NIH Funding Opportunity: Longitudinal Study of Neurodevelopmental Consequences of Substance Use: Adolescent Brain Cognitive Development (ABCD)
SFN MEETING SUMMARY
NOVEMBER 17, 2014
6:30PM TO 8:30PM
On November 17, 2014 the Collaborative Research on Addiction (CRAN) at NIH [National Institute on Drug Abuse (NIDA), National Institute on Alcohol Abuse and Alcoholism (NIAAA), and the National Cancer Institute (NCI)] and the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) hosted a satellite session at the Society for Neuroscience to solicit input from the scientific community for the planning of a large-scale prospective project, known as the Adolescent Brain Cognitive Development (ABCD) study, to assess developmental effects of substance exposure on brain development, functional outcomes, and individual trajectories. CRAN leadership--Drs. Nora Volkow, George Koob, and Bob Croyle; and Dr. Alan Guttmacher, Director, NICHD, provided welcoming remarks and an overview of the strategy used to developing this NIH Funding Opportunity, as well as the currently proposed study design. Dr. Michael Charness, Veteran’s Affairs Boston Healthcare System, Harvard Medical School, and Boston University School of Medicine, moderated the discussion that followed, which was guided by the central research questions and responses obtained from a Request for Information published in July 2014 (http://grants.nih.gov/grants/guide/notice-files/NOT-DA-14-014.html). Meeting attendees were asked to share their thoughts and provide feedback based on their experiences and expertise. The following summarizes the discussion.
Category 1: Sampling Strategy
What is the optimal sampling strategies to establish a rich, informative, and representative community-based cohort?
RFI respondents cited sampling and design as the key issues to consider. While they were split on the optimal approach (oversampling high risk cohorts vs. maintain a population-representative sample), their responses highlighted a need to clearly identify who the target population would be.
- Consideration should be given to the environmental, technological, and other changes expected to occur during the study period and their impact on results and their generalizability. For example, a cohort of same age will be confounded by changes in scanner technology, drug availability, etc. Thus, need to weight the proposed sampling and design strategy of a single cohort assessed prior to exposure vs. multiple cohorts, with some already exposed to substances.
- Strong recommendation for over sampling of at risk populations to ensure sufficient subset using drugs [and developing substance use disorders (SUDs)].
- Consideration of attrition, sampling strategy, and how that affects the end result. Research demonstrates that likelihood of conversion to SUDs is probably low, and therefore, the risk of studying normal development as a result (what is the study goal?).
- Do not necessarily need a SUD diagnosis to measure effects of drug use on the brain, many adolescents using marijuana daily, weekly and /or binge drinking. Consideration to look at dose dependent effects and critical to look at healthy development in adolescents with this sort of exposure. Important to look at healthy development, since exposure is quite high in adolescents.
- Consideration for cohort starting age: Initiation of use may be rare at 9 or 10 years of age, but a young starting age range will impact opportunity to address emerging adults (21-25). However, an older starting age range may result in not capturing enough drug-naïve individuals.
- Initiation of use confounded with SES and race/ethnicity. Depending on cohort, there may be use at younger ages (9-10 years).
- Recommendation to stagger sample to follow different ages for the 5yr funding period.
Category 2: Sample Size (~10,000 participants)
What sample size is required for higher-order interaction effects of substance use exposures on measures of brain structure and function?
RFI respondents suggested a need for additional information to accurately ascertain the necessary sample size. Respondents were in agreement that sample size would need to be large to account for issues in recruitment and retention as well as data validation.
- Consideration for consortia ideal for study. Need sites that can easily recruit 1000 adolescents, but design and site selection should also take into consideration rural communities, so as to not limit study cohort only to urban populations. Flexibility in design of the study and individual settings is recommended, with the possibility that some sites may have fewer subjects than others. Attrition is dependent on expertise and experience; therefore, need for experienced longitudinal researchers is underscored.
- Is it possible to have experienced centers serve as hubs, but also allow for hub and spoke model? There is a risk of failure if the majority of the sites do not have expertise in neuroimaging and recruitment.
While an N of 10,000 sounds like a large number; the reality is that if we want to look at a large number of variables including psychopathology, multiple drug combinations and patterns of use, and also want to balance the sample for certain environmental factors and sex differences, and eventually also assess genomics, we need large numbers of subjects.
Category 3: Data-Sharing Arrangement
What is the appropriate data-sharing arrangement to balance need for open access with incentives for the data-collecting Principle Investigators?
RFI feedback consensus on need for a standardized data management and sharing plan with data becoming available to the public after some protected time (~6 months to 1 year) for primary outcome publication (and possibly secondary) by funded investigators.
Note: NIH prioritizes data access and sharing as rapidly as possible.
- Recommendation to release data sooner rather than later.
- Consideration for amount of maturation needed to ensure neuroimaging data are ready to send out into the community, preference to get out high quality data.
- Challenging balance of technological issues and methodologies ensure data analyses are aligned with data sharing plan.
- Recommendation to release data in format that is as raw as possible.
- Critical to have a data dictionary when releasing data.
- Data release and format are important for validation and replication issues. Consideration for the protection of the integrity of the data and need to answer research questions responsibly.
- Consideration for divergence in the field with differences in expertise of collecting data and allowing analyses by entirely different teams.
- Recommendation to look at Human Connectome Project and their data sharing plan as a model that has been done well.
- Consideration for concerns surrounding pooling of data without hypothesis identification.
- Critical that integrity of data is preserved. An experiment is never designed without a data analysis plan. In order to understand working memory and things that change over the course of development, need to be very specific about planned data analyses for proper interpretation of findings.
Recommendation to share raw data so others can explore. While there are things that we currently know, there is a lot that is still unknown. Allowing for exploration will allow for new and innovative discovery. While integrity of data is critical, the field is conscientious and processes are in place with peer review and publication.
Category 4: Validation and Replication Strategy
What strategy should be employed to replicate/validate findings and avoid false positives due to the large number of experiment-wide comparisons?
RFI feedback suggested utilization of statistical methodologies for validation and replication. Many respondents suggested including statisticians as members of the core research group. Some suggested dividing the sample to allow for internal replication.
- Recommendation to include/require highly trained statisticians as members of core research group.
- Recommendation to design study to allow for adaptation to new technologies and ideas with creative ability.
- Importance of statistical design and standard data dictionary/common data elements across entire study.
- This is most important -- do not leave home without highly trained and powered statisticians to deal with these data, otherwise garbage in and garbage out
- Incorporation of design element for study hubs to address certain questions that have not yet been considered, to address interesting research questions, novel imaging approaches, allowing for innovation. Recommendation to have sub studies that look at emerging research questions.
- Consideration of evolution within the field of SUD (such as e-cigarettes, measures for tobacco use) and required need for adaptability. Recommendation for advisory council to provide guidance on changes in the field.
- Recommendation for utilization of toxicology and biomarkers to measure and quantify drug use. Importance of having basic science researches as members of core research group.
Category 5: Cognitive Domains to be Examined by Neuroimaging
What are the most essential domains to assess via neuroimaging techniques? What tasks have high validity, generalizability, and the potential to be translated for study in animal models?
RFI respondents suggested the following cognitive domains to be measured: impulsivity, inhibitory control, reward processing, stress reactivity, working memory, emotional reactivity, and executive function. Note: Structural and functional connectivity required.
- Agreement that domains identified are appropriate, with suggestion to explore repeated measures aspect (floor and ceiling effects). Consideration should be given to change specific items in the task to reduce practice effects. Solution is not ideal, but there may be statistical approaches to facilitate.
- Current methodological limitations of fMRI were noted—does not necessarily measure basic brain function, such as electrical activity; instead, measurement within a voxel of interest might reflect the balance between inhibition and excitation; therefore, interpretation of results can be somewhat arbitrary. If patient demonstrates reduced BOLD signal relative to control – considered impairment; if patient shows greater BOLD signal compared to control—also considered impairment. Need to be thoughtful about the question and interpretation of the data collected.
- What is the primary focus of study: imaging to predict use or identification of changes based on use?
- Consideration of impact of drug exposure on brain development, how different experiences influence the likelihood to use drugs, emergence of psychopathology during adolescence, and if that influences the use of drugs or if drug use affects psychopathology.
- Recommendation to consider alterations in brain metabolites and energetics over time, which may ultimately provide molecular targets for pharmaceutical interventions (magnetic resonance spectroscopy recommended).
- Suggest reducing the number of domains with possibility of different centers evaluating different functions. Important to have a basic battery across all sites that is guided by research question(s). Review objective of study: interest in normative experimentation and impact on development vs. predictors of addiction.
- Recommendation to have core domains covered at every site with protocol and design that allows for additional domains beyond those suggested above.
- Expert panel recommended the use of imaging techniques to characterize structure and connectivity to identify characteristic patterns of activation that can be predictive of substance use disorders or rates of use.
- Importance of social cognition and social learning. Want to know what happens to children in the actual circumstances in which they use drugs; neuroimaging has paradigms to address this (virtual reality).
- Question from attendees on exclusion criteria for neuroimaging –does a person need to be sober or are we interested in understanding exposure to drugs. Leadership feedback--important for people not to be intoxicated at time of imaging.
- Replication is a critical component.
- Critical to consider class of imaging with directional hypotheses and description of brain measures that may change.
- Concern that fMRI measures have small signal and large amounts of data manipulation—creating significant variability from site to site and failure to replicate
- Leadership: Brain activity patterns are very sensitive to a variety of measures, but some methodologies are very solid, reproducible across labs and imaging devices. There is a lot of evidence for these applications (e.g., resting state functional connectivity, DTI and others).
- Question whether there are preliminary data to suggest change in the fMRI domains with drug use. Leadership response—yes-- existing studies have looked at consequences of drug exposure on measures such as impulsivity, executive function, delayed discounting, reward sensitivity, etc.
- Leadership reported that all participants will have structural and resting state imaging; other domains would serve as a menu for fMRI options across sites.
- Psychometrics and reliability crucial here with knowledge about nascent variables. Recommendation to have a working group to narrow list of domains with consideration of floor and ceiling effects with age groups.
- Structural components are the standard, functional connectivity has well validated procedures, different machines in different locations can be used and changes in technology can be accounted for. Consideration for design approach, so replication is built into the study. Critical to consider functional measures and potential impact of peers.
- Recommendation to consider tasks that have been adapted for computerized modeling and imaging.
- Take into consideration limitations of neuroimaging and social context level, need to measure things at population level that cannot be measured in a scanner at an individual level.
Category 6: Neurocognitive and Behavioral Domains to Measure (outside the scanner)
What are the most essential domains to assess outside of the scanner? What are the neurocognitive and other behavioral measures that reflect the function of brain circuits most likely to be affected by substance use across development?
RFI responses indicated the importance of acquiring additional behavioral measures outside the scanner. These include well validated protocols that measure: sensitivity to consequences, decision-making, temporal discounting, memory, impulsivity, emotionality, reward responsivity, executive function, attention, and general intelligence.
- Recommendation to add motivation.
- Include symptoms of substance use disorders.
- Incorporate comprehensive mental health evaluation.
- Temperament and personality to identify at risk kids.
- fMRI assessments of habit learning.
- Measurement of bullying, social isolation, academic measures.
- Utilization of offline web based assessments.
Category 7: Psychosocial and Environmental Factors to Measure
What are the most essential domains to assess psychosocial, environmental, and other risk and protective factors?
RFI respondents highlighted the importance of collecting data around environmental and ecological factors and constructs.
- Include subgroups to get to mechanisms of the environment. Use methods of continuous communication of environmental variables with cell phones or tablets
- Measure environmentally-enriching protective factors. Also measure body fat distribution, exercise, etc.
- Inclusion of prenatal toxin exposures, such as lead, that can affect brain dopamine systems.
- Measurement of peer networks.
- Importance of having developmentalist as member of core study group.
- Inclusion of life transitional factors such as sexual experiences and driving and how brain development influences these behaviors.
- Consideration of gaming and impact on brain development.
Category 8: Obstacles and Challenges
What are the obstacles or challenges that are likely to arise in a project of this magnitude and duration?
RFI respondents identified obstacles that overlapped with other inquiry areas (e.g. recruitment and retention, role for environment vs. genetics, impact of different and multiple substances, overall management and coordination, ethics such as incidental findings and how to address early or problematic alcohol, tobacco, or other drug use).
- Incidental findings have to be found by individual that is appropriately trained to find them.
- Consideration for IRB: what happens if adolescent drug use is detected--do you report it to parents, since there is a danger that if parents are informed, kids will not participate? In the example noted, the psychiatrist informs kids ahead of time that drug use will be reported to their parent.
- Leadership recommendation to consider waiver from parents on right to be informed and decision be left to the clinical team.
- Certificate of Confidentiality carries protection from having to inform any parent, agency or official of information obtained in health related research, and even provides protection from disclosure if subpoenaed. However, NIH customarily requires a provision in the consent form that if researcher becomes aware of a serious health or life threatening risk to self or others, disclosure may be made voluntarily (including informing appropriate medical personnel).
- Question whether obligation to inform varies by state and IRB compliance.
- Response from leadership-- Certificate of Confidentiality overrides state provisions. Provisions are dependent on age of the child and emancipation ages that vary by state. The state substitutes for the parent when a minor is not considered emancipated, such as children in foster care, and the state is then the party that must agree to the provisions in the parent’s consent form. The latter may be pre-negotiated with the responsible state agency on a group basis.
- Also stated (Leadership) that the appropriate response to detection of a substance use disorder or other medical problem is to recommend treatment. The study is observational, but that does not remove the ethical obligation to recommend treatment if an illness is detected. Treatment intervention with any individual will not compromise the study and would also not detract from the validity of the study.
Back to ABCD Study Page