In this review, which does not cover other naturally occurring or synthetic TRPV1 agonists, we discuss the potential utility of topically administered capsaicin for the management of pain in classical peripheral neuropathies and other hypersensitivity disorders, some of which are currently considered as idiopathic. Furthermore, we seek to elucidate the molecular and cellular basis of capsaicin treatment, and clarify misunderstandings, particularly with respect to the involvement of substance P depletion.

In an attempt to evaluate whether pain relief could be achieved by a single exposure to a much higher concentration of topical capsaicin, 10 patients with intractable pain syndromes were treated with a compounded high-concentration 5–10% w/w cream.⁶ Patients were provided regional anaesthesia for tolerability and airborne contamination of treatment rooms occurred. Based on encouraging results, a high-concentration capsaicin-containing (8%) patch designated NGX-4010 and then given the trade name Qutenza™ was developed and evaluated.⁷

The capsaicin 8% patch is designed to rapidly deliver capsaicin into the skin while minimizing unwanted systemic or environmental exposure of capsaicin to patients and health-care providers. Phase 1 data suggested that a single 60-min patch application was adequate to induce nociceptor defunctionalization, as measured by reversible reduction in intra-epidermal nerve fibres (ENFs), marked by the structural nerve marker protein gene product (PGP) 9.5 immunostaining, and small, reversible alterations in cutaneous nociceptor function.^8,9 Phase 3 studies demonstrated efficacy against PHN^10,11 (Fig. 1) and painful HIV-AN (associated neuropathy).¹² For both neuropathic pain syndromes, efficacy was observed to last for 12 weeks. Blinding was provided by a control patch which contained sufficient capsaicin to induce pain and erythema in a substantial number of subjects.

When activated by a combination of heat, acidosis, or endogenous/exogenous agonists, TRPV1 may open transiently and initiate depolarization mediated by the influx of sodium and calcium ions. In the nociceptive sensory nerves which selectively express TRPV1 (mostly C- and some Aδ-fibres), depolarization results in action potentials, which propagate into the spinal cord and brain, and may be experienced as warming, burning, stinging, or itching sensations (Fig. 2).

In contrast to transient activation which follows normal environmental stimuli or inflammatory responses to tissue injury, activation of TRPV1-expressing nerve fibres by exposure to a chemically stable exogenous agonist, such as capsaicin, can generate a biochemical signal with a persistent effect. The TRPV1 channel is highly calcium permeable (with a calcium:sodium permeability ratio that starts at about 8:1 and increases to about 25:1 during prolonged capsaicin exposures),¹⁶ which allows significant amounts of calcium to flow down its steep electrochemical gradient into nerve fibres. Furthermore, as TRPV1 is also expressed on intracellular organelles, external capsaicin application can cause release of calcium from the endoplasmic reticulum¹⁷ and induce additional intracellular calcium release from internal stores via calcium-dependent calcium release.¹⁸ Taken together, these multiple sources of calcium provide a robust intracellular signal which can overwhelm local calcium sequestration mechanisms. Consequently, sustained high levels of intracellular calcium can activate calcium-dependent enzymes such as proteases,¹⁹ and can induce the depolymerization of cytoskeletal components such as microtubules.^20,21 Moreover, osmotic swelling due to the chloride accumulation that must accompany influxes of positively charged ions also occurs.¹ An additional effect of high concentrations of capsaicin, which does not involve TRPV1, is a direct inhibition of mitochondrial respiration. Numerous mitochondria are present in the peripheral terminals of nociceptors and may congregate there in response to nerve growth factor (NGF).²² At concentrations much higher than required to activate TRPV1, capsaicin can compete with ubiquinone to inhibit directly electron chain transport.²³ Consequently, capsaicin can dissipate mitochondrial transmembrane potential,²⁴ and does so with an EC₅₀ of 6.9 μM in sensory neurones.²⁵ In accord with these widely recognized effects, if TRPV1-expressing sensory nerve fibres are exposed to high concentrations of capsaicin or to lower concentrations in a continuous fashion, high levels of intracellular calcium and the associated enzymatic, cytoskeletal, and osmotic changes, and the disruption of mitochondrial respiration lead to impaired local nociceptor function for extended periods²⁶ (Fig. 3).

The term ‘desensitization’ is often used to describe these local effects of capsaicin on sensory nerve function, but is unsatisfactory in several respects. The use of this nomenclature in capsaicin literature arose many years ago from psychophysical studies of human subjects who displayed reduced reactions to painful stimuli applied to skin areas pretreated with capsaicin.¹ Unfortunately, once the capsaicin receptor (TRPV1) was recognized as a unique molecule entity, the psychophysical use of ‘desensitization’ evolved into pharmacological use, which denotes the reduction of responsiveness of receptors, ion channels, or intracellular signalling pathways after prolonged or repeated agonist exposures. In the continued presence of exogenous agonists such as capsaicin, pharmacological desensitization of TRPV1 itself may indeed contribute acutely to analgesic efficacy. However, transient effects on TRPV1 are quite unlikely to account for the persistent pain relief seen clinically after either single treatments with high-concentration capsaicin or repetitive administration of low-concentration capsaicin. Hence, the emerging preferred term for the persistent local effects of capsaicin is ‘defunctionalization’,^26,27 which avoids conceptual confusion with the intrinsic desensitisation of the TRPV1 receptor.

Loss of mitochondrial function due to calcium overload and inhibition of metabolism may render affected nerve processes unable to maintain plasma membrane integrity and thus cause collapse of nerve endings to the depth where the capsaicin exposure was insufficient to irreversibly overwhelm mitochondrial function (Fig. 4). If nerve fibres in skin retract or ‘degenerate’ to the depth at which mitochondrial function was preserved, it is expected that markers for any constituent of those fibres will show reductions. Indeed, many immune-histochemical studies using antibodies to PGP 9.5 or other nerve fibre proteins provide evidence that capsaicin can produce highly localized loss of nociceptive nerve fibre terminals in the epidermis and dermis²⁸ (Fig. 5). However, it is important to recognize that defunctionalization and nerve terminal degeneration are distinct phenomena, each with potentially different time-courses. In the simplest case, a loss of electrical excitability (the loss of function) may occur via depolarization block or sodium channel inactivation, independently of a loss of axonal integrity. Other important physiological roles such as fast axonal transport of growth factors can also be compromised by capsaicin without axonal collapse.²⁹ Even though a general parallel between capsaicin-induced functional and structural changes is expected, there may be differences in the time-courses. For instance, in a recent study of the time-course of recovery of cerebral-evoked potential amplitudes to cutaneous heat-pain laser stimuli after topical capsaicin treatment, functional recovery occurred before the conventional cutaneous nerve markers PGP 9.5 and TRPV1 showed significant recovery³⁰ (Fig. 5). In contrast, fibres marked by GAP-43, which is expressed by regenerating nerve fibres, did show good correlation with the functional responses.

Rapid delivery of capsaicin may promote, rather than reduce, the tolerability of topical capsaicin. Some of the defunctionalization mechanisms discussed above can occur very rapidly and in vitro loss of capsaicin responsiveness may develop within 20 s.³⁹ By driving cutaneous nociceptors to a defunctionalized state quickly, the inevitable pungency may be greatly mitigated. Indeed, in clinical studies with capsaicin 8% patch, <2% of patients asked for early removal of the patch due to intolerance.⁷

Direct correlations between aberrant activity of peripheral nociceptors and pain reported by patients have not often been observed due to the technical complexity of measuring electrical activity in small-diameter nerve fibres in patients. A technique known as microneurography, which measures action potentials extracellularly, can be used, but this diagnostic procedure is somewhat invasive and may cause discomfort.⁴⁵ In patients with painful small-fibre polyneuropathy, both polymodal and mechanically insensitive C-fibres were hyperexcitable, as indicated by reduced receptor thresholds, spontaneous discharges, and exaggerated responses to stimulation.⁴⁶ Interestingly, it was suggested that the clinical and electrophysiological profiles of these patients resembled the effects of experimental capsaicin application to the skin. Studies of hyperactive nociceptors in patients with erythromelalgia (burning pain of the feet, in some patients due to mutations in Na_V 1.7 voltage-dependent sodium channels) or diabetic neuropathy showed spontaneous activity in nociceptive fibres, sensitization of mechano-insensitive C-fibres, and a reversal of the proportion of the two main subtypes of C-fibres (which indicates a loss of function of polymodal nociceptors).⁴⁷ Microneurography studies also suggest that hypersensitivity of C-fibres—particularly those classified as ‘silent’ or mechanically insensitive under normal conditions—can contribute to tactile allodynia.⁴⁸ Therefore, tactile allodynia may not arise only from inappropriate spinal integration of electrical signals arising from A-fibres.

Immunohistochemical analyses have indicated that the density of ENFs in the epidermis is decreased in a wide range of neuropathic pain syndromes, including PHN,^55–57 painful diabetic neuropathy (PDN),^58,59 painful HIV-associated neuropathy (HIV-AN),⁶⁰ complex regional pain syndrome,⁶¹ small-fibre neuropathy,^62–64 metabolic syndrome,⁶⁵ and Fabry disease.⁶⁶ Moreover, data suggest a positive correlation between the extent of ENF loss and the severity of pain in PHN,⁵⁷ PDN,^58,59 and HIV-AN.⁶⁰ Thus, in most pain syndromes considered to be neuropathic, sensory neurone axon density in the target tissue (commonly the skin) is decreased. In contrast, in some other neuropathic pain conditions, such as in proximal inflammatory or compressive disorders affecting the spinal nerve root or dorsal root ganglion (e.g. Sjogren's syndrome), which do not involve significant loss of cell bodies or axons distal to the dorsal root ganglion, there may be changes which are not length-dependent, with regionally preserved nociceptor innervation of target organs. Spared nerve fibres may also sprout in the skin in neuropathic conditions such as PHN, although overall density of ENFs is generally reduced. Why density changes of cutaneous nerve fibres can lead to chronic pain and particularly hypersensitivity will be discussed below (Fig. 6).

It is now widely accepted that nociceptors may develop hyperexcitable electrophysiological properties, due to exposure to relatively abnormal concentrations of neurotrophins such as NGF or glial cell line-derived neurotrophic factor (GDNF), or pro-inflammatory cytokines, as hypothesized and discussed previously.^67–69 In chronic pain syndromes associated with denervation (Fig. 6b), pain intensity may correlate with reduced nociceptor immunostaining because when there are only a small number of intact nociceptive endings, it is more likely those endings will have access to an abnormally high supply of the neurotrophins produced by the skin and ensheathing Schwann cells. Although NGF has an important role in controlling the survival and development of small-diameter neurones—both sensory and sympathetic—it has become clear that NGF also serves as an important signal for neuroimmune and inflammatory processes in mature organisms.⁶⁹ In normal human skin, NGF-immunoreactivity is predominantly in basal keratinocytes.⁷⁰ Production of NGF may be up-regulated by inflammation or denervation of skin; rats show a rapid and prolonged increase (5- to 10-fold) of NGF mRNA in denervated skin, distal nerves, and basal keratinocytes.⁷¹ In response to enhanced NGF supply, intact nociceptors may respond by becoming hyperactive, sprout, or both. As a direct excitatory stimulant, NGF causes immediate excitation of nociceptors,⁷² resulting in prolonged hyperalgesia and allodynia.^73,74 In addition to this direct and rapid effect, retrograde transport of NGF to sensory neurone cell bodies may lead to the up-regulation of pro-excitatory proteins such as TRPV1 and voltage-activated sodium channels,⁷⁵ and down-regulation of anti-excitatory proteins such as voltage-activated potassium channels.⁷⁶ The recent clinical successes of the anti-NGF neutralizing antibody tanezumab provide direct evidence for the role of NGF in chronic pain syndromes.⁷⁷ Similar success has been achieved by relocating painful injured nerves from NGF-rich (subcutaneous) to NGF-poor (muscle) regions, with appropriate changes in NGF levels.⁷⁸

Although not as extensively studied as NGF, oversupply of other neurotrophins produced in the skin such as GDNF⁷⁹ or artemin⁸⁰ can also hypersensitize cutaneous nociceptors. We postulate that much of the hyperactivity of nociceptors in peripheral neuropathic pain syndromes of ‘static denervation’ (Fig. 6b), ‘dynamic denervation’ (Fig. 6c), and ‘hyperinnervation’ (Fig. 6d) types is due to the relatively hyperstimulating hypertrophic environment to which intact or regenerating cutaneous nociceptors are exposed.

Pro-inflammatory cytokines can also directly activate and modify gene expression in sensory neurones, and there are several sources of these molecules in close proximity to peripheral nerves. Schwann cells, which have often been thought of as having only a passive support role for peripheral nerves, are able to secrete pro-inflammatory cytokines,⁸¹ including via a purinergic P2X7-mediated mechanism.⁸² Wallerian degeneration is a post-traumatic process of the peripheral nervous system whereby damaged axons and their surrounding myelin sheaths are phagocytosed by infiltrating macrophages or leucocytes. During the process of infiltration of inflamed or damaged peripheral nerves, these immune system cells are known to secrete pro-inflammatory cytokines.⁴⁹ Pain initiation by pro-inflammatory cytokines may be particularly important in ‘dynamic’ denervation neuropathies such as PDN or HIV-AN (Fig. 6c).

According to the concept depicted in Figure 6b, diabetic and some types of traumatic neuropathy would fall under the rubric of ‘classical denervation neuropathy’, because either there is long-term ‘dying back’ of nerve fibres such as in established PDN or severed large nerve trunks. This would also apply to severe PHN with loss of cell bodies in the DRG, or in chronic PHN when any surviving nerve fibres in skin will have sprouted to their maximum capacity. For early PHN, early PDN, HIV-AN, cancer-chemotherapy-induced neuropathy, partial nerve injury, and post-nerve repair, and others, denervation and nerve sprouting is a dynamic process which can fluctuate depending on the metabolic health of the sensory neurones and the degree to which they are exposed to toxins, etc. (Fig. 6c). Sensory neuronal cell bodies which survive, and their axons which sprout, may lead to hypersensitivity disorders. Within this group, there are chronic painful syndromes associated with overall increased nociceptor density; examples include vulvodynia,^83,84 burning mouth syndrome,⁸⁵ interstitial cystitis,^86,87 notalgia paresthetica,⁸⁸ rectal hypersensitivity,⁴³ gastro-oesophageal reflux disease,⁸⁹ inflammatory and irritable bowel diseases,^90–92 post-surgical breast pain,⁹³ and allergic rhinitis.⁹⁴ These conditions do not reflect nerve injury or disease per se, but may fall under the definition of ‘neuropathic’ pain as pain arising as direct consequence of a lesion or disease affecting the somatosensory system.⁹⁵ Denervation may be followed by aberrant re-innervation, proliferation with or sprouting of distal axons associated with these pain syndromes (Fig. 6d).

Vulvodynia constitutes a very interesting exemplar of chronic ‘dynias’ or idiopathic pain syndromes with regional changes in skin innervation. It is characterized by painful burning sensations, allodynia, hyperalgesia, and itching, usually localized in the region of the vulvar vestibules.⁹⁶ Vulvar tissue arises from the same urogenital progenitors as bladder; hence it might not be surprising to find parallels involving hyperproliferation of nociceptors in bladder tissue.⁹⁷ In vulvodynia patients, the hypersensitivity of vulvar C-fibres is well documented,^98,99 and immunohistological evaluation of small-diameter nociceptive nerve fibres shows increased densities relative to normal subjects.⁸³ Moreover, TRPV1 expression appears to be significantly increased in these proliferated nociceptors.⁸⁷ Patterns of enhanced TRPV1 expression similar to vulvodynia occur in rectal hypersensitivity syndrome, which includes faecal urgency and incontinence as symptoms.⁴³ Increases in TRPV1 expression appear to correlate with decreases in heat and distension sensory thresholds.

If the primary cause of chronic musculoskeletal pain lies deep within joints and topical capsaicin does not provide substantial transdermal delivery, then the apparent efficacy of topical capsaicin in lower back pain and OA could be postulated to be by CNS modulatory mechanisms. Another possibility is that alterations in, or sensitization of, cutaneous nociceptors could be a contributing factor. Such alterations might be driven by unusually high concentrations of cytokines or growth factors in joints which could either diffuse some distance away from the primary site of inflammation or promote excitatory phenotypes in nearby collateral axons. To assess sensory function in the skin overlying the joints of patients with rheumatoid arthritis, capsaicin was injected intradermally over those joints.¹⁰⁸ Capsaicin-induced axon reflex vasodilatation was significantly greater over affected joints when compared with age-matched normal controls and there was also a correlation between axon reflex vasodilatation and visual analogue pain score apparent in the rheumatoid group.

The theoretical concern most relevant to chronic pain management is that in peripheral neuropathies associated with cutaneous denervation, defunctionalization of cutaneous nociceptors could leave areas of skin without sufficient protective sensation to prevent or avoid injury. In addition, cutaneous C-fibres may play a role in blood flow regulation and wound repair, as the substance P and neurokinin A released by C-fibres can stimulate growth of fibroblasts and epithelial cells, while CGRP can stimulate growth of keratinocytes and epithelial cells.¹¹⁹ However, these theoretical effects have not been observed in clinical practice. The safety and efficacy of multi-week applications of low-concentration capsaicin creams (0.025 to 0.075% w/w) have been evaluated in six large double-blind trials with comparators and in other smaller studies with no reports relating to either loss of protective sensations or impaired wound healing.¹¹⁰ When the effects of topical capsaicin on sensory nerve fibre function have been examined specifically in diabetic neuropathy patients, there have also been neither detectable changes in sensory function nor adverse effects relating to cutaneous perfusion.^120,121 Similarly, in an experimental model which uses a 48 h application of 0.1% w/w capsaicin cream under occlusion to induce nearly complete loss of immunostaining for ENFs and a very robust effect on dermal nociceptors, no effect on wound healing has been reported in PDN and HIV-AN patients after repeated 3 mm punch biopsies in capsaicin-treated areas on the leg.^28,122

Limited safety data are currently available for capsaicin 8% patch for diabetic neuropathy, which may have led the EMA to restrict use of this treatment to non-diabetic neuropathy patients.⁷ Treatment with capsaicin 8% patch did not result in changes suggestive of detrimental effects on sensory function in patients with painful diabetic neuropathy¹²³ or painful HIV-AN¹² at 12 weeks after treatment. Similarly, neurosensory testing in HIV-AN and PHN patients treated with up to four treatments over a period of 48 weeks revealed no evidence of impairment.¹²⁴ The reversibility of both sensory function and ENF innervation after capsaicin 8% patch treatment has been documented in healthy subjects, with small reductions in tactile and sharp pain sensations returning to normal within 12 weeks and ENF density to within a few per cent of normal by 24 weeks.⁹ Because of the high selectivity of capsaicin for the TRPV1 receptor and the selective expression of TRPV1 in nociceptive sensory nerves, other skin sensory nerve endings may remain intact and functional even with pronounced defunctionalisation and reduction of cutaneous nociceptors.²⁶ Capsaicin-insensitive nerve endings include those which arise from Aβ-fibres which transduce tactile and proprioceptive stimuli, as well as a subpopulation of C-fibres and the majority of Aδ-fibres, which are primarily responsible for mediating pin-prick.¹²⁵ Non-TRPV1-expressing nerve fibres are also capable of transducing thermal stimuli, as they express TRP receptors such as TRPV2 (which is activated at 52°C) and TRPV3 (which is activated at 39°C).¹²⁶ It should be remembered that ENF density measurements are a diagnostic tool, one of the several used to evaluate symptomatic small-fibre neuropathies. Age- and length-related reductions of ENF density occur in healthy individuals who display no symptoms of sensory loss.¹²⁷ Therefore, perhaps it should not be surprising that the complex cutaneous innervation allows for selective defunctionalization of a subpopulation of cutaneous nociceptors (i.e. those capsaicin-sensitive) without significant loss of protective sensation. A recent study using topical capsaicin in an occlusive 48 h application reported elimination of immunostaining for both sensory and autonomic nerve markers.¹²⁸ However, such cutaneous nerve fibre losses have not been observed with either the high-concentration patch^8,9 or low-concentration creams.¹²⁹

In conclusion, directed TRPV1 agonist therapies, in which nociceptor defunctionalization is restricted to discrete target organs such as the skin, may be an attractive treatment to control localized pain or hypersensitivity. There is good evidence for hyperactivity and proliferation of cutaneous nociceptive C-fibres in numerous pain syndromes, and thus topical TRPV1 receptor agonist-mediated defunctionalisation warrants evaluation as an approach for their management. Although the ‘substance P depletion’ hypothesis was attractive in the 1980s, subsequent advances in our understanding of cutaneous innervation and the responses to topical capsaicin have rendered this hypothesis irrelevant. Reduction of substance P content in the skin is just one of many consequences of defunctionalisation, and there is no evidence that this process is related causally to pain relief.