to 75 years) and 75% of the study population was male. Mean follow-up was 9.5 months (range 3 to 24 months).

    The 2 retrospective cohort studies by Goodyear et al. (2024) and Li et al. (2024) compared acute (primary) TMR with delayed (secondary) TMR. They included 103 and 32 people, respectively, with upper or lower limb amputations. The reasons for amputation were mixed and included cancer, infection, trauma, and ischaemia. In the study of 103 people, the proportion of males was 58%, the mean age at surgery was 53 years and the mean time from TMR to survey was 18 months for the acute group and 23 months for the delayed group (p=0.31). In the study of 32 people, the proportion of males was 81%, the median age was 39 years, and the median follow-up was 24 months for acute TMR and 21 months for delayed TMR.

    The 2 retrospective cohort studies by Chang et al. (2021 and 2024) compared TMR with traction neurectomy and muscle implantation. The 2021 study was also included in the systematic review by Tham et al. (2023). It included 200 procedures (100 TMR) on below-knee amputations. The most common reason for amputation was infection and there were none because of cancer or trauma. The mean age was 59.7 years in the TMR group and 58.8 years in the control group, and 68.5% were male. The mean follow-up was 9.6 months in the TMR group and 18.5 months in the control group (p<0.01). The 2024 study by Chang et al. included 99 people with through- or above-knee amputations. The main reasons for amputation were infection and ischaemia. The mean age was 60 years in the TMR group and 65 years in the control group, and the proportion of males was 68% and 57%, respectively. The mean follow-up was 9.5 months in the TMR group and 14.3 months in the control group (p=0.10).

    The retrospective study by Smith et al. (2024) compared the rate of revision surgery in people who had primary transtibial amputation with (n=29) or without (n=83) TMR. The mean age of the cohort was 47 years, 88% were male and the median follow-up was 1.2 years.

    Table 2 presents study details.

    Figure 1 Flow chart of study selection

    Table 2 Study details

    Study no.

    First author, date

    country

    Characteristics of people in the study (as reported by the study)

    Study design

    Inclusion criteria

    Intervention

    Follow up

    1

    Tham JL (2023)

    Countries of included studies not reported.

    10 studies (n=1,099 upper and lower limbs; 448 TMR procedures)

    The mean age of people who had TMR ranged from 35.0 to 59.6 years.

    The proportion of males in the studies ranged from 56 to 86%.

    The number of patients ranged from 13 to 100 in the TMR group, and 3 to 438 in the control group.

    Most studies included amputations from any cause, but 1 study only included people who had amputations because of cancer.

    Systematic review and meta-analysis of 3 selected studies

    Search date: June 2022

    Of the 10 included studies, 1 was a randomised controlled trial, 6 were cohort studies and 3 were case series.

    Adults with neuropathic or chronic limb pain after amputation who had TMR.

    Studies were included if they reported quantitative outcome measures of pain or function at various time intervals, surgical complications, and medication usage.

    • TMR

    • Comparator: other interventions for chronic postamputation pain, standard care, or no treatment (control)

    In 1 study, RPNI was done when the remnant nerve was too short or there was no suitable muscle target for TMR.

    The TMR was done either at the same time as the amputation or later, as a secondary procedure.

    Mean 17.9 months (range 9.6 to 24.0).

    2

    Dumanian G, 2019

    US

    Included in Tham et al. (2023)

    n=28 (30 limbs)

    Mean age (years):

    • TMR=39.6

    • Control=45.3

    Male (%)

    • TMR=86%

    • Control=57%

    71% in both groups were described as Caucasian.

    26 lower limbs, 4 upper limbs

    Trauma was the reason for most (90%) amputations.

    Randomised controlled trial (single blinded for 1 year).

    Treatment allocation was done in the operating room using sealed envelopes and a random number generator.

    Study enrolment: 2014 to 2017.

    At 1 year, people in the standard care group who still had symptoms could choose to have TMR.

    People with chronic pain associated with major limb amputations above the wrist or ankle, older than 18 years old, and with no previous neuroma treatments for pain after their initial amputation.

    • TMR (n=15 limbs)

    • Standard neuroma treatment of neuroma excision and muscle burying (n=15 limbs)

    TMR was delayed (most amputations were 5 or more years earlier)

    Selection of nerves to be treated for both groups was determined preoperatively by the location and distribution of pain found on physical examination.

    Mean (months)

    • TMR=17.7

    • Control=19.3

    3

    Shammas R, 2022

    US

    Matched sample

    n=96 (31 TMR procedures)

    Mean age 58.5 years (range 31.6 to 90.9)

    Male 64.6%

    All amputations were below-knee.

    Ethnicity was described as White in 67.7% of people, and Black in 27.1% of people.

    85% had diabetes and 77% were ambulatory.

    The main indications for amputation were infection (67%) and ischaemia (27%).

    The rate of peripheral vascular disease was lower in the TMR group (26% compared to 48%).

    Retrospective propensity score-matched study.

    Procedures were done between January 2018 and June 2020.

    Adults who had a below-knee amputation with or without TMR at either of 2 centres were included.

    People who were younger than 18 years old, were converted to an above-knee amputation, had an incomplete amputation, or had bilateral amputations were excluded from analysis.

    Below-knee amputation with or without TMR at the same time.

    A small proportion of TMR procedures (19%) used a 2‑incision approach.

    An incisional vacuum-assisted closure at the time of closure was commonly used after TMR.

    60 days

    4

    O'Brien A, 2022

    US

    Included in Tham et al. (2023)

    n=81 (83 limbs)

    Mean age=52 years (range 18 to 85)

    Male=42%

    Ethnicity was described as White in 91% of people, African American or Black in 6% of people and other in 3% of people.

    Lower limb (including hip disarticulation)=81% (67/83)

    Upper limb (including shoulder disarticulation)=19% (16/83)

    Reason for amputation:

    • Cancer=52%

    • Infection=14%

    • Trauma=20%

    • Ischaemia=1%

    • Other=14%.

    Prospective single-centre case series

    Patients were surveyed at in-person follow-up appointments or telehealth appointments.

    Study period: October 2015 to December 2020

    Adults (aged 18 or over) who had major limb amputation with TMR and without cognitive impairment.

    Patients were included if they had completed at least one survey.

    Patients were excluded if they were enrolled in concurrent trials for neuropathic pain or had metastatic or recurrent cancer or open wounds at the time of survey.

    Major amputation (below-knee, above-knee, hip disarticulation, transradial, transhumeral, and shoulder disarticulation levels) and TMR at the same time.

    Common nerve transfers and targets were based on previously published studies.

    Range 3 months to 4.6 years.

    Of the 81 enrolled people, 23 (28%) completed surveys at 18 months or later; the cohort completing surveys at 18 months or later had a mean follow-up time of 2.4 years (range 1.5 to 4.6 years).

    5

    Kang N, 2022

    UK

    n=36 (40 TMR procedures)

    Mean age=49 years

    Male=75% (27/36)

    Lower limb=72.5% (29/40)

    Upper limb=27.5% (11/40)

    Reasons for amputation:

    • trauma (64%)

    • peripheral vascular disease (8%)

    • tumour (6%)

    • infection (8%)

    • unknown (14%).

    Mean ASA score=2.33

    Retrospective single-centre case series

    Treatment period: 2013 to 2020

    People with intractable neuroma pain or phantom limb pain after major amputation of an upper or lower limb.

    TMR was delayed; the mean duration from amputation to TMR was 11 years.

    Mean and median=38 weeks

    Range 3 to 24 months

    6

    Goodyear E, 2024

    US

    n=103 (105 limbs)

    Mean age at surgery=53 years

    Male=58% (61/105)

    Lower limb=82% (86/105)

    Upper limb=18% (19/105)

    Reason for amputation in acute TMR group:

    • Cancer=48%

    • Infection=12%

    • Trauma=22%

    • Ischaemia=3%

    • Other=15%.

    Reason for amputation in delayed TMR group:

    • Cancer=3%

    • Infection=31%

    • Trauma=41%

    • Ischaemia=9%

    • Other=16%.

    Retrospective cohort study

    Study period: October 2015 to December 2020

    Patients were excluded if they were younger than 18 years, had a cognitive impairment, were enrolled in another neuropathic pain trial, or died within 6 months of TMR.

    Acute (n=73 procedures; 71 patients) or delayed (n=32) TMR. Acute TMR was defined as TMR done within 14 days of major extremity amputation. Delayed TMR was defined as TMR done secondary to the development of a symptomatic neuroma.

    The mean time to survey was 18 months for the acute TMR cohort and 22.9 months for the delayed TMR cohort.

    7

    Li A, 2024

    Australia

    n=32 (38 major limb amputations)

    Median age=39 years (IQR 29 to 57)

    Male=81.3%

    Aetiology:

    • Trauma=75%

    • Ischaemia=13%

    • Complex regional pain syndrome=3%

    • Infection=6%

    • Malignancy=3%

    Upper limb only=22%

    Lower limb only=72%

    Upper and lower limbs=6%

    Retrospective, single-centre cohort study

    TMR procedures were done between January 2018 and December 2021

    Acute TMR was offered as the standard of care for all people for whom it was suitable, referred to the plastic surgery unit for major limb amputation.

    People who had TMR for pain control following major limb amputation (at the level or proximal to the wrists and ankles) and completed at least 1 postoperative patient reported outcome measure questionnaire up to 30 June 2022.

    People were considered for the study if they had suitable nerves and targets, no proximal nerve injury, a clean wound with a good vascular supply, no anaesthetic contraindications to the extra operating time required for the procedure (up to 2 hours) and were expected to survive the injury or disease process leading to the amputation. Delayed TMR was offered to patients with significant peripheral nerve component to their chronic amputation pain.

    Acute (n=16 patients, 22 limbs) or delayed (n=16 patients, 16 limbs) TMR.

    TMR was classified as acute if it was done during the same admission as the primary amputation and delayed if it was done during a later admission. The median time between initial amputation and TMR was 0.4 months for acute TMR and 61.9 months for delayed TMR (p<0.001).

    Median length of follow-up was 24 months for acute TMR and 21 months for delayed TMR (p=0.65).

    8

    Chang B, 2021

    US

    Included in Tham et al. (2023)

    n=200 (100 TMR procedures)

    Mean age (years)

    • TMR=59.7

    • Non-TMR=58.8

    Male=68.5%

    All amputations were below knee.

    The most common reason for amputation was infection (76%).

    No amputations were done in the setting of acute trauma or cancer resection.

    The mean Charlson Comorbidity Index was 5.3, and 57% of people had a Charlson Comorbidity Index greater than 5.

    Retrospective, single-centre, cohort study

    TMR procedures were done between January 2018 and December 2019. The control group had amputations between January 2015 and December 2017.

    The first 100 patients who had primary TMR at the time of below-knee amputation were included.

    The 100 patients who had below knee amputation immediately before the initiation of the TMR protocol were included as a cohort for comparison of outcomes.

    • Below knee amputation with TMR (n=100)

    • Below knee amputation with traction neurectomy and muscle implantation of all identified nerves (n=100).

    For the TMR group, mean follow-up was 9.6 months, compared to 18.5 months for the non-TMR group (p<0.01).

    9

    Chang B, 2024

    US

    n=99 (41 TMR procedures)

    Mean age (years)

    • TMR=60.4

    • Non-TMR=65.0

    Male:

    • TMR=68.3%

    • Non-TMR=56.9%

    All amputations were through- or above-knee.

    Aetiology was infection in 46% of TMR group and 60% of non-TMR group, and dysvascular in 34% of TMR group and 31% in non-TMR group.

    The mean Charlson Comorbidity Index was 5.5 in the TMR group and 4.8 in the non-TMR group.

    Retrospective, single-centre, cohort study

    TMR procedures were done between January 2018 and December 2019.

    Patients who had primary TMR at the time of through- or above-knee amputation were included.

    Patients who had a through- or above-knee amputation immediately before the initiation of the TMR protocol from January 2014 to December 2017 were included as a cohort for comparison of outcomes.

    • Through- or above-knee amputation with TMR (n=41)

    • Through- or above-knee amputation with traction neurectomy and muscle implantation of all identified nerves (n=58)

    For the TMR group, mean follow-up was 9.5 months, compared to 14.3 months for the non-TMR group (p=0.10).

    10

    Smith T, 2024

    US

    n=112 primary amputations (29 with TMR) and 51 revision amputations.

    Mean age at time of amputation was 47 years (range 24 to 81).

    Male=88%

    Indications for primary amputation included trauma, completion or revision traumatic amputations or failed limb salvage (n=72, 64%), osteomyelitis or infection (n=31, 28%), oncology (n=3, 3%), and other indications (n=6, 5%).

    Indications for revision included 11 (22%) for wound breakdown or dehiscence, 11 (22%) for infection, 9 (18%) for neuroma, 6 (12%) for scar revision or redundant tissue or debulking, 5 (10%) for heterotopic ossification or bony prominence, 4 (8%) for instability, 4 (8%) for myodesis or myoplasty failure, and 1 (2%) for seroma.

    Retrospective, single-centre, cohort study.

    Amputations were done between January 2014 and April 2021.

    People who had primary or revision transtibial amputation.

    Revision transtibial amputations that were done on primary amputations done outside of the study site were not included in analysis of revision rates for patients with and without TMR.

    • Primary transtibial amputation with TMR (n=29)

    • Primary transtibial amputation without TMR (n=83)

    • Revision transtibial amputation (n=51)

    Primary TMR was defined by its occurrence at the time of initial amputation or in the setting of the same initial hospitalisation.

    Of the 51 revision amputations, 23 (21%) were in patients who had primary transtibial amputation at the study institution.

    Median=1.2 years (range 0 to 7.2 years).

    Table 3 Study outcomes

    First author, date

    Efficacy outcomes

    Safety outcomes

    Tham, 2023

    Residual limb pain - NRS

    Pooled mean difference for TMR compared with control was −2.68, 95% CI −3.21 to −2.14; p<0.0001; 4 studies [Mioton 2020, O'Brien 2021, O'Brien 2022, Valerio 2019]; I2=0%)

    Phantom limb pain - NRS

    Pooled mean difference for TMR compared with control was −2.17, 95% CI −2.70 to −1.63; p<0.0001; 4 studies as above; I2=51%)

    Residual limb pain – PROMIS

    Intensity

    Pooled mean difference for TMR compared with control was −13.39, 95% CI −14.59 to −12.19; p<0.0001; 3 studies [Mioton 2020, O'Brien 2021, Valerio 2019]; I2=61%)

    Behavioural

    Pooled mean difference for TMR compared with control was −11.97, 95% CI −13.78 to −10.16; p<0.0001; 4 studies [Mioton 2020, O'Brien 2021, O'Brien 2022, Valerio 2019]; I2=87%)

    Interference

    Pooled mean difference for TMR compared with control was −12.18, 95% CI −13.56 to −10.80; p<0.0001; 4 studies as above; I2=70%)

    Phantom limb pain – PROMIS

    Intensity

    Pooled mean difference for TMR compared with control was −11.21, 95% CI −12.97 to −9.45; p<0.0001; 3 studies [Mioton 2020, O'Brien 2021, Valerio 2019]; I2=61%)

    Behavioural

    Pooled mean difference for TMR compared with control was −10.45, 95% CI −12.39 to −8.52; p<0.0001; 4 studies [Mioton 2020, O'Brien 2021, O'Brien 2022, Valerio 2019]; I2=76%)

    Interference

    Pooled mean difference for TMR compared with control was −11.48, 95% CI −12.82 to −10.13; p<0.0001; 4 studies as above; I2=62%)

    Functional outcomes

    Mioton et al. used the Orthotics Prosthetics Users Survey (OPUS) for upper limb amputees who had TMR, which increased from 53.7 (SD 3.4) to 56.4 (SD 3.7), p<0.01 at 1 year postoperatively. Neuro-QoL was used for lower limb amputees who had TMR, which increased from 32.9 (SD 1.5) to 35.2 (SD 1.6), p<0.01 at 1 year postoperatively. Both the Neuro-QoL and OPUS showed higher functional scores in the TMR group, respectively.

    No studies reported the length of hospital stay or patient satisfaction.

    Complications

    Complication rates were reported in 4 studies and ranged from 0 to 16%. They predominantly included wound site or stump infections.

    Hoyt et al. (2021): no TMR related complications were reported in the acute treatment group and there was 1 minor wound complication in the delayed TMR group (n=59).

    Alexander et al. (2019): wound complications needing surgery=16% (5/31), including 1 person who needed a conversion from below knee amputation to above knee amputation because of a non-healing stump wound; 2 people had neuroma excisions and TMR of symptomatic neuromas that developed in pure sensory nerves that were not included in the initial nerve transfer.

    Chang et al. (2021) noted a statistically significantly lower rate of infections needing surgery in those who had TMR compared with those who had traditional traction neurectomy (16% TMR versus 30% control, p=0.02).

    Dumanian et al. (2019) reported no surgical complications (0%).

    Dumanian G, 2019

    Worst pain score (NRS) in the last 24 hours (0 to 10), at baseline and 1 year (primary outcome); mean (SD)

    Phantom limb pain

    • Baseline: TMR=5.8 (3.2), control=3.9 (2.7)

    • 1 year: TMR=2.6 (2.2), control=4.1 (3.0)

    • Change: TMR=3.2 (2.9), control=-0.2 (4.9)

    • Mean (Adjusted 95% CI) difference of change scores=3.4 (‑0.1 to 6.9)

    Residual limb pain

    • Baseline: TMR=6.6 (2.0), control=6.9 (2.5)

    • 1 year: TMR=3.7 (2.0), control=6.0 (2.8)

    • Change: TMR=2.9 (2.2), control=0.9 (3.3)

    • Mean (Adjusted 95% CI) difference of change scores=1.9 (‑0.5 to 4.4)

    Worst pain score (NRS) in the last 24 hours at last follow-up, intention to treat; mean (SD)

    Phantom limb pain

    • Baseline: TMR=5.8 (3.2), control=3.9 (2.7)

    • Last follow-up: TMR=2.3 (2.3), control=4.4 (3.3)

    • Change: TMR=3.5 (3.1), control=-0.5 (5.3)

    • Mean (Adjusted 95% CI) difference of change scores=4.0 (0.8 to 7.2)

    Residual limb pain

    • Baseline: TMR=6.6 (2.0), control=6.9 (2.5)

    • Last follow-up: TMR=3.6 (2.1), control=5.7 (3.0)

    • Change: TMR=3.0 (2.1), control=1.2 (3.5)

    • Mean (Adjusted 95% CI) difference of change scores=1.8 (‑0.3 to 4.0)

    Note: 3 people crossed over from standard care to TMR arm after 1 year

    Proportion of people with no or mild phantom limb pain at longest follow-up

    • TMR=72%

    • Control=40%, p value not stated

    Proportion of people with no or mild residual limb pain at longest follow-up

    • TMR=67%

    • Control=27%, p value not stated

    PROMIS pain scales at 1 year, mean (SD)

    Phantom limb pain - intensity

    • Baseline: TMR=52.4 (11.2), control=48.3 (9.5)

    • 1 year: TMR=38.0 (7.2), control=45.8 (10.9)

    • Change: TMR=13.7 (10.7), control=2.0 (17.9)

    Mean (Adjusted 95% CI) difference of change scores=11.7 (-0.3 to 23.7)

    Phantom limb pain – behaviour

    • Baseline: TMR=58.3 (11.8), control=58.5 (9.7)

    • 1 year: TMR=50.7 (9.9), control=52.0 (8.4)

    • Change: TMR=7.6 (9.7), control=6.5 (14.9)

    Mean (Adjusted 95% CI) difference of change scores=1.1 (-8.3 to 10.5)

    Phantom limb pain – interference

    • Baseline: TMR=60.2 (12.5), control=57.9 (11.0)

    • 1 year: TMR=50.4 (9.8), control=52.8 (8.9)

    • Change: TMR=9.8 (8.9), control=5.1 (16.0)

    Mean (Adjusted 95% CI) difference of change scores=4.7 (-5.0 to 14.3)

    Residual limb pain - intensity

    • Baseline: TMR=55.7 (7.6), control=55.0 (5.5)

    • 1 year: TMR=44.5 (8.2), control=49.5 (8.3)

    • Change: TMR=11.5 (8.3), control=5.7 (8.1)

    Mean (Adjusted 95% CI) difference of change scores=5.8 (-0.9 to 12.4)

    Residual limb pain – behaviour

    • Baseline: TMR=61.5 (3.7), control=61.9 (4.3)

    • 1 year: TMR=56.8 (7.0), control=56.6 (6.5)

    • Change: TMR=4.7 (7.1), control=5.3 (10.4)

    Mean (Adjusted 95% CI) difference of change scores=‑0.5 (-7.2 to 6.1)

    Residual limb pain – interference

    • Baseline: TMR=64.4 (7.0), control=65.8 (5.1)

    • 1 year: TMR=56.8 (6.6), control=57.4 (8.6)

    • Change: TMR=7.6 (9.2), control=8.5 (11.0)

    Mean (Adjusted 95% CI) difference of change scores=‑0.9 (-8.5 to 6.7)

    Functional outcomes

    Analysis of the lower extremity Neuro-QoL results (n=24) showed little difference between the 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.

    There were no surgical complications.

    Shammas, 2022

    Mean length of surgery, minutes

    • TMR=188.5 (SD 63.6)

    • No TMR=88 (SD 28.2), p<0.001

    Mean estimated blood loss, ml

    • TMR=154 (SD 143.3)

    • No TMR=136.3 (SD 118.9), p=0.04

    Mean length of hospital stay, days

    • TMR=7.5 (SD 6.4)

    • No TMR=8.1 (SD 8.8), p=0.92

    Major complications within 60 days of amputation (primary outcome;defined as those that needed readmission, transfer to the intensive care unit, or reoperation, or a cause of death related to the amputation procedure)

    • TMR=29.0% (9/31)

    • No TMR=24.6% (16/65), p=0.69

    RR=1.20 (90% CI 0.57 to 2.55), p=0.35

    Readmission

    • TMR=25.8% (8/31)

    • No TMR=18.5% (12/65), p=0.21

    Reoperation

    • TMR=19.4% (6/31)

    • No TMR=10.8% (7/65), p=0.13

    Admission to intensive care unit

    • TMR=6.5% (2/31)

    • No TMR=10.8% (7/65), p=0.41

    Mortality (all cause)

    • TMR=3.2% (1/31)

    • No TMR=9.2% (6/65), p=0.34

    Minor surgical complications

    • TMR=25.8% (8/31)

    • No TMR=20.0% (13/65), p=0.65

    RR=1.21 (90% CI 0.61 to 2.41), p=0.33

    Wound healing complication

    • TMR=45.2% (14/31)

    • No TMR=33.8% (22/65), p=0.25

    Infection

    • TMR=9.7% (3/31)

    • No TMR=6.2% (4/65), p<0.001

    Kang, 2022

    Mean change in NRS at 12 months after TMR

    Upper limb – neuroma pain (n=10)

    Change=5.30 (SD 4.62), 95% CI 8.61 to 2.00, p=0.006

    All pain had resolved at 12 months.

    Upper limb – phantom limb pain (n=10)

    Change=4.40 (SD 4.03), 95% CI 7.29 to 1.52, p=0.007

    There was a slight increase at 3 months from 6.40 to 7.30 with a subsequent decrease to 3.50 at 6 months and stabilisation at 2.0 by 12 months. Of the 10 patients, 1 was pain free at 12 months and 5 had only mild pain (NRS between 1 and 3).

    Lower limb – neuroma pain (n=20)

    Change=4.35 (SD 4.53), 95% CI 6.47 to 2.23, p<0.0001

    Of the 20 procedures, 10 resulted in complete resolution of neuroma pain and 2 patients had mild residual pain.

    Lower limb – phantom limb pain (n=20)

    Change=2.60 (SD 3.98), 95% CI 4.46 to 0.74, p=0.009

    4 patients reported full resolution of pain at 12 months, and 3 noted marked decreases of pain to minimal levels (NRS between 1 and 3).

    Medication use

    Of the 18 patients who were taking pregabalin before the procedure, 9 had discontinued it after 1 year. There was a mean reduction of 352 mg in daily intake over the 12 months of follow-up (p<0.01).

    Patient satisfaction

    Data on satisfaction was reported for 22 patients (61%, 9 upper limb and 13 lower limb patients). 91% of these responses indicated overall satisfaction with the procedure at 12 months (9 upper limb and 11 lower limb patients). However, out of 22 patients, only 50% (6 upper limb, 5 lower limb patients) felt they would have agreed to a prophylactic (preventive) TMR procedure.

    Complications

    There were 46 complications in 28 out of 40 TMR procedures (70%). Of the 36 people, 13 had more than 1 complication.

    Complications occurred in 23 lower limb procedures (79%) and 5 upper limb procedures (45%).

    Number of complications

    Upper limbs

    • Infection, n=4

    • Paraesthesia, n=1

    • Wound dehiscence, n=1

    • Haematoma, n=1

    • Seroma, n=1

    Lower limbs

    • Unmasking of neuromas, n=12 (symptoms of unmasking typically became apparent within a few weeks of surgery, and 4 patients needed an additional TMR procedure)

    • Infection, n=7

    • Bursa, n=6

    • Paraesthesia, n=4

    • Wound dehiscence, n=3

    • Haematoma, n=2

    • Ulceration, n=2

    • Seroma, n=1

    • Lymphatic discharge, n=1

    The authors noted that they now recommend not to wear a prosthesis for at least 6 weeks after surgery in the lower limbs and longer if there is any delay to wound healing, to reduce the risk of complications.

    O'Brien, 2022

    NRS at 3 months after TMR (worst pain in a 24-hour recall period), n=53

    • PLP: mean=2.51 (SD 2.91), median=2 (IQR 0 to 5)

    • RLP: mean=2.08 (SD 2.94), median=1 (IQR 0 to 3)

    NRS at 6 months after TMR, n=49

    • PLP: mean=2.37 (SD 3.2), median=1 (IQR 0 to 3)

    • RLP: mean=1.86 (SD 2.8), median=0 (IQR 0 to 3)

    NRS at 12 months after TMR, n=43

    • PLP: mean=2.05 (SD 2.71), median=0 (IQR 0 to 4)

    • RLP: mean=1.79 (SD 2.9), median=0 (IQR 0 to 2)

    NRS at 18 months or more after TMR, n=23

    • PLP: mean=0.96 (SD 1.69), median=0 (IQR 0 to 2)

    • RLP: mean=1.04 (SD 2.03), median=0 (IQR 0 to 2)

    PROMIS Interference score at 3 months after TMR, n=53

    • PLP: mean=46.38 (SD 9.43), median=40.7 (IQR 40.7 to 52.3)

    • RLP: mean=45.55 (SD 9.31), median=40.7 (IQR 40.7 to 47.9)

    PROMIS Interference score at 6 months after TMR, n=49

    • PLP: mean=45.29 (SD 8.6), median=40.7 (IQR 40.7 to 47.9)

    • RLP: mean=43.68 (SD 7.19), median=40.7 (IQR 40.7 to 40.7)

    PROMIS Interference score at 12 months after TMR, n=43

    • PLP: mean=45.23 (SD 9.2), median=40.7 (IQR 40.7 to 40.7)

    • RLP: mean=44.81 (SD 8.82), median=40.7 (IQR 40.7 to 40.7)

    PROMIS Interference score at 18 months or more after TMR, n=23

    • PLP: mean=42.4 (SD 4.93), median=40.7 (IQR 40.7 to 40.7)

    • RLP: mean=44.13 (SD 7.21), median=40.7 (IQR 40.7 to 40.7)

    PROMIS Behaviour score at 3 months after TMR, n=53

    • PLP: mean=47.83 (SD 8.73), median=50.1 (IQR 36.7 to 53)

    • RLP: mean=45.97 (SD 9.44), median=50.1 (IQR 36.7 to 51.1)

    PROMIS Behaviour score at 6 months after TMR, n=49

    • PLP: mean=46.7 (SD 9.45), median=50.1 (IQR 36.7 to 53.9)

    • RLP: mean=44.59 (SD 9.38), median=36.7 (IQR 36.7 to 51.1)

    PROMIS Behaviour score at 12 months after TMR, n=43

    • PLP: mean=46.62 (SD 9.64), median=50.1 (IQR 36.7 to 52.1)

    • RLP: mean=44.7 (SD 9.8), median=36.7 (IQR 36.7 to 53.9)

    PROMIS Behaviour score at 18 months or more after TMR, n=23

    • PLP: mean=45.61 (SD 9.06), median=50.1 (IQR 36.7 to 53.9)

    • RLP: mean=43.07 (SD 9.15), median=36.7 (IQR 36.7 to 53.9)

    Unadjusted pairwise analysis demonstrated a statistically significant difference in mean PLP NRS scores between 3 months and 18 months (mean difference −1.38, p=0.004), 6 months and 18 months (mean difference −1.14, p=0.02), and 12 months and 18 months or later (mean difference −1.02, p=0.04).

    There were no statistically significant differences in unadjusted pairwise comparisons for any time points (3 months onwards) regarding the NRS scores for RLP and the PROMIS pain interference or pain behaviour scores for PLP or RLP.

    No safety outcomes were reported.

    Goodyear, 2024

    Subsequent neuroma development (in the same nerve distributions included in original TMR operation)

    • Acute TMR=1.4% (1/73)

    • Delayed TMR=18.8% (6/32), p<0.05

    Neuroma development in a new distribution that was not previously included in TMR operation

    • Acute TMR=5.5% (4/73)

    • Delayed TMR=9.4% (3/32), p=0.433

    In multivariate analysis, those who had delayed TMR had 28.9 times greater odds of developing a subsequent neuroma compared with acute TMR, when controlling for age, sex, and extremity involved (95% CI 2.4 to 347.3; p=0.008).

    Complications

    The rate of major complications was 11.0% (8/73) for acute TMR and 12.5% (4/32) for delayed TMR (p=0.136).

    The rate of minor complications was 17.8% (13/73) for acute TMR and 15.6% (5/32) for delayed TMR (p=0.559).

    Major complications – acute TMR

    • Infection=5.5% (4/73)

    • Haematoma=1.4% (1/73)

    • Dehiscence=4.1% (3/73)

    Major complications – delayed TMR

    • Abscess=6.3% (2/32)

    • Haematoma=3.1% (1/32)

    • Dehiscence=3.1% (1/32)

    Minor complications – acute TMR

    • Superficial dehiscence=11.0% (8/73)

    • Cellulitis=5.5% (4/73)

    • Abscess=1.4% (1/73)

    Minor complications – delayed TMR

    • Superficial dehiscence=6.3% (2/32)

    • Cellulitis=3.1% (1/32)

    • Abscess=6.3% (2/32)

    Li, 2024

    Comparison of PROMs between paired baseline and follow-up surveys for delayed TMR

    Overall PLP score at 6 months, median (IQR)

    • Baseline=5.5 (2.8 to 6.3)

    • Follow-up=3.4 (1.5 to 4.5), p=0.01

    Overall PLP score at 12 months, median (IQR)

    • Baseline=5.5 (4.5 to 6.0)

    • Follow-up=3.5 (0.5 to 4.8), p=0.02

    Overall PLP score at 18 months, median (IQR)

    • Baseline=5.8 (4.5 to 6.8)

    • Follow-up=3.0 (1.0 to 4.0), p=0.03

    Overall PLP score at 24 months, median (IQR)

    • Baseline=5.5 (2.5 to 6.8)

    • Follow-up=3.5 (2.3 to 3.8), p=0.15

    Overall RLP score at 6 months, median (IQR)

    • Baseline=5.6 (4.8 to 6.5)

    • Follow-up=3.5 (2.8 to 4.8), p=0.02

    Overall RLP score at 12 months, median (IQR)

    • Baseline=5.8 (5.3 to 6.0)

    • Follow-up=4.0 (2.8 to 5.3), p=0.07

    Overall RLP score at 18 months, median (IQR)

    • Baseline=5.5 (4.5 to 6.0)

    • Follow-up=3.3 (0.0 to 4.5), p=0.01

    Overall RLP score at 24 months, median (IQR)

    • Baseline=5.8 (4.5 to 6.5)

    • Follow-up=4.0 (2.0 to 5.8), p=0.13

    Worst PLP score at 6 months, median (IQR)

    • Baseline=7.5 (4.0 to 9.0)

    • Follow-up=5.5 (2.0 to 8.0), p=0.06

    Worst PLP score at 12 months, median (IQR)

    • Baseline=8.0 (4.0 to 9.0)

    • Follow-up=7.0 (0.0 to 8.0), p=0.19

    Worst PLP score at 18 months, median (IQR)

    • Baseline=8.0 (4.0 to 9.0)

    • Follow-up=4.0 (2.0 to 7.0), p=0.14

    Worst PLP score at 24 months, median (IQR)

    • Baseline=7.0 (3.0 to 9.0)

    • Follow-up=6.0 (3.0 to 7.0), p=0.32

    Worst RLP score at 6 months, median (IQR)

    • Baseline=8.0 (7.0 to 10.0)

    • Follow-up=6.0 (4.0 to 7.0), p=0.02

    Worst RLP score at 12 months, median (IQR)

    • Baseline=8.0 (7.0 to 9.0)

    • Follow-up=7.0 (3.0 to 7.0), p=0.15

    Worst RLP score at 18 months, median (IQR)

    • Baseline=8.0 (7.0 to 8.0)

    • Follow-up=7.0 (0.0 to 8.0), p=0.13

    Worst RLP score at 24 months, median (IQR)

    • Baseline=8.0 (7.0 to 9.0)

    • Follow-up=5.0 (2.0 to 8.0), p=0.06

    Pain interference score at 6 months, median (IQR)

    • Baseline=7.7 (5.6 to 9.0)

    • Follow-up=4.4 (2.0 to 6.9), p=0.051

    Pain interference score at 12 months, median (IQR)

    • Baseline=7.2 (5.8 to 9.5)

    • Follow-up=2.8 (0.0 to 7.1), p=0.12

    Pain interference score at 18 months, median (IQR)

    • Baseline=7.7 (6.0 to 10.0)

    • Follow-up=6.0 (0.0 to 8.0), p=0.16

    Pain interference score at 24 months, median (IQR)

    • Baseline=7.9 (6.0 to 10.0)

    • Follow-up=7.1 (4.9 to 8.6), p=0.52

    Pain catastrophisation score at 6 months, median (IQR)

    • Baseline=34.0 (19.0 to 44.0)

    • Follow-up=10.5 (10.0 to 13.0), p=0.003

    Pain catastrophisation score at 12 months, median (IQR)

    • Baseline=30.0 (19.5 to 42.5)

    • Follow-up=5.5 (0.0 to 24.5), p=0.04

    Pain catastrophisation score at 18 months, median (IQR)

    • Baseline=33.0 (18.0 to 50.0)

    • Follow-up=1.0 (0.0 to 38.0), p=0.08

    Pain catastrophisation score at 24 months, median (IQR)

    • Baseline=34.0 (18.0 to 50.0)

    • Follow-up=27.5 (13.0 to 36.0), p=0.42

    Comparison of acute versus delayed TMR

    Overall PLP score at 6 months, median (IQR)

    • Acute TMR=0.0 (0.0 to 1.6)

    • Delayed TMR=3.4 (1.6 to 4.6), p<0.001

    Overall PLP score at 12 months, median (IQR)

    • Acute TMR=0.0 (0.0 to 0.8)

    • Delayed TMR=3.5 (0.5 to 6.0), p=0.002

    Overall PLP score at 18 months, median (IQR)

    • Acute TMR=0.0 (0.0 to 0.9)

    • Delayed TMR=3.0 (1.0 to 4.0), p=0.001

    Overall PLP score at 24 months, median (IQR)

    • Acute TMR=0.0 (0.0 to 0.0)

    • Delayed TMR=3.4 (2.5 to 3.6), p<0.001

    Overall RLP score at 6 months, median (IQR)

    • Acute TMR=0.5 (0.0 to 1.8)

    • Delayed TMR=3.5 (2.3 to 4.8), p<0.001

    Overall RLP score at 12 months, median (IQR)

    • Acute TMR=0.0 (0.0 to 1.8)

    • Delayed TMR=4.0 (0.8 to 6.0), p=0.002

    Overall RLP score at 18 months, median (IQR)

    • Acute TMR=0.0 (0.0 to 1.3)

    • Delayed TMR=3.3 (0.5 to 4.5), p=0.01

    Overall RLP score at 24 months, median (IQR)

    • Acute TMR=0.0 (0.0 to 0.0)

    • Delayed TMR=3.8 (1.4 to 5.3), p<0.001

    Worst PLP score at 6 months, median (IQR)

    • Acute TMR=0.0 (0.0 to 3.5)

    • Delayed TMR=5.5 (2.5 to 8.0), p=0.002

    Worst PLP score at 12 months, median (IQR)

    • Acute TMR=0.0 (0.0 to 2.0)

    • Delayed TMR=7.0 (0.0 to 8.0), p=0.006

    Worst PLP score at 18 months, median (IQR)

    • Acute TMR=0.0 (0.0 to 1.5)

    • Delayed TMR=4.0 (4.0 to 6.0), p=0.004

    Worst PLP score at 24 months, median (IQR)

    • Acute TMR=0.0 (0.0 to 0.0)

    • Delayed TMR=6.0 (4.0 to 6.5), p<0.001

    Worst RLP score at 6 months, median (IQR)

    • Acute TMR=1.0 (0.0 to 5.0)

    • Delayed TMR=6.0 (3.5 to 7.0), p=0.006

    Worst RLP score at 12 months, median (IQR)

    • Acute TMR=0.0 (0.0 to 3.0)

    • Delayed TMR=7.0 (1.0 to 9.0), p=0.006

    Worst RLP score at 18 months, median (IQR)

    • Acute TMR=0.0 (0.0 to 1.5)

    • Delayed TMR=7.0 (2.0 to 8.0), p=0.01

    Worst RLP score at 24 months, median (IQR)

    • Acute TMR=0.0 (0.0 to 0.0)

    • Delayed TMR=5.0 (1.0 to 8.0), p=0.002

    Pain interference score at 6 months, median (IQR)

    • Acute TMR=1.5 (0.0 to 2.9)

    • Delayed TMR=4.5 (2.0 to 6.4), p=0.04

    Pain interference score at 12 months, median (IQR)

    • Acute TMR=0.0 (0.0 to 2.4)

    • Delayed TMR=4.7 (0.0 to 7.4), p=0.048

    Pain interference score at 18 months, median (IQR)

    • Acute TMR=0.0 (0.0 to 1.5)

    • Delayed TMR=5.9 (2.0 to 6.6), p=0.03

    Pain interference score at 24 months, median (IQR)

    • Acute TMR=0.0 (0.0 to 0.0)

    • Delayed TMR=6.4 (3.9 to 7.9), p<0.001

    Pain catastrophisation score at 6 months, median (IQR)

    • Acute TMR=0.0 (0.0 to 6.0)

    • Delayed TMR=10.0 (3.0 to 13.0), p=0.006

    Pain catastrophisation score at 12 months, median (IQR)

    • Acute TMR=0.0 (0.0 to 5.0)

    • Delayed TMR=11.0 (8.0 to 32.0), p=0.003

    Pain catastrophisation score at 18 months, median (IQR)

    • Acute TMR=0.0 (0.0 to 0.0)

    • Delayed TMR=8.5 (0.0 to 33.0), p=0.03

    Pain catastrophisation score at 24 months, median (IQR)

    • Acute TMR=0.0 (0.0 to 0.0)

    • Delayed TMR=18.5 (10.0 to 34.0), p<0.001

    Mixed-effect models for pain outcomes – regression coefficients for acute versus delayed TMR, (95% CI)

    Overall PLP scores=-2.18 (-3.4 to ‑0.93), p=0.001

    Overall RLP scores=-1.81 (-2.9 to ‑0.72), p=0.001

    Worst PLP scores=-3.02 (-4.8 to ‑1.2), p=0.001

    Worst RLP scores=-2.42 (-4.1 to ‑0.74), p=0.01

    Pain interference=-2.23 (-4.1 to -0.38), p=0.02

    Pain catastrophisation=-7.88 (-16.0 to 0.25), p=0.06

    There was no statistically significant difference in complication rate between acute and delayed TMR (p=1.00).

    In the acute TMR group, 1 person developed a haematoma needing surgical evacuation and debridement.

    In the delayed TMR group, 1 person had delayed wound healing, and another developed an infection, both needing surgical debridement.

    All complications were grade 3B according to the Clavien–Dindo classification.

    Chang, 2021

    Proportion of people who were pain free at follow-up

    • TMR=71%

    • Non-TMR=36%, p<0.01

    Mean pain score for those people who had pain

    • TMR=3.2

    • Non-TMR=5.2, p<0.01

    Proportion of people with RLP

    • TMR=14%

    • Non-TMR=57%, p<0.01

    Proportion of people with PLP

    • TMR=19%

    • Non-TMR=47%, p<0.01

    Proportion of people taking opioids within 1 month of last follow-up

    • TMR=6%

    • Non-TMR=24%, p<0.01

    Mean dose in those taking opioids (morphine equivalent per day)

    • TMR=47.6

    • Non-TMR=37.4, p=0.66

    Proportion of people taking neuroleptic medication at last follow-up

    • TMR=42%

    • Non-TMR=48%, p=0.20

    Proportion of people who were ambulatory

    • TMR=90.9% (80/88); none were nonambulatory because of pain

    • Non-TMR=70.5% (55/78), p<0.01; 9 could not ambulate because of uncontrollable pain

    Two people in the TMR group developed symptomatic neuromas and PLP that needed a secondary TMR, 1 involving the sural nerve and 1 involving the saphenous nerve.

    Surgical complications

    Stump wounds or infections that needed operative debridement and revision

    • TMR=16%

    • Non-TMR=30%, p=0.02

    Mean number of additional interventions when revision was needed

    • TMR=2.3

    • Non-TMR=2.0

    Mortality at 12 months

    • TMR=4.9%

    • Non-TMR=6.0%, p=0.80

    Chang, 2024

    Proportion of people who reported any type of pain

    • TMR=41.5%

    • Non-TMR=67.2%, p=0.01

    Proportion of people who reported RLP

    • TMR=26.8%

    • Non-TMR=44.8%, p=0.04

    Proportion of people who reported PLP

    • TMR=19.5%

    • Non-TMR=43.1%, p=0.01

    Overall pain severity in those who reported pain (range 0 to 10)

    • TMR=4.9

    • Non-TMR=5.5, p=0.64

    Proportion of people taking narcotics more than 1 month after amputation

    • TMR=9.8%

    • Non-TMR=25.9%, p=0.05

    Proportion of people taking neuroleptic medication

    • TMR=56.4%

    • Non-TMR=60.3%, p=0.70

    Proportion of people with a minimum 3-month follow-up who were ambulatory with a prosthetic

    • TMR=41.9%

    • Non-TMR=22.7%, p=0.11

    Surgical complications

    Superficial wounds needing woundcare only

    • TMR=22.0%

    • Non-TMR=19.0%, p=0.72

    Need for operative stump revision

    • TMR=14.6%

    • Non-TMR=32.8%, p=0.22

    2.4% of people who had TMR were offered a revision TMR procedure for a symptomatic stump neuroma and PLP.

    Mortality at 3 months

    • TMR=12.2%

    • Non-TMR=3.4% (p value not reported)

    Mortality at 12 months

    • TMR=26.5%

    • Non-TMR=15.5% (p value not reported)

    Smith, 2024

    Reoperation after primary amputation

    • TMR=21.4% (6/28)

    • No TMR=22.7% (17/75), p=0.11

    Rate of revision for symptomatic neuroma

    • TMR=3.6% (1/28)

    • No TMR=4.0% (3/75), p=0.97

    Mean VAS score for pain at 2 weeks

    • TMR=3.26 (n=27)

    • No TMR=3.33 (n=64), p=0.91

    Mean VAS score for pain at 6 weeks

    • TMR=2.21 (n=28)

    • No TMR=1.83 (n=64), p=0.43

    Mean decrease in VAS score between 2 and 6 weeks

    • TMR=0.96, p=0.06

    • No TMR=1.5, p=0.0002

    No safety outcomes were reported

    Procedure technique

    The site of amputations varied and included upper and lower limbs, so the nerves involved in the TMR procedures varied within and between studies. The TMR was done as a delayed procedure only in 2 studies (Dumanian 2019 and Kang 2022). It was done at the same time as the amputation in 5 studies (Shammas 2022, O'Brien 2022, Chang 2021, Chang 2024 and Smith 2024). Two studies compared outcomes between acute and delayed TMR (Goodyear 2024, Li 2024). In the 10 studies included in the systematic review by Tham et al. (2023), the procedure was done either at the same time as the amputation or as a later secondary procedure.

    Efficacy

    Residual limb pain (RLP)

    In the systematic review of 10 studies by Tham et al. (2023), the pooled mean difference in NRS for RLP for TMR compared with control was -2.68 (95% CI ‑3.21 to ‑2.14, p<0.0001; 4 studies, I2=0%) The pooled mean difference for PROMIS intensity score was -13.4 (95% CI -14.6 to -12.2, p<0.0001; 3 studies, I2=61%). The pooled mean difference for PROMIS behavioural score was -12.0 (95% CI -13.8 to -10.2, p<0.0001; 4 studies, I2=87%) and the pooled mean difference for PROMIS interference score was -12.2 (95% CI -13.6 to ‑10.8, p<0.0001; 4 studies, I2=70%).

    In the randomised controlled trial of 28 people (4 upper and 26 lower limbs), the worst pain score (NRS) for RLP reduced from 6.6 at baseline to 3.7 at 1 year in the TMR group and from 6.9 to 6.0 in the control group. The mean difference of change scores was 1.9 (adjusted 95% CI -0.5 to 4.4). The proportion of people with no or mild RLP at last follow-up was 67% in the TMR group and 27% in the control group (p value not stated). The PROMIS intensity score for RLP reduced from 55.7 at baseline to 44.5 at 1 year in the TMR group and from 55.0 to 49.5 in the control group. The mean difference of change scores was 5.8 (adjusted 95% CI -0.9 to 12.4). The PROMIS behaviour score for RLP reduced from 61.5 at baseline to 56.8 at 1 year in the TMR group and from 61.9 to 56.6 in the control group. The mean difference of change scores was -0.5 (adjusted 95% CI -7.2 to 6.1). The PROMIS interference score for RLP reduced from 64.4 at baseline to 56.8 at 1 year in the TMR group and from 65.8 to 57.4 in the control group. The mean difference of change scores was -0.9 (adjusted 95% CI -8.5 to 6.7; Dumanian 2019).

    In the prospective case series of 81 people (83 upper or lower limbs), the mean worst pain score (NRS) for RLP was 2.08 at 3 months after TMR (n=53), 1.86 at 6 months (n=49), 1.79 at 12 months (n=43) and 1.04 at 18 months or more (n=23). The changes in scores from 3 months onwards were not statistically significant. The mean PROMIS interference score for RLP was 45.6 at 3 months after TMR and the mean PROMIS behaviour score was 46.0. The changes from 3 months onwards were not statistically significant (O'Brien 2022).

    In the retrospective cohort study of 32 people (38 upper or lower limbs) who had acute or delayed TMR, the overall RLP score for delayed TMR was 3.5 at 6 months follow-up compared with 5.6 at baseline (p=0.02). The score was 4.0 at 12 months (p=0.07), 3.3 at 18 months (p=0.01) and 4.0 at 24 months (p=0.13). The median scores for acute TMR were statistically significantly lower than for delayed TMR for all timepoints. The worst RLP score for delayed TMR was 6.0 at 6 months follow-up compared with 8.0 at baseline (p=0.02). The score was 7.0 at 12 months (p=0.15), 7.0 at 18 months (p=0.13) and 5.0 at 24 months (p=0.06). Again, the median scores for acute TMR were statistically significantly lower than for delayed TMR for all timepoints (Li 2024).

    In the retrospective cohort study of 200 people who had below-knee amputation with TMR or traction neurectomy and muscle implantation, 14% of those in the TMR group had RLP at follow-up compared with 57% of those in the control group (p<0.01; Chang 2021). In the cohort study of 99 people who had through- or above-knee amputation with TMR or traction neurectomy and muscle implantation, 27% of those in the TMR group had RLP at follow-up compared with 45% of those in the control group (p=0.04; Chang 2024).

    Phantom limb pain

    In the systematic review of 10 studies by Tham et al. (2023), the pooled mean difference in NRS for PLP for TMR compared with control was -2.17 (95% CI ‑2.70 to ‑1.63, p<0.0001; 4 studies, I2=51%) The pooled mean difference for PROMIS intensity score was -11.2 (95% CI -13.0 to -9.45, p<0.0001; 3 studies, I2=61%). The pooled mean difference for PROMIS behavioural score was -10.4 (95% CI -12.4 to -8.52, p<0.0001; 4 studies, I2=76%) and the pooled mean difference for PROMIS interference score was -11.5 (95% CI -12.8 to ‑10.1, p<0.0001; 4 studies, I2=62%).

    In the randomised controlled trial of 28 people (4 upper and 26 lower limbs), the worst pain score (NRS) for PLP reduced from 5.8 at baseline to 2.6 at 1 year in the TMR group and increased from 3.9 to 4.1 in the control group. The mean difference of change scores was 3.4 (adjusted 95% CI -0.1 to 6.9). The proportion of people with no or mild PLP at last follow-up was 72% in the TMR group and 40% in the control group (p value not stated). The PROMIS intensity score for PLP reduced from 52.4 at baseline to 38.0 at 1 year in the TMR group and from 48.3 to 45.8 in the control group. The mean difference of change scores was 11.7 (adjusted 95% CI -0.3 to 23.7). The PROMIS behaviour score for PLP reduced from 58.3 at baseline to 50.7 at 1 year in the TMR group and from 58.5 to 52.0 in the control group. The mean difference of change scores was 1.1 (adjusted 95% CI -8.3 to 10.5). The PROMIS interference score for PLP reduced from 60.2 at baseline to 50.4 at 1 year in the TMR group and from 57.9 to 52.8 in the control group. The mean difference of change scores was 4.7 (adjusted 95% CI -5.0 to 14.3; Dumanian 2019).

    In the retrospective case series of 36 people (40 upper or lower limbs), the mean change in NRS for PLP at 12 month follow-up was 4.4 (95% CI 7.29 to 1.52, p=0.007) for upper limbs and 2.6 (95% CI 4.46 to 0.74, p=0.009) for lower limbs. People with upper limb amputation had a temporary worsening of PLP within the first 3 months. Of the 10 people with upper limb amputation, 1 was pain free at 12 months and 5 had only mild pain (NRS between 1 and 3). Of the 20 lower limb procedures, 4 people were pain free at 12 months and 3 had only mild pain (Kang 2022).

    In the prospective case series of 81 people (83 upper or lower limbs), the mean worst pain score (NRS) for PLP was 2.51 at 3 months after TMR (n=53), 2.37 at 6 months (n=49), 1.05 at 12 months (n=43) and 0.96 at 18 months or more (n=23). Unadjusted pairwise analysis demonstrated a statistically significant difference in mean PLP NRS scores between 3 months and 18 months (mean difference −1.38, p=0.004), 6 months and 18 months (mean difference −1.14, p=0.02), and 12 months and 18 months or later (mean difference −1.02, p=0.04). The mean PROMIS interference score for PLP was 46.4 at 3 months after TMR and the mean PROMIS behaviour score was 47.8. The changes from 3 months onwards were not statistically significant (O'Brien 2022).

    In the retrospective cohort study of 32 people (38 upper or lower limbs) who had acute or delayed TMR, the median overall PLP score for delayed TMR was 3.4 at 6 months follow-up compared with 5.5 at baseline (p=0.01). The score was 3.5 at 12 months (p=0.02), 3.0 at 18 months (p=0.03) and 3.5 at 24 months (p=0.15). The median scores for acute TMR were statistically significantly lower than for delayed TMR for all timepoints. The worst PLP score for delayed TMR was 5.5 at 6 months follow-up compared with 7.5 at baseline (p=0.06). The score was 7.0 at 12 months (p=0.19), 4.0 at 18 months (p=0.14) and 6.0 at 24 months (p=0.32). Again, the median scores for acute TMR were statistically significantly lower than for delayed TMR for all timepoints (Li 2024).

    In the retrospective cohort study of 200 people who had below-knee amputation with TMR or traction neurectomy and muscle implantation, 19% of those in the TMR group had PLP at follow-up compared with 47% of those in the control group (p<0.01; Chang 2021). In the cohort study of 99 people who had through- or above-knee amputation with TMR or traction neurectomy and muscle implantation, 20% of those in the TMR group had RLP at follow-up compared with 43% of those in the control group (p=0.01; Chang 2024).

    Neuroma pain

    In the retrospective case series of 36 people (40 upper or lower limbs), the mean change in NRS for neuroma pain at 12 month follow-up was 5.30 (95% CI 8.61 to 2.00, p=0.006) for upper limbs and 4.35 (95% CI 6.47 to 2.23, p<0.0001) for lower limbs. Of the 10 people with upper limb amputation, all pain had resolved at 12 months. Of the 20 lower limb procedures, 10 resulted in complete resolution of neuroma pain and 2 people had only mild pain (Kang 2022).

    Neuroma development

    In the retrospective cohort study of 103 people (105 upper or lower limbs) who had acute or delayed TMR, neuroma development in the same nerve distributions included in the original TMR procedure was reported after 1% (1 out of 73) of acute TMR procedures and 19% (6 out of 32) of delayed TMR procedures (p<0.05). Neuroma development in a new nerve distribution was reported after 6% (4 out of 73) of acute TMR procedures and 9% (3 out of 32) of delayed TMR procedures (p=0.433). In multivariate analysis, those who had delayed TMR had 29 times greater odds of developing a subsequent neuroma compared with acute TMR, when controlling for age, sex, and extremity involved (95% CI 2.4 to 347.3; p=0.008; Goodyear 2024). In the retrospective cohort study of 200 people who had below-knee amputation with TMR or traction neurectomy and muscle implantation, 2 people in the TMR group developed symptomatic neuromas and PLP that needed a secondary TMR, 1 involving the sural nerve and 1 involving the saphenous nerve (Chang 2021). In the retrospective study of 112 primary amputations, the rate of revision for symptomatic neuroma was 3.6% (1 out of 28) in the TMR group and 4.0% (3 out of 75) in the non-TMR group (p=0.97; Smith 2024).

    Unspecified pain

    In the retrospective cohort study of 200 people who had below-knee amputation with TMR or traction neurectomy and muscle implantation, 71% of those in the TMR group and 36% of those in the control group were pain free at follow-up (p<0.01). The mean pain scores for those people who had pain were 3.2 in the TMR group and 5.2 in the control group (p<0.01; Chang 2021). In the cohort study of 99 people who had through- or above-knee amputation with TMR or traction neurectomy and muscle implantation, 42% of those in the TMR group and 67% of those in the control group reported any type of pain (p=0.01). The overall pain severity in those who reported pain was 4.9 in the TMR group and 5.5 in the control group (p=0.64; Chang 2024). In the retrospective study of 112 primary amputations, the mean VAS score for pain at 2 weeks was 3.26 in those who had TMR and 3.33 in those who did not have TMR (p=0.91). At 6 weeks, the scores were 2.21 and 1.83, respectively (p=0.43). The mean decrease in VAS score between 2 and 6 weeks was 0.96 in the TMR group (p=0.06) and 1.5 in the non-TMR group (p=0.0002; Smith 2024).

    Medication use

    In the retrospective case series of 36 people (40 upper or lower limbs), 9 of the 18 people who were taking pregabalin before the procedure had discontinued it after 1 year. There was a mean reduction of 352 mg in daily intake over the 12 months of follow-up (p<0.01; Kang 2022). In the retrospective cohort study of 200 people who had below-knee amputation with TMR or traction neurectomy and muscle implantation, 6% of those in the TMR group and 24% of those in the control group were taking opioids within 1 month of the last follow-up (p<0.01). The proportion of people taking neuroleptic medication at last follow-up was 42% in the TMR group and 48% in the control group (p=0.20; Chang 2021). In the cohort study of 99 people who had through- or above-knee amputation with TMR or traction neurectomy and muscle implantation, 10% of those in the TMR group and 26% of those in the control group were taking narcotics more than 1 month after amputation (p=0.05). The proportion of people taking neuroleptic medication was 56% in the TMR group and 60% in the control group (p=0.70; Chang 2024).

    Functional outcomes

    In the systematic review of 10 studies, 1 study reported that the OPUS score for those who had upper limb amputations and TMR increased from 53.7 to 56.4 at 1 year follow-up (p<0.01). The Neuro-QoL score for those with lower limb amputations and TMR increased from 32.9 to 35.2 (p<0.01). Both outcomes showed higher functional scores in the TMR group (Tham 2023). In the randomised controlled trial of 28 people, the lower extremity Neuro-QoL results (n=24) showed little difference between the groups at 1 year. When crossover data were included and at final follow-up, the mean score increased from 39.9 to 45.2 in the TMR cohort showing functional improvement (Dumanian 2019).

    In the retrospective cohort study of 200 people who had below-knee amputation with TMR or traction neurectomy and muscle implantation, 91% of those in the TMR group and 71% of those in the control group were ambulatory at last follow-up (p<0.01). In the control group, 9 people could not ambulate because of uncontrollable pain (Chang 2021). In the cohort study of 99 people who had through- or above-knee amputation with TMR or traction neurectomy and muscle implantation, 42% of those in the TMR group and 23% of those in the control group with a minimum of 3 months follow-up were ambulatory with a prosthetic (p=0.11; Chang 2024).

    Patient satisfaction

    In the retrospective case series of 36 people (40 upper or lower limbs), 22 people reported data on satisfaction. Of those, 91% indicated overall satisfaction with the procedure at 12 months but only 50% felt they would have agreed to a prophylactic TMR procedure (Kang 2022).

    Operative time

    In the propensity score-matched study of 96 people who had below-knee amputation with or without TMR, the mean length of surgery was statistically significantly longer in the TMR group (189 minutes) than the group without TMR (88 minutes; p<0.001; Shammas 2022).

    Safety

    General complications

    Complication rates ranged from 0 to 16% in the 4 studies that reported them in the systematic review of 10 studies. Complications were mostly wound site or stump infections (Tham 2023).

    Readmission

    Readmission was reported for 26% (8 out of 31) of people who had TMR at the same time as a below-knee amputation and 19% (12 out of 65) of people who had amputation without TMR (p=0.21; Shammas 2022).

    Reoperation (when reported as a safety outcome)

    Reoperation within 60 days was reported for 19% (6 out of 31) of people who had TMR at the same time as a below-knee amputation and 11% (7 out of 65) of people who had amputation without TMR (p=0.13; Shammas 2022). Operative stump revision was reported for 15% of those in the TMR and 33% in the control group (p=0.22) in the cohort study of 99 people who had through- or above-knee amputation with TMR or traction neurectomy and muscle implantation (Chang 2024).

    Admission to intensive care unit

    Admission to an intensive care unit was reported for 7% (2 out of 31) of people who had TMR at the same time as a below-knee amputation and 11% (7 out of 65) of people who had amputation without TMR (p=0.41; Shammas 2022).

    Mortality

    All-cause mortality was 3% (1 out of 31) for those who had TMR at the same time as a below-knee amputation and 9% (6 out of 65) for those who had amputation without TMR (p=0.34; Shammas 2022). Mortality at 12 months was 5% in the TMR group and 6% in the control group (p=0.80) in the cohort study of 200 people (Chang 2021). Mortality at 3 months was 12% in the TMR group and 3% in the control group (p value not reported) in the cohort study of 99 people (Chang 2024).

    Wound healing complication

    Wound healing complications were reported for 45% (14 out of 31) of people who had TMR at the same time as a below-knee amputation and 34% (22 out of 65) of people who had amputation without TMR (p=0.25; Shammas 2022). In the case series of 36 people (11 upper and 29 lower limbs), wound dehiscence, haematoma and seroma were each reported after 1 upper limb procedure. In the lower limb group, there were 3 reports of wound dehiscence, 2 reports each of haematoma and ulceration and 1 report each of seroma and lymphatic discharge (Kang 2022). Haematoma was reported after 1% (1 out of 73) of acute TMR procedures and 3% (1 out of 32) of delayed TMR procedures in the cohort study of 103 people. In the same study, dehiscence was reported in 4% (3 out of 73) of acute TMR procedures and 3% (1 out of 32) of delayed TMR procedures and superficial dehiscence was reported in 11% (8 out of 73) and 6% (2 out of 32), respectively (Goodyear 2024). One person had a haematoma that needed surgical evacuation and debridement after acute TMR and 1 person had delayed wound healing that needed debridement after delayed TMR, in the cohort study of 32 people (Li 2024).

    Infection

    Infection was reported for 10% (3 out of 31) of people who had TMR at the same time as a below-knee amputation and 6% (4 out of 65) of people who had amputation without TMR (p<0.001; Shammas 2022). In the case series of 36 people (11 upper and 29 lower limbs), infection was reported after 4 upper limb procedures and 7 lower limb procedures (Kang 2022). Infection was reported after 6% (4 out of 73) of acute TMR procedures and abscess was reported after 6% (2 out of 32) of delayed TMR procedures in the cohort study of 103 people. Minor abscess was reported in 1% (1 out of 73) of acute TMR procedures and 6% (2 out of 32) of delayed TMR procedures (Goodyear 2024). One person had an infection that needed surgical debridement after delayed TMR, in the cohort study of 32 people (Li 2024). Stump wounds or infections that needed surgical debridement and revision were reported in 16% of the group who had TMR and 30% of the group who did not have TMR (p=0.02) in the cohort study of 200 people (Chang 2021).

    Paraesthesia

    In the case series of 36 people who had TMR (11 upper and 29 lower limbs), paraesthesia was reported after 1 upper limb procedure and 4 lower limb procedures (Kang 2022).

    Unmasking of neuroma

    In the case series of 36 people who had TMR (11 upper and 29 lower limbs), unmasking of neuromas was reported after 12 lower limb procedures. This typically happened within a few weeks of surgery and 4 people needed an additional TMR procedure (Kang 2022).

    Cellulitis

    Minor cellulitis was reported after 6% (4 out of 73) of acute TMR procedures and 3% (1 out of 32) of delayed TMR procedures in the cohort study of 103 people (Goodyear 2024).

    Anecdotal and theoretical adverse events

    Expert advice was sought from consultants who have been nominated or ratified by their professional society or royal college. They were asked if they knew of any other adverse events for this procedure that they had heard about (anecdotal), which were not reported in the literature. They were also asked if they thought there were other adverse events that might possibly occur, even if they had never happened (theoretical).

    They listed the following anecdotal or theoretical adverse events:

    • Neuroma in continuity

    • Worsening of pain

    • Insensate stump

    • Loss of function of target muscles

    • Muscle wasting leading to change in stump shape, making it necessary to refit the socket.

    Eleven professional expert questionnaires for this procedure were submitted. Find full details of what the professional experts said about the procedure in the specialist advice questionnaires for this procedure.

    Validity and generalisability

    • Most of the evidence was from the US but there is some data from the UK.

    • The evidence includes people who had upper or lower limb amputation for a variety of reasons. There were different levels of amputation and the TMR procedures involved different nerves.

    • The reason for amputation may have an impact on the outcomes. In particular, the presence of peripheral vascular disease may be a confounding factor. In 3 studies, the main reasons for amputation were infection and ischaemia and these all noted a high level of comorbidity in the study population (Shammas 2022, Chang 2021, Chang 2024).

    • Some TMR procedures were done prophylactically at the same time as the amputation rather than to treat refractory pain afterwards. Two studies were designed to compare the 2 approaches (Goodyear 2024, Li 2024).

    • Most of the studies were small and many were retrospective.

    • The randomised controlled trial by Dumanian et al. (2019) was stopped early with recruitment of 28 patients rather than the intended sample size of 200.

    • Although the mean follow-up was less than a year in several studies, there were some reports with longer follow-up.

    • It may be difficult for patients to distinguish RLP from PLP.

    • The authors of Dumanian et al. (2019) noted that the 3 supplemental PROMIS item banks that were included in their outcome measures had not yet been validated in people living with chronic postamputation pain.

    Ongoing trials

    Comments panel open