Clinical manifestations

Presentation and course

Pain, hyperalgesia, and allodynia. Given the syndrome’s name, the cardinal symptom is pain – pain disproportionately severe in intensity to the inciting event (282; 171; 175). Burning is the pain’s most distinctive feature (450; 371), as in the name “causalgia.” Other adjectives used are “shooting,” “lancinating,” “electrical,” “tearing,” “stinging,” “squeezing,” “pulsating,” and “aching,” often deep in the limb (191; 38; 05). However, among patients meeting the 1993 IASP CRPS criteria, the special criterion of “burning” pain did not aid diagnostic accuracy (140; 173), despite its presence in 81.1% (173). Birklein and colleagues found the most frequent pain descriptor in patients with CRPS I was “tearing” (25.4%), followed by “stinging” (17.2%), “burning” (16.4%), etc. (40). In CRPS II patients, the percentages were 30.4%, 21.7%, and 26.1%, respectively. The reason for the marked discrepancy between the incidences of burning pain in the study by Birklein and colleagues and that of Harden and colleagues is unclear.

Spontaneous (stimulus independent) pain occurred in 74.6% of patients with CRPS I and 91.3% of patients with CRPS II (40). It begins immediately after injury or shortly thereafter (although a delay in onset after the inciting event is also possible) (287), but its characteristics may change over time. Initially spontaneous, constant, and intense, the “burning,” “electrical” qualities gradually metamorphose into “dull,” “boring,” “aching,” or “tingling” chronicity, with intermittent acute pain, occurring most often after mechanical stimulation (191). Pain is often exacerbated by external stimuli, movement of the affected limb, and temperature changes (442; 40; 191; 450).

A close relationship between burning pain and hyperesthesias (allodynia and hyperalgesia) in CRPS exists: 93.4% in CRPS I and 95.7% in CRPS II (40); 75% (442); and 65.1% (173). Nonetheless, these are wide variations in incidences of hyperesthesias despite each study (except that of Veldman and colleagues) including patients meeting the 1993 IASP CRPS criteria, and significant numbers of patients, otherwise diagnosed with CRPS, were without hyperesthesias. Sieweke and colleagues found marked mechanical hyperalgesia without static hyperalgesia or hyperalgesia to heat (388). It is important to note that sensory deficits (hypoesthesia) may occur in the same distribution as allodynia or hyperalgesia or in a glove and stocking distribution distally, in no particular nerve distribution. Some patients have no pain and/or hyperalgesia: 6.6% to 7% in CRPS I and 4.3% in CRPS II (442; 40). But, the 1993 IASP CRPS criteria require continuing pain, allodynia, or hyperalgesia. Allodynia occurs when normally non-painful stimuli are painful and hyperalgesia when normally painful stimuli are more painful (63). Therefore, in those patients without pain, allodynia, or hyperalgesia, the criteria are not satisfied.

Vasomotor and sudomotor (edema, color, temperature, and sweating) changes. Vasomotor and sudomotor phenomena are diagnostic requisites at some point in the course of CRPS. They include regional edema, erythema, increased skin temperature, and increased sweating (compared to the opposite limb), most often seen early or acutely – with reduced sweating and skin temperature and pallor or cyanosis becoming manifest with chronicity. Veldman and colleagues found discolored skin in 91%, altered skin temperature in 92%, and edema in 69% of patients with CRPS I (442). Birklein and colleagues found side-to-side temperature differences in 90%, and when skin color or edema was included, 98% of patients had abnormalities (40). However, these are not constant features and vary with disease duration. For instance, edema was seen in 92% of patients within the first five weeks, but its incidence decreased to 55% with durations of 25 to 468 weeks. Sweating changes were present in 54%. Warmer limbs and erythema occurred in acute CRPS, but colder and cyanotic limbs were more common with chronicity. Changes in skin color in total were seen in 79% of patients. Veldman and colleagues found that the longer the interval between CRPS onset and the first examination, the more often patients were found with a cold limb (442). However, skin temperature differences between bilateral limbs were less than 1°C in 131 of 296 patients (44.3%) with CRPS I or II who otherwise met the 2012 IASP research criteria (81). Cool temperature was not seen in 88 patients (29.7%), and there was no correlation between symptom duration and skin temperature change or difference. A perfusion index, based on pulse oximetry, has been found to be more sensitive (78.26%) than limb temperature measurements (greater than 1º C, 34.78%) in affected limbs in detecting subjective thermal/vasomotor symptomatology (83).

Movement abnormalities. Movement abnormalities in CRPS were not required in the 1993 IASP criteria but are equal to other criteria in the 2003/2012 revisions. Their earlier absence was because they were believed to be either a secondary defensive mechanism to protect the limb from painful stimuli or of psychogenic origin (429). However, multiple publications describing weakness, bradykinesia, dystonia, myoclonus, tremor, and motor neglect, along with accumulating evidence for central nervous system dysfunction, adaptation, and reorganization in patients with motor symptoms, suggest that movement abnormalities are intrinsic to the CRPS picture. Schwartzman and Kerrigan described 43 of 200 CRPS patients with major motor symptoms, eg, weakness, spasms, tremor, inability to initiate movements, and painful dystonic posturing, most often accompanying pain, sudomotor, and vasomotor manifestations (378). Veldman and colleagues found weakness in 95%, tremor in 49%, muscular incoordination in 54%, and severe muscle spasms in 25% of affected limbs (442). Birklein and colleagues found motor dysfunction in 97% of affected limbs: 79% of patients experienced weakness, 87% limited range of motion, 46% exaggerated tendon reflexes, 48% enhanced tremor, 30% myoclonic jerks and dystonic muscle contractions, and 45% difficulties initiating movement (40). Myoclonic jerks and dystonia were almost twice as common in CRPS II than in CRPS I. Van Rijn and colleagues found movement abnormalities in 121 of 185 patients with CRPS I (65.4%), with dystonia in 90.9% of those affected (431). Schilder and colleagues found that voluntary motor control was impaired in patients with CRPS on both the affected and unaffected sides, as compared to patients with Parkinson disease and normal controls, suggesting involvement of central processing circuits (370).

Perhaps the most dramatic and controversial movement disorder in CRPS is dystonia. Dystonia has varied etiologies and consists of recurrent or sustained muscle spasms and involuntary simultaneous contractions of agonist and antagonist muscles resulting in abnormal postures of various body parts, including the extremities. Dystonia can be generalized or focal and variable or fixed. In most dystonias fixed positions do not occur until advanced stages. Focal dystonias affecting distal limb musculature can be triggered by trauma (407; 433).

In the context of limb trauma and pain, dystonia is associated with CRPS, most commonly a fixed distal dystonia initially in the affected limb but with the potential to spread to other limbs (372). Dystonia can develop at the onset of CRPS or any time thereafter, even more than five years later (431). Compared to CRPS patients without dystonia, those with dystonia are significantly younger with longer disease duration and with greater risk of second extremity affection once dystonia is present in the first extremity (431; 433). Non-motor symptoms of CRPS can improve as dystonia develops (33).

The absence of confirmatory diagnostic tests and a well-understood pathophysiology, and the similarity to other “psychogenic” dystonias, underlie the belief of some observers that the dystonia of CRPS is psychogenic (180). Verdugo and Ochoa described 58 (out of 686) CRPS patients with movement abnormalities without structurally based central or peripheral nervous system abnormalities, each exhibiting at least one “pseudoneurologic” sign (443). Interestingly, all of their patients with movement disorders had CRPS I; none of the 44.8% of the total with CRPS II had movement abnormalities. In contrast, Birklein and colleagues found movement abnormalities more commonly in patients with CRPS II (40). Schrag and colleagues found evidence of psychogenic dystonia in 13 of 18 prospectively-studied patients meeting the CRPS clinical criteria, out of 103 patients with fixed dystonia (373). Other movement disorder experts believe that CRPS dystonia is of psychogenic origin (293; 345; 237; 180).

However, accumulating clinical, neurophysiological, and pharmacotherapeutic data suggest that CRPS dystonia is not of psychogenic origin. Van der Laan and colleagues did not find increased psychopathology in CRPS dystonia secondary to fracture (426). Reduced central inhibition has been found in CRPS dystonia (429) in keeping with current concepts of dystonia as a central circuit disorder (286; 431). Impaired reciprocal inhibition and markedly reduced stretch reflexes were found on EMG, as was vibratory inhibition of the H-reflex, in CRPS dystonia – each finding suggesting impairment of interneuronal circuits that mediate presynaptic inhibition (421). Munts and colleagues used computer modeling to simulate the fixed dystonia seen in 85 CRPS-I patients, concluding that aberrant force feedback regulation from Golgi tendon organs involving an inhibitory interneuron may be present (302). There is also clinical and neurophysiological evidence for the presence of a motor axonopathy in CRPS patients with dystonia (309). The relationship of CRPS dystonia to other dystonias and their psychogenic/organic explanations is well reviewed by Munts and Koehler (301). Complicating the binary division of organic versus functional movement disorders are the findings that patients with functional movement disorders exhibit structural gray matter abnormalities in important components of the limbic and sensorimotor circuitry (276). Long ago, it was observed that “the abnormalities of movements go off when the pain is relieved” (304).

Galer and colleagues described motor or cognitive neglect in 84% of 224 patients, a finding often associated with parietal cortical dysfunction, which may explain the difficulty in initiating movements and bradykinesia (141; 142). There is also spatial neglect in patients with CRPS (298). Kolb and colleagues studied upper limb neglect in CRPS patients, normal controls, and patients with non-CRPS pain (220). They found similar patterns of neglect in the CRPS and non-CRPS pain patients; however, the neglect in patients with CRPS differs from classical neglect seen in parietal lobe stroke patients. Cortical reorganization, involving primary motor (both ipsilateral and contralateral), supplemental motor, and posterior parietal cortices, has been found on fMRI and correlates with kinematic evaluations of motor impairment (262). Cortical reorganization may account for other cognitive and body scheme disturbances in CRPS (eg, impairment in upper extremity position localization) (250) and spatial perception of tactile stimuli (415); agnosia for object orientation (267; 351); finger misperception, impaired hand laterality recognition, and astereognosis (232); and attention bias away from near-space on the affected side (61). Similar findings were seen by Brun and colleagues, although they suggest that kinesthesia and body perception be evaluated separately (58).

Other significant symptomatology. Bone and joint changes, consisting of demineralization and resorption, are common in longstanding disease. As with dystonia, many or all of CRPS’s other clinical features have been attributed to psychological factors; however, reviews concur that the best available evidence supports an interaction rather than a causal connection between psychological variables and the physiologic presentation (27; 137). Depression and stress, which can increase circulating catecholamines, exacerbate CRPS pain (53). Positive and negative correlations amongst various CNS and PNS metabolites were associated with psychological disorders such as depression, anxiety, suicidal ideation, and anger in patients with CRPS (202). Neuropsychological studies have found deficits in executive function and working memory/verbal fluency in 65% of patients with CRPS (253). Vasovagal syncope or orthostasis is seen in patients with CRPS, especially those with lower limb involvement; it is attributed to impaired sympathetic vasoconstriction and venous pooling (394). Skin ulceration, infection, and edema can be problematic (358).

Clinical variations, staging, and spread. CRPS is not homogenous; variations in clinical picture occur over time and, in fact, can change from hour to hour. This had led earlier reviewers to postulate three definite sequential stages: acute, dystrophic, and, ultimately, an atrophic stage (45; 379; 56). A subsequent study, using cluster analysis, found three subgroups of patients with CRPS based on patterns of signs and symptoms, not on sequential staging, in part because pain duration did not vary among these subgroups: (1) a relatively limited syndrome mostly with vasomotor signs, (2) a relatively limited syndrome with neuropathic pain/sensory abnormalities, (3) a florid syndrome with high levels of signs/symptoms in all categories (56).

Several studies classified patients with CRPS into warm and cold subgroups according to skin temperature differences between affected and unaffected sides (120; 57). CRPS beginning with cold skin temperatures most often remains cold, whereas warm CRPS usually turns cold with chronicity (greater than 6 months). There were clinical differences between warm and cold patients: cold patients more often reported a history of serious life events, dystonia, cold-induced pain, and sensory loss. Warm patients had mechanical hyperalgesia and other clinical features suggestive of inflammation. Although accepting that complete clinical distinctions between warm and cold CRPS are impossible, they might represent peripheral and central nervous system pathophysiology (120) or inflammatory or noninflammatory mechanisms (57), respectively. It is hypothesized that the vasomotor disturbances in cold CRPS may be secondary to inflammation-induced endothelial dysfunction (223).

One shortcoming of the 2012 IASP CRPS criteria is that they are dichotomous: either the patient has CRPS or does not; gradations in severity over time are not evaluated. Subsequently, a continuous index, the CRPS Severity Score (CSS), was devised and validated in a multi-center study of 156 patients (176). The score consists of the presence or absence of 16 items, two signs and two symptoms for each of the four diagnostic features of the 2012 IASP CRPS criteria. New patients exhibited more variability in their CSS values than stable patients. It is hoped that the CSS may be of value in assessing new therapies (176).

Spread of CRPS pain to contiguous areas of the affected extremity and to contralateral or ipsilateral limbs has been noted (269). Van Rijn and colleagues described spread of CRPS to another limb in 45% of 174 patients with single limb onset (430). This could occur spontaneously or after separate trauma to the second limb, either contralaterally or ipsilaterally. The severity of CRPS did not differ significantly in the second limb. Spread of CRPS was associated with a younger age of onset and more severely affected phenotype. The authors postulate that this spread is due to spinal or cortically mediated mechanisms. Edinger and colleagues found that 5 of 20 patients with chronic CRPS developed total body pain (122). Similarly, Birley and Goebel found that widespread pain developed in 21 of 190 patients with CRPS (11.1%) (43). Martinez-Lavin and colleagues described three patients who developed fibromyalgia after CPRS (274). Alternatively, the Royal College of Physicians states that “spread is rather rare, occurring in about 7% of all cases” (358). The reason for these widely disparate estimates is unclear.

Pediatric CRPS. CRPS is rare in school-age children and teenagers but, when present, differs little clinically from adult CRPS, albeit with a better prognosis for rapid and complete recovery (460; 459). However, a retrospective assessment of quality of life in adults who had developed CRPS I in childhood found no difference compared with that of individuals with adult-onset CRPS I (405), and another retrospective study found relatively poor outcomes, both physically and emotionally, after a mean follow-up period of 37 months (23). Case series demonstrate consistent features: 70% to 95% of patients are female; onset is usually in the 8-to-16-years age group; the lower extremities are more commonly affected, especially feet and ankles (57% to 100%) (460; 459; 259; 382; 205; 23); and movement disorders such as dystonia, tremors, and myoclonus, although rare, can occur alone or together (03). Most quantitative sensory testing studies document abnormal sensory patterns similar to those of adult patients, although heat and cold allodynia may occur less often (382). The rapid recovery of most children has facilitated documentation of changes in brain function during recovery (240). However, in some adolescents, persistent pain can induce anxiety and emotional and body scheme disturbances requiring aggressive therapy (397). No current diagnostic tools have been validated for pediatric CRPS (283).

Treatment algorithms differ for children, with more physical therapy and behavioral and psychological support and less use of medication and surgery (454). This reflects not only the increased caution of pediatric medicine, but also the better prognosis in children. Medications may be targeted to enable participation in physical therapy (259). Full or partial recovery occurs in the vast majority of patients (about 80% to 90%), but recurrence is documented (20% to 50%) (384; 241; 259; 205). Seven pediatric patients had good outcomes with spinal cord stimulation (313). In a case series of 10 pediatric patients, multidisciplinary management, including capsaicin 8% patches, epidural catheterization for bupivacaine infusions, and spinal cord stimulation, was found efficacious (354). Alternatively, a meta-analysis of the effectiveness of invasive treatment procedures in children and adolescents found weak evidence for their efficacy because of outcome validation issues (470). Intensive interdisciplinary psychophysical pain treatment has been shown to be of great benefit clinically and is associated with a reversion of resting state CNS baseline changes towards normal in pediatric patients (390; 129). Inpatient regional anesthesia, with inpatient rehabilitation, was found helpful in pediatric patients who failed other therapies (92; 116).

Both amitriptyline and gabapentin were effective in 34 children with CRPS I, decreasing pain intensity scores and improving sleep (52). There were no significant differences between the two drugs.

Prognosis and complications

The prognosis of CRPS varies widely. That said, defining recovery in CRPS is problematic as many patients with longstanding CRPS no longer meet all the diagnostic criteria yet still are symptomatic (257). The factors important in 252 patients’ own definition of recovery were cessation of pain, resolution of movement restrictions, and no need for pain medications.

It is probable that many patients develop the clinical picture shortly after injury or operation that resolves spontaneously with adequate physical therapy. In this group, the prognosis is excellent. A population-based review of CRPS I found an overall good prognosis: 74% of patients recovered and usually within the first year (364). CRPS developed within eight weeks of a traumatic event in 35 patients with type I, 19 of whom (54%) improved to the point of not meeting diagnostic criteria at three months, indicating partial remission, but they did not further improve over the next nine months (59).

Patients seen in a pain center with more severe symptoms and a longer duration of illness fare worse. De Mos and colleagues found in patients with the syndrome for more than one year (mean 5.8 years) that 30% recovered, 54% were stable, and 16% reported severe progression (100). Schwartzman and colleagues did not find any spontaneous remission in 656 patients with CRPS of at least 1-year duration (377). A review concluded that in prospective studies symptom improvement was significant in the first 6 to 13 months, but the results from retrospective and cross-sectional studies indicate that there is a cohort of patients with persistent pain, often with motor symptomatology (25). A prospective study of 59 patients with recent onset CRPS I followed at 6 and 12 months found a significant reduction in symptoms, greatest in the first six months, but at one year nearly two thirds of patients continued to meet the IASP-1993 criteria and one fourth met the Budapest research criteria. Only 5.4% were symptom-free at 12 months (24). These data suggest that the syndrome can be divisible into acute and chronic conditions, with a reasonably good prognosis for improvement within the first year of onset and a poor prognosis thereafter.

Several factors predict a poorer outcome: (1) longer duration of illness, (2) a cooler extremity (428), (3) the initial degree of pain sensitization, and (4) the severity of the initiating injury (after distal radius fracture) (355). It is not clear if early aggressive treatment is more effective because it is not known whether those patients whose CRPS lasts longer than 6 to 12 months constitute a group with an intrinsically poorer prognosis. However, given current knowledge, early aggressive treatment makes sense (243).

Complications of the disease generally pale in comparison to the disease itself, which, in its later course may essentially incapacitate the limb, with atrophy of nearly all muscle groups, osteoporosis, and hairless, paper-thin skin. Complications at this stage and, to a lesser extent, at earlier stages, include infection, particularly cellulitis. Deep venous thrombosis may occur if the limb is completely immobilized due to pain.

Clinical vignette

A 64-year-old male, who underwent an unsuccessful venipuncture in his right antecubital fossa, experienced a “shooting, electrical” pain radiating down his arm into his fingers with the needle in situ. The pain persisted and over several months he developed allodynic symptoms in the lateral antebrachial cutaneous nerve (LACN) distribution without vasomotor, sudomotor, or motor/trophic phenomena. Two separate examinations 7 and 16 months post-venipuncture were entirely normal until 5 to 10 minutes afterward when he began to experience itching and burning in the LACN distribution followed by development of 15 to 20 3- to 4-mm erythematous, pruritic bumps resembling hives solely in that nerve’s distribution at points where he was tested with a pin. Similar pin testing in the distributions of adjacent nerves and the opposite LACN did not evoke such responses and those areas remained normal. About 10 months post-venipuncture his pain and allodynic symptoms spread to his right leg and foot and the right foot felt warmer than the left.

Strictly speaking, at the 7-month examination he would not have met the 1993 CRPS criteria, except for the “hive” formations, and even at the 16-month examination he would not have satisfied the 2003 CRPS criteria. Yet he had persistent burning pain, spreading allodynia, and the perception of temperature asymmetry. He satisfied the criterion for a third diagnostic category – CRPS NOS (171; 175), but this category is even more problematic.

Biological basis

Etiology and pathogenesis

Our understanding of the pathophysiology of CRPS, although still fragmented, has expanded enormously (101; 53; 273). Several major nonexclusive, probably interrelated mechanisms, are being considered: (1) small fiber neuropathy, (2) sympathetic nervous system dysfunction, (3) peripheral and central nervous system inflammation, (4) autoimmunity, (5) central sensitization, and (6) central nervous system reorganization. Efforts have been made to parse patients into groups representing different etiologies, such as peripheral (edema, skin color changes, skin temperature changes, sweating, trophic changes), central (minor injury, motor signs, allodynia, stocking/glove sensory deficits), and mixed forms of CRPS (110).

Small nerve fiber involvement. Peripheral nerve trauma is requisite in CRPS II, and peripheral nerve damage has been demonstrated in CRPS I (308). Pathological study of amputated legs from eight patients with advanced CRPS I revealed partial injuries, particularly to small-fiber (A delta and C) axons, neurogenic changes in muscle, and marked thickening of capillaries as occurs in diabetic small-fiber polyneuropathy (425). Skin samples from two additional amputated limbs of patients with CRPS I revealed partial loss of epidermal, sweat gland, and vascular small fibers and altered patterns of neuropeptide expression in remaining fibers (06). A larger study of skin biopsies from affected limbs in 19 patients with CRPS I revealed average losses of 29% of intra-epidermal nerve fibers (IENF) as compared to unaffected control sites (311). IENF density was reduced in both affected and contralateral limbs in patients with CRPS when compared to healthy controls (294; 344). Further, Morellini and colleagues found bilateral reductions in the proximity of dermal mast cells to surviving nerve fibers. The authors hypothesize that loss of dermal nerve fibers in CRPS might result in loss of chemotactic signals, thus halting mast cell migration toward surviving nerve fibers, prolonging inflammation, and delaying tissue repair (294). These studies suggest that in CRPS, patients’ limb traumas have initiated chronic focal axonal degeneration, predominantly affecting small fibers, and that CRPS I, like CRPS II, is associated with nerve injury.

Damage to some small fibers appears to trigger ectopic activity in surviving fibers and in central postsynaptic targets in the dorsal horn and higher centers via early long-term potentiation, chronic facilitation, and disinhibition (463). The final result is spontaneous pain, allodynia, hyperalgesia, and spread of symptoms to nearby areas.

Alternatively, Reimer and colleagues argue against the concept of minimal nerve injury; they attribute improved sudomotor and vasomotor functions and pain and mechanical detections on quantitative sensory testing, along with increased thermal and mechanical pain sensitivity ipsilaterally and contralaterally, to central neuroplasticity induced by nociceptive sensitization, (ie, pain-induced hypoesthesia) (346).

In regards to other nerve fiber abnormalities, Geertzen and colleagues reported decreased 12 to 14 um myelinated fiber counts in the sural nerves of the amputated lower limbs of eight patients with therapy-resistant CRPS (144). Although microscopic evidence of myelinated fiber degeneration and regeneration was widespread, myelinated fiber counts in other lower extremity nerves were normal in these patients, as were myelinated fiber counts in upper limb nerves in three patients with upper limb amputations. The significance of the reduced sural nerve myelinated fiber counts is unclear. The study of amputated limbs is difficult to review due to inconsistent reporting features (194).

Sympathetic nervous system (autonomic) dysfunction. There are at least three different mechanisms by which the non-pain abnormalities that define CRPS – swelling, color, sweating, temperature, and trophic changes – can be attributed to the sympathetic nervous system: (1) sympathetic nervous system overactivity, (2) sympatho-afferent fiber coupling, and (3) end-organ hypersensitivity to norepinephrine. Historically, sympathetic nervous system overactivity (hyperactivity, excessive outflow) was assumed: “The causalgia of Weir-Mitchell must be related to a traumatic lesion of the sympathetic system in the limbs” (247). However, directly assessing sympathetic nervous system function in a limb is complex. There is no such entity as “global limb sympathetic function” in the generic sense any more than there is global limb motor or global limb sensory function. With microelectrode insertion in appropriate nerve fascicles, spontaneous activity in small nerve fibers, the majority of which subserve blood vessel control in skin and muscle, can be measured directly. Surprisingly, in patients with CRPS, average skin sympathetic activity is not different from controls, and there is no microelectrode electrophysiological single nerve fiber evidence of activation of C-nociceptors by sympathetic efferent fibers in cutaneous nerve fascicles in CRPS I and II patients, even in those clinically responsive to diagnostic sympathetic blocks (67). This finding would also seem to negate the presence of sympatho-afferent nerve fiber coupling in the skin, which has been supported by more indirect measurements (20). Additionally, sympatho-afferent coupling in deep somatic structures, especially in acute CRPS, has been proposed (369). Unfortunately, the results in patients with CRPS II in the study by Campero and colleagues are invalidated by the patient selection process because 10 of 11 patients had polyneuropathies (67).

If increased sympathetic activity itself is not part of CRPS, an alternative hypothesis might be end-organ supersensitivity to norepinephrine. Such supersensitivity would lead to the appearance of sympathetic overfunction in the affected part, even if true sympathetic outflow were normal or reduced. Several lines of evidence support alpha-adrenergic receptor hypersensitivity in limbs affected by CRPS, including the following neuropathologic alterations: (1) inappropriate expression of neuropeptide Y on innervation to sweat glands and superficial arterioles, (2) significant vascular hypertrophy and a loss of vascular endothelial integrity, (3) a loss of calcitonin gene-related peptide (CGRP) expression on remaining innervation to sweat glands and blood vessels, (4) the presence of numerous abnormal thin caliber myelin basic protein-negative/neurofilament-positive axons innervating hair follicles, and (5) a decrease in vascular, epidermal, and sweat gland innervation (06).

During conditions of sympathetic nervous system activation, pain intensity increases in patients with CRPS (20; 53; 163). Patients with CRPS experience increased pain after intradermal norepinephrine injections, whereas control patients do not (07). Additionally, activation of alpha-1a adrenoreceptors by IgG serum autoantibodies was found in patients with longstanding CRPS (118). Expression of alpha1-adrenoreceptors on nerve bundles in CRPS-affected limbs was increased in patients with CRPS (CRPS II greater than CRPS I) and greater in acute than in chronic CRPS (117). When the alpha1-adrenoreceptor agonist phenylephrine was injected intradermally, prolonged pain and pinprick hyperalgesia at the injection site was associated with this increased receptor expression, suggesting that alpha1-adrenoreceptor upregulation in dermal nerve bundles contributes to pain or hyperalgesia in a subgroup of patients with CRPS (117). Norepinephrine, via alpha1-adrenoreceptors, stimulates dendritic cell migration and inflammatory cytokine release (360). Some patients with CRPS have been found to have increased resting heart rates and decreased heart rate variability and baroreflex sensitivity, suggesting sympathetic/parasympathetic imbalance (406; 21; 360). The sympathetic nervous system may be responsible for IL-6 upregulation after fracture and immobilization and CRPS-like symptoms (84). Lower parasympathetic activity may contribute to the production of inflammatory mediators and worsen CRPS-like symptoms (84; 360).

Other studies have not confirmed alpha-adrenergic hypersensitivity and found that sympathetic vasoconstrictor activity is often reduced rather than increased in acute, subacute, and chronic CRPS (along with reduced sweat responses) (449; 335). Finally, many patients with CRPS do not respond or respond minimally to sympathetic blockade or sympathectomy.

In summary, the role of the sympathetic nervous system in CRPS is currently unclear; it is possible that many of the clinical features previously attributed to sympathetic overactivity may be explained by other factors, such as neuroinflammation (41). Alternatively, the sympathetic nervous system and neuroinflammation are hypothesized to be closely related (360).

Neurogenic inflammation. There is increasing evidence that inflammation in both the peripheral and central nervous systems follows some peripheral nerve injuries and may be related to the clinical manifestations of CRPS (361; 393; 29; 84; 360; 337; 456). Although normal nociceptive transmission does not provoke an inflammatory response, for still unknown reasons neurogenic inflammation may involve efferent conduction through nociceptive C fibers back to the afflicted tissue (287). This retrograde activity may result in the release of proinflammatory neuropeptides, including substance P, CGRP, neurokinin A, adrenomedullin, neurokinin B, vasoactive intestinal peptide, neuropeptide Y, brain derived neurotrophic factor, and gastrin-releasing peptide (256), along with cytokines, adenosine triphosphate, and reactive oxygen scavengers (338; 29) and may be required for IL-beta, TNF alpha, and nerve growth factor production by keratinocytes (84). In this way, some peripheral nerve injuries activate the innate immune system, including dendritic/Langerhans cells, mast cells (84; 360), and Schwann cells with upregulation of cell-surface antigens (eg, glial fibrillary acidic protein) and release multiple proinflammatory mediators (361). Some of these substances may act upon nociceptive A delta fibers inducing peripheral nerve sensitization (251; 256). CGRP activates receptors in smooth muscle and on endothelial cells with resulting vasodilation and substance P and neurokinin A promote vascular permeability (256; 64). In turn, this may cause hyperemia, edema, and migration of leukocytes. CGRP may also provoke sweat gland function and hair growth (256; 64; 287).

Patients with CRPS have been found to have elevation of certain blood pro-inflammatory cytokine mRNA levels (TNF and IL-2) and reduced mRNA levels of anti-inflammatory cytokines (IL-4 and IL-10) (417). IL-1beta and IL-18 are also elevated, the former possibly being a mediator of painful symptomatology (84). A meta-analysis of the presence of inflammatory factors in CRPS concluded that there are different inflammatory profiles in acute and chronic CRPS (317; 115; 223). In acute CRPS blood studies demonstrated a large effect size for IL-8 and small effect sizes for soluble TNF receptors I and II. Relationships to other cytokines, and between cytokine levels and acute CRPS severity, remain uncertain. In chronic CRPS, increases in blood, blister, or CSF levels of multiple proinflammatory cytokines were found along with decreases in substance P, several selectins, sGP130 in blood, and sICAM-1 in CSF. The authors conclude that a proinflammatory state exists in CRPS, but with several caveats: (1) Because very few studies used control subjects with chronic pain due to other causes, it is not clear that the findings are specific for CRPS rather than chronic pain states in general; and (2), it is not clear that the heightened inflammatory responses found in these studies have a causal role in CRPS. Varenna and Crotti hypothesize that bone is a possible source of these inflammatory mediators in acute CRPS (438). Altered metabolism of substance P in patients with CRPS or in patients with limb trauma suggests that this may contribute to the development of persisting pain (222).

Somewhat in conflict with the conclusion that there is a proinflammatory state in chronic CRPS are the findings by Lenz and colleagues that although a proinflammatory cytokine profile was seen in the blister fluid of patients with early CRPS, cytokine levels were comparable to those of non-CRPS patients after six months independent of clinical outcome or symptom resolution. They could not conclude that the reduction in proinflammatory cytokine levels over the 6-month period was treatment induced or a natural course of the disease (246). Mast cell accumulation was not seen in chronic CRPS (314).

Birklein and colleagues found similar results when studying the skin biopsies of 55 patients with early and chronic CRPS (37). In acute CRPS (less than 3 months duration), there was increased mast cell accumulation, and epidermal keratinocytes were activated, resulting in epidermal thickening and increased TNF-alpha and IL-6 expression, whereas in chronic CRPS the reverse occurred; keratinocyte proliferation and epidermal thickness decreased, and mast cell numbers were unchanged.

Neurogenic inflammation has been identified in affected and, to a lesser degree, non-affected limbs of patients with CRPS (451; 245; 181; 41); it produces axon reflex vasodilatation and plasma extravasation to cause edema. Nociceptive changes also include upregulation of sodium and transient receptor potential channels resulting in reduced thresholds for activation and repetitive firing of dorsal root ganglion neurons and consequent peripheral sensitization (29). There is clinical evidence for oxidative stress in CRPS I patients, also associated with inflammation (123).

An autopsy study of the spinal cord of a patient with long-standing CRPS demonstrated significant glial (microglial and astrocyte) activation, plus neuronal loss in the posterior horn, with decreased posterior horn size, most prominent at the level of original injury (L4 to S2) but extending throughout the spinal cord (98). Neuroinflammatory gene upregulation has been reported in the dorsal root ganglion in a rat model of CRPS (75).

Neurogenic inflammation may play a role in immobilization-induced CRPS-like clinical changes. Pepper and colleagues found increased levels of IL-6, TNF-alpha, and the mast cell marker tryptase in the skin of hands affected by wrist fracture, subsequent surgery, and cast immobilization (319). An experimental study of limb casting (immobilization) alone in rats demonstrated clinical changes and upregulation of neuropeptide signaling, similar to those seen after fracture with casting (168).

Peripheral nerve injury also induces entry of glial and neuronal-derived cytokines, trophic factors, neurotransmitters, reactive oxygen species, and immune-competent leukocytes into the dorsal horn, the ultimate effect being neuronal sprouting and central sensitization via multiple mechanisms (239; 361; 29). In the needlestick-nerve-injury model, hyperalgesic rats had greater endoneurial edema both ipsilaterally and contralaterally, and significant axonal loss in uninjured ipsilateral nerves, supporting similar spread of symptoms and signs clinically in patients with CRPS I (217).

A role for the gut microbiome in CRPS neuroinflammatory states has been proposed (91). Metabolomics as a research tool in chronic pain is in its infancy (202; 203; 409).

Autoimmunity. There is considerable interest in the role of autoimmune mechanisms in these neuroinflammatory features in association with the clinical and physiological findings in at least some patients with CRPS, especially when chronic. If autoimmunity is causative, it would be via novel mechanisms, which account for CRPS’s regional or local phenomenology, its occasional propensity to spread, its initiation by trauma—either to a specific nerve as in CRPS II or to limb bones or soft tissues as in CRPS—resulting in the formation of as yet unknown regional neoantigens, the development of central sensitization (vide infra), and the difference between acute CRPS (resolution in patients within the first year) and the chronic form.

An early autoimmune finding was surface-binding autoantibodies against an inducible autonomic nervous system autoantigen in 30% to 40% of patients with CRPS (219), followed by identification of these autoantibodies as active against the muscarinic 2 receptor and beta2 adrenergic receptor (218). There may be connections between these autoantibodies and the aforementioned clinical vasomotor and sudomotor features.

Among the various animal models of CRPS developed over the years, two have been used to study autoimmune activity: the hind paw plantar incision model and the tibial fracture/cast immobilization model in rats and mice. Tekus and colleagues used the incision model in mice to study passive transfer of serum IgG, finding increased substance P and clinical features resembling CRPS in mice receiving serum from CRPS patients, but not in those receiving serum from healthy controls (concentrations of CGRP, TNF-alpha, IL-1, and IL-6 were normal) (410). Li and colleagues demonstrated, in the fracture/cast model, but not in the incision model, that B-cell depletion with anti-CD20 antibody (rituximab), or the lack of B-cell production in transgenic B-cell deficient mice, attenuated the development and maintenance of nociceptive and vascular CRPS-like changes in the limb that had surgery (252). Additionally, they found none of the typical increase in IgM levels in the skin and sciatic nerve in the fractured/casted limb of B-cell deficient mice, which raised the possibility of a role of complement activation secondary to IgM production in the development and maintenance of these CRPS-like changes. B-cells were not found to affect the presence of pro-inflammatory or anti-inflammatory cytokines. The results suggest to these authors that although B-cells do not directly modulate nociceptive sensitization, they support CRPS-like changes in the murine fracture/cast model by producing IgM antibodies, which in turn activate complement and the formation of C5b-9 membrane attack complexes with consequent cellular damage. Some of the differences in these findings may relate to the murine model used and the timing of the study evaluations.

T-cell activity, as measured by plasma soluble interleukin-2 receptor (sIL-2R) levels, was found to be increased in 80 patients with CRPS I of 5 to 36 months’ duration (31), raising the possibility of T-cell-mediated inflammation in CRPS and the potential of sIL-2R as a new marker of disease activity. Serum CD14+ monocytes are elevated in patients with CRPS and correlate with allodynia severity (350).

Goebel and Blaes proposed that this potential form of autoimmunity in CRPS be considered “injury-triggered, regionally-restricted, autoantibody-mediated autoimmunity with a minimally destructive course” (159).

Two reviews (one including a new hypothesis) (84; 360) highlight the various inflammatory and autoimmune dynamic and interacting components involved in the onset and development of acute CRPS following trauma. Essential to CRPS onset is the activation of various cells of the innate immune system, especially dendritic/Langerhans cells, provoked by multiple endogenous products such as heme, hemoglobin, interleukins (ie, IL-1alpha, IL-33), and heat shock proteins released during tissue trauma. Dendritic cell activation is also associated with tissue hypoxia, a possible component of CRPS. Activated mature dendritic/Langerhans cells migrate to draining lymph nodes followed by the release of multiple proinflammatory and pronociceptive mediators, eg, cytokines, complement fragments, TNF alpha, and prostaglandins, and associated with sympathetic hyperactivity. There they drive an adaptive immune response (84; 360) as part of a “neuro-immuno-cutaneous system” involved in the regional production of neuropathic pain. Regional activation of keratinocytes, osteoclasts, and monocytes is also thought to play a role in CRPS development. These activated cells also are thought to contribute to central sensitization.

Central sensitization. “Central sensitization represents an enhancement in the function of neurons and circuits in nociceptive pathways caused by increases in membrane excitability and synaptic efficacy as well as to reduced inhibition and is a manifestation of the remarkable plasticity of the somatosensory nervous system in response to activity, inflammation, and neural injury” (239). There are “novel inputs to nociceptive pathways…development of or increases in spontaneous activity, a reduction in the threshold for activation by peripheral stimuli, increased responses to suprathreshold stimulation, and an enlargement of…receptive fields” (239). Stimuli that would otherwise be innocuous or undetectable elicit discomfort (allodynia) or pain in the affected nerve distribution (primary hyperalgesia) or outside it (secondary hyperalgesia) (463). Multiple mechanisms serve to increase excitability and reduce inhibition (239), including increased neuronal burst firing in the ventroposterolateral nucleus of the thalamus, which correlates with behavioral manifestations of neuropathic pain (361; 29). Central sensitization can explain a number of clinical features of CRPS, particularly mechanical allodynia and hyperalgesia, stimulus-induced pain, and spread of pain to other anatomic sites outside the injured zone.

Central nervous system reorganization (neuroplasticity). Over the past two decades, cortical reorganization has been demonstrated in patients with CRPS utilizing various technologies (380; 403). Changes in the contralateral primary (S1) and secondary (S2) somatosensory cortices to non-painful tactile stimulation of the affected side – increased cortical inhibition, shrinkage of cortical hand representation, increased 2-point discrimination thresholds – were seen with somatosensory evoked potential and fMRI BOLD signal studies (204; 265; 332; 331; 330). These findings correlated with chronic pain intensity and reverted toward normal as CRPS resolved (266; 331). In CRPS patients, pinprick hyperalgesia and mechanical allodynia evoked widespread cortical activation – contralateral S1, motor and parietal association cortices, bilateral S2, insular, frontal, and cingulate cortices – whereas cortical activation to non-painful stimuli on the unaffected limb was far more restricted (263; 264). Patients were reported to feel pain not only during active movement of the affected limb, but even with the thought of movement only (263). Significant cortical adaptive changes (attenuation) were also seen in the motor systems of CRPS patients both without and with dystonia (228; 262; 150; 216). Reduced functional connectivity between the insula and the postcentral and inferior frontal gyri and cingulate cortices suggests a possible role of the insula in pain processing (211).

Brain anatomy has been studied using structural MRI whole brain voxel-based morphometry. Barad and colleagues found decreased gray matter volume in the dorsal insula, orbitofrontal cortex, and in parts of the cingulate cortex (18). Gray matter volume was increased in the dorsal putamen and hypothalamus. Pain duration was associated with decreased gray matter in the dorsolateral prefrontal cortex, and pain intensity was positively correlated with posterior hippocampus and amygdala volume and negatively correlated with dorsolateral prefrontal cortex volume. The authors concluded there is widespread dysregulation of pain processing in CRPS, affecting multiple systems, which may be the underlying neurobiology to central sensitization. Alternatively, Pleger and colleagues found increased gray matter density in the dorsomedial prefrontal cortex (where emotional correlates of pain are coded) and in the contralateral primary motor cortex (329). Using diffusion tensor imaging, Hotta and colleagues found white matter abnormalities in the corpus callosum and corona radiata of the contralateral cerebral hemisphere (193).

Chronic CRPS signs and symptoms may result from cortical reorganization (39). However, the opposite alternative is equally possible, and both may play a role (299). In favor of cortical reorganization resulting from peripheral dysfunction, Strauss and colleagues found that after applying local anesthetic cream to the forearm of 12 patients with CRPS I, spatial temporal resolution and motor function improved, along with increased short-latency intracortical inhibition on the affected/treated side, suggesting that temporary deafferentation of an area neighboring the CRPS-affected area can modulate cortical changes (401). Even magnetoencephalography has demonstrated somatosensory misrepresentation in the brain of a patient with CRPS with spread to other areas (62).

In multiple studies, the P.A.I.N. group at Boston Children’s Hospital has reported resting state gray matter, amygdala, and multiple functional connectivity changes in pediatric patients with CRPS at baseline that revert towards normal with intensive interdisciplinary psychophysical pain treatment (390; 129).

A debate has arisen as to the presence, form, and validity of the CNS changes. A meta-analysis found a high risk of bias across studies and concluded there was no evidence regarding M1 spatial representation and only limited evidence for bilateral M1 disinhibition in CRPS of the upper limb (111). Di Pietro and colleagues, using fMRI in 17 patients with upper extremity CRPS, found S1 representation of the affected hand was no different than that of either hand in controls, but S1 representation of the healthy hand of patients with CRPS was larger than that of control hands (113). They concluded that although CRPS is associated with larger representations of the healthy hand and not smaller representations of the affected hand, this finding is not related to compensatory healthy hand use nor pain duration; it has no clear etiology at present (114). Alternatively, van Velzen and colleagues found no differences between 19 female patients with CRPS I and normal controls regarding resting state brain structure or function and found that previous MRI results indicating aberrant neuroplasticity are inconsistent and poorly reproducible (435). Di Pietro and associates, in response, challenge this view, pointing out that van Velzen and colleagues conducted a structural resting state investigation, whereas much of the literature on the subject documented evidence of functional changes when doing a task or delivering an external stimulus (112; 435).

Neuroplasticity has been found to differ in early and late stages of CRPS (386). The early patient group showed reduced gray matter volume and cerebral perfusion in areas associated with spatial body perception, the somatosensory cortex, and the limbic system. The late patient group exhibited increased perfusion in the motor cortex, and gray matter volumes in areas associated with pain processing were negatively associated with pain levels.

Psychological factors. A prospective, multicenter trial including 596 patients with a fracture looked at many psychological factors that might be linked to the development of CPRS in 7% of the patients, but none of the psychological scores differed from those in the general population or were predictive of the occurrence of CPRS (26). A systematic review by the same author found a link only with more life events (27).

Epidemiology

By consensus definition, CRPS occurs consequent to injury or immobilization, although in large case series, it has been reported to occur spontaneously in 7% to 10% (99; 104). In a retrospective review of 134 CRPS patients, the most common injuries were sprains/strains, surgical procedures, and fractures, often associated with physician-imposed physical immobilization (09). The syndrome is rare in young children and the elderly; median age of onset is near 40, with the highest incidence between ages 45 and 55 (125; 325). The incidence and prevalence are only estimates because of diagnostic imprecision, nomenclature, and variable presentation of patients across multiple specialties. Sandroni and colleagues generated the first community-based data in one county in the Midwestern United States: CRPS I incidence was reported as 5.46 per 100,000 person-years and the prevalence as 20.57 per 100,000, with a median age of onset of 46 years (364). De Mos and colleagues, in a second epidemiological study, from the Netherlands, found an incidence of 26.2 per 100,000 person-years, greatest in the 61- to 70-year-old group, with a mean age of onset of 52.7 years (99). The explanation of the 5-fold disparity in incidence and substantially older age of onset is unclear, although cultural, geographic, and sampling differences between populations have been raised by Sumitani and colleagues and by Bruehl (54; 402). Females comprise 75% to 80% of patients and fractures are the most common precipitant. Distal radius or Colle fracture have been reported to have a CRPS incidence as high as 36.7% (163; 287), but the incidence appears to be declining in the recent years in the Netherlands (166). CRPS II (CRPS with major nerve injury) is much less common, with an estimated incidence of 0.82 per 100,000 person-years at risk and prevalence of 4.2 per 100,000, without a sex effect (364). It occurs in approximately 2% to 5% of peripheral nerve injuries (442). CRPS developed in 68 of 291 (26.2%) patients with surgically treated traumatic hand injuries within 40.1 days post-operation (367). Patients with crush injuries and a pain score of greater than five were at highest risk.

Other epidemiological factors associated with CPRS include Caucasian race, high household income, private insurance, depression, headache, and drug abuse (125; 287). Conversely, there appear to be no links to diabetes mellitus, anemia, obesity, and hypothyroidism (125). The environmental influences of trauma and immobilization may obscure genetic contributions. In the Netherlands, HLA-DQ1 is more prevalent amongst patients with CRPS than in the general population (420), and an association between CRPS I and Ehlers-Danlos syndrome, an inherited disorder of collagen, has been proposed but not quantified (399). Certain genetic polymorphisms may render patients more or less susceptible to the development of CRPS after trauma or surgery (184; 392). Rare family clusters have been reported; familial CRPS may be more severe with earlier onset, more affected extremities, and more dystonia (103; 385).

Prevention

Fortunately, complex regional pain syndrome rarely develops without a known predisposing event such as trauma or immobilization. After a fracture, cast tightness should be carefully assessed, and the limb recast if pressure within the cast becomes excessive due to swelling. The cast should be removed and the limb mobilized as soon as it is safe from an orthopedic standpoint. If range of motion appears less than expected, physical therapy should aggressively increase range of motion. Before removal of the cast, motion should be encouraged across uninvolved joints of the same limb. There is controversy over the value of vitamin C in reducing disability (including CRPS) after distal radial fracture. There is level II evidence that the use of vitamin C (500 to 1000 mg for 50 days) from the date of injury reduces the development of CRPS I in patients with wrist fractures (322). This therapy is endorsed by the American Academy of Orthopedic Surgeons (268) and is supported in three meta-analyses and systematic reviews (76; 04; 151). Regional vitamin C during the Bier block for distal radius fracture surgery was also reported to be effective (08). Vitamin C was found to be effective in preventing CRPS after knee arthroplasty (197). A prospective, randomized trial of vitamin C after foot and ankle surgery found it to be effective in the prevention of CRPS (183). Alternatively, Ekrol and colleagues found no benefit of 500 mg of vitamin C for 50 days in patients with distal radial fracture, including the rate of CRPS development, also at level II (124). Also, a meta-analysis of RCTs did not find a benefit (131).

In patients with a previous history of CRPS anywhere in the body, the risk of new CRPS is increased; therefore, surgery should be restricted to operations with unequivocal indications and no available alternative conservative treatment and postponed until CRPS I signs are minimal, if possible (322). There is level III evidence that perioperative administration of salmon calcitonin may help prevent development of CRPS I postoperatively (322).

CRPS has been described in patients with hemiparesis due to stroke, especially in the upper extremities, due to biomechanical factors and microtrauma (72). Care should be taken post-stroke not to injure any joint or joint capsule, especially the shoulder. All caregivers should be informed of the loss of sensation on the affected side and the consequent absence of forewarning by the patient with passive movement.

Differential diagnosis

Peripheral neuropathies, particularly small fiber neuropathies, secondary to genetic abnormalities, diabetes mellitus, uremia, alcoholism, etc. can include symptoms of burning pain, allodynia, hyperalgesia, and sudomotor and vasomotor changes. However, symptoms are symmetrical, tend to begin in distal lower limbs, and affect upper extremities later and not as severely. If large (A alpha beta) nerve fibers are affected, there are loss of reflexes, weakness, and vibratory and proprioceptive deficits. Trauma is absent, and most peripheral neuropathies progress gradually. However, in a study of the specificity of the 1993 IASP CRPS criteria, Galer and colleagues found that up to 37% of patients with painful diabetic neuropathy met the clinical criteria of CRPS (140). Focal or entrapment neuropathies such as carpal tunnel, cubital tunnel, or thoracic outlet syndromes are not usually associated with trauma and progress gradually, and symptoms and signs are confined to specific nerve(s) distributions.

Vascular abnormalities such as deep vein thrombosis or thrombophlebitis can cause swelling, pain, temperature changes, and discoloration (erythema) of the affected extremity. Unilateral arterial insufficiency can also be painful with clinical signs similar to CRPS. Ultrasonography is necessary to diagnose these vascular conditions. Lymphedema secondary to surgery, radiation, or infection develops insidiously, and pain is usually of an aching quality. It can be diagnosed with lymphoscintigraphy. Also, focal joint inflammatory diseases such as gout, pseudogout, tenosynovitis due to arthropathy, as well as scleroderma must be considered (156). Other orthopedic conditions such as bone infections, stress fractures, mal-fixations, or ligament damage may be considered (358). Of note, three patients presenting with CRPS were found to have an underlying peripheral nerve vasculitis responsive to immunomodulative therapy (342).

Diagnostic workup

There are no gold-standard diagnostic tests in CRPS; the diagnosis is clinical. However, the following tests can be helpful in supporting the clinical diagnosis or ruling out other clinical entities.

Tourniquet ischemia testing. Relief of pain during a period of ischemia has been reported to have a highly predictive value, but low sensitivity, as a screening test (236).

Neurophysiological studies. Nerve conductions and electromyography are often used to document peripheral nerve damage. However, surface electrode nerve conduction studies only measure large (fast-conducting) myelinated A alpha beta fiber activity in mixed peripheral nerves, not the small A delta and C fibers primarily affected in CRPS. Abnormal nerve conduction studies can reveal a generalized or focal, eg, entrapment, neuropathy, thereby determining large-nerve-fiber damage, but coincident small-fiber dysfunction is inferred. Normal nerve conduction studies do not rule out the presence of nerve damage and cannot be used to differentiate between CRPS I and II, as has occurred in some studies (55; 56; 173; 171; 356). The same points also apply to electromyography in which abnormal “neurogenic” patterns only identify axonal injuries to large myelinated motor axons.

Quantitative sensory testing. Quantitative sensory testing is a more formal and sensitive method of testing various sensory thresholds (most commonly thermal, mechanical, and vibratory) than the clinical examination. Similarly, the findings are subjective and dependent on patient motivation and alertness (139) and are not specific to CRPS. Maier and colleagues studied 1236 patients with a variety of neuropathic pain syndromes with qualitative sensory testing and characterized the findings: thermal and mechanical hyperalgesias were most frequent in CRPS (261). The method is not recommended as a routine laboratory test for CRPS (356); it is mainly a research tool at present.

Radiological studies. Three-phase bone scintigraphy (TPBS) and, to a lesser extent, x-ray and bone densitometry, have been used in the diagnostic workup of CRPS patients, as they can reveal focal subchondral or subperiosteal osteoporosis. Some studies have found three-phase bone scintigraphy to be diagnostically helpful (77), particularly in phase 3 and in the upper limb (464), whereas other studies found that not to be the case (292; 457). These may be underutilized for identifying patients whose bone remodeling is a source of pain that can be treated with nasal calcitonin or bisphosphonates (169; 89).

Computerized axial tomography and MRI are less often useful but can identify neurogenic edema of affected bones, joints, and deep tissues. MR neurography and ultrasound are emerging techniques for visualizing nerves (01; 447). They can help identify entrapment neuropathies amenable to surgery (412; 164). MRI/PET fusion may be useful in detecting increased [18F] FDG uptake in affected areas, even when MRI alone is without abnormality (469).

In a small study, osteoprotegerin (OPG), an osteoclastogenic inhibitory factor produced by osteoblasts, was increased in patients with CRPS 12 weeks post upper extremity fracture and related to ipsilateral phase III technetium uptake in TPBS (226). The authors suggest that bone turnover may relate to CRPS pathophysiology, and OPG might be a useful biomarker in CRPS. Bone markers in CRPS were reviewed (221). Infrared thermography has not been very useful in some studies (77), but others have found it helpful in both diagnosis and in assessing treatment results (108).

Punch skin biopsy. This increasingly popular method involves removing a small piece of skin under local anesthesia. This is sectioned and immunolabeled with a pan-axonal marker that permits evaluation and quantitation of the unmyelinated sensory axons, A delta and C fibers. Neurodiagnostic skin biopsy, currently the best diagnostic test for small-fiber polyneuropathy (126), has provided critical research evidence of small-fiber nerve injury in CRPS I, obviating its distinction from CRPS II (311). As normative densities differ at different body locations and with high inter-individual variability, skin biopsy is not useful clinically except for rare situations in which comparison of biopsies from patients’ CRPS-affected and matching unaffected areas confirm focal small-fiber losses. Further, commercially available standard biopsy methods were not successful in detecting intradermal axon pathology and showed no correlation between quantitative sensory testing changes and epidermal nerve fiber densities in 43 patients with CRPS I (208).

Sympathetic nerve blocks. Sympathetic nerve blocks are a controversial subject with a complicated history. For an affected upper limb, the stellate ganglion is blocked with local anesthetic with or without corticosteroids, and for a lower limb the lumbar sympathetic chain is blocked. Following Evans’ enthusiasm, for decades these were used not only for treatment, but a positive response was required for diagnosis of reflex sympathetic dystrophy (132; 133). Few, if any, early studies of sympathetic nerve blocks are considered adequate today. They were unblinded and generally uncontrolled; the importance of the placebo response, which is particularly strong for therapeutic interventions, was unrecognized (87). Later, patients with temporary post-block pain reduction were diagnosed with sympathetically maintained pain; those not responding had sympathetically independent pain. Given the technical and interpretational limitations of this method, and evidence that somatic axons co-innervate most structures traditionally thought only autonomically innervated, use of sympatholytic procedures has waned.

Other testing for autonomic nervous system dysfunction in CRPS, such as deep breathing testing, orthostasis testing, and sympathetic skin response testing, has also not been found to be useful (242).

Management

Detailed meta-analyses have assessed the current status of various therapies in CRPS and should be referred to for completeness (90; 177; 458; 119; 358). An interdisciplinary approach, including psychotherapy and cognitive behavioral therapy, tailored to the individual patient, is advocated because of the frequency of psychological symptoms such as anxiety, depression, and insomnia (358). In general, non-CRPS surgery on an affected limb should be avoided or deferred for one year after the active process has resolved, if possible (Royal College Physicians 2018).

Functional restoration/physiotherapy. Restoration of limb function is the primary goal of CRPS management and may be all that is needed in early, mild, or pediatric cases. Pharmacotherapy should be considered adjunctive. Desensitization, physical therapy to increase range of motion and limb strength, education, a progressive-loading exercise program with pain-avoidance behavior management, gait, postural and balance normalization, relaxation techniques, edema control strategies, vocational support, correction of body perception disturbance, conflict allodynia re-education, general aerobic conditioning, and occupational therapy share the primary goal of increasing function, as well as secondary goals of minimizing effects of disuse such as contractures, osteoporosis, weight gain, skin and muscle atrophy, depression, and disability. Pain management may improve participation (145; 171). However, reviews suggest that standard physiotherapy guidelines should be interpreted cautiously, as there is little evidence of efficacy (94; 441; 391). Nonetheless, the consensus of several expert panels supports the use of physical and occupational therapies in restoring patient function (156; 441). Somatosensory rehabilitation was found to be effective in reducing allodynia in upper limb CRPS (315).

Graded 3-stage motor imagery (GMI) consists of limb laterality recognition, imagined movement, and mirror movement phases over six weeks (295). The mechanisms are unknown but are thought to be cortical and possibly related to neglect (141; 142; 298). GMI had level II evidence in CRPS (295; 94) or “very low-quality evidence” (391). Other studies found a failure of GMI to relieve the pain of CRPS (200) or limited efficacy in improving upper extremity function (235), even with simultaneous transcranial direct current stimulation (234).

Two small studies report the utility of mirror therapy in reducing CRPS I pain in the paretic arm of stroke patients (66; 324). Another preliminary study of mirror visual feedback therapy utilizing virtual reality technology reported positive results in patients with CRPS (365). However, reviews of mirror therapy studies have found evidence of efficacy to be weak (297; 357; 391). Combining mirror therapy with immersive virtual reality synchronous with heartbeat, without tactile stimulation or movement of the affected limb, was found to reduce pain and improve limb function in 24 patients with CRPS (395). There is preliminary evidence that exposure therapy targeting pain-related fear, performed by highly trained clinicians, has value in reducing CRPS I-related disability (105; 296). Self-administered tactile and thermal desensitization therapies are also of benefit (249; 358).

In addition to rehabilitation, patients with CRPS I of the shoulder who were refractory to standard therapies benefited over a 6-month period after one week of continuous interscalene blockade with bupivacaine (106).

Rehabilitation interventions were the subject of a metaanalysis (383).

Evidence-based pharmacotherapy for CRPS. The uncertainties about CRPS definition, pathogenesis, and diagnosis and its rarity make it difficult to conduct clinical trials. The United States has no drugs approved by the Food and Drug Administration for a CRPS indication. Although some patients improve without treatment, particularly pediatric and acute cases, most patients with chronic severe symptoms should consider pharmacotherapy. Several meta-analyses of controlled clinical trials concluded that studies conducted for other forms of peripheral neuropathic pain had better methodology, and few medications were rated at better than level III for CRPS (dimethylsulfoxide 50% cream, calcitonin, and bisphosphonates) and then only in small trials (213; 322). Basically, CRPS pharmacotherapy uses the same medications found effective for other neuropathic pain disorders.

Treatment options for acute CRPS. Small numbers of randomized clinical trials have been conducted for acute CRPS, evaluating agents that reduce bone resorption such as calcitonin and several bisphosphonates, as well as corticosteroids, NSAIDs, free radical scavengers, and vasoactive medications.

Calcitonin is a neuroactive peptide with centrally mediated analgesic effects independent of bone activity and may be effective through both release of beta endorphin in the central nervous system and through inhibition of bone resorption (473). A randomized trial in 24 patients treated subcutaneously for three weeks showed more rapid relief than those in physical therapy only (152), as did a subsequent trial with intranasal delivery (153). However, a third prospective, randomized, double-blind trial demonstrated no efficacy (34). The review by Duong and colleagues found several studies that did not demonstrate calcitonin efficacy, with no studies of calcitonin in CRPS performed after 2009 (119). Therefore, the use of calcitonin is not currently recommended (455).

Small studies and case series support intravenous bisphosphonates in early CRPS in adults and children (02; 440; 271; 134; 446); patients with the warm disease subtype and fracture as a predisposing event fared best (439). A RCT supported the use of intravenous neridronate in acute CRPS I (436). Pamidronate was reportedly helpful in a case series of chronic and acute patients (231; 352; 130). Two meta-analyses of intravenous therapies in CRPS concluded that bisphosphonates were of benefit in CRPS (465; 78). One of the four studies cited by Chevreau and colleagues was an oral study of alendronate (78); another opined that bisphosphonates were the treatment of choice in early CRPS I, and a short-term course of calcitonin was helpful in the more chronic stages (458). A single 60 mg intravenous dose of pamidronate should be considered in suitable patients with CRPS less than six months duration (358). Intramuscular neridronate was also reported to be effective in a randomized, double-blind, placebo-controlled trial (437). Varenna and Crotti suggest that bisphosphonates can exert an antiinflammatory effect in early CRPS, inducing leucocyte necrosis via proapoptotic properties (438).

Long-term bisphosphonates, usually for osteoporosis, rarely may increase fracture risk (381), and intravenous use has been associated with jaw osteonecrosis (16) although not thus far when used in CRPS therapeutic regimens (438).

The successful use of teriparatide, a synthetic analog of human parathormone, in two patients with CPRS has been reported (143).

Corticosteroid efficacy during the acute phase of CRPS has been claimed in several series over several decades, but the methodology of the studies may be inadequate to be definitive according to some authors (225; 82; 49; 167; 35; 14; 198; 316; 233; 423). For example, oral prednisone was more effective than placebo in 23 patients (82). Forty-five patients treated with prednisolone were studied retrospectively and were found to have significantly reduced symptoms and signs and improved quality of life (14). Lower doses of steroids may be as effective as higher doses (316). A single study in the chronic phase was less effective (19), suggesting that the opportunity for use may be during the acute phase.

Free radical scavengers and antioxidants have been used based upon the hypothesis that toxic oxygen and hydroxyl free radicals may be involved in the pathophysiology of CRPS (473). Topical dimethylsulfoxide has shown positive results (472). Dimethylsulfoxide may be more effective in the warm type I presentation and in the lower extremity whereas, N-acetylcysteine may be more effective in the cold type (323). Mannitol was not shown to be effective (320). As noted above under Prevention, there is controversy over the value of vitamin C in reducing disability (including CRPS) after distal radial fracture. There is level II evidence that the use of vitamin C (500 to 1000 mg for 50 days) from the date of injury reduces the development of CRPS I in patients with wrist fractures (322). This therapy is endorsed by the American Academy of Orthopedic Surgeons (268) and is supported in two metaanalyses (76; 04). Alternatively, Ekrol and colleagues found no benefit of 500 mg of vitamin C for 50 days in patients with distal radial fracture, including the rate of CRPS development, also at level II (124). Also, a metaanalysis of randomized controlled trials did not find a benefit (131).

There is little evidence for efficacy of nonsteroidal antiinflammatory drugs in CRPS. Neither a randomized, double-blind, placebo-controlled trial of parecoxib (50) nor a prospective randomized trial of piroxicam (206) showed any benefit. Similarly, aspirin was not effective (127).

Vasoactive and sympatholytic medications have been tried in CRPS. The alpha sympathetic blocker phenoxybenzamine was effective in reducing pain in the acute phase (149; 270). Clonidine patches were useful in reducing hyperalgesia (96). The calcium channel blocker nifedipine was also found to be useful in the acute stage, as was phenoxybenzamine, but both were less effective in the chronic stage (300). The use of such treatments has fallen out of favor, perhaps as the pathoetiology of CRPS has shifted away from sympathetically mediated causes.

Treatment options for pain in chronic CRPS. The lack of strong therapeutic trials in chronic CRPS forces physicians to be guided by the results of high-quality randomized clinical trials for other neuralgias, such as postherpetic neuralgia after shingles and painful diabetic neuropathy. The postherpetic neuralgia trials may be most relevant because postherpetic neuralgia is also a focal (or regional) neuralgia that follows a monophasic insult, and most patients are otherwise healthy (182). Many patients with CRPS will require polypharmacy but should start with a single medication and increase it to the full recommended or maximal tolerated dose. If not clearly effective, it should be discontinued before a second medication is tried. In general, if several drugs are used, they should be from different classes or treating different symptoms. Rarely, a second medication is needed to counteract adverse effects of an otherwise useful medication. Sympathetic blockade or chemical and surgical sympathectomy have long been advocated in CRPS; however, meta-analyses have shown no long-term benefit for these procedures (71; 400).

A Cochrane Review concluded that there was a scarcity of data to support the use of sympathetic blockade in the treatment of CRPS, and the data that are available did not suggest effective pain reduction (312). Alternatively, two more studies indicate that in small numbers of patients a single thoracic sympathetic block can provide positive therapeutic responses 1 and 12 months later, when combined with standardized pharmacologic and physical/occupational therapies, as compared to control subjects (353), and multiple blocks (4 to 19) can cause pain reduction in unblinded patients (452). Kim and colleagues applied pulsed radiofrequency to the cervical sympathetic chain in 12 patients with CRPS and found reduced symptoms lasting a mean of 31.4 days (210). Stellate ganglion block was reported to reduce intraprocedural and postoperative analgesic requirements in patients with CRPS (340).

There are currently four classes of medications with documented efficacy against neuralgias in multiple randomized clinical trials: tricyclics and serotonin-noradrenalin reuptake inhibitors, opioids, gabapentinoids, and topical or systemic local anesthetics. Systemic local anesthetics are underutilized for patients with severe refractory pain including CRPS. Lidocaine cannot be given orally but can be delivered by continuous subcutaneous injection using external metered pumps of the kind used for insulin (73). Topical local anesthetics, available as viscous lidocaine gels, creams, and sprays, may be helpful. The strongest evidence is for the 5% lidocaine patch, whose use is supported by studies in patients with peripheral neuropathic pain, including CRPS (279). Lidocaine may exert its analgesic effect in neuropathic pain via suppression of ectopic activity in injured afferent A- and C-fibers (215).

Injections of local anesthetics into the wound scars of trauma related CRPS have been reported (284). Local lidocaine injections combined with oral citalopram has been reported to be effective (135). Erector spinae blocks at T2 was reported to improve upper extremity pain (138).

Gabapentin was not shown to be effective in a reported trial (422), but three other trials did support the use of gabapentin, and one supported the use of amitriptyline (199). Pregabalin use in pediatric CRPS has been reported in case reports only (199).

Treatment options for other symptoms in chronic CRPS. Treatment of dystonia is complicated, because prolonged tonic postures can lead to fixed contractures that require orthopedic therapy, such as serial casting or tendon release. Baclofen is the current treatment of choice. Because it is sedating, many patients cannot tolerate the oral high doses often needed to lessen dystonia. Administration by intrathecal pump is an effective option that is underutilized, although pharmacological and mechanical complications are common (339; 432). There is no evidence of efficacy of long-term use of muscle relaxants such as benzodiazepines or cyclobenzaprine. Botulinum toxin injections are useful for focal dystonias limited to small areas but are impractical and too expensive for widespread dystonias. Amantadine can be considered for tremor. Rare patients with CRPS have edema severe enough to distort their tissues and prevent adequate tissue oxygenation and nutrition. This can lead to skin ulceration and infection; treatment is usually by limb elevation, compressive garments, or exercise as tolerated.

Promising pharmacological therapies. The NMDA receptor antagonist ketamine has achieved widespread attention and use for refractory CRPS because of the increasingly recognized role of increased glutamate release and NMDA receptor function in spinal dorsal horn physiology in nerve injury models of neuropathic pain (471). High-dose infusions (up to 100 mg for 4 hours daily for 10 days) in a randomized double-blind placebo-controlled study produced significant pain reduction (375). Another study of low-dose intravenous ketamine (22.2 mg/h/70 kg) in chronic CRPS I patients produced significant pain relief but without functional improvement (389). Infusions of ketamine from 5 to 20 mg/h/70 kg over 100 hours also produced prolonged pain relief (93). Administration of anesthetic doses over a 5-day period was beneficial in an open-label phase II study of 20 patients. Complete remission occurred at one month in all patients, at three months in 17 patients, and at six months in 16 patients, although complications ensued (209). Differing length of time of infusion may be needed for lower versus upper extremity symptoms (214). Ketamine may be more likely to be effective in patients with sympathetically mediated pain proven by sympathetic blockade (85). At present, there is tempered enthusiasm (374; 327) and tempered concern (28) for the use of ketamine in neuropathic pain states and CRPS. Several meta-analyses suggested that ketamine is a promising therapy (15; 465; 79), whereas others found weak evidence or a lack of evidence for efficacy in CRPS (86; 88). Three reviews discuss the limited evidence in favor of ketamine infusions and the lack of agreement regarding optimal infusion dosage, rate, and duration (162; 260; 254). Complicating ketamine therapy in CRPS is the risk of liver damage (306; 185) and impairment in cognitive executive functions (212), which may be related to prolonged or frequent infusions within a short time interval. Topical ketamine is under consideration to bring higher local tissue concentrations in the affected limb than can be achieved with systemic administration (366).

The increasing evidence for a significant role of neuroinflammation, the immune system, and autoimmunity in the pathophysiology of CRPS has prompted consideration of new classes of therapies. The finding of local overproduction of TNF-alpha in CRPS has prompted the suggestion that anti-TNF-alpha therapy should be investigated (448). Infliximab, a TNF-alpha inhibitor, reduced cytokine levels and pain in two patients with CRPS (195). This effect was shown again in 11 out of 15 patients with CRPS in 2023 (424). Unexpected improvement occurred in a CRPS patient being treated for multiple myeloma with thalidomide (which inhibits proinflammatory cytokines such as TNF) (341) and in two open-label thalidomide trials in patients refractory to other therapies (336; 376). Unfortunately, lenalidomide, a less toxic but more anti-inflammatory thalidomide derivative, was not found efficacious in a phase II study of 184 subjects (272).

Although a preliminary short-term trial of low-dose (0.5 g/kg) intravenous immunoglobulin (IVIG) in 13 patients with CRPS was positive for pain reduction (154), and long-term administration of subcutaneous immunoglobulin in four of these patients resulted in remission of pain and improvements in function in two of them 12 months after the final infusion (161). A multicenter randomized placebo-controlled trial of low-dose intravenous immunoglobulin in 111 patients, given on days 1 and 22 post-randomization, failed to show efficacy over a 6-week period (158). Birklein and Sommer hypothesize that this lack of efficacy of intravenous immunoglobulin may be due to the low dose, its failure to cross the blood brain barrier to affect immune-mediated central nociceptive sensitization, and because peripheral immune mechanisms may not be present in chronic CRPS (42).

There is a report of a patient with acute CRPS responding to high-dose intravenous immunoglobulin (2 g/kg) (278). Plasma exchange therapy was retrospectively studied in 33 CRPS patients; 30 reported significant pain reduction following the initial series of exchanges, and 24 patients that received maintenance therapy with plasma exchange or immunomodulating agents continued to experience reduced pain (13).

On the basis of presumed focal neuroinflammation in CRPS, Thor and colleagues injected three CRPS patients, 2 to 6 months post-injury, at tender chronic constrictive injury points (perineurally), with 2 to 5 mL of 5% dextrose (414). All three patients had resolution of their CRPS and were able to participate in physical therapy.

Botulinum toxin, used for years to selectively weaken focal muscle groups in various conditions including movement disorders and spasticity, works by blocking acetylcholine exocytosis at cholinergic synapses. It also inhibits non-cholinergic neurotransmitter (eg, glutamate) and neuropeptide (substance P and CGRP) release from sensory nerve terminals by blocking transient receptor potential vanilloid 1 (TRPV1) exocytosis (12; 416), prompting its evaluation for neuropathic pain conditions. Regional intradermal injections of botulinum toxin improved spontaneous pain, brush allodynia, and cold pain thresholds at the painful sites of 25 patients with focal painful neuropathies (343) and, when used in conjunction with sympathetic blockade with bupivacaine, extended the duration of analgesia in a subset of patients with CRPS (70). In a randomized trial against local anesthetic only in lumbar sympathetic block, it was found to reduce pain and increase the temperature in the affected limb for three months (468). However, botulinum toxin was not found to be effective and was poorly tolerated in 14 CRPS patients with allodynia (362). Dosing may be a key factor in achieving improvement, and treatment results may vary depending on whether the pain is of muscular (dystonia or spasticity) or neural origin (neuropathic) (387). These findings await confirmation.

In a mouse model, metformin, an AMP-activated protein kinase activator, was shown to reduce mechanical allodynia, decrease extremity edema, and reverse burrowing deficits compared to saline (95).

Treatment with hyperbaric oxygen was found to be effective in a single patient (36).

A cannabis inhaler was shown to reduce pain in a randomized, double-blind, placebo-controlled trial for CRPS (10).

Neurosurgical treatments. Rare patients with CRPS with ongoing nerve injury from compression or a lesion within or near a nerve (eg, tumor, vascular malformation) may benefit from surgical exploration, but such lesions need be firmly localized first. Ablative neurosurgery (cutting the nerves or roots innervating a painful area) is currently used in patients with short life expectancies, as pain relief tends to be transient after this procedure, whereas neurofunctional loss is permanent, and anesthesia dolorosa can develop in denervated body parts. However, Dev and colleagues report a positive outcome in 49.5% of patients with CRPS who underwent fluoroscopy-guided lumbar sympathetic neurolysis; a short duration of pain and concurrent cold intolerance predicted success (107). Another surgical technique, regional subcutaneous venous sympathectomy, resulted in significant improvement in limb function in 12 of 16 patients with CRPS II (170).

Much better results have been obtained for implanted neural stimulators, which augment the function of surviving neurons. Spinal cord stimulation (SCS) has been used for more than half a century to treat CRPS pain. The device is placed by an often minimally invasive technique with insertion of a stimulator electrode into the dorsal epidural space close to the spinal cord, most frequently in the cervical or thoracic areas. Wire leads may be placed through needles or paddle electrodes through minimally invasive laminotomies. A pulse generator is placed in a subcutaneous pocket. There is experimental evidence that SCS is associated with activation of several spinal serotonin receptors--5-HT(2A), 5-HT(3), and 5-HT(4) (396), and it may reduce the affective component of pain due to decreased cortical connectivity between somatosensory and limbic areas and increased connectivity between somatosensory areas and the default mode network (102). Spinal cord stimulation has also been found to have positive immunomodulatory effects with decreases in both pro- and anti-inflammatory cytokines, chemokines, and growth factors in patients with CRPS (230). Spinal cord stimulation can improve sensory symptoms such as allodynia, hyperalgesia, and quantitative sensory testing (229).

Reviews and meta-analyses demonstrate long-term reductions in pain intensity, increased function of affected extremities, and improvements in quality of life with spinal cord stimulation (408; 363; 444; 445). Spinal cord stimulation is usually considered when the disease is not resolving despite maximal medical and rehabilitation therapy. However, a meta-analysis concluded that spinal cord stimulation can be both clinically beneficial and cost effective if used earlier (eg, before chronic opioid therapy) in adults (334) and in children (17). Stimulation may be effective, even in severe cases with high opioid consumption (128), or in patients with allodynia and hyperalgesia (229). Some studies suggest that spinal cord stimulation may not be effective within the first year of CRPS (427) or after three years (207). Another study of the long-term follow-up of 84 patients with CRPS found spinal cord stimulation effective in 63% (147). Another study found a 70% reduction in painful symptoms in all 33 patients with CRPS I treated with spinal cord stimulation and suggested that the usual presurgical therapeutic trial may not be necessary in such patients (349). In another study, 70% of patients continued to use their stimulators at follow up after eight years but were not able to wean pharmacologic support after stimulator implantation (186). Subthreshold high-frequency spinal cord stimulation (10,000 Hz) has also been found to be promising (444; 68). Patients with CRPS may report lesser benefit than other patients during the first days of test electrode placement (275). Spinal cord stimulation may reduce lymphedema and promote skin ulcer healing in some patients (Royal College Physicians 2018). Spinal cord stimulation may be combined with dorsal root ganglion stimulation as well (see below) (148). Unfortunately, randomized clinical trials are few and complications such as infections and mechanical problems with lead position can be problematic. The International Consortium of Investigative Journalists/Associated Press and other media organizations have reported that spinal cord stimulators “account for the third-highest number of medical device injury reports to the U.S. Food and Drug Administration, with more than 80,000 incidents flagged since 2008” (453).

A neurostimulation system utilizing quadripolar epidural leads was used to stimulate dorsal root ganglia (DRG) in 11 subjects with lower extremity CRPS; significant reductions in self-reported pain persisted for 12 months (419). Dorsal root ganglia stimulation was effective in patients with lower extremity CRPS refractory to spinal cord stimulation (160; 467), suggesting that failure of spinal cord stimulation does not negate possible dorsal root ganglia responsivity and, when the two procedures were compared head-to-head, dorsal root ganglia stimulation was found preferable and more successful for lower extremity CRPS (97; 418). Cervical dorsal root stimulation has also been reported to be effective (69). Dorsal root ganglion stimulation has been reported to lose less efficacy over time than spinal cord stimulation (248). Both spinal cord stimulation and dorsal root ganglion stimulation have been shown to be cost effective compared to best medical management, with perhaps a slight edge in quality of life for dorsal root ganglion stimulation (280). Dorsal root ganglion stimulation has been used in pediatric patients as well (328).

Peripheral nerve stimulation is another option, especially in patients who qualify for dorsal column stimulation but are not able to achieve persistent relief (179). Some jury rigging of the electrical leads is needed, and the leads not infrequently shift position (196). A pilot study of ultrasound-guided percutaneous implantation of a cylindrical lead close to the suprascapular nerve or cervical nerve roots in 26 patients with upper limb refractory neuropathic pain (16 with CRPS) found substantial long-term benefit in 61.5% of patients (48). In 165 patients with CRPS treated with peripheral nerve stimulation, pain scores decreased by 1.87, opioid use decreased from 62% to 41%, and 51% of patients reported functional improvement 12 months after surgery (80). Direct stimulation of the brachial plexus and the sciatic nerve have also been reported to be effective (136; 65).

Chronic motor cortex stimulation by implanted electrode or repetitive transcranial magnetic stimulation has shown beneficial effects in neuralgia (51; 244; 11; 255) but has not been sufficiently evaluated for CRPS. One meta-analysis found that invasive brain stimulation produces greater pain reduction than noninvasive brain stimulation, but the advantage was small (255). One small study found 10 daily sessions of repetitive transcranial magnetic stimulation (TMS) to be effective in some CRPS-I patients, but only during the stimulation period. No significant effect was seen one week after the last stimulation (326). Transcranial magnetic stimulation’s characterization of cortical neuroplasticity and its therapeutic efficacy in very limited studies is reviewed (303).

Some have advocated scar excision for CRPS associated with wounds, often after successful local anesthetic infiltration into the suspect scar (333; 284).

Limb amputation for chronic refractory CRPS is rare and controversial. Extensive consultations with a multidisciplinary team are necessary after the CRPS diagnosis is affirmed, complaints are at least one year (368) or two years (358) in duration, the aforementioned treatments have failed, and the consequences of amputation have been well discussed and considered by the patient (358; 368). One study reported persistent postamputation stump pain in 28 of 34 limbs (109). Other studies of 36 cases suggest that although approximately three-quarters of patients suffer from postamputation phantom, stump, and residual limb pains, and validation of this therapy is weak, many patients found amputation beneficial (227; 44). Nineteen patients who underwent amputation did better postoperatively when compared with 19 patients in whom amputation was considered but not performed (285). Amputations in nine patients were successful (greater than 50% pain relief); three had intermediate success (30% to 50% pain relief); and eight were unsuccessful (less than 30% pain relief). Recurrent CRPS in the stump was reported in six (32%) patients (all in the unsuccessful group), and 17 (89%) experienced some degree of phantom limb pain. Among 47 patients from another institution, 77% reported improved mobility, and 73% reported significant pain reduction (146). There was recurrence in four patients, three of whom had recurrence in another limb. Post-amputation follow-up is strongly advised (358).

Outcomes

It is not clear if early aggressive treatment is more effective, because it is not known whether those patients whose CRPS lasts longer than six months constitute a group with an intrinsically poorer prognosis. However, given current knowledge, aggressive treatment early makes sense and has its advocates (243).

It should be emphasized that many studies in this review used small numbers of patients, sometimes used different diagnostic criteria (although most recent ones used the Budapest criteria without remarking as to whether the clinical or research criteria were satisfied), and employed a wide range of outcome measures (165). Therefore, validation of results is problematic in many references.

For example, one study reported that decreases in exosomal micro RNA mi-RNA-223 heralded persistent pain, but the study included only 33 patients based upon clinical characteristics (347). Latour and colleagues reported benefit to stellate ganglion cryoablation based upon Budapest criteria, but in only six patients (238).