Cardiac Autonomic Nervous Dysfunction in Diabetic Patients With a Mitochondrial DNA Mutation: Assessment by heart rate variability

We investigated HRV in 10 unrelated patients who had type 2 diabetes associated with the 3243-bp mutation. They had no clinical signs of MELAS but were suspected of having the mutation because of maternally inherited diabetes and/or hearing impairment. They were proved to have the mutation in blood leukocytes by a molecular test using a restriction endonuclease, ApaI, as we previously reported (8). Diagnosis of diabetes had been made according to the World Health Organization criteria in 1985. All 10 DM-Mt3243 patients underwent 24-h Holter monitoring because of chest pain (5 patients), palpitation (4 patients), and/or DM-Mt3243 (4 patients). HRV was analyzed, and the results were compared with those of 55 ordinary diabetes patients and 45 control subjects. The ordinary diabetes group consisted of 55 consecutive patients with type 2 diabetes (age range 50–75 years) who underwent Holter monitoring at Tokyo Saiseikai Central Hospital from 1996 to 1999 for chest pain (15 patients), palpitation (14 patients), suspected arrhythmia (13 patients), and/or long duration of diabetes (18 patients). The control group comprised 45 consecutive nondiabetic subjects (age 50–75 years) who had Holter monitoring from 1998 to 1999 for chest pain (15 patients), palpitation (13 patients), suspected arrhythmia (5 patients), and/or medical checkup (12 patients). All subjects in the three groups underwent echocardiography. In all subjects, results of stress ²⁰¹Tl imaging or exercise testing were negative. Any subjects taking cardioactive drugs, such as β-adrenergic blockers and antiarrhythmic drugs, were excluded from the study.

Two-channel Holter monitoring was performed with a Marquette 8500 recorder (Marquette Electronics, Milwaukee, WI). HRV over a 24-h period was analyzed using a commercially available computerized system (Marquette Electronics). All tapes were visually processed by an experienced technician with manual imposition of the mean R-R interval for all ectopic beats. Any tapes that had >5% ectopic beats in a 24-h record were excluded. In the time domain analysis of HRV, the SD of all R-R intervals (SDNN), the SD of 5-min mean R-R intervals (SDANN), the mean of all 5-min SDs of R-R intervals (SDNN index), the root-mean-square difference of successive R-R intervals (RMSSD), and the proportion of adjacent R-R intervals varying by >50 ms (pNN50) were calculated. In the frequency domain analysis, the recording period was divided into 2-min segments, and total (0.01–1.00 Hz), low frequency (LF) (0.04–0.15 Hz), and high frequency (HF) (0.15–0.40 Hz) spectra were calculated at each segment. The Marquette system provides spectral plots as amplitudes (ms) instead of power (ms²). These values were computed hourly and averaged over 24 h.

Any differences among the three groups were evaluated by ANOVA with the Fisher’s least significant differences test for parametric variables, by the Kruskal-Wallis’ test for nonparametric variables, and by the χ² test for categorical variables. Any differences between two groups were evaluated by the unpaired Student’s t test for parametric variables, by the Mann-Whitney’s U test for nonparametric variables, and by the χ² test for categorical variables. A P value <0.05 was considered statistically significant. Results are presented as the means ± SD.

Of the 10 diabetes-Mt3243 patients, 5 patients (50%) and 4 patients (40%) underwent Holter monitoring because of chest pain and palpitations, respectively, but 4 patients (40%) had no cardiac symptoms. The percentages of patients having chest pain and palpitations were not different from those with ordinary diabetes (27 and 25%) and nondiabetic control subjects (33 and 29%) (P = NS).

Age, sex, and blood pressure were similar in the three groups (Table 1). Between DM-Mt3243 and ordinary diabetes groups, duration of diabetes and blood glycemic levels were not different. Family history of diabetes in mothers was present in 7 of 10 DM-Mt3243 patients (70%) versus 13 of 55 ordinary diabetic patients (24%) (P < 0.025). Hearing impairment was found in eight DM-Mt3243 patients (80%), three of whom needed a hearing aid. Regarding diabetes complications, peripheral neuropathy, which was defined as absence of Achilles tendon reflex, was found in 6 DM-Mt3243 patients (60%) versus 20 ordinary diabetic patients (36%) (P = NS). On echocardiography, thickness of the left ventricular wall was greater in DM-Mt3243 patients than in those with ordinary diabetes and control subjects (10.4 ± 1.5 vs. 9.3 ± 1.2 and 8.8 ± 1.3 mm, respectively; P < 0.05). LVH with wall thickness >11 mm was found in four DM-Mt3243 patients (40%) versus six ordinary diabetic patients (11%) and two control subjects (4%), but none of these patients had marked LVH (>15 mm).

Regarding HRV over a 24-h period (Table 2), the mean R-R intervals in DM-Mt3243 patients, ordinary diabetic patients, and control subjects were 782 ± 105, 834 ± 97, and 862 ± 98 ms, respectively. The mean R-R interval was shorter in DM-Mt3243 patients than in control subjects. In time domain analysis, there was no difference in SDNN and SDANN among the three groups. However, DM-Mt3243 and ordinary diabetic patients had significantly smaller SDNN index and pNN50 than control subjects. Compared with ordinary diabetic patients, those with DM-Mt3243 had smaller SDNN index (31 ± 10 vs. 44 ± 17 ms, P < 0.02), but rMSSD and pNN50 were not different. In frequency domain analysis, total, LF, and HF spectra were significantly smaller in DM-Mt3243 patients and those with ordinary diabetes than in control subjects. Compared with ordinary diabetic patients, DM-Mt3243 patients had smaller total (17.3 ± 5.7 vs. 25.9 ± 11.5 ms) and LF spectra (8.8 ± 3.4 vs. 14.4 ± 7.2 ms) (P < 0.05), but HF spectra (6.6 ± 2.4 vs. 9.1 ± 5.5 ms) were not significantly different. Notably, the LF/HF spectra ratio was lower in DM-Mt3243 patients than in ordinary diabetic patients and control subjects (1.37 ± 0.40 vs. 1.70 ± 0.46 and 1.68 ± 0.41 ms, P < 0.05), but this ratio was not different between ordinary diabetes patients and control subjects.

The circadian rhythms of LF and HF spectra in the three groups are shown in Fig. 1. LF and HF spectra were significantly smaller throughout the 24-h period in DM-Mt3243 patients and during the daytime in ordinary diabetic patients than in control subjects. Compared with ordinary diabetic patients, those with DM-Mt3243 had smaller LF during the daytime and smaller HF spectra only for a short time in the evening (P < 0.05).

To evaluate the association between peripheral neuropathy (PN) and HRV, DM-Mt3243 and ordinary diabetic patients were divided into two subgroups according to the presence or absence of PN (absent Achilles tendon reflex) (Table 3). Among the DM-Mt3243 patients, HRV parameters were not different between patients with and without PN. However, SDNN index and LF spectra tended to be smaller in DM-Mt3243 patients with PN than in ordinary diabetes patients with PN. In DM-Mt3243 patients without PN, SDNN index and LF spectra were smaller than in ordinary diabetic patients without PN (P < 0.05). DM-Mt3243 and ordinary diabetic patients were also divided into two subgroups by the presence or absence of LVH (Table 4). Among DM-Mt3243 patients, SDNN index and LF spectra tended to be smaller in patients with LVH than in those without LVH. Notably, HF spectra were smaller in patients with LVH than in those without LVH (P < 0.02). Compared with ordinary diabetic patients with LVH, DM-Mt3243 patients with LVH had smaller SDNN index and LF spectra (P < 0.05). Although DM-Mt3243 patients without LVH tended to have smaller SDNN index and LF spectra than ordinary diabetic patients without LVH, the LF/HF ratio was smaller in DM-Mt3243 patients without LVH than in ordinary diabetic patients without LVH (P < 0.02).

The 3243-bp mutation was originally detected in MELAS. In addition to brain and muscles, the involvement of peripheral nerves was reported to be common in patients with MELAS (2). Sural nerve biopsy showed axonal degeneration and marked loss of myelinated fibers. Markedly increased mitochondria with abnormal configuration of cristae were shown in Schwann cells (2). Some patients with symptoms of autonomic dysfunction, such as gastrointestinal dysmotility, were also reported (3). In the heart, LVH is recognized to be the clinical feature of cardiac involvement in MELAS (4), but cardiac autonomic nervous dysfunction has not been elucidated.

Approximately 1% of type 2 diabetes is associated with the 3243-bp mutation (1). The mechanism of this mutation that causes MELAS in some patients and diabetes in others remains unclear. In the heart, we recently reported that diabetes-Mt3243 patients have LVH more often than ordinary diabetic patients, but marked LVH, like hypertrophic cardiomyopathy, is rare (9). Regarding cardiac autonomic nervous function, Ishida et al. (10) reported two DM-Mt3243 patients with marked LVH who showed no uptake of ¹²³I-metaiodobenzylguanidine (MIBG) in myocardium. Imaging with ¹²³I-MIBG, a norepinephrine analog, can be used to assess cardiac sympathetic nervous function. The same group (11) recently showed reduced myocardial uptake of ¹²³I-MIBG in 10 DM-Mt3243 patients compared with 19 diabetic patients without the mutation. They suggested that the 3243-bp mutation would cause cardiac sympathetic nervous dysfunction in DM-Mt3243 patients. However, parasympathetic nervous function has not been evaluated.

HRV is another common tool to assess cardiac autonomic nervous function in various diseases, such as diabetes and myocardial infarction (12). Among HRV parameters, SDNN index and the LF component correlate with each other and reflect both the sympathetic and parasympathetic modulations; RMSSD, pNN50, and the HF component reflect primarily the parasympathetic modulation (13,14). The LF/HF ratio is a measure of sympathovagal balance and may reflect the sympathetic modulation (15). In our study, ordinary diabetes had significantly smaller SDNN index, pNN50, and LF and HF spectra than control subjects. However, the LF/HF ratio was not different. These results are compatible with those of the study by Pagani et al. (5) and suggest that cardiac autonomic function in patients with diabetes is impaired in both sympathetic and parasympathetic nerves. Aso et al. (6) also documented that sympathetic nervous damage in diabetes occurs much earlier than previously believed and runs parallel with parasympathetic damage.

Using HRV, we investigated cardiac autonomic nervous function in diabetes-Mt3243 patients compared with ordinary diabetic patients and control subjects. We found that DM-Mt3243 patients had significantly smaller SDNN index and LF spectra than patients with ordinary diabetes, which suggests that cardiac autonomic nervous function is more severely impaired in DM-Mt3243 than in ordinary diabetes. Regarding the parasympathetic parameters, RMSSD and pNN50 were not different between diabetes-Mt3243 patients and those with ordinary diabetes, but HF spectra were smaller in DM-Mt3243 patients only for a short time during the day. Notably, DM-Mt3243 patients had a lower LF/HF ratio than ordinary diabetic patients and control subjects. This ratio was similar in ordinary diabetic patients and control subjects. Therefore, sympathovagal imbalance in DM-Mt3243 patients is different than in ordinary diabetic patients and control subjects. Cardiac autonomic dysfunction in DM-Mt3243 patients may be more severe in sympathetic nerves than in parasympathetic nerves.

In our study, LVH was present in 4 of 10 DM-Mt3243 patients (40%) compared with 11% of ordinary diabetic patients and 4% of control subjects. As we previously reported (9), DM-Mt3243 patients are more likely to have LVH than patients with ordinary diabetes. Our study also showed that HRV parameters, especially HF spectra, were smaller in DM-Mt3243 patients with LVH than in those without LVH. Cardiac autonomic dysfunction in DM-Mt3243 seems to be more severe in patients with LVH than in those without LVH. Reduced HRV was reported to be predictive of cardiovascular death in diabetic patients as well as patients with myocardial infarction (12,16). LVH is recognized to be associated with cardiovascular morbidity (17). Therefore, DM-Mt3243 patients, especially those with LVH, may be at much higher risk for cardiovascular death than ordinary diabetes. However, to elucidate the prognosis and the clinical usefulness of HRV in patients with DM-Mt3243, a further study with long-term follow-up is needed.

Our study was not without limitations. First, due to the small number of DM-Mt3243 patients, further studies in a large number of patients are needed to confirm our data. Moreover, our data in Japanese may not be applicable to patients of other ethnicities. Second, cardiac autonomic function was evaluated only by HRV. Classic bedside procedures such as Valsalva maneuver were not used in our study. Peripheral neuropathy was also diagnosed only by the absence of Achilles tendon reflex. Other signs, such as pain and numbness, were not evaluated. Finally, of the 10 DM-Mt3243 patients, 6 underwent Holter monitoring for chest pain and/or palpitations. Although the percentages of patients having chest pain and palpitations were not different from those in ordinary diabetic patients and control subjects, these may have caused some selection bias and may have confounded the results.

We conclude that patients with diabetes associated with the 3243-bp mutation have more severely impaired cardiac autonomic nervous function than patients with ordinary diabetes. Moreover, they have sympathovagal imbalance that is different from that in patients with ordinary diabetes.