Aphasia Neuroplasticity Review

Griffis et al. (2017b)

Reference

AuthorsGriffis JC, Nenert R, Allendorfer JB, Vannest J, Holland S, Dietz A, Szaflarski JP
TitleThe canonical semantic network supports residual language function in chronic post-stroke aphasia
ReferenceHum Brain Mapp 2017b; 38: 1636-1658
PMID27981674
DOI10.1002/hbm.23476

Participants

LanguageUS English
Inclusion criteria
Number of individuals with aphasia43
Number of control participants43
Were any of the participants included in any previous studies?Yes (data were collected as part of "several separate studies")
Is age reported for patients and controls, and matched?Yes (mean 53 ± 15 years, range 23-90 years)
Is sex reported for patients and controls, and matched?Yes (males: 25; females: 18)
Is handedness reported for patients and controls, and matched?Yes (right: 41; left: 2)
Is time post stroke onset reported and appropriate to the study design?Yes (range 1-14 years)
To what extent is the nature of aphasia characterized?Not at all
Language evaluationBNT, semantic fluency, phonemic fluency
Aphasia severityNot stated
Aphasia typeNot stated
First stroke only?Yes
Stroke typeNot stated
To what extent is the lesion distribution characterized?Individual lesions
Lesion extentMean 105.2 ± 76.3 cc
Lesion locationL
Participants notes

Imaging

ModalityfMRI
Is the study cross-sectional or longitudinal?Cross-sectional
If longitudinal, at what time point(s) were imaging data acquired?
If longitudinal, was there any intervention between the time points?
Is the scanner described?No (Siemens Allegra 3 Tesla or Philips 3 Tesla; model not stated)
Is the timing of stimulus presentation and image acquisition clearly described and appropriate?Yes
Design typeBlock
Total images acquired165
Are the imaging acquisition parameters, including coverage, adequately described and appropriate?Yes (whole brain)
Is preprocessing and intrasubject coregistration adequately described and appropriate?Yes
Is first level model fitting adequately described and appropriate?Yes
Is intersubject normalization adequately described and appropriate?Yes
Imaging notes

Conditions

Are the conditions clearly described?Yes
ConditionResponse typeRepetitionsAll groups could do?All individuals could do?
semantic decisionButton press5NoNo
tone decisionButton press6UnknownUnknown
Conditions notesGroup performance below chance; several patients at 0 which is difficult to understand in a 2AFC task

Contrasts

Are the contrasts clearly described?Yes

Contrast 1: semantic decision vs tone decision

Language conditionSemantic decision
Control conditionTone decision
Are the conditions matched for visual demands?Yes
Are the conditions matched for auditory demands?Yes
Are the conditions matched for motor demands?Yes
Are the conditions matched for cognitive/executive demands?Yes
Is accuracy matched between the language and control tasks for all relevant groups?Unknown, not reported
Is reaction time matched between the language and control tasks for all relevant groups?Unknown, not reported
Behavioral data notesTone decision accuracy not reported
Are control data reported in this paper or another that is referenced?Yes
Does the contrast selectively activate plausible relevant language regions in the control group?Yes
Are activations lateralized in the control data?Yes
Control activation notesTemporal activation is mid MTG and AG rather than pSTS
Contrast notes

Analyses

Are the analyses clearly described?Yes

Voxelwise analysis 1

First level contrastSemantic decision vs tone decision
Analysis classCross-sectional correlation with language or other measure
Group(s)Aphasia
CovariateSemantic decision accuracy
Is the second level contrast valid in terms of the group(s), time point(s), and measures involved?Yes
Is accuracy matched across the second level contrast?Accuracy is covariate
Is reaction time matched across the second level contrast?Unknown, not reported
Behavioral data notes
Type of analysisVoxelwise
Search volumeWhole brain
Correction for multiple comparisonsClusterwise correction based on cluster_threshold_beta
SoftwareSPM12/in-house
Voxelwise p.01
Cluster extent126 voxels (size not stated)
Statistical detailsLesion volume covariate
Findings↑ L dorsolateral prefrontal cortex
↑ L angular gyrus
↑ L precuneus
↑ L mid temporal
↑ L anterior temporal
↑ L posterior cingulate
↑ L cerebellum
↑ L brainstem
↑ L hippocampus/MTL
↑ R IFG pars orbitalis
↑ R angular gyrus
↑ R precuneus
↑ R anterior temporal
↑ R occipital
↑ R brainstem
↑ R hippocampus/MTL
↓ L somato-motor
Findings notesBased on figure and table; larger activations are compelling; smaller activations are not due to lenient correction approach

Voxelwise analysis 2

First level contrastSemantic decision vs tone decision
Analysis classCross-sectional correlation with language or other measure
Group(s)Aphasia
CovariateAverage of semantic and phonemic fluency
Is the second level contrast valid in terms of the group(s), time point(s), and measures involved?Yes
Is accuracy matched across the second level contrast?Unknown, not reported
Is reaction time matched across the second level contrast?Unknown, not reported
Behavioral data notes
Type of analysisVoxelwise
Search volumeWhole brain
Correction for multiple comparisonsClusterwise correction based on cluster_threshold_beta
SoftwareSPM12/in-house
Voxelwise p.01
Cluster extent126 voxels (size not stated)
Statistical detailsLesion volume covariate
Findings↑ L IFG
↑ L dorsolateral prefrontal cortex
↑ L SMA/medial prefrontal
↑ L angular gyrus
↑ L precuneus
↑ L posterior STS
↑ L mid temporal
↑ L anterior temporal
↑ L posterior cingulate
↑ L brainstem
↑ L hippocampus/MTL
↑ R SMA/medial prefrontal
↑ R precuneus
↑ R anterior temporal
↑ R occipital
↑ R posterior cingulate
↑ R hippocampus/MTL
↓ R posterior STS
Findings notesBased on figure and table; larger activations are compelling; smaller activations are not due to lenient correction approach

Voxelwise analysis 3

First level contrastSemantic decision vs tone decision
Analysis classCross-sectional correlation with language or other measure
Group(s)Aphasia
CovariateBNT
Is the second level contrast valid in terms of the group(s), time point(s), and measures involved?Yes
Is accuracy matched across the second level contrast?Unknown, not reported
Is reaction time matched across the second level contrast?Unknown, not reported
Behavioral data notes
Type of analysisVoxelwise
Search volumeWhole brain
Correction for multiple comparisonsClusterwise correction based on cluster_threshold_beta
SoftwareSPM12/in-house
Voxelwise p.01
Cluster extent126 voxels (size not stated)
Statistical detailsLesion volume covariate
Findings↑ L IFG pars orbitalis
↑ L SMA/medial prefrontal
↑ L angular gyrus
↑ L precuneus
↑ L posterior cingulate
↑ L hippocampus/MTL
↑ R IFG pars orbitalis
↑ R SMA/medial prefrontal
↑ R precuneus
↑ R anterior temporal
↑ R posterior cingulate
↑ R cerebellum
Findings notesBased on figure and table; larger activations are compelling; smaller activations are not due to lenient correction approach

Voxelwise analysis 4

First level contrastSemantic decision vs tone decision
Analysis classCross-sectional correlation with language or other measure
Group(s)Aphasia
CovariateLesion volume
Is the second level contrast valid in terms of the group(s), time point(s), and measures involved?Yes
Is accuracy matched across the second level contrast?Unknown, not reported
Is reaction time matched across the second level contrast?Unknown, not reported
Behavioral data notes
Type of analysisVoxelwise
Search volumeR hemisphere
Correction for multiple comparisonsClusterwise correction based on cluster_threshold_beta
SoftwareSPM12/in-house
Voxelwise p.01
Cluster extent126 voxels (size not stated)
Statistical details
Findings↑ R IFG pars opercularis
↑ R dorsolateral prefrontal cortex
↑ R dorsal precentral
↑ R SMA/medial prefrontal
↓ R orbitofrontal
↓ R anterior temporal
↓ R cerebellum
↓ R thalamus
Findings notesBased on figure and table; larger activations are compelling; smaller activations are not due to lenient correction approach

ROI analysis 1

First level contrastSemantic decision vs tone decision
Analysis classCross-sectional aphasia vs control
Group(s)Aphasia vs control
Covariate
Is the second level contrast valid in terms of the group(s), time point(s), and measures involved?Yes
Is accuracy matched across the second level contrast?No, different
Is reaction time matched across the second level contrast?Unknown, not reported
Behavioral data notesSemantic decision accuracy not matched, but tone decision accuracy not reported
Type of analysisRegions of interest (ROI)
ROI typeFunctional
How many ROIs are there?5
What are the ROI(s)?(1) overall canonical semantic network (CSN); (2) L CSN; (3) R CSN; (4) mirror L CSN in R; (5) out-of-network CSN in R
How are the ROI(s) defined?Control data
Correction for multiple comparisonsFamilywise error (FWE)
Statistical detailsCircular because ROI defined in one group
Findings↓ L IFG
↓ L dorsolateral prefrontal cortex
↓ L SMA/medial prefrontal
↓ L angular gyrus
↓ L precuneus
↓ L mid temporal
↓ L anterior temporal
↓ L occipital
↓ L posterior cingulate
↓ L cerebellum
↓ R IFG
↓ R dorsolateral prefrontal cortex
↓ R SMA/medial prefrontal
↓ R angular gyrus
↓ R precuneus
↓ R anterior temporal
↓ R occipital
↓ R posterior cingulate
↓ R cerebellum
Findings notesResults are for whole networks of regions, so individual regions cannot be assured; out-of-network R regions not listed since they were not significant in ROI 5 (only in ROI 4)

ROI analysis 2

First level contrastSemantic decision vs tone decision
Analysis classCross-sectional correlation with language or other measure
Group(s)Aphasia
CovariateLesion volume
Is the second level contrast valid in terms of the group(s), time point(s), and measures involved?Yes
Is accuracy matched across the second level contrast?Unknown, not reported
Is reaction time matched across the second level contrast?Unknown, not reported
Behavioral data notes
Type of analysisRegions of interest (ROI)
ROI typeFunctional
How many ROIs are there?5
What are the ROI(s)?(1) overall canonical semantic network (CSN); (2) L CSN; (3) R CSN; (4) mirror L CSN in R; (5) out-of-network CSN in R
How are the ROI(s) defined?Control data
Correction for multiple comparisonsFamilywise error (FWE)
Statistical details
FindingsNone
Findings notes

ROI analysis 3

First level contrastSemantic decision vs tone decision
Analysis classCross-sectional correlation with language or other measure
Group(s)Aphasia
CovariateSemantic decision accuracy
Is the second level contrast valid in terms of the group(s), time point(s), and measures involved?Yes
Is accuracy matched across the second level contrast?Accuracy is covariate
Is reaction time matched across the second level contrast?Unknown, not reported
Behavioral data notes
Type of analysisRegion of interest (ROI)
ROI typeFunctional
How many ROIs are there?1
What are the ROI(s)?CSN
How are the ROI(s) defined?Control data
Correction for multiple comparisonsOne only
Statistical detailsLesion volume covariate
Findings↑ L IFG
↑ L dorsolateral prefrontal cortex
↑ L SMA/medial prefrontal
↑ L angular gyrus
↑ L precuneus
↑ L mid temporal
↑ L anterior temporal
↑ L posterior cingulate
↑ L cerebellum
↑ R IFG
↑ R dorsolateral prefrontal cortex
↑ R SMA/medial prefrontal
↑ R angular gyrus
↑ R precuneus
↑ R anterior temporal
↑ R posterior cingulate
↑ R cerebellum
Findings notesCorrelation calculated for the whole network of regions, so correlation of individual regions cannot be assured

ROI analysis 4

First level contrastSemantic decision vs tone decision
Analysis classCross-sectional correlation with language or other measure
Group(s)Aphasia
CovariateAverage of semantic and phonemic fluency
Is the second level contrast valid in terms of the group(s), time point(s), and measures involved?Yes
Is accuracy matched across the second level contrast?Unknown, not reported
Is reaction time matched across the second level contrast?Unknown, not reported
Behavioral data notes
Type of analysisRegion of interest (ROI)
ROI typeFunctional
How many ROIs are there?1
What are the ROI(s)?CSN
How are the ROI(s) defined?Control data
Correction for multiple comparisonsOne only
Statistical detailsLesion volume covariate
Findings↑ L IFG
↑ L dorsolateral prefrontal cortex
↑ L SMA/medial prefrontal
↑ L angular gyrus
↑ L precuneus
↑ L mid temporal
↑ L anterior temporal
↑ L posterior cingulate
↑ L cerebellum
↑ R IFG
↑ R dorsolateral prefrontal cortex
↑ R SMA/medial prefrontal
↑ R angular gyrus
↑ R precuneus
↑ R anterior temporal
↑ R posterior cingulate
↑ R cerebellum
Findings notesCorrelation calculated for the whole network of regions, so correlation of individual regions cannot be assured

ROI analysis 5

First level contrastSemantic decision vs tone decision
Analysis classCross-sectional correlation with language or other measure
Group(s)Aphasia
CovariateBNT
Is the second level contrast valid in terms of the group(s), time point(s), and measures involved?Yes
Is accuracy matched across the second level contrast?Unknown, not reported
Is reaction time matched across the second level contrast?Unknown, not reported
Behavioral data notes
Type of analysisRegion of interest (ROI)
ROI typeFunctional
How many ROIs are there?1
What are the ROI(s)?CSN
How are the ROI(s) defined?Control data
Correction for multiple comparisonsOne only
Statistical detailsLesion volume covariate
Findings↑ L IFG
↑ L dorsolateral prefrontal cortex
↑ L SMA/medial prefrontal
↑ L angular gyrus
↑ L precuneus
↑ L mid temporal
↑ L anterior temporal
↑ L posterior cingulate
↑ L cerebellum
↑ R IFG
↑ R dorsolateral prefrontal cortex
↑ R SMA/medial prefrontal
↑ R angular gyrus
↑ R precuneus
↑ R anterior temporal
↑ R posterior cingulate
↑ R cerebellum
Findings notesCorrelation calculated for the whole network of regions, so correlation of individual regions cannot be assured

Complex analysis 1

First level contrastSemantic decision vs tone decision
Analysis classCross-sectional aphasia vs control
Group(s)Aphasia vs control
Covariate
Is the second level contrast valid in terms of the group(s), time point(s), and measures involved?Yes
Is accuracy matched across the second level contrast?No, different
Is reaction time matched across the second level contrast?Unknown, not reported
Behavioral data notesSemantic decision accuracy not matched, but tone decision accuracy not reported
Type of analysisComplex
Statistical detailsCorrelations between activation magnitudes in the L and R canonical semantic network (CSN) were compared between groups. However, this analysis is circular because the CSN ROIs were defined based on controls only.
FindingsNone
Findings notes

Complex analysis 2

First level contrastSemantic decision vs tone decision
Analysis classCross-sectional aphasia vs control
Group(s)Aphasia vs control
Covariate
Is the second level contrast valid in terms of the group(s), time point(s), and measures involved?Yes
Is accuracy matched across the second level contrast?No, different
Is reaction time matched across the second level contrast?Unknown, not reported
Behavioral data notesSemantic decision accuracy not matched, but tone decision accuracy not reported
Type of analysisComplex
Statistical detailsCorrelations between activation magnitudes in the L CSN and R mirrored CSN were compared between groups. However, this analysis is circular because the CSN ROIs were defined based on controls only.
FindingsOther
Findings notesCorrelations between activations in the L CSN and the mirrored L CSN in the R hemisphere were stronger in patients than controls.

Complex analysis 3

First level contrastSemantic decision vs tone decision
Analysis classCross-sectional aphasia vs control
Group(s)Aphasia vs control
Covariate
Is the second level contrast valid in terms of the group(s), time point(s), and measures involved?Yes
Is accuracy matched across the second level contrast?No, different
Is reaction time matched across the second level contrast?Unknown, not reported
Behavioral data notesSemantic decision accuracy not matched, but tone decision accuracy not reported
Type of analysisComplex
Statistical detailsCorrelations between activation magnitudes in the L CSN and R out-of-network homotopic regions were compared between groups. However, this analysis is circular because the CSN ROIs were defined based on controls only.
FindingsOther
Findings notesCorrelations between activations in the L CSN and R out-of-network homotopic regions were stronger in patients than controls.

Complex analysis 4

First level contrastSemantic decision vs tone decision
Analysis classCross-sectional aphasia vs control
Group(s)Aphasia vs control
Covariate
Is the second level contrast valid in terms of the group(s), time point(s), and measures involved?Yes
Is accuracy matched across the second level contrast?No, different
Is reaction time matched across the second level contrast?Unknown, not reported
Behavioral data notesSemantic decision accuracy not matched, but tone decision accuracy not reported
Type of analysisComplex
Statistical detailsThe difference in activation between the L CSN and R CSN was compared between patients and controls. However, this analysis is circular because the CSN ROIs were defined based on controls only.
FindingsNone
Findings notes

Complex analysis 5

First level contrastSemantic decision vs tone decision
Analysis classCross-sectional aphasia vs control
Group(s)Aphasia vs control
Covariate
Is the second level contrast valid in terms of the group(s), time point(s), and measures involved?Yes
Is accuracy matched across the second level contrast?No, different
Is reaction time matched across the second level contrast?Unknown, not reported
Behavioral data notesSemantic decision accuracy not matched, but tone decision accuracy not reported
Type of analysisComplex
Statistical detailsThe difference in activation between the L CSN and mirror L CSN in the R was compared between patients and controls. However, this analysis is circular because the CSN ROIs were defined based on controls only.
FindingsOther
Findings notesThe difference was smaller in patients.

Complex analysis 6

First level contrastSemantic decision vs tone decision
Analysis classCross-sectional aphasia vs control
Group(s)Aphasia vs control
Covariate
Is the second level contrast valid in terms of the group(s), time point(s), and measures involved?Yes
Is accuracy matched across the second level contrast?No, different
Is reaction time matched across the second level contrast?Unknown, not reported
Behavioral data notesSemantic decision accuracy not matched, but tone decision accuracy not reported
Type of analysisComplex
Statistical detailsThe difference in activation between the R CSN and out-of-network homotopic regions in the R was compared between patients and controls. However, this analysis is circular because the CSN ROIs were defined based on controls only.
FindingsOther
Findings notesThe difference was smaller in patients.

Complex analysis 7

First level contrastSemantic decision vs tone decision
Analysis classCross-sectional correlation with language or other measure
Group(s)Aphasia
CovariateInteractions of semantic fluency and naming measures by lesion size
Is the second level contrast valid in terms of the group(s), time point(s), and measures involved?Yes
Is accuracy matched across the second level contrast?Unknown, not reported
Is reaction time matched across the second level contrast?Unknown, not reported
Behavioral data notes
Type of analysisComplex
Statistical detailsFor the 4 R hemisphere regions that were more activated in patients with larger lesions (SPM analysis 4), analyses were carried out to determine whether the semantic fluency or naming measures were differentially impacted by activation depending on whether lesions were larger or smaller.
FindingsOther
Findings notesFor 1 of the 4 regions (R SMA), there were significant interactions such that in patients with larger lesions, more activation was associated with higher semantic fluency scores and higher BNT scores, while in patients with smaller lesions, more activation was associated with lower fluency and BNT scores. There was a similar relationship with semantic fluency in the R IFG pars opercularis but only at p(FDR) = 0.07.

Notes

Excluded analysesAncillary whole brain analyses without lesion volume covariate (Supporting Figure 3); Figure 3b and 3c, which are derivatives of included analyses