Review
Diabetic neuropathy: clinical manifestations and
current treatments
Brian C Callaghan, Hsinlin T Cheng, Catherine L Stables, Andrea L Smith, Eva L Feldman
Diabetic peripheral neuropathy is a prevalent, disabling disorder. The most common manifestation is distal Lancet Neurol 2012; 11: 521-34
symmetrical polyneuropathy (DSP), but many patterns of nerve injury can occur. Currently, the only eff ective University of Michigan,
treatments are glucose control and pain management. While glucose control substantially decreases the development Ann Arbor, MI, USA
of neuropathy in those with type 1 diabetes, the eff ect is probably much smaller in those with type 2 diabetes. Evidence (B C Callaghan MD,
H T Cheng MD, C L Stables PhD,
supports the use of specifi c anticonvulsants and antidepressants for pain management in patients with diabetic A L Smith MS,
peripheral neuropathy. However, the lack of disease-modifying therapies for diabetic DSP makes the identifi cation of Prof E L Feldman MD)
new modifi able risk factors essential. Growing evidence supports an association between components of the metabolic Correspondence to:
syndrome, including prediabetes, and neuropathy. Studies are needed to further explore this association, which has Dr Brian C Callaghan, 109 Zina
implications for the development of new treatments for this common disorder.
Pitcher Place, 4021 BSRB,
Ann Arbor, MI 48104, USA
bcallagh@med.umich.edu
Introduction
prominent than motor involvement. Many patients with
Neuropathy, or damage to the nerves of the peripheral neuropathy experience a sensation of their socks being
nervous system, is a debilitating yet surprisingly com-
bunched up or their shoes not fi tting correctly. They even
mon and complex disorder. The prevalence of neuropathy
have the apparent paradox of numbness and exquisite
is greater than 2% in the general population1,2 and about sensitivity at the same time. Which symptom pre-
15% in people over the age of 40 years.3 By far the most dominates varies substantially from patient to patient.
common cause of neuropathy is diabetes.4 In fact, the
The constellation of symptoms associated with DSP has
prevalence of neuropathy in patients with diabetes is many downstream eff ects that can aff ect patients' quality
about 30%, and up to 50% of patients will eventually of life, both physically and mentally.6 DSP-associated
develop neuropathy during the course of their disease.5
numbness often causes balance problems, which can lead
Diabetes can damage the peripheral nervous system in to falls. Neuropathy is one of three main risk factors for
various ways, but the most common presentation is a falls in patients with diabetes, along with retinopathy and
distal symmetrical polyneuropathy (DSP). Other patterns
vestibular dysfunction.7 In fact, patients with diabetic DSP
of injury include small-fi bre predominant neuropathy, are two to three times more likely to fall than those with
radiculoplexopathy, and autonomic neuropathy, among diabetes and no neuropathy.7 Additionally, patients with
others. Since DSP is the most common neuropathy severe DSP are at risk of ulcerations and lower-extremity
subtype and is the best studied, this will be the main amputations, with 15% developing an ulcer during the
focus of our Review. Currently, the only treatments course of their disease.8 Diabetes is the leading cause of
available to patients with diabetic DSP are improved lower-extremity amputations, roughly 80 000 of which are
glucose control and pain management, which we discuss
undertaken in the USA every year in patients with the
in depth.
disorder.9 Indeed, patients with diabetes are 15 times more
In view of the limitations in current clinical care, the likely than people without diabetes to have this life-
identifi cation of new modifi able risk factors for the changing complication.9 Overall, diabetic DSP can
development of neuropathy is essential. Top candidates severely aff ect quality of life, especially in those with pain.6
include components of the metabolic syndrome, such as
This common, disabling disease also profoundly aff ects
hypertriglyceridaemia, obesity, hyperglycaemia, hyper-
the health-care system. Costs associated with diabetic
tension, and dyslipidaemia. Establishing whether a neuropathy in the USA are estimated to be between
causal relation exists between these components, 4*6 and 13*7 billion dollars, with most of the expense
including prediabetes, and the development of neur-
attributed to those with type 2 diabetes.8 Therefore,
opathy might lead to new disease-modifying therapies.
neuropathy is associated with a quarter of the total costs
of diabetes care in the USA.
Clinical manifestations of diabetic neuropathy
Neuropathic pain is one of the most disabling symp-
DSP accounts for such a large proportion of all peripheral
toms in patients with DSP. This symptom is diffi
cult to
nerve manifestations attributed to diabetes that some treat and therefore causes substantial suff ering and
physicians use the terms diabetic DSP and diabetic societal burden.10 Diabetic neuropathic pain (DNP)
neuropathy interchangeably. Patients with DSP typically occurs in about 10% to 20% of the diabetic population
have one or more of the following symptoms: numbness,
overall, and in about 40% to 60% of patients with
tingling, pain, or weakness. These symptoms begin in the
documented neuropathy.11-13 However, these numbers are
feet and spread proximally in a length-dependent fashion
probably underestimates, because a study showed that
(the so-called stocking-and-glove distribution). The symp-
about 12% of patients with DNP had never mentioned
toms are symmetrical with sensory symptoms more this symptom to their doctors.11 Similar to other types of
www.thelancet.com/neurology Vol 11 June 2012
521

---

Review
A
B
C
D
Figure 1: Patterns of nerve injury in diabetic neuropathy
Clinicians should be aware of all potential patterns of nerve injury because they have implications for the assessment and treatment of patients with diabetes. For example, patients with diabetes can
have radiculopathy without a disc herniation or degenerative changes in the spine. This knowledge could prevent unnecessary spine surgery in cases where imaging results are equivocal. Furthermore,
patients with diabetes can have more than one pattern of nerve injury, and the clinician needs to ask patients about specifi c symptoms such as autonomic involvement, which is often overlooked. The
following patterns of nerve injury are shown in the fi gure: distal symmetrical polyneuropathy (DSP), small-fi bre predominant neuropathy, and treatment-induced neuropathy (A); radiculoplexopathy
and radiculopathy (B); mononeuropathy and mononeuritis multiplex (C); and autonomic neuropathy and treatment-induced neuropathy (D). Small-fi bre predominant neuropathy has the same
pattern as DSP but neurological examination and electrodiagnostic studies give quite diff erent results, which can help the clinician to distinguish between these types of neuropathy. Diabetic
radiculoplexopathy can be responsive to immunotherapy and, in contrast to most nerve injury in patients with diabetes, usually improves with time.16,17 Treatment-induced neuropathy is an
under-recognised disorder.18 Unlike the other peripheral manifestations of diabetes, this disorder is caused by overaggressive control of glucose levels.
neuropathic pain, DNP is characterised by burning, twice the risk of death than patients without autonomic
electric, and stabbing sensations with or without neuropathy.15 Diabetic radiculoplexopathy can aff ect
numbness. Frequently, patients develop allodynia either the lumbosacral (more common) or the cervical
(painful sensations to innocuous stimuli) and hyper-
plexus. Patients present with pain and weight loss
algesia (increased sensitivity to painful stimuli). However,
followed by weakness in the distribution of the aff ected
less than half of patients are treated for pain, despite the plexus. Pathological assess
ment reveals evidence of
availability of many eff ective therapies.11 Fortunately, ischaemic injury and microvasculitis, but whether
several screening methods are available to aid the immuno suppressive drugs are eff ective is unclear.16,17 To
clinician in identifying patients who would benefi t from date, only one randomised controlled trial has assessed
treatment.14
diabetic lumbar radiculo plexopathy and no signifi cant
Other types of peripheral nerve injury in patients with
eff ect was reported for the primary outcome, although
diabetes include small-fi bre predominant neuropathy, secondary outcome measures did show an improvement
autonomic neuropathy, radiculoplexopathy (diabetic with intravenous methylprednisolone.17
amyotrophy), radiculopathy, mononeuritis multiplex,
Diabetes is one of the few causes of non-compressive
mononeuropathy, and treatment-induced neuropathy radiculopathy. Patients with diabetes can also present
(fi gure 1). Small-fi bre predominant neuropathy is an with mononeuritis multiplex without an underlying
increasingly recognised pattern of involvement and rheumatological cause. Furthermore, patients are at
typically is an early manifestation of peripheral nerve increased risk of mononeuropathy, which can be
injury (fi gure 2). In fact, patients often progress from a secondary to compressive or ischaemic mechanisms.
small-fi bre predominant neuropathy to DSP. Small-fi bre The most commonly involved nerves are the oculomotor
predominant neuropathy can be diffi
cult to diagnose and median nerves. Whether the mechanism of these
because the examination (absent refl exes, impaired four peripheral nerve manifestations (radiculoplex-
vibration, weakness) and electro diagnostic testing can opathy, radiculopathy, mononeuritis multiplex, mono-
be normal. Autonomic neuropathy (a type of small-fi bre neuropathy) is the same is unclear. Although poor
neuropathy) is also common in patients with diabetes. glucose control is associated with an increased risk of
Symptoms include gastroparesis, constipation, urinary neuropathy, the treatment of diabetes can also cause
retention, erectile dysfunction, and cardiac arrhythmias.
neuropathy. Treatment-induced neuropathy presents as
Importantly, pa tients with auto nomic neuropathy are at acute pain or autonomic involvement, usually after
522
www.thelancet.com/neurology Vol 11 June 2012

---

Review
patients are given insulin, but it can occur after any quick
A
B
C
establishment of glucose control.18 Pain and autonomic
features can improve signifi cantly with time, and this
pattern of nerve injury underscores the fact that even
rapid improvements in glucose control can lead to
neuropathy. A substantial number of patients with
diabetes also have asymptomatic neuropathy.19 Thus,
50 m
50 m
50 m
while DSP accounts for most neuropathic mani festations
in patients with diabetes, there are other important Figure 2: Small-fi bre predominant neuropathy on skin biopsy
patterns of neuropathy for physicians to consider.
(A) Skin biopsy from a 41-year-old man without neuropathy, assessing intraepidermal nerve-fi bre density (stained
with protein gene product 9*5, 50 m sections). Two nerves are seen crossing the dermal-epidermal junction.
Pathophysiology of type 1 and type 2 diabetic
(B) Skin biopsy from a 50-year-old man with diabetic neuropathy, assessing intraepidermal nerve-fi bre density.
No nerves are seen crossing the dermal-epidermal junction. (C) Sural nerve biopsy from a 44-year-old man with
neuropathy
diabetic neuropathy. Biopsy shows axonal loss of small-diameter and large-diameter nerves.
Hyperglycaemia is a key factor underlying diabetic
neuropathy, but other changes also contribute. In
A
type 2 diabetes, dyslipidaemia is thought to play a major
Mechanisms of cell damage
Type 1
part.20 Changes in insulin signalling are also key; in
Hyperglycaemia
Dyslipidaemia
Type 2
Triglycerides
patients with type 1 diabetes levels of both insulin and
Both
Glucose
LDL
C-peptide are reduced, whereas in type 2 diabetes
HDL
FFA
neuronal insulin sensitivity is thought to be reduced.21,22
Protein
AGE-LDL
Oxidised
Several recent reviews discuss the mechanisms of
AGEs
LDL
diabetic neuropathy in depth.20,23-25 Therefore, we will
briefl y outline the main mechanisms (fi gure 3) and
Insulin
RAGE
LOX1
TLR4
consider how the disease states in patients with type 1 and
type 2 diabetes are diff erent, and why this might aff ect
Insulin+C-peptide
Hexosamine
Glucose
Inflammatory signals
FFAs
treatment effi
cacy.
pathway
Glycolysis
NADPH oxidase
PI3K
Insulin
signalling
olyol
Hyperglycaemia
P
pathway
Akt
Excess intracellular glucose is processed by increased
Electron
Osmotic
Insulin
fl ux through one or more glucose metabolism pathways,
transport
Cholesterol
stress
resistance
overload
and prolonged hyperglycaemia can lead to cellular
damage in several ways. First, excess glycolysis can lead
Oxysterols
ROS ( )
Mitochondrial
Loss of
to overload of the mitochondrial electron transport
complex
neurotrophic
chain and the generation of reactive oxygen species
DNA damage
ER stress
dysfunction
Apoptosis
signals
(ROS).25 Second, increased fl ux through the polyol
pathway can increase cellular osmolarity, reduce
B Cell damagenerve dysfunction
NADPH levels, and lead to oxidative stress.26 Finally,
increased fl ux through the hexosamine pathway is
associated with infl ammatory injury.23
Another consequence of hyperglycaemia is the gener-
Macrophage
ation of advanced glycation end products (AGEs),27 via
activation
attachment of reactive carbohydrate groups to proteins,
Vascular endothelial cells
lipids, or nucleic acids. These groups tend to impair the
Neurons
Glial cells
biological function of proteins, thus aff ecting cellular
function.28 Extracellular AGEs also bind to the receptor Figure 3: Mechanisms of diabetic neuropathy
Factors linked to type 1 diabetes (orange), type 2 diabetes (blue), and both (green) cause DNA damage,
for AGE (RAGE), initiating infl ammatory signalling endoplasmic reticulum stress, mitochondrial complex dysfunction, apoptosis, and loss of neurotrophic signalling
cascades, activating NADPH oxidases, and generating (A). This cell damage can occur in neurons, glial cells, and vascular endothelial cells, as well as triggering
oxidative stress.29 Long-term infl ammatory responses are
macrophage activation, all of which can lead to nerve dysfunction and neuropathy (B). The relative importance of
also triggered, including the upregulation of RAGE and the pathways in this network will vary with cell type, disease profi le, and time. AGE=advanced glycation end
products. LDL=low-density lipoprotein. HDL=high-density lipoprotein. FFA=free fatty acids. ROS=reactive oxygen
activation of NFB.30
species (red star). ER=endoplasmic reticulum. PI3K=phosphatidylinositol-3-kinase. LOX1=oxidised LDL receptor 1.
RAGE=receptor for advanced glycation end products. TLR4=toll-like receptor 4.
Dyslipidaemia
The incidence of dyslipidaemia is high in patients with have systemic eff ects such as promoting infl ammatory
type 2 diabetes.31 Dyslipidaemia is linked to diabetic cytokine release from adipocytes and macrophages.34
neuropathy,32 and several underlying mechanisms have Plasma lipoproteins, especially LDLs, can be modifi ed by
been identifi ed. Free fatty acids have been shown to oxidation or glycation, and these modifi ed LDLs can bind
directly cause injury to Schwann cells in vitro,33 but also to extracellular receptors (including the oxidised LDL
www.thelancet.com/neurology Vol 11 June 2012
523

---

Review
receptor LOX1,35 toll-like receptor 4,36 and RAGE29),
unsurprisingly, neuropathy is diffi
cult to halt or reverse.
triggering signalling cascades that activate NADPH Although hyperglycaemia contributes to the vicious
oxidase and lead to oxidative stress.35 Add
itionally, cycles of oxidative stress, infl ammation, and cellular
cholesterol can be oxidised to oxysterols, which have damage in patients with type 2 diabetes, reducing
been shown to cause apoptosis in neurons.23,37
hyperglycaemia alone might not be enough to stop the
cycle from continuing.
Impaired insulin signalling
While insulin is not involved in glucose uptake into Metabolic syndrome
neurons, it has been shown to have neurotrophic eff ects, Our knowledge of how components of the metabolic
promoting neuronal growth and survival.38,39 Reduction syndrome damage nerves is rapidly improving. In
of this neurotrophic signalling due to insulin defi ciency addition to dyslipidaemia and insulin resistance,
(type 1 diabetes) or insulin resistance (type 2 diabetes) is another central component of the metabolic syndrome,
thought to contribute to the pathogenesis of diabetic visceral adiposity, might be especially detrimental
neuropathy.21 In neurons, insulin resistance occurs by because it causes increased concentrations of free fatty
inhibition of the PI3K/Akt signalling pathway, similar to acids in the plasma and also induces a proinfl ammatory
insulin resistance in muscle and adipose tissue.24
state by the secretion of adipokines (also contributing to
Disruption of this pathway can also lead to mitochondrial
the development of insulin resistance).44 Hypertension,
dysfunction and oxidative stress, further promoting another aspect of the metabolic syndrome, might also
neuropathy.21
be connected to neuropathy, although the link is less
In patients with type 1 diabetes, reduction in C-peptide
well-established. The renin-angiotensin system, which
can lead to nerve dysfunction in several ways, including
controls blood pressure, is upregulated in obesity, and
reduction of sodium-potassium ATPase activity, endo-
might contribute to the development of type 2 diabetes
thelial nitric oxide synthase (eNOS) activity, and endo-
(partly through promotion of insulin resistance and
neurial blood fl ow.22 Treatment with C-peptide can slow proinfl
ammatory cytokine secretion from adipose
the progression of neuropathy in patients with type 1 tissue).45 Angiotensin-converting enzyme (ACE)
diabetes.40
inhibitors have been shown to improve diabetic
The mechanisms outlined above lead to many cellular neuropathy in animal studies,46,47 but the mechanism of
disturbances, including mitochondrial dysfunction, action is unclear. Microvascular dysfunction in the
endoplasmic reticulum stress, DNA damage, and nerve and decreased endoneurial perfusion are also
apoptosis. Another layer of complexity is added when thought to contribute to neuropathy.48 Although these
you consider that these processes of cell stress or damage
elements might be regulated by metabolic factors,
occur in several diff erent cell types within the nerves, upregulation of the renin-angiotensin system might
including neurons (in axons and at nerve terminals), also contribute.48
glial cells, and endothelial cells of the microvasculature.
These mechanisms are probably linked on several
Furthermore, many of these changes will trigger the levels. Indeed, in terms of the mechanisms linking the
activation and recruitment of macrophages,41 feeding metabolic syndrome and type 2 diabetes to neuropathy,
back into infl ammatory mechanisms of cell stress and these pathways might be more accurately described as a
death. Ultimately, these diff erent forms of cellular stress network in which hyperglycaemia, insulin resistance,
cause dysfunction or death of the nerve, which manifests
dyslipidaemia, systemic infl ammation, and activation
as clinical neuropathy.
of the renin-angiotensin system all feed into a self-
Tight glucose control can reduce neuropathy in patients
perpetuating cycle of oxidative stress, infl ammatory
with type 1 diabetes, as we discuss below, but is not signals, and disruption of normal cellular function.
as effi
cacious in patients with type 2 diabetes.42,43 This Thus, even in the absence of overt diabetes, other
disparity is probably related to diff erences in the under-
aspects of the metabolic syndrome could be suffi
cient
lying mechanisms: hyperglycaemia and a reduction in to cause neuropathy. One of the main challenges for
insulin signalling in patients with type 1 diabetes, researchers is to determine which aspects of this
compared with a combination of hyperglycaemia, network of mechanisms can be blocked at which times
dyslipidaemia, and insulin resistance in patients with to eff ectively limit or prevent progression of the
type 2 diabetes. Diff erences in the duration of pro-
neuropathy.
neuropathic changes before the onset or diagnosis of
diabetes could also contribute to diff erences in the Prediabetes and neuropathy
progression of neuropathy between the two diseases. Whereas the link between diabetes and neuropathy is
Type 2 diabetes does not typically develop rapidly; it well-established, there remains scientifi c uncertainty
occurs after many years of obesity and other aspects of regarding the eff ects of prediabetes (impaired fasting
the metabolic syndrome. Tight glucose control will not glucose or impaired glucose tolerance) on neuropathy.
necessarily reduce dyslipidaemia, systemic infl ammation, Increased prevalence of impaired glucose tolerance has
and insulin resistance, and after years of these insults, been reported in patients with idiopathic neuropathy
524
www.thelancet.com/neurology Vol 11 June 2012

---

Review
compared with controls,49,50 and increased prevalence of
Trial size
Length of
Clinical
Other
Enhanced
neuropathy has been seen in patients with impaired
study (years)
outcome
outcomes
glycaemic
glucose tolerance compared with controls.51 Additionally,
control
Smith and colleagues52 followed up a cohort of patients
superior?
with impaired glucose tolerance and neuropathy who
Type 1 diabetes
underwent an extensive diet and exercise regimen. They
Holman et al56
74
2*0
No
QST
Yes
reported that these patients had an increase in nerve
Lauritzen et al57
30
2*0
No
QST
No
fi
bre density over time, in contrast to historical
Dahl-Jorgensen et al58
45
2*0
No
NCS
Yes
controls.52 This fi nding suggests that treatment of
Jakobsen et al59
24
2*0
No
QST
Yes
impaired glucose tolerance could improve neuropathy
DCCT42
1441
5*0
Yes
NCS
Yes
outcomes, although the study did not have a control
Reichard et al60
102
7*5
No
NCS, QST
Yes
group for comparison. By contrast, Hughes and
Linn et al61
49
5*0
Yes
None
Yes
colleagues53 did not fi nd a signifi cant association
Type 2 diabetes
between impaired glucose tolerance and neuropathy in
Kawamori et al62
50
4*0
No
NCS
Yes
a small case-control study.53 Similarly, Dyck and
UKPDS43
3867
10*0
No
QST
Yes
colleagues54 reported no diff erence in the prevalence of
Tovi et al63
38
1*0
Yes
None
No
neuropathy in patients with impaired glucose tolerance
Azad et al64
153
2*0
Yes
None
No
compared with matched controls in a population-based
Shichiri et al65
110
8*0
No
NCS, QST
Yes
study in Olmsted County, MN, USA.
Gaede et al66
160
8*0
No
QST
No
Since the data linking prediabetes with neuropathy
Duckworth et al67
1791
5*6
Yes
None
No
are confl icting, a comprehensive study is needed to
Ismail-Beigi et al68
10 251
3*7
Yes
None
No
investigate prediabetes to establish whether one of the
metabolic drivers underlies the onset and progression
QST=quantitative sensory testing. NCS=nerve conduction studies. DCCT=Diabetes Control and Complications Trial
of neuropathy. The answer has direct implications for
research group. UKPDS=UK Prospective Diabetes Study.
potential therapies for many patients with neuropathy.
Table 1: Clinical trials investigating eff ects of enhanced glucose control on neuropathy
Currently, one-third of adult Americans meet criteria for
prediabetes.55 Since less than 5% of these people have
received a formal diagnosis from their health-care produced less defi nitive results.43,62-68,70 Only four of these
providers and only a small percentage are being treated studies investigated the eff ects of glucose control on
for this disorder, establishing a causal relation between clinical impairment secondary to neuropathy. In 2010,
prediabetes and neuropathy would change the clinical the Action to Control Cardiovascular Risk in Diabetes
management of a substantial number of patients.55
(ACCORD) study group compared the eff ectiveness of a
lower haemoglobin A1c goal (less than 6*0 compared
Glucose control in type 1 and type 2 diabetes
with 7*0-7*9) on the Michigan Neuropathy Screening
Research done over the past 20 years has added to our instrument.68 In more than 5500 patients followed up for
knowledge of glucose control as a modifi able risk factor a median of 3*7 years, researchers reported a 7%
for the development of neuropathy in patients with reduction in the incidence of neuropathy, which was not
diabetes (table 1). 17 randomised controlled trials have statistically signifi
cant. In 2009, Duckworth and
studied the eff ects of enhanced glucose control on colleagues67 followed up 1791 military veterans for a
neuropathy over at least a 12-month period.56-71 Seven of median of 5*6 years and identifi ed a non-signifi cant 5%
these studies focused on patients with type 1 diabetes, reduction in neuropathy. Studies by Azad and colleagues64
but only two reported outcomes related to clinical and Tovi and colleagues63 assessed fewer patients and
impairment.42,56-61 In 1993, the Diabetes Control and their results produced relative risks with large CIs (no
Complications Trial (DCCT) study group42 followed up signifi cant diff erences). However, three of four studies
over 1400 patients for 5 years and reported a 60% that investigated nerve conduction or quantitative
reduction in the incidence of neuropathy in those sensory testing showed signifi cant results in favour of
receiving more frequent insulin dosing. Similarly, Linn glucose control. One such study, done by the United
and colleagues61 followed up 49 patients for 5 years and Kingdom Prospective Diabetes Study (UKPDS) group in
reported about a 70% reduction in neuropathy in those 1998, is the second largest and longest randomised trial
with enhanced glucose control. Both of these studies in patients with type 2 diabetes.43 The main neuropathy
revealed a large, signifi cant reduction in neuropathy with
outcome measure in this study was vibration threshold,
tighter glucose control. Furthermore, only one of the measured with a biothesiometer. This group followed up
seven studies did not show a signifi cant benefi t of tighter
3867 patients for 15 years and reported relative risks of
glucose control.
0*95 (95% CI 0*76-1*18) at 3 years, 0*88 (0*72-1*08) at
By contrast with the robust results obtained from 6 years, 0*84 (0*68-1*04) at 9 years, 0*92 (0*70-1*20) at
patients with type 1 diabetes, the eight randomised 12 years, and 0*60 (95% CI 0*39-0*94) at 15 years (the
controlled trials of patients with type 2 diabetes have only signifi cant result) in patients receiving enhanced
www.thelancet.com/neurology Vol 11 June 2012
525

---

Review
glucose control. Overall, these eight studies support only
that exist in the disease mechanisms and complications
a modest reduction in neuropathy in patients with type 2
of the two types of disease.
diabetes receiving enhanced glucose control, which is in
stark contrast to the substantial eff ect in those with type 1
Pain management in DNP
diabetes. Possible explanations for this diff erence include While glucose control is the only disease-modifying
the diff erent outcome measures used, the diff erent therapy for diabetic neuropathy, pain management is
treatment regimens, the higher incidence of neuropathy the other mainstay of treatment that can substantially
in controls with type 2 diabetes, and the diff erence in improve the quality of life of patients with neuropathy.
baseline glucose control in these clinical trials. However,
Over the past two decades, tremendous eff ort has been
despite the similarities between type 1 and type 2 made through randomised, placebo-controlled trials to
diabetes, these trials highlight the substantial diff erences improve the treatment of DNP. Data from these trials
Dose
Study design
Result
Number needed to treat Class of
Class of
evidence
evidence
(EFNS)
(AAN)
Pregabalin
Trial size
146
300 mg
Parallel, 8 weeks
Pregabalin>placebo75
3*9
I
I
338
75 mg, 300 mg, 600 mg
Parallel, 5 weeks
Pregabalin (300 mg, 600 mg)>placebo76
3*6 (300 mg group);
I
I
3*3 (600 mg group)
246
150 mg, 600 mg
Parallel, 6 weeks
Pregabalin (600 mg)>placebo77
4*2 (600 mg group)
I
I
338
Fixed or fl exible: 150-600 mg
Parallel, 12 weeks
Fixed and fl exible>placebo78
3*6
**
II
167
600 mg
Parallel, 13 weeks
Pregabalin (600 mg)>placebo79
**
I
I
396
150 mg, 300 mg, 600 mg
Parallel, 12 weeks
Pregabalin (600 mg)>placebo80
6*3 (600 mg group)
**
**
Gabapentin
Trial size
165
<3600 mg
Parallel, 8 weeks
Gabapentin>placebo81
4
I
I
40
900 mg
Crossover, 2x6 weeks
Gabapentin=placebo82
**
**
II
Lamotrigine
Trial size
59
<400 mg
Parallel, 8 weeks
Lamotrigine>placebo83
4
I
II
360
200 mg, 300 mg, 400 mg
Parallel, 19 weeks
Lamotrigine=placebo84
**
I
I
53
200 mg vs 75 mg amitriptyline
Parallel, 6 weeks
Lamotrigine=amitriptyline85
**
II
**
Valproate
Trial size
52
600-1200 mg
Parallel, 4 weeks
Valproate>placebo86
**
II
II
39
500 mg
Parallel, 16 weeks
Valproate>placebo87
**
II
II
31
1500 mg
Parallel, 4 weeks
Valproate=placebo88
**
I
**
Amitriptyline
Trial size
29
150 mg
Crossover, 2x6 weeks
Amitriptyline>placebo89
2*1
**
II
24
25-75 mg
Crossover, 2x6 weeks
Amitriptyline>placebo90
**
**
II
19
75 mg
3-phase, crossover
Amitriptyline>maprotiline>placebo91
**
**
I
amitriptyline and maprotiline
Desipramine
Trial size
20
Mean 201 mg
Crossover, 2x6 weeks
Desipramine>placebo92
2*2
**
II
Venlafaxine
Trial size
244
150-225 mg
Parallel, 6 weeks
Venlafaxine>placebo93
4*5
I
I
60
75-150 mg
Parallel, 8 weeks
Venlafaxine>placebo94
**
II
**
Venlafaxine + gabapentin
Trial size
11 and 42
**
Parallel, 2x8 weeks
Venlafaxine + gabapentin>placebo in
**
**
II and III
patients unresponsive to gabapentin95
(Continues on next page)
526
www.thelancet.com/neurology Vol 11 June 2012

---

Review
Dose
Study design
Result
Number needed to treat Class of
Class of
evidence
evidence
(EFNS)
(AAN)
(Continued from previous page)
Duloxetine
Trial size
457
20 mg, 60 mg, 120 mg
Parallel, 12 weeks
Duloxetine (60 mg, 120 mg)>placebo96
4*3 (60 mg group);
I
II
3*8 (120 mg group)
348
60, 120 mg
Parallel, 12 weeks
Duloxetine (60 mg, 120 mg)>placebo97
11 (60 mg group);
I
I
5 (120 mg group)
334
60 mg, 120 mg
Parallel, 12 weeks
Duloxetine (60 mg, 120 mg)>placebo98
6*3 (60 mg group);
I
II
3*8 (120 mg group)
Oxycodone
Trial size
159
10-100 mg
Parallel, 6 weeks
Oxycodone>placebo99
NA
I
II
338
10-80 mg + 100-3600 mg
Parallel, 12 weeks
Oxycodone + gabapentin>placebo +
**
I
I
gabapentin
gabapentin100
Morphine
Trial size
57
120 mg morphine, 60 mg morphine +
Crossover, 4x4 weeks
Morphine + gabapentin>morphine>
**
I
II
2400 mg gabapentin, 3600 mg
gabapentin>placebo101
gabapentin
Tramadol
Trial size
127
100-400 mg (mean 210 mg)
Parallel, 6 weeks
Tramadol>placebo102
3*1
**
II
45
200-400 mg
Crossover, 2x6 weeks
Tramadol>placebo103
4*3
**
II
311
37*5 mg + 325 mg paracetamol
Parallel, 8 weeks
Tramadol/paracetamol>placebo104
**
I
**
Dextromethorphan
Trial size
19
400 mg
Crossover, 2x9 weeks
Dextromethorphan>placebo105
3*2
I
I
14
Mean 381 mg
Crossover, 2x6 weeks
Dextromethorphan>placebo106
4
II
II
Capsaicin
Trial size
252
0*075% four times a day
Parallel, 8 weeks
Capsaicin>placebo107
NA
**
**
22
0*075% four times a day
Parallel, 8 weeks
Capsaicin>placebo108
**
**
I
Isosorbide
Trial size
22
30 mg
Crossover, 2x4 weeks
Isosorbide>placebo109
**
**
I
Glyceryl
Trial size
48
**
Crossover, 2x4 weeks
Glyceryl>placebo110
**
II
**
ABT-594
Trial size
266
150 mg, 225 mg, 300 mg twice a day
Parallel, 7 weeks
ABT-594>placebo111
**
I
**
Botulinum toxin
Trial size
18
Fifty units of subtype A in 1*2 mL 0*9%
Crossover, 12x12 weeks
Botulinum toxin>placebo112
**
II
**
saline given intradermally into each
foot, each injection 4 U subtype A
Levodopa
Trial size
25
100 mg + 25 mg benserazide three
Parallel, 28 days
Levodopa + benserazide>placebo113
**
II
**
times a day
EFNS=European Federation of Neurological Societies task force. AAN=American Academy of Neurology. NA=not available.
Table 2: Published class I and class II evidence for pharmacological treatment of diabetic neuropathic pain
www.thelancet.com/neurology Vol 11 June 2012
527

---

Review
on a meta-analysis of seven trials with class I evidence for
EFNS72
AAN74
a systematic review. The recommended dose is
Pregabalin (300-600 mg a day)
A
A
900-3600 mg a day.
Gabapentin
A
B
Lamotrigine is classifi
ed as ineff
ective or with
Lamotrigine
A/B*
B
discrepant results with level A to B evidence by the EFNS
Oxcarbazepine
A/B*
B
because of one negative class I study and one class II
Lacosamide
A/B*
B
study that showed comparable effi
cacy to amitriptyline.84
Sodium valproate
A/B*
B
Lamotrigine is also not recommended by the AAN on the
Tricyclic antidepressants
A
B (amitriptyline)
basis of two class I studies that failed to show benefi t
Serotonin-norepinephrine reuptake inhibitor
A
B (venlafaxine, duloxetine)
compared with placebo.83 Similarly, both guidelines state
Opioids
A (oxycodone)
B (morphine, oxycodone)
that oxcarba zepine and lacosamide are not eff ective with
Tramadol
A
B
level A/B (EFNS) or level B (AAN) evidence.
Dextromethorphan
B
B
Sodium valproate is classifi ed as ineff ective or with
Topical capsaicin
A/B*
B
discrepant results with level A/B evidence by EFNS. By
Isosorbide spray
A
B
contrast, this drug is classifi ed as eff ective with level B
ABT-594
A
**
evidence by the AAN for doses of 500-1200 mg a day.
Botulinum toxin
B
**
The EFNS justifi ed its decision to classify this drug as
Levodopa
B
**
ineff ective or with discrepant results because both
Lidocaine patch
**
C
positive studies were published from the same group86,87
and one negative study has been published.88 The
EFNS=European Federation of Neurological Societies task force. AAN=American Academy of Neurology. A=established
negative study was not discussed in the AAN guideline.
as eff ective. B=probably eff ective. C=possibly eff ective. *Drug classed as ineff ective or with discrepant results. Drug
not recommended.
Of note, the two positive studies did not report a
signifi cant placebo eff ect or signifi cant side-eff ects that
Table 3: Comparison of EFNS and AAN guidelines for pharmacological treatment of diabetic
are usually attributed to this drug. The two guidelines
neuropathic pain
disagree on whether the current evidence supports or
refutes the eff ectiveness of sodium valproate for the
support the use of specifi c pharmacological treatments treatment of DNP.
for DNP. Taking into consideration the effi
cacy of these
interventions, several guidelines have been generated. Antidepressants
The 2006 and 2010 guidelines from the European Tricyclic antidepressants (TCAs) are classifi ed as eff ective
Federation of Neurological Societies (EFNS) task force72,73
with level A evidence by the EFNS on the basis of two
and the 2011 guidelines from the American Academy of class I meta-analyses;114,115 however, the EFNS guidelines
Neurology (AAN), the American Association of Neuro-
do not recommend a specifi c drug within the TCA class.
muscular and Electrodiagnostic Medicine, and the By contrast, the AAN states that amitriptyline (25-100 mg
American Academy of Physical Medicine and Rehabili-
a day) is supported by level B evidence on the basis of one
tation74 are the most thorough and recent guidelines on class I and two class II studies.89,91,116 The AAN guidelines
this topic. According to these guidelines, several classes state that there is insuffi
cient evidence with regard to
of drugs are considered to be eff ective for the treatment other TCAs because only class III evidence is available
of DNP (table 2). Overall, these two guidelines are in for these drugs (level U evidence).
close agreement on most of the drugs assessed (table 3).
The use of serotonin-norepinephrine reuptake inhibi-
The consensus from these guidelines provides tors such as venlafaxine and duloxetine is supported by
information for the best evidence-based practice for the both the EFNS (level A evidence) and the AAN (level B
treatment of DNP.
evidence) guidelines. The reason for the discrepancy in
the level of evidence is that the EFNS describes three
Anticonvulsants
class I studies96-98 for duloxetine whereas the AAN
Pregabalin is classifi ed as eff ective with level A evidence classifi es only one of these studies as class I.97 Similarly,
by both the EFNS and the AAN guidelines. This the AAN classifi es only one of the two studies of
recommendation is based on four class I studies75-77,79 that
venlafaxine as class I.97 The recommended doses are
all showed superiority of pregabalin compared with 75-225 mg a day for venlafaxine and 60-120 mg a day for
placebo. The eff ect size was small in the highest-quality duloxetine.
studies. The recommended dose for pregabalin is
300-600 mg a day.
Opioids
Gabapentin is also classifi ed as eff ective with level A Controlled-release oxycodone is recommended by the
evidence by the EFNS, although the AAN classed it as a EFNS as eff ective with level A evidence on the basis of
level B drug because only one class I study showed two class I studies.99,100 Similarly, tramadol (200-400 mg
benefi t and one negative class II study had been a day and tramadol 37*5 mg with 325 mg paracetamol)
published.82 By contrast, the EFNS guidelines are based was listed by the EFNS as an eff ective treatment with
528
www.thelancet.com/neurology Vol 11 June 2012

---

Review
level A evidence on the basis of three class I studies,
including two systematic reviews.104,117,118 By contrast, the
Confirmed diabetic neuropathic pain
AAN recommends the following opioids as eff ective
with level B evidence: controlled-release oxycodone
(mean 37 mg a day and up to 120 mg a day) on the basis
Tricyclic antidepressants
Serotonin-norepinephrine
Voltage-gated calcium
First line
reuptake inhibitors
channel ligand
of one class II trial,99 morphine (up to 120 mg a day) on
2
the basis of one class II trial,101 and tramadol 210 mg a
No effect
Partial effect
day on the basis of two class II studies.102,103
Try another first-line drug
Try combination of first-line drugs
Other drugs
Dextromethorphan (400 mg a day) is classifi ed as
If all three classes and
probably eff ective with level B evidence in both the EFNS
combination therapy fail
and the AAN guidelines on the basis of one class I and
one class II study.105,106
Second line
Opioids
Tramadol
Topical capsaicin treatment (0*075% four times a day) Figure 4: Algorithm for the treatment of diabetic painful neuropathy
is supported with level B evidence in the AAN guidelines
First-line and second-line treatments for diabetic painful neuropathy.
on the basis of one class I108 and one class II study.119
However, the EFNS guidelines classifi ed capsaicin with
level A/B evidence as ineff ective or with discrepant drugs head to head or assess the eff ect on quality of life.
results on the basis of a systematic review of fi ve class I-
Studies are needed to further clarify the role of these
II studies.
drugs in the treatment of DNP.
Isosorbide dinitrate spray is backed by level B evidence
in the AAN guideline on the basis of one class I trial.109
Treatment algorithm
Similarly, the EFNS cited the same study but also Several reviews have recommended treatment algo-
reported a study that used glyceryl trinitrate spray rithms for DNP on the basis of drug effi
cacy and safety
(class I)110 and stated that treatment with nitrate (fi gure 4). We reviewed algorithms from Jensen and
derivatives is supported with level A evidence on the colleagues122 and the EFNS73 for the treatment of DNP,
basis of these two studies.
and from Dworkin and colleagues123 for the treatment of
The nicotine derivative ABT-594 is listed by the EFNS neuropathic pain. Importantly, no evidence supports one
as eff ective with level A evidence on the basis of one treatment algorithm over another.
class I study.111 This treatment is not discussed in the
All three algorithms recommend gabapentin,
AAN guidelines.
pregabalin, TCAs, venlafaxine, and duloxetine as fi rst-line
Botulinum toxin and levodopa are classifi ed as probably
treatments. The choice of agent is largely dependent on
eff ective with level B evidence by the EFNS on the basis the comorbidities of the patient and side-eff ect profi les of
of one class II study for each drug;112,113 however, neither is
the drugs. This is especially important for the treatment of
discussed in the AAN guidelines.
DNP because none of the drugs was designed specifi cally
The lidocaine patch is classifi ed as probably eff ective for neuropathic pain and, therefore, each has other
with level C evidence by the AAN on the basis of two class
indications such as the treatment of seizures or
III studies,120,121 but it was not discussed in the EFNS depression. Dworkin and colleagues124 also recommend
guidelines.
topical lidocaine for patients with localised neuropathic
The data for sodium valproate and capsaicin cream are pain and in those who might be adversely aff ected by CNS
confl icting with one guideline providing evidence for side-eff ects. All three algorithms recommend titrating a
their use and one citing evidence against their use. The fi rst-line drug to a maximum tolerated dose before
more subtle diff erences in levels of evidence are probably
switching to a diff erent fi rst-line drug or combination
due to the fact that the AAN guidelines required a therapy. Only once all these options have failed is a second-
completion rate of greater than 80%, which was not line agent recom mended. All of these drugs are supported
required by the EFNS. Therefore, many trials were by level A evidence in the EFNS guidelines and by level A
downgraded from class I to class II because of this or B evidence in the AAN guidelines.
stringent criterion and, as a result, only pregabalin was
All three algorithms support opioid analgesics and
shown to have level A evidence in the AAN guidelines. tramadol as second-line treatments. Although these
With increasing knowledge of the proper conduct and drugs are also backed by level A evidence in the EFNS
reporting of clinical trials over time, there is probably a guidelines and level B evidence in the AAN guidelines,
bias in favour of newer drugs. Furthermore, the levels of concern exists over their long-term use because of their
evidence do not take into account the number needed to addiction potential, side-eff ect profi
le, and waning
treat or the number needed to harm. Rather, the levels of eff ectiveness over time.
evidence are based on the number of high-quality studies
None of the recommendations incorporates cost into
that show benefi t. Unfortunately, few studies compare the decision, but this factor is also an important
www.thelancet.com/neurology Vol 11 June 2012
529

---

Review
consideration not only for the patients, but also for the confounding factors to this association. In 2005, Tesfaye
health system. TCAs are the most aff ordable of the fi rst-
and col leagues131 followed up 1172 patients with type 1
line agents. Gabapentin and venlafaxine are cheaper than
diabetes for a median of 7*3 years and reported that
pregabalin and duloxetine, respectively.
BMI and smoking were independent risk factors for
neuropathy. They also identifi ed associations with
Metabolic syndrome: implications for future
hypertension and LDL in minimally adjusted models.
treatments
In the same year, De Block and colleagues132 did a cross-
Currently, glucose control and pain management are sectional study in 592 patients with type 1 diabetes.
the backbones of treatment for diabetic neuropathy. Their study revealed no association between BMI, lipid
However, glucose control is not the only answer because abnormalities, triglycerides, or hypertension and
patients with adequate glucose control continue to develop
neuropathy. In 2009, Van Acker and colleagues6
neuropathy and their neuropathy worsens over time. investigated 1111 patients with diabetes with a cross-
Furthermore, pain management is not a disease-
sectional study design. They showed that obesity, HDL,
modifying therapy. Therefore, the discovery of modifi able and triglyceride levels were all independently associated
risk factors for neuropathy is essential, with components with neuropathy. Moreover, Wiggin and colleagues32
of the metabolic syndrome representing one possibility. reported that patients with diabetes and progressive
Over the past 10 years, interest in the possible role of these
neuropathy had higher triglyceride levels than did
components in the development of neuropathy has patients who did not have progressive neuropathy.
increased. In 2001, Isomaa and colleagues125 compared Although most studies have shown an association
85 patients with metabolic syndrome and type 2 diabetes between some components of the metabolic syndrome
with patients without metabolic syndrome, controlling for
and neuropathy, all of these studies have been done in
age, sex, glycaemic control, and duration of diabetes. They
patients with outright diabetes, most have used a cross-
showed that patients with metabolic syndrome had a sectional design, and the defi nition of neuropathy has
higher prevalence of neuropathy, but that in multiple diff ered between studies. In view of the confl icting
logistic regression models, the metabolic syndrome and results, further studies are needed to adequately defi ne
its components were not associated with neuropathy. the role of the metabolic syndrome in the development
Costa and colleagues and the Metascreen investigators and progression of neuropathy. Identifying which
used cross-sectional designs to independently show that metabolic syndrome components are associated with
the metabolic syndrome was associated with neuropathy neuropathy could lead directly to future clinical trials,
in patients with diabetes.126,127 In 2007, Cull and colleagues128
because pharmacological treatments are currently
used the UKPDS cohort of 5102 patients with available for all of the components. For example, if
type 2 diabetes, who were followed up for a median of hypertension is associated with neuropathy then a trial
10*3 years to assess the association of metabolic syndrome
to determine whether treatment of hypertension
with neuropathy by using four diff erent defi nitions of the improves neuropathy outcomes would be essential.
syndrome. They showed that the metabolic syndrome was
associated with a combined macrovascular endpoint, but Conclusions and future directions
not with a combined microvascular endpoint, which Diabetes can injure peripheral nerves in various
includes neuropathy, nephropathy, and retinopathy. In distributions. The most common pattern is DSP, which is
2008, Smith and colleagues129 compared patients with characterised by numbness, tingling, pain, or weakness
idiopathic neuropathy and normoglycaemia with those that aff ects the nerves in a stocking-and-glove pattern,
who had impaired glucose tolerance, and reported that beginning in the distal extremities. DSP leads to
each group had the same prevalence of the separate substantial pain, morbidity, and impaired quality of life.
components of the metabolic syndrome. This result Societal, personal, and health-care costs associated with
suggests that components of the metabolic syndrome diabetic neuropathy are high. Unfortunately, few
other than impaired glucose tolerance might have a role interventions are available for the remediation of non-
in the development of neuropathy. However, these studies
painful symptoms, and glucose control is the only proven
have almost all been done with patients who have diabetes,
disease-modifying intervention for these patients.
have used cross-sectional designs, and have used Although pain is a common feature, it is often under-
inconsistent defi nitions of neuropathy.
reported and undertreated. However, many eff ective
Many researchers have assessed the eff ect of the therapies exist for DNP, including medicines designed to
individual components of the metabolic syndrome on treat seizures and depression. Evidence-based con sensus
neuropathy. In 1994, Straub and colleagues130 did a guidelines have been created to guide the use of these
cross-sectional study of 91 patients with type 2 diabetes,
pain interventions.
and stratifi ed them on the basis of body-mass index
Many areas of research into diabetic neuropathy are
(BMI).126 Patients with a BMI greater than 26*5 had a yet to be fully explored, but there are promising lines of
worse clinical neuropathy score than those with a lower
investigation that could lead to improved prevention
BMI. However, this study did not account for any and treatment of the disorder. The magnitude of the
530
www.thelancet.com/neurology Vol 11 June 2012

---

Document Outline

• Diabetic neuropathy: clinical manifestations and current treatments
  • Introduction
  • Clinical manifestations of diabetic neuropathy
  • Pathophysiology of type 1 and type 2 diabetic neuropathy
    • Hyperglycaemia
    • Dyslipidaemia
    • Impaired insulin signalling
    • Metabolic syndrome
  • Prediabetes and neuropathy
  • Glucose control in type 1 and type 2 diabetes
  • Pain management in DNP
    • Anticonvulsants
    • Antidepressants
    • Opioids
    • Other drugs
    • Treatment algorithm
  • Metabolic syndrome: implications for future treatments
  • Conclusions and future directions
  • Acknowledgments
  • References

---