Introduction

Impaired insulin sensitivity and glucose tolerance have been reported in α-synucleinopathies such as Parkinson disease. The prevalence of impaired glucose tolerance ranges from 50 to 80 % in this population. Studies using national survey data found higher rates of type 2 diabetes mellitus in adults with Parkinson disease compared with those without this condition [6, 21]. Neurodegenerative conditions and states of insulin resistance such as obesity shared common pathophysiological pathways. Both have induced pro-inflammatory cytokines and mitochondrial dysfunction, and have documented impaired signaling to a hormone very similar to insulin, the insulin growth factor 1 (IGF-1), which is a neurotropic peptide involved in brain development, maturation, function, and neuroprotection [27].

Multiple system atrophy (MSA) is another progressive neurodegenerative movement disorder. It is characterized by autonomic failure, neurogenic orthostatic hypotension, parkinsonism, cerebellar, and urological dysfunction [9]. In MSA patients, pathological α-synuclein is distributed to glial cells in the basal ganglia and cortex. These patients have impaired IGF-1 signaling in the brain, which is associated with disease progression [17]. It is unknown, however, whether MSA patients have reduced insulin sensitivity. Similarly, there has been a paucity of metabolic data in patients with pure autonomic failure (PAF), a condition characterized by α-synuclein deposits in peripheral autonomic nerves. Therefore, the purpose of this study was to test the hypothesis that insulin sensitivity is impaired in MSA and PAF patients compared with healthy subjects. As a secondary objective, we determined whether residual autonomic activity and the presence of supine hypertension contributed to insulin sensitivity in these patients.

Methods

Study participants

We studied 19 patients with probable MSA, 26 patients with PAF, and 8 healthy controls of similar age. MSA and PAF patients were recruited from the Autonomic Dysfunction Center at Vanderbilt University Medical Center. All patients had severe autonomic failure defined as an absence of autonomic reflexes per standardized autonomic function tests [4]. Exclusion criteria included a history of diabetes mellitus according to the American Diabetes Association guidelines [2]. Healthy controls were recruited from the Vanderbilt community, were nonsmokers, and were excluded if pregnant, had evidence of systemic illness, or were taking medications known to interfere with the regulation of blood pressure or blood volume. Patients and healthy controls were placed on a low-monoamine, caffeine-free diet containing 150 mEq sodium and 70 mEq potassium per day, for at least 3 days before evaluation. The Vanderbilt Institutional Review Board approved this study, and all participants provided written informed consent.

Screening

MSA and PAF patients were admitted to the Clinical Research Center at Vanderbilt University. Medications known to stimulate the autonomic nervous system (i.e., midodrine), affect blood pressure (i.e., anti-hypertensives), or affect blood volume (e.g., fludrocortisone) were discontinued for at least 5 half-lives before admission. Only three patients in this cohort (2 MSA and 1 PAF) were on levodopa–carbidopa. Although this medication has been shown to negatively affect metabolic parameters, it was not discontinued because of safety concerns.

Healthy controls were studied on an outpatient basis. They were recruited from the community through massive email advertisement. All subjects were nonsmokers, did not have any evidence of systemic illness, and were not taking medications known to interfere with the regulation of blood pressure and/or metabolism.

A full medical history, physical examination, and standardized autonomic function tests were conducted for all subjects. The autonomic function tests were performed to evaluate the integrity of autonomic reflex arcs and to confirm the diagnosis of autonomic failure using the 2010 American Autonomic Society criteria [7]. Patients with MSA were diagnosed based on Gilman’s criteria [9]. Patients were diagnosed with supine hypertension if the average supine blood pressure obtained during 12-h blood pressure monitoring was equal to or above 150/90 mmHg as previously defined [12].

Blood analyses

Plasma glucose was measured with a glucose analyzer (YSI Life Sciences, Yellow Springs, OH, USA). Plasma insulin concentrations were determined by radioimmunoassay (Millipore, St. Charles, MO, USA). Plasma norepinephrine (NE) was determined by high-performance liquid chromatography with electrochemical detection [10].

Measurement of insulin sensitivity

The Homeostasis Model Assessment (HOMA2) was used to determine insulin sensitivity. This measurement takes into account variations in hepatic and peripheral glucose resistance, increases in the insulin secretion curve for plasma glucose concentrations above 180 mg/dL, and the contribution of circulating proinsulin [15]. The model estimates steady-state beta cell function (%B) and insulin sensitivity (%S), as percentages of a normal reference population, using fasting plasma glucose and insulin values [19]. We used a HOMA2 calculator available online (https://www.dtu.ox.ac.uk/homacalculator/).

Spectral analysis of heart rate and blood pressure variability

Continuous blood pressure was measured using a NEXFIN device (BMEYE, Amsterdam, The Netherlands) and continuous ECG was measured using the VITAL-GUARD 450c monitor (Ivy Biomedical Systems, Brandford, CT, USA). Data was obtained while the patient was resting in the supine position and in a quiet environment for at least 5 min. Physiological data were recorded using the WINDAQ data acquisition system (DI220, DATAQ, Akron, OH, 14 Bit, 1000 Hz), and were processed off-line using a custom written software in PV-Wave language (PV-wave, Visual Numerics Inc., Houston, TX, USA), developed by one of the authors (AD). Detected beat-to-beat values of R–R intervals (RRI) and blood pressure were interpolated and low pass filtered (cutoff 2 Hz). Data segments of at least 180 s were used for spectral analysis. Linear trends were removed and power spectral density was estimated with the FFT-based Welch algorithm. The total power (TP) and the power in the low (LF: 0.04 to <0.15 Hz) and high (HF: 0.15 to <0.40 Hz)-frequency ranges were calculated according to the task force recommendations [1].

Measurements of baroreflex sensitivity

Cross spectra, coherence, and transfer function analysis were used to capture inter-relationships between R–R interval and systolic blood pressure (SBP). The baroreflex gain was determined as the mean magnitude value of the transfer function in the low-frequency band, with a negative phase and squared coherence value greater than 0.5.

Statistical analysis

The differences in patient’s demographic and clinical variables between groups (MSA, PAF, and controls) were assessed based on descriptive statistics such as medians and quantiles for continuous variables and proportions for categorical variables. For continuous outcomes, we used non-parametric Kruskal–Wallis tests to determine differences among groups and Mann–Whitney U test for pairwise comparisons. Multiple linear regression was used to test for overall linear relationships between the dependent variable (insulin sensitivity) and independent variables (age, presence of supine hypertension, and sympathetic and parasympathetic activity), and to assess the significance of these relationships after adjustments for each covariate. Residual analysis was performed to assess the adequacy of the model fit. Logarithmic transformations of insulin sensitivity, sympathetic activity (LFSBP), and parasympathetic activity (HFRRI) were used to normalize the data and improve model fit. LFSBP and HFRRI were measured in mmHg2 and ms2, respectively. All tests were two-tailed, and a P value of <0.05 was considered significant. Analyses were performed using SPSS for Windows (version 22.0; SPSS Inc., Chicago, IL, USA).

Results

Patient’s characteristics

The clinical characteristics of study participants are shown in Table 1. Weight, height, body mass index (BMI), and fasting glucose levels were similar among groups. Age, fasting insulin, and supine and upright NE were significantly different among groups, Table 1. Pairwise comparison showed that patients with MSA were 10 years younger compared to PAF patients (0.012), and had elevated fasting insulin levels compared to healthy controls (P = 0.034). Both MSA and PAF patients had lower supine NE levels compared to healthy controls (P = 0.008; P < 0.001, respectively). Similarly, upright NE was lower in MSA and PAF patients compared to healthy controls (P < 0.001; P < 0.001, respectively). As expected, PAF patients had lower supine and upright plasma NE compared with MSA (P = 0.004; P = 0.049, respectively).

Table 1 Clinical characteristics
Full size table

None of our patients had renal failure. However, there were significant differences in BUN and creatinine levels among groups (P < 0.001; P < 0.001, respectively). PAF patients had elevated creatinine levels compared with healthy controls, 1.12 mg/dL (IQR, 0.90 to 1.23) versus 0.73 mg/dL (IQR, 0.65 to 0.83), P < 0.001. Similarly, MSA patients had elevated creatinine levels compared with healthy controls, 0.96 mg/dL (IQR, 0.90 to 1.18) versus 0.73 mg/dL (IQR, 0.65 to 0.83), P = 0.005. There were no differences in creatinine levels between PAF and MSA (P = 0.174).

Cardiovascular and autonomic function

Patients with MSA and PAF had higher supine SBP compared with healthy controls (P = 0.001); 47 % of MSA and 70 % of PAF patients met the definition of supine hypertension (average SBP/DBP ≥150/90 mmHg) during an overnight blood pressure monitoring. The results of the autonomic function tests are presented in Table 2. MSA and PAF patients had a profound decrease in SBP upon standing, −83 mmHg (IQR, −92 to −42) and −79 mmHg (IQR, −90 to −58). These patients had a greater decrease in systolic blood pressure (SBP) during phase II of the Valsalva maneuver compared to healthy controls, and the SBP overshoot during phase IV was absent; the pressor responses to isometric handgrip exercise or pain stimulus (cold pressor test) were impaired, and sinus arrhythmia (S/A) was markedly reduced indicating impaired parasympathetic responses. Thus, autonomic testing indicated severe sympathetic and parasympathetic involvement in autonomic failure patients.

Table 2 Autonomic function tests
Full size table

The results for spectral analysis of blood pressure and heart rate variability are shown in Table 3. Indices of sympathetic activity (LFSBP, mmHg2) and parasympathetic activity (HFRRI, ms2) were different among groups. MSA and PAF patients had lower sympathetic and parasympathetic activity compared to healthy controls, Table 3.

Table 3 Spectral analysis of blood pressure and heart rate
Full size table

Insulin sensitivity and beta cell function measurements

There was a statistically significant difference in insulin sensitivity measured by the HOMA 2 index among groups (P = 0.048). Patients with MSA have reduced insulin sensitivity compared to healthy controls, 64 % (IQR, 43 to 117) versus 139 % (IQR, 83 to 212), P = 0.032, respectively. However, there were no differences in insulin sensitivity between MSA versus PAF: 64 % (IQR, 43 to 117) versus 88 % (IQR, 67 to 148), P = 0.103, respectively; or between PAF versus healthy controls: 88 % (IQR, 67 to 148) versus 139 % (IQR, 83 to 212), P = 0.150, respectively, Fig. 1. Beta cell function was similar among groups (P = 0.103).

Fig. 1
figure 1

Difference in insulin sensitivity between patients with multiple system atrophy (MSA), pure autonomic failure (PAF), and healthy controls

Full size image

The relationship between beta cell function and insulin sensitivity among different groups of patients (MSA, PAF, and controls) is presented in Fig. 2. For comparison, we also included insulin sensitivity and beta cell function data obtained in 115 moderately obese subjects with a median BMI of 36.1 IQR (33.3 to 40.7) kg/m2. Even though patients with MSA were lean, their reduced insulin sensitivity was comparable to obese subjects. In MSA and obese subjects, beta cell function was elevated to compensate for the decrease in insulin sensitivity.

Fig. 2
figure 2

Relationship between insulin sensitivity % and beta cell function % in patients with multiple system atrophy (MSA), pure autonomic failure (PAF), moderately obese subjects, and healthy controls. Graphic represents mean and 95 % CI

Full size image

Effect of residual autonomic activity and other variables on insulin sensitivity

We assessed the relationship between the logarithm of insulin sensitivity and age, the presence of supine hypertension, and sympathetic and parasympathetic activity [as measured by log (LFSBP) and log (HFRRI), respectively]. Insulin sensitivity was positively correlated with age (r = 0.248, P = 0.043) and parasympathetic activity [log HFRRI (r = 0.311, P = 0.015)], but negatively correlated with the presence of supine hypertension (r = −0.260, P = 0.036) and sympathetic activity [log LF SBP, (r = −0.177, P = 0.112)], Table 4. Adjusted R 2 for this model was 0.32, F (4, 44) = 6.253, P < 0.001, indicating that approximately 32 % of the variability in insulin sensitivity was explained by our model. As shown in Table 4, two predictors suppressed insulin sensitivity: sympathetic activity (log LF SBP) and the presence of supine hypertension (yes = 1; no = 0).

Table 4 Predicting insulin sensitivity (N = 44)
Full size table

Our model indicated a 24 % increase in insulin sensitivity per one-year increase in age, a 43 % reduction in insulin sensitivity if the patient had supine hypertension, a 17 % reduction in insulin sensitivity for 1 % increase in LFSBP (sympathetic activity), and a 13 % improvement in insulin sensitivity for 1 % increase in HFRRI (parasympathetic activity).

Discussion

The main new finding of this study was that patients with MSA, a form of primary autonomic failure characterized by central autonomic compromise with preserved peripheral noradrenergic nerves, have reduced insulin sensitivity compared with healthy controls. Furthermore, the reduced insulin sensitivity observed in lean MSA is comparable to subjects with moderate obesity. In contrast, no differences were found in insulin sensitivity in patients with peripheral autonomic impairment or PAF compared with healthy controls. The factor that significantly contributed to the decrease in insulin sensitivity in MSA was the presence of residual sympathetic activity, as measured by low-frequency variability of systolic blood pressure. Impaired insulin sensitivity in MSA was also associated with the presence of supine hypertension, which we have previously shown is also due to residual sympathetic activity [23]. On the other hand, parasympathetic activity, as assessed by high-frequency heart rate variability and age, had a favorable effect on insulin sensitivity.

Several explanations may account for the reduced insulin sensitivity in MSA patients; it may be possible that the reduced insulin sensitivity observed in MSA is related to a decrease in physical activity, a sedentary lifestyle, or the severity of the disease. It is well known that patients with autonomic failure become hypotensive on standing [3]. and during exercise [26]. Furthermore, patients with MSA developed pyramidal and extrapyramidal symptoms with severe rigidity and decreased postural reflexes. An adopted protective behavior against fainting or falls has been to decrease physical activity. Of note, in a previous study [24], we reported a significant (82 % reduction) in physical activity in autonomic failure patients, as measured by an actigraphy monitor, compared with age-matched normal volunteers.

It is well known that the autonomic nervous system plays a significant role in the regulation of glucose homeostasis. Metabolically, active tissues such as adipose, muscle, liver, and pancreas are innervated by parasympathetic and sympathetic fibers [16]. Previous studies have shown that acute stimulation of the sympathetic nervous system with lower body negative pressure reduces insulin sensitivity [11]. Conversely, acute and chronic sympathetic inhibition has been shown to improve insulin sensitivity [8, 13]. It may possible that the unrestrained residual sympathetic activity in MSA contributes to reduced insulin sensitivity. We previously reported that the residual sympathetic activity in MSA still modulates some metabolic function such as energy expenditure, [24] autonomic/endocrine circadian rhythm [20], and blood pressure [23]. The residual sympathetic activity could negatively influence insulin sensitivity by decreasing glucose delivery to target tissue through a blood flow mechanism. Indeed, in our cohort, the presence of supine hypertension, [23] induced by unrestrained residual sympathetic tone in MSA, and measurements of sympathetic activity (LFsbp) significantly predicted the decrease in insulin sensitivity.

In our cohort, residual parasympathetic activity had a favorable effect on insulin sensitivity. It has been reported that parasympathetic activity is reduced in chronic inflammatory conditions associated with impaired insulin sensitivity such as obesity. Weight loss resulting from increased physical activity, dieting or bariatric surgery reduces inflammation, ameliorates the metabolic complications, and improves heart rate variability (a measurement of parasympathetic tone) [14, 17].

It is noteworthy that our results showed that aging had a favorable effect on insulin sensitivity. This is unexpected considering previous studies that reported an inverse association between age and insulin sensitivity [5, 25]. In our cohort, MSA patients were 10 years younger than PAF patients; this age imbalance may contribute to this observation.

Our findings raised several questions. For instance, in patients with Parkinson disease, concurrent diabetes mellitus can accelerate progression of both motor and cognitive symptoms [22]. Current randomized clinical trials with drugs approved for the treatment of type II diabetes mellitus (pioglitazone and exenatide) are being conducted to determine their neuroprotective effects. If proven successful, similar trials could be conducted in MSA, a disease characterized by rapidly progressive neurodegeneration.

Alternatively, because of the association between increased parasympathetic activity and insulin sensitivity, one could postulate that interventions that increase parasympathetic tone, such as central acetylcholinesterase inhibitors and/or vagal stimulators, may have a beneficial effect on insulin sensitivity and may possibly provide neuroprotection in MSA patients.

Limitations

Our study has several limitations. In our study, MSA patients were 10 years younger compared with PAF patients and healthy controls, and groups were not matched by gender. We did not measure body composition. In a previous publication, we did not observe a significant difference in body composition between autonomic failure patients and healthy controls [24]. The gold standard for the evaluation of insulin sensitivity is the hyperinsulinemic euglycemic clamp. This technique, however, cannot be safely applied in patients with autonomic failure because of the risk of hypotension during insulin infusion [18]. Finally, PAF patients have a tendency to reduced insulin sensitivity. The lack of a significant difference may be explained by the small sample size. PAF, however, is a rare disease, with unknown prevalence in the general population.