www.elsevier.com/locate/jns

Journal of the Neurological Sciences 220 (2004) 137–139

Experimental models of painful diabetic neuropathy

Nigel A. Calcutt*

Department of Pathology, University of California San Diego, La Jolla, CA 92093, USA

Diabetes mellitus is one of the most rapidly growing Pain or aberrant sensations (paresthesias and dysethesias)

health concerns in the developed world due to the persis-

tence of type 1 (insulin-deficient) diabetes and the dramat-

ic increase in type 2 (insulin- resistant) diabetes that is

secondary to changes of diet and lifestyle in the population

[1]. Hyperglycemia is the primary pathophysiologic mech-

anism common to both types of diabetes mellitus, and

elevated blood and tissue sugar levels have been associated

with the development of the secondary complications of

diabetes that lead to the increased morbidity and mortality

seen in these patients. Of the secondary complications of

diabetes, neuropathy is the most frequently encountered

and can be expected to show some manifestation in over

half of all diabetic patients during their life span. Diabetics

are more prone to mononeuropathies arising from nerve

entrapment or other forms of damage [2]. Polyneuropathies

also occur in diabetes, and the majority of such cases

present as a distal symmetrical polyneuropathy with a

‘‘stocking-and-glove’’ distribution. The primary complaint

of such patients is often one of sensory loss in the

extremities that in the worst cases can lead to tissue

damage and infection, necessitating limb amputation. Con-

duction slowing in the large myelinated sensory and motor

fibers occurs at early stages and is considered a prognostic

indicator for peripheral neuropathy [3]. The pathologic

features of diabetic neuropathy include consequences of

Schwann cell disruption such as nodal widening and

segmental demyelination, axonal degeneration and loss,

basement membrane thickening and microvascular lesions.

In late stages of neuropathy, there is an almost complete

absence of large and small fibers in peripheral nerve which

corresponds to loss of both sensory and motor function. It

has been well documented that the degree and duration of

hyperglycemia are primary risk factors for diabetic neu-

ropathy but at present, the etiologic mechanisms leading to

nerve damage are not well defined so that the only

therapeutic strategy available is to control blood sugar

and hope for the best.

0022-510X/$ - see front matter D 2004 Elsevier B.V. All rights reserved.

doi:10.1016/j.jns.2004.03.015

* Tel.: +1-619-534-5331; fax: +1-619-534-1886.

E-mail address: ncalcutt@ucsd.edu (N.A. Calcutt).

are also reported by some diabetic patients and can precede

or accompany the progressive degenerative neuropathy and

sensory loss. The pain may be persistent or intermittent and

can arise spontaneously or be evoked by light touch

(allodynia). It is often described as having a burning or

tingling quality. These inappropriate sensations are disrup-

tive to normal daily function and impede sleep so that the

general quality of life is dramatically reduced. Like a

number of other neuropathic pain states, painful diabetic

neuropathy does not always respond to commonly used

analgesics such as non-steroidal anti-inflammatory drugs or

opioids at doses below those producing disruptive side

effects. Tricyclic antidepressants and anticonvulsants such

as gabapentin may show efficacy in some patients [4], but

there is clearly a further need for novel therapeutic strategies

based upon an appreciation of mechanisms that initiate and

maintain pain in diabetic patients.

Both genetic and drug-induced models of diabetes are

available in a variety of species and have allowed the study

of mechanisms by which hyperglycemia impairs the ner-

vous system. Diabetic rats and mice quickly develop the

functional disorder of nerve conduction slowing and if they

can be kept alive for long enough, they show mild structural

abnormalities that may be precursors to the overt neuro-

pathic changes seen in diabetic patients. Such findings have

prompted the widespread use of diabetic rodents as models

of early functional and structural disorders associated with

diabetic neuropathy, and numerous studies over the last 40

years have revealed both potential pathogenic mechanisms

and novel therapeutic strategies. However, they have yet to

be confirmed in clinic studies. The use of diabetic rodents to

model painful neuropathy has been a more recent develop-

ment that is restricted by many caveats regarding how pain

perception in animals can be measured and interpreted. The

development of ethically and scientifically acceptable pain

tests in animals has required characterization of surrogates

such as electrical and neurochemical activity in the nervous

system and behavioral responses to sensory stimuli. The

emerging interest in these models has prompted a growing

appreciation of the mechanisms of normal pain perception

N.A. Calcutt / Journal of the Neurological Sciences 220 (2004) 137–139138

(nociception) and abnormal persistent pain following a

nerve injury (neuropathic pain) which in turn has led to

the application of similar tests to animal models of diabetes.

Such studies show that diabetic rodents display physiologic,

neurochemical, and behavioral indices suggestive of altered

pain perception which may make them useful for investi-

gating etiologic mechanisms linking hyperglycemia with

painful neuropathy (reviewed in Ref. [5]). It is hoped that

this knowledge will help produce targeted prophylactic

drugs and also allow the models to be used for screening

potential therapeutic agents designed to alleviate existing

pain in diabetes.

Behavioral studies in diabetic rats often focus on the

response to a painful or non-painful sensory stimulus,

thereby measuring hyperalgesia and allodynia, respectively.

The simplest of such tests measures the time to withdrawal

of a limb such as the tail or a paw from a noxious heat

source, with a faster withdrawal time being interpreted as

hyperalgesia and a slower one as hypoalgesia. The literature

is somewhat confusing over the effects of diabetes on such

thermal latency tests, and this may be in part due to

methodological details such as use of different species,

limbs, and methods of heat application. It is our experience

that insulin-deficient diabetic rats with the plantar surface of

the hind paw exposed to a temperature ramp rising from 30

jC at a rate of 1 jC/s over 20 s, exhibit a transient thermal

hyperalgesia during the first few weeks of diabetes. It

progresses to thermal hypoalgesia within 2–3 months. This

may model the progression from painful to degenerative

painless neuropathy in diabetic patients, although at present

it is not clear that thermal hypoalgesia in diabetic rats

coincides with any loss of epidermal thermal nociceptors,

as appears to be the case in diabetic patients [6]. Our recent

studies show that a range of drugs, including aldose reduc-

tase inhibitors and neurotrophic factors, can prevent both

thermal hyperalgesia and thermal hypoalgesia without af-

fecting the general metabolic status of the rats.

Nociceptive pathways that respond to touch and pres-

sure are also altered by diabetes. Limb removal or vocal-

ization after application of pressure to the paw or tail

shows a shorter latency in diabetic rodents, illustrating

mechanical hyperalgesia. The paw of diabetic rats is also

more sensitive to the light touch by von Frey filaments so

that these animals display a tactile allodynia similar to that

reported by diabetic patients who complain that the touch

of clothes or bed sheets is painful. Tests of mechanical

hyperalgesia and tactile allodynia are now widely used to

assess the efficacy of drugs aimed at alleviating painful

neuropathy.

Another test originally developed to assess the interac-

tions of peripheral and spinal nociceptive processing mech-

anisms involves injection of the irritant formalin into the

paw to produce a peripheral inflammation which evokes a

characteristic flinching of the paw that can be used to

quantify pain-associated behavior. Diabetic rats show a

marked increase in the frequency of paw flinching after

paw formalin injection, illustrating hyperalgesia to a chem-

ical nociceptive stimulus. Further, because later stages of the

flinching response to formalin in normal rats are known to

be driven by spinal amplification of peripheral nociceptive

input that is mediated by local prostanoid production in the

spinal cord [7], the increased flinching of diabetic rats

during this period prompted us to investigate whether

diabetes might alter aspects of spinal nocicpetive processing

as well as peripheral sensory input. We have found that the

hyperalgesic behavioral response to direct application of

substance P to the spinal cord is prolonged in diabetic rats

[8,9]. Spinal release of the nocicpetive modulator prosta-

glandin E is also prolonged following paw formalin injec-

tion in diabetic rats, and this is accompanied by an increase

in spinal cord levels of the protein cyclooxygenase 2 (COX-

2), an enzyme involved in prostanoid synthesis [10]. These

findings, plus studies by others showing increased postsy-

napatic receptors for excitatory neurotransmtters in the cord

of diabetic rodents [11] and enhanced electrophysiologic

activity in spinal neurons [12,13] have prompted us to

consider that an exaggerated amplification of spinal noci-

ceptive processing contributes to allodynia and hyperalgesic

behaviors during diabetes. What induces this enhanced

spinal processing is not yet clear, but it is worth noting that

neurochemical studies suggest that excitatory peripheral

input to the spinal cord is reduced rather than increased in

diabetic rats [9]. It is plausible that the spinal cord may

develop a denervation hypersensitivity analogous to that

seen following muscle denervation.

The utility of the animal models of diabetes for studying

mechanisms of painful neuropathy can only be validated

when an agent targeted at a specific mechanism identified

by animal studies and shown to be effective in animal

behavioral models translates to being a useful drug that

prevents or alleviates pain in diabetic patients. At present,

this validation has not occurred. However, the reverse

progression, from human efficacy to rat model, can be of

use in supporting the continued study of animal models. A

number of agents that show some efficacy against painful

diabetic neuropathy such as lidocaine and gabapentin are

reasonably effective in the models described above. Such

findings support the use of the models and provide opti-

mism that novel agents are being identified based on the

growing appreciation of how hyperglycemia alters pain

processing in rodents. They include selective COX-2 inhib-

itors, sodium channel antagonists, and antagonists of spinal

excitatory neurotransmitters, and may soon make the tran-

sition to useful drugs in the clinical management of painful

diabetic neuropathy.

References

[1] Zimmet P, Alberti KG, Shaw J. Global and societal implications of the

diabetes epidemic. Nature 2001;414:782–7.

[2] Malik RA. Focal and multifocal neuropathies. Curr Diab Rep 2002;

2:489–94.

N.A. Calcutt / Journal of the Neurological Sciences 220 (2004) 137–139 139

[3] Arezzo JC, Zotova E. Electrophysiologic measures of diabetic neurop-

athy: mechanism and meaning. Int Rev Neurobiol 2002;50:229–55.

[4] Morello CM, Leckband SG, Stoner CP, Moorhouse DF, et al. Ran-

domized double-blind study comparing the efficacy of gabapentin

with amitriptyline on diabetic peripheral neuropathy pain. Arch Intern

Med 1999;159:1931–7.

[5] Calcutt NA. Potential mechanisms of neuropathic pain in diabetes. Int

Rev Neurobiol 2002;50:205–28.

[6] Kennedy WR, Wendelschafer-Crabb G, Johnson T. Quantitation of

epidermal nerves in diabetic neuropathy. Neurology 1996;47:1042–8.

[7] Svensson CA, Yaksh TL. The spinal phospholipase– cyclooxyge-

nase–prostanoid cascade in nociceptive processing. Annu Rev Phar-

macol Toxicol 2002;42:553–83.

[8] Calcutt NA, Freshwater JD, O’Brien JS. Protection of sensory func-

tion and antihyperalgesic properties of a prosaposin-derived peptide

in diabetic rats. Anesthesiology 2000;93:1271–8.

[9] Calcutt NA, Stiller C, Gustafsson H, Malmberg AB. Elevated sub-

stance-P-like immunoreactivity levels in spinal dialysates during the

formalin test in normal and diabetic rats. Brain Res 2000;856:20–7.

[10] Freshwater JD, Svensson CI, Malmberg AB, Calcutt NA. Elevated

spinal cyclooxygenase and prostaglandin release during hyperalgesia

in diabetic rats. Diabetes 2002;51:2249–55.

[11] Li N, Young MM, Bailey CJ, Smith ME. NMDA and AMPA gluta-

mate receptor subtypes in the thoracic spinal cord in lean and obese–

diabetic ob/ob mice. Brain Res 1999;849:34–44.

[12] Chen SR, Pan HL. Hypersensitivity of spinothalamic tract neurons

associated with diabetic neuropathic pain in rats. J Neurophysiol

2002;87:2726–33.

[13] Pertovaara A, Wei H, Kalmari J, Ruotsalainen M. Pain behavior and

response properties of spinal dorsal horn neurons following experi-

mental diabetic neuropathy in the rat: modulation by nitecapone, a

COMT inhibitor with antioxidant properties. Exp Neurol 2001;167:

425–34.