Excitability and cation loading

In adults with type 2 diabetes diagnosed within five years and no clinical DPN at entry, elevated inward cation current in peptidergic sensory neurons precedes substantial terminal loss. Limiting the specified currents from entry would avert at least 20% of five-year net physical terminal loss in a baseline-defined group accounting for at least 10% of reference-care incident DPN.

The target is the unmyelinated peptidergic neuron's skin arbor, connected peripheral axon and DRG cell body. The defined entry routes are plasma-membrane Nav1.7, Nav1.8, Nav1.9, voltage-gated calcium channels, TRPA1 and TRPV1. Their inward sodium and calcium fluxes are measured separately in mol per unit membrane area per day. Pump flux, ER calcium release and mitochondrial calcium uptake are not themselves the initiating exposure.

R1's boundary. During a 30-day observation ending at entry, at least 10% of mapped target neurons in a qualifying group must have either sodium or calcium influx through the listed channels above both the matched nondiabetic 95th percentile and twice its positive median on at least seven days. The person-weighted proportion is used, so a person with more recorded neurons does not dominate. The excess must precede that neuron's first ≥5% net physical terminal deficit maintained for 30 days. Ordinary turnover is not that deficit. Reference measurements match age, sex, skin site, temperature and external sensory stimulation.

These are explicit exposure and timing definitions, not measured thresholds for diabetic injury. They focus this assessment on an elevated-current state already detectable in early diabetes. A route arising only after established terminal degeneration is outside this prevention proposition. Neither spontaneous firing alone nor a larger fluorescence transient establishes R1.

Q — selective limitation. An ideal neuron-restricted controller reduces opening of the listed plasma-membrane channels whenever necessary to keep each rolling 24-hour inward sodium or calcium flux density at or below its matched nondiabetic 95th percentile. It only reduces channel opening; it does not force a normal voltage, firing pattern, calcium concentration or ATP level. Below the cap it has no direct action. The comparator receives contemporary diabetes care without Q.

Q begins at entry and continues for five years. It has no direct action on ion pumps, intracellular calcium stores, mitochondrial enzymes, glia, vessels or systemic metabolism. Glycemia, insulin exposure, circulating substrates and other initiating insults remain comparable by design. Changes in membrane voltage, ion gradients, sensory behavior, neuropeptide release, calcium handling, energy demand, inflammation and repair are allowed and count in the result. Q may impair useful activity or trophic signaling; those costs are included. This is a defined ideal perturbation, not a claim that an existing analgesic achieves it.

Groups and physical endpoint. The finite candidate groups are the whole eligible population, males, females, baseline BMI ≥30 kg/m², baseline HbA1c ≥7.5%, and a baseline high-current group meeting R1's neuronal exposure criterion within each person. No further intersections or future responder groups are permitted.

Map target fibers with stable subtype identity and anatomy, then measure their actual destruction and successful replacement. A change in CGRP, PGP9.5 staining, firing or pain is not physical loss. For each person, five-year net loss is destruction minus successful replacement, divided by baseline terminal amount. With equal weighting of humans:

B = (mean reference net loss − mean Q net loss) / mean reference net loss.

The reference mean must be positive. Incident DPN means new bilateral length-dependent nerve signs with objective small-fiber loss or abnormal conduction, excluding another cause. Coverage is C = P(group | incident DPN under reference care).

All requirements concern one of the finite groups with the same connected neurons and compatible sequence. More precisely, R1 means at least one group meets the exposure condition; each later requirement means at least one of the groups meeting all predecessors also meets the added condition. This prevents stitching exposure in one group to benefit in another. A complete NO must exclude all still-qualifying groups.

Source assessment

Best judgment: about 14%; uncertainty range (sensitivity): 0.5–62%. Excess channel current can damage sensory neurites, especially when energy supply is constrained. The difficult transfer is from that capability to a common, preventable cause of early human diabetic terminal loss. These are literature-informed subjective probabilities, not a treatment effect or an empirically calibrated forecast. Reviewed 14 September 2026. Probability method · Collection index.

Reasoning from the biology

Membrane current is a rate of ion movement. Firing frequency, channel expression, intracellular concentration and cumulative ion movement are different quantities. Persistent sodium current can add considerable sodium entry between action potentials; conversely, frequent brief spikes need not produce sustained calcium accumulation when extrusion, sequestration and energy supply keep pace.

Three sodium ions exchanged for two potassium ions cost one ATP through the sodium-potassium pump. Increased entry therefore adds a restoration burden. If sodium accumulates, the sodium-calcium exchanger can reverse and import calcium. In parallel, depolarization can open calcium channels and TRP channels can pass calcium directly. Calcium can stimulate mitochondrial metabolism at one exposure and disrupt it at another.

An illustrative 3Na:1Ca exchanger calculation at 37°C gives reversal potentials of −39 mV at 10 mM intracellular sodium and −95 mV at 20 mM, assuming extracellular sodium 140 mM, extracellular calcium 1.2 mM and intracellular free calcium 100 nM. At a membrane potential of −60 mV, the equilibrium boundary is about 12.9 mM intracellular sodium. Thus sodium accumulation could reverse exchanger direction without a change in channel abundance. These assumed concentrations are not measured diabetic terminal exposures, and equilibrium direction does not specify flux or toxicity. Calculation (background note outside this collection).

The causal alternatives are:

excess inward current → greater restoration demand and/or calcium entry → altered maintenance, destruction or replacement

and

metabolic or terminal injury → altered channels, firing and calcium responses.

They can coexist as feedback. Calcium accumulation, ATP insufficiency and downstream protease activation are plausible mediators, but none is imposed as an individually necessary requirement of the current-limitation claim. A protective Q effect could travel through more than one downstream route. This assessment does not identify an exclusive calcium-to-death pathway. Conversely, a calcium-handling intervention that protects while current entry remains unchanged does not by itself establish Q's effect.

Evidence that moves the judgment

Human diabetic C-fibers can fire spontaneously, but the evidence is small and late. Becker studied seven people, ages 54–75, with established mixed small- and large-fiber diabetic neuropathy: three with spontaneous cutaneous pain and four without. The percentages of spontaneously active recorded C-nociceptors were 72% and 15%. These are fiber percentages, not proportions of independent patients. Six patients used metformin and none used insulin.

This establishes an abnormal electrical phenotype in surviving fibers. It does not measure daily sodium/calcium flux, identify all recorded fibers as peptidergic, establish their pre-loss state, or show that suppressing their activity preserves skin terminals. Mouse TRPA1 and methylglyoxal experiments in the same report inform a possible upstream trigger, not an independent human replication. Becker et al., 2023.

Human channel genetics supports excitability more directly than structural prevention. Among 189 sequenced patients with definite DPN, Blesneac found rare Nav1.7 variants in 10 of 111 painful cases and none of 78 painless cases. Two selected variants, M1852T and T1596I, altered channel inactivation in expression experiments. The painful and painless groups differed in disease characteristics; within painful DPN, carriers had shorter diagnosed diabetes duration. This is a selected case comparison, not incident DPN risk or a channel-suppression experiment. Blesneac et al., 2018.

A 2025 UK Biobank preprint challenges treating every previously nominated gain-of-function variant as highly penetrant. It identified 1,381 heterozygous and one homozygous I228M carrier. The reported adjusted odds ratios were 0.97 (95% CI 0.87–1.08) for chronic pain and 1.04 (0.73–1.49) for neuropathic pain. Phenotypes were questionnaire/prescription based; very rare variants were inadequately represented, and skin structure was not measured. This weakens a blanket genetic extrapolation, without refuting rare severe variants or a diabetes-dependent interaction. Newton et al., preprint 2025.

Correction of an A1632G mutation in patient-derived sensory neurons also reversed abnormal electrical properties. The cellular assays and differentiation batches are not independent human donors, and the reported correction does not supply a five-year structural endpoint. Shim et al., 2025.

A rare mutant supplies a real calcium-dependent neurite injury experiment. Estacion expressed human G856D or wild-type Nav1.7 in adult DRG neurons. After 18 days in culture, 50 mM KCl plus 25 mM 2-deoxyglucose for four days produced more fragmentation in mutant cultures. In the rescue experiment, fragmented neurites decreased from 30% to 18% with 0.5 μM KB-R7943, and to 13% with calcium-free medium plus 100 μM EGTA: relative reductions of 40% and 57%.

The study used at least four wells per condition from triplicate experiments; wells, fields and neurites do not establish that many independent animals. The combined depolarization and glycolytic blockade is much more severe than measured early human diabetic exposure. KB-R7943 has other actions, although several characterized off-target effects require higher concentrations. Calcium removal is broader than selective reverse-exchanger inhibition. The experiment demonstrates a possible interaction and a physical neurite endpoint, not a common human exposure-response curve. Estacion et al., 2015.

A longer culture experiment separates axonal degeneration from soma death. Rolyan's G856D cultures showed day-30 neurite length 70.5% of wild-type, with degeneration 30.9% versus 11.6% and approximately 95% soma viability in both conditions. Calcium differences were prominent in thin neurites rather than cell bodies. Each preparation pooled two young adult rats, with at least three cultures; image counts are not independent rat counts.

The “ATP” interpretation used an indirect magnesium-sensitive signal. The reported 78% thin versus 22% thick distribution among degenerative observations does not establish a fourfold per-fiber risk without the denominators of all thin and thick fibers. This supports compartment-specific vulnerability, while leaving exposure magnitude and transfer to intact diabetic terminals uncertain. Rolyan et al., 2016.

Bioenergetic rescue is suggestive but has reporting problems. Lee used DRG neurons from 6–8-week-old mice, splitting each preparation between wild-type and I228M constructs. In seven-day cultures, reducing glucose from 25 to 2.7 mM worsened mutant neurite length; ATeam measurements supported altered ATP-related signals. The mitochondrial drug dexpramipexole, 2 μM for seven days, increased mutant neurite length by about 29%, based on three animals and 12 culture measurements per arm.

The published figure gives a mean difference 0.2875, SE 0.09296, interval −0.009469 to 0.4803, and p=0.0053. That interval crosses zero and is not centered on the stated difference. The paper mentions nested analyses, but the animal-linked values needed to repair this inconsistency are unavailable in the recovered report. The direction is useful; its displayed inferential precision is not reliable. Total neurite length after replating also combines growth and degeneration. Lee et al., 2020.

Endogenous I228M mice resist a simple current-to-degeneration story. In the 2021 study, young mice had hyperexcitable DRG neurons but plantar skin density 19.6 versus 17.0 fibers/mm, with four wild-type and four homozygous mice, p=0.2. This is not an equivalence result. Chen et al., 2021.

The 2023 follow-up found substantial age-dependent thermal/pruritic insensitivity and reduced sodium conductance in older homozygotes. Skin density remained 35.31, 30.24 and 32.89 fibers/mm in 5 wild-type, 8 heterozygous and 7 homozygous mice, respectively. Figure digitization gives homozygous minus wild-type −2.44 fibers/mm, approximate Welch interval −8.01 to 3.12; the heterozygous contrast is −5.14 (−10.91 to 0.63). These exploratory intervals do not exclude important density reductions, but there is no demonstrated collapse of terminal density matching the striking sensory deficit. Lower C-LTMR marker expression also does not establish loss of peptidergic C-fibers. Wimalasena et al., 2023.

Diabetic mitochondrial calcium intervention supplies the strongest structural bridge. George deleted the mitochondrial calcium uniporter, MCU, in Nav1.8-lineage neurons. After ten weeks of high-fat feeding, digitized skin counts were about 17.2 fibers/mm in heterozygous versus 37.0 in homozygous-deletion mice, five animals each. The difference was 19.75 fibers/mm, approximate interval 12.00–27.51, exact permutation p=0.00794. Accounting for the corresponding regular-diet means suggests about 96% closure of the diet-associated density deficit.

Both high-fat groups developed obesity and glucose intolerance, although small metabolic samples do not establish identical lifetime exposure. The deletion was developmental and broader than peptidergic fibers. Crucially, MCU changes intracellular handling rather than selectively suppressing initiating plasma-membrane current. The in-vivo calcium study found more responding neurons; overall transient area/amplitude comparisons were weaker, and a higher amplitude appeared after restricting to responses above a fluorescence cutoff. Resting fura ratios were not detectably different. This supports a calcium-sensitive diabetic structural route, but not inevitable sustained calcium overload caused by hyperactivity. George et al., 2022.

Analgesic response is not structural validation. The LENSS crossover trial exposed 24 patients to lacosamide and 23 to placebo, using 200 mg twice daily for eight-week phases. A one-point pain improvement occurred in 58.3% versus 21.7%; the reported sensitivity-analysis OR was 5.65 (1.83–17.41). Entry skin biopsy helped diagnose selected Nav1.7-associated small-fiber neuropathy; it was not evidence of regenerated terminals under treatment. de Greef et al., 2019, trial registry.

The reviewed primary studies do not supply a human early-diabetes intervention linking quantified current limitation to stable peptidergic terminal preservation. The missing measurement is not a negative trial. Further hits confined to pain, current recordings or channel expression would not resolve that gap.

Probabilities of the necessary claims

All rows condition on the preceding requirements and common evidence. The conjunction is the full proposition; the number of rows does not create an independence assumption.

See the claims and probabilities table above.

R1

Human spontaneous firing and diabetic animal sensitization favor exposure; daily flux, pre-loss timing and target-subtype prevalence are unmeasured.

R2

Mutant neurite rescue and diabetic MCU protection favor a structural contribution, but intact-mutant mice, severe culture exposures, compensation and loss of useful signaling limit transfer.

R3

Assumes a smaller physical benefit is real; the issue is its share of net loss after parallel injury and replacement, not whether currents can injure again.

R4

Acquired excitability changes need not require rare mutations, but a large structural response could remain confined to uncommon metabolic/channel combinations.

Central inputs [65,50,65,65]% give:

0.65 × 0.50 × 0.65 × 0.65 = 0.1373125, or 13.73125%.

A separate whole-claim assessment favors a real but far-from-established prevention route: physical mutant injury and diabetic MCU protection justify more than a negligible belief, while neither measures Q's effect in early human disease. Roughly one chance in seven fits that assessment. Differences of a few probability points are not supported by these data.

Uncertainty and consequential updates

The secondary electrical phenotype interpretation uses [30,15,35,30]%, giving 0.4725%. It treats most observable hyperactivity as a marker of existing injury, severe culture rescue as a narrow interaction, and physiological activity limitation as weak or costly in intact humans.

The common early current-dependent route interpretation uses [90,85,90,90]%, giving 61.965%. It transfers the channel–calcium–structure connection to a prevalent acquired diabetic state and assumes compensation seldom prevents a large net benefit. The absence of the decisive human flux/structure comparison permits this interpretation, but does not make it the preferred one. The full sensitivity width is about 61.5 percentage points.

Two other coherent interpretations matter. Rare but strong susceptibility, [70,80,85,30]%, gives 14.28%. Common exposure with a small structural share, [85,45,40,85]%, gives 13.01%. Similar headlines can therefore conceal very different biology. These scenarios have no empirical weights and do not define a probability distribution over probabilities.

A hypothetical upward update—representative pre-loss human flux measurements plus a selective current intervention showing durable physical preservation across a common baseline group—would support [85,80,80,80]%, or 43.52%. A hypothetical downward update—accurate early flux measurements mostly below the exposure boundary plus adequate current suppression with little physical benefit—would support [40,20,50,50]%, or 2%. An analgesic result alone would not supply either update.

The chosen exposure cutoff, seven-day frequency, 5% smaller-effect boundary, 20% materiality and 10% coverage are definition sensitivities. Relaxing them broadens the proposition; tightening them narrows it. No available dataset identifies a numerical probability curve over those choices. Sampling intervals in the mouse experiments do not become intervals on these elicited probabilities.

Ideal experiments that would resolve each claim

These are conceptual resolving experiments, requiring longitudinal measurement of identified living human axons and a selective controller beyond current capabilities. They do not count as evidence. Independent humans are the inference units. Each design evaluates the finite groups; complete exclusion means exclusion of every group still satisfying the predecessors. A precision interval crossing the stated boundary remains unresolved, regardless of its p-value.

R1 — excess current before substantial terminal loss; no predecessors. Sample early-T2D people representatively and matched nondiabetic reference people. During the 30 days ending at entry, directly measure inward sodium and calcium flux through each listed channel class in mapped peptidergic somata and connected axons, normalized to membrane area. Record external stimulation and temperature, and track physical arbors with stable subtype identity. Repeated measurements determine whether each neuron exceeds both reference thresholds on seven days before its sustained 5% deficit. Calculate the equally person-weighted fraction of qualifying neurons within every allowed baseline group.

YES requires at least one group with fraction ≥10%; NO requires all groups below 10% under sufficiently complete and precise observation. A fraction of 12% can satisfy R1 without establishing any injury from that current. Uncertain channel attribution, sparse sampling, lost subtype identity or an unobserved prior structural deficit leaves it unresolved. Existing human recordings supply abnormal firing in established DPN; they do not supply this flux-and-order measurement.

R2 — smaller physical benefit from Q; assumes R1. Within the independently defined groups meeting R1, randomize independent humans at entry to five years of Q or reference care. Verify directly that the rolling flux caps are achieved in the target membranes and that there are no direct systemic or intracellular-handling actions. Reproduce the current perturbation with an independent channel-gating implementation; restoring the original current pattern in a suitable mechanistic comparison distinguishes intended current effects from an implementation artifact. Keep other initiating exposures comparable, while allowing every downstream mediator and cost to respond.

Map the same physical terminal territories over five years, separating destruction from replacement. Compute B from person-level normalized losses. YES requires B≥0.05 in at least one R1 group; NO requires B<0.05 in all such groups, with positive reference loss and adequate precision. Reducing reference loss from 30% to 28% gives B=6.7% and resolves YES, despite failing R3. Pain relief without physical preservation does not satisfy R2. Incomplete engagement, nonpositive reference loss, altered subtype staining without anatomy, or intervals crossing 5% remain unresolved. Mutant cultures supply physical neurite rescue and MCU mice supply diabetic skin protection, but neither implements the specified human Q.

R3 — material effect; assumes R1–R2. Continue the same human structural comparison in groups already meeting the exposure and smaller-benefit conditions. Estimate five-year B precisely around 0.20, including all replacement, competing injury and adverse downstream effects. The added proposition is effect size, not a new mechanism assay.

YES requires B≥0.20 in a group meeting R1–R2; NO requires every such group's B<0.20. With 30% reference loss, 24% or less under Q is YES; 27% is a real 10% benefit and NO for R3. Uncertain denominators or intervals crossing 20% are unresolved. The large MCU mouse contrast supports possible materiality, while the intervention, developmental timing and human transfer remain missing.

R4 — case coverage; assumes R1–R3. In a population-representative early-T2D reference-care cohort, ascertain the allowed group memberships at entry and incident DPN over five years. Use standardized bilateral examination and objective nerve measurements, with independent adjudication of other causes and complete follow-up. Link the group definitions to those meeting the preceding causal requirements without selecting people by treatment response.

Calculate C as the proportion of all reference-care incident cases belonging to each qualifying group. YES requires at least one such group with C≥0.10; NO requires all below 0.10. A selected mutation clinic with excellent structural benefit cannot settle this population quantity. Missing incident cases, imprecise subgroup prevalence or incomplete matching to the causal groups leaves it unresolved. Current channel-variant and microneurographic samples do not supply the required denominator.

Relationship to neighboring hypotheses

Methylglyoxal can initiate TRPA1/Nav changes, energy insufficiency can both cause and follow ionic imbalance, and MCU-dependent calcium entry can affect mitochondrial dynamics. Those overlaps are compatible with a shared chain. They are not independent confirmations or effects to add. Methylglyoxal, energy insufficiency and mitochondrial dynamics define different interventions.

Q could protect without a specific ATP-deficit threshold, mitochondrial fragmentation or somal apoptosis; conversely, those mechanisms can act without elevated current through the finite channel set. The propositions therefore do not automatically imply one another. The most consequential unresolved link here remains selective current limitation to human physical terminal preservation.