Data of the REST-meta-MDD Project from DIRECT Consortium

(Note: Part of the content of this post was adapted from the original DIRECT Psychoradiology paper (https://academic.oup.com/psyrad/article/2/1/32/6604754) and REST-meta-MDD PNAS paper (http://www.pnas.org/cgi/doi/10.1073/pnas.1900390116) under CC BY-NC-ND license.)Major Depressive Disorder (MDD) is the second leading cause of health burden worldwide (1). Unfortunately, objective biomarkers to assist in diagnosis are still lacking, and current first-line treatments are only modestly effective (2, 3), reflecting our incomplete understanding of the pathophysiology of MDD. Characterizing the neurobiological basis of MDD promises to support developing more effective diagnostic approaches and treatments.An increasingly used approach to reveal neurobiological substrates of clinical conditions is termed resting-state functional magnetic resonance imaging (R-fMRI) (4). Despite intensive efforts to characterize the pathophysiology of MDD with R-fMRI, clinical imaging markers of diagnosis and predictors of treatment outcomes have yet to be identified. Previous reports have been inconsistent, sometimes contradictory, impeding the endeavor to translate them into clinical practice (5). One reason for inconsistent results is low statistical power from small sample size studies (6). Low-powered studies are more prone to produce false positive results, reducing the reproducibility of findings in a given field (7, 8). Of note, one recent study demonstrate that sample size of thousands of subjects may be needed to identify reproducible brain-wide association findings (9), calling for larger datasets to boost effect size. Another reason could be the high analytic flexibility (10). Recently, Botvinik-Nezer and colleagues (11) demonstrated the divergence in results when independent research teams applied different workflows to process an identical fMRI dataset, highlighting the effects of “researcher degrees of freedom” (i.e., heterogeneity in (pre-)processing methods) in producing disparate fMRI findings.To address these critical issues, we initiated the Depression Imaging REsearch ConsorTium (DIRECT) in 2017. Through a series of meetings, a group of 17 participating hospitals in China agreed to establish the first project of the DIRECT consortium, the REST-meta-MDD Project, and share 25 study cohorts, including R-fMRI data from 1300 MDD patients and 1128 normal controls. Based on prior work, a standardized preprocessing pipeline adapted from Data Processing Assistant for Resting-State fMRI (DPARSF) (12, 13) was implemented at each local participating site to minimize heterogeneity in preprocessing methods. R-fMRI metrics can be vulnerable to physiological confounds such as head motion (14, 15). Based on our previous work examination of head motion impact on R-fMRI FC connectomes (16) and other recent benchmarking studies (15, 17), DPARSF implements a regression model (Friston-24 model) on the participant-level and group-level correction for mean frame displacements (FD) as the default setting.In the REST-meta-MDD Project of the DIRECT consortium, 25 research groups from 17 hospitals in China agreed to share final R-fMRI indices from patients with MDD and matched normal controls (see Supplementary Table; henceforth “site” refers to each cohort for convenience) from studies approved by local Institutional Review Boards. The consortium contributed 2428 previously collected datasets (1300 MDDs and 1128 NCs). On average, each site contributed 52.0±52.4 patients with MDD (range 13-282) and 45.1±46.9 NCs (range 6-251). Most MDD patients were female (826 vs. 474 males), as expected. The 562 patients with first episode MDD included 318 first episode drug-naïve (FEDN) MDD and 160 scanned while receiving antidepressants (medication status unavailable for 84). Of 282 with recurrent MDD, 121 were scanned while receiving antidepressants and 76 were not being treated with medication (medication status unavailable for 85). Episodicity (first or recurrent) and medication status were unavailable for 456 patients.To improve transparency and reproducibility, our analysis code has been openly shared at https://github.com/Chaogan-Yan/PaperScripts/tree/master/Yan_2019_PNAS. In addition, we would like to share the R-fMRI indices of the 1300 MDD patients and 1128 NCs through the R-fMRI Maps Project (http://rfmri.org/REST-meta-MDD). These data derivatives will allow replication, secondary analyses and discovery efforts while protecting participant privacy and confidentiality.According to the agreement of the REST-meta-MDD consortium, there would be 2 phases for sharing the brain imaging data and phenotypic data of the 1300 MDD patients and 1128 NCs. 1) Phase 1: coordinated sharing, before January 1, 2020. To reduce conflict of the researchers, the consortium will review and coordinate the proposals submitted by interested researchers. The interested researchers first send a letter of intent to [e-mail]. Then the consortium will send all the approved proposals to the applicant. The applicant should submit a new innovative proposal while avoiding conflict with approved proposals. This proposal would be reviewed and approved by the consortium if no conflict. Once approved, this proposal would enter the pool of approved proposals and prevent future conflict. 2) Phase 2: unrestricted sharing, after January 1, 2020. The researchers can perform any analyses of interest while not violating ethics.The REST-meta-MDD data entered unrestricted sharing phase since January 1, 2020. The researchers can perform any analyses of interest while not violating ethics. Please visit Psychological Science Data Bank to download the data, and then sign the Data Use Agreement and email the scanned signed copy to [e-mail] to get unzip password and phenotypic information.   ACKNOWLEDGEMENTSThis work was supported by the National Key R&D Program of China (2017YFC1309902), the National Natural Science Foundation of China (81671774, 81630031, 81471740 and 81371488), the 13th Five-year Informatization Plan (XXH13505) of Chinese Academy of Sciences, Beijing Municipal Science & Technology Commission (Z161100000216152, Z171100000117016, Z161100002616023 and Z171100000117012), Department of Science and Technology, Zhejiang Province (2015C03037) and the National Basic Research (973) Program (2015CB351702).Given the huge size of the files, please use FTP to download the data. You may need to resume from break-point sometimes. You need to sign the Data Use Agreement and email the scanned signed copy to [e-mail] to get unzip password and phenotypic information.(注：部分内容依据CC BY-NC-ND许可协议引自以下两篇文章：《The DIRECT consortium and the REST-meta-MDD project: towards neuroimaging biomarkers of major depressive disorder》 (https://academic.oup.com/psyrad/article/2/1/32/6604754) 与《Reduced default mode network functional connectivity in patients with recurrent major depressive disorder》 (http://www.pnas.org/cgi/doi/10.1073/pnas.1900390116) )重性抑郁障碍 (Major Depressive Disorder, MDD) 是造成全球健康负担的第二大原因(1)。然而，目前仍缺乏可辅助诊断的客观生物标志物，现有一线治疗的疗效也较为有限(2, 3)，这在一定程度上反映了我们对 MDD 病理生理机制认识的不足。深入阐明其神经生物学基础，有助于推动开发更有效的诊断方法和治疗手段。静息态功能磁共振成像 (resting-state functional magnetic resonance imaging, R-fMRI) 是近年来广泛应用于揭示临床疾病神经生物学机制的重要工具(4)。尽管已有大量研究尝试通过R-fMRI来解释MDD的病理生理机制，但迄今为止尚未发现能够稳定支持临床诊断或预测治疗效果的影像标志物。此外，既往研究结果之间往往不一致，甚至相互矛盾，严重阻碍了相关发现向临床实践的转化(5)。导致研究结果不一致的原因主要有两点，其中一个原因是样本量较小导致统计效力不足(6)。小样本研究更容易产生假阳性结果，从而降低研究发现的可重复性(7, 8)。近期有研究表明，要想获得可重复的全脑关联性发现，样本量可能需要达到数千例(9)，这凸显了构建大规模数据集的重要性。另一个原因是分析过程的灵活性较高(10)。Botvinik-Nezer等人(11)发现，不同研究团队在处理相同的fMRI 数据集时，由于采用了不同的分析流程，会得到不同的结果，这提示了“研究者自由度” (即预处理与分析方法的差异) 对研究结果的影响。为了应对上述问题，我们于2017年发起并成立了抑郁症脑影像大数据联盟(Depression Imaging REsearch ConsorTium，DIRECT)。经过多次会议讨论，来自中国17家医院的研究团队达成共识，共同建立了DIRECT联盟的首个项目——REST-meta-MDD数据库，共享了25个研究队列的数据，其中包括1300名MDD患者和1128名健康对照者的R-fMRI数据。基于既往研究经验，各站点统一采用基于DPARSF(12, 13)的标准化预处理流程，尽量减少预处理方法差异带来的影响。此外，由于R-fMRI指标容易受到头动等生理噪声的干扰(14, 15)。结合以往关于头动对功能连接影响的研究(16)及其他基准研究发现(15, 17)，DPARSF默认采用Friston-24回归模型进行个体水平进行头动校正，并在组水平对平均帧间位移(Framewise Displacement, FD)进行校正。在REST-meta-MDD项目中，来自17家医院的25个研究团队在各自机构伦理委员会的批准下，同意共享MDD患者及其健康对照者的R-fMRI指标数据(见补充表；下文将以“站点”指代各队列)。联盟共汇集了2428例既往采集的数据(1300名MDD患者和1128名健康对照)。各站点平均贡献MDD患者52.0±52.4名(13-282名)，健康对照45.1±46.9名(6–251名)。MDD患者中女性占多数(826名女性，474名男性)。在562名首次发作的患者中，有318名未服药患者(first episode drug-naïve, FEDN)、160名服用抗抑郁药的患者、84名患者服药情况不详。在282名复发性MDD患者中，有121名服用抗抑郁药、76名未接受药物治疗、85名患者服药情况不详。此外，有456名患者的发病类型(首次或复发)和药物治疗情况不详。为了提升研究透明度与可重复性，我们已将分析代码公开发布于：https://github.com/Chaogan-Yan/PaperScripts/tree/master/Yan_2019_PNAS。此外，1300名MDD患者和1128名健康对照者的R-fMRI指标数据将通过R-fMRI Maps项目(http://rfmri.org/REST-meta-MDD)对外共享。在保护参与者隐私的前提下，这些数据可供研究者开展重复验证、二次分析及探索性研究。根据REST-meta-MDD项目的联盟协议，数据共享分为两个阶段：(1)第一阶段(协调共享，2020年1月1日前)：为避免研究冲突，联盟对申请者提交的研究方案进行审核与协调。申请者须先向[e-mail]发送意向书，联盟将反馈给申请者目前已批准的课题列表。申请者在确认无冲突的前提下，提交具有创新性的研究方案，经审核通过后研究方案将纳入课题列表。(2)第二阶段(开放共享，2020年1月1日后)：研究者可在不违反伦理规范的前提下自由开展相关分析。REST-meta-MDD 数据已于2020年1月1日起进入开放共享阶段。研究者可访问Psychological Science Data Bank下载数据，签署数据使用协议后将签署扫描件发送至 [e-mail]，即可获取解压密码及表型信息。由于数据文件较大，建议使用 FTP 方式下载，并视情况开启断点续传功能。 致谢本研究得到以下项目资助：国家重点研发计划(2017YFC1309902)、国家自然科学基金(81671774、81630031、81471740、81371488)、中国科学院人才计划和“十三五”信息化专项(XXH13505)、北京市科学技术委员会(Z161100000216152、Z171100000117016、Z161100002616023、Z171100000117012)、浙江省科学技术厅(2015C03037)以及国家重点基础研究发展计划(973计划，2015CB351702)。鉴于文件体积很大，请使用 FTP 下载数据。下载过程中有时可能需要使用断点续传。您需要签署《数据使用协议》（Data Use Agreement），并将签字后的扫描件通过电子邮件发送至 [e-mail]，以获取解压密码和表型信息。 REFERENCES1.        A. J. Ferrari et al., Burden of Depressive Disorders by Country, Sex, Age, and Year: Findings from the Global Burden of Disease Study 2010. PLOS Medicine 10, e1001547 (2013).2.        L. M. Williams et al., International Study to Predict Optimized Treatment for Depression (iSPOT-D), a randomized clinical trial: rationale and protocol. Trials 12, 4 (2011).3.        S. J. Borowsky et al., Who is at risk of nondetection of mental health problems in primary care? J Gen Intern Med 15, 381-388 (2000).4.        B. B. Biswal, Resting state fMRI: a personal history. Neuroimage 62, 938-944 (2012).5.        C. G. Yan et al., Reduced default mode network functional connectivity in patients with recurrent major depressive disorder. Proc Natl Acad Sci U S A 116, 9078-9083 (2019).6.        K. S. Button et al., Power failure: why small sample size undermines the reliability of neuroscience. Nat Rev Neurosci 14, 365-376 (2013).7.        J. P. A. Ioannidis, Why Most Published Research Findings Are False. PLOS Medicine 2, e124 (2005).8.        R. A. Poldrack et al., Scanning the horizon: towards transparent and reproducible neuroimaging research. Nat Rev Neurosci 10.1038/nrn.2016.167 (2017).9.        S. Marek et al., Reproducible brain-wide association studies require thousands of individuals. Nature 603, 654-660 (2022).10.      J. Carp, On the Plurality of (Methodological) Worlds: Estimating the Analytic Flexibility of fMRI Experiments. Frontiers in Neuroscience 6, 149 (2012).11.      R. Botvinik-Nezer et al., Variability in the analysis of a single neuroimaging dataset by many teams. Nature 10.1038/s41586-020-2314-9 (2020).12.      C.-G. Yan, X.-D. Wang, X.-N. Zuo, Y.-F. Zang, DPABI: Data Processing & Analysis for (Resting-State) Brain Imaging. Neuroinformatics 14, 339-351 (2016).13.      C.-G. Yan, Y.-F. Zang, DPARSF: A MATLAB Toolbox for "Pipeline" Data Analysis of Resting-State fMRI. Frontiers in systems neuroscience 4, 13 (2010).14.      R. Ciric et al., Mitigating head motion artifact in functional connectivity MRI. Nature protocols 13, 2801-2826 (2018).15.      R. Ciric et al., Benchmarking of participant-level confound regression strategies for the control of motion artifact in studies of functional connectivity. NeuroImage 154, 174-187 (2017).16.      C.-G. Yan et al., A comprehensive assessment of regional variation in the impact of head micromovements on functional connectomics. NeuroImage 76, 183-201 (2013).17.      L. Parkes, B. Fulcher, M. Yücel, A. Fornito, An evaluation of the efficacy, reliability, and sensitivity of motion correction strategies for resting-state functional MRI. NeuroImage 171, 415-436 (2018).

搜集汇总

背景与挑战

背景概述

该数据集是DIRECT联盟REST-meta-MDD项目的组成部分，旨在通过大样本静息态功能磁共振成像（R-fMRI）数据研究重度抑郁症（MDD）的神经生物学基础。数据集包含来自中国17家医院的25个队列的2428个样本（1300名MDD患者和1128名正常对照），采用标准化预处理流程以减少方法异质性，并自2020年起开放无限制共享，支持研究人员进行复制和二次分析。