Impairment in Baroreflex Sensitivity in Recent-Onset Type 2 Diabetes Without Progression Over 5 Years

Cardiovascular autonomic neuropathy (CAN) assessed by heart rate variability (HRV) affects ∼20% of individuals with diabetes (1), 12–17% of patients with newly diagnosed type 2 diabetes (2,3), and 6–11% of those with prediabetes (2) and carries an increased risk of cardiovascular mortality (4–6). Clinical consequences, such as resting tachycardia, orthostatic hypotension, exercise intolerance, intra- and perioperative hemodynamic instability, and a higher risk of silent myocardial ischemia, contribute to substantial morbidity of patients with CAN (1,6). Established risk factors for CAN include poor glycemic control in type 1 diabetes and a combination of hypertension, dyslipidemia, obesity, and poor glycemic control in type 2 diabetes (1). In patients with recent-onset diabetes from the German Diabetes Study (GDS) baseline cohort, we demonstrated associations of lower vagus-mediated HRV with insulin resistance (7) and reduced cardiorespiratory fitness in both diabetes types (8) as well as hepatic steatosis (9) in type 2 diabetes, suggesting that these factors may contribute to the early development of CAN (10).

The baroreflex system plays an important role in regulating short-term fluctuations of arterial blood pressure (BP). Arterial baroreceptors in the carotid sinuses and aortic arch sense changes in BP and modulate efferent autonomic neural activity to the central nervous system accordingly. A rise in sensed BP leads to an increase and decrease in the discharge of vagal and sympathetic neurons, respectively, resulting in decreased heart rate, cardiac contractility, and peripheral vascular resistance, whereas a fall in BP causes the opposite effects (11). Traditionally, baroreflex sensitivity (BRS) has been assessed using manipulations such as injections of vasoactive drugs, Valsalva maneuver, and the neck chamber technique (11). More recently, noninvasive BRS assessment has allowed for examining larger cohorts by the relatively simple pulse photoplethysmography technique using simultaneous recordings of spontaneous beat-to-beat fluctuations of BP and heart rate, which can be analyzed in the time domain and frequency domain (11–15).

Impairment of baroreflex mechanisms resulting in a chronic adrenergic activation often accompanies cardiovascular disease. A blunted baroreflex gain is predictive of increased cardiovascular risk in patients with myocardial infarction and heart failure (11,16). Evidence has also emerged suggesting that BRS could be a marker of CAN in patients with longer-term diabetes (17–22). However, the temporal sequence of BRS early in the course of both type 1 and type 2 diabetes remains unclear. We therefore sought to determine 1) in a cross-sectional analysis whether various time domain and frequency domain measures of spontaneous BRS are altered in individuals with recent-onset type 1 and type 2 diabetes compared with control groups with normal glucose tolerance and associated with metabolic parameters, such as insulin sensitivity or insulin secretion, and 2) in a prospective setting the progress of BRS over the next 5 years.

Patients recently diagnosed with diabetes (known diabetes duration ≤1 year) and glucose-tolerant control subjects were recruited consecutively from the baseline cohort of the GDS, a prospective observational study investigating the natural course of metabolic alterations and the development of chronic diabetic complications (ClinicalTrial.gov reg. no. NCT01055093). The study was approved by the local ethics committee of Heinrich Heine University, Düsseldorf, Germany, and informed written consent was obtained from all participants before participation. The study design and cohort profile of the GDS were described in detail previously (23). The present cross-sectional analysis included 206 consecutive participants with type 1 diabetes, 381 consecutive participants with type 2 diabetes, and two corresponding age- and sex-matched control groups with glucose-tolerant individuals (control 1, n = 65; control 2, n = 83). The prospective analysis included 84 individuals with type 1 diabetes and 137 participants with type 2 diabetes who were monitored for 5 years.

Continuous plethysmographic arterial measurements of spontaneous changes in systolic BP and R-R intervals (R-Rs), measured on the middle finger, were recorded using a Finometer MIDI device (Finapres Medical Systems, Enschede, the Netherlands) over 10 min in the supine position during spontaneous breathing. BRS parameters were calculated using the BeatScope Easy software (Nevrokard BRS Analysis v6.3.0.; Nevrokard, Izola, Slovenia). The sequence method was used to calculate time domain BRS indices for positive, negative, and all sequences to obtain BRS(+) slope, BRS(−) slope, and BRS sequence all, respectively. Spontaneously occurring parallel trends of three or more R-Rs and systolic BPs were determined, while a change of 1 mmHg for consecutive systolic BPs and 5 ms for consecutive R-Rs were required for a shift (24). Spectral analysis (autoregressive method) was used to determine frequency domain BRS parameters in the low-frequency (LF) (0.04–0.15 Hz) and high-frequency (HF) (0.15–0.4 Hz) bands by computing the square root of the ratio between R-R and systolic BP spectral components (α coefficient), i.e., BRS-αLF, BRS-αHF, and BRS-αMean (mean of BRS-αLF and BRS-αHF) (25). Because the correct interpretation of the α coefficient requires a high coherence for the phase link between R-R and BP variability signals (25), the α coefficient was computed only if coherence (K²) was >0.5 (25), leaving the following group samples: control 1, n = 55; control 2, n = 67; type 1 diabetes, n = 146; and type 2 diabetes, n = 223. Cross-correlation BRS (xBRS) was computed by cross-correlation and regression between systolic BP and R-Rs over 10-s sliding windows, a time span sufficient to fully accommodate a 10-s variability in rhythm or several cycles at ventilatory frequencies. The correlation coefficient was calculated six times per window. The highest value of all calculated cross-correlations was selected, and the corresponding regression slope was taken to determine xBRS, provided that it was positive and its probability of being a random regression was <0.1% (26). The BRS-to-SD ratio was used as a simplified index representing the ratio of the SD of R-Rs and the SD of systolic BP, as previously described (27).

R-Rs were measured in the supine position during a hyperinsulinemic-euglycemic clamp over 3 h using a digital Spider View Holter recorder with seven electrodes to record three-channel electrocardiograms (Sorin Group, Munich, Germany), as previously described (7). Time domain HRV measures included the SD of differences between adjacent normal-to-normal (NN) intervals, SD of NN averages over 5 min (SDANN), SD of all NN intervals, the number of pairs of adjacent NN intervals differing by >50 ms in the entire recording divided by the total number of NN intervals, and the root mean square of successive differences. Frequency domain HRV indices included the very LF band (0.003–0.04 Hz), LF band (0.04–0.15 Hz), and HF band (0.15–0.4 Hz) and the LF-to-HF ratio.

Cardiovascular autonomic function tests were performed during spontaneous breathing over 5 min (coefficient of R-R variation, spectral analysis), at deep breathing (expiration-to-inspiration ratio), after standing up (maximum-to-minimum 30:15 ratio), and in response to a Valsalva maneuver (Valsalva ratio) using a VariaCardio TF5 system (MIE Ltd, Leeds, U.K.), as previously described (28). Age- and sex-dependent lower limits of normal for HRV tests were defined at the fifth percentile obtained from 218 (142 male, 76 female) glucose-tolerant individuals. Orthostatic hypotension was defined as a decrease in systolic or diastolic BP within 3 min after standing up of >28 or >15 mmHg in men and >27 or >11 mmHg in women, respectively. This definition differs from a commonly used one of ≥20 or ≥10 mmHg for systolic or diastolic BP, respectively (1). Borderline CAN was assumed if the results of two of seven indices were abnormal, whereas definite CAN was diagnosed if three or more of seven indices were abnormal (28).

Participants were examined in the supine position in the morning after an overnight fast, with arms and legs abducted from their body. BIANOSTIC-AT double-size electrodes (Data Input, Pöcking, Germany) were fixed on the dorsum of the hand and foot of the dominant side of the body (29). Fat-free mass and fat mass were measured using Nutriguard-S (Data Input, Darmstadt, Germany) by determining resistance and reactance (30).

Cardiorespiratory fitness was assessed as previously described (8). In brief, all participants underwent an incremental exhaustive exercise test on an electronically braked cycle ergometer (Ergometrics 900; Ergoline, Bitz, Germany) at 60 rpm. Respiratory gas exchange measurements were determined by open-air spirometry (Masterscreen CPX; Jaeger/VIASYS, Hoechberg, Germany). Work rate was increased by 10 W/min, and exhaustion was reached on average after 12–15 min. Heart rate and a 12-lead electrocardiogram were recorded continuously, while arm BP was recorded every 2 min during the test. Cardiorespiratory fitness parameters included VO_2max, VO₂ at anaerobic threshold, VO₂ at the respiratory compensation point, and VCO_2max.

A modified Botnia clamp was performed, consisting of an intravenous glucose tolerance test, followed by a hyperinsulinemic-euglycemic clamp test with frequent measurements of blood glucose, C-peptide, and insulin to determine whole-body insulin sensitivity (M-value) (mg glucose ∗ (body weight in kg)⁻¹ ∗ min⁻¹), which was calculated from the difference between mean glucose infusion rates during steady state in the last 30 min of the clamp with glucose space correction. First-phase C-peptide secretion was calculated as the incremental area under the curve (iAUC) until 10 min (iAUC 0–10 min), second-phase C-peptide secretion as the iAUC between 10 and 60 min (iAUC 10–60 min), and the total C-peptide secretion as the sum of both (iAUC 0–60 min) (23).

Blood samples were taken before injecting 1 mg glucagon (GlucaGen; Novo Nordisk, Mainz, Germany) intravenously and 6 min thereafter. Glucagon-stimulated insulin secretion was determined as the difference between the C-peptide concentrations at 6 min and 0 min (∆C-peptide) (23).

Plasma glucose, cholesterol (total, HDLs, LDLs), serum triglycerides, and creatinine were measured on a Hitachi 912 analyzer (Roche Diagnostics, Mannheim, Germany) as previously described (23). Estimated glomerular filtration rate was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation (31).

Data are presented as mean ± SD, median (first and third quartiles), or percentages. Categorical variables were compared using the χ² test and expressed as percentages of participants. Continuous data were assessed using the parametric t test or nonparametric Mann-Whitney U test for cross-sectional and the Wilcoxon test for prospective data. Correlations between two variables were determined using Spearman rank correlation for nonparametric data and Pearson correlation analyses for parametric data. For multiple linear regression analyses, dependent variables with skewed distribution were ln-transformed before analyses. The analyses were adjusted for sex, age, BMI, smoking, hypertension, fat mass, and VO_2max. All statistical tests were two-sided, and the level of significance was set at α = 0.05. All analyses were performed using SPSS 22.0 software (IBM Corp, Armonk, NY).

The data that support the findings of this study are available from the GDS, but restrictions apply to the availability of these data, which were used under license for the current study and therefore are not publicly available. Data are, however, available from the authors upon reasonable request and with permission of the GDS.

The demographic and clinical characteristics of the baseline cohort are given in Table 1. Compared with control 1 subjects, individuals with type 1 diabetes had higher HbA_1c and fasting glucose levels, while BMI, fat-free mass, M-value, fasting and stimulated C-peptide, ΔC-peptide, iAUC 0–10 min, iAUC 10–60 min, iAUC 0–60 min, and VO_2max were lower (all P < 0.05). Compared with control 2 subjects, participants with type 2 diabetes showed higher BMI, waist circumference, fat mass, heart rate, systolic BP, HbA_1c, fasting glucose, fasting C-peptide, triglycerides, hs-CRP, and albuminuria and were more frequently treated with antihypertensive agents, while HDL cholesterol, M-value, ΔC-peptide, iAUC 0–10 min, and VO_2max were lower (all P < 0.05). No differences between the corresponding diabetes and control groups were noted for the remaining variables.

The median levels of the time domain and frequency domain HRV indices in the four groups at baseline are reported in Table 2. After adjustment for age, sex, BMI, smoking, hypertension, fat mass, and VO_2max, SDANN and SD were lower in both diabetes groups compared with the corresponding control groups (all P < 0.05). No differences between the groups were noted for the remaining seven HRV indices.

The median levels of the BRS indices obtained in the four groups at baseline are presented in Table 3. No differences were observed between participants with type 1 diabetes compared with control 1 subjects. In the group with type 2 diabetes, all three spectral analysis indices (BRS-αLF, BRS-αHF, and BRS-αMean) were lower compared with control 2 subjects after adjustment for age, sex, BMI, smoking, hypertension, fat mass, and VO_2max (all P < 0.05). Further adjustment for definite CAN did not alter these results. Table 4 reports the full multiple linear regression models including CAN for the three frequency domain BRS indices in control 2 participants and patients with type 2 diabetes at baseline. No differences between the group with type 2 diabetes and control 2 were found for the remaining five BRS indices.

The three frequency domain BRS indices showing reduction in patients with type 2 diabetes compared with control 2 participants were assessed further with respect to their possible associations with metabolic parameters (Table 5). After adjustment for sex, age, BMI, smoking, and antihypertensive medication, BRS-αLF, BRS-αHF, and BRS-αMean were positively associated with the M-value and inversely associated with fasting blood glucose and HbA_1c (all P < 0.05). There were no associations of BRS indices with fasting insulin, fasting C-peptide, and glucagon-stimulated incremental C-peptide or with first-phase insulin secretion (intravenous glucose tolerance test) in patients with type 2 diabetes or with any metabolic parameter in control 2 subjects (data not shown).

All BRS indices were inversely correlated with age in both diabetes and control groups (control 1: r = −0.48 to −0.71, control 2: r = −0.38 to −0.57, type 1 diabetes: r = −0.30 to −0.45, type 2 diabetes: r = −0.20 to −0.33). Figure 1 shows representative plots of the correlations between xBRS and age in the groups with type 1 and type 2 diabetes and their corresponding control groups.

In patients with type 1 diabetes, HbA_1c increased from baseline to 5 years from 6.4 ± 1.1 to 6.9 ± 1.0% and in those with type 2 diabetes from 6.3 ± 0.8 to 7.0 ± 1.1% (both P < 0.05). Furthermore, hs-CRP decreased from 22.3 ± 16.3 to 15.4 ± 21.8 nmol/L in participants with type 1 diabetes and from 105 ± 738 to 26.5 ± 24.2 nmol/L in those with type 2 diabetes (both P < 0.05). Table 6 reports the follow-up of BRS indices over 5 years in the diabetes groups. In patients with type 1 diabetes, only the BRS-to-SD ratio declined after 5 years (P < 0.05), while the remaining seven indices did not change. In participants with type 2 diabetes, xBRS, BRS(+) slope, BRS-αHF, and BRS-αMean decreased after 5 years (all P < 0.05 vs. baseline). However, after adjustment for the 5-year follow-up period using age-dependent equations for the BRS indices obtained from glucose-tolerant control subjects, statistical significance was lost for each of the five BRS indices that showed an unadjusted significant decline.

The results of this study point to an early baroreflex dysfunction, detected by frequency domain rather than time domain BRS indices, which is associated with both hyperglycemia and insulin resistance rather than insulin secretion in patients with well-controlled recent-onset type 2 diabetes. Four of eight BRS indices decreased further over the next 5 years in participants with type 2 diabetes, but this could be explained by an aging effect rather than by diabetes, because the decline was no longer statistically significant after adjustment for the 5-year follow-up period. By contrast, spontaneous BRS is not disturbed within the 1st year after diagnosis of type 1 diabetes and remains unaltered over the next 5 years.

There are no published data available that would allow a direct comparison with ours, since previous studies neither focused on BRS comparing recent-onset type 1 diabetes and type 2 diabetes nor did they include multiple BRS indices, measures of insulin sensitivity and secretion, or a prospective follow-up. Moreover, most of the published studies included relatively small sample sizes and focused on longer-term type 1 diabetes (19–21) or mixed groups of patients with type 1 and type 2 diabetes with a mean diabetes duration ranging from 9 to 28 years (18,22). On one hand, a decline in BRS with increasing diabetes duration was reported in individuals with longer-term type 2 diabetes, but no control group was included for comparison (32). On the other hand, although blunted BRS could be demonstrated in patients with type 2 diabetes compared with age-matched lean control subjects, a similar impairment was present in obese individuals without diabetes, suggesting that BRS dysfunction may also be due to the coexistence of obesity per se rather than type 2 diabetes (33). In fact, evidence has emerged suggesting that spontaneous BRS may also be impaired in persons with the metabolic syndrome (12) and prediabetes (14). Collectively, these studies suggest that diminished BRS may be detected not only in patients with longer-term type 1 and type 2 diabetes but also in individuals with the metabolic syndrome and obesity.

Only a few studies have assessed spontaneous BRS in patients with recently diagnosed type 2 diabetes. Michel-Chávez et al. (34) found no BRS alterations in the supine position in 30 patients with recently diagnosed type 2 diabetes whose diabetes duration was <2 years. In contrast, Gerritsen et al. (35) and Wu et al. (14) demonstrated reduced BRS using a single BRS index in individuals with newly diagnosed type 2 diabetes detected by screening, but measures of insulin sensitivity were not reported. Since in our study baseline mean HbA_1c was 6.5% in participants with type 2 diabetes, we hypothesize that even a slight degree of hyperglycemia is sufficient to induce alterations in BRS, particularly in view of the unknown duration of unrecognized hyperglycemia preceding the diabetes diagnosis. Our results in the baseline GDS cohort with type 2 diabetes are in line with some previous studies (14,35) and extend on these data by adding insulin resistance, assessed by the gold standard measure (M-value), as a risk factor for early BRS impairment in type 2 diabetes. Likewise, in obese individuals without diabetes, lower BRS was associated with the HOMA of insulin resistance (36), a surrogate measure of insulin resistance, and healthy individuals with higher HOMA of insulin resistance also showed reduced BRS (37). Of note, the relationship between insulin resistance and BRS was statistically mediated by cerebral blood flow in central autonomic regions, including the insula and cingulate cortex. Thus, activity within the central autonomic network may link insulin resistance to blunted BRS (37). However, experimental data obtained from a chronic hyperinsulinemic model suggest that chronic hyperinsulinemia rather than insulin resistance could contribute to baroreflex dysfunction, possibly due to insulin-mediated central effects of sympathoexcitation and vagal withdrawal (38).

The only noninterventional study that assessed BRS prospectively is the FinnDiane Study including 80 patients with type 1 diabetes with a mean diabetes duration of 8.8 years. After a 5-year follow-up, only BRS-αHF of six time and frequency domain indices deteriorated after adjustment for time of follow-up. When the six parameters were averaged, BRS declined after 5 years, but statistical significance for the change was lost when adjusted for time of follow-up. Low BRS at baseline did not progress to CAN but predicted an increase in the nighttime systolic BP (13). We extend on these data by demonstrating that BRS remains preserved over the first 5–6 years after the diagnosis of type 1 diabetes and that the early BRS alterations observed in recent-onset type 2 diabetes do not progress over 5 years in excess of the physiologic decline associated with the aging process.

The putative mechanisms underlying the BRS alterations observed herein deserve comment. Under physiological conditions, an evoked rise in arterial pressure reduces heart rate via excitation of arterial baroreceptor afferent nerves that activate nucleus tractus solitarius (NTS) neurons in the brainstem, which in turn excite cholinergic neurons in nucleus ambiguus to activate cardiac parasympathetic efferent nerves and neurons that produce γ-aminobutyric acid in the caudal ventrolateral medulla to inhibit presympathetic neurons in the rostral ventrolateral medulla (39). In obese Zucker rats, treatment with metformin or pioglitazone enhanced baroreflex control of heart rate by improving the NTS response to raising arterial pressure, suggesting that impaired glucose homeostasis in prediabetic, insulin-resistant male obese Zucker rats contributes to reduced baroreceptor-mediated NTS activation and bradycardia even before the onset of overt type 2 diabetes (39). Moreover, the attenuation of arterial BRS in diabetes and cardiovascular disease has also been attributed to changes in the arterial vascular walls, mechanosensitive ion channels, and voltage-gated ion channels. Some endogenous factors (such as angiotensin II and superoxide anion) can modulate these morphological and functional alterations through intracellular signaling pathways in impaired arterial baroreceptors. It has been suggested that arterial baroreceptors could serve as a potential therapeutic target to improve the prognosis of patients with diabetes and cardiovascular diseases (16).

The therapeutic consequences of BRS impairment in the early phase of type 2 diabetes can be derived from experimental and clinical studies suggesting favorable effects of nonpharmacological and pharmacological interventions on BRS function. In previously sedentary, otherwise healthy, middle-aged adults, 2 years of high-intensity exercise training improved integrated cardiovascular regulation by enhancing the BRS and dynamic Starling mechanism (40). The only randomized clinical trial in individuals with type 2 diabetes demonstrated improved BRS after 52 weeks of exercise training, which was confirmed by several nonrandomized trials over 12–24 weeks (41). Among the pharmacological interventions, pioglitazone treatment for 12 weeks augmented arterial BRS in relation to decreased muscle sympathetic nerve activity in patients with type 2 diabetes and acute myocardial infarction (42), while treatment with the calcium channel antagonist lacidipine for 4 weeks improved 24-h BRS in hypertensive patients with type 2 diabetes (43). From the therapeutic perspective, it is noteworthy that interventions known to enhance insulin sensitivity, such as exercise training, insulin sensitizers, or drugs targeting the renin-angiotensin system (44), also enhance BRS.

The strengths of this study are its prospective design, the relatively large baseline sample size of individuals with well-controlled type 1 and type 2 diabetes, the comprehensive phenotyping, including gold standard measurement of insulin sensitivity, and the inclusion of multiple frequency and time domain techniques to determine BRS.

Limitations of this study are the relatively small prospective 5-year cohorts of individuals with type 1 or type 2 diabetes and lack of prospective analysis of the glucose-tolerant control subjects. To account for this limitation, we used the equations for the physiologic age-dependent decline in BRS indices over 5 years in glucose-tolerant control subjects to compute age-adjusted BRS values, which were added to those obtained at 5 years.

Another limitation is that the group with type 2 diabetes and the corresponding control subjects were not matched for BMI and systolic BP. However, rigorous adjustment for these and other potential confounders, such as fat mass or VO_2max, still unveiled baroreflex dysfunction in recent-onset type 2 diabetes.

In conclusion, this study demonstrates an early baroreflex dysfunction in individuals with well-controlled recent-onset type 2 diabetes, which is associated with both hyperglycemia and insulin resistance rather than insulin secretion and does not progress over the next 5 years in excess of aging. Moreover, frequency domain BRS indices appear to be more sensitive than time domain BRS indices in detecting early baroreflex dysfunction in type 2 diabetes. In contrast, spontaneous BRS remains preserved within the first 5–6 years after the diagnosis of type 1 diabetes. It is conceivable that favorable modulation of BRS accompanied by improved insulin sensitivity can be translated into a reduction of cardiovascular end points and improved prognosis in people with type 2 diabetes, but this remains to be demonstrated in large-scale controlled clinical trials.

Acknowledgments. The authors appreciate the voluntary contribution of all study participants. The authors thank the staff of the Research Group Neuropathy, Institute for Clinical Diabetology at the German Diabetes Center, Düsseldorf, Germany, especially F. Battiato, M. Schroers-Teuber, and J. Schubert for their excellent work.

Funding. The GDS was initiated and financed by the German Diabetes Center, which is funded by the German Federal Ministry of Health (Berlin, Germany), the Ministry of Innovation, Science, Research and Technology of the state North Rhine-Westphalia (Düsseldorf, Germany), grants from the German Federal Ministry of Education and Research (BMBF) to the German Center for Diabetes Research e.V. (DZD), and partly funded through an European Foundation for the Study of Diabetes award to A.S. and D.Z. supported by Novartis.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. J.-L.K., G.J.B., and D.Z. researched data and wrote the manuscript. A.S., O.-P.Z., K.M., J.S., and M.R. researched data, contributed to the analysis, and revised the manuscript. D.Z. designed the study. D.Z. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

The GDS group consists of A.E. Buyken, J. Eckel, G. Geerling, H. Al-Hasani, C. Herder, A. Icks, J. Kotzka, O. Kuss, E. Lammert, D. Markgraf, K. Müssig, W. Rathmann, J. Szendroedi, D. Ziegler, and M. Roden (speaker).