Compression and mechanical injury

In adults with type 2 diabetes diagnosed within five years and no clinical DPN at entry, ordinary-life compression or stretch reaches specified lower-limb nerve sites before their connected terminals undergo sustained loss. Compared with matched people without diabetes at the same mechanical exposure, the diabetic population has greater load-induced structural loss. Reducing the specified loading from entry would preserve at least 20% of five-year net peptidergic C-fiber terminal loss in a baseline-defined group representing at least 10% of reference-care incident DPN.

This is a claim about a mechanically vulnerable diabetic population and a causal effect of loading. It does not identify which diabetic metabolic change causes the vulnerability, or claim that merely changing a diabetes diagnostic label would alter tolerance. Increased loading alone, with identical structural response at a common load, does not satisfy the full proposition.

Sites and exposure. The finite sites are the common fibular nerve at the fibular neck; superficial fibular nerve at its fascial exit; deep fibular nerve under the ankle extensor retinaculum and extensor hallucis brevis tendon; tibial nerve in the tarsal tunnel and its medial/lateral plantar branches beneath abductor hallucis; and sural nerve at the posterolateral ankle.

At a mapped site, define exposure as focal compressive stress on the nerve surface ≥20 mmHg above adjacent tissue hydrostatic pressure, or axial elongation ≥6% above its unloaded local length, for at least 30 cumulative hours in year one. At least 10% of people in a qualifying group must have that exposure on a pathway serving at least 10% of their mapped target units. Exposure must precede the connected arbor's first ≥5% physical net deficit lasting 30 days. These boundaries identify a measurable loading state; they are not established human injury thresholds.

Q — mechanical load limitation. For five years, an ideal controller at those sites caps focal compressive stress at 10 mmHg above adjacent hydrostatic pressure and axial elongation at 3%. Values below the caps are unchanged. It does not transect nerves, remove epineurium, directly alter neuronal metabolism, or denervate the skin. Whole-body activity, joint movement, footwear and systemic treatment are comparable at assignment. Local perfusion, axonal transport, inflammation, repair and subsequent behavior can respond, and their benefits and costs count. Q is not a named surgical procedure: surgery adds tissue injury and may not achieve these caps.

Territories and outcome. Follow physical terminal entries in fixed bilateral plantar hallux, first and fifth metatarsal-head, first dorsal web-space and lateral-foot territories, with each nerve-to-skin connection mapped. A label change is not disappearance. Establish peptidergic identity independently of current peptide abundance.

Net loss is cumulative terminal disappearance minus new established entries, including later disappearance of replacements, normalized to baseline amount. Average territories within each person and people equally. Let L₀ and L_Q be five-year peptidergic losses under contemporary diabetes care and Q. With L₀ >0, B = (L₀ − L_Q)/L₀. Let C = P(group | incident DPN under reference care). Incident DPN requires new bilateral length-dependent nerve signs with objective small-fiber loss or abnormal conduction and no better alternative cause. An isolated focal entrapment diagnosis does not enter that denominator automatically.

The finite candidate groups are the entire eligible population, males, females, BMI ≥30 kg/m², people reporting ≥4 hours/day of weight-bearing work at baseline, and people meeting the exposure criterion during a 30-day baseline observation. There are no additional intersections or groups selected by future response. Each successive claim asks whether a candidate satisfies every preceding requirement; a NO must exclude all candidates satisfying the previous prefix.

The tolerance comparison in R2. Use a nondiabetic, normoglycemic reference population with the same joint distribution of age, sex, BMI, height, baseline skin-fiber density and habitual activity as the candidate diabetic group. Replay a common distribution of pressure/elongation histories, sampled from that diabetic group's baseline ordinary activity, in both populations. Within each population, compare the complete replay with its Q-capped version for one year.

Let A_D and A_N be the resulting causal differences in pan-small-fiber net loss, expressed as fractions of each person's baseline terminal amount. R2 requires A_D >0 and A_D − A_N ≥0.01. This specifies greater structural sensitivity at equal mechanical exposure. The comparison is effect modification by the stated populations; it does not independently prove that diabetes itself, rather than another remaining population difference, created that sensitivity. Molecular attribution is outside the claim.

Source assessment

Best judgment: about 6.5%; uncertainty range (sensitivity): 0.06–52%. Compression can damage nerves, and human decompression can be followed by small-fiber regrowth. What remains uncertain is a diabetes-associated increase in structural susceptibility at the same load, followed by enough durable peptidergic preservation to affect incident DPN. These are literature-informed subjective probabilities, not measured surgical success rates. Reviewed 14 September 2026. Probability method · Collection index.

Biological logic

A swollen nerve in a stiff tunnel can experience greater pressure during ordinary movement. At a fixed pressure, altered microcirculation or transport reserve might also produce more damage. Those are different propositions: changed geometry can increase delivered exposure without decreasing tolerance. Force, pressure, deformation, perfusion and injury must not be substituted for one another.

Compression could reduce perfusion, obstruct transport or injure axons directly. These are alternative mediators; none is mandatory individually. Conversely, a diabetic nerve can resist an acute conduction block or show slower degeneration while recovering poorly later. More stiffness or worse baseline conduction does not determine the sign of the structural loading interaction.

Skin-fiber loss downstream of an entrapment is anatomically possible. A diffuse symmetric DPN endpoint nevertheless needs a distributed burden or a sufficiently common subgroup with bilateral affected territories. Height, body weight and a positive Tinel sign do not measure that contribution. Reduced foot pressure can also prevent an ulcer without preserving nerve terminals.

Evidence that moves the judgment

Human focal compression supplies actual structural evidence. In the Oxford carpal-tunnel cohort, 60 people underwent release, 13 opted out, and 20 healthy volunteers supplied a baseline comparison. Diabetes was excluded. Operated participants had mean epidermal density 4.20 before and 5.35 fibers/mm six months later, versus 8.03 in healthy controls. The paired test has 58 people, t=3.5, P=0.001; the figure's broader group caption gives 59.

Nonoperated patients declined 6.94→5.61 fibers/mm, a paired change −1.33, approximate 95% interval −2.42 to −0.24, n=13. This is meaningful evidence that terminal degeneration and regrowth can accompany a focal human compression disorder. It is not just a pain response.

The difference between the reported changes is approximately 2.48 fibers/mm, but treatment was self-selected. Operated patients were older, more symptomatic, and had lower baseline density. No diabetes group or peptidergic-specific count was included. The study therefore supports compression-related human structural plasticity without estimating diabetic tolerance, a randomized surgical effect or five-year prevention. Baskozos et al., 2020.

A simple regression-to-the-mean benchmark could generate that change difference from the 2.74-fiber baseline gap only if the common six-month conditional-mean slope were about 0.095. Substituting the study's technical repeat-count reliability of 0.913 gives only 0.24 fibers/mm, but technical repeatability is not six-month biological stability. Thus neither automatically dismissing the result as regression to the mean nor treating it as randomized rescue is justified. Calculations (background note outside this collection).

The human diabetes comparison does not demonstrate lower small-fiber tolerance. At carpal-tunnel surgery, 35 diabetic and 31 age/sex-matched nondiabetic patients had wrist skin biopsies. The study found no significant IENFD difference by diabetes, diabetes type or mild electrophysiological neuropathy. Hairy versus glabrous skin, sex and age affected the result. This is a cross-section at the wrist, not a matched compression dose or downstream foot trajectory. Without a precise effect interval, it cannot establish equal susceptibility. Thomsen et al., 2009.

Diabetic nerves can show an earlier perfusion response to pressure. In six diabetic and six control Wistar rats, sciatic-nerve transverse modulus was 210.7 versus 116.3 kPa, a 1.81-fold difference. Blood perfusion began to decrease at 24.1 versus 47.1 mmHg, about half the control pressure. This supports altered mechanics and a possible vascular route. It does not measure irreversible axonal or terminal injury at those pressures; the accessible primary abstract lacks the dispersion needed for a structural-effect interval. Chen et al., 2010.

Direct injury studies oppose a universal decrease in tolerance. An STZ-rat experiment applied 150 mmHg for 30 minutes to peroneal nerves. Across three combinations of diabetes duration and follow-up, teased-fiber abnormalities were more frequent in controls; differences were significant at three and 14 days, but not 24 days. The investigators explicitly considered increased tunnel constriction as an alternative to increased intrinsic acute susceptibility. The accessible abstract does not supply the independent animal n or complete effect estimates, so no precision is invented. The assay concerns mainly myelinated fibers and acute high pressure, not ordinary human C-terminal loading. Dyck et al., 1989.

A September 2026 study assigned 20 male ZDF diabetic and 20 lean rats to sciatic-sleeve compression or sham. Five per nominal group were examined after six weeks; the remainder underwent release/sham and were examined six weeks later. Three diabetic animals died or reached humane endpoints and were excluded, so 40 is not the final histological n.

At six weeks, nondiabetic compressed nerves showed more conspicuous loss of axonal area and myelin thickness, while diabetic structural changes emerged later. Diabetic function improved partly after release but did not normalize. This challenges an invariant “diabetes makes compression damage faster” rule. It does not prove protection: the same sleeve geometry does not guarantee equal nerve pressure, histology measures myelinated cross-sections, and there was no continued-compression arm after week six to isolate decompression from delayed injury. The small groups and absent direct pressure data leave R2 unresolved. Garcia Fleury et al., 2026.

Randomized DPN trials support selected symptom benefit more than terminal preservation. The DNND primary abstract reports 78 randomized participants from 2,987 screened, with unilateral decompression and contralateral sham in surgical participants plus an observation group. At 12 months, surgical participants improved by about 4.46 pain points relative to observation, but the decompressed and sham legs improved equally. At 56 months, the decompressed leg did have lower pain than the sham leg: difference 1.57 points, 95% interval 0.46–2.67.

The late difference is favorable evidence and should not be erased by the earlier sham result. However, it concerns pain in selected established disease, with fewer participants at long follow-up, not new physical C-terminal preservation. The reported abstract allocation/counts and earlier follow-up descriptions are not sufficient to reconstruct a complete participant-level attrition analysis. Neither an improvement in both legs nor a later local difference identifies the source of a five-year structural effect. Rozen et al., 2024.

In a separate 42-person randomized-side study, operated-leg pain at one year was 3.5 versus 5.3 in the untreated leg, from a baseline near 6.1. Its nerve-conduction analysis found no overall multivariate benefit. These are reports from the same trial, not independent positive and negative cohorts. A 22-person, 12-surgery/10-observation trial also favored surgery on pain, without a terminal endpoint. The nonsignificant conduction result does not bound small-fiber preservation tightly enough to serve as equivalence. Macaré van Maurik, pain, conduction, Best et al..

Recent positive studies do not close the structural comparison. A 2025 randomized study compared traditional release, modified release and modified release plus murine NGF in 12/16/14 patients. All arms received surgery. The combined arm received 18 µg perineural NGF and daily intramuscular NGF for 14 days; its superior functional outcomes cannot isolate mechanical release. Follow-up was 12 weeks and outcomes included pain, conduction, sensory tests and ultrasound size, not IENFD. Primary report.

A later uncontrolled series reported large improvements after quadruple release in 26 people, 45 limbs. Limbs are not 45 independent human donors, and no sham or structural terminal comparison was available. Primary report.

A 2026 long-term cohort supplies a dramatic favorable observation but severe selection limits. Of 281 surgical and 86 medical patients, only 76 and 31 were alive and available after more than 13 years: 27.0% and 36.0%. The remainder included 150/36 deaths and 55/19 lost participants. Reported pain improvement was 5.63 versus 1.03 points; analyzed surgical participants had 0 ulcers versus 10/31 medical participants.

The zero is an observed result in the analyzed cohort, not a known zero among all 281 originally treated people. Even ordinary binomial sampling would allow about 4.8% risk above zero for 0/76; selection makes that interval insufficient for the original population. Treatment was not randomized, and imputing no pain improvement to deceased people does not resolve the different question of ulcer occurrence before death. Ulcer reduction can also result from care, loading changes or perfusion without nerve regrowth. Liao et al., 2026.

A registered German study and the DECO program address symptoms, function and foot complications. The German protocol's biopsies are ligament/fascia/fat for remodeling, not a completed skin-terminal rescue result. An unavailable result is not a negative structural trial. German primary protocol; DECO primary protocol.

Probabilities of the necessary claims

Each probability after R1 conditions on all earlier requirements being true. Low/best/high entries are the rows of complete coherent scenarios below. Their ranges are sensitivity judgments, not independently sampled distributions.

See the claims and probabilities table above.

R1 is plausible because nerves traverse constrained moving tissues, but pressure, duty cycle and pre-loss timing have not been measured adequately in this population. Swelling or ultrasound area alone cannot resolve it. R2 is the bottleneck: perfusion vulnerability supports it, while direct acute and chronic compression studies show inconsistent structural timing and direction.

Given actual excess load-induced small-fiber injury, peptidergic retention becomes more plausible, supporting R3. Human focal regrowth prevents a near-zero structural judgment. R4 remains uncertain because relief of a local stress can improve symptoms without preserving a material fraction of long-term terminals. R5 is not the prevalence of entrapment among specialist patients: it is coverage of new bilateral DPN in early diabetes.

The product is 0.70 × 0.40 × 0.65 × 0.65 × 0.55 = 0.065065, about 6.5%. Exposure plus excess susceptibility jointly receive 28%; the remaining target/magnitude/coverage requirements jointly receive 23.24% conditional on that. The row count supplies no independent evidence or automatic penalty.

Uncertainty and numerical updates

The lower scenario, [35,10,30,30,20]%, gives 0.063%. It interprets most mechanical findings as focal disease or increased delivered load, with no consistent excess structural response and little contribution to ordinary incident DPN.

The upper scenario, [95,80,90,90,85]%, gives 52.326%. It assumes repeated human exposure reveals a vulnerability missed by acute myelinated-fiber assays, that delayed degeneration matters, and that bilateral distributed injury is common enough for substantial retained benefit.

The sensitivity range is about 52.3 percentage points wide. Its endpoints are not confidence limits or empirical quantiles. No defensible scenario weights, variance or calibrated intervention-success base rate are available. The central estimate gives the contradictory direct-compression evidence substantial weight while retaining the human structural positive. Small numerical differences within this broad uncertainty are not an intervention ranking.

A coherent effective but rare entrapment interpretation, [85,70,85,85,15]%, yields 6.4483%. A greater load without lower tolerance interpretation, [90,15,80,80,70]%, yields 6.048% for this complete proposition. The latter could support a broader load-reduction claim while failing the required excess-susceptibility comparison.

Hypothetical matched-load human structural evidence plus durable peptidergic benefit could move the rows jointly to [90,80,85,80,75]%, giving 36.72%. Precise equal tolerance and absent retained benefit could give [65,15,35,40,35]%, or 0.47775%.

Changing only R2 from 40% to 80% gives 13.013%, and reducing it to 15% gives 2.43994%. Certainty that R2 is true, holding every other judgment fixed, permits at most 16.26625%. That is an arithmetic ceiling for one resolved link, not the expected value of another experiment.

Ideal experiments that would resolve each claim

These ideal comparisons specify truth with unlimited measurement and selective load-control capability. They are not proposed clinical procedures. Use the population, finite groups, anatomical sites, reference matching and physical endpoints above. Track connected neurons and terminals without injuring them. Count independent people, observe all follow-up and allow downstream costs and compensation. A later conditional is evaluated only among groups meeting its entire earlier prefix.

Complete NO requires exact population quantities excluding every remaining candidate. Inadequate pressure control, unknown counterfactual load, incomplete nerve mapping, marker-only data or imprecise results are unresolved, not NO.

R1 — actual preceding exposure. Measure local pressure above adjacent hydrostatic pressure, unloaded length, strain and cumulative duration during ordinary life, while mapping each site to stable target terminals. YES: at least one candidate meets the ≥10% people/≥10% units exposure criterion for ≥30 hours in year one before the connected sustained deficit. NO: no candidate does. Ultrasound enlargement and occupational questionnaires approximate anatomy/activity but cannot resolve this dose and order.

R2 — excess structural sensitivity at a common dose. In R1-qualified diabetic groups and their matched nondiabetic references, randomize people to replayed common loading histories or the corresponding Q-capped histories. Follow pan-small-fiber physical net loss for one year. YES: some group has A_D >0 and A_D − A_N ≥0.01. NO: every R1-qualified group fails at least one condition. Matching force histories removes “the diabetic nerve was simply compressed more” as an explanation of this contrast; the result remains effect modification between the specified populations, not a unique metabolic mediator. Rat pressure/perfusion measurements approach dose; existing structural studies do not supply the human endpoint or matching.

R3 — retained peptidergic contribution. In groups satisfying R1–R2, compare five years of Q with reference care under ordinary activity. Determine stable-identity peptidergic disappearance, replacement and subsequent loss in all mapped territories. YES: some group has L₀ >0 and B ≥0.05. NO: none meets both conditions. Human carpal-tunnel regrowth supports possibility but excludes diabetes; pain and conduction trials do not resolve this structural subtype comparison.

R4 — material magnitude. Use the same intervention and complete five-year counts. YES: a group meeting R1–R3 has B ≥0.20. NO: every such group has B <0.20. If reference care loses 30% of baseline terminals, Q must preserve at least six baseline percentage points. Added surgery, NGF or ulcer protection cannot be substituted for this Q-only structural effect.

R5 — incident-DPN coverage. In a complete reference-care inception population, ascertain new bilateral DPN and the baseline membership of every case. YES: at least one R1–R4 group contains ≥10% of these cases. NO: none does, or there are no reference cases. Exclude isolated focal diagnoses that do not meet the DPN definition. Referral proportions, ulcer-free surgical survivors and the prevalence of positive Tinel signs do not supply this denominator.

Overlap and boundary

Compression may act through ischemia, axonal transport or matrix remodeling. Those routes are alternatives here, not extra mandatory links. Matrix glycation could increase external load without producing R2's excess sensitivity at a common load.

A positive focal surgical outcome does not establish prevention of incident DPN. Conversely, a null pain or conduction trial cannot exclude meaningful terminal preservation without the specified physical measurement. This assessment prices the defined combination of exposure, comparative susceptibility and durable structural contribution.