The Impact of Co-occurring Post-traumatic Stress Disorder and Substance Use Disorders on Craving: A Systematic Review of the Literature

The frequent co-occurrence of post-traumatic stress disorder (PTSD) and substance use disorders (SUDs) leads to manifestations of both conditions that are more severe and more resistance to treatment than single disorders. One hypothesis to explain this synergy is the impact of intrusive memories on craving which, in turn, increases the risk of relapse among patients with substance use disorders. The aim of this systematic review is to examine this possibility by assessing the impact of PTSD and its symptoms on craving among dual disorder patients. Using PRISMA criteria, four databases were comprehensively searched up to June, 2021, in order to identify all candidate studies based on broad key words. Resulting studies were then selected if they examined the impact of PTSD or PTSD symptoms on craving, and if they used standardized assessments of PTSD, SUD, and craving. Twenty-seven articles matched the selection criteria and were included in this review. PTSD was found to be significantly associated with increased craving levels among patients with alcohol, cannabis, cocaine, tobacco, and other substance use disorders. Exposition to traumatic cues among dual disorder patients was also shown to trigger craving, with an additive effect on craving intensity when exposure to substance-related cues occurred. In addition, certain studies observed a correlation between PTSD symptom severity and craving intensity. Concerning mechanisms underlying these associations, some findings suggest that negative emotional states or emotion dysregulation may play a role in eliciting craving after traumatic exposure. Moreover, these studies suggest that PTSD symptoms may, independently of emotions, act as powerful cues that trigger craving. These findings argue for the need of dual disorder treatment programs that integrate PTSD-focused approaches and emotion regulation strategies, in addition to more traditional interventions for craving management.


INTRODUCTION
The diagnosis of Post-Traumatic Stress Disorder (PTSD) first appeared in the 3rd edition of the Diagnostic and Statistical Manual of Mental Disorders in 1980 (1), based largely on clinical descriptions of soldiers returning from the Vietnam war (2). Its definition has evolved considerably over recent decades, including its removal from the anxiety disorders in DSM 5 as well as the creation of a distinct diagnostic category for this disorder (3). PTSD is characterized by a variety of symptoms that persist over the months or years following a traumatic event and that notably include intrusive memories, avoidance of cues associated with the event, alterations of cognition and mood, and a state of hyperarousal. Although diverse mental disorders are frequently associated with PTSD, substance use disorders (SUDs) are particularly prevalent (4)(5)(6). The principal hypotheses that have been formulated to explain these associations include self-medication (implying that PTSD is the primary condition and that substance use disorders occur later), the notion that addiction to substances may constitute a risk factor for the occurrence of traumatic events (whereby PTSD is a secondary condition), and finally the possibility that both disorders share common vulnerability factors (7). Regardless of which mechanism best explains these forms of dual disorder, the combination of PTSD and addiction leads to poorer prognosis, increases in suicide attempts, greater social disability, poorer treatment adherence, and reduced medication efficacy when compared to patients without comorbidity (8)(9)(10). In line with these results, a recent review of clinical investigations also documented a strong relationship between the diagnosis of PTSD and increased substance use and relapse in dual disorder individuals (11), but again the exact mechanisms underlying this association remain unclear.
Craving has been studied extensively over the years and particularly over the past two decades due to acknowledgment of its crucial role in addiction (12,13). Craving refers to the intense, urgent, and unwanted desire to consume a substance (14) and it is now considered to be a core component of addiction with important diagnostic implications following its inclusion in DSM-5. Based on findings that demonstrate a prospective link between craving episodes and substance use, craving is increasingly viewed as a central construct in the etiology and course of different forms of addiction, and it is a strong predictor of treatment outcome (15)(16)(17)(18)(19). Among the diverse factors that may affect craving, a large body of research has highlighted the major role of substance-related cues and stress (20)(21)(22)(23)(24)(25)(26)(27)(28)(29). These investigations have shown the ability of substance-related cues and stress exposure to elicit craving among individuals with alcohol, opiate, cocaine, tobacco, and cannabis dependence. Moreover, laboratory studies have also shown that exposure to stress-related events among individuals with alcohol use disorder (AUD) reliably elicits craving in a manner that is as powerful as alcohol-related cues (30,31). Although similar patterns of reactivity have been shown among individuals with PTSD and alcohol use disorder after exposures to personalized trauma cues via "trauma scripts, " such scripts were found to provoke greater craving that non-trauma scripts and to be more salient in eliciting alcohol craving (31,32). These results could suggest that the intrusive memories experienced by persons with PTSD and the significant stress they induce may therefore constitute major triggers of craving as well as explain reductions in treatment efficacy in this population. This pattern of findings is consistent with the findings that patients in SUD treatment who report higher PTSD scores also report higher scores on craving, depression, anxiety and stress (33), with a potential relationship between PTSD severity, SUD severity and craving levels.
One hypothesis to explain the synergy of PTSD and SUDs as a dual disorder is therefore the impact of intrusive memories or trauma-related cues on craving which, in turn, increases the risk of relapse among patients with substance use disorders. Examining this relationship across different forms of substance addiction should help elucidate the mechanisms underlying the general increase in clinical severity in this population, and the literature on this topic is now of sufficient size to permit a reliable summary that should more fully respond to the goals of precision psychiatry and personalized medicine (34). The aim of this systematic review is to address this issue by assessing all published investigations of the impact of PTSD and its symptoms on craving, among dual disorder patients.

Research Design
The study involved a systematic review of the literature based on the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines (35).

Information Sources
This review was based on the following databases: PUBMED/MEDLINE, Psychinfo, Cochrane, and Wiley Online Library. The search was performed for all years up to June, 2021.

Eligibility Criteria
The following criteria were used to select investigations for this review:

Studies
Published in English-Language peerreviewed journals. 2. Studies concerning patients, with no restrictive criteria regarding age, sex, ethnic origin, or place of residence. Studies had to include participants with PTSD and SUD comorbidity, defined, or explored according to standardized questionnaires. 3. Studies including measures of craving, and assessing the impact of PTSD or PTSD symptoms on craving occurrence or severity. It was not necessary that craving was designated as the primary outcome of the study in order for it to be included in this review.
Studies were excluded if they were based on animal models, or if they were limited to conference abstracts, dissertations, book chapters, or incomplete articles.

Study Selection
Two authors independently examined all titles and abstracts. Relevant articles were obtained in full-text and assessed for inclusion criteria separately by the two reviewers based on the inclusion and exclusion criteria previously mentioned. Disagreements were resolved via discussion of each article for which conformity to inclusion and exclusion criteria were uncertain and a consensus was reached. The reference lists of major papers were also manually screened in order to ensure comprehensiveness of the review. All selected studies were read in full to confirm inclusion criteria, study type, and study population.

Quality Assessment
Two reviewers (LJ and MF) assessed the quality of data in the included studies using the Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies (S2C) from National Institutes of Health (36). This tool is comprised of 14 questions with responses to each being "yes, " "no, " or "other" (not applicable, NA or nor reported, NR). We rated the overall quality of each included study as "good, " "fair, " or "poor."

Collecting Data
Sample characteristics (including socio-demographic data, comorbidity, and treatment status), and information on study design and methods of assessment of PTSD, SUD, and craving were extracted. Table 1 presents these data extracted from the selected studies.

Study Selection
A total of 247 articles were identified through the search of the databases. After review of titles and abstracts, 52 articles were selected for further examination. After reading the full text, 27 met inclusion criteria for this review. This process is described in the PRISMA flowchart (Figure 1). The selected articles were published between 2002 and 2021.

Quality Assessment
A summary of risk of bias is presented in Table 2. Eleven studies were considered to be of "good" quality, six were "fair" quality and nine of "poor" quality.

Study Characteristrics
Twenty-seven studies fulfilled criteria for inclusion in this review, of which 12 focused on alcohol, 4 on tobacco, 1 on cannabis, 1 on cocaine, and 9 on various substance use disorders (three studies on AUD and/or Cocaine Use Disorder and six studies on different types of SUD). Among the 27 included studies, 13 were experimental studies, 13 were observational studies and 1 was a randomized controlled trial.
A detailed description of all studies included and their main results can be found in Tables 3, 4.

Effects of Traumatic Cue-and Stress-Exposure on Craving Across SUD Subgroups
The 13 experimental studies selected for this review consisted, for most part, of exposing participants with comorbid SUD and PTSD to traumatic memories, non-specific stressors, and substance-related cues, and then evaluating their responses across SUD subgroups. Seven experimental studies involved AUD, two involved tobacco use disorder, one involved cocaine use disorder, and three studies included patients suffering from AUD and Cocaine Use Disorder.

Alcohol Use Disorder
The primary finding was that exposure to a traumatic memory (in the form of a script recounting a traumatic life event) generated a significantly greater increase in craving than neutral exposure and similarly to exposure to an alcohol-related cue (31,(38)(39)(40). The studies by Coffey et al. (31) and Nosen et al.'s (38) went further, showing that the combination of exposure to a traumatic script followed by an alcohol-related cue generated greater craving than each type of exposure when considered separately. Two studies showed that exposure to a traumatic memory increased craving more than a non-specific stressor (39,40).
Only one study compared subjects with the comorbidity AUD and PTSD to subjects with AUD alone (41). This study found no significant difference between the two groups in terms of craving intensity after exposure to a non-specific stressor. Ralevski et al. (40) found no correlation between the intensity of craving provoked by the different scripts (traumatic, non-specific stress, and neutral) and the severity of PTSD symptoms, measured by the Clinician Administered PTSD Scale (CAPS). Finally, Schumacher et al. (42) showed that subjects who had suffered early childhood trauma (<13 years of age) presented a more severe AUD, traumatic intrusion symptoms, and post-exposure craving (traumatic script and alcohol-related cues), than subjects who suffered their first trauma later in adolescence (13-18 years of age).

Results
Presented according to substance type FIGURE 1 | PRISMA flowchart of selected abstracts and articles.

Tobacco Use Disorder
Two experimental studies addressed tobacco use disorder. Beckham et al. (43) showed that the increase in craving, negative affect and traumatic symptoms were stronger after being exposed to traumatic scripts than stressful event scripts and neutral scripts. Subjects with PTSD experienced a more significant increase in craving and negative affect compared to the other group. Cigarette use was associated with a reduction in craving (which was greater in magnitude after a traumatic script), as well as reduction in negative affect, and traumatic symptoms, independently of group type. Dedert et al. (44) found that subjects with PTSD presented more severe withdrawal symptoms and a higher craving level for two dimensions (anticipation of reinforcing effect, anticipation of withdrawal symptoms and negative affect release) during acute withdrawal. Participants with PTSD reported lower craving reductions after smoking.

Cocaine Use Disorder
One experimental study by Tull et al. (45) assessed the effect of exposure to a personalized trauma cue on cocaine craving in patients with cocaine use disorder with or without PTSD. Subjects with PTSD had significantly higher craving for cocaine than other participants after the traumatic script.

Multiple Type of Substance Use Disorder: Alcohol and/or Cocaine Use Disorder
Three experimental studies exposed subjects with alcohol and/or cocaine addiction and a history of traumatic events to combinations of traumatic or neutral scripts and substancerelated or neutral cues (32,46,47). All studies showed that exposure to traumatic memories and substance-related cues increased craving and negative affect significantly more than neutral exposures. The studies of Saladin et al. (47) and McHugh  Experimental study. Participants were exposed to traumatic script and alcohol-related cues, then craving was assessed. The authors analyzed the correlations between age of first trauma (before 13 years old vs. 13-18 years old), severity of PTSD and AUD, and post-exposure craving.
PTSD diagnosis with CAPS according to DSM-IV. Presence of A PTSD criterion of DSM-IV with National Women's Study (NWS) PTSD module. PTSD severity assessed with the Impact of Event Scale-Revised (IES-R).
Self-evaluation with analog visual scale (0-10) 86% of participants had their first trauma <13 years old, and 71% had their first episode of drunkenness ≥13 years old. There was no correlation between age of first trauma and first drunkenness. Subjects who had experienced trauma in childhood (<13 years), had more severe AUD, PTSD symptoms (specifically intrusive symptoms), and post-exposure craving than subjects who experienced their first trauma later (13-18 years). However, there was no significant difference regarding alcohol consumption in response to negative emotions.

(Continued)
Frontiers in Psychiatry | www.frontiersin.org Self-evaluation with analog visual scale from 0 to 10 Exposure to traumatic script and to alcohol cues led to significantly superior responses (more craving, emotional distress, salivation, and arousal), from neutral expositions. The association between trauma script exposure following with alcohol cue was associated with the more intense craving level.
Both the traumatic and alcohol-related scripts induced significantly higher craving for alcohol than the neutral script. The peak craving induced by exposure to the traumatic script was significantly greater than that induced by the Trier test, which was itself greater than that obtained after exposure to a neutral script. There was no correlation between craving intensity and endocrine response (ACTH and cortisol) after the different tests. Anxiety's level following the Trier Test and the exposure to the traumatic script was significantly higher than the alcohol-related and neutral exposures.

(Continued)
Frontiers in Psychiatry | www.frontiersin.org Experimental non-controlled, non-randomized study. Subjects were exposed to traumatic, alcohol, crack, or neutral cues. After each exposition, subjects reported their craving level.
PTSD diagnosis with CAPS according to DSM-IV. 66% subjects with AUD and 56% with crack use disorder had PTSD (61% of total sample). Self-evaluation of traumatic severity symptoms with IES-R.  There was a positive correlation between anxiety sensitivity and PTSD severity. Subjects with PTSD had higher anxiety sensitivity. Traumatic exposure resulted in increased craving and negative affect. Anxiety sensitivity was positively correlated with post-traumatic exposure negative affect but not with craving. PTSD severity was positively correlated with post-traumatic exposure negative affect and craving.
et al. (46) showed a positive correlation between the severity of PTSD symptoms and the intensity of craving after traumatic exposure. Finally, McHugh et al. (46) observed a positive correlation between the level of anxiety sensitivity (tendency to react with fear to signs and symptoms of anxiety) and the severity of traumatic symptoms and negative affect, but not craving after the exposures.

Association Between PTSD Symptoms and Craving Across SUD Subgroups
A total of 14 studies investigated the association between PTSD symptoms and craving across SUD subgroups: 12 were observational studies (five for AUD, two for Tobacco Use Disorder, one for Cannabis Use Disorder, and four for different types of Substance Use Disorders), one was an experimental study and one was a Randomized Controlled Trial among AUD individuals.  51) compared the psychosocial functioning of subjects with comorbid PSTD and AUD with those with SUD or PTSD alone. The authors found that social functioning was more impaired (less education, lower income, more unemployment), as well as more severe depressive symptoms and cravings, in the comorbid subjects. In a 28-day study using a daily monitoring with an Interactive Voice Response (IVR), Simpson et al. (52) found that PTSD severity was positively correlated with craving level on the same day but not the following day. In a more specific way, some traumatic symptoms (startle, irritability), were positively correlated with craving levels on the same day, whereas other symptoms (nightmares, emotional blunt, hypervigilance), predicted craving increases on the following day. On the other hand, craving intensity on a given day was not correlated with PTSD symptom severity on the following day.

Alcohol Use Disorder
Two studies assessed the impact of changes of PTSD symptoms overtime on craving after specific treatment approaches. In a randomized clinical trial conducted by Kaczkurkin et al. (53), 165 comorbid subjects were randomly assigned to four different treatment groups: Naltrexone + exposure therapy, Naltrexone alone, exposure therapy + placebo, and placebo alone. At baseline, participants with greater levels of PTSD symptom severity endorsed a significantly greater percentage of days drinking and alcohol craving. The percentage of days drinking was positively correlated with alcohol craving. Cross-sectional study. Comparison of psychosocial variables between subjects with comorbid AUD and PTSD, and subjects with PTSD or AUD alone.
PTSD diagnosis with the PSS-I and the Structured Interview For PTSD (SIP) according to DSM-IV.
Self-evaluation with the Penn Alcohol Craving Scale (PACS).
Comorbid subjects had less employment, less college education, and lived alone more often than PTSD or AUD group and had lower income than PTSD group. Comorbid subjects were not different in terms of alcohol consumption compared to subjects with AUD alone, but experienced more craving. Comorbid subjects had more depressive symptoms than subjects with PTSD alone.  PTSD diagnosed with the PCL-C according to DSM-IV.
Self-evaluation with the Obsessive Compulsive Drinking Scale (OCDS).
Severity of PTSD symptoms was positively correlated with craving and alcohol consumption in the past 3 months. Craving intensity was also correlated with the frequency and quantity of alcohol consumption. Lower verbal learning and memory were correlated with more intense alcohol consumption. Higher level of impulsivity was correlated with stronger craving.

(Continued)
Frontiers in Psychiatry | www.frontiersin.org Cross-sectional observational study, using data collected from a randomized controlled trial. The study assessed the relationship between PTSD negative cognitions, negative affect, and alcohol craving.
PTSD diagnosis with the CAPS according to DSM-5.
Self-evaluation with PACS. Alcohol craving was positively correlated with the number of massive drinking days, severity of PTSD symptoms, negative affect, and trauma-related cognitions. Negative emotions mediated relationship between trauma-related cognitions and alcohol craving.

(Continued)
Frontiers in Psychiatry | www.frontiersin.org Cross-sectional observational study assessing motivational processes influencing tobacco addiction in smokers with PTSD or depressive episode.
Current PTSD diagnosis using CAPS according to DSM-IV.
Self-evaluation with the Brief WISDM.
Subjects with PTSD had higher mean Fagerstrom scores than other participants. Subjects with PTSD or depressive episode had greater craving than the control group. No significant difference in craving was found between subjects with PTSD and depressive episode.

(Continued)
Frontiers in Psychiatry | www.frontiersin.org Cross-sectional observational study. Analysis of the relationship between PTSD diagnosis, type of addiction (alcohol or other substances), addiction severity, and craving intensity. Participants were classified by addiction type: • Group A (AUD alone; n = 182) • Group D (SUD other than alcohol; n = 154) • Group AD (AUD + SUD; n = 123) And by PTSD status: • PTSD group (score positive on the IDCL and PDS) • Subthreshold PTSD group (score positive on the IDCL or PDS) • Subjects who have been exposed to a traumatic event without PTSD • Subjects who have never been exposed to a traumatic event.
Diagnosis of current PTSD according to DSM-IV by IDCL and Posttraumatic Diagnostic Scale (PDS).
Assessed with ASI Prevalence of PTSD in groups AD and D was significantly higher than group A. Subjects with PTSD had a higher addiction severity score on ASI, a greater number of inpatient admissions to addiction care, a shorter mean time of abstinence between relapses, and experienced more frequent craving than other participants. Addiction severity was higher in subjects with a SUD other than alcohol.

(Continued)
Frontiers in Psychiatry | www.frontiersin.org Cross-sectional observational study, using from data collected previously in a randomized controlled trial during rehab treatment. Evaluation of relationship between PTSD symptoms and craving according to the substance (alcohol, psycho stimulants, opiates, cannabis).
Diagnosis with the PCL-C according to the DSM-IV-TR.
Self-evaluation with PACS adapted for other substances.
For alcohol (n = 131), craving was correlated with global PTSD severity, more specifically with hypervigilance symptoms. Concerning psychostimulants (n = 66), craving was correlated with global PTSD severity, more specifically with avoidance syndrome. Concerning opiates (n = 36): craving was correlated with global PTSD severity, but not with specific PTSD symptoms. Concerning cannabis (n = 24), there was no correlation between craving and PTSD.

(Continued)
Frontiers in Psychiatry | www.frontiersin.org Observational study, Evaluation of correlation between PTSD symptoms and craving level.
PTSD assessed with the PCL-5 according to the DSM-5.
Screening for life trauma event with the Life Events Checklist (LEC-5).
Self-evaluation with the Mannheimer Craving Scale (MaCs) PTSD symptoms were positively associated with craving level but were not predictors of craving level the following day.
Participants in the Naltrexone + exposure therapy group had a greater decrease in craving than in the Naltrexone alone group. Subjects with higher initial PTSD severity had a more rapid decrease in craving intensity over time. Finally, the percentage of drinking days and the severity of traumatic symptoms at time t predicted the intensity of craving at time t + 1, with alcohol craving being dependent on the amount of PTSD symptoms and alcohol use reported at the previous time point. Coffey et al. (30) compared the effectiveness of exposure therapy and relaxation-based therapy in 43 subjects with comorbid AUD and PTSD using a laboratory-based experiment. A first session conducted before the beginning of treatment showed an increase in craving and emotional distress of the participants after exposure to a traumatic script and alcohol-related cues. Follow-up analyses after six sessions of group exposure therapy found a reduction of both PTSD symptoms and alcohol craving overtime.

Tobacco Use Disorder
In 2014, Dedert et al. (54) used Ecological Momentary Assessment (EMA) to follow smokers with PTSD in daily life over 14 days. Participants were allowed to smoke freely during the first week, then had to begin withdrawal without any substitute or pharmacological treatment in the second week. Compared with the pre-withdrawal phase, abstinence was associated with reductions in PTSD symptoms and craving, but not negative affect. During withdrawal period, an increase in traumatic symptom intensity predicted an increase in craving at next EMA evaluation, but the reverse association was not observed. Rosenblum et al. (55) compared three groups composed by 162 US Army veteran daily smokers: a PTSD group (52 subjects with PTSD alone or with a comorbid depressive episode); a depressive episode group (52 subjects with depressive episode without comorbidity); and a control group (58 subjects with no psychiatric disorder). The PTSD group (with or without depression) described higher craving than the control group without any psychiatric disorder.

Cannabis Use Disorder
Boden et al. (56) explored the links between PTSD and different characteristics associated with cannabis use (motivation, relational problems, withdrawal symptoms, and craving) in veterans with cannabis use disorder with and without PTSD. Patients with PTSD used cannabis more frequently as a coping strategy and reported a significatively higher level of craving in several components (compulsive, anticipating release of emotional distress, and use planification). Traumatic symptom severity was positively correlated with the emotional component of craving (release of emotional distress).

Multiple Type of Substance Use Disorder
Two studies assessed the course of craving during SUD treatment according to PTSD symptoms severity at baseline. Wieferink et al. (33) assessed outcomes of standard, non-integrated SUD treatment among 297 SUD outpatients (AUD, or Cannabis or Cocaine Use Disorder) with higher (≥48) or lower (<48) PTSD symptom severity based on the Self-Reported Inventory for PTSD (SRIP). At baseline, there was no difference in the number of days of use between subjects, however, subjects with more severe PTSD symptoms had significantly higher levels of craving and anxiety-depressive symptoms. After 3 and 6 months of treatment, there was a decrease in the number of days of use for all subjects, a significantly greater decrease in craving for subjects with more severe traumatic symptoms, and a significant decrease in anxiety-depressive symptoms for subjects with severe traumatic symptoms only. However, patients with higher levels of PTSD symptoms still reported significantly higher scores on depression, anxiety and stress after 6 months of SUD treatment. Peck et al. (57) assessed the impact of a 6 weeks therapeutical program including cognitive processing therapy, Alcoholics Anonymous (AA) meetings, and group discussion with 72 American veterans suffering from PTSD and a substance use disorder (AUD, Cannabis Use Disorder, Cocaine Use Disorder, Opiate Use Disorder, Amphetamine Use Disorder, or Anxiolytic Use Disorder). Baseline dysfunctional cognitions associated with the trauma were positively correlated with PTSD and craving severity. However, PTSD severity was not correlated with craving levels. Cognitive processing therapy was associated with significant improvements in erroneous cognitions, trauma-cued substance craving, and depressive or trauma-related symptoms. Reduction in depressive or trauma-related symptoms was partly explained by the therapy's impact on erroneous cognitions, contrary to craving.
In a cross-sectional study, Driessen et al. (58) focused on the relationship between the type of addiction (alcohol or drug addiction or both), the severity of the addiction and of craving, and the presence or not of comorbid PTSD. Participants with PTSD had a higher addiction severity score, were more often hospitalized, had a shorter abstinence time between relapses and experienced craving more frequently than other participants. Somohano et al. (37) assessed the correlation between severity of different symptoms of PTSD and craving according to four classes of substances: alcohol, psychostimulants (cocaine, amphetamines), opiates and cannabis. Concerning subjects with alcohol use disorder (n = 131), global PTSD severity and hypervigilance levels were associated with craving intensity. For participants with psychostimulant use disorder (n = 66), craving levels were correlated with global PTSD severity and more precisely with avoidance syndrome intensity. Among subjects with opiate use disorder (n = 36), global PTSD severity was correlated with craving levels, but with no association to specific symptoms. Finally, for subjects with cannabis use disorder, no variable was associated with craving. An observational study led by Vogel et al. (59), highlighted a positive correlation between craving levels and PTSD symptoms over 6 days following admission for detoxification among comorbid patients (PTSD with alcohol, cannabis, sedatives or mixed use disorder). However, no correlation was found concerning PTSD symptoms at Day 1 and craving the following day.

Association of Negative Affect With Trauma Exposure and Craving Alcohol Use Disorder
Several studies focused on the role played by negative affect during different forms of exposures (31,38,40). The results were similar to those obtained for craving and showed that exposure to a traumatic script and an alcohol-related cue generated a more intense level of negative affect than during a neutral exposure. Nosen et al. (38) reported that in a traumatic context (exposure to a traumatic script), the intensity of craving was correlated with the severity of negative affect. The study of Coffey et al. (30), through a laboratory-based experiment, found a decrease of both craving and emotional distress after traumafocused imaginal exposure, suggesting that negative emotions should constitute a mechanism of alcohol craving induced by trauma exposure. Lyons et al. (48) examined more specifically the mediation role of negative affect on the association between PTSD cognitions and craving among 136 treatment-seeking veterans with PTSD and AUD. Mediation models demonstrated that negative affect mediated the association between specific posttraumatic cognitions related to the self, the world, the self-blame, and craving controlling for PTSD/AUD symptom severity and gender. Posttraumatic cognitions were associated with increased negative affect, which in turn was related to increased craving. Finally, one observational study (50) examined sex differences in trauma cognitions and their relationship to symptoms of AUD including craving. Specifically, negative cognitions about the self were associated with increased craving in men, but not in women, a finding that could be related to greater subjective negative emotions related to traumatic experiences in men. In this perspective, higher craving levels could be explained as a result of maladaptive coping of traumarelated negative emotions.

Tobacco Use Disorder
The study of Dedert et al. (54) investigated whether craving for cigarettes was driven by PTSD symptoms and negative affect among smokers with PTSD attending to quit, using an EMA procedure. Negative emotions were identified as predictors of craving during the withdrawal period (54). Increased PTSD symptoms and negative affect predicted an increase in craving at the next EMA evaluation, even on days with low levels of craving, but the reverse association was not observed.

Cocaine Use Disorder
In the experimental study of Tull et al. (45), in male subjects only, the experience of negative emotions (shame, guilt) in response to the traumatic script mediated the relationship between traumatic symptoms and craving for cocaine. The experience of selfconscious negative affect in response to the trauma script accounted for the relation between PTSD diagnosis and cocaine craving following trauma script exposure.

DISCUSSION
Twenty-seven studies fulfilled criteria for inclusion in this review, of which 12 focused on alcohol, 4 on tobacco, 1 on cannabis, 1 on cocaine and 9 on various substance use disorders. The results showed that regardless of substance type, PTSD and SUD dual disorder was associated with more intense craving levels and was characterized by a prospective link between PTSD symptom severity and craving episodes. Exposure to traumatic memories in experimental studies was associated with emotional distress whose severity was correlated with craving intensity (31,38).
Whatever for alcohol (31,38), tobacco (43), or cocaine (47), experimental results showed that exposure to traumatic cues among subjects with PTSD and substance use disorder comorbidity triggered craving in the same way as exposition to substance cues. There also was an additive effect of the association of both forms of exposure on craving, a finding that is consistent with literature showing an association between exposure to stress and craving among patients with substance use disorder (26, 60). However, beyond stress exposure, there appears to be a specific effect of traumatic memories on craving. According to the study by Beckham et al. (43), exposition to traumatic cues triggered significatively greater craving compared with exposure to non-traumatic stress cues. This result could explain the lack of difference in craving levels between subjects with and without PTSD, after exposition to a cold pressor task (neutral stress) in the investigation by Brady et al. (41). In this way, persons with these comorbidities are repeatedly exposed to traumatic memories and therefore to more intense craving, which could increase the risk of relapse. Moreover, the study by Boden et al. (56) lends support to this interpretation by highlighting the correlation between traumatic symptom severity and craving intensity. Finally, studies using EMA observed a prospective link in the association between PTSD symptoms and craving, showing notably that craving daily variation was a reaction to traumatic symptoms intensity. Such results are generally supportive of selfmedication theory, as aggravation of PTSD symptoms would then trigger greater craving and lead to substance use as a means of assuaging traumatic symptoms.
In line with this interpretation, some studies in this review also highlighted the role of negative affect associated with traumatic exposure in the risk of relapse and thereby indicating that substance use may constitute a coping strategy to deal with negative affect. Experimental studies among subjects with alcohol and tobacco use disorder (31,38,43) showed a correlation in evolution of negative affect and craving after exposition to a traumatic factor. Moreover, negative emotions were identified as predictors of craving after exposition to traumatic cues as well as in daily follow-ups during withdrawal (54). This literature has also demonstrated a salient association between PTSD, addiction and negative emotions, and points specifically to the mediation role of negative emotions and the relationship between traumatic symptoms and alcohol use (61). According to Zvolensky et al. (62), smokers experience greater negative affect if they have comorbid PTSD. Individuals with comorbidity would also use emotions to assuage emotional distress, in accordance with the principle of negative reinforcement. This dysphoric state could be explained by a decrease in dopaminergic D2 receptor density in the reward network (ventral striatum) among persons with substance use disorders (63) and a higher number of DAT dopamine transporters in persons with PTSD (64) that are correlated with craving intensity. Perturbations of the stress axis could also be implicated in these dual disorders, as anomalies of the stress response among subjects with substance use disorders is associated with the activation of extra hypothalamic corticotropin-releasing factor (CRF) synthesis, excessively activating the amygdala (the brain structure implicated in emotional reactions such as fear) (65,66). Such hyperactivity has also been observed among individuals with PTSD (67) and it is associated with the presence of enduring negative emotional states (anxiety, irritability, dysthymia). While the prefrontal cortex has a major impact on emotional regulation by the inhibition of the amygdala (68), SUD and PTSD are both associated with a hypoactivation of this area (69,70) that could explain the major emotional dysregulation among these cases of dual disorder (71,72). Thus, a negative emotional state or emotional dysregulation characterized by significant fluctuations in daily life could constitute a clinical feature of these dual disorders leading to greater craving frequency and/or intensity, although this hypothesis requires further investigation.
Another important observation of this review that could further understanding for mechanisms underlying PTSD and craving is the impact of early trauma. Schumacher and colleagues (42) demonstrated that patients with dual disorders and early trauma (<13 y.o.) experienced more severe PTSD symptoms, more craving after exposure, and more severe AUD. This is consistent with previous studies highlighting a link between age that the trauma was experienced and PTSD severity (73). The link with craving intensity could be partly explained by the fact that early trauma leads to deficit in inhibitory control during stress exposure, which might facilitate the use of substances as coping strategy (74). Indeed, deficits in inhibition capacities during adolescence is known to be associated with a greater risk of both substance experimentation and the development of substance use disorders (75).
Several limitations of this systematic review should be considered in interpreting its findings. A first concern is the heterogenous nature of the selected studies. Based on the Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies (S2C), most studies included in this review could be qualified as being of "Good" or "Fair" quality (n = 17 or 65%). However, nine studies (35%) were classified as "Poor" quality, and this may partly explain the considerable variation observed in study methods. The majority of investigations classified as "Poor" quality were observational studies while most of the "Good" quality studies were cross-sectional in nature. Moreover, the studies used a large variety of questionnaires to assess SUD, PTSD, and craving. Substance use was mainly reported using self-report questionnaires and only six studies used objective methods of assessment (urine or breath tests). Nevertheless, the majority of experimental studies on cuereactivity administered single-item instruments, mainly visual analog scale which are considered as an acceptable and valid tool in experimental paradigms, while observational studies used different multi-items questionnaires. The heterogeneity of selfreport measures of craving and research design in observational studies could explain some variability in the findings. Finally, few studies addressed sex differences in the analysis, although some results indicated specific relationships between traumarelated cognitions and emotions (self-depreciation and selfblame) and craving among males. More comprehensive analyses are needed to examine the impact of sex (and gender) on the underlying relationship between PTSD and SUD across different SUD subgroups. Despite these limitations, the results strongly underscore the strong relationship between PTSD and substance craving and the necessity to concomitantly treat SUD and PTSD as dual disorder.
Concerning treatment approaches, recent studies assessed several classes of pharmacological agents in the treatment of this dual disorder based on neurobiological mechanisms implicated in both disorders when considered individually (76). Moreover, the positive effect on alcohol use and traumatic symptoms was demonstrated with the association of Disulfiram and Naltrexone in a randomized trial (77), and the use of Desipramine led to an improvement of alcohol use and PTSD symptoms (78). Promising results have also been found with treatments using noradrenergic (Prazosine, Propranolol), GABA and glutamatergic system (Memantine, N-Acetyl-Cysteine, and Topiramate). The results of this review also suggest the importance of improving regulation of negative emotions associated with traumatic memories, and treatments of erroneous or dysfunctional cognitions linked with the traumatism. Furthermore, the potential mediation role of post-traumatic cognitions on negative affect and craving raises the issue to consider posttraumatic cognitions and negative emotions as a salient target for craving reduction. On this issue, several therapies targeting emotion regulation and dysfunctional cognitions linked with traumatism such as Prolonged Exposure and Cognitive Processing Therapy were found to be efficacious for substance use, craving and PTSD symptoms (53, 57, 79). The study of Coffey et al. (30) using trauma-focused exposure therapy led to reductions in negative affect and craving, although the potential link between negative post-traumatic cognitions, negative affect and craving was not specifically assessed. Integrated treatment combining prolonged exposure and naltrexone among individuals with comorbid PTSD and AUD demonstrated better outcomes in terms of alcohol craving compared to exposure alone or naltrexone alone. The necessity of global treatment approaches for comorbid patients, including pharmacological treatment, psychotherapies, and psycho-social treatment has also emphasized (80), but further studies are needed in other SUD populations to generalize these findings and examine the temporal changes of emotion dysregulation, traumarelated emotions such as guilt and shame, on subsequent craving and substance use.

CONCLUSIONS
Findings from the current study further inform our understanding of the synergetic relationship between PTSD and SUDs that lead to craving that is greater than that observed with either disorder alone. PTSD symptoms can act as powerful craving cues with an additive effect when combined with exposure to substance-related cues, thereby constituting a salient risk factor for relapse. The craving elicited by PTSD may differ according to specific PTSD symptoms and the effects of specific forms of substance use, although this possibility requires further investigation. Daily life studies using Ecological Momentary Assessment appear to be particularly adapted to investigating the temporal relationship between different PTSD symptoms, emotional states and the clinical expression of addiction, and hold considerable promise for the development of more personalized interventions in dually-diagnosed individuals. Since the majority of the studies included in our review concern alcohol and tobacco, it would be also interesting to expand this research to other substances as well as to behavioral addictions. Specifically, no studies examined the association between craving and MDMA or psychedelic drugs, that is a major issue in view of novel treatment approaches of PTSD.
Our data therefore challenge our current clinical practice in the treatment of patients suffering from dual diagnosis, and argue for the integration of an additional trauma-focused strategies into addiction facilities, notably including cognitive-behavioral therapies based on prolonged exposure. While all individuals suffering from SUD should be systematically assessed for trauma history and PTSD, the present data suggest that PTSD treatment should not be delayed until abstinence has been achieved. The direct relationship between PTSD symptoms and craving argues for the need of these integrated therapies in the goal of providing the most comprehensive and efficacious treatment possible.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author/s.