This is another attempt at identifying biological markers for pain in humans from skin biopsies, which ends with a negative result, but yields some interesting information outside the original research question. Preferably, such a biomarker should result from a noninvasive or minimally invasive test, and Martin Schmelz and colleagues (Schley et al., 2012) employ a comprehensive battery of such tests, including bedside examination, quantitative sensory testing, axon reflex flare and the quantification of skin innervation. In contrast to most previous authors who tried to correlate skin innervation with pain, they took great care in separating skin nerve fibres into different layers and correlated their quantity with sensory function like thermal thresholds or the degree of electrically induced pain. They found some – partially unexpected – correlations between structure and function, like correlations between pain thresholds and deep dermal innervation. More expected were the correlation of temperature thresholds with epidermal nerve fibre density and of the area of electrically evoked axon reflex flare with calcitonin gene related peptide (CGRP) staining of dermal nerve fibres. None of the parameters showed a close correlation with the pain measures. Some of the weak correlations identified are even counterintuitive. If cold pain thresholds correlated positively with higher neuronal CGRP staining in the dermal skin, then less CGRP would mean lower pain thresholds, thus more pain, in contrast to what is generally assumed and to the authors’ hypothesis.
This poses several questions: Would the results have been changed if the authors had compared neuropathy patients with pain with neuropathy patients without pain (e.g., Üçeyler et al., 2007)? This might have eliminated some of the confounding factors caused by the disease itself. With the present design, most of the differences between healthy controls and neuropathic pain patients may be explained by the nerve lesion, like loss of distal nerve fibres and loss of the respective function. The parameters responsible for gain of function (pain, allodynia) may thus be missed. On the other hand, the structural measures employed may not be suited to differentiate between a painful and a painless state. For example, nerve fibres may lose the PGP 9.5 immunoreactivity, but may still be present and active, and might be visualized by a GAP-43 stain (Ragé et al., 2010) or sodium channel immunoreaction, or numbers of remaining nerve fibres might play a minor role in determining pain, and the surrounding microenvironment might be of more importance (Üçeyler et al., 2010). Furthermore, it is possible that the pathophysiology of the disorders examined, which include peripheral neuropathy, traumatic nerve lesion, post-herpetic neuralgia and complex regional pain syndrome (CRPS), differs too much to yield consistent findings when lumping them into one group. Studies were more homogenous groups were investigated, yielded at least partially stronger correlation between their morphological finding in skin and pain (e.g., Vlckova-Moravcova et al., 2008; Chao et al., 2010; Casanova-Molla et al., 2012)
The authors conclude that the analysis of neuronal markers, such as sodium channel subtypes in painful neuroma may be a more promising approach to characterize the mechanisms of neuropathic pain. I would go even further and propose that not only the types and numbers of sodium channels, but their activity state may play a major role, as exemplified by the sodium channel mutations underlying erythromelalgia (Fischer and Waxman, 2010). Should we thus stop using those accessible measures in patients to look for patterns that might help us differentiate painful and painless states? No, but we should go to different levels. The main problem is that most of the ‘objective’ measures we have are good at showing loss of function, but are not related to gain of function. There seems to be no way around microneurography or other activity markers for C-fibres. Schley and colleagues have provided a good baseline for such further studies.
Neurologische Klinik, Universitätsklinikum Würzburg, Würzburg, Germany
Tel.: +49 931 201 23763 fax: +49 931 201 23697