# 14. Insufficient target-derived NGF/TrkA support

**Best judgment: about 15% (16.1% by multiplication); interpretation sensitivity: 0.8–62%.** These are subjective probabilities of the causal proposition, not effect sizes. The range joins coherent skeptical and favorable interpretations; it is not a confidence or credible interval. 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, a deficiency of target-derived mature NGF signaling through neuronal TrkA precedes distal terminal loss. Selectively restoring that input to its physiological level would prevent at least **20% of five-year net peptidergic C-fiber terminal loss**, in at least one baseline-describable group accounting for **10% of reference-care incident DPN cases**.

Reference care continues without the selective correction. Incident DPN means new bilateral, length-dependent clinical signs accompanied by objective small-fiber loss or abnormal nerve conduction, excluding another sufficient explanation. The structural endpoint concerns identified peptidergic C-fiber endings in distal-leg/foot skin. Their identity must be established independently of intervention-induced changes in TrkA, substance P or CGRP expression. A surviving soma, a changed pain threshold and a longer branch are different outcomes.

For each person, `L = (terminal disappearances − successful new terminals)/baseline terminal number` over five years. A new ending that subsequently disappears contributes to both counts. Average sampled territories within each person; people are the inference unit. For a group, `B = (mean L_reference − mean L_corrected)/mean L_reference`, requiring positive reference net loss. Let `C = P(group | incident DPN under reference care)`. The complete proposition requires B ≥0.20 and C ≥0.10 in the **same** group and connected neuronal pathway.

Consider the whole eligible population and these separate baseline groups: HbA1c ≥7.5%; intact mature NGF in distal skin interstitial fluid below the matched nondiabetic fifth percentile; or local productive TrkA input below that percentile. Membership precedes intervention and outcome. No group is defined by later pain relief or response to NGF. These thresholds, group family and time window are chosen definitions, not observed biological constants.

### What the deficit and correction mean

The input is **productive signaling from mature NGF-bound TrkA complexes originating at skin terminals**, including their signaling endosomes. Low skin NGF, low NGF mRNA, precursor abundance, receptor expression and weak fold induction after a laboratory pulse are not interchangeable measurements of this input. Ligand provenance, processing and endosomal destination matter.

Measure productive TrkA-mediated activation of immediate SHC/FRS2/PLCγ signaling machinery as molecule flux per neuronal volume and time, distinguishing local from delivered endosomal signaling. A candidate deficit is input below the matched nondiabetic fifth percentile for ≥90 cumulative days in year one, before structural disappearance in tracked terminals. Matching includes age, sex, skin site and neuronal subtype. Candidate compartments are the terminal, its connected axon/endosomal route and its soma. A deficit in any one or their combination is allowed; requirements cannot be assembled from unrelated cells or groups.

**Q** restores the abundance, location, lifetime and native signaling pattern of these mature-NGF/TrkA complexes to the matched nondiabetic median profile when input is below the fifth percentile. It supplies ordinary receptor signaling, with its physical material and ATP costs, without driving above that physiological target or selecting only the downstream outputs subsequently found to help. Correction at each candidate compartment and their joint correction are the finite alternatives. These conceptual interventions require capabilities beyond present delivery methods.

Q can compensate for reduced mature ligand presentation or defective delivery of its signaling complexes. It does not independently repair all axonal transport, activate Akt regardless of receptor input, replace VEGFA or suppress p75/sortilin signaling. Normal TrkA input with a separate downstream lesion is outside this deficiency claim. Other trophic inputs remain available; their compensation and Q's downstream effects on growth, firing, metabolism and terminal turnover are allowed to unfold. A physiological input can fail to help or increase loss.

## Biological reasoning

A target cell must synthesize, process and release NGF where a terminal can bind it. Signaling can act locally and after transport toward the soma. Increased transcript abundance can coexist with poor processing, and abundant receptor with inadequate productive input. Denervation can itself change the tissue producing or consuming NGF.

Adult dependence concerns terminal maintenance and renewal, not necessarily developmental neuronal survival. Other growth factors can compensate. Extra NGF can change peptide expression and nociceptive sensitivity without adding persistent endings. Increased branching may add successful replacements, unstable sprouts or no additional epidermal crossings. The net balance must therefore be measured structurally.

TrkA's growth and excitability signals coexist with receptor feedback and protein turnover. The causal comparison allows all consequences of physiological restoration, including injury from sensitization. Pain need not remain unchanged for structural preservation to count; pain improvement alone does not count as preservation.

## Probabilities of the necessary claims

Each row after R1 is conditional on **all preceding rows**, retaining their eligible groups and correction alternatives. These are additional claims about the same path, not independent marginal probabilities. Their conjunction is the complete proposition.

| Claim | Exact additional requirement | Skeptical / best / favorable | Best cumulative | Evidence and reason for uncertainty |
|---|---|---:|---:|---|
| R1 | At least one specified baseline group has the defined pre-loss deficiency of productive target-derived mature-NGF/TrkA input in a connected candidate compartment. | 25 / **55** / 85% | 55.0% | Human skin observations support relevance, but precursor, transcript and active input differ. Early diabetic/obese mice can instead have increased signaling. No human longitudinal compartment-specific activity measurement settles the sign. |
| R2 | For at least one R1 group/correction, Q produces B ≥0.05 despite remaining trophic support and all downstream consequences. | 40 / **75** / 95% | 41.3% | Adult NGF sequestration reduces epidermal innervation; local NGF-R100W has structural benefit in diabetic mice. Negative NGF rescue, context dependence and the human trial constrain transfer. |
| R3 | At least one retained R2 group/correction has B ≥0.20. | 25 / **60** / 85% | 24.8% | Animal structural effects make substantial benefit plausible once the deficit is causal. They do not measure the five-year human prevention fraction; mixed growth and replacement endpoints leave magnitude broad. |
| R4 | At least one retained R3 group accounts for C ≥0.10. | 30 / **65** / 90% | 16.1% | Anatomical TrkA relevance does not establish how many incident cases have this correctable deficit. Baseline local activity data are missing. |

R2 prices compensation, structural causation and opposing consequences jointly. R3 distinguishes a small structural effect from the material effect: B=0.12 satisfies R2 and fails R3. B=0.30 in a group covering 3% of reference cases satisfies R3 and fails R4. Merging R1 and R2 gives 41.25% and leaves the joint probability unchanged.

## Evidence bearing on the judgment

### Human observations and interventions

[Anand et al. (1996)](https://doi.org/10.1038/nm0696-703) associated depleted skin NGF and substance P with early small-fiber dysfunction and reduced axon-reflex vasodilation. The accessible primary abstract does not give participant counts, diabetes-type breakdown or a structural intervention contrast; these are not inferred. It establishes neither a preceding deficiency nor productive neuronal TrkA input.

Molecular form changes the interpretation. [Yiangou et al. (2002)](https://doi.org/10.1046/j.1529-8027.2002.02024.x) reported a **53-kDa precursor-like band** in six people with subclinical diabetic neuropathy versus six controls. Median optical intensities were 1.5 and 52: a 97.1% relative reduction in this band, not in mature NGF activity. Conversely, [Diemel et al. (1999)](https://doi.org/10.1046/j.1464-5491.1999.00035.x) measured NGF mRNA in 19 diabetic participants and eight controls: means 5.12 versus 1.57, about 3.26-fold higher in diabetes. Thermal deficits did not correlate with mRNA. Altered processing or compensation could reconcile these observations; neither resolves R1. Primary abstracts supply these numbers; full participant-level and processing measurements were not recovered.

The phase II randomized trial included **250 symptomatic patients** treated for six months. Three prospectively identified multiple-endpoint analyses favored rhNGF (p=0.032, 0.008 and 0.005), involving clinical/sensory outcomes rather than a reported terminal-density rescue. [Apfel et al. (1998)](https://doi.org/10.1212/WNL.51.3.695)

Phase III randomized **504 to rhNGF and 515 to placebo**, using 0.1 μg/kg subcutaneously three times weekly for 48 weeks. Approximately 74% had type 2 diabetes, with diabetes for about 15 years and neuropathy symptoms for five years. The primary NIS-LL comparison failed (p=0.25). Improved categories were 159/504 versus 175/515: **−2.4 percentage points**, with a descriptive unadjusted 95% interval of −8.2 to +3.3. Regimen completion was 82.9% versus 89.5%; injection-site pain was common with NGF. There was no terminal-structure endpoint or verified local TrkA correction. This weighs against broadly effective replacement in established DPN, while leaving early selective structural prevention incompletely tested. [Apfel et al. (2000)](https://doi.org/10.1001/jama.284.17.2215)

Blocking NGF can improve pain: the randomized [tanezumab DPN study](https://doi.org/10.1111/pme.12677) favored active treatment on eight-week pain reduction, after the assessment time was changed following a trial hold. Unchanged neuropathy assessments do not establish unchanged five-year terminal structure. Analgesia from blockade and possible structural support from restoration can coexist.

### Adult and diabetic structural comparisons

[Bennett et al. (1998)](https://doi.org/10.1046/j.1460-9568.1998.00139.x) sequestered NGF with local TrkA-IgG for 10–12 days in adult rat saphenous skin. Epidermal PGP9.5 innervation density decreased **44%**. This is adult structural evidence beyond developmental dependence. The abstract's 152 nociceptors are recording units, not independent animals; the structural animal count and uncertainty were not recovered. Severe withdrawal in healthy rat skin does not identify the diabetic human exposure range.

[Christianson et al. (2003)](https://doi.org/10.1016/S0014-4886(02)00017-1) found reduced cutaneous innervation in STZ-diabetic mice. Two weeks of intrathecal **NGF increased branching but did not improve innervation density**; GDNF and neurturin improved both. Qualitative presence of skin neurotrophin mRNA did not establish normal ligand supply. Full arm-level counts and effects were inaccessible; the reported NGF null is not an equivalence bound. Neurite growth cannot automatically be called rescued skin innervation.

The [2025 NGF-R100W/VEGFA study](https://doi.org/10.2337/db24-0989) treated male db/db mice from 16 to 22 weeks, using plantar mRNA delivery with both monotherapies and their combination among five diabetic comparator arms. Figure 5 reports **n=6 per structural group**, with PGP9.5 density increased by NGF-R100W and more strongly by combined treatment; blood-flow comparisons use n=10. Full text and captions were recovered, but numerical density coordinates were not, so a precise rescue fraction is not supplied. This supports repair capacity and interaction, without isolating restoration of an observed endogenous deficit, native NGF signaling or prospective human prevention. A combination exceeding monotherapies is not by itself quantified synergy. The same group's earlier mRNA study concerned **paclitaxel neuropathy**, not diabetes. [Yu et al. (2023)](https://doi.org/10.1002/adhm.202202127)

### Opposing effects and interpretation limits

In young db/db mice, [Cheng et al. (2011/2012)](https://doi.org/10.1016/j.nbd.2011.08.011) found transiently increased epidermal innervation during mechanical allodynia. Anti-NGF from six to eight weeks reduced this increase, without changing controls' density. There were **four mice per group**, with sections averaged per mouse. TrkA-positive fibers rose about 1.5-fold at eight weeks and over twofold at ten weeks, then approached control levels. Total PGP9.5 density also rose, so peptide induction is not the sole possible explanation. Tiny reported p-values should not imply large independent n: an exact pairwise rank test at n=4/4 cannot yield two-sided p<0.0286. The article also reports across-age ANOVA; its individual tests cannot be reconstructed from summaries.

A [2025 diet-induced-obesity study](https://doi.org/10.1016/j.isci.2025.112047) found heightened NGF-associated sensory responses in ear skin before later axon loss. Thirty-minute anti-NGF or PI3K-inhibitor exposure reduced capsaicin-evoked calcium responses. **It did not test sustained NGF blockade against later structural loss.** Its 30-week IENFD comparison reports n=4 per group and acute inhibition panels n=6; fields cannot inflate animal n. Insulin increased keratinocyte NGF expression, opposite to a universal deficiency model.

[Danos et al. (2025)](https://doi.org/10.1091/mbc.E25-01-0005) found that acute NGF stimulation reduced axonal NMNAT2/STMN2 and accelerated post-axotomy fragmentation. These were E13.5-derived mouse sensory cultures at DIV7–8; addback followed deprivation with Bcl-xL supporting survival. The axotomy protein and fragmentation panels each report four preparations. This identifies an opposing maintenance consequence under specific injury conditions, not degeneration of intact adult diabetic terminals. It does not establish that inhibitory matrix removes NGF's growth benefit while retaining that cost.

The [2025 plantamajoside study](https://doi.org/10.1111/jcmm.70571) provides weak mechanistic transfer: its pathway-blockade comparison used an immortalized Schwann cell line at **400 mM glucose**, about 7,206 mg/dL, without an identified osmotic-control comparison. Mouse nerve/behavioral effects do not measure peptidergic terminal rescue under selective NGF correction. A [2026 NGF nanozyme study](https://doi.org/10.1002/adhm.71541) combines NGF with glucose/ROS/oxygen effects and reports late diabetic **sudomotor** repair; its primary abstract does not isolate this route or target fiber.

## Ideal experiments that would resolve each claim

These define truth conditions using ideal measurement and manipulation, rather than propose practical experiments. Independent people are the inference unit. Resolve the finite candidate family with simultaneous precision; a failed manipulation or an interval crossing a boundary is **unresolved**, not a biological NO.

**R1 — pre-loss productive input deficiency.** Follow a representative early-T2 cohort and matched nondiabetic controls from baseline through year one, before and during tracked terminal loss. Trace target-derived mature ligand separately from precursors and measure productive TrkA input at terminals and their connected signaling endosomes/somata, retaining baseline group membership. No intervention is required; routine-care inputs are observed, not mistaken for randomization of diabetes. YES requires at least one specified group/compartment with below-fifth-percentile input for ≥90 cumulative days before loss. NO requires all specified alternatives to fail that exposure/timing boundary. A deficit appearing only after loss, or low precursor abundance with normal productive input, is NO for R1. R1 does not independently claim that glucose rather than another feature of diabetes caused the deficit.

**R2 — structural consequence despite compensation.** Assuming R1, randomize qualifying groups to each specified Q or matched sham before the anticipated loss and sustain correction for five years. Verify the prescribed local mature-NGF/TrkA input. Keep upstream glycemia, lipids, insulin exposure, administered growth factors, mechanical load and non-NGF transport interventions comparable. Allow downstream firing, endogenous trophic feedback, intracellular signaling and growth to change. Track terminal identity, disappearances and successful replacements; record pain and peptide expression separately. YES requires B ≥0.05 for at least one retained alternative with positive reference loss. NO requires every retained alternative to have B <0.05 after successful correction. More staining, pain relief or branching without that structural benefit is NO. Failure to establish the B boundary or positive reference loss leaves this fraction unresolved.

**R3 — material preservation.** Assuming R1–R2, use the same randomized contrasts, connected compartments, five-year endpoint, input controls and freely responding downstream pathways. Resolve B in each retained group. YES requires B ≥0.20 for at least one; NO requires all to remain below 0.20 despite the established smaller benefit. This is a magnitude decision, not a second causal mechanism. Uncertainty crossing 0.20 remains unresolved.

**R4 — coverage of incident cases.** Assuming R1–R3, follow a representative reference-care cohort from the same eligible population for five-year incident DPN, applying the stated clinical definition and baseline group rules. The corrected randomized arms establish benefit but do not define the reference-case denominator. Count independent incident cases in each retained beneficial group without double-counting overlapping groups. YES requires C ≥0.10 for at least one; NO requires every retained group to cover less than 10%. Defining membership by post-treatment response is invalid. An interval crossing 0.10 remains unresolved. Reference care fixes the denominator while allowing disease to follow its course without Q.

## Uncertainty and quantitative sensitivity

The skeptical conditional vector is **[25,40,25,30]%**, the best **[55,75,60,65]%**, and the favorable **[85,95,85,90]%**. Their products are **0.75%, 16.09% and 61.77%**. The skeptical interpretation combines poor correspondence between measured NGF and productive input with limited adult diabetic rescue. The favorable interpretation emphasizes adult terminal dependence and local structural replacement, supposing a common early deficit missed by late systemic treatment.

Two intermediate coherent interpretations locate the uncertainty: mainly functional/marker effects give **4.3%** with [45,45,35,60]%; a genuine early local deficit with transferable repair gives **38.3%** with [75,85,75,80]%. These are scenarios, not draws from a fitted distribution. Their endpoints carry no assigned probability mass; a variance or sampling confidence interval for belief is unjustified.

A longitudinal human result establishing pre-loss productive input deficiency could change R1 from 55% to 80%, raising the whole probability to **23.4%** if remaining conditional judgments stayed fixed. Evidence that mature input is usually preserved could reduce R1 to 25%, giving **7.3%**. These are hypothetical updates, not study-derived effects.

Definition sensitivity is separate. Lowering materiality from 20% to 5% merges R2 and R3; it does not justify retaining the same R3 discount. Changing the deficit percentile, duration, group family or case threshold needs a new elicitation. Present data do not identify numerical probabilities for those alternate definitions. Near-zero reference loss makes B unstable even when absolute growth increases; absolute disappearance and replacement counts must accompany it.

## Overlap and implication

This route overlaps [axonal transport](26_axonal_transport.md), [skin repair](32_skin_repair_environment.md), [insulin/IGF support](12_insulin_igf_trophic_deficit.md), [microvascular supply](15_microvascular_ischemia.md) and [SARM1](27_sarm1_axon_destruction.md). Their probabilities cannot be added. The VEGFA combination informs vascular interaction; the NMNAT2 finding connects a growth input to execution under a different injury condition.

Adult structural NGF dependence and local diabetic repair are credible broader findings. The fragile link is an early, common, deficient **native signaling input** in the intended human fibers whose physiological restoration yields durable net preservation. Evidence supports that possibility, not a universal instruction to increase NGF.