Heart rate variability in the subsyndromal depressive phase of bipolar disorder

Authors


Yeon Ho Joo, MD, PhD, Department of Psychiatry, Asan Medical Center, University of Ulsan College of Medicine, 388-1 Pungnap-2 Dong, Songpa-gu, Seoul 138-736, Korea. Email: jooyh@amc.seoul.kr

Abstract

Aims:  To compare the heart rate variability of bipolar patients in the subsyndromal depressive phase with healthy controls and to evaluate the relationship between severity of subsyndromal depressive symptoms and heart rate variability.

Methods:  Thirty-three bipolar patients in the subsyndromal depressive phase and 59 healthy controls were enrolled. A patient was considered to be in a subsyndromal depressive phase when the Montgomery–Åsberg depression rating scale score was ≤10 and the Clinical Global Impression–Severity scale (CGI-S) was ≤3 for the previous 1 month. After approximately 10 min of supine rest, all participants underwent resting electrocardiograms for 5 min in the supine position using limb leads. Different parameters of heart rate variability were analyzed in the time and frequency domains.

Results:  Bipolar patients had significantly lower standard deviation of all RR intervals (SDNN), proportion of adjacent NN intervals that differ by >50 ms (pNN50), log total power (log TP) and very low frequency power (VLF) compared to healthy controls. There were significant negative correlations between CGI-S score and some heart rate variability parameters, including heart rate variability index, SDNN, root mean square successive difference (RMSSD), pNN50, log TP, VLF, low frequency power (LF) and high frequency power (HF).

Conclusion:  Patients with bipolar disorder in the subsyndromal depressive state have reduced heart rate variability relative to healthy controls, and reduction of heart rate variability appears to be correlated with severity of symptoms in bipolar patients.

RECENT STUDIES OF patients with bipolar disorder have focused on the subsyndromal phase, traditionally regarded as a state of remission, because the social and occupational dysfunctions of these patients continue during this phase. A long-term follow-up study of patients with bipolar I disorder with affective symptoms indicated that the subsyndromal symptoms were primarily depressive rather than manic, and that subsyndromal and minor affective symptoms predominated.1 In addition, subsyndromal depressive symptoms in patients with bipolar disorder are associated with functional impairments, and there is an association between severity of subsyndromal symptoms and extent of impairment.2–4 Evaluations of the severity of depression of bipolar patients in the subsyndromal or syndromal phase typically use the Hamilton Depression Rating Scale (HAM-D)5 or the Montgomery–Åsberg Depression Rating scale (MADRS).6 These scales, however, do not evaluate all aspects of bipolar depression7 and do not measure subsyndromal symptoms. Thus, a physiological method may be more useful than these scales.

Heart rate variability (HRV) analysis provides a non-invasive measure of autonomic activity and can be used to assess the effect of psychopathology and disease and the balance between sympathetic and parasympathetic functions. Healthy people exhibit a high degree of HRV, which reflects their ability to adapt quickly to physical or psychological demands. Reduced HRV is associated with cardiac damage (including myocardial infarction, impaired ventricular ejection, and sudden cardiac death),8 predicts risk after acute myocardial infarction, and is an early warning sign of diabetic neuropathy.9 Reduced HRV may also be an autonomic dysfunction that indicates a maladaptive response to stressors in the environment.10

Recent studies have reported that several psychiatric disorders, including unipolar depression, anxiety disorder, and post-traumatic stress disorder, are associated with reduced HRV11–17 and that an increase in HRV occurs after successful treatment of depression.5,11,18,19 This suggests that some psychiatric disorders are associated with dysfunction of the autonomic nervous system or of biological rhythms. Based on observations of the alterations of sleep patterns, appetite, and activity states of bipolar patients, several reports have suggested that the pathophysiology of bipolar disorder may be associated with impairment of biological rhythms.20–22 Several mechanisms of association of the pathophysiology and impairment of biological rhythms have been suggested. First, patients with bipolar disorder had altered cytokines, such as pro-inflammatory interleukin-6 and tumor necrosis factor-α,23 which are known to interact with the autonomic system (epinephrine, norepinephrine) and neurotransmitter system (dopamine, serotonin).24,25 Second, according to the review by Lenox et al., bipolar patients had a phase advance of REM sleep and temperature relative to the sleep/wake cycle and this dysregulation of the circadian pacemaker affected mood.26 Thus, measurement of HRV may be useful for assessment of the status of patients in the subsyndromal phase of bipolar disorder.

The purposes of the present study were to compare HRV of bipolar patients in the subsyndromal depressive phase with that of healthy controls and to evaluate the relationship between the severity of subsyndromal depressive symptoms and reduced HRV in bipolar patients.

METHODS

Subjects

The present study enrolled 33 patients (10 male, 23 female) who were outpatients at a psychiatric clinic. (ClinicalTrials.gov number, NCT01455038 [ClinicalTrials.gov]) All patients were diagnosed with bipolar disorder based on DSM-IV criteria and were in clinical remission. All patients received treatment as usual (supportive psychotherapy and/or pharmacological treatment). Patients were excluded if they had axis I or II psychiatric disorders other than bipolar disorder, neurological symptoms or signs, history of head trauma, history of cardiovascular diseases, and were taking any medication except a psychiatric medication. Healthy controls had no history of psychiatric disorders and no psychiatric symptoms based on interviews by a board-certified psychiatrist. The control group consisted of 59 healthy subjects (23 male, 36 female) from a health promotion center in a university-affiliated general hospital.

There were no significant differences between the patients and controls in terms of age, gender, educational level, smoking status, or body mass index (BMI; Table 1). The mean MADRS score of the bipolar patients was 3.8 ± 3.7 and the range of the Clinical Global Impression–Severity scale (CGI-S) at baseline indicated that 51.5% of the patients were normal, 33% were borderline, and 15.2% were mildly ill. Among all bipolar patients who underwent echocardiography, one patient did not take any psychiatric medication, 16 patients (48.5%) received monotherapy (lithium, n = 3; lamotrigine, n = 3; any antipsychotic medication, n = 10) and the other patients (n = 16, 48.5%) received polypharmacy (combination of antipsychotics, lithium, lamotrigine or antidepressants).

Table 1.  Baseline subject characteristics
 Bipolar disorderHealthy controlsP
  1. anova; χ2 test.

  2. §Bipolar disorder, n = 31 vs healthy controls, n = 48; bipolar disorder, n = 30.

  3. BMI, body mass index; CGI-S, Clinical Global Impression–Severity; MADRS, Montgomery-Åsberg depression rating scale; YMRS, Young Mania Rating Scale.

n3359 
Age (years)38.8 ± 13.539.8 ± 13.10.715
Males10230.274
Females2336 
Education (years)14.1 ± 3.215.3 ± 2.50.043
BMI (kg/m2)§23.2 ± 4.423.4 ± 3.50.811
Smoker5110.563
Non-smoker1945 
Clinical variables   
 MADRS3.8 ± 3.7  
 YMRS0.7 ± 1.7  
 CGI-S1.6 ± 0.7  
 Age at onset (years)29.6 ± 10.3  
Psychiatric medication   
 No medication1  
 Monotherapy16  
 Lithium(3)  
 Lamotrigine(3)  
 Antipsychotics(10)  
 Antidepressants(0)  
 Polypharmacy16  

We defined presence of subsyndromal depressive phase as MADRS score ≤10 and CGI-S score ≤3 during the previous month. The institutional review board approved this study and all patients and controls gave informed written or oral consent.

Electrocardiography

All participants were instructed not to smoke or drink alcohol or caffeinated beverages for at least 8 h prior to this study. After approximately 10 min of supine rest, an electrocardiogram (ECG) was recorded for 5 min in the supine position using limb leads according to the QECG-3 model (LXC3203; LAXTHA, Daegu, Korea). The participants were asked to breathe normally with eyes open and to lay quietly without moving. The ECG data were stored in a personal computer and analyzed using TeleScan (ver 2.0, LAXTHA) in the time and frequency domains. In the time domain, the following statistical parameters were calculated: normal-to-normal, R-R interval (NN), HRV index (total no. all NN/height of the histogram of all NN), standard deviation of the NN (SDNN), root mean square of differences of successive NN (RMSSD), number of pairs of adjacent NN intervals differing by >50 ms in the entire recording (NN50), and NN50/total number of all NN (pNN50). In the frequency domain, the power spectrum of HRV signals was calculated using Fourier transformation, and the power (area under the curve related to each component) was calculated for four components: total power (TP; variance of NN intervals over the temporal segment), very low frequency power (VLF; <0.04 Hz), low frequency power (LF; 0.04–0.15 Hz), and high frequency power (HF; 0.15–0.4 Hz).

Statistical analysis

Clinical and demographic variables of patient and control groups were compared using the χ2 test for categorical variables and the t-test or analysis of covariance (ancova) for continuous variables. HRV parameters were compared using ancova (covariates: age, gender). The relationship between severity of subsyndromal depressive symptoms and reduced HRV was assessed using non-parametric statistical methods (Spearman's correlation). A two-tailed P ≤ 0.05 was considered statistically significant. All analyses were performed using spss (ver. 14, spss, Chicago, IL, USA).

RESULTS

HRV parameters

Table 2 compares the HRV results of bipolar patients and healthy controls. ancova (covariates: age, sex) indicated that patients had significantly lower SDNN and pNN50 in the time domain and significantly lower log TP and VLF in the frequency domain.

Table 2.  Heart rate variability
 Bipolar disorderHealthy controlsP
  • ancova (covariate: age, sex).

  • HF, power in high frequency range (0.15–0.4 Hz); HR, heart rate; HRV, heart rate variability; LF, power in low frequency range (0.04–0.15 Hz); pNN50, no. pairs of adjacent normal-to-normal intervals differing by >50 ms in the entire recording/total no. normal-to-normal intervals; RMSSD, root mean square of differences of successive normal-to-normal intervals; SDNN, standard deviation of the normal-to-normal interval; TP, variance of NN intervals over the temporal segment; VLF, power in very low frequency range (<0.04 Hz).

n3359 
HRV index11.0 ± 3.512.3 ± 3.80.043
Time domain   
 HR (beats/min)68.6 ± 8.768.0 ± 9.60.821
 SDNN35.5 ± 12.741.0 ± 12.70.020
 RMSSD28.9 ± 18.028.8 ± 13.30.807
 pNN50 (%)30.3 ± 15.737.1 ± 14.80.014
Frequency domain   
 Log TP6.8 ± 0.77.1 ± 0.70.014
 VLF5.8 ± 0.66.5 ± 0.8<0.001
 LF5.4 ± 0.95.6 ± 0.80.123
 HF5.5 ± 1.15.6 ± 0.80.588

Subsyndromal depressive symptoms and reduced HRV in bipolar patients

Pearson correlation analysis indicated significant negative correlations between CGI-S score and HRV index (r = −0.424, P = 0.014), HR (r = −0.156, P = 0.385), SDNN (r = −0.415, P = 0.016), RMSSD (r = −0.347, P = 0.048), pNN50 (r = −0.436, P = 0.011), log TP(r = −0.418, P = 0.016), VLF (r = −0.129, P = 0.475), LF (r = −0.379, P = 0.030) and HF (r = −0.396, P = 0.022). There were no significant correlations, however, between CGI-S score and VLF index or HR. In addition, there was no significant correlation between HRV and age of onset, duration of illness, number of episodes, or duration of hospitalization.

In Table 3 the HRV of controls is compared with that of patients with different CGI-S scores. Most HRV parameters of patients with CGI-S scores of 2 or 3 (but not 1) were lower than those of healthy controls. Except for VLF, however, there were no significant differences of HRV parameters between patients with CGI-S scores of 1 and healthy controls. In addition, we divided the patients into a MADRS score 0 group (n = 13) and MADRS score 1–10 (n = 20) and compared HRV parameters of healthy controls with that of patients according to MADRS scores. When the healthy group was compared to the MADRS groups (MADRS 0 group or MADRS 1–10 group) the results were similar to comparison of the healthy group and the CGI-S groups (CGI-S 1 group or CGI-S 2,3 group). Statistically significant differences were not observed, however, for LF and HF variables between the healthy group and MADRS score 1–10 group, while they were observed for the healthy group versus the CGI-S 2–3 group.

Table 3.  Heart rate variability vs CGI-S score
 HCBipolar disorder (CGI-S)P
CGI-S = 1CGI-S = 2,3HC vs CGI-S = 1HC vs CGI-S 2,3
  • t-test.

  • HC, healthy control; HF, power in high frequency range (0.15–0.4 Hz); HR, heart rate; HRV, heart rate variability; LF, power in low frequency range (0.04–0.15 Hz); pNN50, no. pairs of adjacent normal-to-normal intervals differing by >50 ms in the entire recording/total no. normal-to-normal intervals; RMSSD, root mean square of differences of successive normal-to-normal intervals; SDNN, standard deviation of the normal-to-normal interval; TP, variance of NN intervals over the temporal segment; VLF, power in very low frequency range (<0.04 Hz).

n591716  
HRV index12.3 ± 3.812.4 ± 3.99.4 ± 2.20.9220.004
Time domain     
 HR (beats/min)68.0 ± 9.667.6 ± 7.869.6 ± 9.60.8780.550
 SDNN41.0 ± 12.740.9 ± 13.029.9 ± 9.80.9770.002
 RMSSD28.8 ± 13.328.8 ± 13.322.5 ± 12.70.2550.093
 pNN50 (%)37.1 ± 14.837.0 ± 14.823.1 ± 13.80.9750.001
Frequency domain     
 Log TP7.1 ± 0.77.1 ± 0.76.5 ± 0.70.8260.002
 VLF6.5 ± 0.85.9 ± 0.45.6 ± 0.8<0.001<0.001
 LF5.6 ± 0.85.6 ± 0.85.1 ± 0.90.7440.031
 HF5.6 ± 0.86.0 ± 1.05.1 ± 0.90.0920.043

DISCUSSION

The present findings suggest that patients with bipolar disorder in the clinically remitted depressive state have reduced HRV. In particular, these patients have reduced HRV index, SDNN, pNN50, log TP, and VLF index. Moreover, reduction of HRV appears to be correlated with the severity of symptoms.

Studies of HRV in patients with bipolar mania, bipolar depression, and unipolar depression have consistently reported sympathetic hyperactivity and parasympathetic hypoactivity.27–31 Several previous studies have found a positive correlation between severity of unipolar depression and reduced HRV,32–34 however, to our knowledge, no previous study has investigated this correlation in patients with bipolar depression. The present study indicates a positive correlation between severity of clinically remitted bipolar depression and reduced HRV. In particular, we found a negative correlation between CGI-S score and HRV variables such as HRV index, SDNN, RMSSD, pNN50, log TP, LF, and HF and a significant reduction of HRV in bipolar patients with CGI- S scores of 2 or 3 (but not 1) compared with healthy controls. Together, these findings suggest that HRV reflects subtle depressive symptoms in bipolar patients in the clinically remitted state and that HRV may be a useful indicator of the subsyndromal phase of bipolar depression. In addition, VLF, unlike the other variables of HRV, was significantly decreased in bipolar patients with CGI-S scores of ≤3 relative to healthy controls. In other words, VLF can be considered as a trait marker rather than a state marker.

A limitation of this study is that we could not exclude the possibility that psychiatric medications influenced the HRV of the present patients. Licht et al. reported that reduced HRV in the presence of major depressive disorder might be due to the effect of antidepressants.15 Mujica-Parodi et al. observed a deleterious effect of atypical antipsychotics on HRV, which may exacerbate an underlying vulnerability.35 To clarify the effect of psychiatric medication on HRV, we compared the HRV variables between patients who took a specific drug class and patients who did not take a specific drug class. Excepting the fact that patients who received lithium (n = 22) had reduced LF compared to patients who did not (n = 10), there was no significant difference in all HRV variables (HRV index, HR, SDNN, RMSSD, pNN50, logTP, VLF, LF, and HF) according to specific drug class including lithium, lamotrigine, antipsychotics, and antidepressants using the Wilcoxon rank sum test. (Whether bipolar patients received a specific drug class or not did not cause a statistically significant effect on all HRV variables.) In addition, a review article by Kemp et al. reported that a selective serotonin re-uptake inhibitor, mirtazapine, and nefazodone had no significant impact on HRV.34 Cohen et al. and Henry et al. found no significant effect of medications on HRV in patients with bipolar disorder and schizophrenia.36,37 The statistical method in the present study, however, cannot guarantee that psychiatric medication did not affect the HRV in bipolar patients, and the results of previous studies on the effects of psychiatric medications on HRV have been inconsistent. These results should be interpreted cautiously. Despite these limitations, the present study suggests that HRV appears to be a sensitive marker for bipolar disorder in patients who are in a clinically remitted depressive state.

ACKNOWLEDGMENT

This study was supported by a grant of the Korean Health Technology R&D Project, Ministry of Health and Welfare, Republic of Korea. (A090116)

Ancillary