ENIGMA Major Depressive Disorder

Teleconference, July 2nd 2015

6am Los Angeles / 2pm UK / 3pm Netherlands / 11pm Australia

Dial-in number: +31 20 5356541 / Entry code: 299710

Chairs: Lianne Schmaal

Attendees:

●  Dick Veltman

●  Sinead Kelly

●  Sean Hatton

●  Neda Jahanshad

●  Miguel Renteria

●  Matthew Sacchet

●  Roberto Goya

●  Katherina Wittfeld

●  Henrik Walter

●  Ilya Veer

●  Baptiste Couvy-Duchesne

●  Geoff Hall

●  Theo Van Erp

●  Thomas Frodl

●  Benson Irungu

●  Martin Walter

●  Boris Gutman

●  Philipp Saemann

●  Udo Dannlowski

●  Bernhard Baune

Action Items

●  All sites should look at slide 5 of the ppt and complete all action points

●  Members should contact Lianne if they are interested in participating in a first draft working group to help with the cortical meta-analysis paper write-up (but only if you have enough time to dedicate in the next 1.5-2 months, i.e. tasks will be divided!)

●  All sites with DTI data should contact Lianne if they haven’t already received a link to the DTI form

●  Lianne will follow-up with sites that have Site variables in their Covariates.csv file with a new script for the Dx by Sex interaction analysis

●  Lianne will follow-up with sites what had trouble with BDI/HDRS

●  Lianne will send around a form to collect additional information for moderator analysis

Business

Lianne introduced two new sites joining:

●  Martin Walter (Magdeburg) joining with structural scans and DTI scans - this group has expertise in resting state MRI and the sample is comprised of medicated and unmedicated unipolar patients.

●  Matthew Sacchet/Ian Gotlib (Stanford University): This sample is comprised of 60 depressed, 60 control structural scans. Some of the patients are medicated. This site also has some DTI data, as well as genetic data.

●  Both sites will hopefully be able to join the cortical meta-analysis.

1. Cortical meta-analysis results

●  Lianne said that there are still a few sites she needs information from. She will be in touch with these sites. She invited all sites to have a look at slide 5 on the ppt and make sure to complete the action points.

●  Lianne disseminated the results before OHBM

●  Slide 9 has an overview of the samples

●  Slide 10 has an overview of the analysis - similar to subcortical project

●  The moderator analysis still needs to be done

●  Lianne will follow-up with sites that have site variables included - these sites may need to run part of the script again as there was a mistake in the original script wrt the Dx by Sex interaction analysis.

●  Significant cortical thickness reduction in MDD patients observed - not as strong as SZ or BP (see slide 11 for more detail)

●  Age, sex, site, avg thickness were used as covariates and corrected for multiple comparisons for both left and right hemispheres.

●  Slide 12 (main comparison) shows some overlap between left and right hemispheres. Colour coding represents effect size. There are more widespread effects observed in the cortex (in comparison to subcortical results (mainly hippocampus)).

●  Lianne pointed out that there were increases in cortical thickness in the cuneus and postcentral gyrus.

○  It was asked if these findings are still significant if average thickness isn’t regressed out

○  Lianne said this the result isn’t significant if average thickness isn’t corrected for.

●  Lianne reiterated that separate analyses for left and right hemispheres were conducted, as well as a combined analysis to look at the averages across both hemispheres (results presented for both analyses in slides 11 and 12).

●  Separate analyses for first episode MDD patients and recurrent MDD patients were discussed. When patients were separated based on first-episode or recurrent, only significant effects were observed for recurrent MDD patients. This suggests that recurrent patients are driving the results (slide 15 for more detail)

●  There were also more significant differences for late onset patients vs controls - this is in contrast to the subcortical findings of early onset MDD (slide 16)

○  This is a counterintuitive finding, however, it may be explained by the fact that some of the sites have younger samples (15 -25 years) - cortical thinning in adolescence is typically observed. However, adolescents with MDD may show less cortical thinning.

○  It was asked if the early onset group is much younger than the late onset group.

○  Lianne re-ran the analysis excluding the Australian samples (younger samples) - this didn’t change the results

○  Lianne also discussed the likelihood of early onset patients being more likely to have recurrent MDD - however, this was looked at and no relation was found

○  A suggestion was made to model age separately for the groups (early and late onset) - this might be more precise and may model age more accurately in each group. Lianne concurred that this was a good idea and should be looked at.

○  There were no findings for early onset MDD - perhaps as a result of the inclusion of adolescents. If anyone has any further ideas on how to address this, please contact Lianne

●  Slide 18 and 19 detail the effects of antidepressants on cortical thickness.

○  All effects of antidepressants disappear when recurrence was included in the model.

●  A few sites had trouble with the BDI and HDRS analyses but Lianne will follow-up with these sites

●  There were no significant associations between number of MDD episodes and cortical thickness (see slide 20)

○  The distribution and lack of variation may explain this

○  It was suggested to look at where this non-linearity comes in.

○  Some of the samples may be skewed as they may have a lot of first episode patients.

○  It was suggested to just look at two MDD episodes and up (i.e. exclude first - episodes and just look at this in recurrent MDD). Lianne will have a look at this. This may normalise the distribution of the data

●  There were also no significant results for symptom severity - this is difficult to measure and there were also no significant results for symptom severity and subcortical measures.

○  The classification of patients (e.g. hospital-based, outpatient, general population etc) was suggested.

○  Could look at this as a moderator variable.

○  Lianne will send around a form to collect this data.

●  Surface area: there were no significant differences observed for surface area, not even borderline. This is similar to the bipolar results that Derrek presented at the ENIGMA meeting.

●  Slide 22 outlines the things that are still left to do.

●  Slide 23 outlines a few remaining questions.

○  Lianne suggested presenting the surface area results in the supplementary material. Everyone agreed.

○  There are some hemisphere effects - so reporting left and right separately was agreed.

○  Results are corrected for average thickness and total surface area. It was concurred that the paper should document the uncorrected effects as well and report this in the supplement (but keep the main results corrected).

○  Lianne asked if adolescent and adult MDD should be separated for the analysis. It was highlighted that the diagnosis effects are very interesting and could help formulate hypotheses for longitudinal studies.

○  The suggestion of running the analysis using a cut-off (e.g. 25 years) was discussed - each site could split up their sample based on below and above 25 years and run separate analyses for each. Some sites may not have enough subjects in each group but Lianne will follow-up with this suggestion.

●  Lianne asked if slice thickness should be included as well. Theo suggested looking at voxel size overall. Cortical thickness measures are influenced by spatial resolution so this should be looked at.

●  Lianne said that it might take a little longer to have results now that there are extra analyses, however, she will try to create the scripts next week.

○  All sites said they should be able run the scripts in a week.

○  The group will aim for mid-August for the first draft of the paper. Lianne may form a first-draft working group to help write the paper. Please send Lianne an email if you are interested (but only if you have enough time to dedicate in the next 1.5-2 months, i.e. tasks will be divided!).

2. Update on ongoing projects: Childhood maltreatment (Thomas Frodl) and suicidal ideation (Miguel Renteria)

●  Suicidal ideation: Miguel mentioned that there are now 2000 controls and 1000 cases included in this study, with findings in the amygdala. This will comprise the first paper. Miguel also invited everyone to participate in a cortical thickness analysis for a second paper.

●  Thomas stated that there are now six sites participating on the childhood maltreatment study and he is in contact with other sites. A preliminary mega-analysis was conducted (see slide 28). There were also strong effects for the interaction of childhood adversity and diagnosis (slide 29). However, there was no diagnosis effect in the hippocampus - Lianne said this is interesting as it contrasts with the MDD meta-analysis results. Thomas said that it would also be interesting to do a meta-analysis of the six samples (as well as the mega-analysis) to see if the results are the same. Sean asked if there were findings for different forms of childhood adversity - Thomas said that they didn’t find anything for this. New sites can still join this analysis at the moment (for the next 3-4 weeks). Sites must have CTQ data.

3. New projects: Shape analysis (Boris Gutman / Sean Hatton) and hippocampal subfields (Philipp Saemann)

●  Lianne introduced the new projects and informed the sites that there will be some additional processing steps involved, as well as QC. She mentioned that Boris and Philipp will be in touch with the sites about these projects.

●  Philipp outlined the hippocampal subfields project – in his local sample the CA1 region of the hippocampus is affected in first episode but the result did not reach significance. Philipp is hopeful that there will be significant findings in the larger sample. All sites that participated in the subcortical study, as well as new sites, are invited to join. The FreeSurfer subfield tool of v6.0 should be downloaded. There is a script to extract the volumes and as well as a QC script. This can be conducted on data based on the analysis of recon all from FS v5.3. Philipp assured the sites that they will not need any new clinical phenotypes. The proposal will be sent around in the next few days. 1.5T data is also suitable for this study.

●  Boris outlined the proposed shape analysis and informed sites that if they already have FS parcellations, they can download the shape analysis scripts (Boris will provide the links to the scripts). The aim is to measure subregional volumetric changes of the subcortical structures. It takes about an hour to run per subject. The proposal has been sent around to the sites and an invitation to join the shape analysis project will be disseminated to all sites.

4. Other projects: DTI (Sinead Kelly) and imaging genetics (Lianne Schmaal)

●  Lianne will think about interesting ideas for imaging genetics. If anyone is interested in brainstorming, please get in touch with Lianne.

●  For sites who have DTI data, Lianne has send a link to a form with some questions regarding type of data, timeline, feasibility etc. Please let Lianne know if you haven’t received this link. The DTI project will be based on the ENIGMA-DTI protocols for the harmonization of multi-site DTI data. Sinead and Laura will follow up with the sites about this.

●  There is also another project proposed to look at first-degree relatives of MDD patients (Henrik Walter/Ilya Veer). More about this soon.

5. Conference plans:

·  Please let Lianne know your plans for attending conferences, so perhaps an ENIGMA MDD meeting can be set up during one of the upcoming conferences. The ECNP conference will be held in Amsterdam end of August. If you are planning on attending, let Lianne know.