Abstract

1. Introduction

A growing body of work examines the link between depression and cardiac function, particularly heart-rate variability (HRV) (Carney et al., 2005; Henje Blom et al., 2014; Yaroslavsky et al., 2014). HRV, thought to reflect the impact of the sympathetic and parasympathetic branches of the autonomic nervous system on cardiac function, refers to the measurement of beat-to-beat changes in heart rate (Berntson et al., 1997). While high HRV is associated with healthy cardiac activity—low HRV suggests inadequate parasympathetic or excessive sympathetic activity (Michels et al., 2013; Task Force, 1996). HRV is thought to index the nervous system’s impact on the regulation of physiological arousal in proportionate response to an environmental challenge (Appelhans and Luecken, 2006; Thayer and Lane, 2000). HRV is consistently linked to emotion regulation ability—response to stress is associated with HRV reductions (Berntson et al., 1993; Porges et al., 1994), and low resting HRV is associated with negative emotionality and depressive symptoms (Beauchaine, 2001; Thayer and Lane, 2000).

Research investigating the high co-morbidity of adult heart disease and depression found that HRV partially mediated this link (Bhattacharyya et al., 2008; Carney et al., 2005; Celano and Huffman, 2011). Healthy adult studies generally find an association between low HRV and depressive disorders (Gorman and Sloan, 2000; Licht et al., 2008). Similar findings have emerged for youth (Forbes et al., 2006; Shannon et al., 2007; Tonhajzerova et al., 2010). However, demographic factors also need consideration when examining depression-HRV linkages in youth because changes in age and puberty are associated with changes in HRV (Tanaka et al., 2000), and sex might play a role in the link between HRV and depression (Greaves-Lord et al., 2007).

One relevant translational line of research links cardiac function, depression, anhedonia and stress within a rodent model. Depression induced by stress paradigms, (e.g. hindlimb unloading, chronic mild stress) is related to behavioral changes including decreased sucrose intake (a measure of anhedonia) and decreased spontaneous locomotor activity, suggesting reduced appetitive drive (Grippo et al., 2006, 2005; Grippo and Johnson, 2009). Stress-related physiological alterations associated with rodent models of depression include autonomic and cardiovascular changes (e.g. elevated resting heart rate, elevated sympathetic tone, decreased HRV) and immune function changes (e.g. increased pro-inflammatory cytokines, decreased immune response) (Grippo et al., 2003, 2005, 2008; Grippo and Johnson, 2009; Moffitt et al., 2008). This work is particularly relevant because it dovetails with recent human work linking stress, anhedonia, and depression (Pizzagalli et al., 2007).

Few studies examine cardiac activity and depressive symptoms among adolescents prospectively (Yaroslavsky et al., 2014) or discuss HRV and depressive pathophysiology within a translational framework. Our study had two goals. First, we examined HRV at Time 1 [T1] as a predictor of depressive symptoms one year later [Time 2; T2]. Second, we evaluated the relative importance of T2 anhedonia and other depressive symptom facets (negative mood, interpersonal problems, ineffectiveness, negative self-esteem), reflecting the five-factor structure of the Children’s Depression Inventory (CDI; Kovacs, 1992), to explain variability in T1 HF HRV.

2. Methods

3. Results

Means and SDs for study variables are presented in Table 1. Correlation analyses (Table 1) revealed that relative T1 and T2 HF HRV was negatively associated with several T2 CDI subscales. T1 LF HRV was positively related to T2 anhedonia, but not related to T2 depression overall. Contrary to our previous report, T1 VLF HRV was not related to T2 depressive symptoms. As expected, sex, age, and pubertal status were significantly associated with HRV (Table 1). We controlled for these variables in our models. Females and males were comparable on depressive symptoms, (t(71)= − 1.48, p=.14).

To evaluate the predictive power of relative T1 HF HRV in explaining variability in T2 depressive symptoms, we used hierarchical linear regression, first entering age, sex and pubertal development (see Table 2). While HF has the most support in the literature as a correlate of depressive symptomatology (Carney et al., 2001; Celano and Huffman, 2011), other reports found that VLF predicted concurrent depressive symptoms (Blood et al., 2015; Vaccarino et al., 2008). Thus we considered the predictive power of relative HF, LF, and VLF. Age, puberty and sex were first entered into the model F(3, 72)=1.41 p=.25, R²=.045. Then, HF, LF, and VLF HRV were added with forward selection, yielding a significant model, F(4, 72)=3.79, p=.002, R²=.165, ΔR²=.12. Only sex and HF contributed significantly to the model (ps<.05, see Table 2). Next, with T1 depressive symptoms entered, the overall model was significant in predicting T2 depressive symptoms, F(5, 72)=11.96, p<.001, R²=.47, but HF HRV was no longer significant, p=.108 (Table 2, Model 3). Because both T1 and T2 depressive symptoms significantly relate to T1 HF HRV, this suggests a stable linkage between HF HRV and depressive symptoms.

Table 2

One-year prospective association between total depressive symptoms and HRV.

	B	Std. Error	β	t	Sig.	R	ΔR2
Model 1
(Constant)	−.108	4.995		−.022	.983	.212	.045
Age	2.258	1.425	.196	1.585	.117
Sex	.493	.463	.181	1.064	.291
Puberty	−1.288	1.682	−.134	−.766	.446
Model 2
(Constant)	6.687	5.179		1.291	.201	.407	.165
Age	3.852	1.435	.073	.447	.656
Sex	.200	.446	.335	2.685	.009
Puberty	−1.506	1.585	−.157	−.950	.345
HF	−.117	.037	−.391	−3.133	.003
β in
Excluded
LF	−.059	−.312	.756	−.038	.348
VLF	.059	.373	.711	.045	.489
Model 3
(Constant)	10.33	4.192		2.464	.016	.687	.472
Age	−.809	.393	−.297	−2.061	.043
Sex	1.478	1.211	.129	1.220	.227
Puberty	1.184	1.342	.123	.882	.381
HF	−.052	.032	−.172	−1.630	.108
T1 CDI	.675	.108	.665	6.233	.000

Open in a separate window

^**p≤.01 (n=73).

Next, building on a translational animal model linking anhedonia to decreased HRV, we considered the relative significance of the five T2 CDI depression subscales for explaining variability in T1 HF HRV. We used hierarchical linear regression, entering the five CDI subscales with a forward selection procedure (Table 3). Only T2 anhedonia continued to account for significant variance in T1 relative HF HRV, F(5,72)=10.39, p=.002, R²=.12, ΔR²=.12, suggesting anhedonia is the specific depressive symptomatology that shows the strongest longitudinal link with variance in HF HRV.

Table 3

One-year prospective association between HF HRV and depression subscales.

	B	Std. Error	β	t	Sig.	R	ΔR2
Model 1
(Constant)	47.826	3.195		14.968	<.001***	.374	.14
Anhedonia	−3.215	.947	−.374	−3.395	.001***
β in
Excluded Variables
Negative mood			.074	.600	.551
Interpersonal problems			−.031	−.229	.819
Ineffectiveness			.009	.067	.947
Negative elf-esteem			.086	.747	.458

Open in a separate window

Model 1 regresses CDI subscales onto HF HRV.

^***p≤.001 (n=73).

4. Discussion

This study yielded three important findings. First, we observed a negative association between T1 relative HF HRV and T1 and T2 depressive symptoms, with reduced relative HF associated with greater depressive symptoms. This finding is consistent with previous research linking decreased HF HRV with depressive symptoms among adolescents (Henje Blom et al., 2010; Michels et al., 2013; Tonhajzerova et al., 2010). We also found a positive association between T2 depressive symptoms and T1 relative LF HRV. The literature on LF HRV and internalizing symptoms is mixed (Greaves-Lord et al., 2007; Schmitz et al., 2013) suggesting need for replication. Our findings suggest that youth with greater depressive symptoms may have greater relative sympathetic influence on their HRV. In contrast with previous reports based on concurrently assessed HRV and depression (Blood et al., 2015; Vaccarino et al., 2008), we did not find an association between T1 relative VLF HRV and T2 depressive symptoms. Given the dearth of previous longitudinal research linking depression and VLF HRV, we speculate this may reflect the different neurophysiological mechanisms underlying VLF, LF, and HF HRV and their respective connections to psychological functioning over time (Shaffer et al., 2014), although our modest sample size could have also been a factor.

Second, after considering demographic covariates, we found that only T1 HF HRV significantly predicted variability in T2 depressive symptoms. This finding adds to previous literature documenting a prospective link between autonomic functioning and depressive symptoms among adolescents (Henje Blom et al., 2010; Yaroslavsky et al., 2014). Moreover, consideration of T1 depressive symptoms suggests that the association between depressive symptoms and HF HRV largely reflects a more trait like association, persistent across one year.

Finally, we observed that after considering anhedonia, negative mood, interpersonal problems, ineffectiveness, negative self-esteem, only anhedonia one year later significantly accounted for the variance in relative HF HRV. This finding aligns with rodent models of depression concurrently linking stressor-induced behavioral indicators of anhedonia with decreased HRV, elevated sympathetic tone, and sympathovagal imbalance (Grippo et al., 2006^,, 2008; Moffitt et al., 2008). Although our study has important differences from the rodent studies, i.e. self-report versus behavioral assessment of anhedonia and longitudinal versus concurrent assessment, our results provide the first preliminary evidence for the translational link between anhedonic features of depression and cardiac function among adolescents.

5. Limitations and future directions

We studied depressive symptoms dimensionally. Thus we cannot make strong claims about HRV and clinical depression. Further, although no participants were taking anti-hypertensive medications, we did not exclude youth for hypertension. This and other relevant influences on HRV including physical activity/exercise behavior, should be considered in future research. Finally, our sample size was modest, warranting replication with a larger sample.

The high comorbidity of depression and cardiovascular disease elucidates the need for further research investigating early markers of depression risk, cardiovascular risk and their interplay. Animal models of depression have revealed common mechanisms between depression and cardiovascular function including autonomic function, baroreflex receptor function, and immune function with linkages to stress and anhedonia (Grippo and Johnson, 2009). Future studies should focus on these interrelated factors (i.e. HRV, blood pressure, pro-inflammatory cytokines) in children and adolescents to further examine the association between cardiac function and depressive pathophysiology, particularly anhedonia.

Acknowledgments

Abbreviations

References