Closing the brain–heart loop: Towards more holistic models of addiction and addiction recovery

1 |. INTRODUCTION

Contemporary neurobiological models of addiction coalesce around interactions among specific limbic and cortical brain regions that are thought to control the imbalance of automatic/impulsive relative to deliberate/self-controlled behaviors associated with substance use disorder (SUD). Evidence suggests, however, that information streams from the body link context and bodily states to the brain’s behavioral outputs via both subcortical and cortical paths. The present article argues that the absence of body systems in current neurobiological addiction theories limits their translational and clinical value by missing a fundamental yet malleable source of individual variability. This individual variability can be tied to the seemingly automatic reactions to salient contexts and cues that interfere with more deliberate behavioral responses and can be therapeutically manipulated. We focus on the cardiovascular system as a source of consequential visceral information that is inherently embedded in the experience of addiction.

2 |. THE BODY RELAYS CONTEXT TO THE BRAIN

Alcohol and drug use behaviors are initiated in-the-moment based on internal dynamics, such as cortical and subcortical integration of sensory information with affective states that are often prompted by salient environmental cues.¹ Further, body systems can promote addictive behaviors by, for example, their capacity to serve as “silent” drivers of attention allocation and embedded components of associative learning, motivation, and salience.^2–4 Theoretical models such as Damasio’s somatic marker hypothesis⁵ and Robinson and Berridge’s incentive salience hypothesis⁶ have led to important contributions in this area, but the specific ways in which body systems actively contribute to flexible or inflexible behavior towards alcohol and other drugs continues to receive only limited attention.^7–10

The neurophysiology, functional anatomy, and molecular mechanisms of the bidirectional information streams that connect neural and cardiovascular processes have been well characterized.¹¹ From this literature, the central autonomic network (CAN) was described as a coordinated network of forebrain, midbrain, and brain stem structures that centrally meditate autonomic nervous system functioning.^12,13 This set of structures bears remarkable overlap with those articulated in various contemporary neurobiological models of addiction, and the structures conceptually implicated in core features of addictive behavior.^14–16 Yet, translational addiction models have not gone much beyond a “black box” conceptualization of bodily inputs.

The body continually conveys signals that are components of arousal states and emotional tone to the brain through humoral (e.g., central chemoreceptors in the hypothalamus monitor the hematological milieu), vagal (i.e., cholinergic vagal afferents relay information about physiological states to sensory afferent cell bodies in the nodose ganglia that propagate to the CAN), and spinal pathways (e.g., the lamina 1 spinothalamocortical pathway with its somatotopic organization, conveying visceral information that converges with afferent information from cranial nerves before projecting to the CAN).¹⁷

Rather than being static inputs, however, there is marked complexity in how the brain receives and interprets this information. For instance, afferent signals from the heart may be manipulated or optimized at the level of the brain to alter the way affective stimuli are processed during differing psychological and physiological states.^18–20 Additionally, meta-analytic evidence points to distinct parasympathetic and sympathetic subnetworks to task negative (i.e., the default mode network) and task positive (i.e., the frontoparietal control network, salience network) networks, respectively,¹² which further embeds cardiovascular activation pathways within the functional neural networks that orient an individual towards self-referential thought or externally oriented attention. Moreover, it underscores the importance of the cardiovascular system processes that often operate outside of awareness and contribute to the automaticity of substance use behaviors in addictive disorders.

We therefore posit that CAN-regulated cardiovascular dynamics are not independent parallel processes that regulate physiological processes such as heart rate and blood pressure, but are in fact integrated and necessary components of the brain’s response to any given challenge. To this point, mounting evidence suggests that central control of visceral reactivity is predictive rather than strictly reflexive, such that neural prediction errors contribute to exaggerated and blunted cardiovascular stress reactions that undermine adaptive whole-body responses to challenge.¹⁸ This is a conceptual departure from predominant theories such as Thayer’s heuristic neurovisceral integration model²¹ that have focused on cardiovascular processes as downstream indicators of neural integration, rather than active processes that may enhance neural regulation.

We hypothesize that the neural alterations elicited by addiction also negatively affect central control of cardiovascular processes and arousal. In turn, dysregulation in these processes can then contribute to or exacerbate deficits in brain circuit regulatory mechanisms underlying addiction. A corollary hypothesis discussed in a later section of this article shows that improving central control of cardiovascular processes with interventions such as heart rate variability (HRV) biofeedback provides physiological scaffolding to support behaviorally focused addiction interventions. As a starting point, to illustrate functional overlaps between putative addiction processes and the primary functions of CAN structures (Figure 1), we use Koob and Volkow’s widely discussed neurobiological model of addiction.²² We note, however, that the highlighted functional and structural overlaps are applicable to other neuroscientific models of addiction.^15,23,24

3 |. PERIPHERAL AND CENTRAL CARDIOVASCULAR CONTROL IS ALTERED BY INTOXICATION

In Koob and Volkow’s neurobiological model of addiction,²² the binge/intoxication stage of the addiction cycle is characterized by substance seeking and use, loss of executive control mediated by glutamatergic systems,²⁵ and the manifestation and reinforcement of incentive salience via phasic activation of dopaminergic systems.²⁶ A number of subcortical neural structures, many in the basal ganglia, are implicated as primary drivers of the acutely reinforcing properties of alcohol and other drugs. These include the ventral tegmental area, substantia nigra pars compacta, nucleus accumbens, central nucleus of the amygdala, globus pallidus, and thalamus. As exposure to intoxication–withdrawal cycles builds, the dorsal striatum is recruited to automate stimulus–response habits that drive further substance use related behaviors.

It is noteworthy how these areas, many of which are shared by the CAN, are highly interconnected with brainstem areas that regulate the cardiovascular and respiratory systems. Alterations of cardiovascular processes during the binge/intoxication stage have been extensively documented. HRV (the variability in consecutive R-spike to R-spike intervals in the electrocardiogram waveform), which is commonly used as a biomarker of cardiovascular adaptive capacity, decreases during acute intoxication, as does blood pressure variability.²⁷ Additionally, the sensitivity of the neurally mediated feedback loop that modulates blood pressure known as the baroreflex has also been shown to decrease in response to alcohol intoxication.²⁸ In parallel, high risk drinkers show heightened response of an HRV index that captures baroreflex activation (i.e., the 0.1-Hz HRV index) to alcohol and other drug cues,²⁹ and vascular disruptions have been observed in binge drinkers within the first nine years of drinking.³⁰

This raises the possibility that the allostatic load of repeated intoxication–withdrawal cycles leads to a loss of flexibility in the cardiovascular system in the same way flexibility is thought to be lost in other body systems such as the hypothalamic–pituitary adrenal axis under repeated intoxication–withdrawal.³¹ Thus, less accurate information about physiological context would be relayed to the brain, further undermining basal ganglionic regulatory process, and hypothetically contributing in discoverable ways to the aberrant cognitive, affective, and behavioral features of addiction.

4 |. CARDIOVASCULAR FEATURES ARE EMBEDDED IN NEGATIVE AFFECTIVE STATES

The aversive experience of substance withdrawal, characterized by substance-specific physiological symptoms and negative affect, is also proposed as essential to the addictive cycle. In the withdrawal/negative affect stage of Koob and Volkow’s model,²² the extended amygdala, through its projections to the hypothalamus and brainstem, and in the case of opioid withdrawal the locus coeruleus,³² drives physiological responses to neurological states such as those arising from withdrawal. Evidence for including peripheral physiological feedback in this stage of addiction is strong because extended amygdala components including the central nucleus of the amygdala, bed nucleus of the stria terminalis, and to a lesser extent, the basolateral amygdala are important components of the CAN,¹³ and by definition, the neural structures of the CAN serve by merging feedback from the heart and vasculature with higher order cognitive and affective factors derived in midbrain and forebrain structures, as well as with information from other autonomic systems. In doing so, visceral constituents of withdrawal and negative affect become embedded in the subjective affective experience of these states.¹⁷ Tachycardia/heart palpitations and diaphoresis, for example, are inextricably linked to the negative affective states that arise during acute alcohol or opioid withdrawal. This aligns with current theory that conceptualizes the structures of the dopaminergic extended amygdala as receivers of information that are responsible for evaluating the emotional content of stimuli or states and allocating attention.^9,22

5 |. CARDIOVASCULAR INFORMATION IS INTEGRATED CORTICALLY

Koob and Volkow also propose a substance preoccupation/anticipation stage of addiction characterized by incentive sensitization that is driven by the basolateral amygdala and insular cortex and processing of contextual information represented in the orbitofrontal, prefrontal, and anterior cingulate cortices, as well as the amygdala.²² The insula, prefrontal, and anterior cingulate cortices are reciprocally innervated with midbrain structures, brain stem nuclei, and the periphery in the service of integrating effortful and automatic information processing (Figure 1). To understand how visceral and sensory information become embedded in an addictive process, it is useful to consider the functional roles of these cortical structures, all of which integrate cognitive and emotional information with a variety of bodily sensations. A clear example is the insular cortex, which as the seat of interoception, is considered a primary viscerosensory area that mediates consciousness of feelings through its role in the representation of bodily states and urges.^33,34 Large literatures further implicate this structure in emotional awareness, appetitive drive, and urges to use alcohol and other drugs.^34,35

Diminished capacity of the executive system structures to regulate attention, salience, and craving is believed to be an essential component of addiction.^36,37 The discussion of disruptions in attention and salience, and the experience of craving, raises consideration of the baroreflex feedback loop that delivers cardiovascular information to and from the brain and ensures a continuous bidirectional stream of modulation.³⁸ This recursive feedback loop involves well-documented pathways from the brain to the heart, via the efferent parasympathetic and sympathetic nervous systems, and from the cardiovascular system to the brain (afferent traffic) through baroreceptors—mechanoreceptors in arterial walls that respond to vessel stretching in response to moment-to-moment changes in blood pressure (Figure 2).

Baroreceptors convey information to nucleus tractus solitarius neurons that then propagate information to higher neural centers that control peripheral resistance (i.e., vasculature tone) and heart rate, as well as behavioral responding. The entirety of the baroreflex feedback loop allows us to consider how system variability is generated and used to modulate homeostatic set points and adjust to chronic challenges either adaptively or maladaptively (i.e., allostatic load).^39,40 Experimental evidence suggests baroreflex-mediated afferent signaling about cardiovascular variability participates in substance use behavior maintenance through its influence on attention capture by provocative cues, cue salience, and visceral reaction to cues.² Thus, visceral arousal responses elicited by cues in the environment—such the sight of a regularly attended drinking establishment or substance use location—can influence attention capture and thereby serve to perpetuate or exacerbate preoccupation, salience, anticipation, and craving. Other pathways of heart–brain interaction, such as assessed by heartbeat-evoked potentials⁴¹ and direct modulation of neural blood flow and oxygenation,⁴² further implicate cardiac cycle timing effects and predictive coding in attention and somatosensory perception.

6 |. CLINICAL IMPLICATIONS OF INCLUDING THE CARDIOVASCULAR SYSTEM IN MODELS OF ADDICTION

Cardiovascular disease progression associated with substance use is well studied,^43,44 but there is a need for laboratory and experimental study of the mechanistic role of cardiovascular changes in decision making,⁴⁵ the emotion regulation failures associated with addiction,⁴⁶ and their potential for rebound during recovery. We previously observed that women receiving a 12-week course of cognitive behavioral therapy for alcohol use disorder experienced improvements in cardiovascular regulation from pretreatment to posttreatment, even without stable abstinence.⁴⁷ These improvements were observed across measures of metabolic activity, HRV, and baroreflex sensitivity. This study supported other research showing that improved baroreflex sensitivity may signal improved blood pressure regulation and cardio protection,⁴⁸ as well as increased regulation of emotion, stress, and behavior,⁴⁹ and aligns with a growing literature showing improvements in cardiovascular processes following mindfulness-based interventions and cognitive behavioral therapy for a range of clinical presentations. It is likely that other treatments like dialectical behavior therapy give rise to similar improvements in cardiovascular functioning. Future clinical trials of psychotherapeutic interventions should include cardiovascular measures to explore such effects. Additionally, measures of cardiovascular regulation could serve as much needed objective measures of treatment response in clinical trials.⁵⁰

Observed improvements in cardiovascular regulation in individuals receiving cognitive behavioral therapy further support the possibility that improved cardiovascular–brain feedback may serve as a scaffold to reinforce the goals of psychotherapy and behavior change. It is also possible that restoration of autonomic homeostasis may help individuals better buffer stress and mitigate craving. If, in fact, the cardiovascular system is capable of recovery even after years of chronic, heavy substance use,⁴⁷ treatment adjuncts that directly target cardiovascular regulatory mechanisms such as the baroreflex may accelerate cardiovascular recovery in the service of sustaining substance use behavior change,⁵¹ as well as improving health and quality of life.

The idea that improvements in cardiovascular processes may help increase regulation of emotion, stress, and behavior is consistent with the observation that sympatholytic medications affecting the cardiovascular system are known to reduce stress, craving and substance use. Prazosin and doxazosin, two alpha-1 adrenergic inverse agonists commonly referred to as alpha-blockers and widely used to treat hypertension, reduce stress reactivity, craving, and the use of alcohol and other drugs,^52–56 potentially by dampening high basal levels of sympathetic arousal commonly observed in individuals with SUD.^55,56 Whereas the effects of alpha-blockers on blood pressure are primarily driven by countering norepinephrine-mediated vasoconstriction via alpha-1 adrenergic receptors in the smooth muscle of the vasculature, effects on craving and substance use likely derive from direct actions on alpha-1 adrenergic receptors found throughout the central nervous system.

GABAergic drugs offer another class of medications with the potential to mitigate CAN/autonomic dysregulation in SUD.⁵⁷ Acamprosate (Campral), for instance, is a centrally acting GABA agonist and glutamate antagonist⁵⁸ that is hypothesized to attenuate hyper-glutamatergic states associated with craving⁵⁷ and relatedly effect reductions in psychophysiological arousal, as evidenced by reduced heart rate, galvanic skin conductance,^59,60 and blood cortisol.⁶¹ Acamprosate may also reduce alcohol and other drug use, although findings to date have been mixed.⁶²

Similar effects can also be achieved with HRV biofeedback, a behavioral intervention that originated from the observation that maximal increases in the amplitude of heart rate oscillation are produced when the cardiovascular system is rhythmically stimulated by paced breathing at a frequency of approximately 0.1 Hz (equivalent to six breaths per minute).^63,64 This effect is linked to resonance properties of the cardiovascular system and the physiological phenomenon of respiratory sinus arrhythmia, that is, increases and decreases in heart rate in response to respiratory inhalation and exhalation. HRV biofeedback is empirically supported for treating a variety of disorders and performance enhancement, putatively through effects on baroreceptors and the vagal afferent pathway to frontal brain regions.⁶⁵ A recent meta-analysis found that HRV biofeedback was associated with large effect size reductions in stress and anxiety that were not moderated by clinical diagnosis.⁶⁶ A small but growing body of research also suggests that HRV biofeedback may be a useful complement to current first-line treatments for SUD. Preliminary studies suggested that both short^67,68 and longer term^69,70 courses of HRV biofeedback may help reduce craving levels beyond reductions observed during inpatient addiction treatment as usual, and relapse prevention housing. A recent systematic review suggested that HRV biofeedback and paced breathing proxies were useful for promoting decreases in craving, although larger and better controlled studies are needed.⁷¹

Theoretically, HRV biofeedback enhances communication between neural and cardiovascular systems that modulate physiological arousal in response to salient cues and internal states that trigger substance use, and thus may increase behavioral flexibility towards appetitive stimuli, and improve individuals’ capacity to regulate negative affect. Given the strong psychophysiological component of craving⁷² and HRV biofeedback’s ability to acutely effect major shifts in autonomic activation,⁶⁵ it is possible that HRV biofeedback may mitigate craving by helping individuals restore autonomic homeostasis or executive control, or reduce autonomic allostatic load during craving states. It is also possible that HRV biofeedback may chronically enhance CAN regulatory capacity to buffer against craving or restore cardiovascular flexibility in the more generic service of supporting new learning. These speculations are empirically testable in human experimental studies designed to capture HRV biofeedback’s effects on craving, emotion, cognition, and behavior at multiple, integrated system levels using neuroimaging, electroencephalogram, electrocardiogram, cognitive tasks, neuropsychological tests, and subjective report. The growing availability and scalability (via mobile devices) of HRV biofeedback provides the means for expanded and ecologically valid study of enhanced baroreflex sensitivity and diminished in-the-moment cardiovascular reactivity to triggers as potential mechanisms that may interrupt substance use behaviors in the natural environment.

A recent proof-of-concept functional magnetic resonance imaging study supports this idea. Bates and colleagues²⁰ exposed alcohol drinkers to two blocks of alcohol picture cues with an experimental group completing 5 min of 0.1-Hz breathing between alcohol cue blocks (equivalent to six breaths per minute; a typical breathing rate in HRV biofeedback at which resonance between respiration and HRV occurs) and controls completing a low-demand cognitive task. During exposure to the second round of alcohol cues, the experimental group exhibited significantly less activation in brain regions involved in visual processing (i.e., left inferior and superior lateral occipital cortices, right inferior lateral occipital cortex, bilateral occipital pole, and temporal occipital fusiform cortices), while activation in brain areas implicated in behavioral control, internally directed cognition, and brain–body integration was increased (i.e., medial prefrontal, anterior and posterior cingulate, and precuneus cortices, paracingulate, and lingual gyri). The control group showed no significant changes in brain activation between alcohol cue exposures. These findings provide preliminary evidence that manipulation of the cardiovascular system with 0.1-Hz breathing alters neural activation in a manner theoretically consistent with a dampening of automatic sensory input and strengthening of higher level cognitive processing.

7 |. CONCLUSIONS

There remains vigorous debate about the neurological processes underlying the acute reinforcement of alcohol and other drugs, the learning and stimulus–response habits that drive further substance use, the aversive experience of substance withdrawal, and perseveration of motivational states (e.g., craving).⁴⁵ The cardiovascular system, as well as other sensory and visceral systems, importantly contribute to these processes and the experience of addiction by providing contextual information to the brain, often outside of conscious awareness. This information can derive from and support an overlearned behavioral repertoire, even in opposition to conscious efforts to inhibit such responses. Physiological reactivity serves a critical role in survival and adaptation, but can also become maladaptive when reflexive, predictive, or over-learned responses (e.g., changes in heart rate and blood pressure) help to perpetuate SUD.

We suggest that neurobiological models that include sensory and visceral information streams are more complete because their inclusion allows models to more roundly explain the role of context and affective states in addiction and addiction recovery. Decades of psychophysiological research provides tractable experimental manipulations to test the plasticity of the baroreflex loop, and mobile assessment of physiological signals used to study cardiovascular mechanisms in the natural environment are now widely available. As such, as a field, we are well positioned to capture simultaneous visceral and neural response through well-defined and ecologically valid decisional and emotion regulation tasks that will inform more holistic models of addiction and recovery—work that is ongoing in the cognitive and affective neuroscience communities at large.⁴⁵

Interest is growing in the way the cardiovascular and other physiological systems (e.g., immune and gastrointestinal) are involved in psychopathology, but research in these domains remains “fringe” in both biological and psychological sciences of addiction. In this article, we used Koob and Volkow’s neurobiological model of addiction as an example to point out several natural intersections between the brain, body, and behavior. We emphasized that without sensory and visceral input and reactions, the brain would be disconnected from the bodily context in which it exists. These ideas integrate well with other neuroscientific models and theories of addiction,^5,25,73 that provide an additional level for exploring core biological mechanisms underlying addictive disorders through Research Domain Criteria (RDoC).⁷⁴ These ideas also align with calls for neuro-clinical assessment of individuals with addiction to characterize patients’ individual prognosis and probability of response to behavioral and pharmacological treatments.²⁵

In conclusion, it is difficult to imagine a comprehensive model of addiction that does not consider sensory and visceral information streams and attendant bodily processes as fundamental components of behavioral responding. Expanding existing neurobiological models of addiction to more comprehensively integrate the neurological structures subsumed by the CAN and the sensory and visceral information streams it both controls and is influenced by, as well as connected bodily systems (e.g., cardiovascular, endocrine), will grow the translational and clinical utility of current models. Further, this will support the development of noninvasive, body-based interventions that specifically target the automatic physiological reactivity that so often undermines SUD recovery attempts and serve as a complement to existing cognitive-behavioral and pharmacotherapeutic approaches.