*3.3.5 Comparator*

The majority of studies utilized pre-cue baseline as a comparator for their measures of reactivity (n = 22); a minority only compared reactivity data across cue types (i.e., comparing neutral vs. trauma responses; n = 6). However, many studies used a combination of comparators by comparing to baseline data and across cue types as well (n = 12).

While we have summarized the key components of included studies here, a full summary of each study across the coded variables of interest is available in Appendix A.

#### **3.4 Population considerations**

Populations of interest were largely adults who were assessed for PTSD symptoms/ diagnoses and/or trauma history, and substance use. Participants across studies were more often male (*M* = 61.5%, SD = 24.6), with n = 5 studies recruiting only males [29, 30, 34–36] and one study recruiting only females [28]. Four studies included only veterans [30, 36–38], and 12 included only patients in treatment for SUD ([39]; n = 10), PTSD (n = 1), or both (n = 1). Four studies examined emerging adult college students specifically [13, 40–42] and one study recruited low-income, inner-city adults [43].

All studies included participants with either PTSD (n = 14) or those who had been exposed to a lifetime trauma (n = 10), or both with PTSD and/or trauma histories assessed continuously (n = 4). PTSD was assessed but not required for some studies, with others requiring trauma exposure but not a PTSD diagnosis (see [44, 45]). To assess for PTSD, most studies used some form of a validated structured interview (n = 25), such as the MINI [46], the SCID-5-RV [47], and the Clinician-Administered PTSD Scale [48]. Those studies examining trauma-exposed individuals typically administered a questionnaire to assess trauma history (n = 3), such as the Trauma History Questionnaire [49] or the Life Events Checklist [50], as well as continuous measures of PTSD symptoms, such as the PTSD Checklist for DSM-5 [51].

Substance use among the study populations was similarly measured. Specifically, the majority of studies (n = 18) required an SUD as inclusion criteria [18, 52], with some using inpatients receiving treatment for PTSD, SUD, or both (n = 12; [32, 53]). Fewer studies required less extreme forms of substance use, such as occasional drinking (n = 6) (see [13, 54]) and other cut-off points for use of various substances (n = 3; [55]). To assess for the presence of a SUD, most studies (n = 18) used structured interviews, such as the C-DIS IV [56] or the SCID-5-RV [47], but others used shorter self-report measures, such as the Alcohol Use Disorders Identification Test (AUDIT; n = 10) [57].

#### **3.5 The cue-reactivity paradigm.**

#### *3.5.1 Personalized vs. standardized cues*

Many of the studies employed personalized cues within their cue-reactivity paradigms, either through interviews where they obtained information about a participant's worst traumatic experience and transcribed the interview into an imagerybased cue [58] or utilized the participants' preferred substance as part of an *in vivo* cue [59]. The vast majority of these studies found significant reactivity results in their research, specifically noting that trauma, substance, and/or stress-related cues elicited greater craving responses (i.e., greater change from baseline) compared to neutral cues (n = 24 of 28). Even interviews in which the participant described their worst traumatic experience functioned well as a personalized cue for eliciting reactivity on craving measures [36]. Photo, video, or task-based cues were standardized rather than personalized [28, 54] although one study did take into account participants' preferred substances when selecting substance-related video cues [60]. Studies utilizing standardized cues did find cue-reactivity effects on their outcomes, with some caveats. For example, Trautmann and colleagues [54] found craving increased in response to their trauma film cues only among females. Other studies using standardized, nonpersonalized cues that used control groups found cue-reactivity effects (craving and

neural activation, respectively) only in substance-using [29] and trauma-exposed [60] experimental groups vs. non-using/non-exposed controls.

#### *3.5.2 Task-based cues*

Studies that utilized photographic cues as part of task-based cue paradigms found to support that their paradigms functioned as effective cue-reactivity paradigms, even though craving was not the primary outcome of interest. For example, Garland and colleagues [28] showed participants trauma-related images and asked them to either simply view the photos or reappraise the photos by reinterpreting the photo's meaning to regulate their emotions in reaction to the photo. Following this task, relief craving increased; this increase was associated with the number of adverse childhood experiences to which participants reported having been exposed. Similarly, Beckham et al. [30] utilized a Stroop color-naming attentional [31] task with trauma-related words with a veteran sample of cigarette smokers. Results demonstrated trauma words, relative to neutral words, led to greater cigarette craving as well as more withdrawal symptoms.

#### **3.6 Subjective and physiological craving**

One of our inclusion criteria was the measurement of craving following a cuereactivity paradigm. Accordingly, all studies included a measure of craving, with all studies including a measure of subjective craving. Many studies measured craving using a Visual Analog Scale (VAS) or various Likert-type rating scales. Among those who examined craving changes from baseline by cue type, subjective craving responses were highest following trauma-related cues compared to substance, stress, and/or neutral cues (n = 9). Studies that did not use trauma cues found substancerelated cues elicited greater craving compared to neutral cues (n = 3). In those studies that used trauma cues, substance cues, and neutral cues (n = 9), typically trauma cues elicited the greatest craving, followed by substance cues, and then neutral cues. Interestingly, studies, where trauma imagery cues were paired with *in vivo* substance cues (n = 5), found craving was higher for these combined cues compared to trauma imagery cues alone, as well as compared to neutral imagery and *in vivo* substance cue combinations [18, 58].

While our inclusion criteria did not specifically require an objective assessment of craving, the frequent use of salivation, heart rate, and other measures of physiological reactivity warrants a brief summary of this work. Most studies that included physiological/objective craving measures did so by measuring salivary flow (n = 5). Coffey et al. [18] found a significant increase in salivation following trauma and alcohol cues relative to neutral cues. Nosen et al. [59] found an increase in salivation following alcohol *in vivo* cues as well, and this increase was greatest when paired with trauma imagery cues. Two intervention studies examined craving pre- and post-treatment and found a significant decrease in salivation during trauma cue exposure at posttreatment compared to pre-treatment [53, 61]. Interestingly, one study did not find any significant effect of trauma cue imagery and *in vivo* alcohol cue exposure on salivary flow among depressed individuals, but did among those with PTSD [41]. Finally, one study which used heart rate as an objective measure of craving found *in vivo* alcohol cues significantly increased heart rate relative to neutral water cues among males with comorbid PTSD-AUD [35].

#### **3.7 Treatment outcome studies**

Seven studies examined outcomes of pharmacological or psychotherapeutic treatment in clinical populations, utilizing cue reactivity as a secondary outcome measure or adjunct to symptom measures. Two studies examined the effectiveness of pharmaceuticals as a treatment for comorbid PTSD-SUD. Specifically, in a pre-clinical lab-based study, Stauffer et al. [35] examined the use of intranasal oxytocin (20 IU and 40 IU) vs. placebo in males with comorbid PTSD-AUD. Each participant took part in each condition across three counterbalanced sessions. Following drug or placebo administration, participants were exposed to *in vivo* cues of their preferred alcoholic beverage and water. Both heart rate and subjective craving response increased the following alcohol *in vivo* cue exposure relative to neutral *in vivo* (water) cues, but neither dose of oxytocin reduced cue-induced heart rate nor subjective craving responses relative to placebo. Similarly, Kwako et al. [32] combined the Trier Social Stress Test [62] with personalized *in vivo* alcohol cues and conducted separate sessions involving guided imagery scripts of stress, alcohol, and neutral cues. All experimental cues increased subjective craving responses and blood cortisol when compared to the neutral cues. However, they found no effect of the neurokinin-1 receptor antagonist aprepitant (125 mg/day) vs. placebo on subjective craving in response to stress or alcohol vs. neutral cues; however, participants who received the aprepitant had reduced cortisol levels during the presentation of the stress cue.

Five studies examined the effects of several psychotherapeutic interventions on cue-elicited craving as well as distress, PTSD symptoms, and resilience. Coffey and colleagues [18] examined the effects of trauma-based imaginal exposure vs. relaxation using a cue-reactivity paradigm to assess trauma cue-reactivity (i.e., craving), showing a decrease in craving to the trauma-alcohol cue combination only among those enrolled in prolonged exposure (PE) therapy but not among those in the relaxation condition. However, craving following the trauma-only cue decreased relative to baseline among both intervention groups. Similarly, two studies [53, 61] assessed the merits of PE therapy in comparison to a health/lifestyle therapy using a craving to a cue-reactivity paradigm as an outcome measure; one study [53] found both healthy lifestyle (control) and trauma cue-exposure treatments led to a decrease in craving responses to trauma imagery and *in vivo* substance cues compared to pre-treatment baseline responses. While the other study [61] only included those enrolled in the trauma cue-exposure group in analyses, they too found a decrease in cue-induced craving when exposed to trauma and substance cues from pre- to post-treatment. Additionally, a study [36] that examined trauma cue exposure during cognitive processing therapy, a form of cognitive behavior therapy, in veterans with comorbid PTSD-SUD, also found a decrease in trauma cue-induced craving from pre- to posttreatment, the magnitude of which was associated with a degree of increase in resilience and degree of decrease in PTSD symptoms. Finally, one study [37] used *in vivo* cues as part of the COPE (Concurrent Treatment of PTSD and Substance Use Disorders Using Prolonged Exposure) therapeutic intervention. Specifically, Badour et al. combined PE therapy to trauma cues with CBT for substance disorders and *in vivo* substance cue presentations. They examined cue-induced craving at each *in vivo* substance cue exposure session. Craving significantly decreased across sessions, and this decrease was associated with a concurrent decrease in PTSD symptom severity and distress.

#### **3.8 Neural activation**

Three studies combined fMRI and a cue-reactivity paradigm. One study [34] examined neural activation during the presentation of stress, neutral, and substancerelated cues among cocaine-dependent individuals with and without childhood maltreatment histories. The degree of craving to the stress cues predicted activation of the rostral anterior cingulate cortex to a lesser extent in the participants with maltreatment histories. The authors interpreted this to suggest that childhood maltreatment interferes with a key mechanism for resolving conflict and responding adaptively to stress [34]. Conversely, the degree of craving to the substance-related cues was associated with activation of the supplemental motor area and the visual cortex to a greater extent in those with maltreatment histories. The authors interpreted this latter finding to suggest that childhood maltreatment enhances the anticipatory reward response to substance cue exposure [34]. Further, during substance cue presentation, another study [35] found childhood trauma histories among substance users were significantly associated with increased activation of the frontal striatal circuit and the amygdala. However, a third study [32] did not find any psychological correlates of neural activation during the presentation of substance-related vs. neutral stimuli in a sample of adults with comorbid PTSD-AUD. It is difficult to know if this failure to observe an effect of cue exposure on neural activation was due to an ineffective manipulation since craving responses were not measured.

#### **3.9 Affect**

Fourteen studies included a measure of effect as part of their evaluation of cuereactivity. Eleven of such studies examined both positive and negative affect, and three examined negative affects only. The Positive and Negative Affect Schedule or PANAS [63] was overwhelmingly used as the standardized measure of this variable (n = 10), although other measures were used as well, such as the Affect Grid [64] (n = 2). Among the majority of studies (n = 9), negative affect increased following stress and trauma-related cues [38]. In those studies which also examined positive affect, positive affect tended to decrease following stress and trauma-related cues [42, 33] but this was not always consistent. For example, Coffey et al. [39] did not find any statistically significant differences in positive affect across cue types. Interestingly, one study reported that substance-related cue exposure increased both positive and negative affect, and this ambivalent response was associated with the strongest substance cravings [55].
