A longitudinal resource for studying connectome development and its psychiatric associations during childhood

Abstract

Most psychiatric disorders are chronic, associated with high levels of disability and distress, and present during pediatric development. Scientific innovation increasingly allows researchers to probe brain-behavior relationships in the developing human. As a result, ambitions to (1) establish normative pediatric brain development trajectories akin to growth curves, (2) characterize reliable metrics for distinguishing illness, and (3) develop clinically useful tools to assist in the diagnosis and management of mental health and learning disorders have gained significant momentum. To this end, the NKI-Rockland Sample initiative was created to probe lifespan development as a large-scale multimodal dataset. The NKI-Rockland Sample Longitudinal Discovery of Brain Development Trajectories substudy (N = 369) is a 24- to 30-month multi-cohort longitudinal pediatric investigation (ages 6.0-17.0 at enrollment) carried out in a community-ascertained sample. Data include psychiatric diagnostic, medical, behavioral, and cognitive phenotyping, as well as multimodal brain imaging (resting fMRI, diffusion MRI, morphometric MRI, arterial spin labeling), genetics, and actigraphy. Herein, we present the rationale, design, and implementation of the Longitudinal Discovery of Brain Development Trajectories protocol.

Fig. 1

Participant enrollment and attrition. (a) Adapted CONSORT diagram demonstrating group level attrition by longitudinal follow-up track (either 0/12/24 month or 0/15/30 month follow-up) and characterization time-points, (b) Participant-level characterization time-point completion stratified by age at enrollment, females coded as red and males coded as blue, (c) Participant-level days to follow-up characterization visit from baseline, (d) Participant-level days to retest from most recent characterization visit. FLU1 = Follow-up Visit 1 (mid-point visit), FLU2 = Follow-up Visit 2 (final visit).

Fig. 2

Age, sex, and diagnostic distribution of participants. (a) Age and sex distribution of participants. (b) Frequency of diagnoses given to participants during the baseline characterization visit. Participants could receive more than one diagnosis. Diagnostic data for this figure represent the ‘consensus diagnosis’ factoring all data collected during the protocol visit, including the K-SADS, medical history, and clinically-relevant measures. Females are coded as red, and males are coded as blue.

Fig. 3

Distribution of general intellectual ability, academic achievement, and broad mental health measures. Participant IQ was estimated using the WASI-II (top row left to right: Verbal Comprehension Index Composite; Perceptual Reasoning Index Composite; and Full-Scale IQ Composite). Academic Achievement as measured by WIAT-II (middle row left to right: Word Reading; Numerical Operations; and Spelling Standard Scores. Mental Health as measured by CBCL (bottom row left to right: CBCL Total Score; Internalizing Problems Scale; and Externalizing Problems Scale). Data are represented for all participants at baseline.

Fig. 4

Distribution of attention ratings. Total CBCL Attention subdomain distribution compared to total SWAN distribution (left column). Distributions of Conners Inattention compared to SWAN Inattention (middle column) and Conners Hyperactivity compared to SWAN Hyperactivity (right column). The distribution of each scale was tested for normality using the Shapiro-Wilk test. Finally, Pearson (r) and Spearman (p) correlations between the SWAN and Conners are represented for both the hyperactivity and inattention subscales for all participants, those with SWAN > = 0, and participants with SWAN <0.

Fig. 5

Raw and standardized results of the D-KEFS Color-Word Interference Test. From left to right, D-KEFS Color Word Interference Test Color Naming; Inhibition; and Switching Conditions are represented by age for participants (ages 8+) with females coded as red and males coded as blue. Individual lines demonstrate participant-level longitudinal change. Raw scores and error are reflected in Row 1 while standard scores and error are reflected in Row 2. Larger circle sizes indicate more errors.

Fig. 6

Physical measures distribution with age. (a) Time (in seconds) to completion of the Purdue Pegboard task and (b) kilogram-force Grip Strength with age are represented for all participants (age 6–17) with females coded as red and males coded as blue. Color shading degree indicates Tanner stage. Box plots by Tanner stage for (c) time to completion of Purdue Pegboard task and (d) kilogram-force Grip Strength are represented for males and females with * indicating statistical separation between males and females at that Tanner stage (p < 0.05).

Fig. 7

Correlation matrix of phenotypic measures. Heatmap depicting correlations between a broad sampling of behavioral, cognitive, and physical measures. Correlation values represented with color coding survived multiple comparisons correction (false discovery rate; q < 0.05). BMI, Body Mass Index; WASI, Wechsler Abbreviated Scale of Intelligence-II; VCI, Verbal Comprehension Index; PRI, Perceptual Reasoning Index; WIAT, Wechsler Individual Achievement Test-II; Read, Word Reading; Num, Numerical Operations; Spell, Spelling; CBCL, Child Behavior Checklist; Int, Internalizing Subscale; Ext, Externalizing Scale; SWAN; Strengths and Weaknesses of Attention-Deficit/Hyperactivity Disorder Symptoms and Normal Behavior Scale; Hyp, Hyperactivity Subscale; SRS, Social Responsiveness Scale; MASC, Multidimensional Anxiety Scale for Children; ICU-P, Inventory of Callous-Unemotional Traits – Parent Version; Grip, Grip Strength; Hemo, Serum Hemoglobin; Tanner, Tanner Stage; NIDA, National Institute on Drug Abuse Questionnaire; Corr, Correlation.

Fig. 8

Development curves for intracranial, gray matter, white matter, cerebral spinal fluid, and ventricular volume, mean cortical thickness, and mean surface area. Data from all participants were included in these plots, independent of data quality.

Fig. 9

Quality control measures for structural scans. (a) Average Braindr scores per age group. (b) Correspondence between Braindr and Euler number. (c) Receiver operating characteristic (ROC) curve showing the predictability of the Euler number to identify passed (braindr> = 0.5) or failed (braindr < 0.5) structural images.

Fig. 10

Distribution of age and quality control measures for structural scans (Euler Number) and mean Framewise Displacement (FD) for the Diffusion Scans (DTI) and functional scans. BH-1400: Breath holding task with TR = 1400 ms; CB-1400: Checkerboard stimulation task with TR = 1400 ms; pCASL: pseudo-Continuous Arterial Spin Labeling scan; Rest-1400: Multiband resting state scan with TR = 1400 ms; Rest-645: Multiband resting state scan with TR = 645 ms; Rest-CAP: Single band resting state scan with TR = 2500 ms. Correlations between quality control measures are calculated across the whole sample (text in black), or within sex group (red: females; blue: males).

Fig. 11

Respiration rate during fMRI scans across the sample.

Fig. 12

Median framewise displacement across resting state scans (TR = 645/1400/2500 ms) and different filtering strategies (original results, lowpass at 0.1 Hz, lowpass at 0.31 Hz, and notch filter with a central frequency at 0.36 Hz and a width of 0.07 Hz).