Hereditary Neuropathy with Liability to Pressure Palsies (HNPP)

Treatment

What Works

Dark cinematic study of protective positioning and load management

Tier 1 — Strong Evidence None exists

There is no clinical practice guideline for this condition from NICE, APTA, BOA, EULAR, ACR or JOSPT. No Cochrane review. No meta-analysis. No adequately powered randomised trial. The entire randomised literature for HNPP is a single n-of-1 trial. Saying so plainly is more useful than dressing up weaker evidence.

Tier 2 — Moderate Evidence Moderate

Both of these are about recognising the condition rather than treating it, which is an honest description of where the evidence actually is.

Screen for pain with a validated tool

Baseline, then every 4-6 weeks · painDETECT was the instrument used in the physical-therapy-led study

Pain is present in 52% to 75% of patients across four independent estimates, and pain at the moment of testing is the single significant independent predictor of quality of life.

Refer for multi-nerve electrodiagnostics plus genetic testing

Nerve conduction studies across several nerves including asymptomatic ones · PMP22 deletion testing by MLPA

The diffuse background neuropathy separates this from CIDP and from diabetic nerve damage at p<0.0001. Testing only the painful nerve is how the diagnosis gets missed for decades.

Tier 3 — Emerging / Clinical Reasoning Low

Mechanical exposure management

Elbow padding · wrist splints at night · position changes every 30-45 minutes · cushioning at pressure points

Recommended in every review of this condition, and measured in none of them. It is the default because the mechanism supports it and it costs nothing, not because it has been tested.

Removing high-volume loaded shoulder-girdle work

Push-ups, planks, dips and heavy pressing pulled while symptoms are active, then reintroduced one variable at a time

In 36 young soldiers, 19 presented with a brachial plexus lesion rather than a classic entrapment, and push-ups were the leading identified cause. This is an exposure signal, not a proven contraindication.

Neuropathic pain and central sensitisation management

Extrapolated from general practice · no HNPP-specific outcome data exists

Corticosteroids for a protracted recovery

A neurology decision, listed so the physical therapist can prompt the referral · two adolescents, no control arm

Exercise Prescription

Read this before the numbers. Every set, rep and frequency below is a conventional safe starting point, not a value taken from a study of this condition, because no study of exercise or physical therapy in HNPP has ever been done. They are here because a plan with no numbers is not a plan, and they are labelled because inventing evidence is worse than admitting there is none.

Gentle nerve gliding for the affected limb

2 × 8-10 slow reps · once or twice daily · stop if it produces pins and needles or numbness

Mid-range strength work for the weakened muscles

2-3 × 10-12 at a comfortable weight · every other day · effort is fine, numbness is not

Position changes through the day

Every 30-45 minutes when sitting or working · uncross the legs, elbow off the armrest

General aerobic exercise you enjoy

20-30 minutes · 3-5 × weekly · conversational effort · watch handlebar pressure if cycling

Held back on purpose: heavy push-ups, planks, dips and heavy pressing. Not banned forever. These are the specific movements linked to arm and shoulder nerve injury in people with this condition, so they come out while we learn how you respond, then go back in gradually. Everything else trains at full load. There is no injured tissue here that rest would heal, so blanket deloading buys nothing.

What Doesn't Work

Cubital tunnel decompression. One of five patients improved and two of five got worse. That is the site affected in 85.7% of people with this condition, and it is the one place the surgery performed worst. Five patients is not much evidence, and it is the only comparative evidence there is.
Treating each episode as an isolated local problem and discharging on recovery. This is not a failure of technique. The local treatment usually works, the palsy resolves, and that success is exactly what stops anyone joining the dots. Average age at first symptom is 26. Average age at diagnosis is 33. The longest recorded gap is 40 years.
Using "painless" as a reason to rule it out. The condition's own name teaches clinicians to discount the symptom most of these patients actually have.
Promising that avoiding a list of postures will stop the episodes. Nearly 39% of episodes have no identifiable trigger at all. Managing exposure is worth doing and is not a guarantee.

Safety First

Red Flags

Dark cinematic anatomical study of peripheral nerve compression points

Get seen if any of these apply

Weakness or numbness in a NEW area, when you have had a "trapped nerve" somewhere else before. This is the pattern that makes the diagnosis. Ask for nerve conduction studies across several nerves plus PMP22 genetic testing.
Both arms weak, or shoulder and upper-arm weakness, after loaded shoulder work. Push-ups, planks or heavy pressing, particularly in a young adult. This was the single commonest presentation in the one cohort that recorded what people were doing.
A palsy that has not started to recover. Around 87% recover fully after an episode. One that does not needs neurological review.
Weakness that is getting steadily worse rather than coming in episodes, several nerve areas affected at once, or any change in bladder or bowel control. This is not a squashed nerve.
Prolonged or unexplained nerve problems after any operation. A person who carried this gene and had never had a symptom developed a foot drop after a knee replacement.
Severe pain out of proportion, a tight swollen compartment, or pain when the limb is stretched. That is acute compartment syndrome, a surgical emergency, and nothing to do with this condition.

Where to go: neurology or a peripheral nerve service for a suspected first diagnosis, because confirmation needs electrodiagnostics plus genetic testing and neither happens in a musculoskeletal clinic. Emergency department for progressive weakness, compartment signs, or bladder and bowel change.

Return to Activity

Return to Training

Concrete and binary, not "when it feels ready".

Full recovery of power in the affected area, or a documented plateau accepted by the referring neurologist.
No numbness, tingling or weakness during a full session or in the 24 hours after it.
Loaded shoulder-girdle work reintroduced progressively and tolerated at target volume for 2 consecutive weeks without neurological symptoms.
The provoking exposure identified and changed, or explicitly recorded as unidentified (which is the case for nearly 39% of episodes).
Pain screened with a validated tool, and either absent or under active management.
You can state your own red flags, and you know to declare this diagnosis before any surgery.

The Takeaway

Ask one question, right now, of yourself or of your patient: "Has this ever happened before, somewhere else?" If the answer is yes, this stops being a local trapped nerve and becomes a whole-body question. Write the previous episodes down, with roughly when and where, and take that list to your doctor.

The Verdict

Nerves with thin insulation get switched off by ordinary pressure. And it was never painless.

Think of a nerve as an electrical cable wrapped in many layers of tape. Your body makes a protein whose job is to hold those layers in register so they cannot slide against each other. In this condition you inherited only one working copy of the gene for it, so there is less glue than there should be. Squash or stretch the cable and the layers slip apart, the signal drops out, and the hand or foot stops working.

The cable is not cut, which is why it comes back. But the tape is thin along the whole cable, not just where it hurt this time. That is why it happens again, somewhere new, and why treating each episode as a separate local problem never gets ahead of it.

What this actually is: one inherited condition, not three separate trapped nerves and not bad luck.
The myth that won't die: that it is painless. It has been called "painless palsies" since the 1950s, and when researchers finally handed patients a proper pain questionnaire, three quarters of them had pain.
Start here: tell every clinician you see, and above all any surgeon or anaesthetist before a procedure. Simple padding on an operating table prevents nerve damage that otherwise lasts.

Best for

Anyone who has had more than one trapped nerve, in more than one place, without a clear cause. Especially if a relative has had a dropped foot or a "nerve problem".

Skip if

You have had a single, first entrapment with an obvious mechanical cause. This page is about the recurring pattern, not about one carpal tunnel.

Want the full evidence? Keep scrolling.⌄

Trust Anchor

Conviction: Low Endpoint-stratified

One number would hide the shape of this evidence. Some parts of it are solid and some do not exist at all, so here is every endpoint separately.

Genetic mechanism and the PMP22 deletion as the cause	High
The characteristic electrodiagnostic pattern	Moderate
Pain is common, roughly half to three quarters of patients	Moderate
The condition is substantially under-diagnosed	Moderate
Loaded shoulder-girdle exercise as a provoking exposure	Low-Mod
Cubital tunnel decompression carries net harm	Low
Carpal tunnel release benefits selected patients	Low
Corticosteroids for a protracted recovery	Low
Any physical therapy or exercise intervention	No evidence

What would change the "no evidence" verdict on physical therapy?

A prospective cohort of at least 60 genetically confirmed patients recruited from first-contact musculoskeletal and neuromuscular clinics and followed for 12 months, comparing a structured protocol (education on mechanical exposure, load redistribution across the shoulder girdle, splinting and padding at the two highest-yield sites, and formal neuropathic pain screening) against usual care, with annualised palsy episode count and SF-36 physical component score as co-primary endpoints. A greater than 30% reduction in episodes, or a better-than-threshold SF-36 gain, would move this from NO EVIDENCE to MODERATE. Nothing smaller will settle it, because episodes are infrequent and highly variable between people.

What would change the surgical harm signal?

A multicentre retrospective cohort of at least 100 patients undergoing decompression at any site, with before-and-after strength grading and patient-reported outcome, split by site and by whether symptoms were activity-provoked. If cubital tunnel decompression showed a worsening rate below 10%, the current signal would be downgraded to noise. As it stands it rests on five patients, and it is the only comparative evidence in existence.

If you have collected more than one "trapped nerve" and nobody has joined them up, you deserve better than guessing. The Verdict sends one evidence-graded protocol a week, free.

Join The Verdict, free

The Full Picture — Anatomy, Diagnosis & Evidence

What's Actually Going On

Dark cinematic study of myelinated peripheral nerve structure

One copy of the PMP22 gene is deleted, almost always as part of a 1.4-Mb deletion on chromosome 17. Across 1,734 genetically tested families, that deletion accounted for 93.4% of cases. Charcot-Marie-Tooth disease type 1A is the mirror image of the same genetic accident, a duplication of the identical stretch, which is why both run in the same families.

The protein appears to hold the layers of myelin in register and stop them sliding lengthwise against one another. With one copy missing there is less of it, the layers slip under mechanical load, and a degree of compression or traction that a normal nerve absorbs without consequence produces a block in conduction. Under the microscope the sheath carries sausage-shaped thickenings called tomacula, which is why the condition's other name is tomaculous neuropathy.

Two things follow, and they drive everything on this page. First, the nerve damage in this condition looks remarkably like the damage in a genuinely trapped nerve. That is not an analogy. It is exactly why every episode gets a local diagnosis and local treatment. Second, the background neuropathy is everywhere: nerve conduction is abnormal in nerves that have never caused a symptom, and in gene carriers who have never had an episode at all. The lesion is diffuse. Only the symptoms are focal.

How to Identify It

Dark cinematic study of clinical neurological examination

There is no clever clinical test. There is one question, and it is the whole diagnosis: has a palsy like this happened before, somewhere else? Recurrent episodes occur in 62% of pooled cases, and patients rarely volunteer the previous ones, because they got better.

Nerve conduction studies across multiple nerves, including ones that have never hurt. Sn: not measured | Sp: not measured Slow sensory conduction in 93% of nerves, prolonged distal motor latencies in 78%, prolonged F-waves in 90%, with motor conduction velocity largely normal.
PMP22 deletion testing by MLPA. Sn: not measured | Sp: not measured The reference standard. Misses the roughly 7% of cases caused by sequence variants, which need sequencing instead.
Nerve ultrasound. Sn: not measured | Sp: not measured Shows focal enlargement at entrapment sites, as opposed to the diffuse enlargement seen in Charcot-Marie-Tooth type 1.

"Not measured" above means never measured, for any test, in this condition, and not "we could not find it". No sensitivity, specificity or likelihood ratio has ever been published for any diagnostic test in HNPP against a genetic reference standard.

The one quantified discriminator that does exist is a group comparison rather than a test statistic: mean sensory conduction velocity was 85.6% ± 10.6% of the lower limit of normal, against 114.3% ± 20.1% in CIDP and 108.1% ± 14.8% in diabetic polyneuropathy, both p<0.0001, and removing the carpal tunnel from the analysis did not change it.

Dark cinematic study of peripheral nerve distribution across the limbs

A negative family history does not lower suspicion. It is absent in 35% to 41% of confirmed cases across three separate cohorts, and one of the two published training-induced cases was explicitly sporadic. Where the nerves get hit: ulnar at the elbow in 85.7%, median at the wrist in 84.4%, common peroneal at the fibular head in 36.7%. Common peroneal, ulnar and brachial plexus together account for roughly three quarters of everything.

It is also routinely diagnosed as something else. Published misdiagnoses include CIDP, Guillain-Barré syndrome treated with immunoglobulin before correction, lumbar radiculopathy presenting as low back pain, motor neuron disease, Charcot-Marie-Tooth, and fibromyalgia: in one series, all nine patients whose pain was purely musculoskeletal met the formal criteria for fibromyalgia syndrome.

The Debate

"Painless palsies" (the classical description, still in reviews) vs pain in the majority (systematic review, 2026)

A 2026 systematic review of 53 studies and 655 patients grouped the literature into three diagnostic eras and found reported pain prevalence climbing steadily: a painless paradigm in early work, 34% between 1990 and 2015, and 56% in studies since 2015. Only 4 of the 53 studies (7.5%) used a validated pain instrument, and those consistently reported higher prevalence and more complex symptoms. Three independent cohorts land at 52%, 74% and 75%.

Which to follow: the recent evidence, decisively. The old figure is what you get when nobody asks. The label was not a finding, it was the absence of a question, and the cost is that patients get a fibromyalgia diagnosis and decades of untreated nerve pain.

Decompression is reasonable at a symptomatic site (by analogy with ordinary entrapment) vs the HNPP-specific numbers

Carpal tunnel release improved 6 of 10 patients with this condition, against 4 of 5 in Charcot-Marie-Tooth. Cubital tunnel decompression improved 1 of 5 and worsened 2 of 5, and the authors state it "commonly worsened course". The only feature that predicted benefit from carpal tunnel surgery was whether symptoms were activity-provoked; neither the nerve tests nor the ultrasound predicted anything.

Which to follow: the recent evidence, with the size caveat said out loud. It is 15 operations in total. It is also the only comparative data that exists, and it points at harm at the most commonly affected site in the condition.

Honest Limitations

Every outcome cohort here is a neurology or surgical cohort. None is a physical therapy cohort.

All the outcome data comes from neuromuscular clinics, hand surgery services and genetics laboratories. The patient this page is written for, on their third unexplained trapped nerve with no diagnosis, sits in a first-contact clinic that has published nothing on this condition. Treat the recognition half of this page as the deliverable and the management half as untested.

The exercise evidence is one narrow population doing one movement.

Nineteen of 36 young male conscripts presented with brachial plexus lesions, and push-ups were the leading cause. That group is unusually homogeneous, unusually young, and doing an unusually high volume of one exercise under duress. It transfers as a warning about loaded shoulder work. It does not transfer as a risk estimate for a 55-year-old office worker.

The pain literature measures pain, not treated pain.

Four cohorts establish that half to three quarters of these patients hurt, that the pain is neuropathic in most and centrally sensitised in 84% of those with pain, and that it predicts quality of life better than anything else measured. Not one study reports the outcome of treating it. The most actionable finding in the condition sits on top of a complete outcome vacuum.

The Nuance

The reassuring version of this condition is true and incomplete. 26 of 30 patients recovered fully after an acute attack, and reviews describe the course as benign with spontaneous improvement after most episodes. Every word of that holds up.

And then: when 18 patients with this condition were matched by age and sex against 18 with Charcot-Marie-Tooth type 1A, a condition with far more measurable impairment, there was no difference in quality of life at all. The significant independent predictor of that gap was pain at the moment of testing. In a separate study, patients whose neurological impairment scores were close to normal reported average fatigue of 5.9 out of 10 across two weeks of moment-by-moment sampling. Restless legs affects nearly 39%, and tracks with disability, depression and fatigue.

Both pictures are accurate because they measure different things. "Recovery" in the classical literature means the return of muscle power after an episode. The quality-of-life work measures what the person lives with in between. The endpoint the reassuring framing measures is the one that recovers.

Which leaves the honest position on treatment. The palsies get better on their own most of the time, and the surgical evidence is 15 operations with a worsening signal at the site that gets hit most. The reason for caution about surgery here is mechanical rather than statistical: the nerve is abnormal along its entire length, so releasing one point does not hand back a normal nerve. But what is genuinely unresolved is not whether conservative care beats surgery. It is that nobody has measured conservative care at all. That is a gap in what researchers chose to study, not a low success rate, and the difference matters because the first one is fixable.

Worth noting who did the measuring that exists: the study that broke the "painless" label came out of a School of Physiotherapy, and the study that documented the fatigue came out of a Physical Therapy program. The profession did not fail to treat this condition. It is the profession that worked out what was actually wrong with these patients.

Evidence

Sources

Raasveld FV et al., 2026 — Pain Manag, PMID 42165586. Systematic review, 53 studies, N=655. Pain prevalence 34% (1990-2015) rising to 56% (≥2015); only 4 of 53 studies used validated instruments and those reported more pain.
Beales D et al., 2017 — J Neurol, PMID 29079893. N=43. 32 of 43 (74%) with persistent pain; 75% likely neuropathic, 84% likely central sensitisation.
Yilmaz U et al., 2015 — Muscle Nerve, PMID 25042093. N=32. 24 (75%) with pain; all 9 with purely musculoskeletal pain met ACR fibromyalgia criteria.
Lefour S et al., 2020 — J Neurol Sci, PMID 31862515. N=23. 52% with neuropathic pain; higher modified Rankin scores in those with pain (p<0.007).
Chompoopong P et al., 2022 — Muscle Nerve, PMID 35894586. Carpal tunnel release improved 6 of 10; cubital tunnel surgery improved 1 and worsened 2 of 5.
Andersson PB et al., 2000 — Neurology, PMID 10636123. N=80. Mean sensory conduction velocity 85.6% of lower limit of normal vs 114.3% in CIDP and 108.1% in diabetes, both p<0.0001.
Takahashi S et al., 2017 — J Clin Neuromuscul Dis, PMID 28221301. N=46. Ulnar at elbow 85.7%, median at wrist 84.4%, fibular at fibular head 36.7%.
Kim KE & Yeom J, 2016 — J R Army Med Corps, PMID 26462743. N=36 soldiers. 52.8% presented with brachial plexus lesions; push-ups the leading identified cause.
Kodaira M et al., 2017 — Brain Behav, PMID 28948078. N=1. Bilateral brachial plexopathy after barbell and plank training in an adolescent with no family history.
Cao W et al., 2023 — PMID 38432886. N=34 from 24 families plus review of 1,734 genotyped families. 26 of 30 (87%) fully recovered after an attack.
Bjelica B et al., 2021 — Acta Neurol Belg, PMID 32335868. 18 HNPP vs 18 matched CMT1A. No SF-36 difference despite less impairment; pain the independent predictor (β = -0.93, p<0.001).
Fritz NE et al., 2020 — Ann Clin Transl Neurol, PMID 32856791. N=9. Mean fatigue 5.9/10 despite CMTNS 5.7 ± 2.8.
Chen L et al., 2024 — J Neurol, PMID 39666198. 68 studies, N=124. Mean onset 26.5 y, mean diagnosis 32.7 y, maximum interval 40 y; 38.8% of episodes with no identifiable trigger.
Park JE et al., 2018 — Orphanet J Rare Dis, PMID 29544507. 11,885 newborns. Genetic prevalence 58.9 per 100,000, or 1 in 1,698.
Müller KI et al., 2021 — Brain Behav, PMID 33185984. Clinically ascertained prevalence 8.3 per 100,000 in Northern Norway.
Kramer M et al., 2016 — PMID 26761923. Asymptomatic gene carrier developed foot drop with prolonged recovery after total knee arthroplasty.
Gabreëls-Festen A & Wetering RV, 1999 — PMID 10586258. PMP22 proposed to prevent longitudinal sliding of myelin lamellae; HNPP nerve changes resemble genuine acquired entrapment.
Bjelica B et al., 2026 — J Peripher Nerv Syst, PMID 42052752. Restless legs in 38.9% of HNPP, associated with higher disability, depression and fatigue.

Nerves with thin insulation get switched off by ordinary pressure, and it is not painless.

Hereditary Neuropathy with Liability to Pressure Palsies HNPP