The 1988 San Antonio Conference on Diabetic Neuropathy (¹¹), Boulton et al. (⁴), and the American Academy of Neurology (AAN), American Association of Electrodiagnostic Medicine (AAEM), and American Academy of Physical Medicine and Rehabilitation (AAPM&R) (¹²) have proposed criteria for diabetic neuropathies.

We propose separate definitions for typical DPN (DSPN) and atypical DPNs. DSPN is a symmetrical, length-dependent sensorimotor polyneuropathy attributableto metabolic and microvessel alterations as a result of chronic hyperglycemia exposure (diabetes) and cardiovascular risk covariates. An abnormality of nerve conduction tests, which is frequently subclinical, appears to be the first objective quantitative indication of the condition. The occurrences of diabetic retinopathy and nephropathy in a given patient strengthen the case that the polyneuropathy is attributable to diabetes. Other causes of sensorimotor polyneuropathy need to be excluded. For epidemiologic surveys or controlled clinical trials of DSPN, we advocate the use of nerve conduction (NC) testing as an early and reliable indicator of the occurrence of this neuropathy. To be reliable the test must be done rigorously using appropriate reference values corrected for applicable variables. Volunteered or elicited symptoms and signs and other clinical neurophysiologic abnormalities are also needed to characterize the symptoms, signs, and overall severity of the polyneuropathy. Recent studies emphasize the importance of the proficiency of the clinical neurologic assessment (¹³,¹⁴). Atypical DPNs have been less well characterized and studied.

Estimating the severity of DSPN has not received the attention it deserves. For a given patient with diabetes, it is not sufficient to simply identify patients as having or not having DSPN—severity also needs to be ascertained. We suggest a reliable objective and quantitative measure (i.e., NC abnormality) as the minimal criteria for the diagnosis of DSPN. When NC values have not been assessed, it is not possible to provide a confirmed diagnosis of DSPN—only a possible or probable diagnosis. Since the severity of DSPN is a combination of neuropathy symptoms, signs, neurophysiologic test abnormalities, and other dysfunctions and impairments, the sum scores of various measures of neurologic signs, symptoms, neurophysiologic test abnormalities, or scores of function of activities of daily living may provide an indication of the severity (¹³).

An alternative approach to estimating severity is to indicate severity by grades. Dyck (¹⁵) described the stages of severity:

• Grade 0 = no abnormality of NC, e.g., Σ 5 NC normal deviates <95th percentile or another suitable NC criterion
• Grade 1a = abnormality of NC, e.g., Σ 5 NC normal deviates ≥95th percentile without symptoms or signs
• Grade 1b = NC abnormality of stage 1a plus neurologic signs typical of DSPN but without neuropathy symptoms
• Grade 2a = NC abnormality of stage 1a with or without signs (but if present, <2b) and with typical neuropathic symptoms
• Grade 2b = NC abnormality of stage 1a, a moderate degree of weakness (i.e., 50%) of ankle dorsiflexion with or without neuropathy symptoms.

The presence of symptoms or signs of DSPN may include the following: symptoms–decreased sensation, positive neuropathic sensory symptoms (e.g., “asleep numbness,” prickling or stabbing, burning or aching pain) predominantly in the toes, feet, or legs; or signs–symmetric decrease of distal sensation or unequivocally decreased or absent ankle reflexes.

The presence of a combination of symptoms and signs of neuropathy include any two or more of the following: neuropathic symptoms, decreased distal sensation, or unequivocally decreased or absent ankle reflexes.

The presence of an abnormality of NC and a symptom or symptoms or a sign or signs of neuropathy confirm DSPN. If NC is normal, a validated measure of small fiber neuropathy (SFN) (with class 1 evidence) may be used.

To assess for the severity of DSPN, several approaches can be recommended: the graded approach outlined above, various continuous measures of sum scores of neurologic signs, symptoms or nerve test scores, scores of function of acts of daily living or of predetermined tasks or of disability. Irrespective of which approach is used, it is necessary to ensure good performance of evaluations with monitoring proficiency.

The presence of no signs or symptoms of neuropathy are confirmed with abnormal NCs or a validated measure of SFN (with class 1 evidence).

We recommend that definitions 1, 2, or 3 be used for clinical practice and definitions 3 or 4 be used for research studies.

Cardiovascular autonomic neuropathy (CAN) is defined as the impairment of autonomic control of the cardiovascular system. The prevalence of CAN varies widely from 2.5 to 50%. Factors that influence the prevalence of CAN include the diagnostic criteria used, patient age, and the duration of diabetes (²⁹,³⁰). Additional clinical correlates and predictors of CAN include glycemic control, presence of DPN, nephropathy and retinopathy, blood pressure (BP) levels, obesity, smoking, and cholesterol and triglycerides levels (³⁰,³¹). Intervention studies have documented the protective effect of glycemic control in type 1 diabetes (³²) and a multifactorial strategy aimed at lifestyle change with pharmacological correction of hyperglycemia, hypertension, dyslipidemia, and microalbuminuria (³³).

CAN is significantly associated with overall mortality (³⁴) and in some—but not all—studies with morbidity such as silent myocardial ischemia, coronary artery disease, stroke, diabetic nephropathy progression, and perioperative morbidity. Some pathogenetic mechanisms may link CAN to cardiovascular dysfunction and diabetic complications (³⁴). Thus, CAN assessment may be used for cardiovascular risk stratification in patients with and without established cardiovascular disease; as a marker for patients requiring more intensive monitoring during the perioperative period and other physiological stresses; and as an indicator for more intensive pharmacotherapeutic and lifestyle management of comorbid conditions.

Cardiovascular reflex tests are the gold standard in clinical autonomic testing. These tests have good sensitivity, specificity, and reproducibility and are noninvasive, safe, well-standardized, and easily performed (³⁵). However, a Valsalva maneuver must not be performed in patients with proliferative retinopathy. The most widely used tests assessing cardiac parasympathetic function are based on the time-domain heart rate (HR) response to deep breathing, a Valsalva maneuver, and postural change. Of these tests, HR to deep breathing has the greatest specificity (∼80%). Cardiovascular sympathetic function is assessed by measuring the BP response to orthostatic change and a Valsalva maneuver. The performance of these tests should be standardized, and the influence of confounding variables such as medications, hydration, and antecedent activity should be minimized. Age normative values should be used. The combination of cardiovascular autonomic tests with sudomotor function tests may allow a more accurate diagnosis of DAN (³⁶).

Diabetic patients with features of cardiac autonomic dysfunction such as unexplained tachycardia, orthostatic hypotension, and poor exercise tolerance, or with other symptoms of autonomic dysfunction should be evaluated for the presence of CAN. Screening for CAN should be performed at the diagnosis of type 2 diabetes and 5 years after the diagnosis of type 1 diabetes, particularly in patients at greater risk of CAN due to a history of poor glycemic control, cardiovascular risk factors, DPN, and macro- and microangiopathic diabetic complications.

Diagnostic criteria and staging of CAN are still being debated. We suggest that the presence of one abnormal cardiovagal test identifies possible or early CAN (²⁹); at least two abnormal HR tests are required for a definite or confirmed diagnosis of CAN (³⁰); and orthostatic hypotension (asymptomatic or symptomatic), in addition to HR test abnormalities, identify a condition of severe or advanced CAN (³¹). Progressive stages of CAN are associated with an increasingly worse prognosis (³⁴).

Attenuation (nondipping) or loss of BP nocturnal fall (reverse dipping) on ambulatory BP monitoring have been associated with CAN and attributed to the disruption of the circadian variation in sympathovagal activity (³⁷). In diabetic patients, nondipping or reverse dipping are independent predictors of cardiovascular events and the progression of diabetic nephropathy. Ambulatory BP monitoring may be useful in patients with CAN to detect nondipping and to address 24-h BP control (Table 1).

QT prolongation is an independent predictor of mortality in diabetic patients and is weakly associated with CAN (³⁸) (Table 1). The pathogenesis of QT prolongation is multifactorial and correlates include female sex, nephropathy, coronary heart disease, glycemic control, systolic BP, physical activity, and BMI.

The time-domain HR tests and the BP response to postural change have the reproducibility necessary for clinical trials. These tests were used as end points in the Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications Study (DCCT/EDIC) and other clinical trials.

Frequency-domain indexes obtained by applying spectral analysis to HR variability of short (5–7 min) and longer (24-h) electrocardiogram recordings provide a measure of sympathetic and parasympathetic modulation of HR. HR spectral power in the high-frequency region is a measure of parasympathetic modulation, while spectral power in the low-frequency region provides a measure of both sympathetic and parasympathetic modulation. The low-frequency BP variability may provide a measure of sympathetic modulation. To correctly assess the significance of the different regions, respiration should be measured or controlled breathing performed. These methods, which need standardization, have been widely used in research and as end points in clinical trials.

Sympathetic outflow, at rest and in response to various physiological perturbations, can be measured directly via microelectrodes inserted into a fascicle of a distal sympathetic nerve to the skin or muscle. The technique is invasive and time-consuming. Whole-body sympathetic activity is most accurately assessed by measurements of plasma concentrations of noradrenaline and adrenaline.

Assessment of cardiac vagal baroreflex sensitivity (BRS) combines information derived from both HR and BP in response to pharmacological or spontaneous BP perturbations. Cardiac vagal BRS is a well-established prognostic index in the general and diabetic populations (³⁹) and is frequently used in research studies (Table 1). Cardiac sympathetic BRS can be measured with simultaneous recordings of muscle sympathetic nerve activity.

Scintigraphic studies with radio labeled noradrenaline analogues allow a direct semi-quantitative ([¹²³I]-metaiodobenzylguanidine [MIBG] and single photon emission computed tomography) and quantitative assessment ([¹¹C]-hydroxyephedrine [HED] and positron emission tomography) of cardiac sympathetic integrity. Scintigraphic abnormalities are associated with CAN but may also be present in patients with normal cardiovascular autonomic tests (⁴⁰). No standardized methodology or normative values exist, and available data on reproducibility are limited. Scintigraphic studies are appropriate to explore the effects of sympathetic dysfunction on cardiac metabolism and function and are useful in assessing cardiac sympathetic function in research studies.

Research issues include: 1) the need for multivariate longitudinal studies to clarify the natural history of CAN in diabetes and pre-diabetes, to evaluate the impact of pharmacologic and lifestyle interventions targeting CAN, and to determine the effect of CAN on clinical outcomes; 2) the refinement and standardization of research measures to permit more widespread use; and 3) the need for studies of combined cardiovascular autonomic and other autonomic measures to improve diagnosis and outcome assessment.

Disordered GI motility may be associated with GI symptoms, impaired oral drug absorption, poor glycemic control, malnutrition, abnormal postprandial regulation of BP, poor QoL, and a high rate of hospitalization. The relationships with symptoms and CAN are relatively weak (⁴²). Esophageal transit is delayed in ∼50% of patients with longstanding diabetes and may be associated with regurgitation, dysphagia, and a propensity for pill-induced esophageal erosions and strictures. Gastroparesis affects ∼40% of patients with longstanding diabetes. Symptoms are variable and more common in patients with worse chronic glycemic control and psychological disorders (⁴³).

The rate of GE is a major determinant of postprandial blood glucose changes. In insulin-treated patients, nutrient delivery needs to be matched to the action of the exogenous insulin, and delayed GE is a cause of otherwise unexplained hypoglycemia (⁴⁴).

Postprandial hypotension occurs frequently in diabetes, and its magnitude is related directly to GE rate. The prevalence of disordered small and large intestinal and anorectal motility is high. Diarrhea may result from rapid or slow transit, which is complicated by bacterial overgrowth and/or disordered secretion. Constipation frequently occurs. Fecal incontinence is not uncommon and is related to reduced and unstable internal anal sphincter tone and impaired rectal compliance and sensation.

Studies of GI autonomic neuropathy, whether performed for clinical, epidemiological, or research purposes, may potentially be focused on GI symptoms, QoL, GI motility/transit, glycemic control, and/or postprandial BP. A number of instruments are available to quantify GI symptoms, including the Diabetes Bowel Symptom Questionnaire. Objective GE measurement is advocated for the diagnosis of gastroparesis. Evaluation of solid emptying is probably more sensitive than that of low-nutrient liquid or semi-solid meals. Medications that may influence GE should ideally be withdrawn, glycemia should ideally be <10 mmol/l throughout the test, and other causes of gastroparesis must be excluded. Failure to demonstrate delayed GE does not necessarily imply that symptoms are not due to “diabetic gastropathy,” but it does help guide drug therapy. Scintigraphy is still regarded as the gold standard technique for GE measurement. Standardization of the meal technique has been improved by the recommendation of a low-fat, egg white meal labeled with technetium-99 (99mTc) sulfur colloid (⁴⁵). Breath tests using nonradioactive 13 C-acetate or -octanoic acid as a label are appealing options, at least as a screening tool. They are safe, easy to perform, inexpensive, and correlate well with scintigraphy. Ultrasonography (two-dimensional and three-dimensional) is noninvasive and two-dimensional ultrasound has been validated for measuring emptying of liquids and semi-solids. However, obesity and abdominal gas, together with the necessity for an experienced operator, have limited its widespread use. Surface electrogastrography, used to detect abdominal gastric slow-wave activity, should be regarded as a research tool. A barium meal has no role in quantifying GE.

In the investigation of “diabetic diarrhea” celiac disease, exocrine pancreatic insufficiency and small intestinal bacterial overgrowth must be excluded. Tests of anorectal motor and sensory function are well developed for clinical use.

No study assessing sensitivity and specificity in DPN alone is available. However, several studies in SFN, which included patients with DPN, have been published. In a study of 58 patients with pure SFN (⁵⁹), a cut-off IENF density of ≤8.8/mm at the ankle was associated with a sensitivity of 77.2% and a specificity of 79.6%. In a study of 210 patients with SFN (⁶⁰), which included 65 diabetic patients, the Z-scores and 5th percentile provided the highest specificity (98 and 95%, respectively) but the lowest sensitivity (31 and 35%, respectively) compared with the receiver operating characteristic analysis (specificity 64%, sensitivity 78%). Thus, the diagnostic yield of skin biopsy may depend on the reference and cut-off values selected and the definition of SFN. IENF density correlates inversely with both cold and heat detection thresholds (⁶¹).

The AAN, AAEM, and AAPM&R provide a level C recommendation for the use of skin biopsy to diagnose DSPN, particularly SFN (³⁶). The European Federation of the Neurological Societies and the Peripheral Nerve Society revised guidelines on the use of skin biopsy in the diagnosis of SFN and have concluded that IENF density is a reliable and efficient technique to confirm the clinical diagnosis of SFN with a level A recommendation (⁵⁸). The presence of diffuse IENF swellings, especially if large, may predict a decline in IENF density (⁵⁸)

IENF density is significantly reduced in patients with normal NC, suggesting early damage to small nerve fibers (⁶²,⁶³), and there is an inverse correlation with the Neurological Disability Score (⁶³). Additionally, IENF density is lower in diabetic patients with painful—compared with painless—early neuropathy (⁶⁴). A 1-year diet and exercise intervention program for patients with SFN and IGT led to increased IENF density (⁶⁵). These data suggest that IENF loss is an early feature of diabetes, progresses with increasing neuropathic severity, and may repair with early intervention.

Recently, a skin biopsy study has shown a correlation between sweat gland nerve fiber density, neuropathic symptoms, neurological deficits, and sweat production in diabetic patients (⁶⁶).

ABPM, ambulatory BP monitoring; HRV, heart rate variability; MNSA, muscle sympathetic nerve activity.