Summary of the Expert Panel Meeting on the Development of a National Longitudinal Study of Neurodevelopmental Consequences of Substance Use


Authors: Expert Panel Members

  • Michael E. Charness, M.D.
  • Louise Arseneault, Ph.D.
  • Frank M. Biro, M.D.
  • BJ Casey, Ph.D.
  • Linda Chang, M.D.
  • Raquel Gur, M.D., Ph.D.
  • Rolf Loeber, Ph.D.
  • Robin J. Mermelstein, Ph.D.
  • Adolf Pfefferbaum, M.D.


On May 27-28, 2014, a public expert panel workshop was held to inform the development of a new initiative of the Collaborative Research on Addiction at the NIH, otherwise known as CRAN.  This initiative involves the Institutes which comprise CRAN—the National Institute on Drug Abuse (NIDA), the National Institute on Alcohol Abuse and Alcoholism (NIAAA), and the National Cancer Institute (NCI)—as well as the Eunice Kennedy Shriver Institute of Child Health and Human Development (NICHD). Key staff from each of these Institutes met for several weeks to identify the experts and develop the agenda. In addition, scientific program staff from the National Institute of Mental Health (NIMH) provided technical consultation on the planning committee, and a science writer from NIMH provided background notes of the expert panel discussion for this summary.

Currently known as the National Longitudinal Study of Neurodevelopmental Consequences of Substance Use, this new initiative continues to be under development with an anticipated Notice of Funding Opportunity to be issued by the involved Institutes in 2015.

The goal of the meeting was to begin to develop recommendations on the best large-scale project design and to consider measures to assess developmental effects of substance exposure (particularly nicotine, alcohol, and marijuana but also other drugs), beginning prior to exposure and following participants throughout childhood and into young adulthood.

The charge to the expert panel was given at the meeting opening by the Directors of NIDA, NIAAA, the Director of the Division of Cancer Control and Population Studies, NCI, and the Chief, Child Development and Behavior Branch, NICHD.


Central Research Questions

Early on in the meeting, the panel discussed study aims and formed the following research questions:

  • What is the impact of diverse patterns of use of marijuana, alcohol, nicotine and other substances on the structure and function of the developing brain, as revealed by brain imaging?
  • What are the consequences of substance use on physical health, psychosocial development, information processing, learning, memory, academic achievement, motivation, emotional regulation, and other behaviors? 
  • How does drug use affect the expression of psychopathology, including substance use disorders, and how does the emergence of psychopathology influence drug use?
  • What factors (prenatal exposure, genetic, epigenetic, neurobiological, psychosocial, family history) influence drug use and its consequences during development?
  • In what way does use of each substance contribute to the use of others (gateway interactions)?

General Issues in Study Design

Note: The intent of the workshop was to obtain general design parameters and examples of cognitive neuroimaging and psychological domains that should be collected.  The suggestions of specific imaging protocols, tasks, tests, and screening techniques offered by the panel are seen by the NIH staff as exemplars only and will not necessarily be included in the resulting Notice of Funding Opportunity.

Members of the panel stressed that methods of evaluation should be standardized across all sites and suggested they have the following properties:

  • Be computerized to the extent possible;
  • Require minimal training to produce high reliability among different examiners across sites;
  • Demonstrate validity with repeated testing;
  • Be sensitive at each developmental period throughout the suggested 10-year period of study;
  • Be able to follow an individual’s developmental trajectory (i.e., substance use,  brain development, and other neurobehavioral and functional outcomes);
  • Use test durations that respect the attention span of children at different ages; and include measures that are engaging and age appropriate (fun).


I.    What Kind of Baseline Sample will be Most Informative?

As this is a key to the initiative and essential to the proposed study’s ability to answer the research questions, much discussion time was spent on this topic.  It was also proposed that a sampling expert be consulted to assist in the study design.

  • There was general agreement that the design should be a 10-year longitudinal study of individuals beginning at ~ ages 10-12 and ending at ~19-21. While a sample size of 10,000 had been presented by NIH program staff beforehand, it was suggested that the final N for the study be based on power calculations for adequate response to research questions, with an additional 15% added to account for attrition. The panel also emphasized that investing resources in participant retention is crucial.
  • In addition, there was discussion about the length of time needed for recruitment—for example, a two year period was suggested.
  • The enrollment of siblings, including twins, was also discussed.


Other aspects discussed included the following:

  • Up to, but not more than 20 sites to enroll at least 500 participants per site;
  • Site selection based on capability to enroll participants and conduct all assessments, meeting rigorous standards for imaging, neuropsychological, and behavioral evaluation; and demonstrated ability to retain subjects over long periods of time;
  • Geographical balance to include sites where marijuana is legal, decriminalized, and illegal;
  • Data standardization and harmonization through a central data node or coordinating center for quality assurance and analysis;
  • Broad entry criteria.  Ascertainment from a community-based rather than clinical population – sources could include schools, community centers, and other loci. Other options include point-of-care evaluation by pediatricians and other healthcare professionals during routine checkups;
  • Oversampling of up to 50% to include a high number of predicted heavy users based on entry screening using reliable measures (e.g. family history of substance abuse);
  • Recruit from the community for the “not at risk” sample and more targeted places for at risk samples;
  • Recruitment should account for future attrition; however mechanisms to minimize attrition should be a key characteristic of the study design;
  • At first assessment, collect detailed measures of all potential outcomes that will be assessed at later phases. Collect retrospectively important measures of confounders or other risk factors (e.g., exposure in utero, physical/sexual abuse, etc, if possible from EHRs);
  • Balanced for gender; consider oversampling minorities with known sensitivity to substances (e.g., Native American populations);
  • Baseline evaluation and periodic retesting of multiple modalities, to allow interval analysis and reporting of results throughout the duration of the study;
  • Comprehensive evaluation of brain structure and function.  First two (baseline) imaging studies within 18-24 months of entry; subsequent repeat imaging at intervals of ~two years (total of six imaging studies). More frequent evaluation of drug taking, cognitive function, social development, etc; 
  • Rapid availability of data for analysis by the scientific community with attention to protection of personal health information;
  • Screening for drug intoxication prior to behavioral, cognitive, or functional imaging studies;
  • Minimal or no use of alcohol, marijuana, tobacco, inhalants or other illegal substances at entry; prescribed non-psychoactive medications and stimulants permitted. Prescription of psychoactive medications after enrollment would not disqualify further participation. Careful monitoring of all prescribed medications essential.
  • Exclusion criteria should also include those unable to complete the assessment (e.g., due to language or reading difficulties); moderate and severe TBI at entry, but not following enrollment (TBI should be monitored and analyzed for impact).
  • Proposal for handling of clinically relevant information that emerges from imaging or other psychological/behavioral/functional measures;
  • Engage with the sample, lightly at first to keep contact with participants, and more intensively with onset of substance use;
  • Plan for spin-off studies to explore in more depth specific questions. Include measures for this early on in assessment;
  • Attention to burden for participants—e.g., may be better to have one day of assessment vs. 2.5 days in a year;
  • Collect information to be able to trace participants in the future.


II.   Neuroimaging Data Collection

The next session was devoted to an extensive discussion of the neuroimaging aspects of the study.  Since this is essential, a great deal of time was devoted to this topic.  Notably, four out of the eight expert panelists are leaders in this field and have experience conducting studies on children.

The following is a sample protocol for the basic image acquisition based on the protocols used in the Pediatric Imaging, Neurocognition, and Genetics Study (PING), the Philadelphia Neurodevelopment Cohort (PNC), and the National Consortium on Alcohol and Neurodevelopment in Adolescence (NCANDA) studies.   It was strongly recommended that all imaging be done at 3T and that compatible imaging data be generated across sites.

  • Suggested total time per scanning session would not exceed one hour. The initial and final imaging modules should include two separate sessions of not more than one hour each.
  • Major changes in equipment or software should be preceded and followed by repeated scanning of multiple participants.  Provision should be made for major upgrades to equipment;
  • Members of the panel emphasized the need for experienced human research groups, not just clinical collection sites.  Each site would need to demonstrate ability to build, maintain, and modify all factors in the MR protocols.  It was also stressed that a research agreement with equipment manufacturer is necessary to maintain availability of specific pulse sequences across the timespan of the study.


In the course of the discussion, several additional specific collection parameters were suggested, based on experience from prior studies:

  • Reliability measurement
    • Analysis of imaged data from the same individual scanned at multiple sites (human phantom). 
  • Minimum sample size for trajectory analysis
    • 100 individuals/site imaged at least two times before onset of alcohol or substance use;
    • More than one site per manufacturer to disentangle site effects due to scanner vs. population.

Basic structural and resting fMRI could include, but are not necessarily limited to, the examples below (as suggested by NCANDA investigators):

  • Localizer: 3-Plane Fast Gradient Recalled Echo
  • T1-weighted structural acquisition: 3D Sagittal IRprep SPGR/MP-RAGE
  • T2-weighted structural acquisition: 3D Sagittal Fast Spin-Echo
  • Diffusion Tensor acquisition with PE-Polar spatial distortion correction
  • Dual-echo Fast Spine Echo (FSE) acquisition (for T2 computation)
  • Resting state fMRI (at least 6 min.)
  • Field Map (for Resting state fMRI B0 inhomogeneity correction)


Additional imaging sequences suggested if time permits, include:

  • Susceptibility Weighted Image (SWI) for TBI microbleeds: 3D SPGR
  • 3D Pseudo Continuous Arterial Spin Labeling (PCASL) Perfusion Imaging: Can be done at rest or with a task.


In addition, it was recommended that the image data collection could follow a “tiered” arrangement, where all sites would collect common core scan modalities, but a subset of sites would collect additional data in accordance with their capacities. The following imaging studies could be considered for a second scanning session after accomplishment of the basic protocol. 

  • Task-based functional MRI (fMRI):
    • Memory imaging task that would precisely manipulate both load and duration of items in memory (prefrontal circuitry);
    • Probabilistic associative learning tasks that capture striatal and hippocampal- dependent learning;
    • Computational neuroscience-based tasks that dissociated valence and action, e.g.,  appetitive and aversive Pavlovian and instrumental learning tasks;
    • Tasks that tap attention, cognitive control, motivation and risk aversion; and
    • Tasks that can also be used in animal models.

In general, tasks should be selected to evaluate cognitive control, learning and memory, attention, and/or reward and be sensitive to adolescent-specific development. Networks queried should include frontal lobe, insula, striatum, cerebellum, and limbic system, involving nucleus accumbens, VTA, and amygdala.  Specific tasks should be harmonized across all sites.


III.    Substance Use Data and Biospecimen Collection

This session focused on the domains that are important to include in the longitudinal study as well as the current state of science on biomarkers of substance use.

  • Substance use quantity/frequency to be determined
    • Prevalence, frequency, quantity, and method of administration of substance
    • Link to other measures from existing studies (e.g., National Survey on Drug Use and Health (NSDUH),  Youth Risk Behavior Survey(YRBS), Monitoring the Future (MTF))
    • Yearly or half-yearly assessment period to reduce recall bias
    • Electronic monitoring on different days of the week
    • Seasonable variation of use
  • Substance use symptomatology & disorder
    • Symptom ascertainment and diagnosis at defined intervals
    • More intense measurements during transition periods (e.g., in school, and with respect to brain development) 
    • Measures of abuse and functional impairment
    • Measures of access to and use of services
  • Substance use verification/toxicology (as appropriate, given intermittency of assessment)
    • Assessment of veracity of self-reports
    • Biological markers (tobacco, marijuana vs. other substances)
    • Problems of assessing this for different substances (hair, saliva, urine – limitations of each)
    • Secondary exposure as error
    • Other prescribed medications as sources of variance
    • Dose-response relation to severity of exposure
    • Difficulties to assess quantity of cannabis/purity of drugs
  • Accommodating changes in substance use patterns across waves
    • The key is the ability to do within-individual analyses to narrow down risk factors associated with changes in prevalence, frequency/quantity of various forms of substance use
    • Prospective identification of different trajectories of individuals and groups over time
    • Regular measurement of underlying factors that drive onset:
      • Positive and negative attitudes
      • Positive and negative expectancies
      • Access
      • Pricing
      • Intentions to use
      • Initial sensitivity
      • Use of prescription medications as cognitive enhancers
      • Steroids for performance enhancement
    • Regular measurement of underlying factors that drive onset/frequency of use:
      • Poor inhibitory control/impulsivity
      • Rule breaking
      • Sensation seeking
      • Emotional regulation
      • Anticipation of consequences of behavioral choices
      • Planning ahead
      • Time perspective
    • Changes in prevalence, frequency, quantity, impairment, onset, continuity, periodicity and desistance (quitting or cutting down) of substance use
      • Distinction between experimenters and persisters
      • Measure both the upslope (initiation/escalation) and the downslope (desistance) of substance use by age
      • Downslopes and their causes could be studied in subpopulations
  • Ethical issues/confidentiality in minors
    • Parental consent, minor assent
    • Parent as informant about children
    • Agreement that parent will not request information about their child
    • Confidential setting/procedure for minors
    • Confidentiality of revealed information concerning drug use
    • Appropriate response to indicators of problematic substance use or other mental disorders
    • Consideration of how to address confidentiality/response to suicidal, homicidal ideation
    • Data protection for outside analysts/investigators
    • Problem of deductive identification of participants
    • Confidentiality when using electronic data
  • Genetic/epigenetic samples/testing
    • Whole gene sequencing
    • Selected gene testing
    • Epigenetics: need for repeated samples
    • Data collection:
      • Blood preferable (including through routine draws); saliva if blood is unavailable
    • Genetic (and other measures) from siblings
    • Genetic (and other measures) from biological parent(s)
  • Pregnancy testing
    • Necessary to rule out pregnancy prior to MRI studies
    • Potential for monitoring of pregnancy and prenatal exposure among participants (and their infants)


IV.    Psychosocial/Psychometric and Psychiatric Data Collection

The panel then discussed screening assessments for parents and children of factors that influence or are relevant for substance use and substance use disorders during adolescence and early adulthood. The following were considered essential. 

  • Diagnostic and dimensional measures of psychiatric problems
  • Dimensional measures of problem and risk behavior
    • Conduct disorder, aggression, antisocial behavior
    • Sexual behavior, impulsivity, delay (or lack thereof) of gratification, risk taking, delinquency
  • Measures of psychosocial stressors
    • Trauma/abuse, other adverse childhood experiences, life/home disruptions
  • Measures of affect/emotion
    • Personality
    • Temperament
    • Positive and negative emotional reactivity and regulation
  • Dimensional measures of social behavior
    • Interpersonal relations with peer, family, friends
  • Attitudes (about risk) intentions (to use)
  • Environment
    • Accessibility to substances, second hand exposure
    • Peers: using /selling
    • State-level MJ legalization status; State–level alcohol policies
    • Exposure to prevention programs
  • Academics
    • School attendance
    • IQ
  • Physical Health
    • Body Mass Index (BMI) , height and weight
    • Sleep
    • Exercise

It will be extremely important that all measures used are developmentally appropriate and, as much as possible, measurable across the entire age range of the study without ceiling or floor effects.


V.    Neurocognitive/Task Data Collection

The neurocognitive battery to be included in a multi-site large-scale study should have several features:

  • A neuroscience based computerized battery measuring accuracy and speed of performance;
  • Duration of ~ 1-hour;
  • Easy to administer and incorporate automated scoring;
  • Developmentally sensitive, especially to adolescents;
  • Assess  major neurobehavioral dimensions sensitive to substance use, such as executive function, reward; social cognition, and sensitive to individual differences among likely participants (i.e. not limited by floor and ceiling effects), and if possible, in the public domain;
  • Some measure of premorbid IQ (e.g., reading level); however does not need to be full IQ testing.  Analysis of school records from parents was also recommended.


  How this proposed study can surpass existing studies

  • Very large N with prospective ascertainment and long term follow-up—allowing for analyses of individual drug use patterns and brain development trajectories.
  • Best imaging and technology currently available across a broad distribution of sites (at least 3T).
  • Precise neurobehavioral phenotypes that are developmentally sensitive, and tap into core features of substance use (e.g., impulsivity, cognitive control, motivation, learning and memory).
  • Analytic Strategies
    • Formalize discovery (use subset of data for discovery, identify classifiers, test replication);
    • Data standardization and sharing throughout the study, allowing multiple investigators to conduct secondary analyses;
    • Consider parallel mechanistic studies using animal models to control for variables that cannot be controlled in human studies.



The stated goal of this 1.5 day meeting was to develop general recommendations on the design of a study that can best meet the goal of assessing the pharmacological effects of substance exposure on the developing brain from before first use in childhood into young adulthood.  While the intent was to outline the significant cognitive and imaging domains that need to be included, discussion elicited specific instruments and measures in some places. It is important to note that when the Notice of Funding Opportunity is published (in early FY2015), it is not likely that specific instruments will be prescribed or recommended. Applicant teams will have latitude to describe and justify specific approaches in terms of tasks, measures, and scanning techniques.

The next steps in the development of this initiative include the publishing in the NIH Guide of a formal Request for Information (RFI) that will elicit further comments from scientific and other interested communities on the optimal study design.  In addition, a satellite symposium is planned to solicit additional public comment at the annual Society for Neuroscience meeting, which will take place in Washington, DC in November, 2014.