J Diub Comp, 6:58-63

What Causes Neuropathic

Pain?

Andrew Boulton

ABSTRACT Although the etiology of diabetic neuropathic pain remains enigmatic, two types of neuropathic pain are now recognized: nerve trunk pain following a discrete often vascular lesion as seen in ocular mononeuropathy and hyperalgesic pain as seen in diffuse involvement of peripheral sensory fibers. Our understanding of the causes of hyperalgesic pain has been furthered by studying structural changes in sural nerve biopsies from subjects with symptomatic neuropathy or by assessing biochemical changes that lead to positive symptoms. Recent evidence suggests that both small and large fibers are involved in painful neuropathy, and axonal atrophy may contribute to

pain generation. The observation that acute painful neuropathy may follow either periods of unstable glycemic control or sudden improvement of control (“insulin neuritis”) suggests that blood glucose flux may precipitate pain. Sudden changes in glycemia may contribute to the generation of impulses in nociceptive fibers or even induce relative hypoxia and axonal atrophy. Pain might arise from ectopic impulses in dorsal root ganglion cells or may even be of spinal or central origin. Thus, a combination of structural and functional changes in peripheral nerve is most likely to lead to the generation of neuropathic pain. (Journal of Diabetes and Its Complications 6;1:58-63, 1992.)

INTRODUCTION

gists are among those who have suffered from diabetic neuropathic pain. R. D. Lawrence, founder of the British Diabetic Association, experienced diabetic neuropathy; his comment was that he did not like peripheral neuritis because it interfered with work. The potential mechanisms for neuropathic pain* are listed below:

I

n an attempt to address the question, What causes diabetic neuropathic pain?, the best that presently can be done is to review some of the evidence provided by investigation and take license to offer expert speculation. No one can claim to be an expert in this subject, because in reality, the true cause or causes of pain in diabetic peripheral neuropathy remains enigmatic. If the clinical classification of the diabetic neuropathies (see Table 1) are examined, it is evident that most of the polyneuropathies and the mononeuropathies can be associated with painful syndr0mes.l This discussion will concentrate on sensorimotor neuropathies that give rise to the dysesthetic pain common to many diabetic patients. Many well-known diabetolo-

Reprint requests to be sent to: Dr. Andrew Boulton, The University of Manchester, Department of Medicine, The Royal Infirmary, Manchester Ml3 9WL, England. 0 1992 Journal of Diabetes and Ifs Complications

nerve trunk pain axonal degeneration/regeneration neuroma properties -ectopic impulse generation -ephapses small fiber disease-A delta and C dorsal root ganglion involvement peripheral blood flow abnormalities glycemic flux Nerve trunk pain, experienced in some of the mononeuropathies, is probably related to ischemic involvement of the nervi nervorum supplying the nerve sheath. The remaining causes listed relate to dysesthetic pain. Axonal degeneration/regeneration has 1056-8727/92/$5.00

WHAT CAUSES NEUROPATHIC PAIN?

1 Diab Comp 7992; 6:l

TABLE 1. CLINICAL DIABETIC NEUROPATHY CLASSIFICATION Mononeuropathies

Polyneuropathies Sensory

Isolated Cranial Truncal Multiple

acute chronic sensorimotor Autonomic Proximal motor Truncal

been proposed by many workers, especially if involving small fibers, the small myelinated fibers, and unmyelinated C fibers. Neuroma properties, including ectopic impulse generation and ephaptic transmission

also have been proposed. Ephaptic transmission relates to electrical cross talk between nociceptive afferents and perhaps sympathetic efferent fibers. There may be involvement of cells in the dorsal root ganglion area, giving rise to neuropathic pain. Additionally, abnormalities of peripheral blood flow and glycemic flux may be contributory. The characteristics of the main types of neuropathic pain are listed in Table 2.3 The common dysesthetic pain is usually described as being of a burning, tingling, or electric nature not previously experienced by the patient. The distribution is cutaneous, and it is variable in intensity, usually described as being shooting and lancinating with nocturnal exacerbation. There are no specific relieving or exacerbating factors. A possible cause is increased firing of damaged or abnormally excitable nociceptor fibers or sprouting fibers. More familiar to the patient is trunk pain, described as aching or “knife-like,” similar to toothache pain. This is the kind of pain that certain mononeuropathies

59

can be associated with. The distribution is deep, de-

scribed as being within the nerve itself. It is often continuous, though it may wax and wane. This pain is relieved by rest, especially of the affected area, and worsened by movement or stretching of the involved nerve. This is believed, as suggested above, to be caused by increased firing due to physiologic stimulation of nociceptive fibers in the nervi nervorum. The topic for the rest of this discussion, however, will be restricted to dysesthetic pain. Neurogenic pain that is dysesthetic is frequently characterized as a burning, shooting pain, often associated with a sensory deficit. Patients often describe the sensation to be as though something were gnawing away at their feet continually. Pavy,4 writing in Washington in 1887, described this very well as being ‘of a burning and unremitting character’. Hyperesthesia is common, and there is often autonomic instability. Dysesthetic pain is thought to be resistant to narcotic analgesics; although, that is now somewhat controversial. In the evaluation of neuropathic symptoms, there is a major problem. Huskisson,’ who has written much on the assessment of pain, suggests that neuropathic pain, or any pain, must be evaluated directly because the pain, as he described, is a personal psychological experience that an observer could play no legitimate part in assessing. Physicians must avoid changing the interpretation of symptoms by placing labels in medical terms onto the pain that the patient describes. Perhaps it is because of the great difficulty that sufferers have in describing their pain that the very high coefficients of variation have been reported for repetitive assessments of painful symptoms. For this reason, this perhaps is not the type of evaluation that should be followed in clinical trials. For example, Attali and Valensi,6 in a paper presented last summer to the European Association for the Study of Diabetes,

TABLE 2. PAIN CHARACTERISTICS Dysesthetic

Nerve Trunk

Description

burning, tingling, electric

aching, knife-like

Recognition

unfamiliar

familiar-like

Distribution

cutaneous

deep, nerve trunk itself

Constancy

variable-shooting, exacerbation

Relief/exacerbation

no specific features

Possible causes

t firing of damaged/abnormally excitable nociceptors or sprouts

Adapted from Asbury & Fields, 1984.3

lancinating nocturnal

toothache

continuous, may wax and wane

better with rest, worse with movement, stretch t firing due to physiologic stimulation of nociceptor fibres

60

BOLJLTON

1 Did Comp 1992; 69

describe a coefficient of variation of over 150% for the repeated assessment of painful symptoms. Methods of studying the causes of neuropathic pain include looking at experimental diabetes, experimental neuroma, human nerve morphology, percutaneous microneurography, intraneural microstimulation, and the effects of pharmacological intervention. Each of these will be discussed briefly. METHODS

OF STUDYING

PAIN

Structural Changes and Neuropathic Pain. I will offer only a few comments about structural changes. Morphologic theories regarding changes in human nerves include axonal atrophy, axonal degeneration and regeneration, sprouting, ephaptic connections, and small fiber neuropathy. However, some important inconsistencies should be noted: unmyelinated fiber loss in patients without pain may be similar to that in patients with pain, ephaptic transmission of impulses has yet to be demonstrated in diabetes, and sprouting is not simply confined to painful neuropathies. Nevertheless, the fact that sprouts may be seen in patients without pain does not necessarily mean that they are not involved in the generation of pain. It may be that sprouting, together with some functional change, leads to pain. The functional change may not be present in those patients who do not experience pain. Combining 50 sural nerve biopsies from Sheffield and Manchester, Malik (unpublished observations) recently presented the following results: included were 10 nondiabetic control subjects, 10 non-neuropathic diabetic subjects, 15 patients with positive (painful) symptoms, and 15 patients with numbness or absence of symptoms (negative) but with peripheral nerve disease. As expected, in both groups of neuropathic patients (with and without pain), there was significant reduction in peroneal nerve conduction velocity as compared with controls. Thermal discrimination threshold, though significantly abnormal in both neuropathic groups, did not differ between these groups. Of course, both of these patient groups had thresholds significantly higher than the non-neuropathic diabetic controls. Similarly, vibration perception thresholds were elevated substantially, both in patients with painful and painless syndromes. In myelinated fibers, the fiber density was significantly reduced in both of unassociated neuropathic groups. Assessment Schwann cell profiles, which is a sensitive measure of axonal degeneration in unmyelinated fibers, showed significantly increased numbers of unassociated Schwann cell profiles both in patients with painful and painless syndromes. Perhaps most interesting is the combination of increased density of unmyelinated axons with decreased axonal diameter in the subjects

without pain, suggesting that regeneration is occurring in this group rather than in the group with pain. This, of course, is not in total agreement with earlier findings; ‘8’ however, previous investigation was conducted mainly on myelinated fibers. Blood Glucose Flux and Pain. Another important consideration is that of glycemic control. Many studies have shown that neuropathy is associated with suboptimal glycemic control. In addition, acute painful neuropathy may follow an acute metabolic upset. Accordingly, several groups have observed that stable glycemic control may improve painful symptoms.9-12 Some years ago, a group in the United States looked at pain perception in hyperglycemic states.13 Findings suggested that blood glucose flux is associated with decreased pain tolerance or even modulates opioid receptors. It was also suggested that if this was the case, then, in treatment of painful neuropathy, it may be the stability in blood glucose control-rather than the level-that is important. Support for this theoretically could be provided by observations of patients with insulinoma, who experience both blood glucose flux and neuropathic pain. r4 The studies of Morley et al.13 in 1984 examined normal subjects following an intravenous infusion of 50 g glucose. Associated with the transient hyperglycemia were demonstrated a transient decrease in maximum tolerated pain and a decrease in pain threshold. However, ainore recent study, from the pain center at Liverpool, looked at this in a more controlled manner. In a double-blind, cross-over study, Chan et al. l5 infused eight nondiabetic subjects with either glucose or saline and subsequently tested them for perception of a noxious heat stimulus. The previous study conducted by Morley was not double blind and, rather than a thermal stimulus, employed an electric stimulus, which is more likely to excite many fiber types. No change in pain thresholds was detected whether the subject was given glucose or saline, suggesting that acute hyperglycemia does not effect pain perception threshold. Again, this does not mean that hyperglycemia or blood glucose flux is not relevant to the etiology of neuropathic pain. This is simply a study of effect on pain perception threshold. The studies conducted on the Bio-breeding (BB) rat by Burcheil et al.“j and published in 1985, showed spontaneous discharges in afferent axons were more common in hyperglycemic animals. These discharges occurred in potentially nociceptive C fibers and occasionally in A delta fibers, and generally originated in the region of the dorsal root ganglion in this animal model. This might explain the syndrome of painful/ painless leg, with spontaneous activity proximally in an axon with a distal nonconducting segment, giving rise to spontaneous pain. In such a patient, there

] Diab Comp 1992; 6:l

would be peripheral anesthesia because of the distal nonconducting segment.

WHAT CAUSES NEUROPATHIC

PAIN?

61

This increase in blood viscosity was associated with a reduced basal tissue plasminogen activator. Both of these changes, of course, promote coagulation. The Blood Flow and Neuropathic Pain. Diabetic neuroconclusions were that acute changes during hypoglypathic patients with painful symptoms frequently cemia may lead to local impairment of perfusion and have warm extremities perhaps caused by arterioveperhaps transient hypoxia. Many patients report that nous shunting, as we and others have described.17J8 neuropathic symptoms are more pronounced during Could, therefore, an increased blood flow be important periods of blood glucose flux, either hypo- or hyperin the causation of neuropathic pain? A number of glycemia. It is tempting, therefore, to speculate that variables were measured in a study from King’s Collocal biochemical or hypoxic changes during such glylege Hospital, London, published at the end of 198419 cemic instability might induce increased firing in and conducted in patients with acute painful neuropthese silent or sleeping nociceptors that leads to the athy and those with chronic severe, but painless, sendysesthetic symptoms of neuropathy. Such a theory sory neuropathy. The skin temperature and the restwould link the blood glucose flux and blood flow abing blood flow were significantly elevated in both normality theories of the etiopathogenesis of neurogroups of subjects. Importantly, in the acute painful pathic pain. neuropathy group, reduction of blood flow using a At this point it would be useful to describe and dissphygmomanometer cuff to reduce the arterial inflow cuss the studies of single fiber recording performed resulted in a reduction in painful symptoms. Clearly, with percutaneous microneurography. In this techthis is a preliminary observation from an uncontrolled nique, the tip of the tungsten electrode (small diamestudy, as the authors pointed out; however, it is a very ter, l-15 pm) is inserted into a peripheral nerve. Any pertinent observation, and further work is needed in superficial nerve can be studied, and recordings have this area. One could speculate on therapeutic implicabeen made from single afferent fibers of all types. Abtions should these findings be confirmed. Poor connormalities of afferent nerve fibers also can be studtrol may lead to increased regional blood flow-peried; however, it should be noted that this is a highly haps a link between metabolic control and the probdifficult technique-it is invasive, and one study may lems of arterial inflow and increased peripheral blood required 3-4 h and not produce any useful results. flow. In this case, there would be pharmacotherapeuIn intraneural microstimulation (INMS), the imtic implications. We know that sympathomimetic planted percutaneous microneurography electrode agents, such as ephedrine, have been shown to recan be used to stimulate neural activity to determine duce high blood flow in the neuropathic leg.20 One the quality and projection of sensations. Then the could speculate that if this is the case, such drugs projected field from INMS can be compared with the might be useful. Conceivably, even surgical intervenreceptive field of microneurography. This is not a unition could be considered to reduce arterial inflow.*’ versally accepted technique.23 Intraneural Techniques for the Study of Pain. BeWhat have we learned by these two techniques? fore examining newer theories, the classical theories In terms of nociception, intraneural microstimulation of pain will be reviewed. Von Frey described pain generally elicits a sharp, pricking, stinging, dull, or as a sensation of the skin that originates in the free burning pain. The sharp pain has been shown to be epithelial nerve endings, while Sherrington described conducted by A delta fibers. The burning or dull pain it as a psychical complex adjunct to protective re- is conducted by C nociceptive fibers, but this depends flexes. A more recent suggestion appearing in an edi- on the characteristics of the stimulus and the site of torial from Pain last year’l proposed that pain might recording. 23 We have learned that pain from inresult from mechanisms other than the firing of spe- traneural microstimulation is reduced by vibrating cific nociceptors. In other words, again, this is against and cooling within a projected area. This is probably the classical Sherringtonian concept. This hypothesis due to central gating. Using INMS, it has been shown suggests that the large myelinated fibers may give rise that many analgesics act centrally, not peripherally. to hyperalgesia. It also raises the question of the po- We have also learned that temporal summation of tential for silent or sleeping nociceptors. These may pain may occur centrally. be unmyelinated fibers in the skin that are normally The controversies alluded to above relate to the folunresponsive to mechanical or thermal stimuli, but lowing three main areas. First, it has been suggested may be reactive to chemical stimuli. In this regard, a that a pressure block in surrounding neurofibers may very interesting paper was recently published in Clini- be induced by impaling a nerve in this way. Second, cal Science.” This group from Edinburgh described a can we actually record and stimulate from single fisignificant increase in von Willebrand factor and bers, or are multiple fibers involved? And third, how acutely increased blood viscosity during hypoglyce- often is it that the stimulated and recorded fibers are mia in diabetic individuals, compared with controls. one and the same?

62

J Diab Comp 1992; 63

BOULTON

In a study conducted in New York by Macke1,24 pain. Both these drugs decrease activity in spontapercutaneous microneurography of the upper limb of neously active fibers of rat neuromas. They also lead to a decrease in sensitivity of the fiber to mechanical eight diabetic patients was examined. He demonstimulation, as suggested by Chabal et a1.30 strated that mechanoreceptor responses were abnorTricyclic drugs are probably the best treatment for mal in up to 30% of the studies, and suggested that microneurography provides a sensitive means of de- neuropathic pain. A very recent study, showed that desipramine, which blocks norepinephrine reuptake tection of early neuropathy. by neurons and potentiates the inhibitory effect in My group previously looked at 15 diabetic patients, nociceptive pathways, is effective in the treatment of 10 of whom had neuropathy, by studying the median diabetic neuropathic pain, and that the pain relief is nerve.= It became apparent that technical difficulties independent of the drug’s action in treating depreswere greater in diabetics than in nondiabetics. One sion.31 Other tricyclics also block serotonin reuptake, suggestion is the difficulty of impaling the nerve might be due to nonenzymatic glycosylation. Side ef- but the fact that desipramine is effective suggests that fects were also more common in diabetic patients. Par- noradrenergic manipulations are clinically more helpful than serotonergic manipulation. Thus, norepiesthesia occurred mainly in the median nerve terrinephrine-mediated inhibition of spinal cord nociceptory, probably due to nerve trunk pain. The receptive tive neurons may lead to analgesia. fields of percutaneous microneurography did appear normal; however, there was a suggestion of increased CONCLUSIONS size of the projected field. Painful as well as mechanical description of sensation was made when stimulatIt is difficult to summarize this discussion because ing the mechanoreceptor fibers. Although this is an much of this is speculation and observation. But I interesting observation, these results are preliminary. would suggest to you that attractive theories on the Something also can be learned from studies conetiopathogenesis of neuropathic pain include neuducted in nondiabetic patients with syndromes simi- roma properties, such as increased spontaneous activlar to diabetic neuropathy. A very interesting case re- ity with reduced threshold, and prolonged after-disport appeared in Brain recently;26 this was a detailed charge, possibly in C polymodal or even sleeping nocstudy of a patient with an acquired painful syndrome iceptors or in cells in the dorsal root ganglion region. of the hand that included burning pain, hyperestheGlycemic flux and altered blood flow may well exacersia, and allodynia-all recognized symptoms of dia- bate some of these changes, perhaps by biochemical betic neuropathy. Percutaneous microneurography, or hypoxic changes at nociceptive endings. Finally, recording with mechanical and thermal stimuli, demephaptic transmission also poses a very attractive exonstrated an exaggerated response which persisted planation. The challenge is to try and prove it, beduring A fiber blockade, suggesting sensitization of cause it most certainly would be rather difficult. C polymodal nociceptors. Units with pathologically Some 8 years ago an editorial asked the question in enhanced receptor responses were described as havLancet,32 “Can we do anything about diabetic neuroping reduced threshold and prolonged after-discharge. athy, or do we just have to document it or commiserIntraneural microstimulation reproduced pain in af- ate with the patient. 7” We have certainly come a long fected and in control skin. An analogous situation way in 8 years. We do have effective treatments for may well exist in painful diabetic neuropathy. neuropathic pain, and there are exciting treatments A study of pain and causalgia was conducted in 17 on the horizon for the prevention and early intervenpatients with previous nerve injuries.” These pa- tion in diabetic neuropathy. tients experienced allodynia, where a light touch on REFERENCES the skin led to pronounced pain. After 30 min of ischemic block, during which the vast majority of pa1. Boulton AJM, Ward JD: Diabetic neuropathies and pain. Clin Endocrinol Metab 15:917-931, 1986. tients experienced relief from allodynia with absent touch, warm sensation was preserved. This result 2. Thomas PK, Scadding JW: Treatment of pain in diasuggested that allodynia was being transmitted by betic neuropathy, in Dyck PJ, Thomas PK, Asbury AK, et al. (eds), Diabetic Neuroputhy. Philadelphia, Saunlarge A beta fibers. This is perhaps analogous to what ders, 1987, pp. 216-222. we were seeing in our preliminary studies of intraneural microstimulation. 3. Asbury AK, Fields HL: Pain due to peripheral nerve PHARMACOLOGICAL NEUROPATHIC

RELIEF OF PAIN

We have known from studies conducted in Denmark that intravenous Iignocaine and oral mexilitene28*29 may give rise to prolonged relief from neuropathic

damage: A hypothesis.

Neurology 341587-1590,

1984.

4.

on diabetes, Washington International Congress Medical Nezw. Philadelphia, September 1887.

5.

Huskisson EC: Measurement 1127-1131, 1974.

Pavy FW: Address

of pain.

Lancet

2:

WHAT CAUSES NEUROPATHIC PAIN?

I Diab Comp 1992; 6:l

6.

Attali JR, Valensi P: Reproducibility of the parameters of nerve function investigation in diabetics. Diabetologia 33(suppl l):A15, 1990.

7.

Llewelyn JG, Thomas PK, Fonseca V, King RHM, Dandona P: Acute painful neuropathy precipitated by strict diabetic control. Acta Neuropathol 72:157-163, 1986.

8.

Britland ST, Young RJ, Sharma AK, Clarke BF: Association of painful and painless diabetic polyneuropathy with different patterns of nerve fiber degeneration and regeneration. Diabetes 39:898-908, 1990.

9.

Archer AC, Watkins PJ, Thomas PK, Sharma AK, Payan: The natural history of acute painful neuropathy in diabetes mellitus. J Neural Neurosurg Psych 46: 491-499, 1983.

10.

Boulton AJM, Drury J, Clarke B, Ward JD: Continuous subcutaneous insulin infusion in the management of painful diabetic neuropathy. Diabetes Care 5:386-390, 1982.

11.

Boulton AJM, Worth RC, Drury J, et al: Genetic and metabolic studies in diabetic neuropathy. Diabetologia 26:15-19, 1984.

12.

Young RJ, Ewing DJ, Clarke BF: Chronic and remitting painful diabetic polyneuropathy. Diabetes Care 11: 34-40, 1988.

13.

Morley GK, Mooradian AD, Levine AS, Morley JE: Mechanism of pain in diabetic peripheral neuropathy: Effect of glucose on pain perception in humans. Am J Med 77:79-82, 1984.

14.

Jaspan JB, Wollman RL, Bernstein L, Rubenstein AH: Hypoglycaemic peripheral neuropathy associated with insulinoma: Implication of glucopenia rather than hyperinsulinism. Medicine 61:33-44, 1982.

the diabetic 1982.

neuropathic

63

foot. Diabetologia 22:9-15,

19.

Archer AC, Roberts VC, Watkins PJ. Blood flow patterns in painful diabetic neuropathy. Diabetologia 27:563-568, 1984.

20.

Edmonds ME, Archer AG, Watkins PJ. Ephedrine: A new treatment for diabetic neuropathic oedema. Lancet 1:548-549, 1983.

21.

McMahon S, Koltzenburg mary afferent neurones 43:269-272, 1990.

22.

Fisher BM, Quin JD, Rumley A, et al: Effect of insulininduced hypoglycaemia on haemostasis, fibrinolysis and haemorrheology in insulin-dependent diabetic patients and control subjects. Clinical Sci 80:525-531, 1991.

23.

Torebjork E, Schady W: Cutaneous mechanoception and nociception in humans and the contribution of microneurography and intraneural microstimulation. Physiol Rev 1991 (in press).

24.

Mackel R: Properties of cutaneous afferents in diabetic neuropathy. Brain 112:1359-1376, 1989.

25.

Masson EA, Schady W, Boulton AJM: Intraneural techniques for the study of peripheral nerves in human diabetes. Diabetic Med 5(suppl 1):33, 1988.

26.

Cline MA, Ochoa J, Torebjork E: Chronic hyperalgesia and skin warming caused by sensitized C nociceptors. Brain 112:626-647, 1989.

27.

Campbell JN, Raja SN, Meyer RA, MacKinnon SE: Myelinated afferents signal the hyperalgesia associated with nerve injury. Pain 32:89-94, 1988.

28.

Kastrup J, Petersen P, Dejgard A, Hilsted J, Angelo JR Treatment of chronic painful diabetic neuropathy with intravenous lidocaine infusion. Br Med 292:173, 1986.

M: The changing role of priin pain (editorial). Pain

15.

Chan AW, MacFarlane IA, Bowsher DR, Wells JCD: Does acute hyperglycaemia influence heat pain threshold? I Neurol Neurosurg Psych 51:688-690, 1988.

29.

Dejglrd A, Petersen P, Kastrup J: Mexilitene for the treatment of chronic painful diabetic neuropathy. Lancet 1:9-11, 1988.

16.

Burchiel KJ, Russell LC, Lee RP, Sima AAF: Spontaneous activity of primary afferent neurons in diabetic BB/wistar rats: A possible mechanism of chronic diabetic neuropathic pain. Diabetes 36:1210-1213, 1985.

30.

Chabal C, Russell LC, Burchiel KJ: The effect of intravenous lidocaine, tocainide, and mexiletine on spontaneously active fibers originating in rat sciatic neuromas. Pain 38:333-338, 1989.

17.

Boulton AJM, Scarpello JHB, Ward JD: Venous oxygenation in the diabetic neuropathic foot: Evidence of arterio-venous shunting. Diabetologia 22:6-8, 1982.

31.

Max MB: Towards physiologically based treatment of patients with neuropathic pain. Pain 42:131-137, 1990.

32.

18.

Edmonds

Editorial: Diabetic neuropathy-where Lancet 1:1366-1367, 1983.

ME, Roberts VC, Watkins PJ: Blood flow in

are we now?