Protective lipid deficiency

In adults with T2D diagnosed within five years and no DPN at entry, the specified local lipid deficit develops before sustained terminal loss. Correcting it toward the corresponding nondiabetic state would prevent at least 20% of five-year net physical peptidergic C-terminal loss in a baseline group containing at least 10% of reference-care incident DPN cases.

Candidate groups are all eligible people, BMI ≥30 kg/m², fasting triglycerides ≥150 mg/dL, HbA1c ≥8%, blood Omega-3 Index <4%, and reported oily-fish intake less than once weekly. Use each group individually, without further intersections or later responder selection. Omega-3 Index is the proportion of red-cell fatty acids represented by EPA and DHA; it is a baseline group marker, not a validated nerve-deficiency boundary.

The finite chemical scope has two branches.

• Membrane composition: cholesterol, phosphatidylcholine, phosphatidylethanolamine including plasmalogens, sphingomyelin, galactosylceramide, sulfatide and cardiolipin. Resolve native molecular species, double-bond positions and stereochemistry. Restrict fatty-acyl/alkenyl chains to C14–C24 with zero to six double bonds; a species must normally contribute ≥1 mol% of its class in the relevant membrane to qualify.
• Locally available lipid substrates or signals: gamma-linolenate, dihomo-gamma-linolenate, EPA, DHA, resolvins D1/D2/E1, protectin D1, maresin 1 and prostaglandin E1, in their native active configurations. This list does not encompass every molecule described as a protective lipid.

The branches are alternatives, and neither dietary deficiency nor a specific biosynthetic-enzyme defect is required. This scope includes inappropriate incorporation even when total fat is abundant.

Local deficit. In membranes, measure each species' molar fraction of all membrane lipids, distinguishing plasma membrane, mitochondria and other organelles. Measure membrane area and class totals too, so fewer surviving membranes cannot masquerade as deficient composition. For free substrates/signals, measure the locally accessible unesterified concentration over its actual time course, not total plasma lipid or a transient assay peak.

In at least one specified compartment, a qualifying species must be ≥20% below its matched nondiabetic median and below the fifth percentile for at least 30 days. The abnormality must affect units serving ≥10% of mapped terminals in ≥10% of a candidate group during follow-up year one, before an additional baseline-normalized net terminal deficit ≥5% persists for 30 days. Ordinary replacement is not that sustained deficit. Match the normal reference by age, sex, anatomical site, cell/organellar identity, temperature and activity. A species normally absent from that compartment cannot be deficient there.

Compartments are the peptidergic axon/terminal and its sensory soma, directly supporting nonmyelinating Schwann and satellite glia, and the local capillary/skin cells serving the mapped terminal territory. A whole-nerve myelin deficit does not automatically identify a deficit in these units.

Q — physiological correction. For membrane classes containing a qualifying deficit, restore their species' fractions of total membrane lipid to a joint profile drawn from the matched normal reference population. Restore all affected classes jointly; do not assemble incompatible individual-species medians or select the normal profile producing the largest benefit. Rescale the remaining lipids proportionally to fill the remaining fraction, preserving their relative composition. Preserve membrane area and protein amount, account for displaced lipids and the required material/energy, and let molecular packing respond. For a qualifying free substrate or signal, restore its local accessible concentration to a matched normal trajectory, including ordinary fluctuations and clearance. Membranes and soluble pools without qualifying deficits receive no correction.

Apply these defined corrections from entry through year five when a qualifying local deficit is present. The ideal comparison includes the required lipid transfer and removal; it does not assume that swallowing oil performs them. Correct localization and chemical composition must be verified. If verified exchange destabilizes a membrane, that injury counts against Q's benefit; a delivery failure is a different result.

A membrane is a composition: adding a depleted constituent displaces other constituents. The membrane branch prices that specified physiological balance change. It cannot separately attribute benefit to adding one lipid versus removing its displaced partners, and it is not evidence of insufficient total dietary lipid.

Keep systemic glucose, insulin, circulating lipid delivery, oxygen supply and other incoming exposures at their reference-care trajectories. Local receptor signaling, metabolism, membrane mechanics, inflammation and repair can respond. Q does not directly add axons, provide a supraphysiological receptor agonist or suppress an injury enzyme. Increased oxidation of newly available polyunsaturated lipids, loss of useful remodeling and other consequences can offset benefit.

Outcome and coverage. Track stable-identity peptidergic terminals in fixed bilateral distal-leg and dorsal-foot territories. Net physical loss is disappearance minus newly established connected replacements, including subsequent loss of replacements. Changes in peptide staining, electrical silence and detached debris do not count. Normalize to entry terminal amount, average territories within people and people equally. For positive reference loss L₀ and corrected loss L_Q, B=(L₀−L_Q)/L₀. Complete trajectories include pre-death losses; missing observations cannot count as preservation.

C is the share of reference-care incident DPN cases in the qualifying baseline group. Require new bilateral length-dependent neuropathy with objective small-fiber loss or abnormal nerve conduction. Pain alone is insufficient. The same group must satisfy all requirements. Established-disease intervention studies inform repair and mechanism, not direct prevention in this population.

Source assessment

Best judgment: about 15%; uncertainty range (sensitivity): 0.5–55%. Correcting a local lipid shortage or a defined membrane-composition deficit could materially preserve human sensory terminals. Randomized corneal evidence and diabetic-animal skin counts support lipid-responsive structure. They establish physiological deficiency less clearly than treatment responsiveness. These are subjective probabilities of the complete proposition, not supplement efficacy or a patient's disease risk. Reviewed 14 September 2026. Probability method · Collection index.

Why the mechanism is plausible, and what competes with it

Lipids provide more than fuel. Their molecular shapes and locations affect membrane permeability, protein organization and mitochondrial machinery. Other lipids become local signals that change inflammation, vascular behavior or growth. A neuron may therefore have plentiful calories while lacking an appropriate membrane balance or signal.

However, a lower lipid measurement can reflect less membrane, altered cell proportions or adaptive remodeling. A higher supplement dose can improve receptor signaling even if the natural ligand supply was adequate. A beneficial compound can also act through several pathways unrelated to replacing the measured deficit.

These explanations are distinguishable. The decisive observation is a pre-loss local deficit plus physical preservation under the defined physiological correction. Neither an enriched lipid pathway nor a favorable symptom score supplies that comparison.

Evidence that moves the judgment

Randomized human evidence supports a structural response in the cornea. In nPROOFS1, 43 people with T1D received omega-3 supplementation (21) or olive-oil placebo (22) for 180 days. The primary corneal nerve-fiber-length difference was 2.70 mm/mm² (95% CI 1.64–3.75) in favor of omega-3. Most small- and large-fiber functional outcomes did not differ.

The supplement provided 1,080 mg EPA plus 720 mg DHA daily. Supplementary Table S4 reports mean individual length changes +2.06±1.73 versus −0.72±1.68, mean±SD, with observed n=21/19. Their unadjusted difference is 2.78 (approximately 1.69–3.87), consistent with the published model. Blood Omega-3 Index rose by 3.3 percentage points (2.4–4.2).

This is a controlled structural positive, not just pain relief. But it measures short corneal fibers, not five-year distal-leg peptidergic preservation, and does not establish a deficient local pool or selective restoration to normal. Britten-Jones et al., 2021, primary supplementary tables.

An association supports a lipid–structure relationship, without proving deficiency. A separate analysis of 47 people, 26 with diabetes and 21 controls, found 0.83 mm/mm² greater central corneal nerve length per percentage-point higher Omega-3 Index (0.16–1.50) after covariate adjustment. Fourteen T1D observations came from the trial's baseline sample, so this is not wholly independent replication. The diabetic subgroup included only ten people with T2D, and no serial nerve outcome. The corresponding association with corneal sensitivity was absent. Blood composition still does not specify the local membrane or mediator. Britten-Jones et al., 2023.

GLA has older randomized functional evidence. Keen randomized 111 people with mild diabetic neuropathy at seven centers to 480 mg/day gamma-linolenic acid or placebo for one year. All 16 clinical/electrophysiological changes favored GLA, with 13 reported significant. They include conduction, sensory thresholds and examination measures, not physical C-terminal counts. Correlated favorable tests should not be counted as 13 independent replications.

The accessible primary abstract does not provide the individual effect intervals or identify a deficient nerve lipid pool. The full efficacy tables and an accompanying 1993 commentary were not recovered; their absence is not negative biological evidence. Keen et al., 1993.

A newer 100-person GLA-versus-alpha-lipoic-acid trial had 73 completers over 12 weeks. It established noninferiority for a pain scale but not the total-symptom-score margin. “Double placebo” describes its double-dummy masking: both groups received an active treatment, so it supplies no untreated/placebo-only efficacy comparison and no structural endpoint. Won et al., 2020.

A resolving lipid improves actual diabetic-mouse skin innervation without establishing natural shortage. After eight untreated diabetic weeks, mice received resolvin D1 1 ng/g/day for 12 weeks. Printed endpoint counts were 20.4±0.3 versus 14.6±0.4 profiles/mm, mean±SEM, with n=13 treated and 16 untreated diabetic mice. The difference is 5.80, approximate Welch interval 4.77–6.83. Controls had 24.7 profiles/mm; the descriptive control-gap closure is 57%, not serial net loss prevented.

Measured serum resolvin D1 did not differ significantly between untreated diabetic and control mice. A serum assay does not prove adequate local signaling, but the treatment result cannot itself prove deficiency either. Some animals were lost during treatment, and the printed ANOVA degrees of freedom disagree with the table's total n; the interval uses its stated endpoint group sizes. Yorek et al., 2016.

Early lipid changes can precede broader nerve abnormalities. Palavicini found, in four to five mice per genotype, one-month-old db/db sciatic nerves with 25% less sulfatide, 18% less galactosylceramide and 27% less P18:0–18:1 plasmalogen PE than controls. These mice were obese and hyperinsulinemic but still normoglycemic. Electrophysiological deficits emerged later.

This supports temporal plausibility, principally in myelin-rich tissue. It does not localize the deficit to unmyelinated terminal-serving units or show rescue by replacing those lipids. Early cardiolipin remodeling was no longer significantly different at four months despite nerve dysfunction. Total tissue composition and single-cell membrane sufficiency are different measurements. Palavicini et al., 2020.

A receptor can be useful without its natural lipid ligand being deficient. Amlodipine improved epidermal fiber density in STZ and db/db mice, with five mice per structural arm, and the benefit was lost after peripheral GPR40 knockdown. The study's human nerve staining involved three DPN samples, without a sufficiently described matched comparator. It measured reduced receptor expression, not reduced ligand availability. Thus even receptor-dependent structural benefit does not resolve the chemical deficit in R1. The study used a pharmacological agonist; its effects are not those of Q. Wei et al., 2025.

The public human plasma null is not an equivalence result. In ADDITION-Denmark's 48 DPN and 49 diabetes-without-DPN participants, the all-plasmalogen-to-diacyl composite ratio was 0.903 (0.791–1.032); the PE composite was 0.873 (0.746–1.021). A 13-metric permutation family gave P=0.333. These data do not establish a circulating separation, but allow reductions of approximately 21–25% in those ratios. They also cannot measure a distal-cell lipid pool. The associated nerve-transcript findings were donor/cell-count sensitive and did not identify a reproducible human peroxisome mechanism. Public study data, numerical support (background note outside this collection).

The uncontrolled En Balance-Plus human omega-3 cohort does not provide another randomized or structural result: the same 40 people received fish oil plus weekly diabetes education, with symptom and metabolic changes. Later publications reuse those participants. The 2026 associated rodent transcriptome concerns nerve constriction, not diabetic neuropathy. 2022 primary study, 2026 primary study.

Conditional probabilities and uncertainty

Let Eᵢ mean that at least one of the six baseline groups satisfies requirements 1 through i together. Row i prices P(Eᵢ | Eᵢ₋₁); the first prices P(E₁). Different groups cannot supply different links. Membrane and soluble-lipid branches are alternatives within this finite union, not probabilities to multiply.

See the claims and probabilities table above.

R1

Early animal lipid remodeling and human blood/corneal associations make a deficit plausible across this finite family. Actual adult human terminal-serving pools and their timing remain unmeasured.

R2

Controlled human corneal growth and animal skin rescue support a structural effect. Their transfer to local physiological replacement and distal human C fibers remains uncertain; pharmacological benefit or adaptive remodeling could explain the positives.

R3

Useful replacement may correct only a small share of injury. Large animal contrasts keep material benefit plausible, but competing injury and increased oxidation could reduce it.

R4

Baseline metabolic/nutritional groups are identifiable, but the case share with a causally limiting local deficit is unknown. A response confined to long-duration T1D or established injury would not qualify.

0.70 × 0.60 × 0.55 × 0.65 = 0.15015, or 15.015% before rounding. Approximately one chance in six is defensible; a distinction between 15% and 20% is not well supported.

The lower scenario [35,20,25,30]% → 0.525% treats observed profiles mainly as systemic markers or useful adaptation and supplementation as extra pharmacology. The upper [90,85,85,85]% → 55.27% assumes pre-loss local deficits occur commonly and physiological correction transfers the controlled structural responses to early human distal nerve.

The roughly 55-percentage-point span is sensitivity to these interpretations, not a statistical confidence interval, calibrated frequency forecast or distribution inferred from study sample sizes. Its breadth is driven by local chemistry, physiological versus pharmacological exposure, tissue transfer and case coverage. It is not a count of missing papers.

R2 and R3 divide one physical question into modest and material effects; their joint conditional is 33%. Merging them leaves the headline unchanged. There is no extra multiplier for “lack of compensation”: compensation is already included in Q's net effect.

Hypothetical upward update: replicated local human deficits followed by selective physiological correction, B≥20% and adequate early-T2D case coverage would support [90,85,80,80]%, yielding 49.0%.

Hypothetical downward update: direct measurements showing normal relevant pools in most candidate groups, or valid correction with only marker/pain changes and precise B<5%, would support [40,25,30,40]%, yielding 1.2%. If only R1 became certain while all later prices stayed fixed, the headline would reach 21.45%. That is a maximum effect of resolving one requirement, not the expected value of measuring one more plasma lipid panel.

Ideal experiments that would resolve each claim

These are ideal causal comparisons, not proposed laboratory work. Shared conditions are the stated early-T2D population, six baseline groups, finite lipid panels, matched normal references, connected compartments, Q and five-year physical endpoint. Independent people, not lipid features, sections or cells, provide replication. Quantify molecular identity, localization, viable membrane composition and net structural trajectories, including adverse consequences.

For each claim, YES requires the true quantity to meet its inclusive boundary in at least one group satisfying all preceding requirements. NO requires every such group to fall short. Finite-data intervals crossing a boundary, incomplete group coverage or a failed manipulation leave the claim unresolved. Use simultaneous precision across the fixed candidate family when interpreting finite observations.

1 — Is there a pre-loss local deficit? Repeatedly measure the named membrane profiles and accessible free lipids in mapped terminal-serving units, alongside physical terminal identities. Distinguish lower concentration from reduced cell/membrane amount, assay oxidation, extraction recovery and cell-composition changes. YES: a group meets the ≥20%-below-median and below-fifth-percentile, 30-day, ≥10% people and ≥10% terminal-coverage definition before sustained loss. NO: adequately measured groups all fail at least one condition. Unresolved: blood-only profiles, lipid identities lacking positional specificity, post-loss tissue, or normal concentrations too poorly measured to establish a deficit. Palavicini supplies timing in whole mouse nerve; the human sources do not supply this comparison.

2 — Is the deficit-containing state causally correctable at physiological exposure? Within R1-positive groups, compare Q with matched sham lipid handling under identical incoming systemic exposures. Verify normal-range free-lipid trajectories and the specified joint membrane fractions and proportional displacement without changing membrane area, directly generating axons or exposing subjects to an unpriced drug. The membrane result concerns this balance change; include any destabilization caused by verified exchange. YES: at least one R1 group has L₀>0 and B≥5%. NO: all have B<5% with verified correction, including harm. Unresolved: supraphysiological agonism, unverified delivery, altered systemic glucose, or preservation inferred only from staining intensity. nPROOFS1 and Yorek support lipid-responsive structure; neither identifies this physiological deficit correction.

3 — Does correction preserve enough of the actual net loss? Continue the same comparison through five years and measure disappearance and completed connected replacement, including loss of replacements. Do not substitute a larger supplement dose, gross sprouting or percentage recovery of a cross-sectional control deficit. YES: one R1–R2 group has B≥20%. NO: every such group has B<20%, including real effects between 5% and 20%. Unresolved: broad intervals, selective attrition/death, incomplete loss-versus-replacement accounting, or an effect measured only in the cornea. This adds a magnitude requirement to R2, not another molecular mechanism.

4 — Does a fully qualifying group cover enough incident disease? In a representative reference-care early-T2D cohort, establish the baseline group definitions and objective incident DPN, including competing deaths. Combine that denominator with the same groups shown to meet R1–R3. YES: at least one fully qualifying group contains C≥10% of cases. NO: all fully qualifying groups contain C<10%. Unresolved: a fraction derived from a selected supplement trial, advanced-DPN referrals, blood-deficiency prevalence alone, or a group chosen by subsequent response. Existing data do not estimate this causal case share.

Boundaries and remaining uncertainty

The membrane branch can overlap fatty-acid composition and Schwann support, but changing an axonal membrane is not identical to changing incoming fatty-acid delivery or glial export. The soluble branch can alter inflammation or repair. Those overlaps do not order the complete probabilities under different interventions and scopes.

Polyunsaturated lipids can also become substrates for oxidative membrane injury. Their presence is not uniformly protective, and the ideal comparison includes that cost. Removing toxic deoxylipids alone would not demonstrate the specified physiological lipid deficit.

Human local lipid measurements, selective physiological correction and incident-case coverage remain the consequential gaps. The registry of the impaired-glucose-tolerance fish-oil trial, NCT05145452, reports termination after 31 enrollments because enrollment was interrupted and pills expired; it supplies no posted efficacy results. That missing comparison is unresolved, not a failed biological treatment.