Open questions — neuropathic pain¶
Last curated: 2026-10-21 (periodic sweep: OQ-4, OQ-5, OQ-7 and OQ-23 sharpened; D4 updated; none opened or closed). Previously 2026-09-30 (OQ-3, OQ-4, OQ-5, OQ-7, OQ-9 and OQ-16 sharpened; OQ-23 opened). Stable IDs persist when questions are sharpened; closed questions move to a closing section.
Priority rule¶
Tier 1 questions could change practice and are designable now. Tier 2 questions are important but depend on methods, cohorts or a Tier-1 result.
Dots not yet connected¶
| # | Dot A | Dot B | The missing junction | Powers |
|---|---|---|---|---|
| D1 | QST sensory clusters (PMID 16697110) | Similar average efficacy across OPTION-DM pathways (PMID 36007534) | No prospectively powered phenotype × drug interaction trial | OQ-1 |
| D2 | Reversible QST phenotype after decompression (PMID 33769367) | Central plasticity framework (PMID 19712899) | No study quantifies when peripheral-source removal ceases to reverse central amplification | OQ-2 |
| D3 | NaV channel causal genetics/mechanisms (PMID 15362153) | Weak broad translation of sodium modulators (PMID 33899639) | No enrichment design joins genotype/excitability to selective therapy | OQ-3 |
| D4 | Persistent CIPN at ≥6 months, 30% (PMID 25261162); chronic CIPN 40% at ≥24 months in a 2026 synthesis (PMID 42713463) | Candidate neurofilament marker (PMID 39242335); plasma pNF-H showed no association with oxaliplatin CIPN in 30 patients (PMID 42873277) | No validated baseline marker predicts painful persistence and safe cumulative dose | OQ-4 |
| D5 | SCS randomized efficacy in painful diabetes (PMID 33818600), now with randomized six-month sensory and nerve-fiber gains (PMID 42713876) | Very-low-certainty long-term implanted evidence (PMID 34854473) | No complete five-year comparison counts infection, revision, explant, medication and cost | OQ-5 |
| D6 | Patient priorities include sleep/function/work (PMID 25354872) | Trials emphasize mean pain intensity (PMID 23042476) | Treatment rankings have not been recalculated under patient-weighted outcomes | OQ-6 |
| D7 | PHN is vaccine-preventable (PMID 15930418) | Wide global PHN risk definitions (PMID 24916088) | No harmonized contemporary preventable-burden estimate | OQ-7 |
| D8 | DRG immune cells initiate/persist pain (PMID 31937758) | Immune programs also resolve pain (PMID 36859719) | Human phase-specific cell-state marker and timed intervention are absent | OQ-8 |
| D9 | Updated SCI neuropathic-pain prevalence is 57% but varies by classification system (PMID 41043376) | CanPainSCI added screening and diagnosis recommendations using GRADE (PMID 35124700) | No representative implementation study tests whether standardized classification changes recognition, treatment or function | OQ-19 |
| D10 | Significant CIPN symptoms were associated with fall/near-fall hazard ratio 2.67 (PMID 27183099) | Exercise consensus recommends etiology-specific rehabilitation despite heterogeneous evidence (PMID 34901069) | No adequately powered fall-prevention trial enrolls on emerging sensory symptoms and measures injuries rather than pain alone | OQ-20 |
| D11 | Forty-one percent of central post-stroke pain begins 1–12 months after stroke (PMID 32451951) | Rehabilitation pathways concentrate assessment earlier | No randomized or implementation comparison tests scheduled sensory surveillance through 12 months | OQ-21 |
| D12 | Chinese exercise consensus recommends exercise for neuropathic pain (PMID 34901069) | PEER judges neuropathic-pain exercise benefit unclear (PMID 35292455) | Etiology-specific responder, function and harm data have not been synthesized under one certainty framework | OQ-22 |
Tier 1¶
OQ-1. Can sensory phenotype select treatment?¶
Retrospective associations exist, but clinical utility requires prespecified stratification, a treatment-interaction test and external replication (PMID 24670811; PMID 36198371).
OQ-2. When is neuropathic pain still dependent on peripheral input?¶
Central sensitization can outlast input, while decompression changes phenotype; no validated perturbation test quantifies dependence (PMID 19712899; PMID 33769367).
OQ-3. Can sodium-channel treatment be rescued by enrichment?¶
Channel biology is strong, but subtype selectivity, target engagement and trial phenotype remain barriers (PMID 15362153; PMID 33899639). Sharpened 2026-09-30: the selective NaV1.8 inhibitor suzetrigine (PMID 41481839) is in two placebo-controlled phase 3 trials in painful diabetic neuropathy (NCT06628908, NCT07231419; estimated completion 2027) after a phase 2 study without a placebo arm (NCT05660538). Those trials enroll by diagnosis, so they test subtype selectivity without enrichment; a null result would not answer this question and a positive one would weaken the case that enrichment is necessary. A human perception-threshold-tracking protocol that separated topical lidocaine and mepyramine, but not phenytoin, from placebo in 20 healthy participants is a candidate target-engagement readout but is untested in patients (PMID 42751727).
OQ-4. Can persistent CIPN be predicted before irreversible exposure?¶
CIPN remains in about 30% at ≥6 months, but proposed axonal markers lack predictive clinical validation (PMID 25261162; PMID 39242335). Sharpened 2026-09-30: a 197-study synthesis estimates chronic CIPN at 44.5% overall and 40.4% at ≥24 months after chemotherapy, with no decline after the first year (very low certainty; PMID 42713463), so the target population for prediction is larger and more persistent than the 2014 figure implied. Neither estimate isolates painful CIPN, which remains the outcome a predictive marker would need to forecast. Sharpened 2026-10-21: the marker evidence divides by target. Plasma phosphorylated neurofilament-heavy chain was not associated with onset, severity or persistence of oxaliplatin CIPN in 30 patients, which its authors attribute to dorsal-root-ganglion rather than axonal injury (PMID 42873277), while a preprint meta-analysis reports a moderate correlation of circulating neurofilament light with CIPN severity (r=0.456, 95% CI 0.341–0.558; low GRADE certainty) and concludes that no stand-alone predictive threshold is supported (Luo 2026, Research Square doi:10.21203/rs.3.rs-10882446/v1, preprint, not peer reviewed). Correlation with concurrent severity is not prediction before exposure. A 901-person cross-sectional study adds clinical predictors that any marker must beat: female sex for dose-limiting CIPN and older age and diabetes for patient-reported neuropathy after treatment (PMID 42814908).
OQ-5. What is the long-term net benefit of implanted stimulation?¶
Short-term selected-population efficacy must be balanced against infection, revision, explant and cost over years (PMID 33818600; PMID 34854473). Sharpened 2026-09-30: PDN-Sensory replicated the six-month pain response and added allocation-blinded gains in neuropathy score and intraepidermal nerve-fiber density (PMID 42713876), so the benefit side of the ledger may include nerve function, not pain alone. The question now has three unmet parts: a sham or otherwise blinded pain comparison, randomized follow-up beyond the six-month crossover point, and ulcer or amputation endpoints rather than surrogate sensory measures. Sharpened 2026-10-21: the cost side has been modeled but not measured. A deterministic three-year model at US Veterans Health Administration contract prices put SCS device costs at $19,723 more per patient than magnetic peripheral nerve stimulation, using revision (34%) and explant (53%) prevalences from a single published 100-patient series and setting costs beside each therapy's published results rather than estimating cost per outcome (PMID 42836090). Neither therapy has been compared with the other in a trial, so the model frames the question rather than answering it.
OQ-6. Do patient-weighted endpoints reorder treatment choices?¶
Qualitative priorities extend beyond pain, while most trials optimize pain intensity (PMID 25354872; PMID 23042476).
OQ-7. How much PHN is preventable now?¶
Vaccination works, but age, product, immune status and PHN definitions impede a current global estimate (PMID 15930418; PMID 36098300). Sharpened 2026-09-30: recombinant-vaccine efficacy against PHN in immunocompetent adults ≥50 is now graded at 91.2% (moderate certainty) and 87.5% at 11 years (low certainty) (PMID 42748460), against a 2024 pooled-RCT analysis that found no significant PHN difference (PMID 39399936). A phase 3 trial of a second recombinant vaccine counted 24 placebo PHN cases under its own definition and 8 under a ≥90-day definition (PMID 42733065). The preventable-burden estimate therefore depends first on fixing the PHN definition, then on coverage and immune-status strata. Sharpened 2026-10-21: the immune-status stratum is now quantified as a large gap, not a caveat. In immune-mediated inflammatory diseases, recombinant-vaccine effectiveness against PHN was 33.6% (95% CI 11.7–50.0), against zoster 59.0% in real-world studies versus 90.5% in trials (PMID 42896729). A transportable estimate needs separate strata at least for immunocompetent older adults and treated IMID populations.
OQ-8. Can neuroimmune therapy be timed to harmful rather than resolving states?¶
Macrophage/glial pathways can initiate, maintain and resolve pain, demanding state-specific human markers (PMID 31937758; PMID 36859719).
OQ-19. Does standardized SCI classification improve outcomes, not just prevalence precision?¶
The updated meta-analysis pooled prevalence at 57% (95% CI 51–64; I²=96.2%) and found that classification system significantly altered the estimate, while the updated CanPainSCI process added three screening/diagnosis recommendations (PMID 41043376; PMID 35124700). A cluster-randomized implementation trial could compare structured classification plus treatment linkage against usual rehabilitation assessment; the primary endpoint should be participation or sleep at 6–12 months, with recognition rate, false-positive referrals and medication/device exposure as co-endpoints.
OQ-20. Can symptom-triggered balance intervention prevent CIPN falls?¶
In a prospective cohort of 116 people beginning taxane or platinum chemotherapy, significant numbness/tingling was associated with a 2.67-fold fall-or-near-fall hazard (95% CI 1.62–4.41), but the hypothesis was generated after data collection and 94% of participants were female (PMID 27183099). A pragmatic trial should trigger gait/balance assessment and strength training at a prespecified symptom threshold, stratify by agent and baseline fall risk, and use injurious falls per person-time—not neuropathy score alone—as the primary outcome.
OQ-21. Should stroke pathways screen for central pain through 12 months?¶
Central post-stroke pain pooled prevalence was 11% (95% CI 7–18); among affected patients, 41% developed it between one month and one year and 5% after one year (PMID 32451951). A stepped-wedge comparison of scheduled somatosensory/pain assessment at 1, 3, 6 and 12 months versus symptom-initiated review could test time to correct diagnosis, inappropriate analgesic exposure, pain interference and participation while measuring the burden of false-positive screens.
OQ-22. Is exercise efficacy etiology-specific or obscured by pooled “neuropathic pain”?¶
One expert consensus used 8 systematic reviews and 21 RCTs to recommend exercise across ten etiologies, whereas the PEER guideline judged evidence of exercise benefit unclear for neuropathic pain (PMID 34901069; PMID 35292455). Resolution requires a prospectively registered individual-participant-data synthesis separating diabetic, chemotherapy, radicular, SCI, stroke, MS, HIV and postsurgical populations, with pain responders, physical function, falls and adherence analyzed jointly.
Tier 2¶
- OQ-9. What representative prevalence meets probable/definite 2016 grading rather than questionnaire positivity? (PMID 24291734; PMID 27115670) Sharpened 2026-09-30: in a grade-stratified multiple-sclerosis meta-analysis, 26 of 32 papers reached only “possible”, one “probable” and five “definite” (PMID 42678174)—the gap is measurable even in a single disease.
- OQ-10. Which combination sequence retains benefit beyond 16 weeks? (PMID 36007534; PMID 23732189)
- OQ-11. Can a multimodal biomarker outperform examination and lesion confirmation externally? (PMID 35659993; PMID 20642627)
- OQ-12. Which acute intervention adds PHN prevention beyond vaccination? (PMID 38050854)
- OQ-13. Can procedure-specific nerve mapping prevent chronic postsurgical neuropathic pain? (PMID 23273105)
- OQ-14. What rTMS protocol and maintenance schedule gives durable benefit? (PMID 34196698; PMID 34826512)
- OQ-15. How should mixed central, peripheral and nociplastic mechanisms be represented in trials? (PMID 28666966; PMID 32584191)
- OQ-16. Can guideline panels reconcile different placements of pregabalin and topical therapy? (PMID 25575710; PMID 32276788) Sharpened 2026-09-30: the 2025 NeuPSIG revision keeps α2δ-ligands as a class first line and places capsaicin and lidocaine second line (PMID 40252663), so the disagreement with the French placement of pregabalin persists after the update.
- OQ-17. Which rehabilitation program reduces falls and restores work in CIPN? (PMID 35149899)
- OQ-18. Can renal-dose and fall-risk controls reduce gabapentinoid harm without undertreating pain? (PMID 30673120)
- OQ-23. (opened 2026-09-30) Is the hierarchical 2016 grading the right reference standard? A 2026 position paper proposes a convergence model giving history, pain characteristics and examination equal weight and treating lesion tests as etiological rather than classificatory (PMID 42550034); a published comment disputes it (PMID 42708623), and the reply concedes boundary-case misclassification while noting that the accuracy of the existing system as a whole has never been formally established (PMID 42765589). No head-to-head evaluation of the two frameworks exists. Sharpened 2026-10-21: a third 2026 proposal questions the importance given to demonstrating a lesion, the specificity of sensory descriptors and the imprecision of “central sensitization”, and proposes distinguishing large- from small-fibre neuropathic mechanisms and a “sensory nociplastic pain” category marked by diffuse, variable neuropathic-like symptoms (PMID 42835403). A test of the grading system now has to cover the boundary between neuropathic and nociplastic pain, not only the possible/probable/definite boundaries. PubMed and Europe PMC searches on 2026-10-21 found no NeuPSIG revision of the 2016 system.
Minimum decisive designs¶
These are design specifications, not claims that a single protocol is mandatory. They state what evidence would materially narrow each question and what result would count against the working premise.
| ID | Population and allocation | Primary decision endpoint | Minimum follow-up | Result that would narrow or close the question |
|---|---|---|---|---|
| OQ-1 | Probable/definite peripheral neuropathic pain; phenotype-stratified randomization across mechanistically distinct drug classes | Prespecified phenotype × treatment interaction on ≥30% pain response plus function | 16 weeks | Replicated interaction with calibration in an external cohort, or precise exclusion of a clinically useful interaction |
| OQ-2 | Persistent pain after a treatable peripheral lesion; randomized diagnostic perturbation or source-directed treatment | Sustained change in pain interference and evoked-pain area | 6 months | A validated perturbation threshold predicting durable response, or evidence that the test adds no value over lesion/examination data |
| OQ-3 | Genotype- or excitability-enriched peripheral pain; selective sodium-channel modulator versus placebo/active class | ≥30% response with target-engagement confirmation | 16 weeks | Reproduced enrichment interaction, or adequate target engagement with no clinical separation |
| OQ-4 | Patients starting neurotoxic chemotherapy; prospective biomarker model with external validation | Painful CIPN at ≥6 months and dose-limiting toxicity | 12 months | Calibration/discrimination sufficient to alter exposure decisions without excess cancer-treatment undertreatment |
| OQ-5 | Device-eligible painful neuropathy; implanted stimulation versus optimized non-device care | Patient-weighted net benefit including pain, function, serious harm, revision and explant | 5 years | Durable positive net benefit under realistic attrition/crossover, or confidence interval excluding the prespecified benefit threshold |
| OQ-6 | Broad neuropathic-pain trial archive with patient-derived outcome weights | Rank stability across pain-only versus patient-weighted benefit-harm utility | Trial duration plus sensitivity analyses | Stable rankings across weights, or reproducible clinically important reordering |
| OQ-7 | Age-, immune-status- and vaccine-stratified population cohorts | Prevented PHN cases per 100,000 person-years using a harmonized ≥90-day definition | 3 years | Transportable preventable-burden estimates with measured coverage and breakthrough disease |
| OQ-8 | Serial human blood/CSF/skin sampling across acute-to-persistent neuropathic pain | Reproducible phase-specific immune state linked to prognosis and intervention response | 12 months | External replication and treatment interaction, or failure of candidate states to predict beyond clinical variables |
| OQ-9 | Representative population sample assessed with 2016 grading | Probable/definite prevalence with non-response weighting | Cross-sectional plus confirmation | Narrow, transportable prevalence intervals and quantified questionnaire misclassification |
| OQ-10 | Partial responders after standardized monotherapy; randomized sequence/combination pathways | ≥50% response, function and withdrawal for harm | 12 months | One pathway’s sustained net benefit or precise equivalence within a prespecified margin |
| OQ-11 | Multi-etiology diagnostic cohort; locked multimodal model versus expert clinical assessment | External discrimination, calibration and net benefit | 12 months | Clinically useful incremental net benefit, or no gain after optimism correction |
| OQ-12 | Acute zoster despite contemporary vaccination; intervention versus placebo | PHN at ≥90 days with acute harms | 6 months | Reproducible relative and absolute risk reduction beyond vaccine protection |
| OQ-13 | Procedure-specific surgical cohorts; mapped nerve-preservation strategy randomized by center | Probable/definite neuropathic pain at 12 months | 12 months | Reduced neuropathic pain without worse oncologic/surgical outcomes |
| OQ-14 | Central or peripheral phenotype-specific rTMS; factorial target/frequency/maintenance design | Pain interference and ≥30% response after treatment cessation | 6 months | Protocol-specific durable benefit or exclusion of the minimum important effect under credible sham |
| OQ-15 | Mixed-mechanism cohorts with lesion-confirmed and nociplastic features measured independently | Model fit, treatment interaction and patient-level calibration | 12 months | A representation that improves prognosis/treatment selection over mutually exclusive labels |
| OQ-16 | Independent guideline panels using one evidence-to-decision framework | Concordance after harmonized certainty, costs and preferences | One update cycle | Residual differences explicitly attributable to values/resources rather than evidence selection |
| OQ-17 | Persistent CIPN with fall/work limitation; rehabilitation versus attention control | Injurious falls, work participation and physical function | 12 months | Clinically important functional gain with acceptable adherence and no injury excess |
| OQ-18 | New gabapentinoid users with renal impairment or fall risk; electronic control plus pharmacist review | Serious falls/respiratory events and ≥30% pain response | 12 months | Harm reduction without loss of responder probability beyond a non-inferiority margin |
| OQ-19 | SCI rehabilitation services randomized to structured classification/treatment linkage or usual assessment | Participation or sleep plus diagnostic yield and false-positive burden | 12 months | Better patient-centered outcome with acceptable workload, or no incremental benefit despite improved labeling |
| OQ-20 | Taxane/platinum recipients crossing a prespecified sensory threshold | Injurious falls per person-time | Through treatment plus 6 months | Reduced falls with preserved cancer-treatment delivery, or precise evidence of no preventive effect |
| OQ-21 | Stroke follow-up pathways using scheduled versus symptom-triggered central-pain screening | Time to correct diagnosis and pain interference | 12–18 months | Earlier diagnosis with better function and acceptable false-positive burden |
| OQ-22 | Individual-participant data across etiology-specific exercise trials | Pain responder, function, fall and adherence interaction by etiology | ≥6 months | Reproducible etiology-specific benefit, or precise absence of clinically useful exercise effects |
| OQ-23 | Multi-etiology cohort with suspected neuropathic pain classified independently under the 2016 hierarchical grading and the convergence model, including the disputed boundary cases | Agreement between frameworks, misclassification against longitudinal expert-panel adjudication, and prognostic validity | 12 months | One framework shows better classification and prognostic validity, or the two are interchangeable outside identifiable boundary cases |
Closed¶
None in this build.