key: cord-0312957-ddvyiqvt
authors: Strike, Lachlan T.; Hansell, Narelle K.; Chuang, Kai-Hsiang; Miller, Jessica L.; de Zubicaray, Greig I.; Thompson, Paul M.; McMahon, Katie L.; Wright, Margaret J.
title: The Queensland Twin Adolescent Brain Project, a longitudinal study of adolescent brain development
date: 2022-05-20
journal: bioRxiv
DOI: 10.1101/2022.05.19.492753
sha: 9de433c43f399f81c5b083acc74a2f02bcccccc2
doc_id: 312957
cord_uid: ddvyiqvt

We describe the Queensland Twin Adolescent Brain (QTAB) dataset and provide a detailed methodology and technical validation to facilitate data usage. The QTAB dataset comprises multimodal neuroimaging, as well as cognitive and mental health data collected in adolescent twins over two sessions (session 1: N = 422, age 9-14 years; session 2: N = 304, 10-16 years). The MRI protocol consisted of T1-weighted (MP2RAGE), T2-weighted, FLAIR, high-resolution TSE, SWI, resting-state fMRI, DWI, and ASL scans. Two fMRI tasks were added in session 2 to probe emotion-relevant neural processes (emotional conflict task) and evoke activity in the Theory of Mind network (passive movie watching task). Outside of the scanner, we assessed cognitive function using standardised tests. We also obtained self-reports of symptoms for anxiety and depression, perceived stress, sleepiness, pubertal development measures, and risk and protective factors. We additionally collected several biological samples for genomic and metagenomic analysis. The QTAB project was established to promote health-related research in adolescence.

Adolescence is critical for understanding brain changes associated with depression 1-3 , as nearly half of lifetime diagnoses begin by age 14 4 . Adolescents who experience depression are more likely as adults to have poor mental and physical health, lower levels of educational attainment, lower salaries, and more relationship difficulties [5] [6] [7] . During adolescence, the brain's cognitive control, emotion, and reward-related circuitries are undergoing substantial development 8, 9 -developmental changes that can be impaired by restricted sleep, which is common among teenagers 10 . In addition, hormonal surges and consequent physical maturation linked to pubertal development in adolescence are thought to affect multiple aspects of brain development, social cognition, and peer relations 11 .

Sex differences in mental health problems emerge in puberty, a developmental period characterised by rapid increases in estrogen in girls and testosterone in boys. Before puberty, depression occurs relatively infrequently in girls and boys. After puberty, there is a sharp increase in the incidence of depression, with adolescent girls around twice as likely to experience depression as boys 12 . Children who start puberty earlier are at greater risk for depressive symptoms and anxiety, especially girls 13 . Adolescent boys, by contrast, are more likely than girls to develop substance use disorders and die by suicide 14 . Importantly, these pubertal hormones cross the blood-brain barrier, influence brain development, and affect various signalling pathways (e.g., neurotransmitter activity) that underlie mood and cognition 15 . Thus, puberty involves transformation across virtually every psychobiological domain-endocrine, neural, physical, cognitive, and socioemotional-and represents a vulnerable time during which depressive symptoms and psychiatric conditions may emerge.

Magnetic resonance imaging (MRI) allows developmental brain changes to be studied in vivo while retaining the advantage of being non-invasive, thus providing an invaluable tool for longitudinal research in a population sample. In recent years, terrific progress has been made towards characterizing typical adolescent structural and functional development 1, 2, 8, 16, 17 and understanding brain structure and function changes during adolescence associated with various mental health disorders [1] [2] [3] 16 . However, there is still little work in this sensitive period of neurodevelopment using genetically informative samples. Further, prior genetic studies have primarily been cross-sectional, and we lack knowledge of how this critical neurodevelopmental transition during adolescence contributes to optimal cognitive and emotional functioning as well as vulnerability to brain disorders and mental illness. Thus, longitudinal studies are vital to characterise the structural and functional integrity of the brain in genetically informative adolescent samples.

To this end, we present the Queensland Twin Adolescent Brain (QTAB) dataset: a multimodal neuroimaging dataset of Australian adolescent twins with mental health, cognition and social behaviour data collected over two time points (i.e., sessions). Comparisons within and between identical (MZ) and non-identical (DZ) twin pairs -further powered by multiple assessments -provide a rich information source about genetic and environmental contributions to developmental associations, and enable stronger tests of causal hypotheses than do comparisons involving unrelated adolescents 18 .

The QTAB project reflected a concerted effort to: 1. Assess brain development in a large population sample of adolescent twins and disentangle the influence of genetic and environmental factors on neurodevelopmental trajectories. 2. Investigate whether neurodevelopmental trajectories are the same for adolescent boys and girls and whether any differences are linked with pubertal maturation and genetic and environmental factors. 3. Concurrently assess cognitive function to examine how neurodevelopmental changes are associated with the acquisition of complex tasks (e.g., working memory) and behaviour (e.g., emotional functioning) and whether shared genetic or environmental factors underpin associations. 4. Concurrently assess psychiatric symptoms and examine whether those with anxiety or depressive symptoms follow a different developmental pathway, whether they exhibit early or delayed brain maturation, and whether genetic and environmental factors influence these atypical developmental trajectories. Here, we make this multimodal QTAB dataset publicly available and provide an overview of the collected imaging, mental health, cognition, and social behaviour data. In doing so, we hope to make a significant and novel contribution to our understanding of the numerous emotional and behavioural health problems that emerge during the developmental period of adolescence. At the same time, we believe the QTAB dataset provides a significant opportunity for combining data with existing adolescent studies such as the Adolescent Brain Cognitive Development (ABCD) 17 and Lifespan Human Connectome Project in Development (HCP-D) 19 , as well as consortia efforts such as ENIGMA Lifespan 20,21 .

Data collection overview. Data from imaging and questionnaire/cognition components were collected concurrently between twin pairs (i.e., while one twin underwent brain imaging, their co-twin completed the mental health/behavioural questionnaire and cognitive testing). An early morning saliva sample was collected at home and brought to the session, along with any available baby teeth. Hair, blood and urine samples were obtained during the session, with faecal samples and accelerometry measures of sleep behaviour collected at home. Neuroimaging and key phenotypes, saliva and hair samples were collected at sessions 1 and 2.

Structural and functional scans were acquired on a 3T Magnetom Prisma (Siemens Medical Solutions, Erlangen) paired with a 64-channel head coil at the Centre for Advanced Imaging, University of Queensland. Session 1 scans were collected in a fixed order: localisers, 3D T1weighted, two runs of resting-state fMRI (rs-fMRI), four runs of diffusion-weighted imaging (DWI), three runs of high-resolution Turbo Spin Echo (TSE) imaging, 3D fluid-attenuated inversion recovery (FLAIR), 3D T2-weighted, Arterial Spin Labelling (ASL), and 3D susceptibility-weighted-imaging (SWI). In session 2, scan order was very similar, with the addition of two task-based fMRI scans midway through the imaging session, i.e., localisers, T1-weighted, two runs of rs-fMRI, four runs of DWI, two runs of TSE, Partly Cloudy fMRI task, emotional conflict fMRI task, ASL, SWI, FLAIR, T2w. Immediately after each scan, the images were checked for apparent artifacts (e.g. head motion), and scans were re-acquired if necessary. Scan parameters are summarised in Table 2 . Protocol optimisations (e.g. increase in the number of TSE volumes to improve hippocampus and amygdala coverage) and sequence reissues (e.g. to the T1w) that occurred across participants or sessions are noted as a protocol code (detailed in Table 2 

Anatomical A 3D whole-brain T1-weighted image was acquired using a work-in-progress Magnetisation Prepared with 2 Rapid Gradient Echoes (MP2RAGE) sequence. The MP2RAGE sequence combines two individual images acquired at different inversion times (inv-1_MP2RAGE, inv-2_MP2RAGE) to produce an image corrected for B1 and T2* inhomogeneities 25,26 . This "uniform" image (UNIT1) improves T1 contrast compared to conventional T1-weighted protocols. However, a by-product of the MP2RAGE protocol is an amplification of background noise (also known as "salt and pepper" noise), which is problematic for automatic registration and segmentation algorithms. The sequence generates a denoised uniform image (UNIT1_denoised), though some background noise is still present in this image. The MP2RAGE sequence was reissued during session 2, producing a denoised uniform image (UNIT1_denoised) without background noise. T2-weighted and FLAIR whole-brain 3D sequences were acquired to examine brain pathology and to improve the visibility of brain regions bordering CSF (e.g. cortical pial surface, periventricular structures). Due to time constraints, the T2-weighted and FLAIR sequences were excluded for some participants (~5% in session 1, ~10% in session 2). T2weighted TSE scans (slab aligned orthogonally to the hippocampus) were acquired to examine the structural properties of the hippocampus and amygdala at the subdivision level (three runs collected at session 1, reduced to two runs at session 2). Susceptibility SWI scans were acquired to study the pathological mechanisms underlying paediatric brain diseases and disorders. In addition to a combined magnitude and phase image (*_swi), the SWI sequence produces magnitude (*_part-mag_GRE), phase (*_part-phase_GRE), and minimum intensity projection (*_minIP) images.

Multi-shell diffusion weighted images were acquired in four runs with opposing phase encode directions (two A>>P and two P>>A runs). Each run consisted of 23 diffusion weighted volumes (which includes b = 0 s/mm 2 (3 volumes), b = 1000 s/mm 2 (5 volumes), b = 3000 s/mm 2 (15 volumes).

Resting-state scans were acquired consecutively over two runs with reversed-phase encoding directions (anterior-posterior (A>>P), posterior-anterior (P>>A)). Participants were instructed to relax with their eyes open, with an abstract visual stimulus displayed to improve resting-state compliance 27 . The stimulus for functional scans was presented on a backprojection screen that the participants viewed via a mirror attached to the head coil. In session 2, the emotional conflict task was based on a previously characterised emotional conflict task 28,29 and probed emotion-relevant neural processes engaging the amygdala and other structures that represent the processing of emotions and faces. In each trial, participants were presented with a face with a fearful or happy expression and the words "happy" or "fear" written across the face to create congruent and incongruent emotional stimuli. Participants were asked to identify the facial emotion while ignoring the word overlayed on the face. Responses were made via a button press (left button = fearful face, right button = happy face). Stimuli were presented for 1,000 ms, with a varying interstimulus interval of 2,000 -4,000 ms (average 3,000 ms; 163 trials in total), presented in a fixed order, counterbalanced across trial types for expression, word, and sex. Participants completed an in-scanner practice task (8 trials) immediately prior to the task. Stimuli were presented using the Cogent toolbox (www.vislab.ucl.ac.uk/cogent_2000.php) implemented in the MATLAB (www.mathworks.com) programming environment.

Also, in session 2, participants watched a short animated film (Partly Cloudy 30 ) to evoke Activity in the Theory of Mind network and enable the tracking of links between cortical and cognitive changes in adolescents' social development 31,32 . Participants were instructed to remain still and watch the movie. The stimulus was presented using the E-Prime software (Psychology Software Tools, Pittsburgh, PA).

ASL scans provide a measure of cerebral blood flow without the requirement of contrast agents or radiation. In session 1 we used a work-in-progress 2D pseudo-continuous ASL (PCASL) sequence. The product sequence was available at the start of data collection for session 2. In both sessions, a separate M0 scan was acquired using a shortened version of the ASL sequence to produce an improved measure of M0 for calibration purposes. Table 3 provides a detailed list of the scales and tests used to assess puberty, cognition, anxiety and depressive symptoms, emotional and social behaviours, social support and family functioning, stress, sleep and physical health, early life and family demographic factors, dietary behaviour, and COVID-19 pandemic specific assessments. In general, the same measures were collected in sessions 1 and 2, and for several scales, we included both adolescent and parent versions. Selection of tests and self-report scales for phenotypic characterisation was based on validation for use with adolescents (i.e., gold standard tasks with known reliability and applicability to the age range of the QTAB cohort) and ease of administration (i.e., suitability or adaptation for online assessment with an iPad or computer, and whether scales and tests were freely available or provided at a modest cost [e.g. NIH Toolbox Cognition Battery 33 ]). Tests and scales that were widely used were also prioritised and highly considered.

While clinician assessment of sexual maturation is considered superior to self-report, the validation of self-assessment methods has supported their use as an acceptable alternative for research purposes 34 . In QTAB, we assessed pubertal status using a combination of selfand parent-report based on line drawings corresponding to the Tanner stages of pubertal development 35-37 and complementary questions regarding the emergence of secondary sexual characteristics 38 . Further, to aid in tracking pubertal development, saliva samples were collected at each visit and stored for future analysis of sex hormones (see Biological Samples for more information on saliva collection). 40 . In session 2, a novel inclusion was the assessment of social cognition using the Children's Reading Mind in the Eyes Task (RME) 45 and the Empathy Questionnaire for Children and Adolescents (EmQue-CA) 44 , along with the viewing of the Partly Cloudy movie to capture both cognitive and affective Theory of Mind while in the scanner (described in detail above). In addition, we obtained consent to access their National Assessment Program -Literacy and Numeracy (NAPLAN) scores (https://www.nap.edu.au/home). NAPLAN is a standardised national assessment of reading, spelling, grammar and punctuation, writing, and numeracy skills undertaken by Australian children in grades 3, 5, 7, and 9 in both government and non-government schools. Currently, NAPLAN scores are not provided as part of the QTAB dataset; please contact authors Greig de Zubicaray and Katie McMahon for more information.

We focused on symptoms of anxiety and depression, both of which increase in prevalence during adolescence. Depressive symptoms were assessed by self-report with the Short Moods and Feelings Questionnaire (SMFQ) 47 and Somatic and Psychological Health Report (SPHERE-21) 49 . Anxiety symptoms were measured using the Spence Children's Anxiety Scale (SCAS) 46 . We also assessed traits linked to mental health, which may indicate risk of progression or vulnerability. 59 , which provides both dimensional and overall measures of family functioning 81 . Peer influences and family factors are posited to influence adolescent brain development [82] [83] [84] and moderate adolescent risk of experiencing mental health problems [85] [86] [87] [88] .

We included structured assessments to capture perceived stress ( 65 ) stress. Adolescence is a period of increased vulnerability to stressors and to stress-related psychopathology, including anxiety and depression -relationships that may be mediated through the effects of stress on the developing adolescent brain 3, 89 .

During each session, we assessed sleepiness using the Pediatric Daytime Sleepiness Scale (PDSS) 67 , which has robust psychometric properties in adolescents aged 11 to 15 years and is associated with academic achievement and mood. Participants completed a sleep diary, recording their sleep and wake times (i.e., sleep duration) on weekdays and weekends. In addition, we obtained several anthropometric measures (e.g. height, weight). In session 2, we also asked whether they were a morning or evening person (i.e., to provide a proxy measure of chronotype) and assessed general physical activity and device usage (questions from the literature 90 68 and sleep behaviours across early childhood items, as assessed in the Generation R study 69, 70 ).

In addition, at the end of each session, a sub-sample of participants were given a wrist-mounted accelerometry recording device (GENEActiv, Activinsights, Kimbolton, UK) to wear for two weeks on the wrist of their non-dominant hand. Accelerometry devices detect motion and estimate sleep from decreased movement. Participants completed a sleep diary every morning to consolidate the accelerometry data, providing information on bedtimes, wake times, and restorative sleep 93 . On day 15, the devices were returned via postal service and the data was downloaded using GENEActiv software. Currently, shared actigraphy data is restricted to measures of mean sleep onset, wake time, duration, midpoint, and restorative sleep (session 1 only). The raw actigraphy data for sessions 1 and 2 are held by Kathleen Merikangas, National Institute of Mental Health, and Ian Hickie, University of Sydney, Australia, and may be able to be shared by contacting them.

Perinatal and postnatal information (e.g. maternal smoking and drinking; gestation age, birth weight, breastfeeding) were provided by the mother. Questions were adapted from earlier work 94, 95 and other freely available sources (The World Health Organisation Global Adult Tobacco Survey [GATS, 2 nd Edition]). Family demographic information was also provided by the parent attending session 1. These included self-report of ancestry (based on questions included in the 2016 Australian Census) and socioeconomic indexes from Census neighbourhood classifications (Socio-Economic Index for Areas [SEIFA] 71 ), occupation or education (Australian Socioeconomic Index [AUSE106] 22 ), and social status (The MacArthur Scale of Subjective Social Status 72 ), as well as age at twin birth, and current height and weight.

In session 1 the attending parent completed the Australian Child and Adolescent Eating Survey (AES) 73 for each twin. This online food frequency questionnaire assesses the dietary intake of children and adolescents aged 2 to 17 years. It includes the frequency of consumption of 120 common foods, use of supplements, as well as eating and behaviours. Although not suitable for estimating absolute intakes, individuals can be classified into quintiles of intake for categories including total energy, protein, carbohydrate, sugars, fibre, and vitamin and mineral intake 73 . For a sub-sample of QTAB participants, these dietary intake assessments have been analysed and supplement gut microbiome measures obtained from stool samples 96 .

In August 2020, following the end of the first major lockdown in Brisbane, Queensland, we surveyed stress levels, depressive systems, and concerns with respect to the COVID-19 pandemic for both twins and their parents. We adapted questions from multiple COVID-19 survey sources, including the Swinburne University of Technology (Melbourne) and the NIH Office of Behavioral and Social Sciences Research (OBSSR). We also assessed posttraumatic stress (Child PTSD Symptom Scale for children 8-18 years 97 ), resilience (Brief Resilience Scale 78 ; Posttraumatic Growth Inventory X 98 ) and active and passive social media use (Multi-dimensional Scale of Facebook Use 75 ; modified for all types of social media 74 ). Resilience is implicated in mediating the development of mental illness following trauma 99 , while type of social media use (i.e., active "connection promoting" versus passive "nonconnection promoting") has been shown to mediate symptoms of anxiety and depressed mood in adolescents 74 , and active usage may be protective against negative consequences of social distancing. Further, a parent (usually the mother) reported on pandemic-related concerns and the impact on their work and home situation, personal finances, general wellbeing, and mental health -all factors that may mediate or moderate adolescent responses to the pandemic. Note that all session 1 assessments were completed prior to the onset of the pandemic. 31% of the twins returned for session 2 prior to COVID-19 restrictions and school closures).

While we collected several biological samples, most samples have not yet been processed due to budgetary constraints; however, collection statistics (i.e., yes/no) are provided for each biological sample (see Data Records).

Participants collected a saliva sample upon minutes of wakening on the day of their visit. A saliva kit, including a collection tube, icepack, and step-by-step written instructions, was mailed in advance of the study session. The procedure was explained to the parent on the phone when booking their session visit. Participants were instructed to generate some saliva in their mouth and then slowly drool it into the tube until the desired amount was collected (i.e., 2 ml). Then, they placed it on the icepacks for transport. Upon arrival at the QTAB session, saliva samples were immediately placed on fresh ice and transferred to a -80C freezer. 99% of participants provided a saliva sample at session 1 and 100% at session 2. The long-term aim was to process the samples at the Stress Physiology Investigative Team (SPIT) laboratory at Iowa State University. Estradiol and testosterone concentrations (pg/mL) would be indexed using the Salimetrics Salivary Testosterone Enzyme Immunoassay ELISA kit.

At sessions 1 and 2, a research assistant used scissors to cut 10-50 mg of hair from the posterior vertex region of the scalp (1 to 3 cm -enough to assess cortisol for the previous 3 months). Hair specimens have been stored at room temperature and have not yet been processed. Hair cortisol has recently emerged as a promising biomarker for long-term retrospective HPA activation 100 , with extremes of cortisol concentration levels (i.e., both lowest and highest) predicting depressive symptoms in adolescents 101 and longitudinal change in concentration levels predicting later social anxiety 102 . Puberty is a period of HPAaxis plasticity, and the effects of stress on cortisol regulation may depend on developmental stage/pubertal maturation. A longer-term aim of QTAB was to track how cortisol functioning impacts the brain regions processing emotion and whether HPA reactivity interacts with pubertal stage and subsequent associations with depression. Blood Genomic DNA was extracted from a blood sample provided at session 1 using standard procedures. DNA samples for available twin participants are currently being genotyped using the Infinium Global Screening Array-24 v3.0 BeadChip. Genotyping data is expected to be available in September 2022

The QTAB imaging dataset is publicly available through the OpenNeuro 103 data sharing platform (https://doi.org/10.18112/openneuro.ds004069.v1.0.3). The dataset is organised as per the Brain Imaging Data Structure (BIDS) specification v1.6.0 104 . BIDS specifies a hierarchical organisational format, with participant data stored under sub-folders denoting session (ses-01, ses-02) and then image modality: anat (structural), dwi (diffusion), func (functional), perf (asl), swi (susceptibility). An overview of the data record is available in Supplementary Table 1 . Before sharing, all personally identifiable information was removed from the dataset, including facial features from the 3D whole-brain images (T1-weighted, T2weighted, and FLAIR). We have made all data available, regardless of data quality (see Technical Validation).

Each scan is stored as a compressed NIfTI file (.nii.gz) with an accompanying sidecar JSON file (.json) describing scan acquisition parameters. In addition, for the diffusion scans, gradient orientation information is provided in *_dwi.bvec and *_dwi.bval, and for the asl scans, volume type information (i.e., m0scan, control, label) is provided in *_aslcontext.tsv. For the emotional conflict and Partly Cloudy movie watching tasks, event details (e.g. facial emotion responses for the emotional conflict task, condition timings for the Partly Cloudy task) are provided in *_task-emotionalconflict_events.tsv and *_task-partlycloudy_events.tsv respectively (variables and properties described in accompanying *_task-emotionalconflict_events.json and *_task-partlycloudy_events.json). The defacing masks used to deface the 3D whole-brain images are provided alongside participant anatomical data in *_inv-2_MP2RAGE_defacemask.nii.gz. Lastly, the derivatives folder at the top-level of the dataset directory contains quality checking metrics for imaging data (see Technical Validation).

All available QTAB phenotypic data is stored alongside the imaging data in the OpenNeuro repository. Key demographic data (i.e., age, sex, zygosity, multiple birth, birth order, handedness) is provided at the top-level of the dataset directory in participants.tsv (variables and properties described in participants.json), with phenotypic data stored under the toplevel folder phenotype. Multiple assessments and scales of a common domain are grouped as per the phenotypic domains detailed in Table 3 Table 2 , with notations identifying the key measurement domain. Before sharing these phenotypes, we removed all participant identifiable information, including occupation, postcode, and special categories of personable data (i.e., free-response text). Collection statistics (i.e., yes/no) for biological samples (blood, hair, saliva, urine, stool, baby tooth) are stored under the phenotype folder in *_biological_samples.tsv (variables and properties described in *_biological_samples.json).

Metagenomic data for a subset of QTAB participants used in a study by Yap et al. 96 is publicly available through a separate repository (https://doi.org/10.14264/e803a68).

Imaging Data 3D whole-brain (T1-weighted, T2-weighted, and FLAIR) and TSE scans were visually checked and rated by one author (LTS). Scans were rated using a three-category scale (pass, warn, fail) and scan quality ratings are available in the derivatives/visual_qc folder. There was a higher percentage of warn and fail ratings for TSE, T2-weighted and FLAIR scans than T1-weighted scans (Fig. 2a) , likely due to increased head movement associated with the acquisition of the TSE, T2-weighted, and FLAIR scans towards the end of the imaging session. The T1-weighted, T2-weighted, and FLAIR images were visually checked to ensure that facial features were successfully removed.

We used the MRtrix3 script dwipreproc 105 (which implements the FSL tool eddy 106,107 ) to calculate volume-to-volume motion estimates (Fig. 2b ) and the percentage of detected outlier slices (i.e., slices affected by severe signal dropout; Fig. 2c ); estimates available in the derivatives/mrtrix3 folder. The median across participants of average absolute motion was 0.77 mm and 0.76 mm at sessions 1 and 2, respectively. This finding is comparable with the same metric reported in a subset of the adult HCP 108 (median 0.83 mm). The median percentage of total outlier slices was 1.18% and 1.01% at sessions 1 and 2. The same metric was 1.89% and 0.39% in the developing/neonatal and adult HCP datasets 108 , respectively.

Image quality metrics (IQMs) were calculated for task and resting-state fMRI scans using MRIQC 109 . Framewise displacement (FD) head motion IQMs are displayed in Fig. 3d , and all IQMs are available in the derivatives/mriqc folder. The median across participants of average FD ranged from 0.16mm (Partly Cloudy) to 0.21mm (session 1 Rest). This finding is comparable with the same metric reported in a dataset of 3-12-year-old children 32 (Partly Cloudy task, median = 0.29 mm) and a dataset of 8-17-year-old children 110 (lexical processing tasks, median across all tasks and sessions = 0.17 mm). The ASL and SWI scans provided have not yet undergone quality checking or pre-processing. However, all scans were checked for apparent artefacts during the scanning session, with affected scans re-acquired (time permitting). Fig. 3 focuses on the design of QTAB and the non-imaging phenotypes. We chose an accelerated longitudinal design (Fig. 3a) so that in the first three years of the study, we worked to recruit every willing participant who was a twin between the ages of 9-14 years and who lived close to the study centre at the University of Queensland, Brisbane. We were successful in recruiting 422 twins across this age range. The long-term aim was to follow them prospectively and gather data on at least one further time point. For the 304 participants returning for session 2, the inter-session interval ranged from 1.1 to 2.5 years (M=1.7±0.3). The interval was partly influenced by pauses in data collection due to the pandemic, availability due to orthodontic treatment, and 5 years of funding. With an accelerated design 111 , as the participants age into new categories, they contribute data in every cell, i.e., the sample sizes get larger and larger as the adolescents age into these categories.

The advance of pubertal development can be seen in Fig. 3b . At session 1, more than half of the sample were classified as pre-or early-pubertal. In contrast, at session 2, approximately three-quarters of the sample had progressed to mid-or late-pubertal status, with a small number classified as post-pubertal. Developmental advances in processing speed (Fig. 3c) were also evident throughout the inter-session interval. The NIHTB-CB processing speed task records the number of correctly answered items in 85 seconds. On average, participants correctly answered an additional 8 items at session 2 compared to session 1. A trend for cross-sectional age-related increases in processing speed was also found, with older participants, in general, being able to answer more items within the given timeframe correctly. Within the 5 year funding period, the accelerated longitudinal design maximised the benefits of cross-sectional and longitudinal data collection (a), and developmental changes in pubertal status (b) and processing speed (c) occurred between sessions 1 and 2. The QTAB cohort was highly representative of community norms in cognitive ability (d), with the expected genetic relationship, i.e., identical twins were more alike than non-identical twins (e). In addition, the tools chosen to measure depression (f) and anxiety (g) were sensitive enough to capture individual differences and identify at-risk individuals. Variability in a depression risk factor, i.e., daytime sleepiness (h), was consistent with other community samples.

We calculated a measure of general cognitive ability using the NIHTB-CB Total Composite Score. Age-corrected standard scores for QTAB are shown in Fig. 3d . They are very close to the US national average scores (115 and 85 indicate performance 1 SD above and below the national average of 100), with a QTAB cohort mean score of 103.5 and SD of 17.2. In addition, 7.9% of the QTAB cohort had scores of 130 or greater (top 2% based on normative NIHTB-CB data), indicating that the cohort is oversampled for high cognitive performance, and 1.4% had scores of 70 or below (bottom 2% nationally of US scores), which suggests very low cognitive functioning and may be indicative of difficulties in school or general functioning. We also found that cognitive ability is more similar between QTAB identical co-twins (MZ twins) than non-identical co-twins (DZ twins), as expected for a trait that is influenced by genetic inheritance 112 .

Distributions of symptom levels for depression and anxiety in QTAB are shown in Fig.  3f and Fig. 3g , respectively. Depression symptoms were obtained from the SMFQ, for which a cut-off of 8 has been suggested as a screen for depression in children aged 8-16 years 47 . Approximately 17% of the QTAB cohort scored 8 or higher (Fig. 3f) . This is consistent with national surveys of child and adolescent health in Australia (e.g. 20% of adolescents aged 11-17 years reported experiencing high levels of psychological distress in the Young Minds Matter 2013-14 Survey 113 ). SCAS sum score (Generalized anxiety) was higher in girls than boys and younger compared to older participants, consistent with other studies 46, 114 . Of the 54 individuals aged 13-14 years, the mean score (M=13.3) is consistent with that reported for a community sample of 875 adolescents aged 13-14 years (M=13.5) 46 Fig. 3h shows the QTAB distribution for daytime sleepiness, as assessed using the PDSS. Excessive daytime sleepiness has been associated with poor stress management and higher levels of depressive mood in adolescents aged 14-19 years 115 . A cut-off above 17 has previously been used to identify excessive sleepiness in 618 children aged 10 to 12 years 116 , identifying 18% of the sample. With the same cut-off threshold, 17% of the QTAB cohort would be classified as having excessive daytime sleepiness.

For analyses of genetic (co)variance in the QTAB dataset, we suggest investigators familiarise themselves with the classic twin model 117 . Investigators not interested in genetic (co)variance should nevertheless consider the correlated nature of twin data (i.e., the nonindependence of participants) as it may violate statistical test assumptions 118 . Mixed models, which use random effects to model the correlation among twin pairs 119 , and structural equation modelling using the classic twin design 18 , are widely used approaches in controlling familial relatedness. Example code is provided online (https://github.com/QTAB-STUDY/twin-data-models).

The amplified background noise in the T1w MP2RAGE uniform image (*_UNIT1) can cause registration and segmentation issues. One technique for dealing with this problem is to input brain-extracted (i.e., skull-stripped) MP2RAGE uniform images to automated processing pipelines (e.g. FreeSurfer, fMRIPrep). We found that creating a brain mask based on the second inversion time image (*_inv-2_MP2RAGE) and applying this mask to the MP2RAGE uniform image resulted in successful brain extractions. Another approach is to remove background noise from the MP2RAGE uniform image using the inversion time images (https://github.com/JosePMarques/MP2RAGE-related-scripts). Multiple TSE scans were collected to average the TSE runs to improve image quality; see Shaw et al. 120 for implementation of TSE alignment and averaging using the QTAB dataset.

Diffusion and rs-fMRI scans were acquired using reversed-phase encoding directions to correct geometric distortions in the images. We recommend using the FSL tools topup and eddy to correct for distortions and movement in the diffusion scans and topup and applytop to correct the rs-fMRI scans. Similarly, we recommend using the reversed-phase encoding field maps and the FSL tools topup and FEAT to correct the task fMRI scans. Processing pipelines implementing these tools are available online (https://github.com/QTAB-STUDY/pre-processing). For the Partly Cloudy movie watching task, participants viewed a PAL format DVD of the movie. The event time codes provided (i.e., onset and duration for mental, pain, social, and control conditions; *_task-emotionalconflict_events.tsv) correspond to the timings provided by past studies 31,32 , converted to PAL timing 121 .

Two participants (twin pair sub-0109 and sub-0113) required an increased number of slices for their session 1 structural scans (T1w, T2w, FLAIR) for adequate brain coverage. Three participants (twin pair sub-0200 and sub-0419, sub-0373) have a reduced number of slices in their session 1 structural scans (T1w, T2w, FLAIR) due to operator error. Four participants have task fMRI but not corresponding field map scans (sub-1207, sub-7877, sub-8742, sub-9549). One participant has an incomplete T1w acquisition at session 2 (sub-0271, missing UNIT1, inv-1 images, but has UNIT1_denoised, inv-2 images). During session 1, the T1w MP2RAGE sequence name changed (MP2RAGE_wip900C_VE11C to MP2RAGE_wip900D_VE11C); however, there was no change to the sequence parameters. Data acquisition during session 2 was interrupted by a COVID-19 related lockdown of approximately 3 months, during which schools were closed. Approximately 31% of session 2 families were assessed before the lockdown, with the remainder assessed post-lockdown (see variable lockdown_ses02 in 10_covid-19.tsv).

Reversed scored questionnaire items have been re-coded in the OpenNeuro dataset. Parent responses for several scales (i.e., APQ, FAD, LTE, PaSS, PSES, PSS, BRS, and COVID-19 Experiences and Worries) are for the family (i.e., scores are the same for cotwins within a family). Australia is a multicultural country, as reflected in the QTAB ancestry measure. This measure reflects self-perceived group identification and may differ from a person's genetic ancestry, as obtained from genotyping. We also note that 29 participants have higher puberty scores at session 1 than session 2 (see the PDS_scores_ses01_greater_ses02 variable in 01_puberty-ses01.tsv). We believe this disparity reflects improved self-report puberty measurement with increased age. We suggest replacing session 01 PDS_scale_score, PDS_category_score, Gondal_score, and Adrenal_score variables with the corresponding session 02 variables. Due to copyright restrictions 122 , scale items from questionnaires are not included in the OpenNeuro dataset. However, we provide detailed instructions for linking item variables to the published questionnaire items (see 00_non_imaging_phenotypes_overview.pdf in the phenotype folder of the OpenNeuro dataset; also provided in Supplementary File 1). Where necessary, we contacted scale/questionnaire authors to obtain permission to use their measure and share the data collected.

DICOM format MRI data was converted to a BIDS compatible dataset using HeuDiConv 123 (v0.9.0; https://github.com/nipy/heudiconv). Facial features were removed from structural scans using BIDSonym 124 (v0.0.5; https://github.com/peerherholz/bidsonym) and the FSL 125 (v5.0.1; https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/) tools flirt and fslmaths. Code used in data organisation and defacing is available online (https://github.com/QTAB-STUDY/dicom-tobids). Head movement and outlier metrics for diffusion scans were calculated using the MRtrix3 105 

study) for generously sharing database information for recruitment. Recruitment was further facilitated through access to Twins Research Australia, a national resource supported by a Centre of Research Excellence Grant (ID: 1079102) from the NHMRC. Lastly, we thank the many researchers worldwide for providing access to their assessments. All publications of findings using data from the QTAB project should consider including the following credit

Strike -data curation, visual quality checking, pre-processing and analysis, writing -original draft

Hansell -designed the data acquisition protocol, data curation, writing -original draft

Chuang -designed the data acquisition protocol

Miller -data collection, study management

de Zubicaray -designed the data acquisition protocol, writing -review & editing, funding acquisition

Thompson -designed the data acquisition protocol, writing -review & editing, funding acquisition

McMahon -designed the data acquisition protocol, writing -review & editing, funding acquisition

Wright -designed the data acquisition protocol, writing -review & editing, funding acquisition

Adolescent brain development and depression: A case for the importance of connectivity of the anterior cingulate cortex

Convergent neurobiological predictors of emergent psychopathology during adolescence

The impact of stress on the structure of the adolescent brain: Implications for adolescent mental health

Lifetime prevalence of age-of-onset distributions of DMS-IF in the National Comorbidity Survey Replication

Social Determinants of Mental Health: Where We Are and Where We Need to Go

Adult mental health outcomes of adolescent depression: A systematic review

Systematic Review and Meta-Analysis: Adolescent Depression and Long-Term Psychosocial Outcomes

Development of the emotional brain

Emotional and cognitive changes during adolescence

Adolescent sleep restriction effects on cognition and mood

Puberty Initiates Cascading Relationships Between Neurodevelopmental, Social, and Internalizing Processes Across Adolescence

Sex differences in adolescent depression: do sex hormones determine vulnerability?

Girls' Pubertal Timing and Tempo and Mental Health: A Longitudinal Examination in an Ethnically Diverse Sample

Adolescent suicide as a global public health issue

Impacts of stress and sex hormones on dopamine neurotransmission in the adolescent brain

Social connectedness, mental health and the adolescent brain

The conception of the ABCD study: From substance use to a broad NIH collaboration

The utility of twins in developmental cognitive neuroscience research: How twins strengthen the ABCD research design

A self-report measure of pubertal status: Reliability, validity, and initial norms

Wechsler Intelligence Scale for Children

Event-and time-triggered remembering: the impact of attention deficit hyperactivity disorder on prospective memory performance in children

Delis Kaplan Executive Function System: Technical Manual

A cognitive neuroscience-based computerized battery for efficient measurement of individual differences: standardization and initial construct validation

Visual perception of line direction in patients with unilateral brain disease

Assessing Empathy across Childhood and Adolescence: Validation of the Empathy Questionnaire for Children and Adolescents (EmQue-CA)

A Comparison of Children's Ability to Read Children's and Adults' Mental States in an Adaptation of the Reading the Mind in the Eyes Task

Psychometric properties of the Spence Children's Anxiety Scale with young adolescents

Development of a short questionnaire for use in epidemiological studies of depression in children and adolescents

Development of a simple screening tool for common mental disorders in general practice

Validation and psychometric properties of the Somatic and Psychological HEalth REport (SPHERE) in a young Australian-based population sample using nonparametric item response theory

Measuring Impulsivity in Children: Adaptation and Validation of a Short Version of the UPPS-P Impulsive Behaviors Scale in Children and Investigation of its Links With ADHD

Strengths and difficulties questionnaire as a dimensional measure of child mental health

Initial Validation and Refinement of the Hierarchical Inventory of Personality for Children in the Australian Context

Red Flags" for autism screening: The Short Autism Spectrum Quotient and the Short Quantitative Checklist for Autism in toddlers in 1,000 cases and 3,000 controls

An examination of the response styles theory of depression in third-and seventh-grade children: a short-term longitudinal study

Children's Attributional Style Questionnaire Revised: Psychometric examination

Temperament profiles associated with internalizing and externalizing problems in preadolescence

Psychometric characteristics of the Multidimensional Scale of Perceived Social Support

Assessment of parenting practices in families of elementary school-age children

A psychometric study of the McMaster Family Assessment Device in psychiatric, medical ,and nonclinical samples

The trials of childhood: the development, reliability, and validity of the Daily Life Stressors Scale

Does bullying cause emotional problems? A prospective study of young teenagers

A comparison of the Gatehouse Bullying Scale and the peer relations questionnaire for students in secondary school

Adverse childhood events and psychosis in bipolar affective disorder

Neural correlates of prenatal stress in young women

The Parental Stress Scale -Initial Psychometric Evidence

Psychometric properties of the List of Threatening Experiences--LTE and its association with psychosocial factors and mental disorders according to different scoring methods

The pediatric daytime sleepiness scale (PDSS): sleep habits and school outcomes in middle-school children

Construction and validation of an instrument to evaluate sleep disturbances in childhood and adolescence

The Developmental Course of Sleep Disturbances Across Childhood Relates to Brain Morphology at Age 7: The Generation R Study

Manual for the ASEBA Preschool forms & Profiles. (Usniversity of Vermont

Reproducibility and comparative validity of a food frequency questionnaire for Australian children and adolescents

Active and Passive Social Media Use and Symptoms of Anxiety and Depressed Mood Among Icelandic Adolescents

Toward an Integrated and Differential Approach to the Relationships Between Loneliness, Different Types of Facebook Use, and Adolescents' Depressed Mood

A global measure of perceived stress

The brief resilience scale: assessing the ability to bounce back

Translating Cognitive Vulnerability Theory Into Improved Adolescent Depression Screening: A Receiver Operating Characteristic Approach

The Multidimensional Scale of Perceived Social Support

The McMaster Approach to Families: theory, assessment, treatment and research

Positive parenting predicts the development of adolescent brain structure: a longitudinal study

Observed Measures of Negative Parenting Predict Brain Development during Adolescence

Studying individual differences in human adolescent brain development

Parenting During Early Adolescence and Adolescent-Onset Major Depression: A 6-Year Prospective Longitudinal Study

Parenting style and mental disorders in a nationally representative sample of US adolescents

Frequent peer problems in Australian children and adolescents

Association of different forms of bullying victimisation with adolescents' psychological distress and reduced emotional wellbeing

Stress and the developing adolescent brain

A physical activity screening measure for use with adolescents in primary care

Adolescents living the 24/7 lifestyle: effects of caffeine and technology on sleep duration and daytime functioning

Adolescent sleep patterns and night-time technology use: results of the Australian Broadcasting Corporation's Big Sleep Survey

Genetic and environmental contributions to sleep-wake behavior in 12-year-old twins

Brisbane adolescent twin study: Outline of study methods and research projects

Long-term stability and heritability of telephone interview measures of alcohol consumption and dependence

Autism-related dietary preferences mediate autism-gut microbiome associations

The child PTSD Symptom Scale: a preliminary examination of its psychometric properties

The Posttraumatic Growth Inventory: A Revision Integrating Existential and Spiritual Change

Resilience as a translational endpoint in the treatment of PTSD

Hair Cortisol Concentration as a Biomarker of Sleep Quality and Related Disorders

Hair cortisol and depressive symptoms in youth: An investigation of curvilinear relationships

Associations of saliva cortisol and hair cortisol with generalized anxiety, social anxiety, and major depressive disorder: An epidemiological cohort study in adolescents and young adults

The OpenNeuro resource for sharing of neuroscience data

The brain imaging data structure, a format for organizing and describing outputs of neuroimaging experiments

MRtrix3: A fast, flexible and open software framework for medical image processing and visualisation

Incorporating outlier detection and replacement into a non-parametric framework for movement and distortion correction of diffusion MR images

An integrated approach to correction for offresonance effects and subject movement in diffusion MR imaging

Automated processing pipeline for neonatal diffusion MRI in the developing Human Connectome Project

Advancing the automatic prediction of image quality in MRI from unseen sites

A longitudinal neuroimaging dataset on multisensory lexical processing in school-aged children

Accelerated longitudinal designs: An overview of modelling, power, costs and handling missing data

The heritability of general cognitive ability increases linearly from childhood to young adulthood

Australian Institute of Health and Welfare. Australia's Youth: In Brief

A measure of anxiety symptoms among children

Proposal of cutoff points for pediatric daytime sleepiness scale to identify excessive daytime sleepiness

Assessment of vitamin D, exercise, and lipid profile associated with excessive daytime sleepiness in school children

Methodology for genetic studies of twins and families

Regression models for twin studies: a critical review

The use of linear mixed models to estimate variance components from data on twin pairs by maximum likelihood

Non-linear realignment improves hippocampus subfield segmentation reliability

Commentary: Copyright Restrictions Versus Open Access to Survey Instruments

nipy/heudiconv v0.9

BIDSonym: a BIDS App for the pseudoanonymization of neuroimaging datasets

We are forever grateful to the twins and their families for their willingness to participate in our studies. We thank Liza van Eijk, Victoria O'Callaghan, Islay Davies, Ethan Campi, Kimberley Huang, Eleanor Roga, Michael Day, Aiman Al-Najjar, Zoie Nott, Tom Shaw, Nicole Atcheson, and Sarah Daniel for data acquisition. We thank Naomi Wray for funding the collection of metabolic samples, including detailed dietary data, Ian Hickie and Kathleen Merikangas for funding and support of actigraphy data, and Sarah Medland and ENIGMA GWAS for funding genotyping. Special thanks to Anjali Henders, Leanne Wallace, Lorelle Nunn and the many laboratory assistants at the Human Studies Unit (part of the Program in Complex Trait Genomics based at the Institute of Molecular Bioscience, University of Queensland) for the processing and storage of biological samples. Thanks also to Julie Henry for helpful discussion of social cognition measures. The QTAB project was funded byThe authors declare no competing interests. PMT received a research grant from Biogen, Inc., for work unrelated to the current manuscript.