Agenda TC ENIGMA-MDD Meeting June 13 2013

Agenda TC ENIGMA-MDD Meeting June 13 2013

Minutes TC ENIGMA-MDD conference callMarch 27 2014

1. First results meta-analysis

Preliminary results of the meta-analysis on differences in subcortical volumes between MDD patients and controls based on 13 studies (NESDA, SHIP, SHIPtrend, CLING, CBASP, Dublin3T, Dublin1.5T, Sydney, QTIM, Rotterdam, Edinburgh, Oxford, MPIP) with total of 1677 patients and 7123 controls.Results of two studies (Andy Simmons/Geoff Hall) haven’t been received yet.

Summary results:

- Smaller (left/right/mean) hippocampus and left thalamus volumes in patients were found, although the effect sizes were not very large.

- Field strength (3T vs 1.5T) and mean age of onset per study did not moderate subcortical volume differences between patients and controls.

- Proportion of antidepressant was a significant moderator for hippocampus volume, showing larger reduction in these volumes in studies with a higher proportion of antidepressant users (might be an indication of severity?).

- Severity moderated differences in hippocampal volume between patients and controls: studies with lower (!) severity show larger volume reductions in MDD patients. However, as can be observed in the figure on slide 12 this effect is mainly driven by 2 studies. There is not a lot of variation in the mean severity across studies.

- Separate meta-analysis on the association between severity and subcortical volumes in each study (instead of taking mean severity of each study as a moderator) showed the opposite effect: higher severity is associated with larger reductions in left hippocampal volume and left lateral ventricle volume.

- Separate meta-analysis for the association between age of onset and subcortical volumes shows a positive relationship between AO and amygdala volume (i.e. the younger the patients at time of onset, the smaller the amygdala volume).

- Separate meta-analyses on first-episode patients vs controls and recurrent patients vs controls showed that (left) thalamus volume was reduced in first episode patients, whereas hippocampus and amygdala volume was reduced in recurrent patients.

Remaining analyses:

-Meta-analysis for first episode patients vs recurrent patients (scripts will be send)

-Severity MA’s separately in first episode patients and recurrent patients

-Examine mean differences , variances of each subcortical volume and % reduction of significant structures between patients and controls and look at heterogeneity measures

On basis of the paper of David Glahn in Biol Psychiatry 2012, we discussed the idea to also include the ventral diencephalon as a subcortical structure in the meta-analysis. Although this structure is very heterogeneous including that don't have much contrast in T1 images (e.g. mamillary bodies, tuber cinereum/infundibulum, some hypothalamic nuclei and sometimes fragments of the optic tracts), it was the highest ranked neuroimaging phenotype in the study of David Glahn, so could be worth looking at. Derrek tested the reliability of Freesurfer segmentation of this structure and it seems to be good (but it definitely looks like versions >5.0 have better reliability). Scripts for this structure will be send to the sites.

Timeline: given the interest of Nature Neuroscience in ENIGMA disease group papers and the fact that the SCZ and BD working groups just submitted their papers there, it was decided that we will submit our MDD paper on subcortical volumes to NN as well. However, to profit from the momentum of the SCZ and BD papers it would be best to submit the paper sooner rather than later. Therefore, only remaining analyses as stated above will be performed after which the results will be incorporated in an article as soon as possible.

2. Severity measures

For all future projects, our consortium would greatly benefit from a way to combine the different severity questionnaires that were used by the different sites. We have at least 5 uniquely used questionnaires: Hamilton, PHQ-9, BDI, MADRS, IDS (although converting either IDS or MADRS would be enough to obtain converted scores for all studies).

The feasibility of applying item response theory (IRT)methodsin order to develop a common metric for the different depression questionnaires, as was done by Matthias Rose in Wahl et al 2014 in Journal ofClinical Epidemiology, was discussed with Matthias (thanks for joining the call!). The metric that Matthias has already contains 11 questionnaires, but not the Hamilton, IDS and MADRS. By using data (scores on individual items) from the MARS study (Philipp Saemann), Dublin studies (Thomas Frodl) and Goetingen study (Roberto Goya/Oliver Gruber) it seems to be possible to map these remaining questionnaires to Matthias’ common metric.

3. Secondary proposals

Secondary proposals on childhood maltreatment (Thomas Frodl) and suicidal thoughts (Miguel Renteria) were discussed. In addition, Danai Dima joined the call and informed the sites about their normal structural variation project (includes only healthy controls, individual data (per subject) on subcortical volumes and age and sex are required). The project leaders of these secondary proposals will email the sites to ask whether they want to participate and to request info.

Additional ideas for new projects can be proposed by filling out the ‘ENIGMA-MDD_secondary_proposal’ form that can be found on our (password) protected website: (at the end of the page there is a link to the MDD members page). Password: enigma@mdd

The website will be updated shortly (to include all secondary projects that have been proposed to date).

An overview of potential interesting (clinical) phenotypes can be found in the online spreadsheet:

or alternatively go to gmail.com

- Login with loginname: enigmamdd password: metaanalysis
- Go to 'drive' (top row in browser, next to 'gmail')
- Go to 'shared with me' (left side of screen), where you can find the document 'ENIGMA-MDD overview data'

3. Cortical measures: update on QC protocols

Neda and Derrek have created QC protocols for the cortical measures of Freesurfer. A few versions have been sent around, but the version to be used is the one that includes cortical thickness and surface area measures. QC protocols for vertexwise volumes are currently also tested (there might be a Freesurfer version issue for this measure), but QC protocols for this potential additional measure will be send around at a later stage. Everyone can start the QC on cortical measures using the QC protocols!

Previous notes about the cortical QC protocols:

The idea is that there will be no editing during the segmentation process. Derrek tested the protocols on some datasets and the quality of cortical measures (without editing) is quite good (in general exclusion of around 5-10% of scans in each dataset). The test-retest reliability is also quite good, as long as Freesurfer versions 5.0 and later are used. Studies that used earlier versions (4.x versions) need to redo the segmentation using a Freesurfer 5.x version.

Some ideas have already been communicated: ‘suicidal thoughts in MDD’ (Miguel Renteria, QTIM Brisbane) and ‘childhood maltreatment’ (Thomas Frodl, Dublin). Miguel and Thomas will fill out the secondary proposal form for these projects.