Skip to content

Biomarkers

TL;DR — No externally validated biomarker independently diagnoses neuropathic pain or selects a drug as of a targeted PubMed search on 2026-08-30. Skin biopsy confirms small-fiber loss, not pain; QST quantifies perception, not lesion location; EEG and imaging show heterogeneous group signals (Lauria 2010, PMID 20642627); (Mussigmann 2022, PMID 35659993). A useful biomarker must add value beyond history, examination and lesion confirmation and validate externally.

Use cases

Diagnostic, prognostic, predictive, pharmacodynamic and safety biomarkers require different validation; correlation with pain intensity satisfies none by itself.

The intended use determines the study design. A diagnostic marker must distinguish neuropathic pain from clinically relevant mimics; a prognostic marker predicts outcome under usual care; a predictive marker requires a marker-by-treatment interaction; and a pharmacodynamic marker must change with target engagement on a biologically plausible timescale (Edwards 2023, PMID 36198371). Reusing a cross-sectional case-control association for all four roles creates circular “validation.”

Intended use Required comparison Minimum performance report Frequent invalid shortcut
Lesion confirmation Suspected neuropathy versus disease and healthy controls Sensitivity, specificity, likelihood ratios with prespecified cutoff Comparing only extreme cases with healthy volunteers
Pain-state discrimination Painful versus painless neuropathy with similar lesion severity Calibration and discrimination in an external cohort Correlating marker with pain score in painful cases only
Prognosis Baseline marker versus future pain/function Adjusted effect, calibration, missingness and time horizon Calling cross-sectional severity prognostic
Treatment prediction Randomized treatment-by-marker interaction Interaction estimate and confidence interval Separate significance in one subgroup but not another
Target engagement Pre/post change versus control and dose Reliability, temporal response and exposure-response Assuming symptom improvement proves target engagement

Skin biopsy

Age-adjusted distal-leg intraepidermal fiber density has Level A support for small-fiber neuropathy diagnosis with laboratory quality control (Lauria 2010, PMID 20642627).

Skin biopsy measures structure: intraepidermal nerve-fiber density below age- and sex-adjusted normative limits supports small-fiber neuropathy, provided biopsy site, fixation, immunostaining, counting rules and laboratory norms are controlled (Lauria 2010, PMID 20642627). It does not establish that the lesion is painful, localize ectopic firing or determine whether pain is centrally maintained.

Diagnostic criteria remain non-uniform. A systematic review found substantial variation in reference standards and combinations of symptoms, signs, QST and biopsy, which changes apparent sensitivity and specificity (Haroutounian 2021, PMID 32989823). The correct report therefore states which criteria were used rather than labeling biopsy as a binary “pain test.”

QST

DFNS provides standardized norms but remains psychophysical and pathway-wide (Rolke 2006, PMID 16697110).

The DFNS protocol maps 13 parameters spanning thermal detection/pain, mechanical detection/pain, wind-up, vibration and pressure, normalized by body site, age and sex (Rolke 2006, PMID 16697110). QST can identify loss and gain patterns, but attention, expectation, medications and central processing contribute. It cannot by itself distinguish a peripheral lesion from an abnormal central perceptual state (Backonja 2009, PMID 19692807).

Longitudinal responsiveness is also endpoint-specific. In entrapment neuropathy, sensory phenotypes changed after disease-modifying intervention, demonstrating that QST is not necessarily a stable trait marker (Kennedy 2021, PMID 33769367). That plasticity is useful for monitoring but complicates one-time treatment stratification.

Electrophysiology

Nerve conduction confirms large-fiber dysfunction; evoked potentials and microneurography answer narrower questions. Normal NCS does not exclude small-fiber disease (Devigili 2020, PMID 32654574).

Routine NCS measures large myelinated-fiber function and can grade distribution and severity, while laser-evoked potentials interrogate small-fiber pathways and microneurography can record spontaneous activity in selected research settings. Each measures a different level of the system; concordance is informative, but discordance should not automatically be called measurement error (Truini 2023, PMID 37253688).

EEG

Fourteen heterogeneous studies showed candidate theta/beta changes but inconsistent intensity relations (Mussigmann 2022, PMID 35659993).

The EEG literature includes resting power, peak alpha frequency, connectivity and source-localized measures. Across the 14 studies, small cohorts, mixed etiologies, inconsistent preprocessing and medication confounding prevented a clinically deployable signature (Mussigmann 2022, PMID 35659993). Neurofeedback response does not retrospectively validate a baseline EEG feature as a diagnostic biomarker (Mussigmann 2025, PMID 40947911).

Imaging

Functional imaging maps networks and group differences; reverse inference, medication and small samples prevent diagnostic use (Peyron 2019, PMID 30318262).

Pain-related BOLD and connectivity differences are distributed across salience, sensorimotor, affective and descending-control networks rather than confined to a neuropathic-pain “center” (Peyron 2019, PMID 30318262). A classifier developed in a selected case-control dataset must be tested against nociceptive pain, painless neuropathy, depression, sleep disturbance and medication effects before it can claim clinical specificity.

Molecular candidates

Neurofilament light may index axonal injury in CIPN, but pain prediction and clinical cutoffs remain unvalidated (Andersen 2024, PMID 39242335). Metabolomics is exploratory (Teckchandani 2021, PMID 33065770).

The 2026 search found candidate RNA signatures and proposed assay workflows, but the newest clinical review states that the composite pain-biology score remains theoretical and lacks in-vivo, in-vitro or clinical validation (Soin 2026, PMID 42534835). This is positive evidence of an active validation gap, not evidence that molecular candidates are absent.

Inflammatory-marker studies demonstrate the multiplicity problem. A sciatica review found 16 studies/1,212 participants and numerous cytokines/chemokines; severe pain was associated with higher hsCRP in one analysis (adjusted OR 3.4, 95% CI 1.1–10), while heterogeneity and risk of bias prevented firm conclusions (Jungen 2019, PMID 30967132). An isolated significant cytokine is therefore a candidate, not a validated panel.

Corneal confocal microscopy

Corneal confocal microscopy offers rapid, non-invasive quantification of small-fiber density, length and branching. In a cross-sectional study of 53 painful DSPN, 63 painless DSPN and 46 controls, density and length were reduced in both neuropathy groups; only branch density differed between painful and painless disease—55.8 versus 43.8 branches/mm² (P<0.05) after covariate adjustment (Püttgen 2019, PMID 31390004).

That result supports a regeneration/branching hypothesis but is not a diagnostic cutoff for neuropathic pain. The groups overlapped substantially (SD 29.9 and 28.3 branches/mm²), the design was cross-sectional, and corneal anatomy is a surrogate for somatic peripheral nerve state rather than the generator of foot pain (Püttgen 2019, PMID 31390004).

Validation table

Requirement Minimum evidence
Analytical validity Reliable assay across labs
Clinical validity Prespecified independent cohort
Specificity Disease and pain controls
Prediction Treatment interaction, not association
Utility Better decisions/outcomes
Transportability Multiple sites/populations

Development-to-utility pathway

Stage Required deliverable Stop/go criterion
Discovery Prespecified assay and unbiased sampling Association survives multiplicity control
Technical replication Inter-/intra-laboratory reliability Error is small relative to biological separation
Clinical validation Blinded external cohort with relevant controls Discrimination and calibration meet intended-use threshold
Incremental value Comparison with history, examination and lesion tests Marker improves decision metrics, not only AUC
Predictive validation Prospective randomized interaction Treatment-by-marker interaction replicates
Utility trial Marker-guided versus standard care Better patient outcome or lower burden without missed disease
Surveillance Drift, subgroup performance and harms Performance remains calibrated across sites and time

Evidence interpretation map

The table makes the evidence role and inferential boundary explicit; it is not a replacement for the full reports.

PMID Year Evidence role What it cannot establish alone
20642627 2010 Standardized skin-biopsy lesion assessment Pain diagnosis or treatment selection
16697110 2006 QST protocol and reference distributions Lesion localization or mechanism-specific therapy
35659993 2022 Systematic EEG evidence map Deployable individual classifier
39242335 2024 Neurofilament/CIPN biomarker synthesis Pain-specific cutoff or utility
31390004 2019 Painful-versus-painless corneal comparison Individual diagnosis; distributions overlap
30967132 2019 Inflammatory-biomarker heterogeneity and effect estimates Validated inflammatory signature

Explicit controversies

  1. Lesion marker versus pain marker. Reduced intraepidermal fibers confirms small-fiber loss (Lauria 2010, PMID 20642627), yet painful and painless neuropathy overlap. Calling biopsy a neuropathic-pain biomarker without specifying “lesion confirmation” overstates its use.
  2. Phenotype as predictor. QST-defined irritable nociceptor phenotype interacted with oxcarbazepine in one trial (Demant 2014, PMID 25139589), but did not predict lacosamide response in a prematurely closed later trial (Carmland 2024, PMID 37565715). The predictive claim remains unresolved rather than disproven.
  3. Corneal regeneration signal. Greater corneal branching in painful than painless DSPN may indicate regeneration (Püttgen 2019, PMID 31390004), but overlap and cross-sectional design also permit confounding or reverse interpretation.
  4. Multimodal machine learning. Combining EEG, imaging, QST and molecular measurements can improve apparent in-sample classification while magnifying overfitting. External, site-held-out validation and comparison with clinical examination are prerequisites (Edwards 2023, PMID 36198371).

Minimum reporting controls

Domain Required report
Case definition Possible, probable or definite neuropathic pain
Etiology Lesion/disease and diagnostic evidence
Distribution Focal, length-dependent, dermatomal, at-level or below-level
Baseline phenotype Negative and positive sensory signs
Comparator Placebo/sham, active care or natural history
Exposure Dose, duration, adherence and co-interventions
Benefit Mean change plus ≥30% and ≥50% responders where applicable
Function Sleep, mobility, participation and patient global change
Harm Adverse events, withdrawals and serious events
Durability Follow-up after treatment and attrition
Subgroups Prespecified interaction test, not within-group significance
Missingness Denominator and imputation method

Reporting cautions

  • Do not infer lesion presence from a symptom descriptor.
  • Do not convert a group-average association into an individual diagnostic rule.
  • Do not treat statistical significance as clinically important benefit.
  • Do not compare NNTs without checking outcome threshold, duration and population.
  • Do not interpret an inactive or completed registry record as proof of efficacy.
  • Do not merge painful and painless neuropathy outcomes.
  • Do not omit adverse-event withdrawals from responder interpretation.
  • Do not call a post hoc subgroup predictive without an interaction test.
  • Do not generalize a focal peripheral result to central neuropathic pain.
  • State when evidence is short-term, indirect or restricted to a selected cohort.

Open questions

  • Can a marker distinguish painful from painless neuropathy?
  • Does multimodal fusion improve external prediction?
  • Which marker captures ongoing ectopic firing?
  • Can biomarkers reduce trial sample size without bias?

References

  1. Lauria G, Hsieh ST, Johansson O, et al.. European Federation of Neurological Societies/Peripheral Nerve Society Guideline on the use of skin biopsy in the diagnosis of small fiber neuropathy. Report of a joint task force of the European Federation of Neurological Societies and the Peripheral Nerve Society. Eur J Neurol. 2010;17:903-12, e44-9. PMID 20642627
  2. Mussigmann T, Bardel B, Lefaucheur JP. Resting-state electroencephalography (EEG) biomarkers of chronic neuropathic pain. A systematic review. Neuroimage. 2022;258:119351. PMID 35659993
  3. Rolke R, Baron R, Maier C, et al.. Quantitative sensory testing in the German Research Network on Neuropathic Pain (DFNS): standardized protocol and reference values. Pain. 2006;123:231-243. PMID 16697110
  4. Devigili G, Cazzato D, Lauria G. Clinical diagnosis and management of small fiber neuropathy: an update on best practice. Expert Rev Neurother. 2020;20:967-980. PMID 32654574
  5. Peyron R, Fauchon C. Functional imaging of pain. Rev Neurol (Paris). 2019;175:38-45. PMID 30318262
  6. Andersen NE, Boehmerle W, Huehnchen P, et al.. Neurofilament light chain as a biomarker of chemotherapy-induced peripheral neuropathy. Trends Pharmacol Sci. 2024;45:872-879. PMID 39242335
  7. Teckchandani S, Nagana Gowda GA, Raftery D, et al.. Metabolomics in chronic pain research. Eur J Pain. 2021;25:313-326. PMID 33065770
  8. Haroutounian S, Todorovic MS, Leinders M, et al.. Diagnostic criteria for idiopathic small fiber neuropathy: A systematic review. Muscle Nerve. 2021;63:170-177. PMID 32989823
  9. Strand N, Wie C, Peck J, et al.. Small Fiber Neuropathy. Curr Pain Headache Rep. 2022;26:429-438. PMID 35384587
  10. Krumova EK, Geber C, Westermann A, et al.. Neuropathic pain: is quantitative sensory testing helpful? Curr Diab Rep. 2012;12:393-402. PMID 22623149
  11. Backonja MM, Walk D, Edwards RR, et al.. Quantitative sensory testing in measurement of neuropathic pain phenomena and other sensory abnormalities. Clin J Pain. 2009;25:641-7. PMID 19692807
  12. Kennedy DL, Vollert J, Ridout D, et al.. Responsiveness of quantitative sensory testing-derived sensory phenotype to disease-modifying intervention in patients with entrapment neuropathy: a longitudinal study. Pain. 2021;162:2881-2893. PMID 33769367
  13. Åkerlund M, Baskozos G, Li W, et al.. Genetic associations of neuropathic pain and sensory profile in a deeply phenotyped neuropathy cohort. Pain. 2025;166:1354-1368. PMID 39471050
  14. Sachau J, Bruckmueller H, Gierthmühlen J, et al.. SIGMA-1 Receptor Gene Variants Affect the Somatosensory Phenotype in Neuropathic Pain Patients. J Pain. 2019;20:201-214. PMID 30266269
  15. Sommer C, Üçeyler N. Small-Fiber-Neuropathien. Fortschr Neurol Psychiatr. 2018;86:509-518. PMID 30125923
  16. Cazzato D, Lauria G. Small fibre neuropathy. Curr Opin Neurol. 2017;30:490-499. PMID 28665811
  17. Kool D, Hoeijmakers JG, Waxman SG, et al.. Small fiber neuropathy. Int Rev Neurobiol. 2024;179:181-231. PMID 39580213
  18. Widerström-Noga E. Neuropathic Pain and Spinal Cord Injury: Phenotypes and Pharmacological Management. Drugs. 2017;77:967-984. PMID 28451808
  19. Widerström-Noga E. Neuropathic Pain and Spinal Cord Injury: Management, Phenotypes, and Biomarkers. Drugs. 2023;83:1001-1025. PMID 37326804
  20. Mussigmann T, Bardel B, Créange A, et al.. Relieving Chronic Neuropathic Pain With EEG-Neurofeedback. Eur J Neurol. 2025;32:e70363. PMID 40947911
  21. Peltier A, Goutman SA, Callaghan BC. Painful diabetic neuropathy. BMJ. 2014;348:g1799. PMID 24803311
  22. Wang T, Liu L, Song D, et al.. Emerging roles of lncRNAs in the pathogenesis, diagnosis, and treatment of trigeminal neuralgia. Biochem Soc Trans. 2022;50:1013-1023. PMID 35437600
  23. Dusch M, Schmelz M. [Erythromelalgia: skin redness and pain]. Schmerz. 2019;33:475-490. PMID 31485751
  24. Edwards RR, Schreiber KL, Dworkin RH, et al.. Optimizing and Accelerating the Development of Precision Pain Treatments for Chronic Pain: IMMPACT Review and Recommendations. J Pain. 2023;24:204-225. PMID 36198371
  25. Colloca L, Ludman T, Bouhassira D, et al.. Neuropathic pain. Nat Rev Dis Primers. 2017;3:17002. PMID 28205574
  26. Püttgen S, Bönhof GJ, Strom A, et al. Augmented corneal nerve fiber branching in painful compared with painless diabetic neuropathy. J Clin Endocrinol Metab. 2019;104:6220-6228. PMID 31390004
  27. Jungen MJ, ter Meulen BC, van Osch T, et al. Inflammatory biomarkers in patients with sciatica: a systematic review. BMC Musculoskelet Disord. 2019;20:156. PMID 30967132
  28. Demant DT, Lund K, Vollert J, et al. The effect of oxcarbazepine in peripheral neuropathic pain depends on pain phenotype: a randomised, double-blind, placebo-controlled phenotype-stratified study. Pain. 2014;155:2263-2273. PMID 25139589
  29. Carmland ME, Kreutzfeldt MD, Holbech JV, et al. The effect of lacosamide in peripheral neuropathic pain: a randomized, double-blind, placebo-controlled, phenotype-stratified trial. Eur J Pain. 2024;28:105-119. PMID 37565715
  30. Truini A, Aleksovska K, Anderson CC, et al. Joint European Academy of Neurology-European Pain Federation-Neuropathic Pain Special Interest Group of the International Association for the Study of Pain guidelines on neuropathic pain assessment. Eur J Neurol. 2023;30:2177-2196. PMID 37253688
  31. Soin A, Khaira M, Soin A, et al. New insight into RNA biomarkers in neuropathic pain: a clinician-neuroscientist roadmap to translational testing and treatment monitoring a clinical review. Front Pain Res (Lausanne). 2026;7:1874336. PMID 42534835