early life adversity dimensions, functional connectivity, and change in child behavior

[avannucci]

Data-driven early life adversity dimensions, functional connectivity, and change in internalizing and externalizing behaviors among children

Contributors

Alexis Brieant, Jack Lovell, Hajer Nakua, & Anna Vannucci

Research Question(s)

1. What is the factor structure of early life adversity in the ABCD sample at baseline?

2. What is the relationship between early life adversity dimensions and whole-brain resting state functional connectivity?
   2a. Which experience-sensitive functional connections are common across all adversity dimensions?
   2b. Which experience-sensitive functional connections are unique to specific adversity dimensions?

3. How do associations between early life adversity dimensions and experience-sensitive functional connectivity relate to changes in internalizing and externalizing behaviors after one year?

Description

Rationale

This project seeks to examine the relationships among early life adversity dimensions, neural alterations in whole-brain functional connectivity, and changes in psychopathology symptoms in a large, demographically diverse sample of children based in the U.S. (i.e., ABCD, baseline ages = 9-11). Several dimensions of early life adversity have been proposed and identified in relatively small samples that have a limited range of exposures (e.g., violence exposure, poverty) (Cohodes et al., 2020; Ellis et al., 2009; McLaughlin & Sheridan, 2016; Sheridan et al., 2020). There is considerable ongoing debate about the underlying structure of early life adversity dimensions (McLaughlin et al., 2020; Smith & Pollak, 2020). Moreover, little is known about the relationship between early life adversity and whole-brain resting state functional connectivity, with the majority of studies focusing on targeted neural circuits (e.g., Fareri et al., 2017; Gee et al., 2013; Peverill et al., 2019; Ramphal et al., 2020) or subcortical seed-based functional connectivity (e.g., Barch et al., 2016; Cheng et al., 2020). Further, alterations of functional connectivity linked to early life adversity have been found to predict changes in externalizing (Barch et al. 2018) and internalizing (Gur et al. 2019) behavior; however, these initial studies focused on a narrow range of specific regions or networks. Investigating whole-brain functional connectivity provides an opportunity to assess whether dimensions of early life adversity and psychopathology have widespread implications for brain function in children. Further, examining the relationship between these three constructs can maximize the theoretical and clinical impact of early risk factors linked to developing psychopathology. Harnessing the power of the ABCD sample and predictive modeling techniques will further allow us to evaluate the generalizability of the observed relationships.

Potential Analyses and Questions

1. Early life adversity dimensions: principal components analysis (PCA) on all the early adversity variables

2. Early life adversity dimensions and whole-brain functional connectivity: connectome-based predictive modeling (CPM) using ridge regression with functional connectivity estimates predicting dimension scores OR canonical correlation analysis (CCA) with adversity dimensions and functional connectivity estimates; Benefits of both CCA and CPM is that a single score will be generated for question, but CCA may be harder to interpret than CPM results
   2a. What do you think would be effective approaches to qualitatively evaluate commonalities and differences among experience-sensitive connectomes across dimensions?

3. Early life adversity dimensions, functional connectivity, and change in internalizing and externalizing problems over one year: Multilevel model using a PLS path model analysis OR Structural equation model; causal mediation model: adversity dimension ⇒ experience sensitive network ⇒ within-person change in internalizing and externalizing problems |

4. ReproNim: Given that this sample is large, it is possible to find many relationships that may not be meaningful. We would like to use resampling techniques to assess the stability and generalizability of the results. What predictive modeling / reproNim method(s) is most appropriate (train-test set, bagging, cross-validation, split-half resampling)?

Publishing Questions

1. We would like to do a four-person first authorship because, as a group, we feel that we have equally contributed to the conceptualization of this project and will continue to contribute equally to the data cleaning, data analysis, and write-up of results as a manuscript. Is this possible? What are your recommendations here?

2. We are confused about who the co-authors and senior authorship would be for this manuscript. We want to iron out these details early on.
   2a. Will the ABCD-ReproNim course team be on the manuscript since you all are teaching us these skills and will be helping us with analyses?
   2b. Should we invite our PIs?

3. We wondered about your recommendations for covering the cost of submitting / publishing an article in certain outlets (e.g. DCN, PNAS, etc.).
   3a. Should we ask if our PIs can split the cost (although they may be hesitant since this project is not tied to their grants)?
   3b. Should we just select journals that do not have submission/publication costs?
   3c. Is there a funding mechanism within the ABCD-ReproNim course to cover these costs?

Suggested keywords/tags

early life adversity; functional connectivity; internalizing behaviors; externalizing behaviors

[tsalo]

Would you mind adding a bit of information about the necessary skills and tasks you could use help with in this project? It should make it easier for prospective collaborators to determine how and if they can contribute. Specifically, our new template has sections named "Tools and algorithms to be used", "Skills we could use help with", and "Link to analysis plan" (if you have been working on a more detailed plan than would fit in this issue).

[avannucci]

yes, we are actually meeting this week to create a more detailed plan of action. the larger project is being split up into at least 3 sub-projects (one for each question), and the way you suggested to frame things for other potential collaborators is a great idea!

…

On Tue, Feb 16, 2021 at 6:10 PM Taylor Salo ***@***.***> wrote: Would you mind adding a bit of information about the necessary skills and tasks you could use help with in this project? It should make it easier for prospective collaborators to determine how and if they can contribute. Specifically, our new template has sections named "*Tools and algorithms to be used*", "*Skills we could use help with*", and "*Link to analysis plan*" (if you have been working on a more detailed plan than would fit in this issue). — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#7 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ANA5LIU5A6PWKL3AGGV3PQDS7L3MVANCNFSM4VGKZP4Q> .

-- *Anna Vannucci, MS* Graduate Student Columbia University Graduate School of Arts and Sciences Department of Psychology (Schermerhorn Hall) Developmental Affective Neuroscience Laboratory 1190 Amsterdam Avenue, MC 5501 New York, NY 10027 anna.vannucci@columbia.edu 860-712-1441 https://danlab.psychology.columbia.edu

[angielaird]

Hi team, great topic! In thinking about CPM vs. CCA, I recommend reading the Marek/Tervo-Clemmens et al. preprint, as well as the Taxali et al. Cereb Cortex paper.

Project Week FAQs, including answer to your publishing questions, can be found here. An email will be sent out to all ABCD-ReproNim students with this link in the next day or so.

Looking forward to hearing more about your proposed plan!

[richford]

Very interesting project! We discussed this a little bit on slack and it's exciting to see the project advance. I'd be happy to serve as the TA-liaison for this project.

Also, @avannucci, I noticed that two other projects (#16 and #19) might have a bit of overlap. All of these projects will have to make decisions about which adversity measures to include, what kind of factor analysis you want to do (if any), how to impute missing data, etc. So I made the completely-optional suggestion that the project leads might want to get in touch to discuss their approaches, realizing that you might all use different approaches in the end.

[avannucci]

Thanks @richford - we would love to welcome you as our TA liaison for the project!

I will reach out to projects #16 and #19 about approaches / harmonizing.

[sarahechang123]

Super interesting proposal - would be happy to collaborate if there is an opportunity to contribute! I'm a 3rd year PhD student in Neuroscience at UCLA, broadly interested in functional neuroimaging and internalizing symptoms across typically developing, at-risk, and adolescent-onset psychiatric disorders. Nice to meet y'all!

[jessicabartley]

hi @sarahechang123 - this was actually a project from last year. Unless @avannucci and team are planing on doing a continuation of it during this year's Project Month (which would be awesome!)... I suggest you take a look at the current year's project proposals for you options. And if you don't find any of those interesting then go ahead and propose something yourself!