Is there cardiac autonomic neuropathy in prediabetes?

Abstract

Although there is considerably more data showing an association between type 2 diabetes mellitus (T2DM) and autonomic neuropathy, accumulating evidence indicates that cardiovascular autonomic neuropathy (CAN) is common in persons with impaired glucose tolerance (IGT). Furthermore, CAN may occur early after a metabolic insult and obesity, especially among mean, and seems to play an important role in the early pathogenesis of CAN. Autonomic symptoms are common in subjects with IGT. In addition to defects in CAN, in subjects with IGT, there is impaired sudomotor function and abnormalities of endothelial peripheral vasoreactivity. At the present time, the only interventions that may be effective in preventing or reversing IGT associated autonomic neuropathy are lifestyle improvement. These include a tailored diet and exercise program. Other approaches that may be beneficial include modulation of oxidative stress and improvement of metabolic regulation in subjects with IGT. Interventions are most likely to be effective early in the course of disease and therefore it is extremely important to have early diagnosis of IGT and autonomic neuropathy.

1. Introduction

Type 2 diabetes mellitus (T2DM) is now a worldwide epidemic and the prevalence continues to rise. An early, potentially deadly and common but often underrecognized complication of T2DM is cardiovascular autonomic neuropathy (CAN). CAN is a serious life-threatening complication of diabetes, which affects about 25% of subjects with type 1 diabetes mellitus (T1DM) and 33% of those with T2DM (Vinik and Ziegler, 2007). The exact prevalence of CAN among individuals with impaired glucose tolerance (IGT) remains unknown and varies widely in published reports due to a wide degree of variability in definition of IGT, population studied, and type of autonomic assessment utilized. Although hyperglycemia has been recognized as a predictor of CAN in diabetic patients, the glucose threshold at which autonomic function begins to become impaired has not been established.

The treatment of diabetic autonomic neuropathy (DAN) can be difficult and it is preferable to prevent its onset or slow the progression. CAN is often present at the time of diagnosis of T2DM. For this reason, identifying patients with IGT and establishing an effective screening protocol for DAN would be of great clinical significance. Patients with IGT, through lifestyle and medication intervention programs, may be able to return to normoglycemia or prolong the amount of time until they progress to T2DM. This could therefore be an effective strategy against the life-threatening complications of CAN.

2. Impaired glucose tolerance

Impaired fasting glucose (IFG) and IGT are two forms of glucose dysmetabolism that are considered as prediabetes and represent an increased risk of developing diabetes and cardiovascular disease. Both IGT and IFG may be associated with impaired autonomic function as discussed below. There is currently insufficient data to determine if the prevalence of autonomic neuropathy and specific autonomic abnormalities are different between IGT and IFG. Thus, for the purposes of this paper, IGT will refer to patients with either defined IGT or IFG or both. However, a few important distinctions will be made. The American Diabetes Association has defined prediabetes as a hemoglobin A1c of 5.7%–6.4%. A diagnosis of IFG or IGT can be made by use of a 2-hour oral glucose tolerance test (OGTT). The patient fasts overnight, and a fasting plasma glucose is measured early the next morning. The patient is then given a 75 g oral dextrose solution to ingest and a plasma glucose measurement is taken 2 h later. IFG is defined as a fasting plasma glucose of 100–125 mg/dl (5.6–6.9 mmol/l) and IGT is a plasma glucose of 140–199 mg/dl (7.8–11.0 mmol/l) 2 h after a 75 g dextrose load (American Diabetes, 2018). For reference, the diagnosis of diabetes requires a hemoglobin A1c > 6.5%, a fasting plasma glucose > 126 mg/dl (7 mmol/l), or a post 2-hour plasma glucose > 200 mg/dl (11.1 mmol/l).

The International Diabetes Federation, estimates that worldwide up to 472 million adults will have IGT by 2030 (IDF, 2015). In 2015, the Centers for Disease Control estimated that a third of adults over the age of 18 years old in the United States had prediabetes based on their fasting glucose or hemoglobin A1c level. Using the same criteria, in 2017, the Centers for Disease Control, National Diabetes Statistics Report reported almost half of adults aged 65 years or older had prediabetes (CDC, 2017). Of note, the Centers for Disease Control and Prevention found that only 11.6% of adults with prediabetes reported being told by a health professional that they had prediabetes. Not only is IGT a marker of risk of developing diabetes, but it is an independent marker of large- and small-vessel injury. Independent of progression to diabetes, individuals with IGT have an increased risk of all-cause mortality (Baron, 2001; Kuller et al., 2000) as well as an increased incidence of ischemic heart disease and cerebrovascular disease (Kuller et al., 2000; Levitzky et al., 2008). Microvascular complications associated with IGT are similar to those seen with diabetes and include retinopathy, nephropathy and neuropathy (Singleton et al., 2003). The neuropathy that is seen in IGT patients is typically a painful, symmetrical, distally prominent, sensory neuropathy (Russell and Zilliox, 2014; Singleton et al., 2001b; Zilliox and Russell, 2011b; Zilliox et al., 2015). In general, there is early involvement of small diameter, unmyelinated nerve fibers (Green et al., 2010), (Divisova et al., 2012). IGT neuropathy resembles the neuropathy seen in the early diabetes with prominent neuropathic pain and autonomic symptoms (Singleton et al., 2003).

3. Impaired fasting glucose vs impaired glucose tolerance

As previously stated, IFG is defined by an elevated fasting plasma glucose concentration and IGT is defined by an elevated 2-h plasma glucose concentration after a 75-g glucose load. Further separations have been made to distinguish isolated IFG, isolated IGT and combined IFG and IGT as well as normal glucose tolerance (NGT, Table 1). IFG was originally defined by the American Diabetes Association to identify individuals who were at increased risk of developing diabetes based upon a fasting plasma glucose measurement and it was meant to correspond to IGT (Genuth et al., 2003). However, IGT and IFG do not define the same individuals. There are epidemiologic differences between IFG and IGT that suggest different underlying pathophysiology. For instance, the prevalence of IGT and IFG is different with IGT being generally more prevalent among women compared to men (Cowie et al., 2006). In addition, the two conditions do not frequently co-exist. IGT is present in approximately half of IFG patients, but IFG exists in fewer (approximately 20%) IGT patients (Alberti and Zimmet, 1998). Accordingly, an oral glucose tolerance test is a more sensitive test than a fasting plasma glucose to detect early glucose dysregulation. It has been shown that performing an oral glucose tolerance test in addition to a fasting blood sugar increases the diagnostic yield by 17% (Carnevale Schianca et al., 2003).

Table 1.

Definition of glycemic control states.

Glucose tolerance state	Fasting plasma glucose level (mg/dl)	2-h plasma glucose level in OGTT (mg/dl)
Impaired fasting glucose (IFG)	100–125	< 200
Isolated IFG	100–125	< 140
Impaired glucose tolerance (IGT)	< 126	140–199
Isolated IGT	< 100	140–199
Combined IFG/IGT	100–125	140–199
Normal glucose tolerance (NGT)	< 100	< 140

Adapted from Nathan et al. (2007).

Both IFG and IGT are associated with progression to diabetes, especially if older or overweight, and with an increased risk for cardiovascular disease. However, IGT is a slightly stronger risk predictor for cardiovascular disease independent of progression to diabetes (Anon, 1997; Balkau et al., 2004; Levitan et al., 2004; Meigs et al., 2002; Ning et al., 2010; Qiao et al., 2002). In addition, IGT is independently associated with microvascular complications of diabetes (Singleton et al., 2003). As discussed further in this paper, the association between IFG and IGT and CAN is not as clear and deserves further examination. The differences described in IFG and IGT reflect different pathophysiologic mechanisms. While insulin secretion is defective in IFG, isolated IGT is due to impaired insulin sensitivity/insulin resistance. While both conditions are insulin-resistant states, they differ in the site of insulin resistance (Abdul-Ghani et al., 2006). Individuals with isolated IFG predominantly have hepatic insulin resistance and normal muscle insulin sensitivity, which leads to excessive fasting hepatic glucose production. In comparison, individuals with isolated IGT have normal to slightly reduced hepatic insulin sensitivity and moderate to severe muscle insulin resistance. In combination with defective late insulin secretion, this leads to prolonged hyperglycemia following a glucose load (Nathan et al., 2007).

4. Impaired glucose tolerance and peripheral neuropathy

Neuropathy is the most common complication of diabetes. Previously, length dependent peripheral neuropathy was thought to be a late complication of diabetes that was associated with poor glucose control. However, more recent work has shown an association between IGT and peripheral neuropathy. In fact, peripheral neuropathy most likely occurs at the earliest stages of glucose dysregulation. Nerve conduction studies performed at the time of diagnosis of T2DM demonstrate that almost 20% of patients already have neuropathy (Cohen et al., 1998). In addition, several population-based studies have found an increased prevalence of peripheral neuropathy in patients with IGT compared to normoglycemic controls (Franklin et al., 1990; Ziegler et al., 2009) and screening of individuals with “idiopathic” neuropathy with an OGTT found that 30–50% of patients had IGT, which is a significantly higher prevalence than the general population (Singleton et al., 2001a; Sumner et al., 2003).

Most patients with IGT related PN have a symmetric, distal sensory polyneuropathy with prominent neuropathic pain. A large prospective study found that 81% had only sensory complaints, and 92% had neuropathic pain as a dominant symptom of their neuropathy (Singleton et al., 2001b). This is clinically similar to early diabetic neuropathy, which has neuropathic pain and autonomic dysfunction due to involvement small unmyelinated nerve fibers that carry pain and temperature sensation as well as autonomic fibers. In fact, intraepidermal nerve fiber density measurement from skin biopsies, which are used to quantify small fiber loss, are reduced and show an altered morphology in patients with IGT and early diabetes associated PN (Smith et al., 2001, 2006).

5. Clinical presentation of autonomic neuropathy

As previously mentioned, autonomic neuropathy is a common complication of diabetes and likely prediabetes that is associated with high levels of morbidity as well as an increased risk of mortality. It can affect multiple organ systems throughout the body: cardiovascular, gastrointestinal, genitourinary, pupillary, and sudomotor. Due to the increased risk of cardiovascular mortality associated with DAN there has been an increased interest in early recognition and diagnosis of DAN. The diagnosis of DAN is classified as clinical (Table 2), or subclinical based upon symptoms and the results of autonomic testing, which is discussed in the next section. DAN is a multisystemic disease that affects individuals differently. There is not a consistent, recognized pattern of involvement or progression. Cardiac autonomic neuropathy (CAN) may present with symptoms such as resting tachycardia, orthostatic hypotension (fainting, lightheadedness, dizziness, blurry vision or neck pain upon standing), or exercise intolerance. There is also a risk of silent myocardial ischemia with CAN. Symptoms of gastrointestinal autonomic neuropathy are often nonspecific and can include esophageal dysmotility and gastroesophageal reflux disease, delayed gastric emptying, constipation, diarrhea, and fecal incontinence. Involvement of the genitourinary system can result in erectile dysfunction, retrograde ejaculation, female sexual dysfunction, or a neurogenic bladder. Pupillary involvement may manifest as pupillomotor function impairment with patient complaint of photosensitivity or impaired night vision. Sudomotor dysfunction with anhidrosis can lead to dry skin and heat intolerance (Table 2).

Table 2.

Clinical manifestations of autonomic neuropathy in IGT and T2DM.

Cardiovascular	Perioperative instability, resting or postural tachycardia, loss of reflex heart variations, hypertension, exercise intolerance, orthostatic hypotension, silent myocardial ischemia, increased peripheral blood flow, lower extremity edema, loss of cutaneous vasomotor reflexes
Gastrointestinal	Esophageal dysmotility, gastroesophageal reflux disease, gastroparesis, poor glycemic control, malnutrition, abnormal postprandial hypotension, constipation, diarrhea, fecal incontinence
Genitourinary	Erectile dysfunction, retrograde ejaculation, dyspareunia from lack of vaginal lubrication, bladder dysfunction with frequent urinary tract infections, overflow incontinence and poor urinary stream
Pupillary	Difficulty with dark adaptation and difficulty driving at night, Argyll-Robertson pupil
Metabolic	Hypoglycemic unawareness, hypoglycemia-associated autonomic failure
Sudomotor	Distal anhidrosis, proximal hyperhidrosis, heat intolerance, dry skin, development of foot ulcers, gustatory sweating

Autonomic symptoms can be used to accurately diagnose autonomic neuropathy. The Composite Autonomic Symptom Score 31 questionnaire has been validated against cardiovascular reflex tests-based diagnosis of CAN (Sletten et al., 2012). In addition, the presence of autonomic symptoms in early diabetic neuropathy can be assessed by the Survey of Autonomic Symptoms (SAS), which is a validated and easily administered tool to measure autonomic symptoms that is sensitive enough to detect mild autonomic neuropathy (Zilliox et al., 2011). In subjects with IGT, there is a strong association between the SAS and certain measures of CAN such as the 30:15 ratio (−0.53; P < 0.01) and also between the SAS and a measure of sudomotor autonomic function, the forearm sweat volume (−0.31; P < 0.05). There is also a strong association with other validated measures of autonomic function, for example the Autonomic Symptom Profile (ASP) of the full Composite Autonomic Symptom Score (0.68; P < 0.001). However, the poor correlation with other measures of cardiac autonomic function, indicates that autonomic symptoms and autonomic testing assess different components of overall autonomic function. The most common autonomic symptoms reported in men and woman were dry eyes or dry mouth, feet feeling colder than the rest of the body, and difficulty in obtaining an erection in men (Table 3). In contrast, symptoms related to gastroparesis were uncommon.

Table 3.

Autonomic symptom frequency in subjects with IGT and neuropathy.

Item	% Affected		Mean TIS (SEM)
	Men	Women	Men	Women
Lightheadedness?	50.00	38.00	1.28 ± 0.35	0.96 ± 0.26
Dry mouth or dry eyes?	77.78	70.83	1.89 ± 0.27	2.00 ± 0.34
Feet pale or blue?	22.22	33.33	0.50 ± 0.23	0.83 ± 0.28
Feet colder than the rest of your body?	66.67	70.83	2.28 ± 0.44	2.26 ± 0.34
Sweating in your feet decreased compared to the rest of your body?	33.33	25.00	1.00 ± 0.39	0.50 ± 0.23
Sweating in your feet decreased or absent (exercise/hot weather)?	16.67	20.83	0.44 ± 0.29	0.50 ± 0.25
Sweating in your hands increased compared to the rest of your body?	5.56	20.83	0.28 ± 0.12	0.38 ± 0.17
Nausea, vomiting, or bloating after eating a small meal?	5.56	16.67	0.06 ± 0.02	0.50 ± 0.26
Persistent diarrhea?	5.56	16.67	0.28 ± 0.15	0.26 ± 0.16
Persistent constipation?	11.11	25.00	0.17 ± 0.09	1.00 ± 0.36
Leaking of urine?	22.22	45.83	0.67 ± 0.33	1.14 ± 0.33
Difficulty obtaining an erection (men)?	55.56	NA	1.82 ± 0.48	NA

From Zilliox et al. (2011). Reproduced with permission.

6. Methods of assessment of autonomic neuropathy

Diagnostic criteria and staging of CAN has been suggested by the Toronto Consensus Panel: (1) the presence of one abnormal cardiovagal test identifies possible or early CAN, (2) at least 2 abnormal heart rate tests are required for a definite diagnosis of CAN, and (3) orthostatic hypotension in addition to heart rate test abnormalities identify severe CAN (Tesfaye et al., 2010). Cardiovascular reflex tests are the gold standard in clinical autonomic neuropathy (Anon, 1996b; Tesfaye et al., 2010). These tests include the heart rate response to deep breathing, heart rate and blood pressure response to a Valsalva maneuver and heart rate and blood pressure response to postural change. These tests are noninvasive, easy to perform, standardized with good sensitivity, specificity and reproducibility. However, they require careful patient preparation to minimize the effects of medications, hydration status and activity. Reduced heart rate variability on tests of cardiovagal function is frequently an early marker of subclinical CAN. Among the cardiovascular reflex tests of heart rate there is not one that has been found to be superior. However, the most specific test for cardiac parasympathetic function is the heart rate response to deep breathing (expiration: inspiration ratio).

Another frequently used measure of autonomic function that is often used in clinical trials is power spectral analysis to assess the heart rate variability (HRV) of successive R-R intervals. The testing can be performed on short (5–7 min) or 24-hour electrocardiogram recordings and less patient participation is required as compared to cardiovascular reflex tests. The analysis includes low and high frequency bands. The high frequency region is considered a marker of vagal activity and the low frequency component includes both sympathetic and vagal activity. As mentioned, HRV measures have been widely used in research. However, they need standardization and better reproducibility. The Toronto Consensus Panel on Diabetic Neuropathy suggested that heart rate cardiovascular tests and HRV time and frequency domain indices as well as baroreflex sensitivity measures (BRS) and scintigraphic studies are suitable as end-point measures in clinical trials (Spallone et al., 2011). Comparisons of power spectral analysis and some time-domain heart rate tests have found that time domain values correlated strongly with high-frequency spectral indexes (Freeman et al., 1991).

Sudomotor dysfunction in autonomic neuropathy is commonly assessed by the quantitative sudomotor axon reflex test (QSART), which detects peripheral, postganglionic sympathetic denervation. Data in subjects with IGT indicates that skin biopsy, which is more commonly used to assess small fiber neuropathy, can also be used for assessment of sweat gland nerve fiber density. The thermoregulatory sweat test requires specialized equipment, but it assesses both central and peripheral aspects of the sympathetic nervous system.

7. Prediabetes and autonomic neuropathy

An association between IGT and autonomic dysfunction has been found by several studies, but there has been little consistency among these studies. The major source of variability among studies has been in terms of patient population and characterization of glucose dysmetabolism as well as the assessment of autonomic function. A recent review of CAN in T2DM by Spallone described the amount of heterogeneity in 14 studies examining the association between autonomic dysfunction and pre-diabetes (Spallone, 2019). In these 14 studies, as well as two additional studies, four included both IFG and IGT, eight included only IGT, four included only IFG, and one study, the KORA S4 study, included a separate category of IFG plus IGT (Ziegler et al., 2015). Seven studies had small sample sizes (< 200 individuals) and eight studies were population based. In terms of CAN testing, standardized procedures and age-related reference values were not always used for cardiovagal reflex testing.

Despite these inconsistencies, findings suggest that decreases in autonomic function, especially reduced HRV indices that measure cardiovagal function, are present at early stages of glucose dysregulation. Altered cardiovagal responses are also a common feature of early DAN (Vinik and Ziegler, 2007). However, three studies did not find a difference in subjects with IGT compared to those with NGT (Annuzzi et al., 1983; Fujimoto et al., 1987; Isak et al., 2008). In studies including both IFG and IGT there was a trend toward greater autonomic dysfunction in IGT versus IFG (Dimova et al., 2017; Wu et al., 2007). There was also more severe autonomic neuropathy in participants with combined IFG and IGT compared to those with isolated IFG or IGT (Ziegler et al., 2015).

In studies that only included participants with IGT, The Hoorn Study described a reduction in HRV. However, only the standard deviation of normal RR intervals (NN intervals: SDNN) was significantly reduced in subjects with IGT compared to normoglycemic subjects. Cardiovascular autonomic function was most strongly associated with age and anti-hypertensive drugs (Gerritsen et al., 2000). A small study (n = 91) comparing autonomic function in patients with IGT compared to subjects with NGT found significant differences between the two groups in heart rate response to deep breathing, Valsalva ratio, blood pressure responses to standing and sustained handgrip but not the 30:15 ratio, and HRV as measured by the triangular index. BMI adjustment had only minor effects on the outcomes (Putz et al., 2009). Another small study by the same group found that 60% of participants with IGT had at least 1 abnormal cardiovascular test compared to none in the NGT group. In addition, significant differences were found in most of the tests, except the 30:15 ratio (Putz et al., 2013). Overall, the changes found in cardiac autonomic function in isolated IGT subjects were subclinical and mild, but detectable.

Two small studies of IGT patients, one in Italy and one in Japanese-American men, examined autonomic function as assessed by heart rate response to deep breathing and found no significant difference compared to individuals with NGT (Annuzzi et al., 1983; Fujimoto et al., 1987). Another study including only IGT patients, examined 50 subjects (25 with NGT and 25 with IGT) and found no evidence of CAN assessed by HRV, heart rate response to deep breathing, heart rate response to Valsalva maneuver, blood pressure response to standing, and hand grip test in the IGT subjects. However, amplitudes of the sympathetic skin response were lower in the IGT group, which is consistent with a possible sudomotor autonomic neuropathy (Isak et al., 2008).

There has been significantly less attention paid to non-cardiac autonomic neuropathy in IGT. A study of postganglionic sudomotor function in subjects with IGT found a statistically significant difference in total sweat volume between participants with IGT and those with NGT. After adjustment for differences in age, gender or level of physical activity, both summary variables of sudomotor function remained significantly different (total sweat volume, P < 0.01; proximal-distal ratio, P < 0.001) (Grandinetti et al., 2007). The percent of subjects with abnormal sudomotor responses in the forearm and foot is shown in Fig. 1.

Fig. 1.

Pattern of QSART abnormalities in subjects with IGT based on normative data for age and gender. From (Zilliox et al., 2011). Reproduced with permission.

Four studies were reviewed that included only participants with IFG, but did not include an oral glucose tolerance test in their protocol to test for IGT (Singh et al., 2000; Panzer et al., 2002; Schroeder et al., 2005; Stein et al., 2007). Three of these studies utilized the older American Diabetes Association criteria of a fasting glucose of 6.1–6.9 mmol/l to define IFG (Panzer et al., 2002; Schroeder et al., 2005; Singh et al., 2000) and one study used the 2003 ADA criteria for IFG and divided the subjects into subcategories of mild IFG (5.6–6.0 mmol/l) and significant IFG (6.1–6.9 mmol/l) (Stein et al., 2007). The Framingham Heart Study included 56 subjects with IFG who were found to have reduced HRV (decreased SDNN and low and high frequency spectral components with LF and HF power) compared with 1779 normal fasting glucose (NFG) control subjects. HRV was found to be inversely associated with plasma glucose levels and was also found to be reduced in subjects with T2DM (Singh et al., 2000). The Atherosclerosis Risk in Communities (ARIC) study followed subjects with NFG, IFG, T2DM and nondiabetic hyperinsulinemia over 9 years. At baseline, the IFG subjects had a lower RR interval, but not a lower SDNN, than subjects with NFG (Schroeder et al., 2005). Another population-based study, derived from the Lipid Research Clinics Prevalence Study, examined the association between fasting plasma glucose and abnormal heart rate recovery as a marker of parasympathetic activity. They found that subjects with IFG had a greater prevalence of abnormal heart rate recovery compared to those with NFG (42 vs 31%). Furthermore, after controlling for age, sex, race, BMI, resting systolic blood pressure, HDL cholesterol, smoking regular physical activity, and resting heart rate, fasting plasma glucose was independently associated with abnormal heart rate recover (Panzer et al., 2002). The Cardiovascular Health Study examined heart rate and HRV indices in subjects with NFG, IFG and T2DM compared to those with metabolic syndrome independent of elevated fasting glucose levels. The subjects with IFG were divided into mild IFG (5.6–6.0 mmol/l) and significant IFG (FG 6.1–6.9 mmol/l). The subjects with significant IFG had reduced HRV indices compared to those with mild IFG or NFG. In subjects with NFG or mild IFG, having at least 2 components of the metabolic syndrome was associated with a greater decrease in HRV and in subjects with significant IFG and T2DM, metabolic syndrome was associated with decreased HRV (Stein et al., 2007). These results suggest that factors associated with increasing non-diabetic fasting glucose levels and metabolic syndrome play a role in the onset of CAN.

The methods of these four studies that included only participants with IFG utilized fasting plasma glucose measurements to classify subjects as normoglycemic, IFG, or T2DM. An oral glucose tolerance test was not performed so the IFG subjects in these studies may also have had IGT and even T2DM and may have had a greater degree of glucose dysregulation than isolated IFG subjects in studies that used both a fasting and post-load 2hr glucose to characterize subjects. A population-based study with a total of 1440 participants from Taiwan examined patients with normal glucose tolerance, IFG, IGT, and T2DM (Wu et al., 2007). Subjects with IGT and T2DM were found to have a reduction in measures of HRV (reduced 30:15 ratio and high frequency spectral component but with an increased low frequency: high frequency ratio) compared to those with NGT. However, after controlling for confounding factors there was no difference in cardiac autonomic function seen in subjects with isolated IFG compared to those with NGT. The authors concluded that, independent of other cardiovascular risk factors, altered cardiac autonomic function is present in both IGT and T2DM subjects, but not IFG subjects (Wu et al., 2007). The KORA S4 study also examined subjects with IFG and IGT and it included a group of subjects with combined IFG-IGT, which is the group with the highest risk of developing T2DM (Ziegler et al., 2015). This study examined 1332 individuals aged 55–74 categorized into 6 different groups: NGT, isolated IFG (i-IFG), isolated IGT (i-IGT), combined IFG-IGT (IFG-IGT), newly detected T2DM (n-T2DM), and known T2DM (k-T2DM). The rates of CAN, which was defined as ≥2 of 4 abnormal HRV indices, were: NGT, 4.5%; i-IFG, 8.1%; i-IGT, 5.9%; IFG–IGT, 11.4%; n-T2DM,11.7%; and k-T2DM, 17.5% (p < 0.05 vs NGT, except for i-IGT). Contrary to the Taiwanese and other studies, in this patient population from Germany, it was found that the prevalence of CAN was increased not only in those with T2DM, but also in those with IFG-IGT and, to a lesser degree, in those with i-IFG but not in those with i-IGT (Ziegler et al., 2015).

A third study of cardiac autonomic function (HRV indices) in subjects with NGT, IFG and IGT found an increased prevalence of CAN, defined as at least 2 abnormal autonomic tests, in IFG and IGT as well as T2DM. Confirmed CAN was found in 12.3% of NGT, 19.8% of prediabetes (13.2% of IFG and 20.6% of IGT), and 32.2% of newly diagnosed T2DM. There was a trend toward lower sympathetic and parasympathetic tone with increasing levels of glucose intolerance. Age, waist circumference and QTc interval were found to correlate with both sympathetic and parasympathetic activity and diastolic blood pressure and 120-minute plasma glucose were also found to correlate with sympathetic power (Dimova et al., 2017).

A study of IGT and autonomic neuropathy from Brazil (Rezende et al., 1997) further confirmed that CAN is common in subjects with IGT. In this study, standardized autonomic tests were performed in 44 subjects with IGT and compared to 43 control subjects. The standardized autonomic tests consisted of heart frequency testing, Valsalva maneuver, postural tests and sinus arrhythmia. Sinus arrhythmia was abnormal in 54.5% of subjects with IGT and in 32.5% of controls (p = 0.0039) and the Valsalva maneuver was abnormal in 34.1% with IGT and in 7% of controls (p = 0.004). The postural test was similar between groups (p = 0.334).

Another study that found abnormal autonomic function in both IFG and IGT was designed to assess autonomic function in insulin resistant individuals (Perciaccante et al., 2006). 80 individuals with insulin resistance based on the homeostasis model assessment-index were enrolled and the participants were divided equally into 4 groups: normoglycemic, IFG, IGT, and T2DM and compared to 25 control subjects in terms of heart rate variability analysis. It was found that there was a reduction in SDNN in all four groups of subjects with insulin resistance compared to the control group and that the group with T2DM had greater autonomic dysfunction that the other three groups. The data support the hypothesis that insulin resistance might cause a global reduction of autonomic nervous system activity. It also suggests that dysautonomia increases if insulin resistance is associated with worsening glucose metabolic impairment (Perciaccante et al., 2006).

There have been other clinical studies that have examined the role that insulin may play in CAN. The ARIC study previously described found a nonsignificant difference in HRV between individuals with and without hyperinsulinemia (Schroeder et al., 2005). Due to the findings of increased risk for cardiac events in individuals with nondiabetic fasting plasma glucose levels, baroreflex sensitivity was measured in 162 nondiabetic subjects to determine if fasting plasma glucose or fasting plasma insulin levels were associated with reductions in baroreflex sensitivity (Watkins et al., 2000). In this study the subjects were not categorized as IFG or NGT. After controlling for age, blood pressure and BMI, elevations in fasting plasma insulin but not fasting plasma glucose levels were associated with reductions in autonomic cardiac control as measured by baroreflex sensitivity. Of note, univariate analysis did show that fasting plasma glucose was negatively correlated with baroreflex sensitivity (Watkins et al., 2000). These results support the hypothesis that even modest, nondiabetic elevations of fasting plasma glucose or insulin are enough to impact cardiac autonomic function. Furthermore, the finding that IFG was not associated with impaired cardiac autonomic function when the effects of age, blood pressure, and BMI are accounted for, suggests that glucose may be less directly related to reduced autonomic control than insulin in this population.

Although most studies of autonomic function in IGT have examined changes in cardiovagal, adrenergic, or sudomotor function, some studies have examined other aspects of autonomic function (Boronikolos et al., 2015; Straznicky et al., 2012). For example, one study did examine the effect of IGT on upper gastrointestinal motility (Boronikolos et al., 2015). Gastric emptying, esophageal motility and gastrointestinal symptoms were examined in volunteers with diabetes (41), IGT (17) and normoglycemia (31). A gastric emptying breath test and high-resolution esophageal manometry were performed. Although impaired gastric emptying is usually a late feature of diabetes, it was delayed in individuals with IGT possibly secondary to acute hyperglycemia.

8. Progression of autonomic neuropathy

Although there is considerable data on the progression of autonomic neuropathy and diabetes, far less is known about the progression of autonomic neuropathy in IGT. This is probably because, first only fairly recently has an association between IGT and neuropathy been recognized and second it is now appreciated that glycemic control represents a continuum with fluctuations between IGT and T2DM being common early in the disease (Anon, 1996a, 2000, 2012). In one small study, the autonomic symptom score (ASS) and the E:I ratio were used to assess progression in autonomic neuropathy in persons with normoglycemia, IGT, and in patients with well-controlled T2DM (Zimmerman et al., 2018). Participants were recruited in 2003/2004 with a follow-up in 2014. The participants’ glucose tolerance was categorized using oral glucose tolerance tests. As expected the ASS was higher at follow-up in the T2DM group than in the NGT group (mean 1.21 ± 1.30 vs. 0.79 ± 0.7; p < 0.05). Interestingly, the ASS also increased in subjects with IGT over the 10-year period, however this was not statistically significant. In contrast the E:I ratio did not deteriorate more than could be expected due to age in well-controlled T2DM and in fact improved over 10 years in the IGT subjects, although this was only in 9 subjects and the variance was high. No relationship was found between E:I ratio and HbA1c or ASS. Even though this was a small cohort of subjects, the data for IGT and well-controlled T2DM subjects combined shows that there is minimal or no progression in the ASS and E:I ratio over 10 years and the autonomic symptoms do not correlate well with other measures of autonomic function as confirmed by another study using a different autonomic symptom score, the SAS (Zilliox et al., 2011).

A further study examined progression of somatic and autonomic neuropathy for up to 4 years (Kannan et al., 2014). The prevalence of autonomic neuropathy was determined in subjects with IGT and no other identifiable cause of neuropathy (Kannan et al., 2014). The tests used included heart rate response to deep breathing, Valsalva maneuver, standing (30:15 ratio), blood pressure response to standing and sympathetic skin response (SSR). An abnormality in ≥2 tests was considered as abnormal. The cohort included 30 age-matched controls and 58 subjects with IGT based on an oral glucose tolerance test and defined using World Health Organization (WHO) criteria. All subjects had normal glycosylated hemoglobin HbA1c, vitamin B12 levels, and thyroid function. Of the 58 subjects, autonomic neuropathy was detected during follow up in 8 (13.8%). Twenty subjects (34.5%) developed T2DM in the follow-up period.

The largest study was the Finnish Diabetes Prevention Study cohort (Laitinen et al., 2011) in which 268 individuals with IGT at baseline, who did not develop T2DM during follow-up, were studied for cardiovascular autonomic neuropathy. The participants were divided into intervention and control groups. At the second annual follow-up visit after the end of a lifestyle intervention, deep-breathing and active orthostatic tests were performed to detect possible parasympathetic and sympathetic dysfunction. Prevalence of parasympathetic dysfunction was 25% and prevalence of sympathetic dysfunction was 6%, with no difference between the former intervention and control group participants or between men and women. Unfortunately, baseline measurements of autonomic function were not obtained in this study. Thus, it was not possible to determine the longitudinal changes in autonomic function (Laitinen et al., 2011).

9. Rate of onset of autonomic neuropathy with IGT

As with progression of autonomic neuropathy in IGT, another question that has not been fully answered is the rate of progression of autonomic neuropathy in IGT. There is some evidence that the onset may be fairly rapid as with somatic neuropathy. For example, in one study, 48 healthy women with prior gestational diabetes (pre-T2DM) were assessed 3 months after delivery (Gasic et al., 2007). Alterations of cardiac autonomic function were determined using heart rate variability using both time, as well as frequency, domain methods with 24-h Holter monitoring. Control subjects included 20 women with normal glucose tolerance during and after pregnancy. Time domain analysis (standard deviation of normal RR intervals) showed a reduced HRV in 25 out of the 48 (52%) women with prior gestational diabetes. There was both sympathetic as well as parasympathetic functional impairment based on the HRV, although sympathetic impairment predominated. The impairment of cardiac autonomic function correlated with HbA1c values and the 2-h blood glucose concentration, using an oral glucose tolerance test, but not did not correlate with insulin sensitivity. Thus, functional impairment of autonomic function may occur very early after impaired glycemic control. The study did not address if the autonomic dysfunction eventually returned to normal after pregnancy and if the prevalence of autonomic dysfunction was associated with number of pregnancies.

10. Pathophysiology of autonomic neuropathy in prediabetes

In T2DM both insulin resistance and hyperglycemia are considered to be responsible for DAN, however the underlying pathophysiology remains unclear. There are contradictory results from studies examining the relative effects of elevated fasting plasma insulin levels vs. fasting plasma glucose on the development of DAN in newly diagnosed T2DM patients (Toyry et al., 1996; Vanninen et al., 1993; Ziegler et al., 1991). Even more uncertain is the relationship between autonomic activity, IFG and IGT.

In IGT many factors are likely to contribute to autonomic dysfunction. In prediabetes, independent correlates of autonomic measures include age, BMI, waist circumference, other components of metabolic syndrome, hypertension and anti-hypertensive drugs, and fasting and 2hr post-load glucose level (Annuzzi et al., 1983; Dimova et al., 2017; Fujimoto et al., 1987; Gerritsen et al., 2000; Laitinen et al., 2011; Schroeder et al., 2005; Singh et al., 2000; Stein et al., 2007; Ziegler et al., 2015). Overall, reduced HRV has been described in patients with IGT. This may suggest that there is a reduction in parasympathetic activity, which is similar to what has been described in early DAN and obesity. However, the findings of higher ratios between low-frequency to high-frequency spectral components of HRV suggest the presence of sympathovagal imbalance with sympathetic predominance (Perciaccante et al., 2006; Wu et al., 2007).

As previously mentioned, the majority of deficits in the CAN seen in prediabetes are cardiovagal. CAN often first manifests in the vagus nerve, the body’s longest parasympathetic autonomic nerve and the one responsible for almost three-quarters of parasympathetic activity (Pop-Busui, 2010). As in diabetic peripheral neuropathy, cardiac autonomic dysfunction in prediabetes tends to progress in a length dependent fashion. Thus, the vagus nerve is among the first to be affected in CAN and the earliest signs, such as resting tachycardia, are due to parasympathetic denervation and the corresponding increase in sympathetic tone.

Central obesity and hypertriglyceridemia appear to be important in the pathophysiology of autonomic dysfunction with IGT (Laitinen et al., 2011; Rezende et al., 1997). In the Finnish Diabetes Prevention Study cohort (Laitinen et al., 2011), subjects with parasympathetic dysfunction were older, had an increased weight, waist circumference, body mass index and had higher triglyceride concentration compared with those with normal parasympathetic function (P < 0.01 for all). Parasympathetic dysfunction was not significantly associated with other characteristics of metabolic syndrome; for example, high cholesterol, glucose and insulin levels or HbA1c. The E:I ratio and measures of obesity were significantly associated in the pooled population and in men but not in women. One example of a related condition that can contribute to autonomic dysfunction along with prediabetes is obstructive sleep apnea (Peltier et al., 2007) and is known to promote sympathetic hyperactivity.

At a more basic science level, several defects in organellar dysfunction have shown an association with neuropathy, though not specifically autonomic neuropathy. Endoplasmic reticulum stress results from abnormal folding of newly synthesized proteins and leads to the impairment of metabolism (Lupachyk et al., 2013). In animal models of obesity and impaired glucose tolerance, in the absence of overt T2DM, an increase in endoplasmic reticulum stress was associated with development of neuropathy and could be reduced by a chemical chaperone, trimethylamine oxide that blunted endoplasmic reticulum stress. Deficits in mitochondrial function and oxidative stress have been observed concomitant with the development of DN (Chandrasekaran et al., 2019). Importantly, alpha lipoic acid (ALA) has been shown to have an effect on treatment of autonomic neuropathy, although efficacy may be increased with concomitant exercise (Henriksen, 2006). The combination of exercise training and antioxidant treatment using ALA in an animal model of obesity-associated insulin resistance provides an interactive effect. This results in a greater improvement in insulin action on skeletal muscle glucose transport than either intervention individually (Henriksen, 2006). Specifically, the effect is due in part to improvements in IRS-1-dependent insulin signaling. The benefit of both ALA and exercise has also been shown for atherogenic cardiovascular risk in humans (McNeilly et al., 2011), supporting the concept that combination of therapies is more effective than a single therapy alone. Further evidence supports the use of ALA in improving endothelial function in human subjects with IGT in which endothelial function is impaired (Xiang et al., 2011). By improving endothelial function, potentially ALA could improve impaired vasoreactivity in human DAN. Despite evidence that antioxidants such as ALA may be effective in the experimental studies, there is insufficient evidence to propose use of ALA in DAN. However, ALA in combination with angiotensin-converting enzyme inhibitors showed an improvement in heart rate deep breathing in subjects with diabetic neuropathy (Ziegler et al., 2016). Thus, optimal control of cardiovascular disease risk factors could contribute to improved efficacy of ALA in patients with higher disease burden. It is more speculative if other therapies that regulate mitochondrial function, for example nicotinamide riboside, or oxidative stress would find a role in the therapy of DAN (Chandrasekaran et al., 2019).

11. Impaired glucose tolerance and sudden cardiac death

Of considerable concern is the risk of sudden cardiac death and other cardiac complications with IGT. By some estimates, up to 60% of subjects with established cardiovascular disease have evidence of IGT and insulin resistance (Nesto, 2004). Individuals exhibiting precursor symptoms of diabetes mellitus or reaching diagnostic thresholds for diabetes are at increased risk of death or complications due to cardiovascular disease (Dinh et al., 2011). The same metabolic defects that underlie IGT and proinflammatory and prothrombotic states, lead to endothelial dysfunction and accelerate atherogenesis. Moreover, increases in sympathetic tone with IGT are associated with changes in cardiac and vascular function that lead to hypertension, left ventricular dysfunction, and CAN. A combination of these events may lead to arrhythmia, silent infarction, and sudden death. The short-term QT interval variability is a non-invasive method for assessment of proarrhythmic risk and may act as an early indicator of increased instability of cardiac repolarization during prediabetic conditions (Orosz et al., 2017). There is evidence that short-term QT interval variability was significantly higher in IGT subjects (Orosz et al., 2017). Furthermore, the changes in metabolic and autonomic function, coupled with increased inflammatory and thrombotic signaling, compromise the ability of myocardial and vascular tissue to remodel after injury. As these events occur early with IGT, early intervention is critical as discussed in the following sections. Thus, intervention would occur while the process remains reversible.

12. Improvement in glycemic control on autonomic neuropathy

There have been no specific interventions tested only in subjects with IGT. Most studies have addressed autonomic dysfunction in subjects with diabetes. In other instances, subjects with IGT may have been included as part of the treatment cohort. As there is a metabolic continuum between IGT and T2DM, it may be possible to extrapolate between the results for T2DM and autonomic neuropathy to IGT with autonomic dysfunction. Intensive glucose control has been shown to prevent or delay the development of CAN in T1DM (Group, 1998; Pop-Busui et al., 2009). The beneficial effects of glucose control on CAN in T2DM are not as clear. The Steno 2 study used a multifactorial medication and lifestyle intervention to treat hyperglycemia, hypertension, dyslipidemia and microalbuminuria in individuals with T2DM and were able to reduce the risk of CAN in T2DM by 60% compared to conventional therapy (Gaede et al., 2003; Pop-Busui et al., 2017).

13. Lifestyle interventions and diabetic neuropathy

While there are several studies that have examined the effect of dietary and exercise interventions on small fiber neuropathy, and specifically the intraepidermal nerve fiber density, there is evidence that lifestyle interventions may be effective in subjects with IGT and autonomic neuropathy. In one study with a majority of subjects with IGT (Zilliox and Russell, 2019), even “Standard of care recommendations” produced an improvement in certain measures of autonomic function. “Standard of care recommendations” were provided prospectively to participants over a one-year period. These general recommendations were that the patient lose 7% of body weight, sustain this loss, and undertake aerobic exercise for 150 min/week, up to 30 min per session. Participants were encouraged to follow the intervention but had to lose weight and exercise by themselves without the assistance of an interventionist. The protocol was performed prospectively, was masked, and standardized but none of the recommendations were enforced or monitored. The technician performing the outcome measures was masked to the patient intervention. There was an improvement in the E:I ratio, a sensitive measure of cardiovagal autonomic function in 50 subjects after 1 year (P = 0.042). The baseline 95% CI was 1.14, 1.22, and at 1 year was 1.20, 1.44. There was also a decrease in the Total Impact Score (TIS) on the SAS (n = 71) (Zilliox and Russell, 2019). The decrease in the TIS reflects the overall severity of reported autonomic symptoms and in general the SAS correlates closely with the IENFD (Russell and Zilliox, 2014; Zilliox et al., 2011).

The Diabetes Prevention Program (DPP) was a randomized, controlled clinical trial conducted at 27 clinical centers around the United States from 1996 to 2001 (Anon, 2000; Carnethon et al., 2006). The trial enrolled 3234 participants; 55% were Caucasian, and 45% were from minority groups at high risk for T2DM, including African American, Alaska Native, American Indian, Asian American, Hispanic/Latino, or Pacific Islander, people ages 60 and older, women with a history of gestational diabetes, and people with a parent, brother, sister, or child who had type 2 diabetes. DPP participants were randomly assigned to one of the following groups: (1) Lifestyle Change Group - Group participants joined a DPP Lifestyle Change Program that provided intensive training and in which they tried to lose 7% of their body weight, maintain that weight loss, and exercising 150 min per week. (2) Metformin Group - Group participants took 850 mg of metformin twice a day and were provided standard advice about diet and physical activity. (3) Placebo Group - Group participants took a placebo twice a day instead of metformin and were provided standard advice about diet and physical activity. Participants had impaired glucose tolerance (World Health Organization criteria plus fasting plasma glucose level > or = 5.3 mmol/l [> or = 95 mg/dl]) and were followed for a mean of 3.2 years after random assignment to intensive lifestyle intervention, metformin therapy, or placebo. In 2980 DPP participants, 12-lead electrocardiograms were measured at baseline and annually (Carnethon et al., 2006). Heart rate, HRV, and QT duration were used to estimate fitness and autonomic nervous system function. Heart rate and QT indexes decreased, and HRV increased over time in the intervention group. In the other arms, the magnitude of decline in heart rate and QT duration were smaller, whereas HRV did not increase. Baseline heart rate was the only index significantly (P < 0.05) associated with diabetes after adjustment for demographics and weight change (hazard ratio for lifestyle and metformin arms = 1.19 and 1.17 per 10.6 beats/min, respectively). A decrease in heart rate and QT index or an increase in HRV were associated with a lower risk of developing diabetes and remained significant after accounting for change in weight and physical activity. Thus, the lifestyle intervention overall was more effective than the metformin intervention arm and improvement in indices associated with autonomic neuropathy showed a correlation with reduced risk of developing diabetes.

Studies of patients with existing diabetic neuropathy and IGT have also shown that diet and exercise interventions can improve measures of neuropathy. The Impaired Glucose Tolerance Neuropathy Trial study was a 12-month natural history trial to examine the effects of a lifestyle modification on measures of diabetic neuropathy (Smith et al., 2006). Subjects with impaired glucose tolerance and pre-existing neuropathy underwent a diet and exercise program that was similar to that used in the Diabetes Prevention Program. After 1 year, there were significant improvements in the foot sweat volume measured by QSART that assesses the sweat evoked response in the foot to acetylcholine. This finding is important because in small fiber neuropathy associated with IGT, there is a decrease in sweat generation in the foot that parallels the loss of distal intraepidermal nerve fiber density (Zilliox et al., 2011) and sweat gland innervation density. The QSART is a measure of small fiber controlled sweat generation (Peltier et al., 2009; Russell and Zilliox, 2014; Zilliox and Russell, 2011a). Thus, the exercise and weight reduction intervention improved sudomotor function corresponding to an implied improvement in sweat gland innervation, although this was not directly measured. Improvement in sudomotor function in the foot was also associated with significant improvements in weight, glucose tolerance, and lipid profile (Smith et al., 2006).

The prevention and treatment of CAN should include effective lifestyle interventions with weight loss, physical activity, smoking cessation, dietary changes, glycemic control, control of cardiovascular risk factors and education. The Toronto Consensus statement concluded that a lifestyle intervention may improve HRV in pre-diabetes and diabetes and recommended that it should be offered as a basic preventive measure (Spallone et al., 2011) and the American Diabetes Association recommends considering lifestyle modifications to improve CAN in patients with pre-diabetes (Pop-Busui et al., 2017).

14. Conclusion

CAN remains underdiagnosed in IGT, although the available evidence suggests that symptoms of autonomic dysfunction are frequent in subjects with IGT. There is also concomitant evidence that autonomic testing shows both CAN and sudomotor dysfunction. It is important to recognize both that the subject has IGT and that the symptoms relate to autonomic dysfunction. This allows early intervention to improve metabolic control before the subject converts to T2DM and the autonomic neuropathy becomes irreversible. The risk factors for IGT induced autonomic neuropathy are similar to those associated with T2DM. New data support the efficacy of CAN prevention at an early stage. Lifestyle interventions incorporating tailored diet and exercise interventions offer of hope of improving autonomic neuropathy when offered early in the course of IGT. However, CAN still needs more effective prevention and disease modifying treatment.

Abbreviations:

IGT

impaired glucose tolerance

T2DM

type 2 diabetes mellitus

T1DM

type 1 diabetes mellitus

NGT

normal glucose tolerance

IFG

impaired fasting glucose

NFG

normal fasting glucose

OGTT

oral glucose tolerance test

DAN

diabetic autonomic neuropathy

CAN

cardiac autonomic neuropathy

HRV

heart rate variability

SDNN

standard deviation of NN (or normal RR) intervals

BRS

baroreflex sensitivity measures

SAS

Survey of Autonomic Symptoms

ASS

Autonomic Symptom Survey

TIS

Total Impact Score

DPP

Diabetes Prevention Program

ALA

alpha lipoic acid

ASP

Autonomic Symptom Profile

CASS

Composite Autonomic Scoring Scale

E:I

expiration: inspiration ratio

HR

heart rate

NS

not significant

QSART

quantitative sudomotor axon reflex test