Aphasia Neuroplasticity Review

Supplementary Table S11. Cross-sectional aphasia compared to control: Methodologically robust analyses

AnalysisFirst level contrastSecond level contrastMatched forStatsNotesFindings
AccRT
Leff et al. (2002):
ROI 1
Higher word rates vs lower word rates CAC
Aphasia with pSTS damage (n = 6) vs control (n = 8)
NANB NANT ROI
Func
One
Number of ROIs: 1; ROI: R pSTS; how ROI defined: the peak voxel for the contrast in the R pSTS from each subject's individual analysis, but the search region is not stated; the controls and patients without pSTS damage were combined, however it is stated in the caption to Figure 2 that the patients with pSTS damage were significantly different to both ↑ R posterior STS
Blank et al. (2003):
Vox 1
Propositional speech production vs rest CAC
Aphasia with IFG POp damage (n = 7) vs control
N NANT Vox
SVC
Behavioral data notes: word rates not reported, but offline speech sample differed; search volume: voxels spared in all patients; software: SPM99; voxelwise p: FWE p < .05 with SVC in R pars opercularis ↑ R IFG pars opercularis
notes: no voxels survived FWE correction without SVC
Blank et al. (2003):
Vox 2
Propositional speech production vs rest CAC
Aphasia without IFG POp damage (n = 7) vs control
N NANT Vox
SVC
Behavioral data notes: word rates not reported, but offline speech sample differed; search volume: voxels spared in all patients; software: SPM99; voxelwise p: FWE p < .05 with SVC in R pars opercularis ↑ R IFG pars opercularis
Blank et al. (2003):
Vox 4
Propositional speech production vs counting CAC
Aphasia with IFG POp damage (n = 7) vs control
N NANT Vox
SVC
Behavioral data notes: word rates not reported, but offline speech sample differed; search volume: voxels spared in all patients; software: SPM99; voxelwise p: FWE p < .05 with SVC in R pars opercularis None
Blank et al. (2003):
Vox 5
Propositional speech production vs counting CAC
Aphasia without IFG POp damage (n = 7) vs control
N NANT Vox
SVC
Behavioral data notes: word rates not reported, but offline speech sample differed; search volume: voxels spared in all patients; software: SPM99; voxelwise p: FWE p < .05 with SVC in R pars opercularis None
Sharp et al. (2004):
Vox 1
Semantic decision vs syllable count decision CAC
Aphasia vs control (clear speech)
AM Y Vox
SVC
Behavioral data notes: interaction of group by task not reported for accuracy; search volume: whole brain; software: SPM99; voxelwise p: FWE p < .05 with SVC in fusiform gyri, temporal poles, L IFG, L orbitofrontal and L SFG ↓ L posterior inferior temporal gyrus/fusiform gyrus
Sharp et al. (2004):
ROI 1
Semantic decision vs syllable count decision CAC
Aphasia vs control (clear speech)
AM Y ROI
Anat
One
Behavioral data notes: interaction of group by task not reported for accuracy; number of ROIs: 1; ROI: L fusiform gyrus; how ROI defined: probabilistic brain atlas ↓ L posterior inferior temporal gyrus/fusiform gyrus
Sharp et al. (2004):
ROI 2
Semantic decision vs syllable count decision CAC
Aphasia vs control (noise vocoded)
NAM Y ROI
Anat
One
Behavioral data notes: patients were more accurate on semantic decisions than syllable decisions, whereas controls were less accurate on noise vocoded semantic decisions than clear syllable decisions (which were the baseline for this analysis); number of ROIs: 1; ROI: L fusiform gyrus; how ROI defined: probabilistic brain atlas None
notes: this analysis suggests that the difference between groups in the L fusiform gyrus disappears when the controls perform a semantic task that is similarly challenging
Zahn et al. (2004):
ROI 1
Semantic decision vs phonetic decision and lexical decision (conjunction) CAC
Aphasia vs control
UNT UNR ROI
LI
One
Behavioral data notes: relative performance on language and control tasks unclear; number of ROIs: 1; ROI: language network LI; conjunction analyses not clearly described; in two patients, a different conjunction was used (lexical decision vs phonetic decision & semantic decision vs phonetic decision) None
notes: LI > 0 in 12 out of 14 controls and 5 out of 7 patients; no significant difference
Crinion & Price (2005):
Vox 1
Listening to narrative speech vs listening to reversed speech CAC
Aphasia without temporal lobe damage (n = 9) vs control
NANB NANT Vox
VFWC
Search volume: whole brain; software: SPM2; voxelwise p: FWE p < .05; cluster extent cutoff: 5 voxels (size not stated) ↓ L dorsal precentral
↓ R somato-motor
Crinion & Price (2005):
Vox 2
Listening to narrative speech vs listening to reversed speech CAC
Aphasia with temporal lobe damage (n = 8) vs control
NANB NANT Vox
VFWC
Search volume: whole brain; software: SPM2; voxelwise p: FWE p < .05; cluster extent cutoff: 5 voxels (size not stated) ↓ L posterior STS
↓ L mid temporal
Crinion & Price (2005):
Cplx 2
Listening to narrative speech vs listening to reversed speech CAC
Aphasia without temporal damage (n = 9) vs control
NANB NANT Cplx
Correlations were computed between activity in each voxel, and post-scan story recall, and were compared between patients without temporal damage and controls, in regions with a main effect of story comprehension. The threshold was p < 0.05 corrected, plus a minimum cluster size of 5 voxels. None
Crinion & Price (2005):
Cplx 3
Listening to narrative speech vs listening to reversed speech CAC
Aphasia with temporal damage (n = 8) vs control
NANB NANT Cplx
Correlations were computed between activity in each voxel, and post-scan story recall, and were compared between patients with temporal damage and controls, in regions with a main effect of story comprehension. The threshold was p < 0.05 corrected, plus a minimum cluster size of 5 voxels. None
Crinion et al. (2006):
Vox 1
Listening to narrative speech vs listening to reversed speech CAC
Aphasia vs control
NANB NANT Vox
VFWE
Search volume: voxels spared in all patients; software: SPM99; voxelwise p: FWE p < .05 None
Crinion et al. (2006):
Vox 2
Listening to narrative speech vs listening to reversed speech CAC
Aphasia without temporal lobe damage (n = 6) vs control
NANB NANT Vox
VFWE
Search volume: voxels spared in all included patients; software: SPM99; voxelwise p: FWE p < .05 None
Crinion et al. (2006):
Vox 3
Listening to narrative speech vs listening to reversed speech CAC
Aphasia with temporal lobe damage (n = 18) vs control
NANB NANT Vox
VFWE
Search volume: voxels spared in all included patients; software: SPM99; voxelwise p: FWE p < .05 None
Warren et al. (2009):
ROI 1
Listening to narrative speech vs listening to reversed speech CAC
Aphasia vs control
NANB NANT ROI
Anat
NC
Number of ROIs: 6; ROIs: (1) L anterior superior temporal cortex; (2) L basal temporal language area; (3) L IFG pars triangularis; (4-6) homotopic counterparts; how ROIs defined: ROIs were defined anatomically in regions that were functionally connected with L anterior superior temporal cortex in controls (1-4) or homotopic to these (5-6); somewhat circular because ROIs were defined only in regions where controls showed significant connectivity (even though ROIs were anatomical) None
notes: L IFG pars triangularis almost reached significance (p = .053) for more activation in patients
Warren et al. (2009):
ROI 9
Listening to narrative speech vs listening to reversed speech CAC
Aphasia with positive anterior temporal interconnectivity (n = 8) vs control
NANB NANT ROI
Anat
NC
Number of ROIs: 6; ROIs: (1) L anterior superior temporal cortex; (2) L basal temporal language area; (3) L IFG pars triangularis; (4-6) homotopic counterparts; how ROIs defined: ROIs were defined anatomically in regions that were functionally connected with L anterior superior temporal cortex in controls (1-4) or homotopic to these (5-6); somewhat circular because ROIs were defined only in regions where controls showed significant connectivity (even though ROIs were anatomical); excluded 3 patients with L IFG damage ↑ L IFG pars triangularis
Warren et al. (2009):
ROI 10
Listening to narrative speech vs listening to reversed speech CAC
Aphasia with negative anterior temporal interconnectivity (n = 8) vs control
NANB NANT ROI
Anat
NC
Number of ROIs: 6; ROIs: (1) L anterior superior temporal cortex; (2) L basal temporal language area; (3) L IFG pars triangularis; (4-6) homotopic counterparts; how ROIs defined: ROIs were defined anatomically in regions that were functionally connected with L anterior superior temporal cortex in controls (1-4) or homotopic to these (5-6); somewhat circular because ROIs were defined only in regions where controls showed significant connectivity (even though ROIs were anatomical); excluded 1 patient with L IFG damage None
Fridriksson et al. (2010):
Vox 2
Picture naming (correct trials) vs viewing abstract pictures CAC
Aphasia vs control
YCT UNR Vox
C-
Search volume: whole brain; software: FSL 4.1; voxelwise p: ~.02 (z > 2); cluster extent cutoff: based on GRFT None
van Oers et al. (2010):
ROI 3
Verb generation vs rest CAC
Aphasia vs control
UNR UNR ROI
Mix
NC
Number of ROIs: 7; ROIs: (1) L anterior language region (IFG); (2) L posterior language region (AG, SMG, STG, MTG); (3) R anterior language region (IFG); (4) R posterior language region (AG, SMG, STG, MTG); (5) frontal LI; (6) temporal LI; (7) whole network LI; how ROIs defined: WFU pickatlas ↓ L IFG
↓ LI (language network)
↓ LI (frontal)
Allendorfer et al. (2012):
ROI 2
Verb generation (overt, event-related) vs noun repetition (event-related) CAC
Aphasia vs control
N UNR ROI
LI
NC
Behavioral data notes: patients less accurate and produced less responses on both conditions, but the difference between groups was greater for verb generation; number of ROIs: 2; ROIs: (1) frontal LI; (2) temporal LI ↓ LI (frontal)
Szaflarski et al. (2014):
ROI 1
Verb generation vs finger tapping CAC
Aphasia vs control
UNR UNR ROI
LI
NC
Number of ROIs: 3; ROIs: (1) frontal LI; (2) temporal LI; (3) language network LI ↓ LI (language network)
↓ LI (frontal)
notes: temporal LI was also marginally significantly reduced (p = .08)
Griffis et al. (2017a):
Cplx 1
Semantic decision vs tone decision CAC
Aphasia vs control
N UNR Cplx
Behavioral data notes: semantic decision accuracy not matched, but tone decision accuracy not reported; Multimodal canonical correlation analysis (mCCA) and joint ICA were used to identify 3 joint ICs (structural/functional) that were differently represented in the patient and control groups. Although there was no correction for multiple comparisons when the functional maps were thresholded, the maps for the three networks each appeared to relate to coherent parts of the semantic network. Other:
The first joint IC comprised preservation of tissue in L posterior temporo-parietal region, activity in the L AG and bilateral midline components of the canonical semantic network, and reduced activity in R frontal, temporal and parietal regions. The second joint IC comprised preservation of tissue in the the L basal ganglia/insula region, and activity predominantly in the IFG pars orbitalis bilaterally. The third joint IC comprised preservation of the L IFG and activity in the L IFG and DLPFC along with bilateral midline regions. The first joint IC was considered to provide more robust evidence for structure-function relationships than the other two, because it was the only one where individual structural and functional mixing coefficients remained correlated even when lesion volume was included as a covariate.

Second level contrast = Which of the 8 relevant classes of analyses is this? Which group or groups of participants are included? If there is a covariate, what is it?; Acc = Is accuracy matched across the second level contrast?; RT = Is reaction time matched across the second level contrast?; Stats = Does the analysis involve voxelwise statistics, region(s) of interest (ROI), or something else (complex)? If voxelwise, how are multiple comparisons across voxels accounted for? If ROI, were the ROI(s) anatomical, functional, laterality indices, mixed, or something else? If there was more than one ROI, how were the ROIs corrected for multiple comparions?; Yellow underline = minor limitation; Orange underline = moderate limitation; Red underline = major limitation; CAC = Cross-sectional aphasia vs control; Y = Yes, matched; YCT = Yes, correct trials only; NAM = No, but attempt made; N = No, different; UNT = Unknown, no test; AM = Appear mismatched; UNR = Unknown, not reported; NANB = N/A, no behavioral measure; NANT = N/A, no timeable task; Vox = Voxelwise; VFWE = Voxelwise FWE correction; VFWC = Voxelwise FWE correction and additional arbitrary cluster correction; C- = Clusterwise correction with with GRFT and lenient voxelwise p; SVC = Small volume correction; ROI = Region(s) of interest; Anat = Anatomical; Func = Functional; LI = Laterality indi(ces); Mix = Mixed; NC = No correction; One = One only; Cplx = Complex.