Introduction

Substance use disorder is a chronic brain disorder that is associated with high relapse rates (Box 1) [1, 2]. Some information or cues related with these substances can promote the relapse problem, such as life stress or returning the context where drugs were taken [3]. Either stress [4] or related contexts [5] can retrieve intense desires to reexperience the effects of addictive drugs and substances among individuals who are abstinent. This intense desire is referred to as drug craving, which is a critical factor for inducing drug-taking behaviors and causing relapse [6, 7]. Therefore, how to reduce drug craving attracts much attention for treatment of drug relapse.

Notably, drug craving does not seem to decrease with chronic abstinence. In 1986, Gawin and Kleber [8] reported that drug craving in individuals with cocaine use disorders increased after 1–10 weeks of abstinence. About 15 years later, this phenomenon was confirmed by drug self-administration models in rats and defined as the incubation of drug craving, in which cue-induced drug craving progressively intensifies during the first several months of abstinence [9,10,11]. The incubation of drug craving is widely observed with many drugs, including cocaine [12,13,14,15], nicotine [16], alcohol [17,18,19], heroin [20, 21], morphine [22], methamphetamine [23, 24], oxycodone [25], and fentanyl [26]. Many studies have investigated the mechanisms, and several excellent reviews [27,28,29,30] have summarized critical factors that modulate incubated craving. The present review mainly focuses on translational opportunities in the field of the incubation of drug craving, highlighting the value of several drug targets that have been well studied in animal studies and addressing the possibility and importance of pharmacological therapies, environment-based interventions, and neuromodulation in reducing the incubation of drug craving. We also propose translational obstacles, such as the scarcity of clinical research and different indicators of incubation in animals and humans, to provide insights into future investigations.

Potential drug targets

Most studies have focused on the molecular mechanisms that underlie the incubation of drug craving, and identified various potential drug targets covering kinds of neurotransmitters, neuropeptides, neurotrophic factors, and epigenetics (Fig. 1; Table 1) [30]. In the following parts, we mainly introduce several well-studied molecules in the field of incubation of drug craving.

Fig. 1: Promising drug agents for preventing the incubation of drug craving.
figure 1

Some agents that target neurotransmitter receptors exert inhibitory effects on incubated craving, including Flupenthixol (D1R and D2R antagonist), SCH39166 (D1R antagonist), raclopride (D2R antagonist), SB-277011A (D3R antagonist), LY376298 (mGluR2/3 agonist), AZD8529 (mGluR2 PAM), TBOA (excitatory amino acid transporter inhibitor), DHPG (mGlu1 agonist), Ro67–7476 (mGluR1 PAM), SYN119 (mGluR1 PAM), Naspm (CP-AMPAR blocker), and naltrexone (opioid receptor antagonist). Additionally, BDNF binds to TrkB and induces the activation of downstream signaling (including ERK, PI3K, and PKC) and phosphorylation of CREB, thereby mediating the incubation of craving during late abstinence. U0126 (ERK phosphorylation inhibitor), wortmannin (PI3K inhibitor), and everolimus (PI3K inhibitor) exert therapeutic effects on the incubation of drug craving. The inhibition of DNMTs or HDACs also prevents the incubation of drug craving. For example, microinjections of RG108 (DNA methyltransferase inhibitor) markedly reduces cue-induced cocaine craving during late abstinence.

Full size image
Table 1 Effect of various agents at different drug targets on the incubation of drug craving.
Full size table

Neurotransmitters

Glutamate system

Glutamate is the primary excitatory neurotransmitter in the central nervous system and plays a critical role in addiction and relapse [31]. Consistent with an increase in glutamate release in the ventromedial prefrontal cortex (vmPFC) after abstinence [32], a microinjection of mGlu2/3 agonist LY379268 in the prelimbic cortex (PrL) markedly reduced glutamates levels and prevented cue-induced cocaine craving during late abstinence [33]. However, the function of glutamate in the infralimbic cortex (IL) on the incubation of cocaine craving shows heterogeneity. Cue-induced cocaine craving during late abstinence was attenuated by increasing glutamate levels in the IL with the excitatory amino acid transporter inhibitor threo-b-benzyloxyaspartate (TBOA) [33], suggesting an association between the role of glutamate in incubation of cocaine craving and brain regions where it is released. Despite the brain region-specific role of glutamate, research has shown that systemic injections of LY379268 reduce cue-induced cocaine craving after 21 days of withdrawal [34], indicating that the targeted inhibition of glutamate by systemic administration can effectively prevent the incubation of cocaine craving.

More attention has been paid on the glutamate receptors in the field of incubated drug craving. Given that many excellent reviews have described detailed mechanisms [35, 36], here we provide only a brief introduction of how these findings would be translated into possible treatments. Glutamate receptors are divided into two types: ionotropic receptors and metabotropic ones [37]. Ionotropic glutamate receptors mainly consist of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPARs) and N-methyl-D-aspartate receptors (NMDARs). Of these, calcium-permeable AMPARs (CP-AMPARs) are critical for mediating incubation of drug craving. CP-AMPARs are mainly composed of GluR1, with high conductance and fast kinetics, contributing to synapses strengthening [35]. During long-term abstinence, the insertion of CP-AMPARs occurs in medium spiny neurons (MSNs) in the nucleus accumbens (NAc), accompanied by an increase in GluR1 levels but no changes in GluR2 [38, 39]. The blockade of CP-AMPARs by 1-naphthylacetylsperimine (Naspm) decreased excitatory postsynaptic currents (EPSCs) in MSNs in the NAc shell (NAcsh), which receives inputs from the basolateral amygdala (BLA) [40] and IL [41]. However, incubated cocaine craving was attenuated only when inputs from the BLA were inhibited [40], suggesting that the role of CP-AMPARs in the incubation of cocaine craving is influenced by upstream projections. CP-AMPARs insertion was also found in the NAc core (NAcc) in “incubated rats” after methamphetamine self-administration training. An intra-NAcc injection of Naspm significantly reduced incubated methamphetamine craving after 40 days of withdrawal [42]. These findings reveal the necessity of CP-AMPARs for the incubated craving of psychostimulant drugs, including cocaine and methamphetamine, but remaining unclear is whether CP-AMPARs play a similar role for other addictive drugs. Direct evidence is lacking, but research has found that an intra-BLA injection of Naspm dose-dependently decreased the rate of alcohol self-administration [43], raising the possibility that CP-AMPARs may also mediate the incubation of other addictive substances.

In addition to directly blocking the insertion of CP-AMPARs via Naspm, indirectly targeting metabotropic glutamate receptors (mGluRs) is more effective and has more translational values, due to the development of agents with fewer side effects. Given that the inhibitory effect of mGluRs on the accumulation of CP-AMPARs during abstinence [36], intraperitoneal and intracranial stimulation of mGluR2/3 in the central nucleus of the amygdala (CeA) significantly reduced cue-induced cocaine craving after prolonged withdrawal [34]. However, the side effects of these traditional mGluRs agonists, such as low tolerance [44], and unclear physiological functions [45], hamper their further clinical use. In recent years, development of novel mGluRs-targeting agonists has become a hotspot in the field of incubation of drug craving, among which positive allosteric modulators (PAMs) have received the most attention [46]. Different from orthostatic agonists that bind to extracellular N-terminal sites of mGluRs, PAMs bind the heptahelical transmembrane domain, allowing it to have higher selectivity for mGluR subtypes, fewer side effects, and better pharmacokinetic characteristics [47, 48]. PAMs also show a role in suppressing incubation of drug craving. After 35 days of withdrawal, cue-induced cocaine craving was attenuated by systemic injection of the mGluR1 PAM SYN119 before a relapse test or every other day from abstinence day 15 to day 33 [49]. Similarly, systemic administration of SYN119 also decreased incubated methamphetamine craving after 40 days of withdrawal [42], demonstrating the inhibitory effect of SYN119 on incubation of craving for psychostimulant drugs. Notably, the function of SYN119 in incubated craving is specific to cocaine and methamphetamine, in which no change in incubated cue-induced sucrose craving was found after a SYN119 injection [49]. These results indicate that SYN199 has a therapeutic effect on incubated craving for addictive drugs rather than natural rewards. In addition to SYN119, AZD8529, a mGluR2 PAM, has also been applied in reducing incubation of methamphetamine craving [50]. Despite the effectiveness of PAMs in preventing incubated craving, previous studies mainly confirmed the influence of mGluR PAMs on the incubation of craving for cocaine and methamphetamine. Remaining unclear is whether the administration of SYN119 or AZD8529 effectively reduces the incubated craving of other classes of addictive drugs, such as opioids, nicotine, and alcohol. Additional preclinical and clinical studies are needed.

Dopamine

Dopamine plays a key role in drug addiction [51], and its role in incubated drug craving has been gradually concerned in recent years. In the vmPFC, dopamine release time-dependently waned during abstinence in cocaine-trained rats [32], along with an increase in the levels of dopamine transporter (DAT) both at early and late abstinence time [52]. Consistently, maximal dopamine uptake rate in the NAc also increased after a prolonged abstinence of cocaine [53]. These results suggest an inverse relationship between enhanced cocaine craving during withdrawal and dopamine levels. When dopamine receptors are targeted, the inhibitory effect on incubation will occur. Microinjections of dopamine D1 or D2 receptor antagonists (1.0 μg SCH39166 per side or 1.0 μg raclopride per side) in the NAcc but not NAcsh markedly reduced cue-induced methamphetamine craving after 15 days of voluntary abstinence, suggesting the necessity of D1 and D2 receptors in the NAcc for the incubation of methamphetamine craving [54]. Similar results were also found that microinjection of SB-277011A (the selective D3 receptor antagonist) into the NAc or CeA decreased incubated cocaine craving [55], demonstrating the importance of mesolimbic dopamine system for incubation of craving for psychostimulant drugs. Unexpectedly, the systemic administration of SB-277011A inhibited cue-induced cocaine craving after 2, 10, and 30 days of abstinence [55], and sucrose seeking after 30 days of withdrawal was also attenuated by a systemic injection of the D1 receptor antagonist SCH 23390 (25 μg/kg) [56]. These results raise the possibility that the role of the dopamine system is not specific to incubated craving during abstinence, but it is necessary for the motivational response to drug- and non-drug-related cues during abstinence.

Neuropeptides

Endogenous opioid system

The endogenous opioid system includes opioid peptides and receptors, contributing to drug craving and relapse [57]. Among various opioid peptides, β-endorphin shows a role in decreasing incubated cocaine craving. In 2013, Dikshtein et al. [58] used an enzyme-linked immunosorbent assay, and found that β-endorphin level in the NAc was higher after exposure to cocaine-related cues on abstinence day 1 than that on day 30. Microinjections of a synthetic β-endorphin peptide in the NAc decreased cue-induced cocaine craving on abstinence day 30, which can be reversed by co-injection of the δ-opioid receptor antagonist naltrindole. This finding not only suggests that higher level of β-endorphin in the NAc can inhibit incubated cocaine craving through binding to the δ-opioid receptor, but also provide a new drug target for treatment, that is the opioid receptors.

The opioid receptors consist of three kinds, μ-, δ-, and κ-receptor [57], and opioid receptors antagonists have been examined potential for treatment of substance use disorder for many years [59], among which naltrexone is a promising agent for reducing incubated craving. Naltrexone is an opioid receptor antagonist, and chronic injections of naltrexone (7.5, 15, and 30 mg/kg/day) via minipumps from abstinence day 1 to day 13 significantly suppressed incubated heroin craving [60]. Additionally, subcutaneous injections of naltrexone at a dose of 1 mg/kg also reduced incubation of heroin craving. However, subcutaneous injection at doses of 15 and 30 mg/kg failed to decrease incubation [61], not only suggesting a dose-dependent effect of naltrexone on incubation of heroin craving, but also indicating an association between the naltrexone effect and the delivery methods. Notably, the therapeutic effect of naltrexone does not apply to all drugs of abuse. Research has showed that incubation of methamphetamine craving did not change by naltrexone injection either immediately before the test or chronically during abstinence [60]. These findings suggest that naltrexone may only be suitable for treatment of incubated opioids craving. But clinical research has found more possibilities of naltrexone in incubation treatment, and we will talk about it in the later section (see “Opportunities with translational values” section below).

Hypocretin/orexin

Hypocretin (Hcrt; also called orexin) plays an important role in reward-driven motivation, and there are some researches reporting its role in drug relapse [62, 63]. During abstinence from cocaine, activated Hcrt-positive neurons significantly increased in the lateral hypothalamic subregion, and this activated state persisted during withdrawal for at least 150 days [64]. In Hcrt knockout mice, although cue-induced cocaine craving remained elevated during the first week of abstinence, craving failed to incubate beginning in the second week [65], indicating that Hcrt mainly mediates cue-induced cocaine craving at late abstinence time points.

Neurotrophic factors

Neurotrophic factors play an important role in cell growth and synaptic plasticity [66], such as brain-derived neurotrophic factor (BDNF), glial cell line-derived neurotrophic factor (GDNF), and fibroblast growth factors (FGFs). As secreted proteins, these neurotrophic factors transmit signals by binding to transmembrane receptors and triggering activation of various downstream signaling pathways (e.g., extracellular-signal-regulated kinase [ERK]/mitogen-activated protein kinase [MAPK] pathway, protein kinase C [PKC] pathway, and phosphatidylinositol 3-kinase [PI3K] pathway), thereby activating secondary transmitter systems and inducing the transcription of related genes to mediate neuroplasticity [67].

BDNF

Studies of incubation of drug craving have reported a critical role of the BDNF signaling pathway. After extended cocaine self-administration, the protein and mRNA levels of BDNF in the ventral tegmental area (VTA) increased from the abstinence day 7 [68], and can persist until day 90 [69]. Along with that, the surface expression of tyrosine kinase receptor B (TrkB), the specific receptor of BDNF, also increased [70]. Similar increases in BDNF levels were also found in other brain regions, such as the NAc and amygdala in “cocaine-incubated rats” [69, 70] and mPFC in “heroin-incubated rats” [71]. However, the role of BDNF in incubation of drug craving is inconsistent in different brain region. For example, heterogeneous effects of BDNF-TrkB pathway on incubation of cocaine craving occur in the NAcc and NAcsh. Li et al. [70] found that cue-induced cocaine craving after 90 days of abstinence was strengthened by TrkB knockdown in the NAcc, whereas opposite result was found after blocking BDNF-TrkB binding in the NAcsh on abstinence day 90.

Downstream targets of neurotrophic factors

In addition to directly targeting BDNF, indirect manipulations of its downstream factors also have potential to decrease incubation of drug craving. Various molecules are successively activated by BDNF, such as ERK, the active state of which relies on direct phosphorylation [66]. After 2 weeks of abstinence, incubated craving for morphine and cocaine strengthened the phosphorylation of ERK in the CeA, together with activation of its downstream transcription factor cyclic adenosine monophosphate (cAMP)-response element binding protein (CREB) [22, 72]. Consistently, microinjections of the ERK phosphorylation inhibitor U0126 in the CeA significantly decreased incubation of morphine or cocaine craving [22, 72], suggesting the potentiality of U0126 in treatment of incubated craving for psychostimulant drugs and opioids. ERK activity in the VTA is also required for GDNF-induced potentiation of cue-induced cocaine seeking after withdrawal [73]. Besides, PI3K is an important downstream pathway triggered by BDNF-TrkB activation [67]. Either intracranial injections of the PI3K inhibitor wortmannin [74] or single oral dosing with the FDA-approved PI3K inhibitor everolimus [75], suppressed incubated cocaine craving in rats. Notably, the effect of everolimus on decreasing incubated cocaine craving lasted 24 h [75], indicating that everolimus has immediate and long-term therapeutic effects on incubation of cocaine craving.

Epigenetics

In recent years, the role of epigenetics in drug addiction and relapse has attracted much attention [76]. Several studies have investigated the mechanisms that underlie the incubation of drug craving from the perspective of epigenetics, identifying activated DNA methylation in the corpus callosum and the higher mRNA expression of some epigenetic enzymes in the dorsal striatum after prolonged abstinence from cocaine [77]. Consistently, intracranial injections of the DNA methyltransferase inhibitor RG108 in the NAc decreased cue-induced cocaine craving after 30 days of abstinence [78], indicating the potential of targeting DNA methylation for suppressing incubation of cocaine craving. Moreover, histone deacetylases (HDACs), which restrain gene transcription [79], have been reported to time-dependently increase during long-term abstinence from methamphetamine [80]. And incubated methamphetamine craving can be attenuated by inhibition of HDAC5 with short-hairpin RNA. These results reveal the importance of epigenetic marks for decreasing the incubation of psychostimulant drugs.

These findings identify some promising targets that may be exploited to explore the function of existing drugs and develop novel agents with fewer adverse effects. More importantly, several potential agents, such as SYN119, naltrexone, and everolimus, showed favorable effects in decreasing incubated craving for some addictive drugs through systemic administration. In the next section below, we describe evidence of the incubation of drug craving in humans and several obstacles that need to be overcome in translational research.

Translation of incubation of drug craving

Measures

To date, the dominant method for detecting drug craving in humans has involved subjective measures. Participants are exposed to neutral and drug-related cues, such as visual and olfactory stimuli, and subjective measures are collected before and after cue exposure. The incubation of drug craving in humans was first found in cigarette smokers [16]. Using a between-subjects design, three groups of participants were abstinent for 7, 14, and 35 days, and a single cue session was conducted on the final abstinence day. In the within-subjects design, one group of participants was abstinent for 35 days, and cues were presented on abstinence days 7, 14, and 35. Subjective measures were collected through questionnaires, such as the Tobacco Craving Questionnaire–Short Form (TCQ-SF) Total and the Brief Questionnaire of Smoking Urges (QSU-B) Factor 1 and 2 subscales. The results showed that craving in response to smoking-related cues on the TCQ-SF increased with abstinence in the between-group analysis, and the incubation of cue-induced craving was weaker in the within-subjects assessment, thus identifying the incubation of nicotine craving in humans. Additional clinical evidence has subsequently emerged, demonstrating an increase in cue-induced craving for methamphetamine [23] and alcohol [17, 19] with abstinence. However, some surprising results on subjective self-reported cue-induced craving have been found in heroin and cocaine, which have already been confirmed in animal studies. Measured using a Visual Analog Scale (VAS), craving for heroin-related cues was not different after 1, 3, 12, and 24 months of abstinence [81]. This is likely attributable to the prolonged interval between abstinence time points, which likely missed the time at which cue-induced craving may be elevated. Cue-induced cocaine craving, which was measured by answering the question, “Rate how much you want (or do not want) cocaine in response to this picture,” exhibited a linear decline from 2 days to 1 year of abstinence [12]. This inconsistency may have resulted from the different methods that were used to quantify craving [12]. The above study defined “cue-induced craving” as the rating when participants viewed the drug cues, whereas other previous studies usually quantified “cue-induced craving” as differences between craving before and after cue sessions. Although these findings demonstrate the availability and convenience of subjective measures for detecting incubated craving in humans, self-reports are limited by some factors, such as compromised self-awareness [82] and social desirability bias [83].

To provide more objective evidence of the incubation of drug craving in humans, electroencephalography (EEG) was applied to detect late positive potential (LPP) to track the incubation of cocaine craving [12]. The LPP is electrocortical positivity that can be elicited by emotional and neutral stimuli, with higher levels in response to emotional cues than in response to neutral cues [84, 85]. In cocaine-dependent patients, the LPP became larger in response to cocaine-related cues and exhibited a robust association with cue-induced cocaine craving that was measured by self-reports [86]. Parvaz et al. [12] used the LPP as an objective marker of cue-induced craving. The study recruited 76 participants with cocaine use disorder and divided them into five groups according to self-reported abstinence length (approximate average abstinence time: 2 days, 1 week, 1 month, 6 months, and 1 year). Age, sex, lifetime duration of cocaine use, and dependence severity were matched between groups. Participants viewed four types of pictures, including pleasant, unpleasant, neutral, and cocaine-related pictures, during which the LPP was detected by EEG. Following EEG recordings, the subjects’ levels of self-reported cue-induced craving were also recorded. The results showed that the mean LPP-indexed drug-cue reactivity was higher at 1 and 6 months of abstinence and lower at 2 days, 1 week, and 1 year of abstinence, exhibiting an inverted-U trajectory during abstinence. This finding is consistent with the incubation of cocaine craving in rodents, suggesting that the LPP is a potential objective marker of incubated cocaine craving. Notably, however, self-reported cocaine craving in response to related cues declined with abstinence in this study. The decline in cue-induced craving may stem from the different statistical methods that were applied. The opposite findings between objective recordings and subjective measurements also raise a question. Does the increase in LPP amplitude during abstinence reflect changes in cue-induced craving? Future investigations should study associations between the LPP and cue-induced craving in abstinent individuals by utilizing more spatiotemporal-specific techniques, such as EEG-informed functional magnetic resonance imaging (fMRI).

Obstacles

Translational research on the incubation of drug craving is already underway, but some important issues need to be considered. The largest problem is the lack of adequate clinical research. It has been 21 years since the term “incubation of drug craving” was first proposed and about 10 years since the first clinical study that verified “incubated cue-induced craving” in humans. However, related clinical research is still rare, which may be attributable to inconsistent and inaccurate incubation time windows for different types of drugs, thereby making it difficult to detect this phenomenon clinically. For the incubation of methamphetamine craving, the highest cue-induced craving occurs at 3 months of abstinence (Fig. 2). For the incubation of cocaine craving, in contrast, cue-induced craving is highest after 2 months of abstinence. Although cue-induced alcohol and nicotine craving was higher at abstinence days 35 and 60, respectively, the exact time of intensification cannot be precisely defined because of the lack of data on longer abstinence periods. This uncertainty makes clinical measurement difficult. Unknown is whether abstinent individuals are at the peak of cue-induced craving for predicting relapse risk. A critical question is which abstinence times are the best phase for drug administration. Rodent studies mostly chose a late abstinence time point as the timing for intervention. The translational value of this single dosing timeframe in animal studies is quite small, because it is still difficult to pinpoint the exact time point of incubation in abstinent individuals. More attention should be paid to results of interventions that are implemented at early abstinence time points. For example, naltrexone exerts a therapeutic effect on incubated craving after chronic administration during early abstinence periods. Future studies should more clearly define the time window of incubation to determine the best time window of treatment.

Fig. 2: Incubation of drug craving in humans.
figure 2

A Incubation of drug craving in humans detected by subjective measures, showing ratings of craving in response to drug-related cues. The data are expressed as the mean number of rating scores at different abstinence time points for nicotine [16], methamphetamine [23], and alcohol [17]. B Incubation of cocaine craving in humans detected by electroencephalography (EEG) [12]. The data are shown as cue-induced late positive potential (LPP) amplitude at different days of abstinence. *Different from abstinence day 6 or 7. #Different from abstinence day 14. The data were extracted by the WebPlotDigitizer (https://automeris.io/WebPlotDigitizer/) from the above cited references, and redrawn by the Graphprism9.4.1.

Full size image

Another obstacle to translational research is that the definition of “cue-induced craving” in animals is not exactly the same as the clinical definition of “drug craving.” In humans, self-reports are widely used to measure the degree to which individuals want drugs when they are exposed to drug-related cues. In animal studies, because it is impossible to measure their level of “desire”, behavioral indicators are used to measure craving, such as nosepoke or lever-press responding. Thus, what we detect in rodents is actually the behavioral response to drug-related cues that is driven by desire or motivation, rather than the desire itself. Moreover, detailed neural mechanisms are heterogeneous between animals and humans [87], raising questions of whether findings from animals can be translated to humans and whether interventions that work in animals will also work in clinical trials. In addition to advancing clinical research, we also need to explore whether there are more objective metrics of incubated drug craving that can be detected in both animals and abstinent humans.

Although investigations focusing on incubation of drug craving in humans, especially these about the treatment, is still scarce, some promising therapies merit more attention because they have a good translational potential. In the next section, we describe more possible translational paths for moving forward.

Opportunities with translational values

In addition to well-studied drug targets in the field of incubation of drug craving, other kinds of interventions have also been emerging, including sleep manipulation [88], environmental improvement procedures [89], exercise [90, 91], voluntary abstinence [20, 92,93,94], and extinction sessions (Table 2) [95]. Of these, environment-based interventions gets more attention in recent years, due to its consideration of social factors which are critical for addiction and relapse [96]. Meanwhile, with the research about neurophysiological signatures of incubated craving in individuals with methamphetamine use disorder [24], neuromodulation also becomes a promising method for reducing incubation of drug craving.

Table 2 Effect of various behavioral therapies on the incubation of drug craving.
Full size table

Pharmacological interventions

Although several promising drug targets have been emerging, related clinical research is scarce, but still give us partially preliminary support in screening effective medications. In 2020, Bach et al. [97] found that naltrexone exerted a promising therapeutic effect on incubation of alcohol craving. 55 male alcohol-dependent patients and 35 healthy controls were recruited. Following about 3 weeks of controlled abstinence, the participants received either 21 days of Intensive Withdrawal Treatment (IWT, including occupational therapy, physical activation, psychoeducation, psychological group therapy, and multiple medical rounds [98]) or IWT with adjunct oral naltrexone. Instead of using subjective measures to detect cue-induced craving, the author applied a fMRI task to measure the neural response to alcohol-related cues. In the control group, the mesolimbic alcohol cue reactivity was enhanced over the first weeks of abstinence, consistent with incubation of alcohol craving. Individual who received naltrexone exhibited lower cue reactivity in the left putamen, along with a decrease in relapse (relapse was defined as that the alcohol consumption exceeded 60 g/day for men) risk. Despite that this study only included male subjects and did not measure self-reported cue-induced craving, the results suggest the potentiality of naltrexone for reducing incubated alcohol craving. Naltrexone has been approved by the FDA for weight loss medications in obese people [99] and the treatment for smoking behavior and alcoholism [100, 101]. But some clinical studies have reported adverse effects of naltrexone, such as vomiting, nausea, and abnormal pain, and during the early administration, some patients have reported fatigue and headache [102, 103]. Various side effects of medications are an inescapable problem, limiting their clinical use. In spite of this problem, it is still worthy to develop novel agents with fewer adverse effects, and the occurrence of mGluRs PAMs is a good example.

Environment-based interventions

Environmental improvements show potential and many studies have found that favorable living conditions, such as environmental enrichment, can reduce cue- and stress-induced drug craving [104, 105]. Incubation of cocaine craving can also be attenuated by environmental improvement procedures [89]. However, this therapeutic effect is unstable and disappears if these enriched environments are removed [89], demonstrating the importance of continually improving living conditions to prevent incubation of drug craving. Moreover, several studies have explored the effects of a kind of environmental enrichment where rodents are allowed to exercise, and cue-induced cocaine or nicotine craving at late abstinence was significantly attenuated [90, 91], suggesting that exercise is helpful for reducing incubated drug craving.

Voluntary abstinence models are good examples of “reverse translation,” such as punishment-induced voluntary abstinence, food-choice voluntary abstinence, and social-choice voluntary abstinence, considering that abstinent individuals do not go through withdrawal without access to anything. However, the effect of voluntary abstinence on incubated drug craving is questionable because after punishment-induced voluntary abstinence, methamphetamine- and oxycodone-trained rats still showed higher cue-induced craving at late abstinence time points [94, 106]. Food-choice voluntary abstinence, in which rats have access to food by pressing a lever, did not decrease incubated methamphetamine craving during late abstinence [50, 107]. Incubated heroin craving in both male and female rats was reduced by food-choice voluntary abstinence, consistent with clinical findings in humans with heroin use disorder [108,109,110], suggesting a role for food in combating heroin craving [92, 93].

Of these, social-choice voluntary abstinence gets the most attention, which is a novel abstinence model with potential implications for preventing incubation of drug craving [111]. Compared with forced abstinence, social-choice voluntary abstinence allows drug-dependent rats to choose between a drug and contacting familiar peers by pressing levers during abstinence. This manipulation markedly reduced incubation of methamphetamine and heroin craving, and the therapeutic effect was independent of gender and abstinence duration [112, 113]. These studies suggest the critical role of social reward in decreasing incubated craving during abstinence, highlighting the importance of positive social interaction for preventing drug craving and treatment for relapse. Despite the lack of research on the influence of voluntary abstinence in abstinent humans, similar interventions for the treatment of individuals with substance use disorders already exist, possibly providing some important insights. For example, the community reinforcement approach (CRA), usually combined with contingency management, applies familial, social, recreational, and occupational reinforcers to help people with substance use disorder achieve full abstinence [114]. During the course of the CRA, there is a “relationship counseling” component, in which clients practice communication with their partner and rebuild a good relationship [115]. By reconstructing a healthy lifestyle, abstinent individuals may gradually rid themselves of addictive drugs. However, the social environment in humans is highly complex, and not everyone would rank social rewards first. Thus, medication combined with behavioral interventions may be more promising. Some indirect evidence shows the possibility of naltrexone combined with the CRA for decreasing drug craving. For example, 272 heroin-dependent patients were recruited and received 10 months of naltrexone combined with a CRA. The results showed that self-reported craving decreased during the 16-month follow-up period [116]. Although this study did not measure cue-induced heroin craving during abstinence, the findings raise the possibility that a modified CRA helps attenuate incubated craving in humans.

Neuromodulation

Neuromodulation methods have drawn much attention in recent years, which aim to probe, target, and remodel impaired brain circuits [117]. To date, the best developed devices for neuromodulation include transcranial direct current stimulation (tDCS) [118] and repetitive transcranial magnetic stimulation (rTMS) [119], which act on superficial brain regions through surface electrodes or stimulation coils that are affixed to the scalp. In contrast, deep brain stimulation (DBS) is an invasive therapy that implants microelectrodes into deeper brain regions [120]. Numerous studies have identified the efficiency of neuromodulation on suppressing drug craving (Box 2) [121, 122]. But unclear is remain whether these techniques can be applied for decreasing incubated cue-induced craving in abstinent individuals.

One advantage of neuromodulation techniques to treat the incubation of drug craving is that animal studies have identified potential targets of action, such as the mPFC, NAc, and amygdala [29]. For example, using the Daun02 chemogenetic inactivation procedure, the selective inhibition of neuronal ensembles in the dorsomedial striatum reduced incubated methamphetamine craving [107]. Additionally, the circuit from BLA-NAcsh [40], or circuit from PrL-NAcc [41], can be modulated by optogenetic procedures to suppress the incubation of cocaine craving. These results suggest the essential role of the NAc in the incubation of drug craving, which may be a promising target for neuromodulation via DBS.

Although direct investigations are still lacking, preliminary experiments [24] have provided some support for future studies that utilize neuromodulation to reduce incubated craving. For example, 156 male methamphetamine-dependent individuals were recruited and divided into five groups based on self-reported abstinence duration: <1 month, 1–3 months, 3–6 months, 6–12 months, and 12–24 months. After rating their drug craving in response to methamphetamine-related videos, resting-state 128-high-density channel EEG signals were recorded. Along with that cue-induced methamphetamine craving got higher at 1–3 months of abstinence, the power spectrum for θ (5.5–8 Hz) and α (8–13 Hz) declined, and β (16.5–26.5 Hz) increased. Notably, an association was found between the incubation of methamphetamine craving and the increase in β activity in the mPFC. Previous studies confirmed the potential role of targeting the mPFC via rTMS or tDCS to treat substance use disorder [123, 124]. Future studies should investigate whether targeting β activity in the mPFC via tDCS or rTMS is useful for decreasing cue-induced craving during abstinence.

Additionally, some indirect studies have shown that the application of neuromodulation techniques can reduce cue-induced drug craving during abstinence. In a multicenter double-blind randomized controlled trial, 262 chronic smokers with tobacco use disorder were recruited and randomly divided into two groups: active rTMS group (who received 3 weeks of daily bilateral active rTMS of the lateral prefrontal and insular cortices followed by once weekly rTMS for 3 weeks) and sham group [125]. Cue-induced craving, measured by a VAS, significantly decreased in the active rTMS group compared with the sham group, and the average weekly reduction of TCQ total score was also significantly greater in the active rTMS group. Although this study did not investigate whether cue-induced craving would be at a lower level at longer abstinence time points, the findings still suggest that rTMS may be applied from early abstinence time period to prevent the incubation of nicotine craving.

Other interventions

In addition to the above three kinds of interventions, other methods are also worthy further investigation. For example, Chen et al. [88] reported that incubated cocaine craving during late abstinence markedly decreased after chronic sleep restriction, supporting the effectiveness of sleep manipulation. And the other promising way is 7-day cue-extinction session performed during abstinence, which suppressed incubation of cocaine craving [95]. Although rare studies focus on the effectiveness of these interventions on decreasing incubation of drug craving, it possibly provides some insights for exploring the mechanism from new perspectives such as sleep and memory. And some potential treatments for preventing drug relapse which target the memory process have already been developed, such as the retrieval-extinction procedures [126, 127].

Compared with pharmacological interventions, behavioral therapies are characterized by safety and no adverse effects. As for neuromodulation, milder side effects than pharmacological interventions and more direct modulation relative to behavioral therapies make it more promising for clinical application. More advanced neuromodulation techniques should also be applied to treat substance use disorder and relapse in future studies, such as transcranial ultrasound brain stimulation [128], closed-loop systems [121], and adaptive DBS [129]. Combined therapies will also be a good research direction, such as neuromodulation plus behavioral therapies and pharmacological therapies plus behavioral interventions. Combinations of multiple treatment modalities might make up for shortcomings of one type of therapy and have superior efficacy.

Conclusions and perspectives

Drug addiction is a complex psychiatric disease that interrupts various systems [130], along with a high rate of relapse even after treatment [1]. The relapse rate for individuals with opioid use disorders is over 60% after 1 year of abstinence, and over 80% for cocaine users [131]. High risk of relapse has become the primary problem for treating substance use disorder. Among kinds of factors inducing relapse, cue-induced drug craving is a critical one, and exhibits an increasing trend during the first several months of abstinence, which is termed as incubation of drug craving. Using subjective measures and EEG, several clinical studies have verified this phenomenon in nicotine, methamphetamine, alcohol, and cocaine. Numerous excellent investigations have explored the mechanisms, not only providing potential pharmacological therapies but also proposing preliminary supports for social-choice voluntary abstinence and neuromodulation techniques.

The following two points should be noted in future research. First, gender differences have received more attention in recent years in the field of incubation of drug craving [132]. Animal studies have found that incubated cue-induced drug craving is also present in abstinent female rats, but evidence of sex differences is mixed in studies of the incubation of cocaine craving. In studies that found sex differences, female rats exhibited higher and longer-lasting incubation [13, 133,134,135]. Remaining unclear is whether possible sex differences exist between abstinent men and women. Most clinical research on the incubation of drug craving has recruited only male participants. Whether gender affects the efficacy of interventions in reducing the incubation of drug craving is also unknown.

Second, there is heterogeneity in the mechanisms that underlie incubated craving between different drugs of abuse. For example, incubated heroin craving in rats and incubated alcohol craving in humans can be attenuated by systemic naltrexone administration, whereas no effect on the incubation of methamphetamine craving was found. This difference may stem from the different mechanisms of action between psychostimulants, opioids, and alcohol. Notably, detailed mechanisms of incubated craving during abstinence may also be influenced by the quality or degree of prior drug exposure. Therefore, in addition to developing interventions, exploring whether there are common or distinct mechanisms that underlie the incubation of craving for different types of addictive drugs is also a future research direction.

In conclusion, the present review summarized promising drug targets that have been well studied in rodents, and highlight the potential of pharmacological interventions, social-choice voluntary abstinence and neuromodulations. We hope that the presented information will be useful for developing more effective treatments, investigating underlying mechanisms from new perspectives, and promoting translational research in the future (Box 3).