{
  "id": "06",
  "title": "Glucose overload and mitochondrial superoxide",
  "summary": "Excess glucose-derived reducing input may drive mitochondrial superoxide production and terminal injury.",
  "source_title": "Glucose-driven respiratory over-reduction and superoxide",
  "page": "06_glucose_mitochondrial_overload.html",
  "source_markdown": "sources/06_glucose_mitochondrial_overload.md",
  "structured_data": "data/06_glucose_mitochondrial_overload.json",
  "snapshot_date": "2026-09-14",
  "source_review_date": "2026-09-14",
  "scope_label": "Early type 2 diabetes",
  "scope_exception": false,
  "source_headline_markdown": "**Best judgment: about 5%; coordinated sensitivity scenarios: 0.06–38%.** These are subjective probabilities of the complete human proposition below, not treatment effects or confidence limits. Reviewed 14 September 2026. [Probability method](probability_method.md) · [Collection index](README.md).",
  "probability_percent": {
    "skeptical": 0.0590625,
    "base": 4.5045,
    "favorable": 37.58445
  },
  "probability_meaning": "Subjective belief in the complete causal proposition. Scenario endpoints are sensitivity products, not confidence limits.",
  "causal_proposition_markdown": "In adults diagnosed with type 2 diabetes within five years and without clinical DPN, unusually high glucose-derived electron supply over-reduces respiratory-chain carriers in peptidergic sensory neurons. The resulting additional mitochondrial superoxide contributes enough to subsequent terminal loss that selectively preventing that source would reduce five-year net loss by at least 20%, in a baseline-describable subgroup accounting for at least 10% of reference-care incident DPN cases.\n\nThe relevant compartments are mitochondria in the neuron, from DRG soma through peripheral axon to skin terminal. The biochemical steps must occur in connected compartments of the neurons whose terminals are followed, beginning while those terminals remain structurally intact. Glucose-derived input includes reducing equivalents supplied through pyruvate oxidation and cytosolic shuttles; blood glucose, glucose uptake and glycolytic gene expression are not measurements of that input. Increased oxidation of other fuels alone does not meet this proposition.\n\nThe allowed baseline groups are the whole eligible population, HbA1c at least 7.5%, or neuronal glucose-derived reducing-input rate above the age- and sex-matched nondiabetic 95th percentile. A group is defined by those measurements, never by treatment response. Rates and redox fractions below are group means across independently sampled people and the same connected neuronal compartments. Repeated measurement across ordinary feeding, sleep and activity is required; a laboratory glucose bolus does not define ordinary exposure.\n\nFor person i, let L_i = (terminal disappearances minus newly established terminals)/baseline terminal count over five years. Track actual structures with stable peptidergic identity, not disappearance of CGRP staining. For a group with positive reference-care mean loss L0, define B = (L0 − L1)/L0. Material preservation means B at least 0.20; B at least 0.05 denotes a smaller contribution. New terminals that later disappear enter both counts. Coverage C is the fraction of all reference-care incident DPN cases belonging to that baseline group. Incident DPN requires new bilateral length-dependent sensory signs with objective small-fiber loss or abnormal nerve conduction, excluding another neuropathy cause. C must be at least 0.10. Neither threshold is a biological constant.",
  "claims": [
    {
      "id": "R1",
      "source_id": "R1",
      "claim": "Glucose-derived reducing input has the specified persistent pre-loss elevation.",
      "claim_markdown": "Glucose-derived reducing input has the specified persistent pre-loss elevation.",
      "conditional_percent": {
        "skeptical": 35.0,
        "base": 60.0,
        "favorable": 85.0
      },
      "cumulative_base_percent": 60.0,
      "source_cumulative_display": "60%",
      "reason_markdown": null,
      "question": "Is glucose-derived electron supply to neuronal mitochondria persistently elevated before terminal loss?"
    },
    {
      "id": "R2",
      "source_id": "R2",
      "claim": "Normalizing that input with G lowers the reduced fraction of at least one named carrier pool (strictly positive mean contrast).",
      "claim_markdown": "Normalizing that input with G lowers the reduced fraction of at least one named carrier pool (strictly positive mean contrast).",
      "conditional_percent": {
        "skeptical": 25.0,
        "base": 50.0,
        "favorable": 80.0
      },
      "cumulative_base_percent": 30.0,
      "source_cumulative_display": "30%",
      "reason_markdown": null,
      "question": "Would normalizing glucose-derived electron supply lower the fraction of coenzyme-Q or complex-I flavin carriers in their reduced state?"
    },
    {
      "id": "R3",
      "source_id": "R3",
      "claim": "Adjusting that carrier state with J lowers respiratory-chain superoxide production (strictly positive mean contrast).",
      "claim_markdown": "Adjusting that carrier state with J lowers respiratory-chain superoxide production (strictly positive mean contrast).",
      "conditional_percent": {
        "skeptical": 30.0,
        "base": 70.0,
        "favorable": 90.0
      },
      "cumulative_base_percent": 21.0,
      "source_cumulative_display": "21%",
      "reason_markdown": null,
      "question": "Would directly correcting the redox state of coenzyme-Q or complex-I flavin carriers reduce respiratory-chain superoxide production?"
    },
    {
      "id": "R4",
      "source_id": "R4",
      "claim": "Preventing that additional source with S reduces five-year net terminal loss by at least 5%.",
      "claim_markdown": "Preventing that additional source with S reduces five-year net terminal loss by at least 5%.",
      "conditional_percent": {
        "skeptical": 25.0,
        "base": 55.0,
        "favorable": 85.0
      },
      "cumulative_base_percent": 11.55,
      "source_cumulative_display": "11.55%",
      "reason_markdown": null,
      "question": "Would selectively preventing the extra respiratory-chain superoxide production caused by carrier over-reduction avert at least 5% of five-year net terminal loss?"
    },
    {
      "id": "R5",
      "source_id": "R5",
      "claim": "The same S benefit reaches at least 20%.",
      "claim_markdown": "The same S benefit reaches at least 20%.",
      "conditional_percent": {
        "skeptical": 30.0,
        "base": 60.0,
        "favorable": 85.0
      },
      "cumulative_base_percent": 6.93,
      "source_cumulative_display": "6.93%",
      "reason_markdown": null,
      "question": "Would selectively preventing that carrier-driven respiratory-chain superoxide production avert at least 20% of five-year net terminal loss?"
    },
    {
      "id": "R6",
      "source_id": "R6",
      "claim": "A group satisfying R1–R5 accounts for at least 10% of reference-care incident DPN cases.",
      "claim_markdown": "A group satisfying R1–R5 accounts for at least 10% of reference-care incident DPN cases.",
      "conditional_percent": {
        "skeptical": 30.0,
        "base": 65.0,
        "favorable": 85.0
      },
      "cumulative_base_percent": 4.5045,
      "source_cumulative_display": "4.50%",
      "reason_markdown": null,
      "question": "Does the benefiting subgroup account for at least 10% of incident DPN cases under reference care?"
    }
  ],
  "sections": [
    {
      "heading": "The causal proposition",
      "markdown": "In adults diagnosed with type 2 diabetes within five years and without clinical DPN, unusually high glucose-derived electron supply over-reduces respiratory-chain carriers in peptidergic sensory neurons. The resulting additional mitochondrial superoxide contributes enough to subsequent terminal loss that selectively preventing that source would reduce five-year net loss by at least 20%, in a baseline-describable subgroup accounting for at least 10% of reference-care incident DPN cases.\n\nThe relevant compartments are mitochondria in the neuron, from DRG soma through peripheral axon to skin terminal. The biochemical steps must occur in connected compartments of the neurons whose terminals are followed, beginning while those terminals remain structurally intact. Glucose-derived input includes reducing equivalents supplied through pyruvate oxidation and cytosolic shuttles; blood glucose, glucose uptake and glycolytic gene expression are not measurements of that input. Increased oxidation of other fuels alone does not meet this proposition.\n\nThe allowed baseline groups are the whole eligible population, HbA1c at least 7.5%, or neuronal glucose-derived reducing-input rate above the age- and sex-matched nondiabetic 95th percentile. A group is defined by those measurements, never by treatment response. Rates and redox fractions below are group means across independently sampled people and the same connected neuronal compartments. Repeated measurement across ordinary feeding, sleep and activity is required; a laboratory glucose bolus does not define ordinary exposure.\n\nFor person i, let L_i = (terminal disappearances minus newly established terminals)/baseline terminal count over five years. Track actual structures with stable peptidergic identity, not disappearance of CGRP staining. For a group with positive reference-care mean loss L0, define B = (L0 − L1)/L0. Material preservation means B at least 0.20; B at least 0.05 denotes a smaller contribution. New terminals that later disappear enter both counts. Coverage C is the fraction of all reference-care incident DPN cases belonging to that baseline group. Incident DPN requires new bilateral length-dependent sensory signs with objective small-fiber loss or abnormal nerve conduction, excluding another neuropathy cause. C must be at least 0.10. Neither threshold is a biological constant."
    },
    {
      "heading": "From electron supply to structural injury",
      "markdown": "At steady state, reducing input is balanced by oxidation or diversion. Increased supply need not accumulate indefinitely: the relevant question is whether it changes carrier reduction enough to increase one-electron transfer to oxygen. Faster respiration can consume the additional input; competing fuels can fall; proton leak can lower membrane potential; glucose can enter other pathways. Conversely, low oxygen consumption can coexist with highly reduced carriers if downstream electron transfer is restricted. OCR alone therefore does not diagnose either overload or its absence.\n\nDefine G as an ideal adult-onset, neuron-restricted adjustment of **glucose-derived reducing input** to its matched nondiabetic trajectory, without lowering systemic glucose or other initiating exposures. Carbon and reducing-equivalent balances must be measured; this is not achieved simply by labeling an intervention a glucose blocker. For the redox claim, at least one of two named carrier pools must show a strictly positive decrease in its reduced fraction under G: the coenzyme-Q pool or the complex-I flavin pool. These are alternative branches, not two compulsory discounts. Hyperpolarization is informative but is not required in addition to a demonstrated carrier-redox mechanism.\n\nDefine J as a direct adjustment of the implicated carrier-redox state to its G trajectory while glucose-derived input remains at reference levels. Define S as an ideal source-specific intervention that sets mitochondrial respiratory-chain superoxide production to the trajectory measured under J, while retaining reference glucose-derived input and electron flow used for ordinary respiration. It changes the defined electron-leak reaction, not the outcome-selected “harmful part.” Normal oxidant signaling outside that source remains. Subsequent peroxide formation, redox signaling, DNA/protein/lipid damage, NAD metabolism, ATP production and repair may respond. A generic antioxidant, mitochondrial poison or change in blood glucose is not S.\n\nThe redox-source claim requires strictly lower respiratory-chain superoxide production under J, with restoration of the reference carrier state reversing that source-rate change. These biochemical claims retain a zero sign boundary: a small carrier change could produce a large downstream effect. Finite equivalence near zero is informative but cannot establish exact absence. The later structural boundary supplies the minimum consequential effect; no arbitrary minimum redox shift is imposed. Sustained input means exceeding the matched 95th percentile for at least 90 cumulative days during year one, before the subsequent losses attributed to this route. Brief recurrent shifts within those days can contribute to the measured daily exposure. A single acute culture response does not settle persistence.\n\nSuperoxide production can rise without injury if defenses remove it. Conversely, oxidative damage may originate from NOX, peroxisomes or other chemistry when mitochondrial electron leak is unchanged. Structural causation is therefore a separate requirement. GAPDH/PARP amplification, apoptosis and any particular downstream damage target are optional mediators; none earns an additional mandatory multiplier."
    },
    {
      "heading": "Probability decomposition",
      "markdown": "Each conditional price assumes all preceding claims in the **same qualifying group and connected neurons**. After each row retain only groups meeting every earlier row. The conjunction is the full proposition, including intervention benefit and case coverage.\n\n| Claim | Additional proposition | Skeptical / best / favorable | Cumulative best |\n|---|---|---:|---:|\n| R1 | Glucose-derived reducing input has the specified persistent pre-loss elevation. | 35 / **60** / 85% | 60% |\n| R2 | Normalizing that input with G lowers the reduced fraction of at least one named carrier pool (strictly positive mean contrast). | 25 / **50** / 80% | 30% |\n| R3 | Adjusting that carrier state with J lowers respiratory-chain superoxide production (strictly positive mean contrast). | 30 / **70** / 90% | 21% |\n| R4 | Preventing that additional source with S reduces five-year net terminal loss by at least 5%. | 25 / **55** / 85% | 11.55% |\n| R5 | The same S benefit reaches at least 20%. | 30 / **60** / 85% | 6.93% |\n| R6 | A group satisfying R1–R5 accounts for at least 10% of reference-care incident DPN cases. | 30 / **65** / 85% | 4.50% |\n\nR1 is discounted by the absence of human neuronal flux measurements and contrary established-diabetes tracer results. R2 remains plausible because acute neuronal uncoupling experiments exist, but their developmental state and exposures matter. Conditional on the carrier shift, R3 is more credible than an unconditioned claim about all diabetic oxidants. R4 concerns the additional source at the natural human dose, not whether a large oxidant challenge can kill neurons. R5 prices magnitude after accepting a small effect; R6 prices commonness after accepting magnitude. Neither repeats the structural-causation discount."
    },
    {
      "heading": "Evidence that constrains the judgment",
      "markdown": "**The foundational result is not a human-neuron experiment.** Nishikawa and colleagues suppressed the high-glucose ROS response and several biochemical pathways in bovine aortic endothelial cells using respiratory perturbations, UCP1 and MnSOD. This establishes a possible biochemical connection in that system, without measuring adult human terminal survival. The accessible primary abstract does not supply an independent culture n or a terminal endpoint. [Nishikawa et al., 2000](https://doi.org/10.1038/35008121).\n\n**There are direct neuronal positives.** Russell and colleagues observed early hyperpolarization followed by depolarization, ROS accumulation, mitochondrial swelling and caspase activation under 45 mM glucose, with respiratory inhibitors suppressing several responses. The control medium already contained 25 mM glucose; mitochondria were about 50% larger at six hours. The accessible primary methods use defined medium without the usual antioxidant supplement. These are short culture responses, not a natural human exposure-response curve or long-term terminal-loss measurement. [Russell et al., 2002, primary full text](https://www.researchgate.net/publication/11056013_High_glucose-induced_oxidative_stress_and_mitochondrial_dysfunction_in_neurons).\n\nIn embryonic-day-15 rat DRG cultures, UCP1/UCP3 expression blocked the 25-to-45 mM glucose response in membrane potential, fluorescent ROS signals and caspase/TUNEL endpoints. Caspase experiments used three occasions with duplicate samples; membrane-potential and DHE assays used five experiments with duplicates, not ten independent animals. The paper also compared eight control with eight four-week STZ-diabetic rats for UCP3 protein. The temporal hyperpolarization-to-depolarization sequence supports a stage-dependent possibility. However, uncoupling changes energetics as well as electron leak, and DCF/DHE fluorescence does not uniquely identify the proposed source or stable peptidergic terminals. [Vincent et al., 2004, primary full text](https://www.researchgate.net/publication/6602000_Uncoupling_Proteins_Prevent_Glucose-Induced_Neuronal_Oxidative_Stress_and_Programmed_Cell_Death).\n\n**Adult neurons do not uniformly reproduce those responses.** Zherebitskaya and colleagues found no additional neuronal-survival loss under 50 versus 10 mM glucose over four weeks in adult-rat cultures, with four replicate cultures; 25 mM glucose also failed to raise their oxidative readouts in previously normal neurons. Neurons from diabetic rats did show glucose-sensitive axonal oxidative changes, and 1 mM NAC improved outgrowth and swellings in small culture experiments. The history of diabetes mattered, and soma survival differed from axonal morphology. This challenges simple direct glucotoxicity without excluding sensitized adult axons or a nonmitochondrial oxidant route. [Zherebitskaya et al., 2009](https://pmc.ncbi.nlm.nih.gov/articles/PMC2682687/).\n\nIn 22-week STZ-diabetic rats, Akude and colleagues found reduced DRG respiratory activity and an altered mitochondrial proteome; respiration used n=5, proteomics 4 control/3 diabetic/4 insulin-treated animals. Diabetic-neuron axons had altered depolarization kinetics and smaller FCCP-induced MitoSOX responses despite greater general oxidative staining. Imaging captions count 18–85 axons rather than independent rats; glucose and insulin differed between the principal culture groups. Oligomycin pretreatment addressed potential-dependent dye loading, but the fluorescence remains an indirect source assay. This is a substantive contradiction to sustained overload in that late model, not proof that no early human overload occurs. [Akude et al., 2011](https://pmc.ncbi.nlm.nih.gov/articles/PMC3012184/).\n\n**Tracer evidence favors tissue-specific routing.** Sas and colleagues used 24-week db/db mice, n=8/group for labeled glucose, and observed lower glucose-derived TCA labeling in sciatic nerve despite higher labeling in kidney. Their oral tracer dose was 2 g/kg; pyruvate administration increased nerve TCA labeling. Nerve transcript increases therefore cannot substitute for glucose oxidation. However, this is bulk nerve in established disease, and labeled metabolite amounts after a bolus depend on pool size and precursor enrichment as well as flux. It supplies neither a calibrated resting neuronal electron-input rate nor an early human negative. The nerve static-metabolite panel reuses earlier Hinder data and is not an independent replication of it. [Sas et al., 2016](https://insight.jci.org/articles/view/86976).\n\n**Perturbing mitochondrial redox can change structural outcomes without identifying the initiating fuel.** In high-fat/low-dose-STZ rats, twelve weeks of dietary MitoQ after four weeks of hyperglycemia increased skin IENF density from 14.0±0.2 to 17.8±0.4 profiles/mm, n=9/arm, SEM; controls were 21.8±0.4. The reconstructed unadjusted treatment contrast is +3.8, 95% Welch interval 2.8–4.8. This recovers about 49% of the endpoint deficit relative to controls, not 49% of longitudinal loss. Obese rats showed 17.5±0.4 versus 17.8±0.2, n=8/9: +0.3, interval −0.7–1.3. The dietary amount was 0.93 g/kg diet, not per kg body weight. Glucose did not improve in treated diabetic rats, but neuronal superoxide-source engagement was not measured; systemic lipid-peroxidation and uncoupling effects remain alternatives. [Fink et al., 2020](https://pmc.ncbi.nlm.nih.gov/articles/PMC7329571/).\n\nSOD2 haploinsufficiency worsened neuropathy in a db/db background but did not produce neuropathy during six months of STZ diabetes in the tested C57BL/6J background, with ten animals/group for nerve testing. Culture triplicates repeated on three occasions are three preparations, not nine independent biological units. These results support context-dependent oxidant defense, not glucose-specific excess production. A later nestin-Cre SOD2 model developed early CNS disease and lethality, preventing its intended adult-diabetic test. [Vincent et al., 2007](https://pmc.ncbi.nlm.nih.gov/articles/PMC2190625/); [Oh et al., 2012](https://pubmed.ncbi.nlm.nih.gov/22516022/).\n\n**Human evidence does not identify this complete route.** Skin from 69 recent-onset T2 participants versus 51 controls had about 60% greater subepidermal SOD2-stained area. This is an antioxidant-response marker, not a neuronal glucose-flux or superoxide-production measurement, and was cross-sectional. [Ziegler et al., 2015](https://pubmed.ncbi.nlm.nih.gov/25933618/). The four-year, 460-person NATHAN 1 alpha-lipoic-acid trial missed its primary composite endpoint; established disease, nonselective pharmacology and no direct terminal-loss/source assay make it a weak negative for this exact proposition. [NATHAN 1](https://pmc.ncbi.nlm.nih.gov/articles/PMC3161301/). No retrieved human selective source intervention or usable genetic perturbation supplies the missing early neuronal chain. Positive vascular/leukocyte studies and broad glycemic interventions cannot supply it by substitution."
    },
    {
      "heading": "Ideal experiments that would resolve each claim",
      "markdown": "These designs specify truth conditions, not available capabilities or proposed studies. They require noninjurious repeated measurement of intact human neuronal compartments and selective adult interventions. A representative early-T2 cohort and matched nondiabetic reference group establish baseline subgroup membership. Preserve ordinary reference diabetes care in every causal arm. Follow independently randomized people for five years and track identified terminals. Match other initiating inputs, including systemic glycemia, lipid delivery, oxygenation, trophic support and pre-existing inflammation; allow downstream responses to change. Exact engagement must be demonstrated, not inferred from drug assignment. Simultaneous confidence intervals across the finite groups/branches must clear the stated boundary; a boundary-spanning interval remains unresolved.\n\n**R1 — persistent glucose-derived input.** With no preceding assumptions, directly measure absolute glucose-derived reducing-input rates in intact soma/axon/terminal mitochondria, correcting isotope enrichment, pool turnover and alternative fuels. Use repeated ordinary-day measurements from entry through year one. YES requires a permitted group's mean input above its matched nondiabetic 95th percentile for at least 90 cumulative days while the tracked terminals remain intact before the subsequent losses attributed to this route. Include all baseline participants and early structural trajectories; do not select only eventual decliners or year-one survivors. NO means no permitted group meets this exposure-duration condition with adequate coverage and precision. Infrequent snapshots, uncalibrated enrichment or only soma measurements with no connection to tracked axons remain unresolved. Existing tracer work supplies bulk animal labeling; it does not resolve R1 in humans.\n\n**R2 — input-dependent carrier reduction.** Assuming R1, randomize G versus a sham in the same group and measure the coenzyme-Q and complex-I flavin reduced fractions in the connected compartments during the R1 exposure periods. Confirm normalization of glucose-derived input without directly altering carrier proteins, oxygen or other fuel supply. For contrast reference minus G, YES requires at least one carrier-pool mean contrast greater than zero; NO requires both contrasts no greater than zero. An interval entirely above zero can decide YES; intervals entirely below zero for both pools can decide NO. Exact zero belongs to NO but ordinarily remains unresolved by a finite experiment. Tiny boundary-spanning intervals cannot be declared a biological null. Restore the original input to confirm the direction and distinguish intervention artifacts. High potential without either specified redox contrast, or a lower OCR by itself, does not decide this row. Existing uncoupling experiments support possible redox control but do not measure this exact human contrast.\n\n**R3 — superoxide from that carrier change.** Assuming R1–R2, directly set the implicated carrier pool to its G trajectory with J while retaining reference glucose input, using controlled electron donation/removal and a measured material balance. Measure absolute respiratory-chain superoxide generation under J and sham, separating production from scavenging and source from fluorescent-probe uptake. Restore the reference carrier state in a separate arm to check specificity. YES requires a positive mean production-rate difference, reference minus J, attributable to the carrier adjustment; NO is a nonpositive difference for every carrier branch that passed R2. Complete state restoration should restore the source trajectory under a selective manipulation; failure of that check leaves specificity unresolved rather than establishing mediation. A confidence interval wholly above zero supports YES; one wholly below zero supports NO. Exact zero belongs to NO but a finite interval spanning zero remains unresolved, even if narrow. Both allowed carriers must be excluded to reject their union. This is a mediation comparison, not an added assumption that all mitochondrial ROS is toxic. Existing studies lack the source-resolved carrier-restoration contrast.\n\n**R4 — a smaller structural contribution.** Assuming R1–R3, randomize S versus sham from cohort entry, adjusting the source during the verified glucose-overload episodes and maintaining the comparison for five years. Verify that S reproduces the J-associated source reduction while retaining reference initiating exposures and normal respiratory electron transfer. A separate method achieving the same source trajectory and a source-restoration arm distinguish source effects from intervention artifacts. Count net terminal losses with stable subtype tracking. YES requires B≥0.05; NO requires B<0.05 at adequate precision, including no benefit or harm. An estimated 6.7% benefit can meet R4 while failing R5. Failed source engagement, interrupted follow-up or pain-only endpoints remain unresolved. MitoQ supplies an animal structural contrast, but not S or the target human population.\n\n**R5 — material magnitude.** Assuming R1–R4, use the same five-year causal comparison and estimate B directly, propagating uncertainty in both group mean losses. YES requires B≥0.20 and NO B<0.20 with the interval wholly on one side. For L0=0.30, L1=0.28 gives 6.7% and fails; L1=0.23 gives 23.3% and meets the criterion. If L0 is not positive the relative-loss proposition is not met; an imprecisely estimated near-zero denominator is unresolved. This row adds magnitude, not another generic test that oxidants can cause damage. No current study measures it for S in early human T2 diabetes.\n\n**R6 — case coverage.** Assuming R1–R5 for at least one permitted baseline group, estimate that group's share of incident cases in the population-representative reference-care arm over five years. YES is C≥0.10; NO is C<0.10 for every otherwise qualifying group with precise coverage estimates. For example, 8 of 100 incident cases fails and 15 of 100 meets the truth boundary, though those finite counts alone may remain statistically inconclusive. Response-selected recruitment and case fractions among treated survivors cannot decide C. Existing selected cultures and animal strains provide no estimate of this human coverage."
    },
    {
      "heading": "Calibration and remaining uncertainty",
      "markdown": "The six best inputs [60,50,70,55,60,65]% multiply to 4.5045%; the digits show arithmetic only. The low scenario [35,25,30,25,30,30]% treats glucose oxidation as constrained in adult neurons and the positive oxidant interventions as acting through other sources or exposures. The high scenario [85,80,90,85,85,85]% allows a common early overload phase with effective selective source rescue that later bulk measurements miss. Their products, 0.059% and 37.58%, are coordinated interpretation scenarios, not independent row distributions or statistical credible limits.\n\nBiochemical timing/source identity is the main uncertainty; population coverage remains largely unmeasured. If only R2 moved from 50% to 80% while the other judgments stayed fixed, the product would be 7.21%; at 20% it would be 1.80%. Real evidence resolving the early connected chain should update several conditionals together. Merging R1–R3 gives 21% in the best scenario and leaves the complete price unchanged; adding prose steps must not lower it mechanically.\n\nThe price concerns the explicit persistence, carrier, source-rate, structural and coverage boundaries above. Allowing a shorter initiating episode broadens the claim; stricter persistence or a 30% terminal benefit narrows it. The biochemical sign boundaries remain distinct from finite measurement resolution. Current evidence does not justify a numerical threshold-response curve. Stronger evidence for general oxidative injury or deficient ATP cannot be transferred wholesale into this narrower glucose-origin claim."
    },
    {
      "heading": "Relationship to other hypotheses",
      "markdown": "[Bioenergetic insufficiency](07_bioenergetic_insufficiency.md) and [oxidative/nitrosative injury](39_oxidative_nitrosative_injury.md) can occur when R1–R3 fail. Conversely, glucose-driven superoxide could impair terminals through non-ATP mediators. These probabilities describe overlapping causal propositions and are not additive shares of DPN. Protection by an intervention shared across files is one experiment, not independent support each time it is cited."
    }
  ],
  "source_claim_table_markdown": "| Claim | Additional proposition | Skeptical / best / favorable | Cumulative best |\n|---|---|---:|---:|\n| R1 | Glucose-derived reducing input has the specified persistent pre-loss elevation. | 35 / **60** / 85% | 60% |\n| R2 | Normalizing that input with G lowers the reduced fraction of at least one named carrier pool (strictly positive mean contrast). | 25 / **50** / 80% | 30% |\n| R3 | Adjusting that carrier state with J lowers respiratory-chain superoxide production (strictly positive mean contrast). | 30 / **70** / 90% | 21% |\n| R4 | Preventing that additional source with S reduces five-year net terminal loss by at least 5%. | 25 / **55** / 85% | 11.55% |\n| R5 | The same S benefit reaches at least 20%. | 30 / **60** / 85% | 6.93% |\n| R6 | A group satisfying R1–R5 accounts for at least 10% of reference-care incident DPN cases. | 30 / **65** / 85% | 4.50% |"
}