Oxidative and nitrosative injury

In adults within five years of T2D diagnosis, without clinical DPN at entry, excessive local exposure to a specified reactive-species panel precedes sustained terminal loss and produces measurable oxidative/nitrative chemistry. Capping that exposure would preserve at least 20% of five-year net physical peptidergic C-fiber terminal loss, in a baseline group containing at least 10% of reference-care incident DPN cases.

The panel is superoxide, hydrogen peroxide, peroxynitrite/peroxynitrous acid, hydroxyl radical, nitrogen dioxide radical and hypochlorous acid/hypochlorite. Acid/base pairs count as one member. It includes mitochondrial, NOX, inflammatory and other local sources. No single source, glucose-overload mechanism, NOS isoform or execution pathway is mandatory. Nitric oxide itself is not directly suppressed.

Q — a local reactive-species cap. Apply selective formation/removal control to these six species in mapped sensory neurons, their axons and terminals, connected Schwann and satellite glial cells, and the keratinocytes, fibroblasts and microvascular cells serving the mapped distal skin territories. Include intracellular organelles and immediate extracellular interfaces. Q begins at entry and operates for five years.

The cap is the compartment-specific 95th percentile of each species' 24-hour integrated reactive activity in nondiabetic people matched for sex, age within five years, BMI within 3 kg/m², smoking and renal-function category. Activity means the local rate available for chemical reaction, integrated over time and normalized to compartment volume, with species-specific reaction calibration. It is not generic fluorescent intensity, enzyme expression or a plasma antioxidant score.

For each participant, use one matched reference person's complete spatial and temporal cap profile, retaining cross-species relationships. A rolling controller reduces only exposure exceeding that profile, leaving lower exposure unconstrained. It acts by defined chemical identity, never by whether a reaction later proves harmful. Reaction products, altered physiological signaling, compensatory production and all downstream costs count. Initial glucose management, circulating nutrients, oxygen supply and other treatment are comparable; their downstream consequences can change.

This idealization requires capabilities beyond current antioxidant drugs, especially for short-lived radicals. It specifies a chemical exposure intervention rather than a switch that recognizes “bad oxidation.” Loss of useful oxidant-dependent repair can make its net effect zero or adverse.

Early exposure and chemistry. R1 requires at least one panel member's daily integrated activity to exceed both its matched 95th percentile and 1.5 times its matched median for ≥30 cumulative days in year one. This must occur in connected units serving ≥10% of mapped terminals in ≥10% of a candidate group, before a ≥5% baseline-normalized net physical terminal deficit lasting 30 days. Ordinary terminal turnover does not itself end the pre-loss window.

R2 requires the same exposed units to develop at least one of six measurable product classes: protein 3-nitrotyrosine, protein carbonyls, cysteine sulfinic/sulfonic modifications, phospholipid hydroperoxides, 4-hydroxynonenal–protein adducts, or DNA 8-oxoguanine. Occupancy is normalized to the corresponding parent molecular pool. It must exceed both 1.5 times the matched median and the matched 95th percentile for ≥30 cumulative days before that sustained terminal deficit, and Q must remove ≥50% of the excess.

This is a requirement for local chemical modification, not an assumption that the modification is damaging. A carbonyl assay alone cannot identify the initiating species. Reversible signaling, altered perfusion, impaired transport, energy failure, lipid injury and execution pathways are possible downstream mediators. The proposition does not require every listed adduct to be pathogenic, or one particular modification to carry the entire structural effect.

Outcome and groups. Track stable-identity peptidergic terminals in fixed bilateral distal-leg and dorsal-foot territories. Count physical disappearance minus newly established terminals, including later loss of replacements. Peptide downregulation and delayed clearance of detached debris do not count as retained connected terminals. Normalize to baseline terminal amount, average territories within people and then people equally. With reference-care and Q losses L₀ and L_Q, require L₀ >0 and define B=(L₀−L_Q)/L₀.

Incident DPN means new bilateral length-dependent neuropathy supported by objective small-fiber loss or abnormal nerve conduction, not pain alone. C is the fraction of reference-care incident cases belonging to the qualifying baseline group. It is not the percentage of terminals responding.

The six candidate groups are all eligible people, baseline HbA1c ≥8%, BMI ≥30, fasting triglycerides ≥150 mg/dL, current smokers, and age ≥60 years. There are no additional intersections or future-response groups. Each later claim requires at least one group satisfying every preceding requirement; a complete NO excludes all remaining candidates. These boundaries define this assessment, not validated clinical cutoffs.

Source assessment

Best judgment: about 27%; uncertainty range (sensitivity): 0.98–69%. Local oxidant injury has credible experimental structural support. Human evidence does not establish the early, sustained exposure or five-year preservation fraction of a defined intervention. These are subjective probabilities of the complete proposition, not treatment response rates. Reviewed 14 September 2026. Probability method · Collection index.

Biological reasoning

Reactive species have different lifetimes, destinations and chemistry. A plasma assay cannot specify mitochondrial-matrix or axonal-membrane exposure. Superoxide dismutation produces hydrogen peroxide; its reduction therefore does not mean all oxidants fell. Nitric oxide can support blood flow while also reacting with superoxide to generate peroxynitrite.

High steady-state exposure does not require production to exceed removal indefinitely. Production and clearance can balance at an elevated concentration. Likewise, a persistent adduct pool reflects formation, repair and turnover; a smaller pool after treatment can reflect altered cell composition or replacement rather than removal of a causal lesion.

The structural question includes destruction and failed renewal. Reducing fragmentation for four hours does not establish preservation over five years, and suppressing a signal needed for repair could offset less destruction. Q is therefore priced by retained connected terminals after all downstream responses.

Broad source alternatives make this proposition more plausible than a claim that glucose-overloaded mitochondria are the unique source. They do not make it certain: the relevant exposure could begin after injury, remain chemically well tolerated, or make too small a net contribution.

Primary evidence

Human timing evidence is suggestive but indirect. A six-year study enrolled 89 diabetic patients, 59% already with polyneuropathy; 72 had repeat nerve testing and 17 died. Baseline plasma superoxide generation predicted median sensory-conduction decline, reported β=−0.997, P=0.036. The mortality association was very imprecise, HR 23.2, interval 1.05–513.

This is longitudinal functional evidence, not a cohort demonstrating incident peptidergic loss from measured local exposure. Multiple markers and outcomes, existing disease and survivor follow-up limit its contribution. The accessible primary abstract does not permit reconstruction of a unique adjusted structural effect. Ziegler et al., 2015.

Antioxidant abundance is not a directional injury measurement. Skin from 69 recent-onset T2D participants and 51 controls showed approximately 60% greater subepidermal SOD2-positive area, 0.24% versus 0.15%, P=0.0005. Endothelial stained area did not differ. This supports an early tissue response, without establishing excess reactive dose, failed defense or a prospective structural consequence. Ziegler et al., 2015.

In the German Diabetes Study, 107 T1D and 215 T2D participants plus 37 controls had altered circulating redox markers. Lower SOD3 was associated with prevalent DSPN; TBARS and glutathione did not provide the same disease association. These are blood measurements in a cross-section. Strom et al., 2017.

Conversely, KORA followed 524 initially DSPN-free people, with 132 incident cases over 6.5 years. Higher SOD3 predicted incidence after age/sex adjustment, OR 2.14 per doubling, interval 1.02–4.48, partly explained by cardiometabolic factors. The wider baseline sample contained 20.5% T2D and 40.3% prediabetes. The result does not imply that SOD3 is neurotoxic: compensation, tissue binding and systemic disease can reverse a biomarker's association. Neither direction resolves Q. Herder et al., 2018.

MitoQ provides a structural positive with an informative comparison. Rats with high-fat feeding plus low-dose streptozotocin received 0.93 g MitoQ/kg diet for 12 weeks, beginning after four weeks of hyperglycemia. IENF density was 14.0±0.2 untreated versus 17.8±0.4 treated profiles/mm, SEM, nine rats per arm. The unadjusted contrast is +3.8, 95% interval 2.82–4.78.

In obese rats, 17.5±0.4 versus 17.8±0.2, n=8/9, gives +0.3, interval −0.69–1.29. Treatment recovered approximately 49% versus 7% of the cross-sectional deficit against control 21.8; these are not prospective loss-prevention fractions. Glycemia was not lowered in the diabetic arm. The experiment supports a modifiable structural route, without demonstrating the panel's local engagement, peptidergic specificity or an exclusively mitochondrial source. Fink et al., 2020; calculations (background note outside this collection).

Peroxynitrite-directed treatment separates structure from sensation. After 28 weeks of streptozotocin diabetes, mice received four weeks of FeTMPS 10 mg/kg/day or epicatechin gallate 20 mg/kg/day. The diabetic groups contained 10 untreated, 16 FeTMPS and 14 epicatechin animals; the skin-figure caption assigns these Table 2 denominators.

Both treatments reduced nitrated proteins and improved sensory function without lowering hyperglycemia. Only FeTMPS reduced 4-HNE adducts and significantly increased skin nerve density. The available numerical text does not support a reconstructed density interval.

This strengthens a chemically modifiable structural mechanism in established animal injury. It also shows why reduced nitration or improved thermal behavior cannot substitute for retained terminals. Different administration routes, incomplete chemical specificity and possible vascular effects prevent attributing the structural difference uniquely to one adduct class. Stavniichuk et al., 2014.

Genetic and pharmacological negatives constrain the inference. SOD2 haploinsufficiency worsened nerve abnormalities on a diabetic db/db background, including reduced skin innervation, but the structural comparison used three animals per group and percent epidermal area rather than standard fiber profiles/mm. The corresponding streptozotocin model did not develop neuropathy after 24 weeks in either genotype. That null lacked the disease contrast needed to estimate protection or exacerbation. Global developmental SOD2 reduction also permits vascular and systemic effects. Vincent et al., 2007.

Niclosamide ethanolamine, 1,500 ppm in chow from age six to 24 weeks, did not prevent conduction deficits or reduced IENFD in male BKS-db/db mice. Forty-eight mice entered the four-group design. Cultured DRG coupling efficiency fell from about 77.4% to 69% in treated diabetic animals, indicating some pharmacological engagement. However, that assay used two donor mice per group, plated into 20 wells; the many wells do not establish a large biological n.

Uncoupling was therefore achieved to some degree without observed structural benefit. The experiment did not show that Q's six local exposures normalized, and its skin null lacks a reconstructed equivalence interval here. It weakens a simple uncoupling remedy, without excluding nonmitochondrial chemical injury. Hinder et al., 2019.

Sensory-axon redox effects can run in opposite directions. In a 2025 zebrafish amputation study, neuronal SOD2 overexpression reduced fragmentation, n=11/8, while GPx1a expression, n=7/4, did not prevent it and accelerated completion/clearance. These are short-term injured larval sensory axons, not diabetic human terminals.

The reported NADPH regrowth means, 69.48±4.31 versus 56.56±5.63 µm, SEM, n=6/arm, imply +12.92 µm, Welch interval −3.03–28.87, and the stated two-sided Student test gives P≈0.098, inconsistent with the text's significant enhancement. This arithmetic does not erase the separate fragmentation observations. Predicted NMNAT modification sites are computational hypotheses, not demonstrated biochemical mediators. The study supports species-specific effects and illustrates why preserved fluorescence must be separated from delayed debris removal. Hsieh et al., 2025.

Human antioxidant trials constrain treatment claims. NATHAN 1 randomized 460 people with mild-to-moderate established DSPN to 600 mg/day alpha-lipoic acid or placebo for four years. Most had T2D, with median diabetes duration around 13 years. The primary NIS-LL+7 advantage was 0.78 points, 95% interval −0.16–1.73, P=0.105. Despite the paper's sign label, its group means make this a placebo-minus-treatment benefit. The interval excludes the intended two-point advantage on that composite.

NIS improved more, with an unadjusted mean advantage 1.29 points, interval 0.04–2.54. This secondary result is not a physical small-fiber endpoint. Nerve-conduction and quantitative-sensory measures did not differ significantly. The primary analysis allowed last available observations; its analysis n is not complete four-year retention. Neither nerve chemical engagement nor terminal density was measured. Slow placebo progression limits the prevention question but does not make the observed endpoint bound disappear. NATHAN 1.

BOND randomized 57 people to benfotiamine or placebo for 12 months. The reconstructed skin contrast was −0.49 fibers/mm, interval −1.87–0.90. Blood thiamine engagement did not demonstrate local reactive-species normalization. Its interval includes a benefit equal to 20% of the observed placebo skin decline, about 0.16 fibers/mm. Thus it neither confirms structural efficacy nor excludes Q's defined effect. BOND, 2026; paired-endpoint calculation (background note outside this collection).

Conditional probabilities

Each row prices an additional prefix requirement given all preceding rows. The same surviving group and connected temporal route must qualify throughout. Lower/best/upper values form coordinated interpretations, not independent marginal ranges or empirical confidence intervals.

See the claims and probabilities table above.

R1

Multiple sources and early human redox responses favor occurrence; direct local trajectories remain absent.

R2

Animal chemistry is persuasive, but bulk markers, repair and timing can mislead.

R3

Structural intervention positives survive; species transfer, useful redox signaling and treatment specificity limit confidence.

R4

A convergence route could be material, but partial recovery and mixed human results leave magnitude uncertain.

R5

Common metabolic exposures favor coverage; no observed human responder distribution establishes it.

R2 can be true with harmless chemical turnover and R3 false. B=10% passes R3 but fails R4. A substantial benefit in a group containing 3% of cases passes R4 but fails R5. The conjunction gives the complete bounded proposition; it does not identify which individual source or adduct dominates.

Ideal experiments that would resolve each claim

Assume exact longitudinal species/activity measurements, molecular occupancy mapping, selective Q, and nondestructive tracking of connected human terminals. Randomize independent people within the finite baseline groups and retain their full structural and vital-status trajectories. Samples, organelles and serial images do not increase donor n. A confidence interval crossing a boundary, inadequate engagement or incomplete follow-up is unresolved, not an ideal NO.

R1 — local exposure and order. Under reference care, measure all six species in every mapped compartment from entry, alongside the matched reference profiles and physical terminal trajectories. YES: at least one candidate meets every stated dose, 30-day duration, 10%-of-people/units and pre-loss boundary. NO: complete coverage excludes that conjunction in every candidate. Raised blood oxidation or measurements first obtained after sustained loss leave this claim unresolved. Animal reactive-chemistry measurements and prospective blood markers supply parts of exposure, not this local human time course.

R2 — chemical modification responsive to Q. Conditional on R1, compare randomized Q with reference care and measure all six product classes in the same exposed units. Separate newly formed modification from changing parent-pool abundance and turnover. YES: at least one R1-qualified group has the defined pre-loss excess, duration and ≥50% excess reduction under Q. NO: none does under complete measurement and verified Q. Lower antibody signal caused only by disappearance of stained cells fails the normalized occupancy criterion. Animal oxidant-product responses support chemical engagement; tissue abundance and antioxidant-treatment effects do not isolate the full normalized human occupancy contrast.

R3 — retained structural causation. Conditional on R1–R2, compare five-year L₀ and L_Q with stable subtype identity and complete connected-terminal accounting. Permit changes in downstream perfusion, repair and metabolism, including losses of useful signaling. YES: an eligible group has L₀>0 and B≥5%. NO: every remaining group has nonpositive reference loss or B<5%. Better sensation, reduced fragment fluorescence or more CGRP expression without physical retention cannot establish YES. MitoQ supplies a surviving rat skin contrast; NATHAN 1 constrains clinical translation but neither is the specified local human Q.

R4 — material size. Use the same randomized trajectories without changing groups or Q. YES: at least one R1–R3-qualified group has B≥20%. NO: every such group has B<20%. This separates a small causal route from the material route being priced; a 49% animal endpoint-gap recovery cannot replace the human denominator. The rat structural response and the clinical composite endpoint cannot determine this five-year human terminal-loss fraction.

R5 — case coverage. Follow the full eligible reference-care population for five years and classify incident DPN and baseline group membership independently of Q response. YES: at least one group satisfying R1–R4 accounts for ≥10% of cases. NO: all such groups fall below 10%. Overlapping groups are evaluated separately, never added to manufacture coverage. KORA's circulating marker association does not classify a group with verified local Q benefit or measure its causal case share.

Calibration and sensitivity

The complete best product is 26.775%, reasonably expressed as about one chance in four. There is no empirically calibrated reference class turning these heterogeneous experiments into a human prevention probability. The numerical detail shows arithmetic only.

The lower scenario [50,35,40,40,35]% → 0.98% treats much local chemistry as a downstream or compensated response and expects useful-signaling costs to offset preservation. The upper [95,95,95,90,90]% → 69.45% treats early local injury as a common convergence mechanism and current drug failures as weak tests of its full correction. The best lies between these interpretations because several physical animal positives survive examination, while direct early human intervention evidence remains missing.

An upward evidence scenario would be reproducible, early human local exposure and modification, followed by selective Q preserving connected terminals in common baseline groups. Coordinated [95,90,90,85,85]% would give 55.60%. Conversely, fully engaged local correction with little retained structural benefit would support [65,45,30,40,40]% → 1.40%. These are illustrative joint reassessments, not mechanical likelihood ratios. Exact ideal resolution of a required claim as NO sets the full conjunction to zero.

Merging R4 and R5 gives a best conditional 52.5%, preserving the same 26.775% joint. Adding alternative NOX and mitochondrial sources as separate mandatory discounts would price a different proposition. Lowering B from 20% to 10%, or C from 10% to 5%, cannot lower the probability, but current data do not identify a numerical change. Altering the reactive panel or cap could change benefit in either direction because useful signaling is also affected; a broader cap is not automatically better.

The strongest experimental evidence supports modifiable oxidative/nitrosative contributions to established animal nerve injury, a broader and better-supported statement than the complete early-human prevention claim. Source-specific mitochondrial overload remains distinct; lipid oxidation, PARP activation and SARM1 can lie on shared paths. These assessments are not independent votes or additive attributable fractions. Glucose overload · PARP · Ferroptosis · SARM1.