The vicious cycle between (neuro)inflammation and alcohol use disorder: An opportunity to develop new medications?

Alcohol-related brain damage is a severe consequence of chronic heavy alcohol drinking, causing progressive cognitive dysfunction. Alcohol-induced neurodegeneration is associated with neuroinflammation. Alcohol may indirectly generate neuroinflammation by enhancing the production of systemic pro-inflammatory cytokines and/or directly acting in the central nervous system, where neural damage stimulates the release of inflammatory molecules. In turn, enhanced neuroinflammation increases voluntary alcohol intake, binge drinking, and withdrawal-related anxiety in rodents, and contributes to negative affect and craving in humans, creating a vicious cycle.

In the current paper in Alcohol Clinical and Experimental Research, Adams et al. conducted a meta-analysis of the association between alcohol use and neuroimmune markers, including cerebrospinal fluid (CSF), positron emission tomography (PET), and post-mortem brain tissue studies. This meta-analysis is critical to establishing the role of neuroinflammation in alcohol use disorder, with a focus on brain-based effects.

1. ELEVATED INFLAMMATORY MARKERS IN THE CSF DUE TO CHRONIC ALCOHOL INTAKE

The authors found only two studies investigating inflammatory markers in the CSF, reporting higher levels of monocyte chemoattractant protein-1 (MCP-1) in individuals with AUD (Umhau et al, 2014) and higher interleukin-6 (IL-6) in regular alcohol drinkers compared to healthy controls (Guest et al, 2014). However, Adams et al. could not conduct a meta-analysis of the two CSF studies because different methods were used in those studies. CSF levels of MCP-1 were positively associated with peripheral markers of alcohol-induced liver inflammation, i.e., serum gamma-glutamyltransferase (GGT) and aspartate aminotransferase (AST) levels (Umhau et al, 2014). MCP-1 is increased in neurodegeneration and neuroinflammation, possibly through and release by neurons, astrocytes, microglia, endothelial cells, and the choroid plexus, where its production is stimulated by peripheral proinflammatory mediators. In alcohol-preferring rats, MCP-1 level is higher in the central amygdala (CeA) and ventral tegmental area (VTA), and inhibition of MCP-1 gene expression in CeA and VTA reduces alcohol intake (June et al, 2015). In post-mortem brains of humans with AUD, MCP-1 protein concentrations were higher in the hippocampus, substantia nigra (SN), amygdala, and VTA, all brain areas that play a key role in the development and maintenance of AUD (He and Crews, 2008). The proinflammatory cytokine IL-6 was increased in the CSF even by low daily alcohol consumption, such as 0.5 drinks/day, compared to abstinent individuals (Guest et al, 2014). Serum IL-6 levels are higher in individuals with AUD and positively correlate with alcohol consumption, craving, and depressive symptoms. However, while significant changes in peripheral IL-6 levels in AUD and withdrawal are well-documented, whether these changes reflect CSF/central changes in IL-6 is unclear. Although, primarily, active microglia are responsible for IL-6 production in the brain, neurons, astrocytes, and endothelial cells could also produce IL-6. This results in elevated CSF levels due to neurodegeneration and neuroinflammation. Further studies should focus on CSF levels of inflammatory cytokines in AUD, their correlation with peripheral levels, symptoms, and alcohol intake, and their role in alcohol-related brain damage and addiction. Efforts to assay neuroinflammatory markers using non-invasive imaging methods should also be expanded.

2. EXAMINATION OF MICROGLIA ACTIVATION USING PET IN INDIVIDUALS WITH AUD

Microglia activation marks the activation of the immune system in the brain. When microglia become active, they increase the expression of 18 kDa translocator protein (TSPO), which is responsible for the translocation of cholesterol through the outer mitochondrial membrane. Therefore, PET studies can investigate microglia activation by using radioligands for TSPO. Radioligand-binding affinity is highly influenced by the rs6971 single nucleotide polymorphism (SNP), with individuals being high-affinity (Ala/Ala), mixed-affinity (Ala/Thr), or low-affinity (Thr/Thr) binders. Therefore, genotyping is a crucial step, to exclude low-affinity binders (LAB) and stratify the study sample based on high-affinity (HAB) and mixed-affinity (MAB) binding. The current meta-analysis included three PET studies, all using the [11C] PBR28 radioligand for TSPO binding, excluding LABs, and comparing individuals with AUD after several days of alcohol withdrawal management to healthy controls. The pooled standardized mean difference showed a lower total volume of distribution (V_T) in the hippocampus of individuals with AUD, suggesting a reduced TSPO expression and microglia activation. Other specific areas showing significantly (p < 0.05) lower V_T or a trend (p ≤ 0.10) for a reduced V_T, were the cerebellum, midbrain, thalamus, striatum, anterior cingulate cortex, and frontal cortex. Alcohol dependence severity and the average number of drinks per day correlated negatively with TSPO binding in the hippocampus. Binding affinity affected these results; Kim et al. found a difference between healthy controls and individuals with AUD among MABs, but not HABs (Kim et al, 2018).

Interestingly, although all three studies hypothesized an increased TSPO radioligand binding in AUD individuals due to enhanced neuroinflammation and microglia activation, their findings were in the opposite direction. In a later preclinical study with alcohol-dependent and non-dependent rats, Tyler et al. compared in vivo [11C] PBR28 radioligand binding using PET and in vitro autoradiography using the same ligand on the brain slices of the same animals (Tyler et al, 2019). Whereas they found no difference in PET results, they described significantly higher in vitro [11C] PBR28 binding in alcohol-dependent rats in the anterior hippocampus and thalamus (Tyler et al, 2019). Based on these results, Tyler et al. hypothesized that chronic alcohol use might upregulate an endogenous TSPO ligand(s) which attenuates [11C] PBR28 radioligand binding in vivo, but its effect is blunted in vitro. Therefore, the results of the above-mentioned PET studies reflect only the TSPO radioligand binding per se, rather than microglia activation-induced TSPO expression. The endogenous ligand competing with the radiotracer or changing the affinity of TSPO, as an allosteric modulator, may be PPIX (the major endogenous porphyrin), diazepam binding inhibitor protein (DBI), or cholesterol (Tyler et al, 2019). Future preclinical and human studies are required to examine the effect of chronic alcohol intake on these endogenous TSPO ligands and how their level correlates with radioligand binding.

The use of PET with TSPO-binding radiotracers should be reconsidered in investigating neuroinflammation in AUD. The SNP-based population variability of binding affinity is also a challenge, requiring genotype pre-screening and exclusion of low-affinity binders. Therefore, the third-generation radiotracer, [₁₁C] ER176, which is insensitive to the rs6971 polymorphism, may be used in future studies. It is also important to reconsider the literature generated by TSPO ligand binding, in light of its limitations, as the findings do not contradict the neuroimmune hypothesis of AUD.

3. ELEVATED PROINFLAMMATORY NEUROIMMUNE MARKERS IN POST-MORTEM BRAIN TISSUE OF INDIVIDUALS WITH AUD

Adams et al. also performed a meta-analysis using data from five post-mortem studies. Three studies investigated neuroimmune markers, namely, IL-1β, High Mobility Group Box Protein 1 (HMGB1), HMGB1/IL-1β heterocomplex, Receptor for Advanced Glycation End Products (RAGE) mRNA and protein, and phosphorylated Nuclear Factor Kappa B (NFκB). There were higher levels of these neuroimmune proteins in the hippocampus and prefrontal cortex of individuals with AUD than in controls.

Activation of RAGE and TLRs by pro-inflammatory cytokines, the cytokine-like HMGB1, or other inflammation-causing agents, including pathogen-associated molecular patterns (PAMP), leads to downstream activation of NFκB. The activated NFκB transfers to the nucleus and induces the expression of TLRs, and pro-inflammatory cytokines, such as IL-1, 2, 6, 8 and TNF-α, thereby amplifying immune signaling. The human post-mortem findings, included in the meta-analysis, from individuals with AUD support the up-regulation of the above-mentioned pro-inflammatory pathway (HMGB1 – RAGE / TLR – NFκB – cytokines) in the CNS. Lifetime alcohol consumption positively correlates with the number of HMGB1, TLR-2, 3, and 4 immunoreactive cells in the orbitofrontal cortex, which is increased due to AUD (Crews et al, 2013). From the multiple TLRs, TLR-4 seems to play an essential role in alcohol-induced microglia and NFκB activation. HMGB1 and TLR-7 protein levels are elevated in the hippocampus and correlate positively with total drinking years in individuals with AUD (Coleman et al, 2017). RAGE expression and protein level are also higher in the orbitofrontal cortex of individuals with AUD, primarily localized in neurons, and negatively correlates with the age of drinking onset (Vetreno et al, 2013). This finding is supported by preclinical experiments showing adolescent binge drinking abolishes the maturational decline of RAGE expression in the orbitofrontal and prefrontal cortex (Vetreno et al, 2013).

The pro-inflammatory cytokine IL-1β shows significantly increased immunoreactivity in post-mortem studies, in the hippocampus of individuals with AUD and, interestingly, correlates positively with blood alcohol concentration (BAC) at death (Coleman et al, 2018; Zou and Crews, 2012). The increased IL-1β in the hippocampus localizes primarily in neurons of the dental gyrus granule layer (Zou and Crews, 2012). Alcohol consumption up-regulates IL-1β, which is activated by the inflammasome. The immunoreactivity of inflammasome proteins NALP1 and NALP3 also increase due to AUD in the hippocampus. Elevated NALP1 seems localized in neurons and astrocytes, whereas NALP3 is in neurons and microglia (Zou and Crews, 2012). HMGB1/IL-1β heterocomplexes enhance IL1β receptor-induced proinflammatory gene expression by making it more pronounced than by IL-1β itself. These heterocomplexes were increased in AUD, post-mortem, measured using co-immunoprecipitation in the hippocampus (Coleman et al, 2018). Preclinical results support the relevance of IL-1 signaling in AUD since IL-1 receptor antagonism in the basolateral amygdala reduces binge-like drinking in mice. Additional post-mortem, in vitro and in vivo studies are critical for a better understanding of the process by which chronic alcohol use leads to activation of neuroimmune pathways and upregulation of neuroimmune markers. Furthermore, the actual contribution of pro-inflammatory pathways to the development and maintenance of addiction needs to be unfolded at the molecular-, cellular- and neurocircuit-level, in order to shed light on the vicious cycle between AUD and neuroinflammation.

4. NEUROINFLAMMATORY PATHWAYS AS THERAPEUTIC TARGETS IN AUD

A limited number of medications are approved for AUD treatment, and modulators of neuroinflammation are promising for drug development and repurposing. Here we discuss pharmacotherapies targeting the neuroimmune system, with a focus on the markers investigated in the meta-analysis by Adams et al. Of note, approved drugs may target neuroinflammatory pathways. The opioid receptor antagonist naltrexone is one of the FDA-approved medications for AUD treatment. Its opioid inactive (+)-isomer inhibits the TLR-4 downstream signaling. Alcohol exposure in adolescent mice results in increased alcohol binge-drinking and higher expression of TLR-4, and related genes, in the nucleus accumbens in adulthood. Treatment of these mice with (+)-naltrexone in adolescence prevented the consequences of early alcohol exposure (Jacobsen et al, 2018). Nalmefene is an antagonist of μ-opioid receptors, and weak partial agonist of κ-opioid receptors. It is approved by the European Medicines Agency (EMA) for AUD treatment. In preclinical experiments, nalmefene prevents up-regulation of IL-1β, TNF-α, MCP-1, and myelin damage induced by intermittent alcohol treatment, while these effects were abolished in TLR4 knockout mice (Montesinos et al, 2017). The effects of naltrexone and nalmefene on endogenous opioids has been emphasized as their mechanism of action, however, it is useful to recognize that immune effects are also present.

NFκB activation is an obvious pharmacological target, since it is a crucial step of the neuroinflammatory pathway in AUD. The phosphorylation and subsequent proteasomal degradation of cytosolic protein inhibitors of NFκB is mediated by the kinase IKKβ. IKKβ deletion in NAcc and CeA, and inhibition by TPCA-1 or sulfasalazine, decrease alcohol intake and preference in preclinical experiments (Truitt et al, 2016). Peroxisome proliferator-activated receptors (PPAR) modulate pathways involved in NFκB activation. Pioglitazone is a PPAR-γ agonist approved for the treatment of type 2 diabetes. Preclinical results show that pioglitazone attenuates alcohol-induced IL-6 and IL-1β expression in the hippocampus and entorhinal cortex and protects against alcohol-induced neurodegeneration. After promising results in rodents, pioglitazone was tested in individuals with AUD. The first double-blind RCT was terminated prematurely because of an increased risk of myopathy in the pioglitazone group. Despite the study being terminated early, analysis of the completers did not confirm the preclinical results, because pioglitazone was associated with increased alcohol cue-induced craving and did not affect CSF levels of proinflammatory cytokines, such as IL-6, TNF-α, or MCP-1 (Schwandt et al, 2020). Currently, there is one ongoing, double-blind, parallel-group RCT examining the efficacy of pioglitazone in reducing alcohol consumption (NCT03864146). Fenofibrate is used in the treatment of abnormal blood lipid levels, and it is an agonist of PPAR-α. In a rat model, fenofibrate dose-dependently decreased alcohol self-administration, and motivational effects of alcohol (Haile and Kosten, 2017). Completed double-blind, placebo-controlled RCT examining 9 day fenofibrate treatment in AUD showed no effect on alcohol craving or alcohol consumption as measured by drinks per week (NCT02158273).

Microglia activation is essential in neuroinflammation; Minocycline is a broad-spectrum antibiotic that crosses the BBB and reduces microglia activation. In mice, minocycline decreased voluntary alcohol intake, attenuated withdrawal-induced anxiety, and alcohol cue-induced relapse (Gajbhiye et al, 2018). In a double-blind, placebo-controlled RCT, 10-day minocycline treatment was well tolerated but did not affect alcohol cue-induced craving (Petrakis et al, 2019).

N-acetylcysteine (NAC) is an over-the-counter antioxidant with anti-inflammatory properties targeting glutamate transport through glutamate transporter-1 (GLT-1). GLT-1 removes excess glutamate from the extrasynaptic space. Excess glutamate can lead to glutamate-induced excitotoxicity, which plays an important role in alcohol-related brain damage. Therefore, restoring the disturbed GLT-1 activity protects against neurodegeneration in AUD. Acute intraperitoneal NAC administration reduced binge alcohol intake, and decreased alcohol self-administration and relapse after abstinence in dependent rats (Lebourgeois et al, 2019). A multicenter clinical trial that examined NAC in cannabis dependence was the basis for a secondary analysis established that showed NAC treatment to be associated with increased abstinence, fewer drinks, and drinking days per week (Squeglia et al, 2018). An ongoing placebo-controlled, double-blind RCT is investigating the effect of NAC on alcohol use, as measured by total number of standard drinks, in adolescents with AUD (NCT03707951), and additional efforts are ongoing on the potential use of NAC in patients with AUD and alcohol-associated liver disease. Clavulanic acid, a well-known β-lactamase inhibitor antibiotic, also increased GLT-1 expression in the CNS and reduced alcohol consumption and preference in alcohol-preferring rats (Hakami and Sari, 2017). These promising preclinical results make clavulanic acid a candidate for drug repurposing in AUD treatment. Finally, growing and exciting translational work points to the potential role of ibudilast, a neuroimmune modulator which selectively inhibits phosphodiesterases (PDE)-3, -4, -10, and -11, and macrophage migration inhibitory factor (MIF). A recent RCT shows that ibudilast reduces heavy drinking and alcohol cue-elicited neural activation (Grodin et al, 2021). The demonstration that ibudilast has peripheral and central immune effects that are associated with drinking outcomes provides an initial proof-of-mechanism that contributes to medications development and precision medicine for AUD.

In conclusion, and as nicely summarized in the recent work in Alcohol Clinical and Experimental Research by Adams et al. (2023), alcohol leads to acute and chronic inflammatory effects, and translational work supports the notion of neuroimmune changes, not only peripherally, but also centrally in AUD. Investigating neuroimmune and neuroinflammatory pathways associated with AUD represents an opportunity to elucidate the central and/or peripheral mechanisms responsible for the development and/or maintenance of AUD. Mapping the inflammatory response across the AUD cycle, including acute and protracted withdrawal phases, has great potential to elucidate intervention opportunities. This gain in knowledge, in turn, may lead to the identification of new treatment targets and the development of new medications and/or to the identification of biomarkers that can help efforts toward early diagnosis and/or predicting response to treatments. The latter is of paramount importance, given the critical need in the field to move forward with precision medicine approaches that can improve the well-being of people living with AUD and alcohol-related medical consequences.