A Fresh Look at the Allostasis Theory of Addiction

Addictive disorders continue to be a major cause of death and disability worldwide, and substance use collectively accounts for 5.5% of global disease burden (1). The immense suffering of patients, families, and others affected by these problems cannot easily be quantified. Decades of scientific efforts and billions of dollars in research funding have only made a small dent in these tragic numbers.

People with clinically significant substance use problems show a motivational reprioritization (2) and as a result systematically forego healthy rewards in favor of drugs. They also continue to use drugs despite knowledge of serious, often fatal negative consequences (3). This would seem to indicate a significantly disordered function of brain circuitry subserving motivation and decision-making. However, our mechanistic understanding of why and how this happens in some people remains very limited (4). Obtaining scientific answers to these questions is critical for at least two reasons: to counter the stigma and social marginalization caused by the widespread perception of addiction as simply a character flaw, and to develop more effective, mechanism-based treatments (4, 5).

There once seemed to be a prospect for simple answers. Mesolimbic dopamine (DA) was established to play a key role for motivated behavior and reward, addictive substances were found to excessively activate this system, and the pursuit of reward mediated by DA was thought to be at the core of addiction (6). While an important starting point, this clearly could not be the whole story. For one thing, if it were, we would all become addicted. Things clearly must be more complex. There has since not been a lack of influential theories to account for the complexity of addiction (3, 7, 8). These theories, based on elegant preclinical data, have sparked lively debates in the field but are not necessarily mutually exclusive. The greater challenge is that their clinical relevance and ability to guide treatment development has remained unclear.

Among contemporary addiction theories, an affective allostasis conceptual framework has gained increasing influence. It emphasizes as a key phenomenon the emergence of negative affect, or “hyperkatifeia,” with development of addiction. This is thought to reflect a pathological recruitment of brain “anti-reward,” or stress- and aversion systems, driven by both acute and protracted withdrawal states. As these kick in, the thinking is that they cause a progressive shift of motivation, from positively to negatively reinforced drug seeking and taking, or from “reward” to “relief” seeking (7). This theory aligns with my clinical experience and has for decades guided my own research (9). But what if the stigma of addiction rears its head here as well and filters our clinical experience? Perhaps prevailing, moralizing attitudes they are met with make it easier for patients to talk about underlying traumatic experiences, negative emotions, and adverse consequences of drinking, rather than the highs that may still be fueling the desire to drink?

In this issue, King and colleagues (10) report provocative findings that would seem to prompt a need to take a fresh look at the allostasis theory, at least as it applies to alcohol use disorder (AUD). The investigators recruited people with or and without AUD. In both groups, they sampled high-resolution ecological momentary assessment (HR-EMA) data “in the wild” during a 3-hour monitoring period of an alcohol drinking episode and, on another occasion, a non-alcohol episode of the same duration. This allowed them to obtain a fine-grained characterization of subjective effects associated with alcohol taking. The specific subjective effects assessed were stimulation, liking, wanting, sedation, and negative affect.

The allostasis theory of addiction produces a clear, testable prediction for this experiment. In people with AUD, the pleasurable subjective effects of alcohol, such as stimulation, liking, and wanting should be attenuated compared with people without AUD. Conversely, the ability of alcohol intake to result in a reduction in negative affect should be increased. In fact, the study found exactly the opposite. Positive alcohol effects were greater in people with AUD. And while alcohol did reduce negative affect, this effect was smaller in magnitude and did not differ between the groups.

AUD is a highly heterogeneous condition, with the pattern of comorbidity being a clinically important aspect of heterogeneity. Comorbidity with internalizing disorders characterized by negative affect, such as depression and anxiety, is common in patients with AUD (11). If anywhere, reduction of negative affect by alcohol intake should presumably be most likely in that population. King and colleagues therefore targeted for recruitment participants with and without depression so that both their AUD and non-AUD groups were roughly balanced for this comorbidity. However, focusing on AUD participants with depression did not change the results. Positive alcohol effects still dominated in people with AUD and were greater than in people without it. Reduction of negative affect was small and did not differ between people with and without AUD.

But maybe an allostatic shift only occurs once AUD has reached a certain severity? To address this, King and colleagues compared the subjective effects of alcohol between participants with moderate-to-severe versus mild AUD. If anything, the positive subjective effects of alcohol were more pronounced in the more severe group, in which participants reported higher stimulation and wanting, although liking was similar between groups. Once again, the decrease in negative affect was small and did not differentiate people with mild AUD from those with the moderate-to-severe form of this condition.

These findings complement results coming from another important line of research by King and colleagues. For more than a decade, this group has followed a unique cohort. As young adults, participants entering this study were challenged with alcohol under controlled laboratory conditions, allowing subjective effects to be examined. Over the following years, participants were repeatedly reassessed, with very few of them being lost to follow-up. In this unique cohort, initial positive response to alcohol was the strongest predictor of the risk for subsequently developing AUD. And when the alcohol challenge was repeated, this positive alcohol response increased, rather than being attenuated over time, in those participants who developed AUD (12).

These studies are thoughtfully and elegantly designed. They are very well executed. The data are unusually clear. Collectively, these two lines of work would seem to pose a serious challenge for those of us whose work is guided by the allostasis theory of addiction. Nevertheless, I believe that important questions remain before it is possible to adjudicate this case.

Perhaps foremost among outstanding questions is whether the results of the present study will generalize to clinical AUD populations. The people with AUD that the authors studied were non-treatment seeking and had no current desire to stop drinking. This is a population that is likely to differ in important ways from people who seek treatment (13). Relatedly, almost half of the studied population had mild AUD. It is unclear to what extent this diagnostic category shares underlying mechanisms with what was described in the 1930s as “alcoholism” by patients themselves in the Big Book of Alcoholics Anonymous (14), as “alcohol dependence” half a century later by scientific pioneers of the alcohol field (15), or by any diagnostic system that has followed. At the low levels of severity that correspond to mild AUD, spontaneous remission is common, and the presence of a clinically relevant disorder can be questioned (4). The other important question that remains is of course to what extent the present findings will generalize to other addictive disorders.

In summary, King and colleagues present robust, provocative, and potentially very important findings in people with AUD. Their findings would seem to fundamentally challenge one of the most widely held contemporary theories of addiction that postulates affective homeostasis and a shift toward negatively reinforced pursuit of drugs to be central phenomena of addiction. Those of us who are inclined to believe that the allostasis theory captures clinically and neurobiologically important phenomena have some thinking to do.

For now, however, it remains entirely possible that the present findings are in fact not inconsistent with the allostasis theory of addiction. Instead, the implication may be that the neurobiological and motivational phenomena this theory rests on emerge relatively late in the development of addiction as a result of the onslaught of heavy drinking over many years to decades. Of course, even at this level of severity, treatment is sadly infrequent due to stigma, socioeconomic factors, and other reasons. But among people who seek treatment, the emergence of affective allostasis may well be a key factor that both maintains alcohol use and also leads patients to ultimately seek treatment. If this interpretation is correct, then the phenomena described by the allostasis theory, while not easily detected in non-treatment seeking populations, may still be critically important in patients seeking treatment for their addiction.