Work in Magdeburg

Wenjing Li2012-11-23

Outline

Age and gender effects

Graph analysis on MDD patients

Cortical thickness – MRS correlation

Outline

Age and gender effects

Graph analysis on MDD patients

Cortical thickness – MRS correlation

Age and gender effects

Aim To investigate the effects of age and

gender on subcortical structures in healthy subjects

Why subcortical structures? Previous studies have reported

subcortical structures are involved in psychiatric disorders.

No studies had reported gender specificity of age effects on subcortical structures by then.

How was this work done?

2010.6 Generation of the idea: gender

differences of age trajectories on brain structures

2010.7 – 2010.10 Data selection Data processing and analysis for age

and gender effects on subcortical structures

2010.11 – 2011.12 First draft finished

How was this work done?

2011.1 – 2011.6 Modifying and polishing First submission to HBM in June

2011.8 Decision of the HBM: Reversible rejection

2011.9 – 2012.4 Reanalysis based on the reviews Re-construct the manuscript Resubmission to HBM in April.

How this work was done?

2012.5 Decision of the HBM: major revision

2012.5 – 2012.6 Revision and resubmission

2012.7 Decision of the HBM: accepted

First version of this paper

Data: 78 subjects, including 38 males and 38 females, age range: 19~70 years

First version of this paper

Reviews for the first version

Reviewer 1: Lots of tests – correction for multiple

comparisons Correction for TBV instead of ICV? Small sample size

Reviewer 2: Recommended to publish but with some

minor problem.

Revision

Correction for multiple comparisons? Combine the left and right subcortical

structures. Adjusted Bonforroni correction

Correction for TBV or ICV? We redid the analysis using TBV as

covariates. Small sample size

Rebuttal from the statistics and results.

Second version　 Gender Age Age*Gender

　 r p r p F p

Absolute Volumes

Caudate -.358 .002 -.278 .015 1.287 .260

Putamen -.438 <.001 -.430 <.001 .009 .925

Pallidum -.491 <.001 -.335 .003 .055 .815

Thalamus -.464 <.001 -.339 .003 1.786 .186

Hippocampus -.246 .032* -.181 .118 7.167 .009

Amygdala -.599 <.001 -.130 .262 .382 .539

Relative Volumes

Caudate .069 .551 -.175 .130 6.019 .017*

Putamen .098 .399 -.364 .001 2.464 .121

Pallidum -.037 .753 -.271 .018* 3.613 .061

Thalamus .089 .444 -.272 .018* .006 .937

Hippocampus .459 <.001 .052 .655 3.381 .070

Amygdala -.132 .257 .065 .578 .720 .399

　 Females Males

　 r square r p r square r p

Absolute Volumes

Caudate .218 -.467 .003 .029 -.171 .305

Putamen .309 -.556 <.001 .233 -.483 .002

Pallidum .248 -.498 .001 .132 -.363 .025*

Thalamus .115 -.340 .037* .276 -.525 <.001

Hippocampus .003 .057 .736 .205 -.453 .004

Amygdala .041 -.203 .223 .064 -.252 .127

Relative Volume

Caudate .169 -.411 .010* .021 .145 .384

Putamen .257 -.507 .001 .026 -.161 .333

Pallidum .243 -.493 .002 .001 -.029 .864

Thalamus .106 -.326 .046* .047 -.217 .191

Hippocampus .083 .289 .078 .020 -.140 .402

Amygdala .001 -.031 .855 .029 .169 .309

　 Females Males

　 Model r square p Model r square p

Absolute Volumes

Caudate Linear .218 .003 - - -

Putamen Linear .309 <.001 Quadratic .333 .001

Pallidum Linear .248 .001 Quadratic .347 .009

Thalamus Linear .115 .037* Quadratic .373 <.001

Hippocampus Quadratic .148 .060 Quadratic .321 <.001

Amygdala Quadratic .137 .075 Quadratic .235 .009

Relative Volume

Caudate Linear .169 .010* - - -

Putamen Linear .257 .001 - - -

Pallidum Linear .243 .002 - - -

Thalamus Linear .106 .046* - - -

Hippocampus Linear .083 .078 - - -

Amygdala - - - - - -

Second version

Outline

Age and gender effects

Graph analysis on MDD patients

Cortical thickness – MRS correlation

Distance penalty

Regions that are spatially close have higher correlation coefficients whereas more distinct regions correlate less strongly.

CIJ = CIJ.*log(distmat).^2; % CIJ is modified by ln.^2 of the distance

Global graph metrics

Local graph metrics

Index Regions

Rank Wilcoxon All regions rank

HC MDD P HC MDD

Betweenness centrality

SupraMarginal_R 49.82±20.26 69.52±15.96 0.0021 55 84

Supp_Motor_Area_R 45.73±21.39 26.62±15.77 0.0029 43 14

Degree Ant_Insula_L 19.64±15.58 39.10±21.54 0.0018 4 34

Strength Ant_Insula_L 14.23±14.96 35.67±22.79 <0.001 1 27

Outline

Age and gender effects

Graph analysis on MDD patients

Cortical thickness – MRS correlation

CTh – MRS correlation

Datasets: 46 healthy controls 20 MDD and 20 healthy controls

Processing: Freesurfer

Measurement: Cortical thickness MRS: glx (glu+gln), naa and ins in

pgACC, dACC and dlPFC

Analysis

Local correlation: Correlate cortical thickness in the MRS

region itself with its corresponding MRS value.

Global correlation: Correlate cortical thickness throughout

the whole brain with the MRS values.

Models (for 46 HC)

Raw model: CTh ~ MRS

Corrected model by ICV CTh ~ MRS + ICV

Corrected model by further adding age and gender CTh ~ MRS + ICV + age + gender

Models (for 20MDD&20HC)

Besides the models using for 46HC, we further add the “Group” to test for the group interaction.

Raw model: CTh ~ Group + MRS

Corrected model by ICV CTh ~ Group + MRS + ICV

Corrected model by further adding age and gender CTh ~ Group + MRS + ICV + age + gender

Other work

Correlation between graph metrics and MRS values.

Extract the mean fALFF values within the detected ROI, and then correlate it with MRS values.

FFT analysis

THANKS!