Response to Andrea Truini: Diagnosing a Lesion or Diagnosing Neuropathic Pain?

Abstract

We thank Professor Andrea Truini for his thoughtful and constructive commentary (Truini 2026) on our position paper related to the NeuPSIG diagnostic algorithm for neuropathic pain (Bouhassira and Attal 2026). We are pleased that we share substantial common ground: diagnostic confidence in neuropathic pain should emerge from the convergence of multiple sources of information rather than from a rigid hierarchy privileging supposedly objective findings over patients' reports. Our disagreement concerns a narrower but fundamental question: should investigations demonstrating a somatosensory lesion contribute directly to the classification of pain as neuropathic?

We entirely agree that ‘imperfect tests are not uninformative tests’. We never argued otherwise. Paraclinical investigations are clinically valuable and often essential for identifying an underlying neurological lesion or disease. Our proposal is to reconsider their role in pain classification, not to dismiss their clinical utility. The critical distinction concerns what these investigations really demonstrate. Nerve conduction abnormalities establish peripheral nerve dysfunction, reduced intraepidermal nerve fibre density indicates small-fibre pathology, and imaging may identify structural lesions. However, none establishes that the identified abnormality causes the patient's pain. This distinction matters because neurological lesions are frequently painless, and structural abnormalities may also occur in conditions not conventionally classified as neuropathic pain. Even a hypothetical test with perfect sensitivity and specificity for detecting a neurological lesion would not necessarily establish whether that specific lesion causes pain. Hence, accuracy for diagnosing a lesion and accuracy for classifying pain are fundamentally different purposes.

Adrea Truini refers to the studies by Peter Dyck and colleagues to support objective anchoring when clinical examination is unreliable. However, these studies evaluated the diagnosis of diabetic polyneuropathy, not whether patients' pain was neuropathic. Nerve conduction studies provide an excellent reference for diagnosing neuropathy, but do not constitute an equivalent reference for classifying pain. The same reasoning applies to small-fibre neuropathy. Diagnostic criteria combining clinical signs, thermal thresholds and intraepidermal nerve fibre density illustrate the value of convergence for identifying neuropathy. They do not establish that these findings reliably distinguish painful from painless neuropathy. Indeed, the weak association between intraepidermal nerve fibre density and pain was a central argument in our position paper.

We acknowledge Andrea Truini's point that some laser evoked potential parameters may be less influenced by psychological modulation than response amplitude, and that methodological standardisation may improve test reliability. Nevertheless, improved reliability in detecting somatosensory dysfunction does not solve the question of its causal relationship with pain.

We also agree that the diagnostic accuracy of neuropathic pain screening questionnaires should be interpreted cautiously. Their performances vary across populations, and positive scores may be observed in non-neuropathic pain conditions.

Our intention was never to replace an objective gold standard with a subjective one. Rather we questioned why patient-reported measures are systematically considered less informative than instrument-derived findings, despite the latter's own limitations. Our proposed algorithm deliberately requires convergence between history and topography, neuropathic pain descriptors, and sensory abnormalities. Symptoms alone cannot establish probable or definite neuropathic pain.

Andrea Truini's strongest objection is that our model allows a classification of definite neuropathic pain without independent confirmation of a somatosensory lesion. This is correct and deliberate. Consider a patient with a characteristic history, neuroanatomically plausible pain distribution, typical neuropathic descriptors, and concordant sensory deficits or allodynia, but normal paraclinical investigations. Should this patient's pain necessarily be classified as less certainly neuropathic than that of someone with less compelling clinical features but with an abnormal investigation?

Our proposal challenges precisely this assumption. It does not disregard neurological lesions: history and topography must remain compatible with somatosensory pathology. However, demonstrating a lesion does not automatically provide stronger evidence concerning the nature of the pain than a coherent clinical phenotype.

Ultimately, our positions may be closer than they initially appear. Andrea Truini agrees that investigations should not constitute a rigid diagnostic gatekeeper, that normal findings should not exclude neuropathic pain when clinical evidence is compelling, and that all findings require contextual interpretation.

The remaining disagreement concerns whether lesion-demonstrating investigations should be included within the pain-classification algorithm itself. We favour distinguishing two related questions: is the patient's pain neuropathic? What neurological lesion or disease may account for it? Paraclinical investigations are essential to the second question and may inform clinical judgement regarding the first one, but combining them risks confusing evidence of neural pathology with evidence of pain causation.

We fully support Andrea Truini's suggestion that future research should quantify the incremental contribution of individual investigations to neuropathic pain classification. Importantly, such studies must avoid circular validation against reference standards that already incorporate the tests under evaluation.

As also recently proposed by Finnerup and Glyfadottir (Finnerup and Gylfadottir 2026), a NeuPSIG working group would provide an excellent opportunity to develop and validate a genuinely integrative framework. Our objective is not to diminish the importance of neurological investigations, but to ensure that diagnostic certainty about neuropathic pain is not confused with certainty about the presence of a neurological lesion.

This correspondence was designed and written by both authors.

This article is related Truini and Attal paper. To view this article, visit https://doi.org/10.1002/ejp.70421 and https://doi.org/10.1002/ejp.70350.