Preventing relapse in alcohol disorder with EEG-neurofeedback as a neuromodulation technique: A review and new insights regarding its application

Highlights

• Alcoholism has a huge relapse rate (70–80% at one year post-detoxification).

• Besides psychotherapy and medication, add-on tools are needed to manage alcohol disorders.

• Neurofeedback can provide patients opportunities to modify specific altered brain activity linked to relapse.

Abstract

Alcohol Use Disorder (AUD) has a disconcertingly high relapse rate (70–80% within a year following withdrawal). Preventing relapse or minimizing its extent is hence a challenging goal for long-term successful management of AUD. New perspectives that rely on diverse neuromodulation tools have been developed in this regard as care supports. This paper focuses on electroencephalogram-neurofeedback (EEG-NF), which is a tool that has experienced renewed interest in both clinical and research areas. We review the literature on EEG-based neurofeedback studies investigating the efficacy in AUD and including at least 10 neurofeedback training sessions. As neurofeedback is a form of biofeedback in which a measure of brain activity is provided as feedback in real-time to a subject, the high degree of temporal resolution of the EEG interface supports optimal learning. By offering a wide range of brain oscillation targets (alpha, beta, theta, delta, gamma, and SMR) the EEG-NF procedure increases the scope of possible investigations through a multitude of experimental protocols that can be considered to reinforce or inhibit specific forms of EEG activity associated with AUD-related cognitive impairments. The present review provides an overview of the EEG-NF protocols that have been used in AUD and it highlights the current paucity of robust evidence. Within this framework, this review presents the arguments in favor of the application of EEG-NF as an add-on tool in the management of alcohol disorders to enhance the cognitive abilities required to maintain abstinence more specifically, with a focus on inhibition and attentional skills.

Introduction

Alcohol Use Disorder (AUD) is a major public health concern, with excessive consumption resulting in 3.3 million deaths worldwide each year (6% of all deaths). It is responsible for 5.1% of the disability-adjusted life years (World Health Organization, 2018). Reinforcing prevention and care hence still appear to be highly relevant goals. Although a multitude of therapeutic approaches are available for alcohol-dependent patients, the relapse rate nonetheless remains astonishingly high (approximately 80% at one year post-withdrawal) (Schellekens, de Jong, Buitelaar, & Verkes, 2015). This highlights the limitations of the current treatments, which are mainly based on psychotherapy and pharmacology (Bari et al., 2018). From a neurocognitive point of view, it is an indisputable fact that addictive substances hijack brain systems that subserve neuronal patterns associated with compulsive behavior of substance use (Sokhadze, Cannon, & Trudeau, 2008). According to the Dual Process Network Theory (Stacy & Wiers, 2010), dependent patients are faced with a conflict between strengthened impulsive habits of consumption and weakened executive processes underlying abstinence (Wiers, Gladwin, Hofmann, Salemink, & Ridderinkhof, 2013). An abnormal bottom-up system generates a “craving” response (i.e., an intense desire to drink; Robinson & Berridge, 2008) and an automatic approach tendency, while an ineffective top-down process gives rise to reduced executive control with impaired inhibition of the dominant response (e.g., Wiers et al., 2013). In present-day society, there is a paradoxical link between the volume allocated to advertising of the substance itself and the message allocated to the prevention of health risks. This inconsistency emphasizes the limited impact of health prevention campaigns, and advertisements may render detoxified patients vulnerable to eventual relapse by fostering attentional biases and by straining inhibiting abilities (e.g., Frings et al., 2018). Therefore, preventing relapse or minimizing its extent is a challenging goal for long-term successful management of addictive behaviors.

The typical care for alcoholic patients consists of a dual treatment comprising medication and psychotherapy. Pharmacological entities can target the GABAergic system (using drugs such as baclofen, acamprosate, or topiramate) or they can act as opioid antagonists (e.g., naltrexone or nalmefene). However, these treatments have a limited impact on drinking control, providing significant but still only low to medium efficacy (for a meta-analysis see Palpacuer et al., 2018). Moreover, numerous psychotherapeutic treatments are provided together with pharmacological entities. Although some of these evidence-based therapies result in significant effects, none of them have proven to be a watershed in AUD recovery to date (Ray et al., 2019). There is, therefore, a need for the development of new add-on tools as care supports. In this view, researchers have considered neural networks as a treatment variable of interest to reduce specific symptoms and/or to optimize performance (Enriquez-Geppert, Huster, & Herrmann, 2013). Transcranial electrical stimulation (Philip, Sorensen, McCalley, & Hanlon, 2019), transcranial magnetic stimulation (Philip et al., 2019), and cognitive training procedure (Azevedo & Mammis, 2017) are currently used for this purpose, promoting a “multimodal” approach suggesting that the typical psychological/pharmacological approach needs to be complemented with other treatments such as neuromodulation (e.g., Luigjes, Segrave, de Joode, Figee, & Denys, 2019) and cognitive revalidation (Bates, Buckman, & Nguyen, 2013). The main objective of the present paper is to focus on the merits of neurofeedback (NF), which is a tool that has experienced renewed interest in both clinical and research areas (Marzbani, Marateb, & Mansourian, 2016). This article outlines the arguments in favor of the application of neurofeedback as an add-on tool in the treatment of alcoholic detoxified patients by enhancing the cognitive abilities needed to maintain long-term abstinence.

The widespread interfaces for the application of NF mostly involve functional Magnetic Resonance Imaging-neurofeedback (fMRI-NF) and electroencephalogram-neurofeedback (EEG-NF) (Dickerson, 2018). As EEG has high temporal resolution (on a millisecond scale) and since NF relies on real-time processes, EEG-NF provides a definite advantage for optimal learning (Zotev, Phillips, Yuan, Misaki, & Bodurka, 2014). For all mental states, there is a corresponding cerebral activity arising from a distributed brain network. This activity can be recorded using an EEG during a sensory or a cognitive task. EEG recordings can be analyzed based on parameters such as the amplitude (the magnitude of the oscillation in microvolts) and the rhythm (the oscillation frequency in Hertz) (Mumtaz, Vuong, Malik, & Rashid, 2018), and we usually analyze either Event-Related-Potentials (ERPs) or Event-Related-Oscillations (EROs). On the one hand, EROs reflect the integrative activity and the communication between populations of neurons associated with information processing of the task stimuli (Kamarajan & Porjesz, 2015). On the other hand, ERPs are voltage deflections arising from these superimposed EROs (Başar & Dumermuth, 1982). ERPs represent the average of the electrical activity emanating from synchronized parallel processing systems, thus generating a compound waveform resulting from different oscillatory responses (Karakas, 2000). As ERPs do not reflect the entire activity of the EEG signal, the focus has now returned to analysis of the genuine EEG signal, namely the oscillatory activity (for an overview see Bastiaansen, Mazaheri, & Jensen, 2011). Here, neurofeedback takes on its full meaning. Indeed, the use of EEG-NF involves the identification of target oscillations in order to reinforce or inhibit specific forms of EEG activity associated with cognitive impairments (Gunkelman & Johnstone, 2005). Thus, EEG-NF protocols rely on electrical activity recorded from the scalp, and they mainly focus on: alpha (8–12 Hz), beta (13–30 Hz), delta (0–4 Hz), theta (4–8 Hz), and gamma (30–50 Hz) frequencies, or combinations of these such as the alpha/theta ratio, the beta/theta ratio, etc. (for a review see Marzbani et al., 2016). By offering access to this wide range of targets, the EEG-NF procedure broadens the scope of the investigations through a multitude of experimental protocols that can be considered.

At the methodological level, NF is a form of biofeedback in which a measure of brain activity is provided as feedback in real-time to a participant (Hammond, 2011). More specifically, the brain oscillatory activity is simultaneously recorded and converted into a visual or an auditory signal that is fed back in real-time to the participant. Thus, the participant learns to visualize, interact with, and manage their own brain activity through an interface that digitalizes and converts the recorded signal into customized and understandable information (Gunkelman & Johnstone, 2005) (see Fig. 1).

Neurofeedback requires the participant to actively participate in their own care by finding personal mental strategies, by learning about their impact on brain activity, and by actively implementing them in a repetitive way. This procedure relies on two fundamental concepts: operant conditioning and brain plasticity (Enriquez-Geppert et al., 2013). The participant is asked to modify this feedback information via mental strategies such as imagining particular events taking place (e.g., thinking about the negative consequences of alcohol consumption), and the expected changes are positively reinforced (Cox et al., 2016). Participants thereby control their responses, see their progress in real-time, and achieve optimal performance in order to control their symptoms or an unwelcome behavior (Cox et al., 2016). In this manner, the neurofeedback is aimed at providing patients opportunities to modify altered brain activity and to recover optimal functioning (Micoulaud-Franchi, Quiles, & Vion-Dury, 2013).