# 16. DAG-dependent vascular PKCβ activity reduces nerve supply

**Best judgment: about 5% (5.10% by multiplication); interpretation sensitivity: 0.02–48%.** These are subjective probabilities of the causal proposition, not treatment effects or confidence intervals. Reviewed 14 September 2026. [Probability method](probability_method.md) · [Collection index](README.md).

## The causal proposition

In adults with type 2 diabetes diagnosed within five years and no clinical DPN, excess signaling-accessible diacylglycerol (DAG) drives excessive **vascular PKCβI or PKCβII** activity before terminal loss. Correcting that activity would avert **at least 20% of five-year net distal peptidergic C-fiber terminal loss**, with at least that much preservation attributable to restored local oxygen/fuel availability, in a baseline group covering **at least 10% of reference-care incident DPN cases**.

The β splice isoforms are specified because they carry the principal nerve-vascular and clinical intervention evidence. A result involving neuronal PKC, another vascular isoform or DAG-independent kinase activation is a different proposition. This does not claim that all PKC signaling is excessive or harmful.

Reference care continues without selective PKC correction. Exclude known vasculitis and severe large-artery ischemia, defined here as ankle systolic pressure <50 mmHg or toe pressure <30 mmHg. Incident DPN means new bilateral length-dependent clinical signs with objective small-fiber loss or abnormal nerve conduction, excluding another sufficient cause. The target is the distal-leg/foot ending of a stably identified peptidergic C-fiber, not pain, a marker-positive profile or mixed-nerve conduction.

For each person, `L = (terminal disappearances − successful new terminals)/baseline terminal number` over five years. A new ending later lost enters both counts. Average sampled territories within each person; people are the inference unit. `B = (mean L_reference − mean L_Q)/mean L_reference`, requiring positive reference net loss. `C = P(group | incident DPN under reference care)`.

Candidate baseline groups are the whole eligible population; HbA1c ≥7.5%; foot transcutaneous oxygen <40 mmHg; or measured nerve/skin microvascular exchange capacity below the matched nondiabetic fifth percentile. The last group requires ideal measurements. Membership precedes treatment and outcome. Overlapping groups are assessed separately, not combined after seeing responses.

### Molecules, compartments and interventions

DAG means native **sn-1,2-diacylglycerols accessible to the cytoplasmic C1-binding domains**, measured as membrane mole fraction, rather than total blood triglyceride or all tissue DAG. Restrict acyl chains to C14–C24, zero to six double bonds; resolve isomers. The initiating pool is above the age-, sex-, site- and cell-type-matched nondiabetic 95th percentile for ≥90 cumulative days in year one before structural disappearance.

Candidate vascular cells are endothelial cells, pericytes and arteriolar smooth-muscle cells supplying the terminal environment, connected axon/Remak unit or DRG soma. A cell type/site alone or their joint correction can qualify. At least **10% of mapped baseline neuronal territories**, averaged first within each person and then across the group, must have the specified connected vascular exposure. One exceptional vascular cell is insufficient. These group, duration and burden rules are chosen definitions.

**D** normalizes the excess accessible DAG pool to its matched 95th-percentile boundary using reversible sequestration with explicit lipid mass balance. It does not generate phosphatidic acid or deplete every membrane lipid. D is used to resolve DAG's contribution to kinase activity, not as a presumed selective five-year vascular treatment.

**Q** caps PKCβI, PKCβII or both at the matched 95th-percentile **native substrate-phosphorylation flux** in the specified vascular cells when activity exceeds that boundary. Flux is measured per cell volume and time, preserving normal localization and substrate access. Q leaves normal activity intact and does not directly inhibit neuronal/glial kinases, sodium channels or unrelated vascular isoforms. It allows all endogenous vascular consequences, including changes in nitric oxide, permeability, cytokines and trophic signals.

The supply readout is local interstitial oxygen partial pressure or free glucose, lactate, pyruvate or β-hydroxybutyrate concentration in the connected neuronal territory. R3's deficit is below the matched nondiabetic fifth percentile for ≥90 cumulative pre-loss days in at least the same 10% person-weighted neuronal fraction. Q must remove ≥50% of cumulative below-threshold neuronal exposure time over equal year-one windows. Changes in bulk flow or stimulus-evoked vasodilation do not substitute for this readout.

### Distinguishing supply-mediated preservation

Vascular PKC correction could protect through inflammatory or trophic changes without improving supply. Define a second ideal intervention **K** that, during Q, restores local oxygen/fuel trajectories to the reference-care distribution at the same baseline characteristics, circulating inputs, time and activity state. K changes those extracellular quantities by controlled exchange; it does not directly alter neuronal receptors or intracellular ATP. It uses reference-care exposure profiles, not an arbitrarily severe ischemic insult. This is an intervention on a measured distribution, not an assumption that both personal counterfactual histories can be observed.

Let `B_supply = (mean L_Q+K − mean L_Q)/mean L_reference`. This controlled supply component asks how much of Q's structural benefit is lost when its supply improvement is removed. It permits interactions and need not add up to a unique natural indirect effect. The material proposition requires **min(B, B_supply) ≥0.20**. Fidelity controls must show that the exchange apparatus itself does not create unrelated injury. Full downstream consequences remain free under each assigned exposure.

## Biological reasoning

An increased DAG pool need not produce increased PKCβ flux: localization, calcium, binding partners and turnover matter. Expression, membrane association and activity are distinct. Whole-nerve measurements can average reduced neural PKCα with increased vascular PKCβII. Correcting a kinase's vascular activity need not reproduce a systemic inhibitor's direct neuronal actions.

Vascular tone, diffusion and permeability can change without limiting oxygen or fuel. Better axon-reflex vasodilation can reflect nerve signaling rather than increased nutritive delivery. Conversely, physiological supply can improve while another injury route continues. ATP depletion, oxidative injury and inflammatory responses are possible downstream routes, not simultaneous mandatory requirements.

## Probabilities of the necessary claims

Each row assumes all predecessors and retains the same connected group, vascular cells and correction alternatives. These are conditional probabilities. Their conjunction expresses the complete proposition; the table does not multiply independent marginal beliefs.

| Claim | Exact additional requirement | Skeptical / best / favorable | Best cumulative | Evidence and uncertainty |
|---|---|---:|---:|---|
| R1 | At least one specified group has the defined pre-loss accessible vascular DAG excess. | 30 / **65** / 90% | 65.0% | Metabolic/vascular DAG evidence exists, but the exact human nerve-vascular pool and timing are unmeasured. |
| R2 | In retained R1 territories, βI/βII phosphorylation flux is above its 95th percentile for the same duration; D removes ≥50% of the excess above that boundary. | 40 / **75** / 95% | 48.8% | DAG-dependent activation is plausible and vessel-rich nerve fractions show βII changes. That does not establish activity in the specified human cells or exclude other activation routes. |
| R3 | In a retained R2 alternative, Q removes ≥50% of the defined low-availability burden. | 20 / **55** / 85% | 26.8% | Animal flow rescue and some human physiology support the connection, but human vascular tests disagree and actual neuronal availability is largely unmeasured. |
| R4 | A retained Q produces B ≥0.05. | 30 / **65** / 90% | 17.4% | There is functional rescue but no convincing human structural PKC result; normal-dose, subtype and prevention transfer remain uncertain. |
| R5 | In a retained R4 alternative, B_supply ≥0.05. | 40 / **75** / 95% | 13.1% | Once a selective vascular correction both improves supply and protects terminals, mediation is plausible. Other vascular outputs can still explain the protection. |
| R6 | A retained alternative has min(B, B_supply) ≥0.20. | 25 / **60** / 85% | 7.8% | The amount of structural preservation through this route has not been measured in humans. |
| R7 | At least one retained R6 group has C ≥0.10. | 30 / **65** / 90% | 5.1% | Common diabetes or vascular pathology does not establish common DAG-dependent, supply-mediated preventable loss. |

Merging R1 and R2 gives **48.75%**, preserving the joint probability. R4 and R5 differ: Q can preserve terminals through a nonsupply vascular effect. If B=0.33 but B_supply=0.03, R4 passes and R5 fails; if both are 0.12, R5 passes and R6 fails. Coverage is a separate denominator, not a second effect-size requirement.

## Evidence bearing on the judgment

### Exposure and animal intervention

[Xia et al. (1994)](https://pubmed.ncbi.nlm.nih.gov/8070612/) measured increased total DAG in diabetic dog retina and aorta, approximately 52% and 74% respectively over two to four months. This is direct metabolic evidence in other tissues, not the specified human nerve-vascular accessible pool. Independent animal counts were not recovered from the primary abstract.

[Yamagishi et al. (2003)](https://doi.org/10.1046/j.1471-4159.2003.02011.x) separated endoneurium from vessel-rich epineurium in STZ-diabetic mice with or without a human aldose-reductase transgene. Endoneurial PKC activity decreased in diabetic transgenics, while epineurial activity increased in both diabetic genotypes; endoneurial PKCα expression decreased and epineurial PKCβII increased. Fidarestat corrected the reported changes. The primary abstract establishes compartment-dependent direction; it does not isolate vascular cell types or provide a terminal-preservation effect.

[Nakamura et al. (1999)](https://doi.org/10.2337/diabetes.48.10.2090) found that four weeks of LY333531 prevented diabetic flow and nerve-function deficits despite **no detected whole-sciatic-nerve PKC activity increase or drug-related activity change**. This supports neither a universal neural PKC increase nor verified local target engagement. [Cameron and Cotter (2002)](https://doi.org/10.1002/dmrr.307) reported that two weeks of 10 mg/kg/day LY333531 corrected a roughly 50% endoneurial-flow deficit after eight weeks of STZ diabetes, together with conduction deficits. Low-dose combinations also improved these outcomes. Abstract-level data lack independent-animal structural counts and do not quantify human preventive magnitude; apparent combined rescue alone does not establish a statistically identified interaction.

Ruboxistaurin is not vascular-cell selective. [Hayase et al. (2007)](https://doi.org/10.1016/j.neulet.2007.02.040) acutely inhibited tetrodotoxin-resistant sodium currents in isolated diabetic DRG neurons. Reported n=10/8 refers to neurons, not demonstrated independent donor animals. [Tahara et al. (2006)](https://doi.org/10.1097/00001756-200604240-00026) reported rapid, reversible normalization of evoked calcium responses through mitochondrial buffering. These experiments supply a direct neural explanation for some functional effects and do not test terminal preservation.

### Human physiology is mixed

[Beckman et al. (2002)](https://doi.org/10.1161/hh0102.102359) found that pretreatment with LY333531 prevented the impaired forearm endothelial response produced by a six-hour hyperglycemic clamp in healthy people. This is an acute human perturbation, not early-DPN structural prevention; participant count was not recovered in the accessible primary abstract.

In contrast, a [2010 crossover trial](https://doi.org/10.1210/jc.2010-0286) in **13 people with T2D and 15 healthy controls** found no improvement in forearm resistance-vessel endothelial function after 14 days of ruboxistaurin. A [separate T2D trial](https://doi.org/10.1007/s10557-008-6144-5), with 52 randomized and 49 completing, reported a more favorable brachial flow-mediated-dilation change after six weeks: a 0.12-mm difference at five minutes, p=0.02, versus p=0.08 at one minute. These differ in vascular bed, stimulus and duration; neither measures oxygen at a peptidergic terminal. Full treatment tables were inaccessible.

### Neuropathy trials and what they actually measured

[Vinik et al. (2005)](https://doi.org/10.1016/j.clinthera.2005.08.001) randomized **205** participants to 32 mg/day (66), 64 mg/day (71) or placebo (68) for one year. **110 had T1D and 95 T2D**, with mean DPN duration 3.4 years. Primary vibration threshold and overall symptom changes did not differ. Symptom benefit appeared among 83 with higher baseline symptoms, and more positive results in a **post-hoc 50-person subgroup** with symptoms and retained sural responses. Those are plausible subgroup signals, not replicated peptidergic structural prevention.

[Casellini et al. (2007)](https://doi.org/10.2337/dc06-1699) randomized **20/20** for six months. The often-cited **78.2% endothelial-dependent and 56.4% C-fiber-mediated skin-flow improvements are within-ruboxistaurin changes**, not randomized between-group effects. The symptom score did favor drug: −66.0% versus −13.1%, p<0.03; a QOL symptom subscore also favored it. The study measured nerve-fiber morphometry and did **not establish improvement** in it or the remaining neurological measures. No numerical structural effect or equivalence interval was recoverable.

A [separate one-year substudy](https://doi.org/10.1016/j.jdiacomp.2007.07.002) included **nine ruboxistaurin and 11 placebo** recipients from a phase III trial. Neither skin-flow responses nor symptoms improved relative to placebo. Endothelium-dependent flow increases were 3.6-fold versus 8.6-fold; C-fiber-mediated increases were 1.7-fold versus 2.0-fold, with nonsignificant between-group comparisons. These results are not equivalence bounds or evidence of definite harm. The substudy is not independent of its parent trial.

[Boyd et al. (2011)](https://doi.org/10.1177/193229681100500326) analyzed complete QOL data for **54 people, 18 each on ruboxistaurin, topiramate and placebo**, drawing on previously reported treatment cohorts. Its ruboxistaurin structural null is explicitly attributed to Casellini's study, so this is not independent replication. Total QOL change was −9.56 versus −5.56: an approximate unadjusted difference of **−4.0 points, Welch 95% interval −15.0 to +7.0**, calculated from the published change SEMs. Within-arm significance does not establish that difference. Its “small-fiber” QOL domain is a questionnaire score, not terminal density. Topiramate's structural findings do not establish PKC-mediated rescue.

[Tesfaye et al. (2007)](https://doi.org/10.2337/dc07-0608) analyzed **262 placebo recipients from two phase III trials**, showing symptom and clinical-score improvement despite worsening of some electrophysiological measures. This is informative about endpoints but supplies **no drug-versus-placebo efficacy contrast**. The [MBBR registry record](https://clinicaltrials.gov/study/NCT00044421), checked for this assessment, lists a completed phase III study and enrollment of 400, with a progression composite primary endpoint and **no posted results**. Missing results must not be labeled known failures or successes.

These trials materially constrain broad ruboxistaurin treatment claims. They do not directly test the complete early-human DAG→vascular PKCβ→local supply→terminal preservation chain. The absence of that experiment is not grounds to disregard their negative structural or functional observations.

## Ideal experiments that would resolve each claim

These specify conceptual resolution with ideal human measurement and intervention capabilities. Independent people determine precision. For every entry, retain baseline groups and the same connected vascular/neural territories; resolve finite alternatives with simultaneous precision. Failed fidelity, imprecise boundaries or incorrect timing are unresolved, not biological NO.

**R1 — accessible DAG excess; no predecessors.** Follow a representative eligible cohort and matched nondiabetic references through year one. Measure native membrane sn-1,2-DAG mole fractions and accessibility in the specified vascular cells, while tracking associated terminal/axon/soma integrity. Assign groups at baseline; do not select later progressors. No causal treatment is required. YES requires a group/cell-site alternative with >95th-percentile exposure for ≥90 cumulative pre-loss days in ≥10% of person-weighted baseline neuronal territories. NO requires every alternative to fail a boundary. Whole-tissue lipid changes without localization cannot resolve it. Observation establishes exposure in diabetes, not a separate glucose-causation assertion.

**R2 — DAG contributes to excessive PKCβ flux; assume R1.** In the same early-human groups and vascular sites, measure βI/βII-specific native substrate phosphorylation flux over the exposure window. Randomize D versus sham while keeping glucose, circulating lipids, insulin, calcium inputs and other initiating exposures comparable; allow signaling feedback. Verify DAG sequestration, membrane integrity and lipid mass balance. Confirm direct DAG coupling with independently controlled C1 binding and physiological restoration, without changing unrelated kinase catalytic capacity. YES requires >95th-percentile flux for the same duration/territorial burden and D reducing ≥50% of its excess above that threshold. NO requires no retained alternative to satisfy both the activity and causal-reduction boundaries. Expression without flux, or PKC activation unaffected by faithful D, does not satisfy R2. Off-target inhibition or an interval spanning the reduction boundary is unresolved.

**R3 — PKCβ correction improves actual local availability; assume R1–R2.** Randomize the specified vascular Q alternatives versus sham before anticipated loss and measure connected interstitial oxygen/fuels throughout year one. Keep arterial content, pressure, medication and activity comparable; allow vascular tone, diffusion, permeability, trophic/inflammatory signaling and consumption to respond. Verify β-specific activity capping in vascular cells and absence of direct neuronal/glial drug effects. YES requires the defined reference availability deficit and ≥50% removal of its cumulative low-exposure burden by a retained Q. NO requires all alternatives to fail despite faithful correction. Increased stimulus-evoked skin flow without restored local availability is NO here. Insufficient target engagement or availability precision is unresolved.

**R4 — net structural causation; assume R1–R3.** Sustain retained Q or sham for five years from pre-loss baseline in randomized independent people. Keep other initiating exposures comparable while allowing every endogenous downstream response. Track actual terminal disappearances and successful replacements using stable subtype identity; estimate L and B with absolute counts. YES requires B≥0.05 for a retained alternative with positive reference loss; NO requires every retained B<0.05 despite verified Q. Pain relief, altered peptides or conduction changes alone are NO. Uncertain identity, negligible reference loss or an interval crossing 0.05 is unresolved.

**R5 — a structural supply component; assume R1–R4.** Add randomized Q+K and apparatus-control arms to the same five-year program. Draw K's matched reference-care supply profiles independently of future structural outcomes. Verify the assigned extracellular trajectories without directly altering intracellular repair, inflammatory receptors or ATP. Retain Q and all upstream matching; allow the full downstream response to each supply trajectory. Estimate B_supply. YES requires B_supply≥0.05 in a retained alternative; NO requires every retained value below 0.05. Persistent Q benefit when its supply improvement is removed can therefore fail R5. Inability to reproduce the reference exposure distribution, apparatus-induced injury or an interval across 0.05 is unresolved. This resolves the defined controlled component, not a unique natural mediation fraction.

**R6 — material preservation through this route; assume R1–R5.** Use the same randomized Q, Q+K and reference-care contrasts, five-year terminal endpoint, input matching and fidelity checks. Estimate both B and B_supply. YES requires min(B,B_supply)≥0.20 in a retained alternative; NO requires all alternatives to have at least one value below 0.20. Precise values B=.30/B_supply=.12 fail here despite passing earlier rows. Uncertainty crossing either boundary is unresolved. No additional execution mechanism is assumed.

**R7 — incident-case coverage; assume R1–R6.** Follow a representative reference-care cohort from the same population for five-year incident DPN. Assign the stated groups at baseline and estimate C among independent incident cases for every retained beneficial group. Do not use corrected arms to redefine the reference denominator or combine overlapping groups. YES requires at least one C≥0.10; NO requires all retained groups below 0.10. Incomplete case ascertainment or an interval across 0.10 is unresolved. Routine care and disease proceed without Q in this descriptive cohort.

## Uncertainty and sensitivity

Conditional vectors **[30,40,20,30,40,25,30]%**, **[65,75,55,65,75,60,65]%** and **[90,95,85,90,95,85,90]%** give **0.0216%, 5.0977% and 47.5349%**. Skepticism joins uncertain local DAG exposure with nonlimiting vascular changes and weak structural transfer. The favorable interpretation supposes an early, localized supply route diluted in late systemic trials. These are coordinated interpretations, not independently sampled uncertainty distributions.

A mainly functional interpretation gives **0.33%**, using [45,55,30,45,55,40,45]%; a strong early-supply interpretation gives **31.8%**, using [85,90,80,85,90,80,85]%. Existing evidence does not assign probability mass or a variance to these scenarios. The lower endpoint is best read as “very low,” not a calibrated frequency to four decimal places.

Hypothetical evidence raising R3 from 55% to 85% raises the total to **7.9%** if other conditional judgments remain fixed; lowering R3 to 20% gives **1.9%**. A credible human pre-loss exposure/availability result would move the judgment more than another within-group symptom improvement.

Changing the DAG catalog, β-isoform scope, duration, territorial burden or group definitions requires a new elicitation. Requiring only B≥20% without supply attribution is a broader hypothesis; its probability is not obtained by silently calling every vascular benefit ischemic. Relaxing materiality to 5% removes R6 rather than retaining an extra discount. No numerical threshold-response evidence supports tighter alternate-definition estimates.

## Overlap and implication

This is a specific molecular contributor to the supply-injury family in [15](15_microvascular_ischemia.md), upstream of possible [energy failure](07_bioenergetic_insufficiency.md). The intervention and mediation definitions differ, so the two numerical propositions are not exact set-theoretic subsets. They share evidence and cannot be added as independent causes. [Polyol](01_polyol_flux.md), [RAGE](04_rage_signaling.md) and [oxidative injury](39_oxidative_nitrosative_injury.md) can interact with vascular PKC. Direct neuronal calcium/channel actions belong with [31](31_hyperexcitability_calcium.md), not as proof of vascular mediation.

The β-inhibitor evidence supports biologically active neurovascular pharmacology. It gives substantially less support to a common DAG-dependent route that prevents material human peptidergic terminal loss through restored supply.