Enhancing Neuro Imaging Genetics through Meta-Analysis Consortium (ENIGMA) –
Obsessive Compulsive Disorder (OCD) Working Group
Secondary Proposal Form for Meta-Analyses
Please complete all fields and return this form by e-mail to:
Odile A. van den Heuvel
Email: oa.vandenheuvel@vumc.nl
1. Policy
Members of the ENIGMA Consortium include investigators from different centers around the world who are actively
engaged in neuroimaging research and who have contributed results from primary analyses of imaging, genetic data,
and/or algorithm development for the purpose of meta-analysis, replication, and/or algorithm testing in a collaborative
manner.
Although the data contributed to the ENIGMA consortium consist of group-level summaries and post-estimation
statistics rather than raw genotype and phenotype data, there is theoretically a minute risk of determining whether a
given individual participated in a study. While the re-identification of samples requires access to the raw genotype
data of the target individual and constitutes scientific misconduct, most groups have opted to appoint a gate-keeper
approach rather than allowing full public access to the results of their analyses or meta-analyses. Within the
ENIGMA-OCD working group any consortium member wishing to access the results of specific analyses or metaanalytic results will be asked to complete a short proposal describing why they wish to access the results files from
each group, and submit that for review.
All consortium members are encouraged to submit such proposals, to follow up on ideas which the group as a whole
cannot pursue, which involve novel analyses, or subsets of the available sites. The ENIGMA-OCD working group will
screen OCD-relevant proposals for scientific interest, and will help enlist members who might be interested in
collaborating. Proposals will be discussed on ENIGMA-OCD working group calls and emails to encourage the
broadest participation.
The proposal will then be posted on an ENIGMA forum page and an email will be sent to all consortium members
alerting them to the posting. ENIGMA members will have 14 days from the time of the posting to opt-out of the
analysis, ask for clarification, voice concerns or objections and/or give feedback to the proposal. No site data will be
shared without the consent of the PI of that site, who may opt to impose specific conditions or limitations on the use
of the data; also ENIGMA PIs and members are not required to take part in any proposed project, they can opt out.
If the author of the proposal agrees to the authorship and publication policies of the consortium the access request
will be granted to the results files for those groups who have not opted-out of the analysis and a member of the
Enigma-OCD working group or one of the Enigma-OCD working group PIs will be assigned as a project liaison. The
liaison will be responsible for providing the data and answering any queries relating to the project, and providing the
contributing site PIs with updates. The results files from each group will be housed at the VU university medical
center, Amsterdam, The Netherlands. If there is no possibility of determining if a particular individual participated in a
study (e.g. limited imaging or genetic markers are requested), results from these markers may be sent by the liaison
to other sites if available. If genome-wide results are requested from individual groups, the person submitting the
proposal may be granted an account on Laboratory of Neuro Imaging (LONI) servers or may visit LONI, if desired, to
make it easier to complete the analysis.
We request that the ‘ENIGMA-OCD Working Group’, and the liaison person will be listed as co-authors. The
ENIGMA-OCD Working Group on the byline, will reference the PIs of each study, in addition to contributors at their
site. In this way the authors contributing data to the consortium will be appropriately acknowledged on any
publication.
1
2. Requestor Information
Date of Submission:
Name:
Fabrizio Piras
Institution/Affiliation:
Email:
24th February, 2015
Neuropsychiatry Laboratory, IRCCS Santa Lucia Foundation, Rome, Italy
f.piras@hsantalucia.it
Have you signed and return the ENIGMA Memorandum of Understanding?
yes
If no, read, check and sign the Memorandum of Understanding and send to oa.vandenheuvel@vumc.nl
2. Results request proposal
Proposal Title: Meta/Mega analyses of DTI data in OCD
Co-author names and e-mail addresses (initial list):
Chiara Chiapponi
Tommaso Gili
Mariangela Iorio
Federica Piras
c.chiapponi@hsantalucia.it
t.gili@hsantalucia.it
m.iorio@hsantalucia.it
federica.piras@hsantalucia.it
Proposed Timeline for Completion of Study:
META-ANALYSIS
Months
0-2: preprocessing, QC and processing protocol definition
3-12: data collection
13-16: data analysis
17-18: manuscript/s preparation
MEGA-ANALYSIS
0-2: data sharing policy and protocol definition
3-15: data collection
16-22: data analysis
22-24: manuscript/s preparation
Please confirm that you have reviewed the ENIGMA website for potential areas of overlap. If you see a
project that may overlap, please list along with any plans for addressing this:
Note: this website is not yet ready for the ENIGMA-OCD working group. Overlap needs to be discussed during the
ENIGMA-OCD teleconferences, and in the meantime, we will prepare the listing of proposals on the website.
2
Please list any conflicts of interest:
No conflicts of interest to declare
Please describe the proposed analyses. Include hypothesis, specific results requested, a brief analysis plan
and methods, and references.
Neuroimaging studies have indicated several neurobiological changes underlying the psychological and behavioral deficits of
OCD. Evidence from functional and structural MRI has supported the notion that abnormalities in key gray matter (GM) regions,
such as the orbitofrontal cortex (OFC), anterior cingulate cortex (ACC), thalamus and striatum play an important role in its
pathophysiology 1. Comment Odile: let’s refer here to VBM meta- and mega-analyses, such as Radua et al. 2010 and de Wit et
al. 2014. Based on these studies the frontal part mainly concerns the dmPFC and the operculum/IFG However, the overall
picture is still rather heterogeneous and recent studies employing whole-brain analyses also indicate more distributed
neuroimaging alterations in patients with OCD, implicating other brain regions such as parietal cortex, dorsolateral prefrontal
cortex (DLPFC) and posterior temporo-parieto-occipital associative areas 2. Instead, the white matter (WM) tracts connecting
the cortical and subcortical regions involved in OCD pathophysiology are relatively unexplored 3. Consistently with current
frontal-striatal, fronto-parietal and fronto-limbic models of the disorder, recent studies 4 showed altered functional connectivity
among GM matter nodes of the cortico-striato-thalamo-cortical circuitry. Moreover, the existence of additional abnormalities in
specific WM tracts (e.g. internal capsule (IC), cingulate bundle (CB), and corpus callosum (CC)) and in different brain regions
(medial frontal and parietal WM), in the OCD population has been recently suggested 5. These WM abnormalities may be
familial 1 and responsive to serotonin reuptake inhibitor treatment (SSRI) 3 and vary according to the severity of different
symptom dimensions 6. Comment Odile: see also Radua et al. 2014 NPP (meta-analyses on WM (VBM) and DTI findings)
Here we propose a meta-analytic and a mega-analytic approach to further study structural brain architecture abnormalities in
the OCD-ENIGMA multisite sample. Comment Odile: please, separate analyses plan for meta- and mega-analyses
META-ANALYSIS
In order to quantitatively summarize the evidence of potential differences in diffusion tensor imaging (DTI) parameters between
OCD and HC, we plan to perform data aggregation by meta-analytical pooling, where no raw data are shared between sites.
We will take advantage of existent data gathering/analysis protocols within the ENIGMA consortium (e.g.
http://enigma.ini.usc.edu/protocols/dti-protocols/) for multi-site processing and extraction of DTI values. Briefly, once FA and
diffusivity maps are computed, the first step will consist of quality control in order to remove scans with abnormalities and
artifacts. Then, data will be processed using the tract-based spatial statistics (TBSS) analytic method 7 modified to project
individual FA values on the ENIGMA-DTI skeleton. Following the extraction of the skeletonized white matter and projection of
individual FA values, ENIGMA-tract-wise regions of interest (ROIs), derived from the Johns Hopkins University white matter
parcellation atlas 8 will be transferred to extract the mean FA across the full skeleton and average FA values for twelve major
white matter tracts, breaking down the corpus callosum into 3 regions, for a total of 15 regions of interest (ROIs). Subsequently,
summary statistics (i.e. Cohen’s d), expressing the magnitude of the potential difference in DTI parameters within ROIs, will be
produced. Comment Odile: which 15 ROIs? And why these ROIs? Is that based on consistency with the other DTI studies in
ENIGMA or based on the OCD-specific models? Is tractography not possible here as well?
MEGA-ANALYSIS
(comment Odile: wouldn’t you prefer at least also the same analyses as in the meta-analyses, in order to study
consistency? E.g. the same ROI approach, in parallel to the here proposed whole brain voxel-based approach?
The mega-analysis will include DTI data of OCD patients recruited worldwide for the OCD-ENIGMA project (n about 800) and
an equal number of healthy controls (HC). Analyses will follow a two-fold path:
i) Standard space group analysis (i.e. TBSS): all collected DTI images (independently from acquisition parameters such as
number of diffusion directions) will be collapsed to create a common anatomical framework in a standard space including all
main WM tracts (skeleton). The latter will be used within a voxel-based approach to highlight microstructural abnormalities in
OCD patients (compared to HC) and potential variations according to symptom severity, medication status and comorbidities
(whenever available).
ii) Subject space group analysis (i.e. fiber tractography), restricted to data acquired using a number of diffusion directions64.
This further analysis will be focused on DTI metrics (e.g. radial and transverse diffusivity, fractional anisotropy and kurtosis) of
WM tracts previously deemed as related to OCD pathophysiology (i.e. anterior and posterior limb of IC; CB; CC; inferior frontooccipital fasciculus; inferior longitudinal fasciculus; superior longitudinal fasciculus; uncinate fasciculus, 9). The abovementioned
tracts will be reconstructed at the single subject level by means of virtual dissection of streamlines passing through two ROIs
defined on the fractional anisotropy maps 10 and summary statistics (e.g. tract volume and length, mean tract FA, number of
streams) will be collected. Again, data will be analyzed in order to highlight differences between OCD and HC as well as the
effect of clinical variables on tracts microstructure.
As the mega-analysis is highly demanding in terms of computational and human resources, its accomplishment is conditional to
grant approval (e.g. International OCD Foundation, EC funded grants, Italian Ministry of Health calls).
References: 1.Menzies, L. et al. Am. J. Psychiatry 165, 1308–1315 (2008). 2.Piras, F. et al. Cortex (2013). 3.Yoo, S. Y. et al.
Acta Psychiatr. Scand. 116, 211–219 (2007). 4.Stern, E. R., Fitzgerald, K. D., Welsh, R. C., Abelson, J. L. & Taylor, S. F. PLoS
One 7, (2012). 5. Piras, F., Piras, F., Caltagirone, C. & Spalletta, G. Neuroscience and Biobehavioral Reviews (2013).6.Koch,
K. et al. J. Psychiatr. Res. 46, 264–270 (2012).7.Smith, S., Jenkinson,
M. & Johansen-Berg, H. Neuroimage 31, 1487–1505
3
(2006).8.Mori, S. et al. Neuroimage 40, 570–82 (2008).9.Piras, F., Piras, F., Caltagirone, C. & Spalletta, G. Neuroscience and
Biobehavioral Reviews 37, 2856–2877 (2013).10.Catani, M., Jones, D. K. & Ffytche, D. H. Ann. Neurol. 57, 8–16 (2005).
Resources: Please describe what resources you can commit to the project - junior researcher time,
troubleshooting, computational server time, helping writing and testing scripts, etc.
META-ANALYSIS
The project will be carried out by a multidisciplinary team including two senior research physicists, PhD level (Chiara
Chiapponi and Tommaso Gili, expert in DTI processing and analysis), a junior research psychologist, PhD level
(Mariangela Iorio, expert in MRI database management, images quality check and DTI processing) and two senior
research neuroscientists (Fabrizio Piras and Federica Piras, head of the neuroimaging section and of the
neuropsychology section of the Neuropsychiatry Laboratory, respectively). All personnel involved have a strong
expertise in scripts writing and testing. Comment Odile: we also have to keep options open for other people (DTI
experts) from the consortium to contribute, and we need people from the DTI working group of ENIGMA
MEGA-ANALYSIS
As above-mentioned, the mega-analysis project accomplishment is conditional to grant approval (e.g. International
OCD Foundation, EC funded grants, Italian Ministry of Health calls) in order to gather the computational and human
resources needed. In particular, we plan to employ two junior researchers (having a fair expertize in DTI data
managing, processing and analysis, as well as advanced programming skills) who will be solely dedicated to the
project. Funding will be requested also for server acquisition and setup. Specific to acquired resources will be:
STORAGE: To provide storage capacity we propose a new storage system. The existing storage system would be
turned over to provide a disk-based archival system, a cost and time effective alternative to tape.
CPU TIME: We have deemed that at least 12 x 1U compute nodes, each with 12 processing cores is needed.
MEMORY: We propose to setup high RAM machines, that means 64 gigabytes of memory for each node.
The team involved in the meta-analysis will supervise the overall project.
4