7.1.1. Definitions of Neuropathic Itch

The epistemology of neuropathic itch (NI) is based on that of neuropathic pain, which has so far worked well, no doubt because most peripheral itch neurons are a subset of peripheral pain neurons. NI is thus defined as perception of itch in the absence of pruritogenic stimuli (formerly known to dermatologists as pruritus sine materia). In other words, there is uncoupling of the stimulus-response curve for itch sensation since dysfunction or disease of itch-signaling neurons is causing them to fire without cause. At a practical level, NI can only be diagnosed after dermatological and systemic causes have been excluded, so dermatological evaluation should usually precede referral to a neurologist.

The International Forum for the Study of Itch (IFSI) recognizes four types of itch causality—dermatologic, systemic, neurologic, and psychiatric (Ständer et al. 2007) plus mixed and unknown causes. Etiology Category III (NI) is defined as “arising from diseases or disorders of the central or peripheral nervous system, e.g. nerve damage, nerve compression, nerve irritation” (Ständer et al. 2007, p. 87). From the neurological perspective, NI is primarily a modality-specific sensory hallucination, and the literature on other types of sensory hallucinations, e.g., tinnitus, the Charles Bonnet syndrome, phantom pain (recently reviewed in Sacks [2012]) can help inform understanding.

At the time of this writing, pain specialists are digesting 2012 revisions by the International Association for the Study of Pain (IASP) to their definition of neuropathic pain, which restricts it to “pain caused by a lesion or disease of the somatosensory system” (http://www.iasp-pain.org/Content/NavigationMenu/GeneralResourceLinks/PainDefinitions/default.htm#Neuropathicpain). This replaces the definition in the IASP’s Classification of Chronic Pain (Merskey and Bogduk 1994, p. 212) that defined neuropathic pain more broadly, specifically as “pain initiated or caused by a primary lesion or dysfunction in the nervous system.” In the 2012 version, “dysfunction” has been removed and a lesion or disease affecting the nervous system is now required. The requirement for a structural abnormality is problematic because in many patients, the causal lesions can often be hard to localize, meaning that they therefore no longer qualify as having neuropathic pain (Jensen et al. 2011, 2012; Oaklander et al. 2012). Regarding pruritus, the causal pathologies are even less defined than for neuropathic pain, so the author favors the less restrictive IFSI definition of neuropathic itch that permits “disorders” as well as “diseases” of the nervous system.

7.1.2. Neuroanatomy and Cell Biology

As is typical in neurology, discovering the anatomical pathways for normal itch sensation has been aided by studying patients with neurological diseases or injuries that caused their itch. The relevant methods include localization by neurological examination, imaging (Chapter 24), and rarely, anatomical or surgical pathology. Studying patients has demonstrated that NI is caused by lesions that specifically affect itch neurons, just as other neurological symptoms reflect the functions of the specific neurons damaged. As always, the cause of the lesion is less important than its location. A stroke, tumor, or multiple sclerosis plaque in the same brain region will generate similar signs and symptoms. As imaging methods advance, it should improve our understanding of the pathways and mechanisms of NI. Better visualization of white matter tracts is particularly relevant for network disorders such as NI. For instance, applying 7T magnetic resonance imaging (MRI) to the spinal cord of a patient with NI after cavernous hemangioma (see below) suggested the hypothesis that his late-onset central itch might be triggered by delayed white matter degeneration (Cohen-Adad et al. 2012).

Peripheral nervous system (PNS) lesions appear to be the most common causes of NI, most likely because these are more common than injuries to the well-protected central nervous system (CNS). Neuropathological study shows clearly that injury to specific neurons is necessary to produce itch, namely, to the subset of sensory small fibers that encode itch signals, which includes C-fiber and A-delta primary afferents with various transduction patterns (Chapter 9). Neuropathic itch is caused by the same type of neurological injuries and diseases that cause neuropathic pain as well. Many patients present with pain alone, or pain plus itch, but some report only itch, for unknown reasons. Lesions that primarily affect motor neurons, amyotrophic lateral sclerosis, for instance, are not associated with NI. Lesions anywhere in the PNS appear capable of causing NI, whether distally (axonopathy) or centered on the sensory ganglion (e.g., shingles) or the proximal nerve roots. Even the most distal PNS lesion can affect the spinal cord and then the brain to cause NI. While cutting a nerve reduces incoming sensory signals (after the injury barrage) and thus would be expected to reduce sensation, in some individuals later compensatory mechanisms paradoxically augment sensations coming from nerve-injured areas. The causes include abnormal firing of nearby preserved primary afferents, loss of tonic dorsal-horn inhibitory circuits (Basbaum and Wall 1976; Woolf and Wall 1982), and alterations in thalamic and cortical circuits (Chapters 23 and 26).

NI involves nonneural cells and mechanisms as well. Some diseases that cause NI also cause itch through other mechanisms (e.g., cutaneous inflammation as during shingles) or the mixed itch of uremic neuropathy. Also, scratching NI-affected skin can trigger additional pruritogenic cutaneous inflammation, irritation, ulceration, or scarring. Even under normal conditions, nonneural cells participate in itch signaling. Small fibers release neuropeptides such as calcitonin gene-related peptide (CGRP) and substance P that have paracrine effects on nearby nonneural cells that augment itch. Keratinocytes express the heat-sensing vanilloid type 3 (TRPV3) receptor (Peier et al. 2002) and several neuronal sodium channels (Zhao et al. 2008) so they may contribute to or modulate normal and/or neuropathic itch. Secretions of injured small fibers stimulate mast cell degranulation, which not only initiates itch-signaling action potentials through histamine, serotonin, and tryptase-activated proteinase-activated receptor-2 receptors on primary afferent nerve endings, but additionally stimulates release of proinflammatory cytokines from T-cells to further augment inflammation and neuronal activation (Chapters 19 and 20).

7.1.3. What Is the Relationship between NI and Psychogenic Itch?

In the past, physicians routinely interpreted itch syndromes without evident cause (pruritus sine materia) as psychogenic. Recognition of NI and of how little we know about itch mechanisms is curbing this habit of blaming the patients rather than our own uncertainty. In practice, psychogenic itch diagnoses (Etiology Category IV itch; [Ständer et al. 2007]) are considered in two types of patients, those with or without major psychiatric diagnoses. For patients without psychiatric diagnoses, it is probably inappropriate for nonpsychiatrists to diagnose psychogenic itch. The psychiatric diagnoses associated with unexplained chronic itch—somatization disorder, obsessive compulsive disorder, and thought disorders—are serious illnesses with specific diagnostic criteria. If they are suspected, a psychiatrist should be consulted to diagnose and treat the patient. Even in patients with established major psychiatric diagnoses, new itch complaints will more often have dermatologic or medical causes. My few patients with obsessive-compulsive disorder plus postherpetic itch (PHI) have successfully avoided picking or scratching their itchy areas, recognizing the danger.

Psychotic patients with itch are the most difficult to evaluate because sensory hallucinations are a central symptom of psychosis. It is unclear whether or not psychiatric illness itself can trigger delusions of itch, although considering the prevalence of auditory and other hallucinations in some of the psychoses, this has be presumed possible. When psychotic patients develop unexplained itch, it is often attributed to their psychosis, despite the lack of any known specific association. Advanced imaging is now identifying abnormal brain structure and activity associated with somatic delusions and hallucinations in such patients, blurring the distinction between psychogenic and central neuropathic itch. MRI has identified reduced left fronto-insular gray matter volume in schizophrenic patients with somatic delusions as compared with schizophrenics without hallucinations and healthy controls (Spalletta et al. 2012). The thalamus, particularly the right ventral-anterior thalamic nucleus that projects to prefrontal-temporal cortex and the inferior frontal lobe, is significantly larger in hallucinating patients as well (Spalletta et al. 2012).

We all harbor illogical explanations and expectations regarding health symptoms, and some are widely accepted, e.g., health benefits of expensive organic food or colon cleanses. The most common irrational explanation for unexplained chronic itch is undiagnosed parasite infestation. It is difficult to determine in any specific patient how illogical (delusional) this is, given that the sensation of itch is thought to have evolved to signal parasite contact. Our brains have likely been hard wired to interpret itchy stimuli as insect contact until proven otherwise. However, some patients with unexplained chronic itch cling to the parasite explanation despite repeated testing showing no evidence of this, and sometimes even after other causes of itch are diagnosed.

Names for this syndrome include primary delusional parasitosis, Ekbom syndrome, and Morgellon’s syndrome. Some of these individuals have related convictions that there are tiny threads or fibers in or on their skin triggering their itch. They are a mixed bunch. Some have delusional explanation for itch that is caused by other identifiable causes (Fleury et al. 2008), whereas in others the causes of itch remain unknown, making it hard to exclude implausible explanations. Intellectual impairment and cultural and religious traditions can also make it hard for patients to evaluate the relative validity of their physicians’ explanations versus those suggested by the Internet, friends, and spiritual leaders. Low-level psychoses and schizophrenia are more prevalent than appreciated, and patients with delusional thinking and other mild psychotic symptoms often remain unrecognized even by physicians. Illogical beliefs and paranoid or conspiracy theories are part of all cultures, with no distinct border between “normal” and “delusional” thinking. Typically, such individuals remain in the community without exhibiting flagrant enough behavior to trigger psychiatric evaluation and diagnosis.

In my limited experience caring for patients harboring delusional explanations for obscure sensory symptoms, I have usually identified a neurological cause (although the patient often disagrees), a psychiatrist has confirmed the presence of mild thought disorder, and antipsychotic medications have lessened the delusions but not the sensory symptoms. Such patients are ill served by our fragmented health care system, which enables them to reject and conceal diagnoses that displease them—such as schizophreniform disorder plus atypical trigeminal neuropathy—and to decline treatment for such rejected diagnoses. Instead, they self-refer to new physicians who unwittingly repeat the cycle of unnecessary diagnostic testing and ineffectual treatments for their delusional diagnoses. Tragically, some patients impoverish themselves to “treat” their delusions, investing thousands of dollars on mold remediation, insect extermination, or unnecessary surgeries or medications.

Regardless of a patient’s own interpretation of the cause of their itch, it is their physician’s responsibility to look for “actionable” causes, including not only dermatological and systemic causes but also mastocytosis, restless leg syndrome, and neuropathic itch.

7.1.4. Animal Models of NI

Because itch is subjective, itch assessment in animals traditionally relied on measuring hair loss or wounds caused by scratching or other forms of self-injury directed toward skin that was experimentally irritated or denervated. Video systems now permit quantitation of even noninjurious scratch bouts. Although it can never be proven that scratching and similar behaviors (rubbing and biting) in animals are motivated by itch, the evidence for this overwhelming because a wide range of nonhuman animals scratch in response to the same provocations (dermatologic, systemic, and neuropathic) that cause human itching and scratching.

Until recently, human NI and associated lesions were usually attributed to psychiatric causes or pain (see trigeminal trophic syndrome below), so it is not surprising that self-injury to denervated skin in nerve-injured experimental animals was and still is often attributed to neuropathic pain rather than itch. Although neuropathic self-injury was noted from the very first nerve-injury experiments, it was first systematically studied in rodents and called “autotomy” by pain scientists Pat Wall, Marshall Devor, and colleagues, who developed a grading system for it (Wall et al. 1979). They and others initially interpreted autotomy as modeling anesthesia dolorosa type neuropathic pain, but from the beginning there was disagreement, especially from neurosurgeons who know that patients with neuropathic pain including anesthesia dolorosa protect rather than injure their painful areas (Levitt et al. 1992). Neurosurgical pioneer Will Sweet also pointed out that although dorsal rhizotomy was one of the main triggers for experimental autotomy in animals, it does not cause pain in humans (Sweet 1981). By the end of his career, Dr. Wall had come to believe that neuropathic autotomy in rodents more likely reflected neuropathic itch rather than pain (P. Wall, pers. comm.). He was additionally influenced by later reports of “autotomy” in nerve-injured humans who reported that their self-injurious scratching was motivated by itch and not by pain (Oaklander et al. 2002). Other scientists use the term “overgrooming” for pathological scratching at experimentally nerve-injured skin. These behaviors develop in some but not all rodents after experimental lesions throughout the PNS and in the spinal cord that damage pain/itch pathways (Yezierski et al. 1998). Genetic (Mogil et al. 1999) and dietary factors (Shir et al. 1997) influence whether or not rats develop autotomy after nerve injuries and demonstrate that NI is a complex disorder with both environmental and genetic causes.

Widespread NI caused by polyneuropathy has been modeled by treating neonatal rodents with systemic capsaicin, which permanently ablates TRPV1+ primary afferents and triggers 80% to 90% of rats to engage in self-injurious overgrooming (Maggi et al. 1987). Interestingly, this is largely restricted to the areas around the eyes, ears, and snout, consistent with the idea that these locations may be enriched in itch receptors, perhaps because their moistness and lack of keratinization attracts insects (Maggi et al. 1987). It is unclear whether or not neuropathies that damage axons only while preserving the cell bodies in the sensory ganglia also produce NI.

Focal NI caused by nerve injury has been modeled by studies of autotomy after sciatic nerve transection in rodents. Action potentials from the damaged nerve appear required to maintain autotomy, because chronic administration of local anesthetics to a sciatic nerve that completely suppresses sciatic action potentials does not trigger autotomy (Blumenkopf and Lipman 1991).

Dermatomal NI caused by ganglion lesions has been modeled by surgical extirpation of specific dorsal root ganglia. The resultant overgrooming centered on the denervated dermatome is associated with ectopic firing within the dorsal horn (Asada et al. 1996) and modulated by descending dopaminergic inhibition (Tseng and Lin 1998).

Dermatomal NI caused by radiculopathy is modeled by the autotomy that develops after lesioning rodent dorsal rootlets or roots. This seems to correspond to the rare phenomenon of self-injury after plexus injuries in humans (e.g., McCann et al. 2004). As in humans, there seems to be a rostro-caudal gradient, or at least for developing autotomy as this develops after dorsal rhizotomy of cervical and thoracic roots, but not lumbosacral nerve roots (Lombard et al. 1979).

NI caused by myelopathy has been linked by my group to a rat model of spinal-cord injury, involving micro-injection of quisqualate excitotoxin into the deep dorsal horn, which creates a small necrotic cavity (Dey et al. 2005). The dermatomal scratching and biting that some injected rats develop was interpreted as modeling neuropathic pain until we suggested it more closely modeled NI in humans with intramedullary cavernous hemangiomas (see below). We later found that injected rats that self-injure have deafferentation of their itchy dermatomes, implying that their small spinal cord injuries cause retrograde degeneration of primary afferent neurons as well as of intrinsic spinal neurons (Brewer et al. 2008). Thus, myelopathy-induced NI may not be purely central. This model also provided the information that only injected rats with lesions in the deep dorsal horn that spared the superficial lamina developed dermatomal self-injury (Yezierski et al. 1998). NI in these rats may thus be maintained by spontaneous activity in second order, lamina I, itch projection neurons responding to both loss of local inhibitory circuits (Ross et al. 2010) but also loss of normal peripheral input.

NI caused by brain lesions has not been systematically modeled in animals.

7.3.1. Widespread NI from Generalized Peripheral Nerve Damage (Polyneuropathies)

Widespread small-fiber predominant polyneuropathies (SFPN) typically include symptoms of burning pain, allodynia, hyperalgesia, sweating, and microvascular changes. Itch is currently not adequately recognized as part of this syndrome (Figure 7.1). SFPN has metabolic (e.g., diabetes), infectious, rheumatologic, toxic, paraneoplastic, autoimmune, and hereditary causes, some of which are curable and thus a diagnosis mandates additional testing for cause (Amato and Oaklander 2004). Most small-fiber polyneuropathies are length dependent, beginning with foot pain that can progress proximally in a stocking-and-glove pattern. Occasional patients present with patchy, proximal, or total-body sensory symptoms caused by widespread ganglionopathy/neuronopathy; described below. If large (A alpha or beta) myelinated nerve fibers are also affected, there can be colocalizing hyporeflexia, weakness, and vibratory and proprioceptive deficits.

FIGURE 7.1

Widespread neuropathic itch caused by small-fiber polyneuropathy. A patient with chronic itch and scratching-induced lesions on her distal legs. She had negligible sweat production from the forearm, proximal leg, distal leg, and foot sites after iontophoresis (more...)

Toxic polyneuropathy caused by infusing hydroxyethyl starch (HES), a synthetic colloid widely used to increase intravascular volume in surgical, emergency, and intensive care patients, can cause widespread NI plus kidney and coagulation disorders. These are attributed to starch deposition in tissues including kidney and peripheral nerve (Kamann et al. 2007). Neuropathy symptoms typically include diffuse itching and sensorineural hearing loss (Klemm et al. 2007). It is debated whether or not newer formulations are safer (Hartog et al. 2011).

7.3.2. Focal NI from Nerve and Nerve-Root Lesions Including Brachioradial Pruritus and Notalgia Paresthetica

Any type of injury to itch-signaling peripheral afferent neurons can trigger itch in and near the receptive field of the injured neurons. As with neuropathic pain, it is entirely unknown why only a few patients with such injuries (which are common) develop clinically significant NI. One association is with neurofibromas, benign tumors that arise from Schwann cells. The itch is attributed to the abundant mast cells within these tumors (Johnson et al. 2000).

Radiculopathies involve damage to cranial or spinal nerve roots. These are established causes of neuropathic itch (see below) and other sensory symptoms including pain, and/or reduced sensations in the territory of affected nerve roots. If severe, patients can develop colocalizing motor deficits. Mechanical compression or irritation at the neural foramen (“pinched nerve”) is the most common cause; others include inflammation or neoplasm (e.g., meningioma and schwannoma) and even Tarlov cysts (Oaklander et al. 2013).

Plexopathies refer to lesions of the intertwining neural networks near the spine where axons within nerve roots re-sort into peripheral nerves. Symptoms reflect often patchy or incomplete combinations of nerve root and nerve damage. The brachial plexus is affected more often than the lumbosacral plexus, and motor fibers are affected along with sensory fibers. Movements that irritate the affected plexus can worsen sensory symptoms. Common causes include autoimmunity (e.g., in brachial plexitis, also known as neuralgic amyotrophy or Parsonage-Turner syndrome [van Alfen and van Engelen 2006]), tumor or radiation therapy, diabetes, and tissue entrapment. Additional causes of brachial plexopathy include Pancoast lung tumors, neurogenic thoracic outlet syndrome, and SEPT9 mutations, which cause recurrent attacks (Jeannet et al. 2001). In children, traction injuries during birth have been associated with self-mutilation of the affected arm (McCann et al. 2004). Although such children are preverbal, and others have interpreted such behaviors as a sign of psychopathology or neuropathic pain, the author continues to interpret them as painless self-injury caused by the conjunction of neuropathic itch and loss of protective nociceptive pain sensation from severe sensory-nerve injury (Brewer et al. 2008; Dey et al. 2005), as in the trigeminal trophic syndrome discussed below.

Mononeuropathy, meaning damage to a single nerve, is most often caused by trauma. While originally associated with military trauma, medical injuries are now a more common cause of traumatic nerve injury in developed countries. Less frequent causes include internal lesions such as impingement, entrapment, scarring, infection (e.g., leprosy), or inflammation. Because some causes are curable or require independent treatment (e.g., tumor and aneurysm), focal peripheral neuropathies of unclear etiology require additional evaluation to identify their cause. Symptoms can spread outside the dermatome of the affected nerve (e.g., in the complex regional pain syndrome) owing to electrical coupling in the skin between nociceptive afferents (Meyer et al. 1985), secondary irritation of nearby neurons within nerve trunks, nerve roots or the DRG, the dorsal horn, and occasionally even the sensory cortex of the brain. Furthermore, there are poorly understood links between contralateral anatomically matched primary afferents that allow some types of unilateral nerve injuries to produce “mirror-image” effects on the contralateral uninjured areas (Oaklander et al. 1998; Oaklander and Belzberg 1997; Oaklander and Brown 2004; Scott et al. 2000). Induction or exacerbation of symptoms by percussing the involved nerve (Tinel’s sign) can help with localization. Diagnostic aids include electrophysiological testing and advanced imaging modalities (see below).

Brachioradial pruritus is the historic term referring to NI originating from the cervical spinal nerves. NI characteristically presents with bilateral or less often unilateral itchy patches (with or without cutaneous stigmata of scratching) on the forearms, elbows, or upper outer arms (Massey and Massey 1986). Because of the location, and because most patients have worsening in the summer and improvement with protection against the sun (Veien and Laurberg 2011), ultraviolet exposure may be a contributory factor, but one third to one half of patients have radiological evidence of degenerative osteoarthritis of the spine compressing their cervical nerve roots (Abbott 1998; Goodkin et al. 2003; Veien and Laurberg 2011). As below, this same phenotype is less often caused by spinal cord lesions (myelopathy) including tumors (Kavak and Dosoglu 2002), so cervical MRI should be considered in patients without electrophysiological or other evidence of a peripheral lesion, particularly in patients with bilateral symptoms or signs or other symptoms suggestive of cervical myelopathy. A single-case report suggests efficacy of topical injection of botulinum toxin-A (Kavanagh and Tidman 2012).

Notalgia paresthetica is another well-known focal neuropathic itch syndrome. This historical term describes usually unilateral itchy areas on the back, near and below the scapula in the mid-thoracic dermatomes (Massey and Pleet 1979). These areas often exhibit the stigmata of chronic scratching as well, and dermatopathological examination reveals postinflammatory hyperpigmentation (Savk et al. 2002). Focal entrapment or irritation has been attributed to the location where nerves turn to traverse through the muscles of the back at an angle, and indeed, electromyographic study showed that seven of nine patients with notalgia paresthetica had evidence of chronic denervation of paraspinous muscles; localizing the causal lesion to the posterior rami of the spinal nerves (Massey and Pleet 1981). Radiological study also provides evidence of degenerative spine disease in up to 60% of studied patients so this same phenotype can equally be caused by radiculopathy (Raison-Peyron et al. 1999; Savk and Savk 2004, 2005), and no doubt by zoster sine herpete. Additionally, given the fact that even intramedullary lesions including spinal cord tumors can present as notalgia paresthetica (Johnson et al. 2000), thoracic MRI should be considered in notalgia paresthetica patients without electrophysiological evidence of a peripheral cause, particularly in those with bilateral symptoms or other signs and symptoms suggestive of myelopathy. Regarding treatment of notalgia paresthetica, there are case reports of efficacy of various neuropathic-pain medications including topical capsaicin and oxcarbazepine (Savk et al. 2001). Exercises to alter muscle configuration and presumably reduce nerve impingement have also been recommended (Fleischer et al. 2011; Savk and Savk 2004). Prospective study of injection of botulinum toxin type A in four patients with notalgia paresthetica, one with meralgia paresthetica and one with neuropathic itch of the foot showed mean reductions of itch by 28% at 6 weeks posttreatment (Wallengren and Bartosik 2010).

7.3.3. Focal NI from Sensory Ganglia Lesions (Neuronopathies/Ganglionopathies) Including Shingles

Attacks on the sensory ganglia (cranial and dorsal root ganglia) cause sensory symptoms centered on the peripheral distribution of these ganglia. Motor function is usually spared because motor axons bypass most sensory ganglia, although motor axons may undergo secondary bystander damage due to proximity to degenerating sensory axons within the nerve roots, plexi, or nerves. Multiple sensory ganglia can also be affected by generalized conditions to cause widespread or patchy symptoms (polyneuropathy). Because sensory ganglia lack a blood–nerve barrier, they are vulnerable to autoimmune attack such as in Sjögren’s, and paraneoplastic syndromes. Given the abundance of itch-associated primary afferents within the DRG (Han et al. 2012), the strong link between ganglionopathies and NI is no surprise.

Shingles (herpes zoster) is by far the most common type of sensory ganglionitis, and it is currently the best characterized cause of focal NI (Figure 7.2). Although clinicians recognized that dermatomal itch was a common sequel of zoster, and there were scattered reports, one well-studied patient who scratched through her skull into the frontal lobe of her brain after V1 shingles, brought much wider awareness of PHI (Oaklander et al. 2002). A global epidemiologic study of 586 adults with prior shingles demonstrated that itch, usually mild or moderate, affected roughly one third of patients with acute shingles or PHN. Furthermore, shingles affecting the face, head, or neck was significantly more likely to trigger PHI than shingles affecting more caudal dermatomes (Oaklander et al. 2003).

FIGURE 7.2

Facial neuropathic itch caused by shingles (trigeminal trophic syndrome). A 21-year old woman with trigeminal trophic syndrome in the right V1 dermatome after cranial irradiation for pituitary adenoma followed by right V1 herpes zoster. Her ulceration (more...)

7.3.4. NI Syndromes Arising from Lesions within the Spinal Cord (Myelopathy)

These are well-recognized causes of NI. It is not always clear whether the causal lesions are affecting the PNS or CNS, because primary afferent neurons extend into the spinal cord. Many peripheral sensory neurons synapse segmentally on projection neurons (e.g., spinothalamic tract neurons; see Chapter 22), but the central axons of some unmyelinated fibers as well as of myelinated fibers ascend in the dorsal columns (Briner et al. 1988). Given the small size of the spinal cord, many lesions likely affect all these compartments (Brewer et al. 2008).

It is important to recognize that NI can be the presenting sign of spinal cord injury because this requires neurological evaluation and treatment if possible. Scattered reports associate NI with various causes of myelopathy including tumors (Johnson et al. 2000; Kavak and Dosoglu 2002), multiple sclerosis (Yamamoto et al. 1981), syringomyelia (Kinsella et al. 1992), and traumatic injury causing the Brown-Séquard syndrome (Thielen et al. 2008). These authors attributed their patient’s NI to axotomy of itch neurons crossing the ventral commisure en route to the contralateral anterolateral spinothalamic tract. Paramedian lesions may similarly downregulate the existing tonic inhibitory control of itch in the spinal cord, e.g., by bhlhb5 expressing neuron (Ross et al. 2010). There are more case reports of rostral than caudal lesions. One cervical ependymoma tumor extending between C4-C7 presented with brachioradial pruritus on both arms, elbows, and forearms as the only symptom (Kavak and Dosoglu 2002).

The best documented association is with inflammatory myelopathies. Among a series of 377 MS patients, 4.5% reported NI (Matthews 2005). Itch was even more common in myelopathy from neuromyelitis optica (NMO), a relapsing autoimmune/inflammatory demyelinating disorder of the spinal cord and brain (including the optic nerves) associated with aquaporin-4 autoantibody (AQP4-Ab) (Elsone et al. 2012). Twenty-seven percent among 44 patients with AQP4-Ab-positive NMO myelitis had NI attributed to their spinal cord lesion. Itch affected the trunk in 67%, the limbs in 75%, and the occiput in 25% of patients, most often accompanied by other colocalizing sensory symptoms (Elsone et al. 2012). About half had continuous pruritus whereas half had intermittent symptoms; median itch intensity was 6 of 10 (Elsone et al. 2012). The authors postulated that itch was more common in NMO than in MS because MS affects mostly white-matter tracts, whereas NMO primarily damages gray-matter neuronal cell bodies, including the lamina 1 second-order dorsal-horn neurons that process itch signals.

The second best documented association between spinal-cord injury and itch is with cavernous hemangiomas of the spinal cord (Cohen-Adad et al. 2012; Dey et al. 2005; Sandroni 2002; Vuadens et al. 1994). The author suggested that features of these rare lesions make them pruritogenic, specially their preferred location in the upper rather than the lower spinal cord and perhaps in the dorsal rather than ventral cord (although dorsal cavernomas may simply be those that are more operable and come to neurosurgical attention), plus the tendency of these lesions to trigger gliosis and bleeding. Cavernomas are probably pruritogenic for the same reasons that they are epileptogenic when in the brain; gliosis and hemosiderin deposition both excite ectopic action potentials (Dey et al. 2005). Of note, our rat model of NI from spinal cord cavernoma (see above) includes both these features, and the cervical syrinx associated with NI contained hemosiderin consistent with prior bleeding as well (Kinsella et al. 1992).

7.3.5. NI Syndromes Affecting the Head and Neck (Cranial Nerves and Ganglia)

Lesions of any cranial nerve that contain somatosensory representation, its ganglion, or its central projections can cause NI and/or neuropathic pain that localizes to the region innervated by that nerve’s sensory axons. In the head and neck, this includes VI (nervus intermedius) and IX in addition to V. At least one case of neuropathic self-injury to a tonsil has been associated with damage to IX (Stefaniu et al. 1973). Although any type of lesion or injury can trigger neuropathic sensory symptoms, the most common causes are vascular compression (e.g., tic douloureux) or zoster, which can affect VI and IX as well as all three branches of V (of which involvement of V₁ or herpes zoster ophthalmicus is most common).

Cranial neuropathies are disproportionately likely to trigger NI as compared to lesions affecting nerves innervating the lower torso, legs, and the sacrally innervated dermatomes. The reasons are unknown, but it has been hypothesized that they might be related to protecting the facial mucosa, a prime target of insect attack (Oaklander et al. 2003). In PHI, after shingles, data from several independent groups confirm that higher proportions of patients with shingles of the head or neck develop PHI than patients with zoster on the torso (Oaklander et al. 2003). A similar gradient in susceptibility to autotomy after dorsal rhizotomy is described in rats (Lombard et al. 1979). Despite its proximity to the brain, neuropathic itch of the face, head, and neck is more likely to be caused by lesions of peripheral rather than central neurons. The discussion above pertaining to ganglionopathies and radiculopathies applies equally to the cranial nerve ganglia.

7.3.6. Trigeminal Trophic Syndrome and Other Forms of Self-Injurious Behavior from NI

Group III itch according to the IFSI classification (Ständer et al. 2007) includes self-injurious scratching or mechanical stimulation of the receptive field innervated by injured sensory nerves. Such self-injury was for many years attributed by most to an aberrant response to neuropathic pain (Mailis 1996; Rapin and Ruben 1976) plus or minus coexisting psychopathology. Few patients were ever actually asked if they felt itch. Some authors mention vague descriptors of itch (e.g., aversive sensations and formication). Some patients had itch and pain, making it difficult to be certain which motivated their self-stimulation. However, given that pain triggers protective behaviors such as withdrawal, whereas itch triggers scratching, pain seems implausible as a motivation for self-injury. Publication of the case of a woman who scratched through her skull because of intractable PHI (without postherpetic pain) (Oaklander et al. 2002), helped demonstrate the self-injurious scratching was a marker of neuropathic itch colocalizing with loss of protective pain sensation. Also, see the discussion above of animal models regarding autotomy.

A contributing factor to such self-injurious behavior is many patients’ (and even their physicians’) unawareness that their cutaneous lesions are the consequence of their itch and not the cause of it. Teaching how to break the itch–scratch cycle is imperative. The confusing nomenclature for the skin changes caused by excessive scratching (e.g., prurigo nodularis, lichen planus, and macular amyloidosis) can obscure the true cause, which, in many cases, is chronic itch. Because scratching can occur while patients are sleeping or inattentive (Savin et al. 1973, 1990; Savin 1975), barriers to scratching (e.g., wearing mittens during sleep), and behavioral modification can be effective. Topical local anesthetics can also be helpful, testifying to the importance of remaining primary afferent signaling in maintaining this condition.

My explanation is that severe damage to primary afferent pathways can leave isolated remaining itch axons firing ectopically (without provocation) but in insufficient numbers to generate the powerful surround fields of inhibition that normally laterally inhibit itch in the dorsal horn (Oaklander et al. 2002). This is consistent with the fact that scratching, which augments nociceptive afferent firing in and around the receptive field, provides transient relief, and with reports that administering minute amounts of pruritogens to the skin using cowhage spicules produces itch disproportionate in intensity to the dose of pruritogen or to the size of the stimulated area (Sikand et al. 2011).

Trigeminal trophic syndrome (TTS) is a historical term referring to self-induced traumatic lesions of the head and face caused by the most severe forms of NI. The worst cases are literally disfiguring and even potentially fatal. Other historical names include trigeminal neuropathy with nasal ulceration, trigeminal neurotrophic ulceration, ulceration en arc, and trophic ulceration of the ala nasi (Jaeger 1950; Ziccardi et al. 1996). Attributed first to loss of a “trophic” substance normally supplied by axons, then to psychopathology during the Freudian era, TTS is now attributed to the disastrous conjunction of intractable neuropathic itch plus colocalizing loss of protective pain sensation caused by severe nerve injury. Some but not all patients have colocalizing neuropathic pain as well.

TTS was reported and studied in numerous publications including large case series by early twentieth century neurosurgeons including Harvey Cushing. It occurred then as an adverse effect of trigeminal rhizotomy, the only logical treatment for trigeminal neuralgia in the premedication era (Becker 1925; Cliff and Demis 1967; Cushing 1920; Harper 1985; Henderson 1967; Howell 1962; Jefferson and Schorstein 1955; Karnosh and Scherb 1940; Kavanagh et al. 1996; Knight 1954; Loveman 1933; McKenzie 1933; Philpott 1941; Schorstein 1943). TTS was also reported after other iatrogenic injuries to the trigeminal ganglion and root including injection of neurotoxins or thermal injury (Harris 1940; Walton and Keczkes 1985). Rare causes include infections such as syphilis, leishmaniasis, and leprosy (Thomas et al. 1991). Shingles currently appears to be the most common peripheral cause (Figure 7.2). The TTS syndrome can also be produced by brain lesions, most often stroke (see below). There are scattered reports of self-injurious scratching inside the mouth or nose. One patient whose “burning mouth syndrome” of unknown etiology apparently included itch, although itch was not explicitly mentioned, used “mechanical manipulation” inside her mouth to “obtain relief” to the point of causing superficial lesions. Her symptoms (and erosions) resolved after gabapentin treatment (Meiss et al. 2002).

In addition to the treatments discussed below for NI, patients with self-injury from NI require additional treatment. Most important is to make sure that patients understand that their lesions are self-induced. Behavioral modification including barriers to protect the affected area and wearing mittens during sleep can impede scratching. TTS injuries may require treatment of infections, wound management, and surgical reconstruction of defects. This must be done with innervated flaps as grafts of denervated tissues from the affected areas are less viable and patients continue to damage them (McLean and Watson 1982). Other strategies to modulate peripheral activity in itch neurons, such as transcutaneous electrical stimulation (Westerhof and Bos 1983) are worth considering.

7.3.7. Neuropathic Itch Syndromes Arising from the Brain

Brain lesions that cause NI validate the concept of “central itch”; that lesions of second- and third-order CNS itch neurons are also capable of triggering it. They demonstrate that NI is a systems or network disorder caused by imbalances between excitatory and inhibitory transneuronal signaling, rather than by consequence of lesions in one specific location. Case reports demonstrate that any type of brain lesion that affects the brain’s itch neurons can cause central neuropathic itch, including TTS (King et al. 1982). Central lesions were said to cause about one fifth of cases of TTS (Spillane and Wells 1959).

Stroke is the most common brain lesion associated with NI, particularly infarctions of the lateral medulla (Wallenberg’s syndrome; Figure 7.3) or slightly more rostral strokes in the lateral pons where itch signals presumably ascend (Curtis et al. 2012; Dick and Gonyea 1990; Fitzek et al. 2006; Massey 1984; Savitsky and Elpern 1948; Shapiro and Braun 1987; Wallenberg 1901). Even today, some stroke specialists are not aware of these associations (Oaklander et al. 2009). Strokes that only cause weakened chewing or reduced facial touch with preserved pain sensation do not cause NI and TTS, only those that damage nociceptive pathways and cause profound sensory loss (Fitzek et al. 2006). Such lesions identify the location of ascending itch pathways (see Chapter 22). Neurophysiological study of nonhuman primates has established that itch-signaling spinothalamic tract neurons project to the ventral posterior lateral nucleus, ventrobasal complex, and the posterior thalamic nuclei (Davidson et al. 2012; Simone et al. 2004). Correspondingly, NI has also been associated with Dejerine-Roussy syndrome (also known as the posterior thalamic syndrome and retrolenticular syndrome), caused by lesions of the ventral posterior thalamus and internal capsule (King et al. 1982). Better known symptoms include hemianesthesia and hemi-pain contralateral to the lesion. One of few studies that correlated sensory disturbances with anatomical localization of infarctions of the posterior cerebral artery (it supplies the ventrolateral thalamic sensory nuclei and white-matter tracts to the somatosensory cortex) found that all patients with sensory findings had thalamic damage, and that among patients with sensory findings, 11 of 15 had infarctions in the ventrolateral thalamus in the territory of the thalamogeniculate or lateral posterior choroidal arteries (Georgiadis et al. 1999). However, as in most such studies, itch was not specifically studied.

FIGURE 7.3

Focal neuropathic itch due stroke (trigeminal trophic syndrome). A woman with trigeminal trophic syndrome (TTS) as part of Wallenberg syndrome: (a) Location of her lateral medullary infarction on imaging and (b) excoriations of the itchy areas of her (more...)

Interestingly, Wallenberg’s original patient with TTS had shingles in his stroke affected area as well, so the exact cause of that patient’s NI remains uncertain (Wallenberg 1901; Wolf 1971). Less frequent brain lesions associated with NI include multiple sclerosis (Matthews 2005; Ostermann 1976; Yamamoto et al. 1981), tumors within or near the brain (Andreev and Petkov 1975; Summers and MacDonald 1988), and infections (Sullivan and Drake 1984). NI or scratch marks that spread beyond the dermatome of one individual trigeminal division suggest a causal lesion in the ganglion or brain rather than in the PNS.

Severe pruritus unresponsive to conventional treatments has been associated with several prion diseases including Creutzfeldt-Jakob disease (CJD) in humans. Scrapie (Petrie et al. 1989), which develops in sheep and goats in western Europe, is named for the characteristic injurious self-directed scratching and biting. This most often affects the rump, unlike the rostral predominance typical of other human and animal forms of NI. NI appears to occur only rarely in sporadic CJD (Shabtai et al. 1996) but was reported by 19% among 31 patients with familial CJD (Cohen et al. 2011). The itch was generalized in three patients, regional in two and localized in one patient. It was transient in one patient and continued throughout the disease in five patients. Diffusion weighted imaging associated the itch with reduced diffusion in several brain areas known to be affected by CJD, but most significantly with the midbrain periaqueductal gray. The authors postulated damage to inhibitory gating mechanism for itch however, because prions are deposited in the skin and in the peripheral nerves in CJD and other human prion diseases (Lee et al. 2005), the itch signals might also have conceivably been generated in the PNS.

In addition to the brain stem and posterior thalamus, higher brain areas participate in itch perception and processing (see Chapters 23 and 24). Positron emission tomography in normal adults implicates primary but not secondary somatosensory cortex (Brodmann 24) as well as the periaqueductal gray and cingulate gyrus in histamine-induced itch (Drzezga et al. 2001; Hsieh et al. 1994; Mochizuki et al. 2003). However, the secondary somatosensory cortex and insula have been implicated in electrically induced acute itch (Mochizuki et al. 2009). There are very rare reports of patients with small somatosensory-cortex lesions and somatotopically appropriate NI, usually associated with seizure (Maciejewski and Drop 2002) or subarachnoid hemorrhage (Canavero et al. 1997), consistent with cortical hyperactivity rather than degeneration.

7.3.8. Phantom Itch in Amputated Body Parts

Phantom itch is well described and even known to the public. It can occur after amputation of any innervated body part. Among the roughly 60% of women with phantom sensations after mastectomy, itch was the most common (Lierman 1988). The prevalence of phantom itch demonstrates the critical role of denervation in generating neuropathic itch; phantom itch is an extreme endophenotype of peripheral NI, which arise in the context of sensory deafferentation. Because the itchy area is missing and thus cannot be scratched, phantom itch is not associated with self-injury. It remains uncertain exactly which neurons fire excessively to generate phantom itch sensation. On the basis of the more plentiful studies of postamputation pain, possibilities include spontaneous activity of isolated peripheral nociceptors, perhaps those that are regenerating or entrapped in scar tissue (as with postburn itch) or neuromas. These can generate “stump” sensations perceived in remaining tissue as well as sensations perceived in prior receptive fields that have been amputated. Edge effects at the border of innervated and noninnervated tissues may also contribute. However, central mechanisms, whether originating in the spinal cord or brain, are paramount in phantom sensations including the Charles Bonnet syndrome after visual loss and tinnitus after hearing loss (Jastreboff 1990; Schultz and Melzack 1991). Phantom itch can only be studied in humans who can report it, but there are few or no studies of patients with phantom itch independent of phantom pain. It is likely that thalamic plasticity, including expansion of the representation of the proximal limb into thalamic region that used to represent the amputated part contributes, as well as unmasking of latent receptive fields (Dostrovsky 1999).

Treatment options primarily involve medications shown effective for phantom pain and other sensations, such as the gabapentinoids and tricyclics. Several methods involve increasing remaining afferent input. This includes behavioral therapies such as wearing a prosthesis to provide visual and functional input (Lierman 1988), mechanically stimulating the area or the contralateral body part and “mirror therapy” (Chan et al. 2007). Neuromodulation may have similar effects by triggering action potentials at various parts of the itch network independently of afferent input (Ahmed et al. 2011). One patient reportedly found relief from phantom itch and pain by scratching or massaging his prosthesis or the leg of another person (Weeks and Tsao 2010). The authors hypothesized that sensory mirror neurons permitted integration of this feedback into his own sensory circuits.

7.4.1. Conservative Treatments

Options to consider in all patients include those that minimize secondary itch and inflammation caused by scratching, such as maintaining good skin hydration, as well as barriers to reduce scratching. Barriers should be used in most patients with self-injury from scratching NI, particularly while patients sleep and scratch (Savin et al. 1973). Options include a lockable helmet to protect the scalp in patients with V1 TTS, rigid surgical braces to protect the limbs and torso, and wearing mittens at night. Video recordings document that patients with severe itch engage in nocturnal scratching while sleeping and are unaware of their actions. Cognitive behavioral therapy may be required to help break the itch–scratch cycle or at least to substitute less destructive methods for scratching. Other behavioral changes to consider include protecting affecting areas from itch-triggering stimuli (e.g., contact with clothing) and instituting low levels of tonic mechanical stimulation. Wearing a close-fitting binder or wrap can be helpful in several ways. Exercises to reduce nerve impingement in notalgia paresthetica have also been recommended (Fleischer et al. 2011; Savk and Savk 2004). Consideration should also be given to ways to reducing barriers to axonal regeneration and healing, among which smoking cessation and regular exercise to improve tissue perfusion are paramount. However, many patients will require drug therapies as well.

7.4.2. Topical Application of Medications

Topical treatments that have no systemic absorption and hence no systemic adverse effects are appealing. Among them, the local anesthetics are paramount. These penetrate mucous membranes well (with some potential for systemic absorption) and thus should be considered for chronic itch affecting the mucosa. They are not administered to the eye because of the potential for suppressing protective pain sensation. On the skin, local anesthetics require an occlusive wrap or patch (e.g., 5% lidocaine patch) for penetration. Rare patients with significant self-injury require daily subcutaneous injection of long-active local anesthetics. Rarely, have caregivers or patients with V1 PHI been taught to perform supraorbital nerve blocks at home.

The utility of topical capsaicin (8-methyl-N-vanillyl-6 nonenamide) is uncertain. Punctate or miniscule experimental doses can cause itch (Sikand et al. 2011). When applied to larger areas, topical capsaicin induces predominantly pain, which inhibits itch. Over-the-counter products containing low-concentration capsaicin (usually 0.025%–0.075% w/w) have marginal efficacy against various pain conditions in clinical studies, but they require dosing three to five times daily. Systematic review found no convincing evidence for the use of low-dose topical capsaicin to treat pruritus in any medical condition, although the quality of most of the trials reviewed was poor (Gooding et al. 2010). Practically speaking, most patients find capsaicin application painful and few continue it. Scientifically speaking, capsaicin causes degeneration of distal TRPV1⁺ axon terminals (Simone et al. 1998), producing the same kind of axonal injuries that are associated with neuropathic pain and itch so there is at least potential for delaying or impeding axonal regeneration. A single-dose high-concentration patch containing 8% w/w is currently licensed in the European Union for treating peripheral neuropathic pain in nondiabetic adults and in the United States for treating PHN, but PHI was not studied. Case reports support utility of other topical treatments such as topical tacrolimus and gabapentin (Nakamizo et al. 2010).

7.4.3. Intradermal Injection of Botulinum Toxin Type A

BTX-A administered by local subcutaneous injection is increasingly used to treat peripheral neuropathic pain. Anti-itch effects are independent of its better known ability to weaken cholinergic neuromuscular transmission. The antipruritic benefits are based on inhibition of cutaneous mediators of neurogenic inflammation including substance P, CGRP, and glutamate, as well as inhibition of vanilloid receptor activity (Ranoux et al. 2008). For this reason, BTX-A is being explored in itch, and there is prospective evidence of efficacy in histamine-induced experimental itch (Gazerani et al. 2009). Isolated case reports (Kavanagh and Tidman 2012) and a very small study of four patients with notalgia paresthetica, one with meralgia paresthetica and one with neuropathic itch on the upper foot that showed a mean reduction of VAS by 28% at 6 weeks posttreatment (Wallengren and Bartosik 2010), suggest efficacy so this is worth considering. As of 2012, an industry-sponsored phase III trial of BTX-A (Xeomin) was registered with the US NIH’s ClinicalTrials.gov.

7.4.4. Systemic Pharmacologic Treatment

There are no controlled trials (Class I–III) of treatments for neuropathic itch; hence the recommendations below are based on Class IV evidence such as case reports or expert opinion. General recommendations include beginning with one of the medications discussed below, then raising the dose until relief is obtained or persistent adverse effects ensue. During a medication trial the dose should usually be raised to the maximum tolerated before discontinuing that medication and trying another; preferably from a different class. If partial relief is achieved from a well-tolerated medication, the second one added should be from a different class. Too often, patients are treated with low doses of multiple medications making it difficult to assess each one’s efficacy or to attribute adverse effects.

Current options include systemic administration of medications that reduce ectopic neuronal firing, such as cation-channel antagonists (e.g., anti-epileptic medications), or those that augment inhibitory circuits (e.g., tricyclics). While many of these are also effective for neuropathic pain, not all are. Opioids—which are effective for neuropathic pain—trigger pruritus, particularly when administered near the spinal cord (Jinks and Carstens 2000). Isolated case reports, mostly in TTS patients, support use of carbamazepine (Bhushan et al. 1999), gabapentin (Macovei et al. 1991), amitriptyline (Finlay 1979), and pimozide (Duke 1983; Mayer and Smith 1993). One patient with intractable trigeminal itching and scratching after stroke plus shingles found efficacy of bupivicaine and clonidine administered intrathecally through a high thoracic catheter (Elkersh et al. 2003). As in neuropathic pain (Challapalli et al. 2005; Ferrini and Paice 2004), continuous systemic administration of low doses of local anesthetics by continuous subcutaneous administration at home, is a powerful treatment option that is underutilized. Brief courses of intravenous local anesthetics can be used for an in-hospital trial with cardiac monitoring to determine if there is a dose that is effective without producing adverse effects (Watson 1973).

7.4.5. Neurosurgical Treatment Options

Prior to the mid-twentieth century development of effective drugs, surgery was the only treatment for morphine-unresponsive neuropathic pain (Warren 1828). The first neurosurgeon, Harvey Cushing, established that ablative neurosurgery (cutting nerves carrying sensation from painful areas) is usually ineffective and in some cases triggers intractable anesthesia dolorosa pain (Cushing 1920). In contrast, surgical decompression or removal of structural lesions causing sensory symptoms including NI is potentially curative, even in medically refractory cases. Occasional rare causes of focal NI such as tumors or vascular malformations may require nerve surgery for definitive treatment. These surgeries are currently underutilized, in part because they requires skilled diagnosis and lesion localization by history, examination, electrophysiological study, or imaging before surgery. Itch is not yet adequately recognized as a potential localizing sign for structural nerve and root injuries.

Augmentive neurostimulation using implanted bipolar electrodes has been proven effective for various neurological symptoms including neuropathic pain, and has been used for decades. Its major advantage compared to medications is that its tiny electrical currents only affect nearby cells, unlike drugs that that can spread through the body to cause diverse adverse effects. Neurostimulation has few if any systemic adverse effects. To date, there has been little or no research on potential benefits for NI. Stimulation locations used for neuropathic pain and that should be considered for refractory NI, include proximal to a focal nerve injury (Campbell and Long 1976), the dorsal column of the spinal cord, the motor cortex, and the thalamus or periaqueductal gray (Nguyen et al. 2011). The major disadvantage of neurostimulation has been the requirement for surgical implantation. However, neurons can increasingly be activated from outside the body by electromagnetic induction or direct current stimulation. Noninvasive transcranial magnetic stimulation (TMS) of the brain is an innovative alternative with much lower risk and cost. Repetitive TMS was FDA approved for treating major depression in 2008, and several small clinical trials suggest efficacy for chronic neuropathic pain (Leung et al. 2009), so TMS should be considered for NI.

7.4.6. Contraindicated Therapies

It is perhaps equally more important in avoiding ineffective or potentially injurious treatments as in finding effective ones. Oral treatments for normal itch (e.g., H1 blockers) are usually ineffective for NI, although worth trying due to their low cost and safety, plus some patients may have a component of conventional itch that can be ameliorated by such treatment, e.g., due to substance P release from ectopically firing injured nociceptive neurons (Hagermark et al. 1978). The uncertain utility of capsaicin as a treatment for NI is discussed above. Although local anesthetics are often effective when applied topically or given systemically, there is no evidence that nerve or nerve root blocks containing local anesthetics (and/or corticosteroids) have long-term benefits that justify their risk and cost, despite transient efficacy (Eisenberg et al. 1997; Goulden et al. 1998).