Complex regional pain syndrome (CRPS) is a collection of painful conditions that are characterized by a continuing regional pain, disproportionate to the usual course of any known trauma or other lesion (Harden et al., 2007). The pain is not in a specific nerve territory and typically has a distal predominance of abnormal sensory, motor, sudomotor, vasomotor, and trophic findings. It commonly occurs after trauma, however in up to 9% of cases, there is no traumatic trigger (Baron et al., 2002). The incidence in Europe is 26/100,000 person-years (de Mos et al., 2007), and ~4% of patients sustaining wrist fractures will go on to develop CRPS in the following 4 months (Moseley et al., 2014). The diagnosis of CRPS is largely one of exclusions and based on clinical assessment. The International Association for the Study of Pain first established a set of diagnostic criteria for CRPS, which were then revised in 2003 and termed the “Budapest” criteria (Harden, 2005).

The pathophysiology of CRPS remains elusive, and consequently treatment options are still inadequate. Several pathophysiological hypotheses have been suggested: Central nervous system (CNS) reorganization (Maihöfner et al., 2007; Lebel et al., 2008; Cohen et al., 2013), inflammation (Huygen et al., 2004; Parkitny et al., 2013), neurogenic inflammation from neuropeptides (Birklein et al., 2001), peripheral nerve pathology (van der Laan et al., 1998; Geertzen et al., 2015), capillary dysfunction (Schattschneider et al., 2006; Tan et al., 2012), and autoimmunity (Blaes et al., 2004; Kohr et al., 2009; Dirckx et al., 2015). Concerning peripheral nerves, a decrease in myelinated nerve fiber density has previously been demonstrated (van der Laan et al., 1998), and an animal model has suggested it affects primarily large fibers (Guilbaud et al., 1993). Another human study has also shown histological evidence of nerve fiber loss and regeneration in CRPS nerves, with specific large nerve fiber loss when compared to controls (Geertzen et al., 2015). Several hypotheses for peripheral nerve damage in CRPS have been proposed: local pressure from oedema (Guilbaud et al., 1993), microvascular ischemia from oxidative stress(Coderre and Bennett, 2008), retrograde degeneration (Geertzen et al., 2015), neurogenic inflammation(Weber et al., 2001), and decreased metallothionein (oxidative stress) (Oki et al., 2012).

Treatment strategies are based on addressing individual symptoms with known existing therapies to alleviate burden of disease (Perez et al., 2010). These strategies include treating pain following the World Health Organization analgesic ladder, neuropathic pain with anticonvulsants, topical dimethyl sulfoxide, ketamine, nerve blocks, physiotherapy, and psychological support (Perez et al., 2010; Birklein et al., 2015). For intractable CRPS where pain therapies have failed, amputation has been shown to have better outcomes than non-amputation (Midbari et al., 2016). Outcomes measured included several widely used pain and depression questionnaires and indexes. In a case series of therapy resistant CRPS, amputation was also associated with positive outcomes however 24% had recurrence (Krans-Schreuder et al., 2012). Another case series highlighted a 77% rate of phantom pain at 1 year post amputation (Bodde et al., 2014). A systematic review on the topic concluded that decision-making for amputation still remains a complex process due to insufficient evidence, and unpredictable outcomes, however, it should not be ignored as an option (Bodde et al., 2011, 2014).

We describe a rare case of CRPS leading to forearm amputation and the subsequent histopathological study of peripheral nerve tissue from the amputate. Our results demonstrate a selective large fiber degeneration which is a novel finding and in keeping with current knowledge on CRPS pathophysiology.

Routine hospital histology of skin, muscle, fat, and forearm nerve specimens showed only mild superficial chronic inflammation in the superficial dermis and several foci of calcified material in the muscle.

We describe ultrastructural changes in nerve fibers in the context of a rare case of CRPS requiring amputation of the distal forelimb. Results indicate that degeneration occurred selectively in large myelinated and small unmyelinated nerve fibers as compared to the medium and small myelinated nerve fibers, which were conserved. This has potential implications for the understanding of the pathogenesis of CRPS.

This study is believed to be the first showing ultrastructural images of large fiber (Aα) degeneration in CRPS using TEM. This was further corroborated by our light microscopy images showing large nerve fiber degeneration in all nerve fascicles. Our results are in line with a recent study of 15 severe CRPS patients who underwent amputation and where histological evidence of peripheral nerve pathology was found on light microscopy in radial, median, ulnar, tibial, and sural nerves (Geertzen et al., 2015). The Geertzen study found fiber loss in 81% of nerve biopsies and more importantly, it demonstrated a loss of large fibers (>12 μm) in sural nerves using light microscopy. We show the presence of histological degenerative features in large myelinated nerve fibers using TEM, thereby providing additional detail to that described by Geertzen et al. Similarly, an animal study of experimental pain, which used chronic loose ligatures of the sciatic nerve, also showed a selective loss of large fibers at 10 weeks, which did not affect small myelinated fibers (Guilbaud et al., 1993).

We propose that degeneration of Aα fibers may lead to an imbalance in nerve signaling if the majority of axonal messaging goes through the smaller healthy Aδ fibers, which may inappropriately trigger pain. As previously stated Aδ fibers are myelinated fibers that transmit pain and temperature and their hypersensitivity will cause symptoms of allodynia to light touch or temperature, or hyperalgesia. The concept of surviving fibers inappropriately firing when neighboring fibers are degenerating has long been established in neuropathic pain (Oaklander and Fields, 2009). In this respect, we propose treatment strategies for CRPS could potentially target the development of new drugs to modulate inappropriate firing of fibers (e.g., Pregabalin) and interventions that could decrease the extent of fiber degeneration.

Furthermore, we postulate that large Aα fiber degeneration contributes to motor symptoms in CRPS. As previously stated Aα fibers have a somatomotor function in the peripheral nervous system. Our patient suffered from motor weakness as CRPS progressed, and our histological findings are in keeping with this clinical picture. Movement disorders such as bradykinesia and dystonia affect around 25% of patients with CRPS (van Hilten, 2010). It could be that the range of large fiber degeneration in the disease predisposes patients to suffer from varying amounts of motor symptoms, depending on severity. Large fiber degeneration may consequently affect the associated innervated muscles. Evidence of atrophic muscular changes in CRPS has been shown in human studies (van der Laan et al., 1998; Hulsman et al., 2009). Muscular atrophy is also a well-known consequence of peripheral nerve degeneration in nerve injuries. Other hypotheses for motor dysfunction in CRPS which have been suggested include alterations in sensorimotor processing in the spinal cord (van Hilten, 2010) and disinhibition of the motor cortex (Schwenkreis et al., 2003).

Of interest, Geertzen et al. (2015) noted fiber regeneration in 86% of biopsies. In their study, they considered clusters of small myelinated fibers as evidence of regeneration. Whilst we also found these clusters in our electron microscopy analyses, we have not considered them as evidence of regeneration as these could also represent groups of smaller non-regenerating fibers. Histological evidence of regenerating fibers are generally viewed as fibers with thin myelin sheaths, as the regenerating fibers will initially lack myelin (Geuna et al., 2009). Whilst these were present in small frequencies in our TEM images, they were not a predominant feature. Furthermore, it is not certain that in CRPS simultaneous occurrence of both degeneration and regeneration takes place along the entire nerve.

Importantly this study has also demonstrated small unmyelinated C fiber pathology in upper limb CRPS nerves. This had previously been shown in 4 out of 8 patients in lower limb sural nerves (van der Laan et al., 1998). Partial C fiber degeneration can explain signs of autonomic dysfunction seen in CRPS, including swelling, skin, and temperature changes. Indeed, the concept of CRPS partly being a small fiber neuropathy has previously been described (Oaklander and Fields, 2009), and our results are consistent with this theory. This hypothesis is supported by studies examining skin biopsies of CRPS patients, which noted a decreased density of normal C fibers (Albrecht et al., 2006; Oaklander et al., 2006). In one of these studies (Oaklander et al., 2006), quantitative sensory testing (QST) was carried out in 344 CRPS patients, and small fiber afferent pathway dysfunction was seen in 39% of patients, whereas 48% had large fiber dysfunction. These QST results would be in keeping with our findings of simultaneous small fiber neuropathy and large fiber degeneration.

Radial nerve myelinated fiber densities in our study and healthy radial nerve densities in O'Sullivan et al. were similar, particularly when comparing the medians. Our results (Table 3) suggest myelinated nerve fiber densities were similar to both disease and disease-free controls. Although our study showed evidence of large myelinated fiber degeneration, this did not affect the fiber density. The reason for this is unclear. Similar findings have previously been reported, where evidence of nerve pathology did not translate to differences in nerve density (Geertzen et al., 2015). Myelinated nerve fiber density ranges considerably between individuals and is also known to decrease with age. In our study the patient's age was similar to the median ages of patients included in both studies used for comparison and therefore an overall comparison can be made. However, the low sample numbers (n = 1) means our myelinated nerve fiber density data should be interpreted with caution. The presence of potential regenerating fibers, as proposed by Geertzen et al. could also mean that nerve fiber loss is compensated for and densities consequently unaffected.

Limitations of this study are the sample size (n = 1) and the absence of a patient matched healthy nerve control, which are both dictated by the nature of the clinical case. Indeed, it would have been unethical and harmful to the patient to take a section of normal nerve from this patient's contralateral limb. Studying peripheral nerves in CRPS has long been challenging due to the limited scope of obtaining diseased nerve samples and healthy age-matched controls. For this reason there are very few studies in the literature that have examined peripheral nerves in CRPS (van der Laan et al., 1998; Oki et al., 2012; Geertzen et al., 2015). Furthermore, morphometric studies of healthy radial, ulnar and median nerves are also scarce, and it is difficult to draw comparisons with the limited data available in the literature. Amputated upper limbs are often secondary to vascular disease, extensive trauma, or oncological reasons where systemic and local factors may render these amputated limbs inappropriate to be truly considered as healthy controls. In addition, there is also no accurate or reliable animal model for CRPS.

Low voltage electrical injuries (<1,000 V) tend to not produce any neurological sequelae and rarely require clinical follow up or hospital admission. For this reason we know little about peripheral nerve changes following low voltage electrical nerve injury. It is unlikely our patient's condition was caused by low voltage electrical injury alone, without an additional pathological process such as that of CRPS. Electrical injury has been known to be a trigger for CRPS (Cohen, 1995). In a case series of 22 patients with low voltage electrical injury, 50% had immediate neurological symptoms, which resolved in 9 out of 11 patients (Grube et al., 1990). Another study of electrical injuries found a higher rate of numbness post injury in the low voltage group as compared to high voltage and said these symptoms appeared on average 2 months after the injury (Singerman et al., 2008). In one case series of mononeuropathies following burn injury (n = 32), low voltage electrical injuries were the cause of 50% of these whilst 43.7% of cases were caused by thermal burns alone (Tamam et al., 2013).

It is possible that our histological findings were due to chronic disuse of the upper limb alone, as our patient's nerves were examined at 10 months post injury. Immobilization and disuse are known factors contributing to CRPS, with one study reporting up to 47% of all CRPS sufferers as having a history of medically imposed immobilization (Allen et al., 1999). Another study found that hand surgery elective patients who wore a cast postoperatively had hypersensitivity and persisting pain for up to 1 month after cast removal (Pepper et al., 2013). This study also showed increased levels of inflammatory mediators in skin biopsies of these patients, similar to that seen in CRPS patients. However, there is relatively little literature on histological changes in disused peripheral nerves. In one animal study, disuse following tenectomy resulted in nerve fiber atrophy for “fast” muscles, whilst hemicordotomy resulted in an increase in diameter of nerve fibers supplying both “fast” and “slow” muscles (Calder and Pollock, 1985). In another study examining disused peripheral nerves following stroke, the frequency of abnormal teased nerve fibers was significantly increased with abnormal internodes frequently “clustered” and showing reduction in myelin thickness (Pollock et al., 1984). This same study also found the mean diameter of myelinated nerve fibers was reduced.

In summary, we demonstrate selective large myelinated fiber degeneration in the upper limb peripheral nerves of a patient with CRPS. We recognize the limitations of a single patient study and lack of control nerve examination; however, our findings have been placed into the context of all previously published literature and present an important hypothesis on the peripheral nerve pathophysiology of CRPS.

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.