Abstract
Background Substance use disorders (SUDs) are pressing global public health problems. Executive functions (EFs) are prominently featured in mechanistic models of addiction. However, there remain significant gaps in our understanding of EFs in SUDs, including the dimensional relationships of EFs to underlying neural circuits, molecular biomarkers, disorder heterogeneity, and functional ability. To improve health outcomes for people with SUDs, interdisciplinary clinical, preclinical and health services research is needed to inform policies and interventions aligned with biopsychosocial models of addiction. Here, we introduce Cognitive Dysfunction in the Addictions (CDiA), an integrative team-science and translational research program, which aims to fill these knowledge gaps and facilitate research discoveries to enhance treatments for people living with SUDs.
Methods The CDiA Program comprises seven complementary interdisciplinary projects that aim to progress understanding of EF in SUDs and investigate new biological treatment approaches. The projects draw on a diverse sample of adults aged 18-60 (target N=400) seeking treatment for addiction, who are followed prospectively over one year to identify EF domains crucial to recovery. Projects 1-3 investigate SUD symptoms, brain circuits, and blood biomarkers and their associations with both EF domains (inhibition, working memory, and set-shifting) and functional outcomes (disability, quality of life). Projects 4 and 5 evaluate interventions for addiction and their impacts on EF: a clinical trial of repetitive transcranial magnetic stimulation and a preclinical study of potential new pharmacological treatments in rodents. Project 6 links EF to healthcare utilization and is supplemented with a qualitative investigation of EF-related barriers to treatment engagement for those with substance use concerns. Project 7 uses innovative whole-person modeling to integrate the multi-modal data generated across projects, applying clustering and deep learning methods to identify patient subtypes and drive future cross-disciplinary initiatives.
Discussion The CDiA program has promise to bring scientific domains together to uncover the diverse ways in which EFs are linked to SUD severity and functional recovery. These findings, supported by emerging clinical, preclinical, health service, and whole-person modeling investigations, will facilitate future discoveries about cognitive dysfunction in addiction and could enhance the future clinical care of individuals seeking treatment for SUDs.
Scope statement Substance use disorders (SUDs) are a leading contributor to psychiatric morbidity, mortality, and functional impairment. Cognitive dysfunction contributes substantially to this impairment, and plays a prominent role in causal models of addiction. Executive dysfunction in particular is linked to the onset and persistence of SUDs. Large-scale, multidisciplinary efforts are necessary to elucidate the nature and course of cognitive deficits and associated functional outcomes in SUDs. The Cognitive Dysfunction in the Addictions (CDiA) integrative, translational research program undertakes to advance the study of cognitive dysfunction in SUDs, with a focus on impairments in executive function. The Program incorporates preclinical, clinical, and health systems level investigation. The proposed investigations are well positioned to determine how experimental findings translate from humans to animals (reverse translational approach), as well as how preclinical findings will translate into the development of novel treatment avenues.
Background
Substance use disorders (SUD)1 affect 162 million people worldwide and are associated with substantial morbidity, mortality, and disability (1–4). The socioeconomic and health impacts associated with the use of alcohol, cannabis, and illicit drugs, substantially contribute to the global burden in years living with a disability and lost to premature death (5,6). In 2020, the World Health Organization heightened their calls for enhanced actions to curb the harmful effects of alcohol use, underscoring the global public health priority of the damaging consequences of alcohol use (7). In the US, 140,000 people die annually from excessive alcohol consumption (i.e., consuming more than 25 grams of ethanol per day) and another 70,000 from drug-related causes, costing the U.S. economy 191.6 billion dollars related to alcohol and 151.4 billion dollars due to use of other substances (8–10). Similarly, in Canada, alcohol and drug use causes almost 20,000 deaths, and costs the Canadian economy nearly 40 billion dollars annually (3). Treatments for SUDs have evidence supporting their efficacy but their availability is limited, highlighting the critical need to increase access to treatments and develop novel and improved interventions for SUD (11).
To advance the understanding of addiction, research on the cognitive and neurobiological factors involved in the development and maintenance of SUD is crucial (12). Contemporary models underscore the centrality of cognition to different phases of SUD development, with a strong emphasis on executive functions (EF) such as response inhibition and decision-making, in conjunction with motivational (e.g., incentive salience, cue reactivity/craving) and affective (e.g., negative emotionality) factors (13,14). While there is ongoing debate about the precise number of “core” processes that define EF, converging evidence suggests that EF comprises at least three factors that are related but distinct on a behavioural and neurobiological level (15–19): 1) inhibition, or the ability to prevent the processing of irrelevant information in working memory and/or inhibit a context-inappropriate behavioural response; 2) working memory (updating), or the ability to monitor the contents of working memory for relevance to the current task and to remove from or add information to working memory; and 3) set shifting, reflecting the ability to switch between multiple operations or task sets. These processes are supported by common and dissociable neural substrates within a distributed network of frontoparietal brain regions (20), subserved by known biological and neurotransmitter systems (21,22), and may be partially genetically influenced (23). EF “deficits” refer to impaired functioning in one or more of these domains, whereas EF “biases” reflect the specific or selective influences of motivational or emotional contexts or materials on EF performance (24).
SUDs can be associated with both deficits and biases in EF accompanied by structural and functional alterations in underlying neural circuitry (25). Problems with EF lead to functional impairment across all stages of addiction. Interventions that bolster EF may mitigate the influence of attentional and behavioural control on drug-seeking or relapse. For example, computerized cognitive retraining may improve working memory capacity (26–28) and reduce impulsive choice and valuation of specific rewards in addiction (29). Further, non-invasive brain stimulation has exhibited promise across a range of SUDs and appears to impact both craving and decision-making processes (30–32). However, executive dysfunction in addiction has been under-investigated (33). Furthermore, the deficits and biases of EF most linked to functional outcomes remain to be identified and targeted for maximal therapeutic benefit.
Studies to date have most commonly used relatively “pure” or homogeneous patient groups (e.g., adults with a specific SUD and with no psychiatric comorbidities), with more limited attention to individual differences or within-group variability (34–36). Yet, adults seeking treatment for addiction represent a highly heterogeneous group, with complex substance use histories and a high prevalence of psychiatric comorbidities (37–39). Since “pure” group studies have limited generalizability to patient populations, there is a critical need to characterize heterogeneity of executive dysfunction in a large inclusive cohort study involving a complex patient population that is characteristic of large tertiary care facilities.
Here, we introduce the Cognitive Dysfunction in the Addictions (CDiA) Program, an integrative, team-science and translational research program focused on EF deficits and biases related to SUDs, which will aid the discovery of shared versus uniquely affected EF domains (40,41). Parallel biomarker studies in human participants and translational mechanistic studies in relevant preclinical models will link domains of EF to putative biological underpinnings, together paving the way for the rational design of targeted and individualized therapeutics for treatment and recovery from SUDs. Linkages to health care administrative databases could further mobilize knowledge to help shape public health policy and effect change at the societal level.
The CDiA Program benefits from the full participation of a Lived Expertise Research Advisory (LERA) committee, which is composed of members of the community who self-identify as having lived experience with SUD. Incorporating the perspectives of individuals with lived/living experience helps to increase the impacts of the research on the communities expected to benefit from the findings (42). In conjunction with the scientists leading the program, the LERA committee contributed to the formulation of the program and has offered ongoing insights into its public-facing materials and study procedures. The LERA committee will play a central role in interpreting the CDiA findings and sharing the results with knowledge users in the community. The CDiA Program also consults with an International Scientific Advisory Committee composed of SUD experts who provide ongoing guidance.
CDiA consists of seven interconnected projects (P1-7) with the following main objectives:
To identify domains of EF linked to functional outcomes in adults seeking treatment for AUD or SUD (P1).
To identify imaging biomarkers (MRI, fMRI) associated with domains of EF and those most predictive of functional outcomes in adult outpatients seeking treatment for AUD/SUD (P2).
To identify biomarkers mechanistically associated with EF and functional outcomes in adult outpatients seeking treatment for addiction (P3).
To assess the impact of repetitive transcranial magnetic stimulation (rTMS) on EF deficits in adult outpatients seeking treatment for comorbid alcohol use disorder (AUD) and major depressive disorder (P4).
A preclinical study of the impact of alcohol on EF deficits and their reversal by novel therapeutic interventions (P5).
To assess links between EF, treatment seeking for SUDs, and healthcare utilization and costs (P6).
To identify subtypes of individuals seeking addiction treatment using cross-disciplinary data types from all projects to map the biopsychosocial drivers of cognitive dysfunction in SUDs (P7).
An overview of the CDiA Program is presented in Figure 1. With the exception of the preclinical project (P5), all projects draw on a novel sample of SUD patients at the Centre for Addiction and Mental Health, allowing for the triangulation of evidence within the same pool of participants. A detailed description of each project, including the background, methods, and analytical approach, is described below.
P1: Identifying domains of EF linked to functional outcomes in adult outpatients seeking treatment for addiction
Background
Dual Process models of addiction suggest that addiction can result from an imbalance between a controlled process that is more intentional (for which EF are central), and an automatic process that is more implicit or involuntary (for which affective states are influential) (43,44). EFs modulating either appetitive (desirable) or aversive (unpleasant) motivational states are compromised in those with SUDs ((45); also see the Addiction Neuroclinical Assessment (13,46,47)). Although many studies have investigated how deficits in EFs relate to outcomes among people with SUDs (48), the existing studies are often limited by small samples, very brief (3-6 month) follow up periods, narrow focus on a single substance in the absence of mental health comorbidities, and limited attention to functional outcomes (48). Further, most studies examine either EF deficits (i.e., general impairments in EF performance) or biases (i.e., EF impairments in the context of emotional or motivational material), without examining their relative importance for SUD outcomes. Thus, there is a need for integrative research incorporating all of these factors, and carefully evaluating links with both clinical and functional outcomes in a heterogeneous clinical sample followed prospectively over time.
Objectives
The primary goals of P1 are to characterize EF deficits and biases and to determine which domains of EF are most predictive of functional and treatment outcomes in a complex population of adults seeking treatment for AUD/SUD in order to test the following hypotheses (H1a-e):
H1a: Deficits and biases in the three EF domains will exhibit stronger baseline associations with indicators of functioning, such as disability and quality of life, relative to other components of cognitive functioning.
H1b: EF deficits across all three domains will be most prominent in those with greater (i) severity, (ii) duration of AUD/SUD, whereas EF biases will be most prominent in those with concurrent mood disorders and elevated suicidality.
H1c: EF deficits and biases can be used to identify clusters of participants that are associated with distinct clinical features.
H1d: Relatively more severe EF deficits and biases at baseline will predict slower improvements in functional outcomes over and above substance use, treatment engagement, and healthcare utilization.
H1e: EF domains will show improvements from baseline to the one-year follow-up. Improvements in EF will be positively associated with improvements in AUD/SUD symptoms and in functional outcomes.
Methods and Planned Analyses
Design and Participants
The CDiA Program aims to recruit a total sample of 400 adults seeking treatment for AUD and/or SUD to participate in this observational, longitudinal study involving clinical and cognitive assessments at baseline and over a one-year follow-up period (see Table 1 for schedule of assessments). Inclusion criteria are: (1) 18 - 60 years of age; (2) meeting diagnostic criteria for a current (past year) AUD or SUD (not counting nicotine- or caffeine-related disorders); and (3) seeking support for substance use concerns. Exclusion criteria include: (1) acute intoxication or withdrawal; (2) active psychosis; (3) acute suicidality; and (4) history of severe head injury, dementia, severe neurodevelopmental disorders, and other medical conditions or medications that could severely impair cognition.
Schedule of Clinical & Cognitive Assessments
Measures
Interview measures include the Diagnostic Assessment Research Tool (DART; (49)) to characterize AUD/SUD and comorbid diagnoses, and the Timeline Follow-Back (TFLB; (50)) to characterize alcohol and substance use over the past 60 days. Self-report items assessing lifetime history of alcohol and substance use are also administered, along with measures of severity of alcohol- and substance-related harms. Other factors relevant to understanding substance use patterns are also assessed, including demographic factors, readiness to change, substance use motives, SUD insight and SUD treatment history (see Table 1 for a full list of measures).
A comprehensive assessment of global cognitive functioning is conducted with the Central Nervous System Vital Signs (CNS-VS; (51)) computerized assessment, which has been validated for use in addiction. CNS-VS contains 10 tests that yield 15 individual domain scores, including composite scores assessing Neurocognitive Index, Composite Memory, Psychomotor Speed, Complex Attention, and Cognitive Flexibility scores. Scores are standardized based on an age-matched normative sample. Further, tasks assessing risk taking and decision-making (Iowa Gambling Task (52)), and probabilistic and delay discounting (53–55) are also administered.
EF assessment: To achieve our goal of measuring both (a) general executive function deficits and (b) biases in these functions associated with emotional and motivational states, both neutral and emotional stimuli are used in our executive functioning tasks. The stimuli for the executive functioning tasks are a pool of happy, neutral, and sad faces from the Karolinska Emotional Directed Faces database (KEDF; (56)). Each task is administered with trial blocks with emotional stimuli (i.e., to measure executive function biases) and trial blocks with neutral stimuli (i.e., to measure executive function deficits). A flanker task (57–60) is used to assess inhibition, an n-back task (61,62) is used to assess working memory updating, and a switch task (63) is used to assess set shifting.
Power Analysis and Data Analytic Plan
Multiple regression will be used with the clinical, functional, and demographic factors entered as simultaneous predictors of each EF domain to test hypotheses about which factors are most strongly associated with specific EF domains (H1a and H1b).
For H1c, we will conduct latent profile analyses using domains of EF as indicators of discrete latent profiles of relative deficits and biases. We will then examine the prediction of membership in these latent profiles from substance use profiles.
For H1d, latent growth curve modeling will be used to model changes in substance use frequency, AUD/SUD severity, and functional outcomes over the follow up time points. Next, rate of change will be examined as a function of baseline EF (H1a) by regressing variability in the slopes for the growth factors on baseline EF deficits and biases. These will be included as simultaneous predictors in the model to examine the specific EF domains and biases that are most predictive of change in the clinical and functional outcomes.
For H1e, we will use a generalized linear mixed effects modelling approach. EF deficits and biases will be modelled as dependent variables and time will be entered as a within-subjects factor to determine if there is significant within-person change from baseline to one year follow up. Changes over time in AUD/SUD symptoms and functional outcomes will be modelled using the same approach, with changes in EFs included in the models as a time-varying predictor of changes in these outcomes.
P2: To identify neuroimaging biomarkers associated with domains of EF and addiction symptoms and most predictive of functional outcomes
Background
Functional magnetic resonance imaging (fMRI) studies in healthy adults suggest that the three dimensions of EF rely on overlapping nodes within a broader frontoparietal circuit, but may also have at least partially dissociable neural substrates within and beyond this circuit (64–67). The ability to maintain a task goal stably, shared across all EF domains, relies on areas of the lateral prefrontal cortex (PFC), which extend from frontopolar regions through the mid-dorsolateral prefrontal cortex (dlPFC)(15,68,69), potentially also including the anterior cingulate cortex (ACC) and frontal operculum (70). Additional studies investigating the three EF domains side by side show overlapping activation in the inferior frontal gyrus, anterior cingulate cortex and bilateral parietal cortices (66). In contrast, behavioural inhibition is associated with more extensive recruitment of temporal regions and may engage the cortico-striatal circuit (71). Set shifting may, in turn, involve more posterior dlPFC regions (e.g., inferior frontal junction) and more extensive parietal areas (e.g., intraparietal sulcus) (72,73). Working memory updating has been additionally mapped onto fronto-striatal connections and requires input from the basal ganglia and cerebellum (74–76). Similar partially dissociable signatures of the three distinct EF dimensions have been identified in studies using resting-state (i.e., task-free) fMRI (64,65), structural MRI (77,78), and diffusion-weighted imaging (DWI)(77).
Meta-analyses of task-based fMRI across SUDs highlight the heterogeneity of paradigms used to evaluate EF, but show altered activation in fronto-parietal and striatal regions across studies (79). Importantly, however, prior studies in SUDs have almost universally focused on a single EF domain, providing relatively limited insight into the complex cognitive architecture that may underlie functional impairment in addiction (80–82). Clinically relevant insight is further limited by a predominance of cross-sectional studies, which do not investigate how neural features supporting EF may contribute to symptom change and functional recovery over time.
Characterizing executive dysfunction heterogeneities and underlying neural circuits longitudinally across large naturalistic cohorts of patients seeking treatment for SUDs is likely to provide novel insight into the pathophysiology of addiction and help identify novel targets for intervention.
Objectives
The primary goals of CDiA Project 2 (P2) are to identify imaging biomarkers (MRI, fMRI) associated with domains of EF, their associations with addiction symptoms and day-to-day functioning, and prospective relationships with symptom changes and functional outcomes in adult outpatients seeking treatment for SUDs. These imaging biomarkers may help guide future development and individualization of biologically informed treatments for naturalistic clinical populations with SUDs.
H2a: Behavioural performance along the three EF domains will map onto partially dissociable neural substrates overlapping with frontoparietal and corticostriatal circuits. The extent of activation and pattern of brain activity associated with each of the three EFs will be associated with greater addiction severity and lower general functioning at baseline.
H2b: Greater disturbances in EF-associated neural circuits at baseline will be associated with lower improvement in addiction severity and functional outcomes on follow-up, above and beyond impairments in other neural circuits.
Methods and Planned Analyses
Participants
All participants included in the total sample are eligible to participate in a 1.5-hour MRI protocol. P2-specific exclusion criteria include the presence of MRI-incompatible metal implants, history of stroke, and claustrophobia. We anticipate approximately 50% of the entire sample (n=200) will be eligible and consent to MRI. Participants are eligible for a longitudinal scan at one-year follow-up; we anticipate 25% (n=50) being available and consenting to the repeat scan to assess longitudinal stability of our main measures.
MRI methods
Our research imaging protocol uses state-of-the-art imaging sequences fully harmonized with MRI acquisition protocols used by large-scale population-representative studies (83) and other ongoing CAMH-based cohort studies (84,85) (see Table 2), which will allow us to leverage our data for comparison or transdiagnostic analyses. To support new research studies, a portion of the scanning time is secured for pilot sequences.
MRI Sequences
Structural measures
A standard T1-weighted structural scan is acquired to obtain detailed measures of cortical morphology (e.g., thickness, surface area, and volume). Diffusion-weighted imaging (DWI) is acquired to measure white matter tract integrity and tissue microstructure.
Neuromelanin-sensitive (NM-MRI) scans are acquired to quantify the integrity of the substantia nigra (86,87), with partial coverage of the locus coeruleus (88). NM-MRI scans were included to capture metrics related to catecholamine signaling implicated in addiction based on molecular imaging and preclinical models (89).
Resting-state BOLD
Participants are asked to remain awake with their eyes open for 10 minutes while lying restfully in the scanner.
Cognitive fMRI paradigms
We administer three EF tasks for a total duration of ∼30 minutes. The tasks assess the same EF domains as in P1 but were specifically selected to be distinct from the paradigms used in P1 to reduce practice effects, which may alter in-scanner performance and brain activation patterns (90,91). The EF cognitive paradigms were adapted from Rieck et al (66) and are presented in a pseudorandom order counterbalanced across participants. To minimize visual distraction and permit joint analysis across all EF dimensions, the tasks were designed to be visually simple and consistent across EF domains, using white or light-colored letter stimuli against a dark grey background.
Proposed Analysis
At baseline, task-based brain activation and brain structure will be investigated in association with SUD symptom severity and functional ability. Multivariate approaches (e.g., Partial Least Squares regression) will be used to investigate relationships between microstructure, function, and performance across the three EF tasks by identifying latent variables (LV) that capture covariance between the set of predictors (microstructure, performance) and outcome variables (functional activation). Individual brain scores reflecting the degree to which participants expressed the LV activation pattern will be correlated with addiction severity and functional ability at baseline and on follow-up, uncovering neural factors crucial for recovery. Age, sex, and primary substance of concern will be used as covariates.
Although it is challenging to provide a power estimate for studies involving complex multivariate analytic approaches in neuroimaging, our proposed sample size is consistent with recent empirically derived recommendations to enhance reproducibility in task-based fMRI research (92). These studies suggest that a sample size of n=120 is sufficient to obtain a test-retest voxel-wise correlation >0.80 across most paradigms. Thus, our proposed sample size of n=200 cross-sectionally and n=100 longitudinally, is in line with these recommendations and, to the best of our knowledge, larger than any prior fMRI study in SUDs. Further, based on standard power analysis tools, a sample size of 200 allows power of >0.80 to detect small effect sizes on the order of r=0.2 or Cohen’s d=0.2.
P3: To identify biomarkers mechanistically associated with EF and functional outcomes in adult outpatients seeking treatment for addiction
Background
The analysis of peripheral biomarkers has emerged as a promising approach to probe potential mechanisms or inform prognosis for SUDs (93,94). These markers are easily accessible and relatively inexpensive when compared to other methods (e.g., MRI or PET) (95,96), which highlights their potential utility in clinical settings. Establishing the relationship between peripheral biomarkers and cognitive dysfunction in SUDs can help to unravel mechanisms that contribute to SUDs, including domains of executive dysfunction, and to understand associated risk and mediating factors.
Substance use can trigger the activation of microglia and astrocytes in the brain, driving their function to a neuroinflammatory response and inducing the production of pro-inflammatory biomarkers (97,98). Several of these molecules can cross the blood-brain barrier, promoting immune activation in peripheral tissues. The persistence of this pro-inflammatory feedback loop between the brain and periphery can lead to reduced neuroplasticity, structural and functional changes in key neural circuits underlying cognitive and emotional processing, leading to executive dysfunction, physical disability, and a higher risk of SUD relapse (99).
Consistent with this, prior work supports the relationship between SUDs and peripheral markers of pro-inflammatory biological processes. For example, individuals with AUD exhibit an activation pattern characteristic of a pro-inflammatory response that persists even after alcohol withdrawal (100–102). Cocaine abstinence was also associated with higher levels of pro-inflammatory cytokines (IL-2, IL-6, and IL-17) in a sex-specific manner (103). Importantly, higher IL-6 levels have been correlated with worse EF performance (104). In individuals with opioid use disorder, higher levels of pro-inflammatory cytokines (IL-1β, IL-6) were associated with worse episodic memory performance and worse treatment response to methadone (105,106). A similar pattern of increased pro-inflammatory activation is also observed in patients with cannabis use disorder (107).
The pro-inflammatory activation observed in SUDs happens in the absence of antigens and is considered sterile inflammation. Sterile inflammation often occurs as a response against damaged cells and molecules. A key player of sterile inflammation is the activation of the inflammasome. The inflammasome is an intracellular multiprotein complex (e.g., the NALP3 complex) that functions as a sensor of Damage-Associated Molecule Pattern (DAMPs). It leads to the activation of proinflammatory caspases and the cleavage and release of proinflammatory cytokines, such as IL-1β and IL-18, and caspase-1.
Inflammasome activation is frequently reported in different neuropsychiatric conditions (108–110). More recently, studies have linked cytokines related to inflammasome activation, in particular IL-1β, with cognitive dysfunction in different neuropsychiatric disorders (111–117). However, there is no information about whether inflammasome activation is a potential trigger and/or contributor of cognitive impairment in general, and EFs more specifically, among adults with SUDs. The overarching goals of P3 are to investigate pro-inflammatory changes and inflammasome activation as a mechanism associated with executive dysfunction in SUDs. We will focus our analysis on peripheral biomarkers in plasma. Blood samples and extracted DNA are banked for potential future integrative analyses (see P7).
Objectives
P3 will mainly focus on the investigation of the inflammasome in SUDs and correlate it with executive dysfunction and brain structural and functional parameters. The first objective is to investigate the impact of inflammasome activation on EF domains in SUDs. Second, we will investigate the impact of inflammasome activation on structural and functional brain parameters in SUDs (integration with P2). Results will provide insight into potentially targetable molecular mechanisms underlying EF in this naturalistic sample.
H1: Higher inflammasome activation (i.e., high expression of NALP3) will be associated with: (a) worse inhibition (b) lower working memory updating, and (c) reduced set shifting ability.
H2: Higher inflammasome activation will be associated with: (a) lower gray matter in the prefrontal cortex; (b) reduced activation of the default brain network; (c) less connectivity in neural circuits underlying EFs.
Methods and Planned Analyses
Blood collection, sample preparation, and storage
P3 draws on the same participant pool recruited for P1. Following consent, blood is collected into two EDTA (8 cc) and one citrate (4.5 cc) tubes. After processing, the plasma and buffy coat aliquots are immediately stored at -80°C until analysis.
Laboratory analysis for plasma biomarkers
We will measure the phosphorylated and non-phosphorylated proteins isoform levels of the NALP3 complex using Western Blot. The ratio between non-phosphorylated and phosphorylated isoforms of NALP3 will be used as a proxy measure of the activation of the inflammasome pathway. Additionally, we will measure the plasma levels of the pro-inflammatory cytokines IL-1β, IL-18, and Caspase-1 using a multiplex immunoassay (bead-based multiparametric assay).
The immunoassays will be analyzed using the Luminex xMAP technology® (Luminex 200). The assay has very low cross-reactivity, and this technology has a higher sensitivity to detect very low concentrations of different analytes. The intra- and inter-assay coefficient of variance for the assay is below 10% for most of the analytes measured.
Proposed analysis
To test the association of the biomarkers with the three EF domains and neuroimaging outcomes, we will use Pearson correlation and linear multiple regression models. The linear models can be expanded to include potential covariates, if necessary. The total sample size of 400 subjects will be sufficiently powered to detect small effect sizes of correlation coefficients of 0.11 (α=0.05, statistical power of 80%) to 0.14 (α=0.05, statistical power of 95%).
P4: To assess the role of repetitive transcranial magnetic stimulation (rTMS) on aberrant executive function in the context of major depressive disorder in adult outpatients seeking treatment for SUD (NCT06299787)
Background
The prefrontal cortex, although well established as an efficacious target for the treatment of major depressive disorder (MDD), has recently come into favour as a therapeutic target for AUD (118). Depressive symptoms are also highly prevalent in individuals with AUD (119–123). A number of cognitive and psychological processes stemming from the prefrontal cortex, a common treatment target for repetitive transcranial magnetic stimulation, are disrupted in both MDD and AUD. In a study by Marzuk et al (124), individuals with MDD, demonstrated poorer performance in several EF assessments. Although numerous other frontal cortical and subcortical regions are involved in MDD, the DLPFC appears to focus on executive and cognitive control of negative emotion through reappraisal and suppression strategies (125).Treatment of SUDs with repetitive transcranial magnetic stimulation (rTMS) is an extremely promising frontier (126). It has been posited that there are underlying neurobiological mechanisms shared across SUDs that involve dysfunctional cortico-striatal circuits which can be modulated with rTMS applied to the DLPFC (127). Given the role of executive dysfunction in SUD, activation of the DLPFC via rTMS may aid in modulating inhibitory behavior, salience attribution and decision-making (128). A comprehensive investigation of the domains of EF (inhibition, updating, and set shifting) associated with both MDD and AUD, has yet to be conducted and will be a major contribution of the CDiA (see P1).
Repetitive Transcranial Magnetic Stimulation (rTMS) is FDA approved for the treatment of MDD and a large body of evidence supports its safety and antidepressant efficacy (118,129,130). Theta Burst Stimulation (TBS) is an efficient form of rTMS associated with both antidepressant effects (131) and alterations in plasticity in the cortex (132). TBS involves stimulation of the cortex with triplet pulses of rTMS applied at 50 Hz (burst) and delivered every 200 msec (5Hz). TBS can be applied as continuous TBS (cTBS) as a 40 second uninterrupted burst (i.e., 600 pulses) or intermittent TBS (iTBS) as a 2 second train of bursts repeated every 8 seconds for a total of 600 pulses. Importantly, TBS takes about 1/6th the time of standard rTMS, translating to a higher impact and consequently more efficient treatment at a population level.
MDD is linked to hopelessness, negative affect, and attentional biases (133,134). These features are connected with aberrant DLPFC function and are part of the negative valence domain, which includes constructs of loss, sustained threat, and frustrative non-reward (135). Rodent studies have shown that TMS-mediated increase in cortical inhibition occurs through the recruitment of dendrite-targeting GABAergic inhibitory neurons, putatively implicating Somatostatin (SST)-positive GABAergic neurons, α5-GABAA and GABAB receptors (to be investigated in CDiA Project 5), since these are the main inhibitory cells targeting pyramidal cell dendrites, and their function in targeted pyramidal cells is mediated by these two GABA receptor subtypes (136). The role of activating this pathway on executive function domains will be investigated in P5.
TMS combined with EEG (TMS-EEG) is a powerful method to assess GABA receptor mediated inhibition, excitability, connectivity and plasticity in non-motor cortical regions including the DLPFC. Single and paired pulse TMS-EEG studies have identified several TMS-EEG indices associated with cortical inhibitory mechanisms associated with GABA receptor mediated inhibition (137–141).
Previous studies have demonstrated that rTMS acts through mechanisms linked to GABAergic inhibitory neurotransmission in the cortex. For example, we have previously reported that rTMS increased GABA receptor mediated inhibition through rTMS and that higher frequencies (i.e., 10 and 20 Hz) produced greater inhibitory change compared to lower frequencies (i.e., 1 and 5 Hz) (142). More specifically, iTBS over the prefrontal cortex alters the N100 (143,144), which is closely associated with GABA receptor mediated inhibition (139).
We have recently demonstrated that in MDD, there are abnormalities in TMS-EEG markers of GABAA and GABAB receptor mediated inhibition (i.e., N45 and N100, respectively) in the DLPFC. Specifically, we reported that the N45 waveform predicted depression illness state with 80% sensitivity, 73.3% specificity, and 76.6% accuracy (area under the curve = 0.829, p < .001) (145). More recent pilot data from our group in patients with depression undergoing a 6-week rTMS course demonstrated that active rTMS changed the N45 and N100 waveforms and this change was related to HRSD-17 symptom improvement. Sham rTMS, by contrast, had no effect on either the N45 or N100 waveforms. Collectively, these results suggest that both rTMS and iTBS target neurophysiological markers of GABA receptor mediated inhibition and this mechanism may be related to MDD treatment response. However, there have not been specific neurophysiological investigations into executive functions and depressive symptomatology in the context of MDD comorbid with AUD. This proposal serves as a unique window into the neurophysiology of this at-risk population, and will further investigate the correlation among rTMS treatments, indices of GABAergic inhibition and mode defined domains of executive function (inhibition, updating, and set shifting).
An emerging neurophysiological index of prefrontal cortical activity underlying executive functioning is theta-gamma coupling (TGC), which measures the degree to which high-frequency gamma oscillations are modulated by low-frequency theta oscillations (146) TGC is thought to underlie working memory, the brain’s ability to select, maintain, and manipulate information in the short-term. In a group of patients with MDD, Noda et al. (147) showed that resting state TGC levels correlated with executive function and was enhanced after 10 sessions of DLPFC rTMS, suggesting that it may be a biomarker for intervention in treatments that target brain regions underlying executive dysfunction. To further investigate TGC as a biological mechanism in MDD, the present project adds an investigation of TGC coupled with the N-back working memory task before and after rTMS treatment. The project bridges gaps in previous studies (147) by including a sham control group, measurement of TGC that is coupled with a working memory task, investigating effects in comorbid AUD, and delivering a full 20 session course of rTMS.
Objectives
P4 will enhance the development of TBS as a new intervention for AUD in the context of depressive symptoms and uses integrated TMS-EEG to identify neurophysiological targets of executive dysfunction in this disorder. The study aims to conduct a double-blind randomized pilot study, consisting of two arms (bilateral TBS and sham) to investigate the effect of bilateral TBS on deficits in executive function in the context of comorbid AUD and MDD. We will leverage our expertise in neurophysiological investigations of the cortex through TMS-EEG to index GABA inhibition in the cortex of individuals with a diagnosis of AUD, who have not used substances within the past month (ie. meeting criteria for early remission), in the context of a current depressive episode and MDD, and index alterations in TMS-EEG indexed GABA receptor mediated inhibition induced by bilateral TBS or sham applied to the DLPFC. We will also determine whether bilateral rTMS to the DLPFC changes TGC and is associated with changes in executive dysfunction in patients with MDD and AUD. We will also assess the direct associations between improvement in executive function domains after bilateral TBS, with improvement in MDD symptoms and conduct exploratory subanalyses by biological sex and age categories. Finally, we will aggregate data from other CDiA projects (including imaging, genetics, inflammatory models, preclinical biomarkers and more) to further examine the role of GABA receptor mediated inhibition in the cognitive dysfunction found in SUD. Specific baseline cross-analyses will be performed between executive function tasks, imaging variables, inflammatory markers and TMS-EEG indices of GABA receptor mediated inhibition.
H1: 4 weeks of daily bilateral DLPFC TBS will lead to greater improvement in executive function tasks, compared to sham TBS.
H2: 4 weeks of daily active bilateral DLPFC TBS will lead to a greater proportion of individuals remaining abstinent from substance use compared to sham rTMS.
H3: 4 weeks of daily active bilateral DLPFC TBS will lead to greater amelioration of depressive symptoms (as indexed by decrease in HRSD-17 score) and suicidal ideation (as indexed by a decrease in Columbia-Suicide Severity Rating Scale score), compared to sham rTMS.
H4: Active bilateral TBS DLPFC stimulation (compared to sham) will lead to an increase in GABA receptor mediated inhibition indices after the course of TBS.
H5: Active bilateral DLPFC rTMS (compared to sham) will lead to a significantly greater increase in TGC and correlate with improvements in executive dysfunction.
H6 (exploratory): 4 weeks of daily active bilateral DLPFC TBS (compared to sham) will lead to improvement in all executive function domains (inhibition, updating, and set shifting) in conjunction with improvement in MDD symptoms (ie. more amelioration of MDD symptoms will be associated with improved executive function).
Methods and Planned Analyses
A total of 40 participants will be recruited from the CDiA pooled sample within 6 to 12 weeks of the baseline assessment for P1. Participants will be randomized in a 1:1 ratio to one of two different rTMS arms. The first arm will include bilateral TBS, applied as cTBS over the right DLPFC followed by iTBS applied over the left DLPFC. The second arm will be sham TBS. All participants will be asked to participate in 4 weeks of treatment, 5 days/week (i.e. weekdays).
TMS-EEG will occur at treatment initiation (week 0) and treatment end (week 4). Baseline and post-treatment clinical measures will be administered to characterize each AUD patient. TBS will be administered using the MagPro X100 stimulator equipped with a Cool-B70 coil and Qooler fluid-cooling device (MagVenture, Farum, Denmark) positioned under MRI guidance using the Brainsight neuronavigation system. With a sample size of 20 participants per group and assumed large effect sizes, we will have 80% of statistical power to detect a significant difference between the active treatment arm and sham at a significance level of 0.05.
Randomization and Blinding
We will randomize participants based on a stratified randomization scheme using a permuted block method with a random number generator, in fixed random sizes.
Active/Sham rTMS Apparatus/Blinding Procedure
We will use an R30 or X100 with a cool A/P B70-type coil (Magventure Inc.) to ensure blinding of both patient and technician. As such, the clinician, researcher, patient and technician will all be blinded. For both active and sham stimulation, the coil is positioned under MRI guidance using real-time neuronavigation, thus providing technician and participant with visual feedback throughout stimulation.
TBS Treatment Procedure
TBS will be delivered to the DLPFC bilaterally, with neuronavigation determined through structural MRI. The structural MRI utilized to derive the brain target will have been completed as part of P2. Should participants not have had imaging performed as yet when they are screened for this project, they will be redirected through Project 3 after baseline assessments of this project prior to the start of rTMS treatment. The coil position selection will start by identifying MNI152 stereotaxic coordinates (x, y, z) of (–38, +44, +26) and (38, 44, 26) in the left and right DLPFC, respectively, which we have shown to be effective targets for rTMS in alleviating symptoms of MDD (148). Bilateral TBS will be delivered as follows: R-DLPFC cTBS: 40s uninterrupted bursts (triplet 50 Hz bursts, repeated at 5 Hz, 40s on, 600 pulses total) followed immediately by L-DLPFC iTBS: (triplet 50 Hz bursts, repeated at 5 Hz, 2s on and 8s off, 600 pulses total). For intensity of stimulation we will use 120% RMT. Our work in over 200 people stimulating PFC suggests that 120% RMT is the maximum that should be used for negligible risk of seizure and other serious adverse effects beyond the expected scalp pain and headache during early treatment sessions (149,150).
Participants
Patients will be included if they meet general inclusion criteria for CDiA and in addition, if they: (1) have a Diagnostic and Statistical Manual for Mental Disorders, 5th edition (DSM–5) diagnosis of AUD based on the MINI; (2) do not exhibit problematic use of any substances (excluding nicotine and caffeine), including alcohol, for >1 month; (3) screened positive for an MDE based on the MINI without psychotic symptoms (4) are agreeable to keeping their current antidepressant medications and medications for AUD constant during the study; (5) are reliably taking SUD agonist therapies if appropriate and managed by their clinical team; (6) are able to adhere to the study schedule (7) meet the TMS safety criteria (151).
Participants are excluded if they meet general exclusion criteria for CDiA, and in addition, if they: (1) have a concomitant major unstable medical illness or any significant neurological disorder; (2) are pregnant or intend to get pregnant during the study; (3) have failed a course of ECT, due to the lower likelihood of response to rTMS; (4) have an intracranial implant (e.g., aneurysm clips, shunts, cochlear implants) or any other metal object within or near the head, excluding the mouth, that cannot be safely removed; (5) require a benzodiazepine with a dose equivalent to lorazepam 2 mg/day or higher (152).
Participants will be discontinued from the study if they cannot safely continue the study based on any of the following criteria: (1) experience clinically significant worsening of suicidality that requires an involuntary inpatient hospitalization; (2) develop clinically significant hypomanic or manic symptoms; (3) relapse into substance use in the month prior to rTMS treatment or during the rTMS treatment (4) miss four rTMS treatments (i.e., 20%); or (5) withdraw consent.
If relapse into substance use is suspected during the course of rTMS treatment, for safety purposes, a physician will be called to assess the participant prior to the initiation of rTMS treatment for the day or as soon as substance use is suspected. Investigations (e.g., urine drug screen or blood draw) will be ordered at the physician’s discretion, with the consent of the participant. The decision to continue with or discontinue rTMS treatment for that day or going forward will be at the discretion of the assessing physician and the participant’s SUD clinical team.
Clinical Assessments
Participants will be initially screened with the MINI and C-SSRS as part of P1. Interested and potentially eligible participants will complete a subsequent interview to address all criteria outlined above to confirm eligibility..
The 17-item Hamilton Rating Scale for Depression (HRSD-17) (153) will be our tertiary clinical outcome measure. The tertiary clinical outcome criteria will be a decrease in HRSD-17 score at treatment end compared to baseline (129,148). The HRSD-17 will be performed at baseline, weekly and at the end of the rTMS treatment course.
The Columbia-Suicide Severity Rating Scale (C-SSRS) will be used to evaluate suicidality. This reliable and valid scale has been used in randomized clinical trials and is able to predict completed suicide (154). The C-SSRS has been reported as an effective measure for diagnosis and treatment across several diagnoses (155–157) with high internal consistency, and interrater reliability (157–161). The C-SSRS will be performed at baseline, weekly and at the end of the rTMS treatment course.
TMS Adult Safety Screen (TASS) (151) will be used at recruitment and baseline to ensure that participants are safe to participate in rTMS treatment trials.
Executive Function Battery To enhance integration of each project, we will re-administer the EF cognitive assessments within one week prior to the start of rTMS and at the end of the 4 week course of rTMS.
Neurophysiological Indices (TMS-EEG) of GABA Receptor Mediated Inhibition in the DLPFC
To evaluate GABA receptor mediated inhibition indices in the DLPFC, TMS will be administered to the left DLPFC using two Magstim-200 stimulators (Magstim Company Ltd., UK) connected via a Bistim module and electrophysiological data will be collected using dedicated hardware and software (Neuroscan, Compumedics, USA). Each TMS session will include the establishment of the individual threshold for stimulation, followed by GABA receptor mediated inhibition paradigms in the DLPFC according to our published methods (139). Recordings will be acquired through a 64-channel EEG. GABA receptor mediated inhibition indices from the DLPFC will be the dependent variables of interest and derived according to published recommendations (139,162). Our neurophysiological measures have been established in several of our previous reports and have a high test-retest reliability (i.e., ICC > than 0.9) (163,164). Data analysis will take place using semi-automated methods developed and validated by our group (139,163). TMS-EEG will be conducted prior to the start of the course of rTMS and within 48 hours after the last rTMS treatment.
To evaluate TGC, participants will undergo the N-back task during a 10 minute EEG recording at baseline and after the treatment course as part of the TMS-EEG sessions. The N-back task assesses working memory updating by asking participants to determine whether the stimulus (a letter) presented to them is the same as, or different than, the stimulus that was presented to them N trials previously. Participants indicate whether the stimulus presented on the monitor is a similar match, or different, from the stimulus presented 2 trials back. EEG signals of TGC are recorded using DC and a low pass filter of 100 Hz at 20-kHz sampling rate, and data processed offline using MATLAB and EEGLAB toolbox, following previously published protocols (146,165). Theta-Gamma Coupling: The measure of TGC is indexed by the modulation index (MI), calculated for each electrode, followed by an average across the right and left frontal electrodes. The MI for all target trials is analyzed as a weighted average based on the number of correct and incorrect responses, as previously described (146,165).
P5: A preclinical study of the impact of alcohol on executive dysfunctions (ED) and their reversal by novel therapeutic interventions
Background
Preclinical research on substance administration played a critical role in understanding the nature of substance-related cognitive dysfunction. However, studies focused on single cognitive domains, and relatively little is known about the dynamic trajectories of executive dysfunction (i.e. emergence, dependence and potential reversal (166,167)). This study employs a reverse-translational approach, using homologous behavioral tests applied to human participants, but here in rats, to measure executive dysfunction in three domains affected by chronic alcohol consumption (Objective 1), potential reversal via pharmacological intervention (Objective 2), and identification of peripheral and central biomarkers for translation to humans (Objective 3).
The investigation targets specific biological pathways implicated in executive dysfunction in AUD, focusing on GABAergic neurotransmission, the inflammasome system, the noradrenergic system, and the opioid system. The GABAergic system plays a critical role in cognitive functions (168), and is being investigated in P4. Here, we will enhance its activity pharmacologically in rats to investigate its impact on executive functions. The activity of the inflammasome, a multiprotein intracellular complex that detects internal biological stressors and activates pro-inflammatory cytokines, is implicated in brain disorders (169) and alcohol consumption (170), and will be measured in P3 for its translational biomarker value. Here, we will block its activity pharmacologically to assess the impact on executive functions. The integrity of the noradrenergic system, to be measured in P2, plays a critical modulatory role in AUDs (171), with complex effects depending on activity at various receptor subtypes (α1 versus α2). Here we propose to target the α2-receptor, using guanfacine, an α2-receptor agonist that demonstrated efficacy at reducing alcohol consumption in rats (172) and is used for the treatment of attention-deficit/hyperactivity disorder (173). Finally, naltrexone is a prescribed medication used to manage alcohol dependence, especially consumption, although it appears to have reduced efficacy in certain patients, depending on gender and genetic differences (174). Naltrexone is predicted to be prescribed in a significant proportion of P1 participants. Its effects on executive functions are not fully characterized and tend to vary from study to study (175,176). Here, we propose to systematically investigate the relationship between its known clinical efficacy on consumption and potential improvement of executive functions in rats, providing a clinically-relevant comparison to the other interventions.
Objectives
The primary objective of P5 will be to characterize the impact of alcohol on three cognitive domains in rats, identify the impact of novel therapeutic approaches, and delineate novel biomarkers predictive of cognitive deficit emergence or treatment efficacy. We will test the following hypotheses:
H1: Alcohol withdrawal increases working memory deficits, impulsivity and cognitive flexibility deficits in adult rats.
H2: Cognitive dysfunctions observed are linked to changes in central biomarker of synaptic functions, changes in various neurotransmitter related activity (glutamate, GABA, dopamine) and peripheral signs of neuroinflammation.
H3a: Pharmacological interventions targeting the GABAergic system will contribute to better working memory function, cognitive flexibility and reduced impulsivity during withdrawal.
H3b: Pharmacological interventions targeting the dopaminergic system will contribute to better cognitive functions and reduced alcohol seeking behaviors.
H3c: Pharmacological interventions targeting the inflammasome will reduce signs of neuroinflammation, indirectly contributing to better cognitive health.
This project will also generate a sample repository (brain, blood, plasma) for investigation and discovery of peripheral and central biomarkers indexing putative mechanisms involved in AUD and executive dysfunction, that could be used for follow-up biomarker analyses.
Methods and Planned Analyses
Male and female Long Evans rats, 3 months old at study initiation, will be utilized in this study. Rats will be habituated to consume a liquid diet (Nestlé® Chocolate Protein Boost) in which ethanol can be diluted. After baseline assessment of executive functions, half of each cohort will be exposed to a diet supplemented with 10% w/v alcohol 5 days per week. Removal on 2 consecutive days per week will induce a withdrawal period. Such cycle will be repeated 4 times, with assessment of cognitive functions during withdrawal after 4 cycles (Figure 4; top panel, orange box)
Measures
Three domains of EF (inhibition, working memory, set-shifting) will be assessed to match assessment in P1. Inhibition will be measured using the 5-Choice Serial Reaction Time Task (5-CSRTT). This test is conducted in operant conditioning boxes with 5 apertures featuring integrated light stimuli (177), and a reward dispenser on the opposite wall. Rats are trained to respond to a stimulus among the 5 apertures, earning a reward on a fixed intertrial interval.
Lengthening this interval challenges rats to wait longer before responding, with premature responses during the intertrial interval serving as a reliable measure of response inhibition (or impulsivity). In alcohol-exposed rats, premature responses are expected to increase, especially when the intertrial interval is extended. Working memory will be assessed using a T-maze spontaneous alternation task (178). After 2 days of habituation, rats explore the maze starting in the maze’s initial arm and then choosing between the two goal arms. The chosen arm is closed off, and the rat returns to the start arm for the next trial. This process is repeated for seven trials. Rats naturally alternate between arms, relying on working memory for novel environment exploration. Alcohol-exposed rats tested during withdrawal are expected to show diminished working memory functions Set-shifting will be evaluated following Floresco et al.’s protocol (179), in which rats undergo initial training in standard operant conditioning chambers with a house light, two retractable levers, and stimulus lights. Training involves a fixed ratio 1 schedule on each lever, reaching a criterion of 50 rewards or 30 minutes per lever. The second phase introduces a 90-trial session where rats must press a lever within 10 seconds to receive a reward. Visual cue discrimination training follows, with stimulus lights above the levers. Rats respond within 10 seconds to the lit lever for a sucrose reward, while responses on the non-lit lever are punished. Criterion: 10 consecutive correct responses for 30 trials over 4 days. After visual discrimination is learned, rats are tested in a “shift to response discrimination” task, where the rewarded lever is consistent, irrespective of the light stimulus. The number of trials needed to learn this new set serves as a proxy for set-shifting. Rules are modified after each testing session to ensure continual adaptation.
Due to the complexity of the inhibition and set-shifting task, these cannot be run in the same animals. Therefore, different animals will be used to run inhibition+working memory, and set-shifting.
Following the assessment of alcohol-induced cognitive function changes, interventions will commence (Objective #2). Treatments involve intraperitoneal administration of drugs targeting α5-GABAA receptors (GL-II-73), the inflammasome (MCC950), the α2-noradrenergic receptor (guanfacine), and the opioid system (naltrexone).
Initially, acute dosing will be explored using a Latin square design to test multiple doses (Table 3). Conducting multiple doses in the same animal necessitates several testing sessions, and all employed tests, such as the Y maze and 5-CSRTT, are stable and unaffected by practice and repetition (177,180–182). The set-shifting task (179) can be adapted to establish new rules at each testing session, enabling regular assessment of set-shifting abilities in the same animals while mitigating the impact of practice on repeated testing.
Blood samples will be drawn weekly and 24 hours after discontinuation of the liquid diet (Figure 4) to verify 1) blood alcohol levels (BALs) exceeding 100mg% during exposure and 2) the absence of alcohol 24 hours after cessation when behavioral testing begins. BALs will be assessed using an ANALOX machine. Serum samples will also be utilized for measuring peripheral inflammasome biomarkers. Inflammasome activity will be determined by the phosphorylated/non-phosphorylated NALP3 ratio in peripheral blood mononuclear cells, along with investigating inflammatory markers (NF-ΚB, ERK, p38MAPK) and cytokines (IL-1β, IL-18, Caspase-1). Luminex xMAP technology® will be employed for biomarker analyses.
After behavioral assessments, rats will be euthanized, and brains dissected for immunofluorescence staining and molecular analyses on regions of interest (prefrontal cortex, VTA, nucleus accumbens and hippocampus). Rostro-caudal coronal sections will be stored at - 20°C for neuroinflammation, noradrenergic, and GABAergic marker analysis. RNA and proteins will be extracted from homogenized tissue, allowing for cDNA conversion, quantitative PCR, and Western Blot analyses. The markers examined include, but are not limited to, SST, PV, α5-GABRA, α1 GABRA, BDNF, and other GABAergic markers. Similarly, qPCR, immunostaining, and Western Blot analyses against sterile inflammation biomarkers (NALP3, NF-ΚB, ERK, p38MAPK, IL-1β, IL-18, Caspase-1) will validate the correlation of peripheral and central biomarkers in a region-specific manner. Investigation of changes in noradrenergic system markers (NET, α1 receptor, α2 receptor) will complement NM-MRI work outlined in P2.
The blood and brain samples collected will form a repository for subsequent biomarker analyses related to executive dysfunction in AUD. As Objective #3 establishes this repository, the analysis of these samples (peripheral or central) will be influenced by findings from other projects. Consequently, additional exploratory investigations of biomarkers will be conducted in these samples when other projects identify relevant pathways or markers in the clinical population.
Proposed Analysis
Behavioral data will be examined to assess the impact of alcohol exposure on executive functions, utilizing ANOVA (or relevant non-parametric alternatives) followed by Tukey’s post hoc comparisons for statistical significance. In the alcohol-exposed groups, it is possible to observe subgroups of animals with higher susceptibility to alcohol than the others, suggesting existence of high-responders and low-responders. For data collected weekly, data will be analyzed with repeated measures to identify animals that are more than 50% of the time over the mean, and animals that are more than 50% of the time under the mean to segregate high-responders and low-responders respectively. If such separation does not exist consistently, then analyses will focus on overall alcohol effect. For the second objective involving drug interactions, separate 2-way ANOVA with diet and treatment as factors will be applied, followed by Dunnett’s post-hoc comparisons (182). In Objective #3, blood sample data will be analyzed using Student’s t-test (Objective #1) or 2-way ANOVA (Objective #2) with diet and treatment as factors, followed by Dunnett’s post-hoc test. Brain analyses with qPCR and Western Blot will use 2-way ANOVA with diet and treatment as factors, followed by Dunnett’s post-hoc comparisons. Pearson or Spearman analyses will evaluate correlations between executive dysfunction and peripheral/central biomarker levels, with p-values adjusted for multiple comparisons using the Bonferroni method due to the potential high number of pairwise analyses.
P6: To assess links between EF and treatment seeking for addiction and healthcare utilization and costs
Background
Among individuals with a SUD, EF can affect retention (183–185) and engagement in treatment as well as treatment outcomes (37,186), including the ability to achieve and maintain abstinence (183–185,187). Given that SUDs are strongly associated with health care utilization (39), it is also likely that EF, which impacts the success of SUD treatment, is a key factor affecting the extent of health care utilization. In particular, among those with EF deficits, a diminished ability to achieve abstinence or low-risk alcohol use may lead to a higher use of emergency services for the acute consequences associated with SUDs (188). This could result in increased health care costs (189), which is problematic when existing resources dedicated to emergency services are limited (189,190). Furthermore, a diminished ability to achieve abstinence or low-risk substance use may also lead to a higher use of non-emergency health care services, including primary care visits and hospitalizations, due to acute and chronic health effects associated with substance use (191,192).
Furthermore, previous research has shown that EF is a factor in engagement in SUD treatment, including more treatment episodes and longer durations of abstinence among those without executive function deficits (187). EF deficits lead to worse recognition of problem use (193–195). Therefore, people with executive function deficits are more likely to view their symptoms as temporary or not serious, and consequently do not seek treatment (38). These factors may also affect whether patients stay in treatment or remain abstinent after treatment. Accordingly, it is likely that executive function deficits affect treatment seeking behaviours, experiences, and barriers among people with SUDs.
Objectives
P6 has two objectives. The first is to qualitatively assess the potential impact of EF on treatment-seeking behaviors, experiences, and barriers, specifically among individuals with AUD. The second objective is to quantitatively assess the association of executive function with health care utilization among individuals with SUD, and to estimate the costs of this utilization.
H1: EF deficits will negatively impact treatment seeking and the treatment process as expressed by participants in treatment for AUD.
H2: Emergency and non-emergency health care utilization and related costs will be higher among participants in treatment for SUD with more severe EF deficits than among those with less severe EF deficits.
Methods and Planned Analyses
Objective 1 (Qualitative component)
A semi-structured, one-on-one 90-minute interview with a subset of participants recruited from the pooled CDiA sample will be conducted by a research coordinator trained in qualitative methods including interviewing and analysis, shortly after their baseline data are collected. Given differences in treatment pathways and trajectories and their potential impacts on treatment experiences, this analysis will be limited to participants with AUD. Interview questions will be related to treatment-seeking behaviors and executive functioning. Participants will be asked about their experiences seeking help for substance use, perceived facilitators and barriers to treatment, perceptions and expectations regarding treatment, past health care utilization, and questions related to how executive function (e.g., memory, attention, impulsivity, etc.) may have played a potential role in their treatment-seeking behavior and experiences.
This study will undertake a thematic analysis of the data obtained, examining and identifying common themes that emerge related to the perceived impact of executive function on treatment-seeking behaviors, experiences, and barriers across the sample. We expect that thematic data saturation will be met with a sample size of approximately 30 participants.
All interview transcripts will be manually reviewed by a member of the research team, and an initial codebook will be prepared based on preliminary analyses. All transcripts will be coded for the initial set of codebook themes using NVivo software. As we anticipate both inductive and deductive themes to emerge from the data, the codebook will be subject to further development and revision based on ongoing analyses and discussion of emergent themes among members of the research team. Data analysis will be finalized once we have met data saturation, and no new themes emerge. Final results will be informed by common themes that emerged and were expressed by multiple participants. To increase data analysis credibility, a secondary independent coder will code a sub-sample of transcripts using the coding framework to assess inter-coder reliability, and any discrepancies between codes or themes will be discussed.
Subsequent to the interview, participants will be asked to complete the Behavior Rating Inventory for Executive Functioning in Adults (BRIEF-A), a standardized validated assessment that captures self-reported perceptions of an individual’s executive functions and self-regulation (196). The results of the BRIEF-A assessment tool will be used to better understand the participants’ executive functioning from their perspective, which will further inform our qualitative analysis.
Objective 2 (Quantitative component)
A health systems evaluation will be conducted to assess the association between EF (as measured in P1) and health care utilization, as well as the costs of this utilization. For this analysis, the database constructed in P1 will be linked with data on health care utilization and the associated costs obtained from the Institute for Clinical Evaluative Sciences (ICES). ICES will provide data on the use of publicly-funded health services in Ontario. Participant data will be linked through ICES to the Hospital Discharge Abstract Database (DAD), National Ambulatory Care Reporting System (NACRS), Continuing Care Reporting System (CCRS), Ontario Health Insurance Plan Claims Database (OHIP), and Ontario Drug Benefit Claims (ODB) database. Records-level ICES data linkages will be performed using each participant’s unique Ontario Health Insurance Program health card number; the resulting database will be stripped of all direct personal identifiers and each entry will be assigned a confidential code number. All data linkages will be performed by an ICES data analyst. A subanalysis will be performed stratifying by conditions fully attributable and those not fully attributable to SUD.
Analyses will focus on the occurrence of service utilization, stratified by the type of service (i.e., non-emergency services and emergency services). The occurrence of health care service use among the cohort (recurrent time-to-event data) will be analyzed using multiple models, including covariates across multi-disciplinary domains. These models will include the Andersen and Gill model (i.e., generalizes the Cox model), Prentice, Williams and Peterson (PWP) model, marginal means/rates model, and frailty model (i.e., random effects model) (197–200). The final model to be used will be chosen based upon which of the previously used models best describes the data and which model assumptions are upheld. Model assumptions include the occurrence of an event being dependent upon the prior number of events during the follow-up period (PWP), there being no time-dependent covariates (marginal means/rates model), and the occurrence of an event cannot be explained by observed covariates alone (frailty model). Each statistical model will include covariates for the domains of EF, demographics, substance use, and health service utilization prior to study participation.
P7: To identify subtypes of individuals seeking addiction treatment using cross-disciplinary data types from all projects to and map the biopsychosocial all-cause drivers of cognitive dysfunction in adult outpatients seeking treatment for AUD/SUD
Background
In SUDs, heterogeneity is the rule rather than the exception (201). This heterogeneity is largely driven by extremely high rates of comorbid psychiatric conditions (202–204) and compounded challenges in the diagnostic process, as emergent cognitive syndromes may be temporary, drug induced, or chronic (205,206). Even when considering the same SUD, symptoms and prognoses vary wildly between patients due to complex combinations of life experience (207), biology (208), and socio-demographic factors (209,210). Given this heterogeneity, it is likely that there exist subpopulations, or subtypes (211–213), of individuals with similar diagnostic classifications but different underlying disease mechanisms and degrees of cognitive impairment. Due to these mechanistic differences, interventions aimed at improving deficits in executive function (214) would not be expected to work equally well in different subtypes, meaning their identification is a priority for future applications of precision medicine in SUDs.
Attempts to define subgroups within and across diagnoses with relatively homogeneous symptom profiles and outcomes have met with limited success; for example, a five-biotype model of alcohol dependence has been proposed using latent class modeling of family history, age at onset, DSM-IV criteria, and data on comorbid illnesses (215). Others have proposed four-and two-group subtyping schemes based on psychiatric co-morbidity, family history, psychopathy, temperament, and other clinical health measures (216). Despite the strong links between symptom dimensions and executive dysfunction (see P1), subtypes have not been formally evaluated for differences in domains of inhibition, set shifting, or working memory updating, leaving a major gap in the field. To their detriment, most efforts have focused on relatively few domains of interest, not examined changes in cognitive performance (specifically executive function) over time, and yielded varying, unstable results. Therefore, it is important to approach the problem from a data-inclusive perspective (217). Using insights from unimodal investigations, such as those proposed in the preceding CDiA projects, it is also possible to assemble optimal parsimonious predictive models at the individual level which could aid in screening efforts and clinical decision making.
Objectives
In this project, we propose to integrate data collected from each of the proceeding projects – including use of insights from animal experiments in P5 – to (1) perform multi-modal subtyping on SUD individuals, (2) evaluate subtypes against known biological risk profiles for cognitive dysfunction and psychiatric illness (such as exploring polygenic risk scores) and other longitudinal outcomes (e.g. TMS treatment outcome, specified in P4), and (3) use machine learning approaches to optimize predictors of cognitive dysfunction over time and to understand the structure of modifiable risk and resilience factors that may present opportunities for intervention. No additional data will be collected for this project; it will be computational in nature.
H1: Multi-modal unsupervised and semi-supervised clustering of SUD patients at baseline will reveal transdiagnostic subtypes of SUD with unique risk factor profiles and levels of executive functioning.
H2: Subtypes from Aim 1 will be predictive of changes in executive function over one year of follow-up and individuals with evidence for subgroup transition will experience different treatment response rates.
H3: Multivariate predictive models of changes in executive function over time, incorporating the available multi-modal feature space of CDiA, will offer better performance than models based on individual modalities.
Methods and Planned Analyses
Cross-sectional data-driven subtyping
This will involve using subject clustering algorithms to identify relatively homogeneous subgroups of patients. Primary measurements from each of P1-6 will be used as input for clustering (see Table 4 for domain definitions). The similarity network fusion (SNF) method will then be used to combine more fine-detailed, high-dimensional input variables across multiple modalities. To address the key overarching theme of this proposal, subgroups based on executive function domains (including and excluding other measures) will be of particular interest, and will be tested for distributional differences in primary prognostic outcomes of functional recovery and sustained health. The semi-supervised association-boosted SNF (abSNF) method will be used to identify subgroups with most significance to executive function performance at baseline. Internal validation of group membership will be performed using bootstrap procedures. Socio-demographic, educational, and life experiential characteristics of each subgroup will be described and placed within the context of existing cross-sectional work.
Summary of assessment domains and measurements across study projects.
Longitudinal stability of baseline subgroups and subgroup transitions
Within- and across-data type clustering will be performed independently at multiple time points and evaluate the frequencies of individual-level membership transitions. This aim is similar in nature to Aim 2 of Project 1. However, the methods are adapted to account for the increased diversity and dimensionality of input data that will be used to identify latent subgroups. In project 1, latent growth curve modelling will be used to determine if individual trajectories of substance use and functional outcomes are predicted by baseline executive functions. Here in P7, we will extend the foundational work of P1 by examining patient trajectories based on data collected across projects 1-6 and determined using different, network-based methodologies. Critically, we will also identify socio-demographic, educational, and life experiential factors determining inconsistency in group membership over time. Subtypes based on integrative data analyses will be evaluated against more traditional clinical subtypes (such as those from P1) to determine if distributions of key demographics and outcomes are modified by the inclusion of multi-disciplinary input features.
Supervised Machine Learning for prognostication
We will apply machine learning methods on features identified by each project across domains to build optimally predictive models of: (1) executive function, (2) social function, (3) suicidality and self-harm, and (4) healthcare system burden (218). Modelling will proceed iteratively, using processes developed by our group (219–221)
Discussion
There is widespread recognition of the need for integrative and longitudinal research incorporating cognitive, biological, and clinical measures of vulnerability and resilience in SUDs. Within the CDiA Program structure, detailed clinical and cognitive characteristics of a treatment-seeking sample will be integrated with biological markers closely linked to both cognitive dysfunction and SUDs, permitting the identification of critical mechanisms and therapeutic targets. Linkages to healthcare utilization and innovative whole-person modeling approaches will further take this initiative from “neuron-to-neighborhood” and foster integrative and translational research consistent with recent calls to action in this field.
A main strength and innovation of the CDiA Program consists of its translational team and unique collaborative research approach on executive dysfunction for functional recovery and sustained health in outpatients seeking treatment for SUDs (Figure 1). Other conceptual advantages of the Program are the adoption of an individual differences approach, and the recognition that transformation of care can only occur when combining expertise from clinical neuroscience, clinical care, clinical research, and epidemiology/health science research perspectives. As depicted in Figure 1, P1-4 are closely linked investigations with a focus on collecting deep data in complex patients seeking treatment for SUDs; these human subjects research projects emphasize ecological validity and will occur in parallel with P5, which offers the opportunity to investigate associations between EFs, proposed biological underpinnings, and associated pharmacological therapeutics in a highly controlled and efficient design. P6 and P7 extend the generativity of this work, with a focus on healthcare utilization and costs, as well as lived experience perspectives, and the application of highly sophisticated, advanced computational modeling to integrate data across all other projects. To further demonstrate the intended integrated approach across projects, we provide here selected examples of potential for individual projects benefiting and feeding back on strategy, approach and deliverables of other projects (see Table 5).
Cross Disciplinary Collaboration Across Projects
Our inclusive approach to participant recruitment will allow us to recruit a large sample representative of the complex clinical populations typically seen in tertiary care settings. We recognize this will also pose analytical challenges related to sample heterogeneity. Our reliance on multi-level measures of transdiagnostic relevance (e.g., EF, functioning, neural circuit function, inflammatory markers) and our state-of-the-art data integrative approaches outlined in P7 place the CDiA Program in a strong position to harness this heterogeneity to improve understanding of “real-world” presentations of SUDs and their cognitive features. This can in turn accelerate impactful discovery science, clinical translation, and societal/policy changes needed to improve outcomes in this population.
It is anticipated that following the recruitment of this initial cohort, subsequent research will investigate the impact of specific interventions (33,222) on impaired EFs and functional recovery. Thus, it is the vision to use these results to support an “experimental medicine” approach aimed at remediating cognitive dysfunction. Although both psychosocial (e.g., cognitive training) and biological (e.g., pharmacotherapy, noninvasive brain stimulation) interventions are forecasted, the first planned intervention will investigate the efficacy of rTMS on outcomes in SUDs. Peripheral and imaging biomarkers identified in our initial cohort will permit such a nuanced investigation of mediators of clinical outcomes, and in conjunction with preclinical investigations, will inform the development of novel therapeutics (223,224). As an early testament to the generativity of CDiA, two substudies have already been launched: the first aims to assess links between lab-based cognitive and biological measures, and real-world metrics of activity measured by wearable technology, and the second seeks to evaluate the impact of a novel rTMS intervention for treatment of cannabis use disorder (NCT05859347).
Taken together, through its core and embedded projects, the CDiA Program aims to improve precision medicine by understanding the heterogeneity of SUDs, subtyping patients with diverse SUD presentations, discovering underlying neurobiological mechanisms, and identifying targets for novel therapies and therapeutics.
Ethics Statement
The study was approved by the Research Ethics Board (REB) at the Centre for Addiction and Mental Health (CAMH), REB #: 010/2021. The participants will provide written informed consent to take part in the study. Animal procedures were reviewed and approved by CAMH Animal Care Committee, in accordance with the guidelines of the Canadian Council for Animal Care.
Authors’ Contributions
PP, LCQ, JDQ,ACR, TDP,YSN, DF, DVC, DV and MES assisted with writing the original draft and review editing. SL assisted with writing the original draft. LQ, KS, WW, DMB, OCM, ANH, VMT, RND SKJ, CR, ES, ASS, BF, PG, JO, ELMV, RNW, KPC and SW assisted with review editing.
DMB, JO, BF, DF, LCQ, DVC, JDF, TDP, ACR, SW, YSN, ES and PP assisted with conceptualization. BF, DF, LCQ, DVC, JDW, TDP, ACR, RND, ELMV worked on investigation. KPC, RNW, PP assisted with software. DF, LCQ, DVC, LQ, RND, RNW, PP assisted with software data curation. DF, LCQ, DVC, JDW, TDP, ACR, SW, YSN, LQ, RND, ELMV, KS, VMT helped with methodology. LCQ, DVC, TDP, ACR provided Supervision. DF, LCQ, DCV, RND assisted with formal analysis. LCQ, ACR, ES, PP, RND, CR helped with project administration. LCQ, DCV, RND, RNW assisted with validation. DF, LCQ, JDW, TDP, ACR, YSN assisted with funding acquisition. DF, LCQ, YSN, RND, ELMV, RNW helps with Resources. DF, LCQ, DCV, YSN, PP, RND, RNW assisted with visualization.
CDiA Program Study Group consists of Lena Quilty, Etienne Sibille, Daniel Felsky, Shannon Lange, Thomas Prevot, Yuliya Nikolova, Anthony Ruocco, Jeffrey Wardell, Erica Vieira, Daphne Voineskos, Isabelle Boileau, Nikki Bozinoff, Daniel Blumberg, Christopher Bowie, Paul Fletcher, Phil Gerretsen, Ahmed Hassan, Colin Hawco, Yarissa Herman, Pamela Kaduri, Paul Kurdyak, Bernard Le Foll, Osnat Melamed, Leanne Quigley, Jürrgen Rehm, Kevin Shield, Matthew Sloan, Victor Tang, Wei Wang and Samantha Wells.
Data Availability
This is a protocol paper, which does not include any data.
COI Statement
DMB received research support and in-kind equipment support for an investigator-initiated study from Brainsway Ltd. He was the site principal investigator for three sponsor-initiated studies for Brainsway Ltd. He also received in-kind equipment support from Magventure for two investigator-initiated studies. He received medication supplies for an investigator-initiated trial from Indivior. He is a scientific advisor for Sooma Medical. He is the Co-Chair of the Clinical Standards Committee of the Clinical TMS Society (unpaid).
TDP and ES are listed as inventors on patents covering the use of molecules proposed to be tested in the described study. TDP also acts as the Director of Preclinical Research and Development of DAMONA Pharmaceuticals, a spin-off company from CAMH aiming at leading the licensed technology to the clinical for reduction of cognitive deficits in brain disorders. ES acts as the CSO and Founder of DAMONA Pharmaceuticals.
Acknowledgements
The CDiA Program is supported by a Discovery Fund grant administered through the Centre for Addiction & Mental Health Foundation. The authors gratefully acknowledge the CDiA Program Study Group for their contributions to this research. The authors thank the participants of the CDiA Program. We also thank the CDiA Lived Expertise Research Advisory, CDiA Scientific Advisory Board, and the clinical team members of the CAMH Addictions Program for their contributions and support of this work.
YSN is supported by a Koerner New Scientist Award and a Paul Garfinkel Catalyst Award administered by the CAMH Foundation. DF is supported by CIHR, the Koerner Family Foundation and the Krembil Foundation. DMB receives research support from CIHR, NIMH (R01MH112815;R21MH128815; R01MH1192850), Brain Canada and the Temerty Family through the CAMH Foundation and the Campbell Family Research Institute.
The neurophysiological analysis for Project 5 was partly supported by the Labatt Family Network for Research on the Biology of Depression.
Footnotes
↵1 SUD refers to alcohol, cannabis, hallucinogen, opioid, stimulant, and tobacco use disorders.
References
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.
- 17.
- 18.
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.
- 28.↵
- 29.↵
- 30.↵
- 31.
- 32.↵
- 33.↵
- 34.↵
- 35.
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.
- 55.↵
- 56.↵
- 57.↵
- 58.
- 59.
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.
- 82.↵
- 83.↵
- 84.↵
- 85.↵
- 86.↵
- 87.↵
- 88.↵
- 89.↵
- 90.↵
- 91.↵
- 92.↵
- 93.↵
- 94.↵
- 95.↵
- 96.↵
- 97.↵
- 98.↵
- 99.↵
- 100.↵
- 101.
- 102.↵
- 103.↵
- 104.↵
- 105.↵
- 106.↵
- 107.↵
- 108.↵
- 109.
- 110.↵
- 111.↵
- 112.
- 113.
- 114.
- 115.
- 116.
- 117.↵
- 118.↵
- 119.↵
- 120.
- 121.
- 122.
- 123.↵
- 124.↵
- 125.↵
- 126.↵
- 127.↵
- 128.↵
- 129.↵
- 130.↵
- 131.↵
- 132.↵
- 133.↵
- 134.↵
- 135.↵
- 136.↵
- 137.↵
- 138.
- 139.↵
- 140.
- 141.↵
- 142.↵
- 143.↵
- 144.↵
- 145.↵
- 146.↵
- 147.↵
- 148.↵
- 149.↵
- 150.↵
- 151.↵
- 152.↵
- 153.↵
- 154.↵
- 155.↵
- 156.
- 157.↵
- 158.
- 159.
- 160.
- 161.↵
- 162.↵
- 163.↵
- 164.↵
- 165.↵
- 166.↵
- 167.↵
- 168.↵
- 169.↵
- 170.↵
- 171.↵
- 172.↵
- 173.↵
- 174.↵
- 175.↵
- 176.↵
- 177.↵
- 178.↵
- 179.↵
- 180.↵
- 181.
- 182.↵
- 183.↵
- 184.
- 185.↵
- 186.↵
- 187.↵
- 188.↵
- 189.↵
- 190.↵
- 191.↵
- 192.↵
- 193.↵
- 194.
- 195.↵
- 196.↵
- 197.↵
- 198.
- 199.
- 200.↵
- 201.↵
- 202.↵
- 203.
- 204.↵
- 205.↵
- 206.↵
- 207.↵
- 208.↵
- 209.↵
- 210.↵
- 211.↵
- 212.
- 213.↵
- 214.↵
- 215.↵
- 216.↵
- 217.↵
- 218.↵
- 219.↵
- 220.
- 221.↵
- 222.↵
- 223.↵
- 224.↵
