Brain Advance Access originally published online on February 4, 2004
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Brain, Vol. 127, No. 3, 628-649, 2004
© 2004 Guarantors of Brain
doi: 10.1093/brain/awh075
What’s in a name: voxel-based morphometric analyses of MRI and naming difficulty in Alzheimer’s disease, frontotemporal dementia and corticobasal degeneration
Departments of 1 Neurology and 2 Radiology, University of Pennsylvania, Philadelphia, Pennsylvania, USA
Correspondence to: Murray Grossman, Department of Neurology, 2 Gibson Hospital of the University of Pennsylvania, 3400 Spruce Street, Philadelphia, PA 19104–4283, USA E-mail: mgrossma{at}mail.med.upenn.edu
 |
Summary |
|---|
 Top Summary IntroductionMethodsResultsDiscussionReferences |
|---|
Confrontation naming is impaired in neurodegenerative conditions like Alzheimer’s disease (AD), frontotemporal dementia (FTD) and corticobasal degeneration (CBD). Some behavioural observations suggest a common source of impaired naming across these patient groups, while others find partially unique patterns of naming difficulty. We hypothesized that a large-scale neural network underlies naming, and that patterns of impaired naming in AD, FTD and CBD reflect cortical atrophy that interrupts this network in a manner that is partially shared and partially unique across these patient groups. We tested this hypothesis by correlating naming impairments with voxel-based morphometric (VBM) analyses of cortical atrophy in structural MRIs of 50 patients. We found significant naming deficits in all patient groups. Naming also correlated with lexical retrieval in all patient groups, including subgroups of patients with FTD. VBM analyses showed significant cortical atrophy, which was shared across AD, FTD and CBD patients in the left lateral temporal cortex; this area correlated with naming accuracy in all groups. Left lateral temporal atrophy thus appears to interfere with a lexical retrieval component of naming in AD, FTD and CBD. Impaired naming also correlated with semantic memory and visual perceptual–spatial functioning in specific groups of patients and, correspondingly, naming correlated with cortical atrophy in partially distinct neuroanatomical distributions in AD, FTD, CBD and subgroups of patients with FTD. These partially unique correlation profiles appear to reflect selective interruption of other components of the naming process, including semantic and visual perceptual–spatial functioning. These findings are consistent with the hypothesis that a large-scale neural network supports naming, and that this network is interrupted in several distinct ways in patients with neurodegenerative diseases.
Key Words: Alzheimer’s; frontotemporal; corticobasal; naming; cortical atrophy
Abbreviations: AD= probable Alzheimer’s disease; CBD = corticobasal degeneration; FTD = frontotemporal dementia; MMSE = Mini-Mental State Examination; NON-APH = non-aphasic patients with frontotemporal dementia; ns = not significant; PNFA = progressive non-fluent aphasia; ROI = region of interest; SD = semantic dementia; VBM = voxel-based morphometry
Received June 23, 2003. Accepted November 4, 2003.
|
Introduction |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
Confrontation naming difficulty is impaired in patients suffering from neurodegenerative diseases such as probable Alzheimer’s disease (AD) (Tippett and Farah, 1994
; Hodges et al., 1996; Lambon Ralph et al., 1997), frontotemporal dementia (FTD) (Cappa et al., 1998; Lambon Ralph et al., 1998) and corticobasal degeneration (CBD) (Kompoliti et al., 1998; Black, 2000; Kertesz et al., 2000). Behavioural studies have attempted to establish the sources of naming difficulty in these patient groups to help with differential diagnosis and severity staging, and to improve our fundamental understanding of the neural basis for this critical element of human communication. However, this body of work has not arrived at a definitive conclusion. While cognitive models of naming have pointed out the large variety of possible naming deficits (Garrett, 1992; Levelt, 1992), qualitative analyses of naming errors document a relatively limited number of possible errors. The kind of deficit that is the most common, i.e. no response, is the least informative. In this study, we examined naming difficulty from a different perspective, i.e. through comparative analyses of correlations between confrontation naming difficulty and grey matter atrophy using voxel-based morphometry (VBM) analyses of high-resolution structural MRI in patients with impaired naming.
Our model of confrontation naming includes several components: (i) a stimulus such as a visual line drawing must be interpreted; (ii) the concept corresponding to this stimulus must be identified in semantic memory; and then (iii) the name that corresponds to the stimulus must be retrieved and expressed (Garrett, 1992; Levelt, 1992). Each of these components is quite complex. Lexical retrieval, for example, may involve components such as: (i) selecting the best name from among several possible choices which label objects with overlapping features (e.g. ‘collie’ versus ‘dog’); (ii) inhibiting alternate names that share many phonological features (e.g. ‘collie’ versus ‘dolly’); (iii) translating an abstract, material-neutral representation of a name into a material-specific form; and (iv) assembling the phonological or graphemic elements so that they can be spoken or written (e.g. the written word ‘collie’).
Naming has been studied extensively in neurodegenerative diseases such as AD—partly because this crucial linguistic process is impaired frequently and early in these illnesses. Naming deficit is also prominent in the subgroup of FTD patients with a fluent form of progressive aphasia that is also known as semantic dementia (SD) and, for example, is considered a diagnostic feature of this condition (Snowden et al., 1989; Hodges et al., 1992a; Neary et al., 1998). In AD, naming difficulty has been associated with impairments in lexical retrieval (Hodges et al., 1991; Cronin-Golomb et al., 1992), semantic (Huff et al., 1986; Lambon Ralph et al., 1997; Hodges et al., 1996) and visual (Tippett and Farah, 1994; Silveri and Leggio, 1996) components of naming. There is partial overlap with the basis for naming difficulty in FTD, as the deficit in this group has also been associated with an interruption of lexical retrieval (Weintraub et al., 1990; Thompson et al., 1997; Lambon Ralph et al., 1998) and semantic (Hodges et al., 1995; Lambon Ralph et al., 2001) processes. Individual case studies and occasional group reports also describe impaired naming in CBD (Kompoliti et al., 1998; Kertesz et al., 2000; P. Moore, K. Dennis and M. Grossman, unpublished data). These authors have emphasized the contribution of lexical retrieval and visual perceptual–spatial limitations.
Given the partially shared basis for impaired naming in these neurodegenerative diseases, one hypothesis is that naming difficulty is due to a deficit in a single critical component of the naming process such as lexical retrieval. A deficit in this process is due to cortical dysfunction within a specific neuroanatomical distribution that is compromised across patient groups, according to this approach. Quantitative differences in naming accuracy between groups are due to the severity of disease in this crucial brain area. Evidence consistent with this ‘single deficit’ hypothesis comes from correlation studies that associate naming difficulty with impaired lexical retrieval in many groups of patients. For example, several studies have emphasized the importance of difficulty with lexical retrieval in mildly impaired patients with AD (Chertkow et al., 1992; Hodges et al., 1992b, 1996; Garrard et al., 1998, 2001) and CBD (P. Moore, K. Dennis and M. Grossman, unpublished data). Subgroups of patients with FTD have distinct clinical presentations (Neary et al., 1998; Grossman, 2002), yet SD patients, progressive non-fluent aphasia (PNFA) patients and non-aphasic patients with FTD (NON-APH) with a disorder of social and executive functioning appear to share naming difficulty and an impairment of lexical retrieval (Weintraub et al., 1990; Hodges and Patterson, 1996; Thompson et al., 1997; Croot et al., 1998; P. Moore, K. Dennis and M. Grossman, unpublished data).
Additional support for this ‘single deficit’ approach comes from MRI studies suggesting shared patterns of cortical atrophy in AD and FTD. Areas of atrophy common to these patients appear to include the temporal neocortex, the frontal neocortex and the hippocampus (Baron et al., 2001; Chan et al., 2001; Galton et al., 2001; Busatto et al., 2003; Karas et al., 2003). Assuming a single source of naming difficulty, a correlative MRI study related impaired lexical retrieval in confrontation naming and category naming fluency tests in a combined group of SD patients and patients with AD to atrophy in a single region—the left anterior temporal cortex (Galton et al., 2001).
An alternate hypothesis is based on the partially distinct profiles of impaired naming in AD, FTD and CBD. This approach suggests that naming difficulty in these patient groups is related in part to the unique anatomical distribution of the cortical atrophy seen in these patients. Consistent with this ‘large-scale network’ approach, patients with AD, FTD or CBD have deficits in partially distinct cognitive components of the naming process. For example, a semantic deficit appears to contribute more prominently to naming difficulty in AD and FTD compared with CBD (Huff et al., 1986; Hodges et al., 1995; Lambon Ralph et al., 1997, 2001), while a visual perceptual–spatial deficit plays a more important role in naming difficulty in AD and CBD compared with FTD (Tippett and Farah, 1994; Silveri and Leggio, 1996; P. Moore, K. Dennis and M. Grossman, unpublished data). Although all FTD subgroups have lexical retrieval difficulty, careful comparisons suggest qualitative differences in the nature of their lexical retrieval deficit as well (Thompson et al., 1997; Lambon Ralph et al., 1998; P. Moore, K. Dennis and M. Grossman, unpublished data). From this perspective, even such components of naming as lexical retrieval and the associated downstream processes are multifactorial in nature and may be compromised in several different ways.
Additional support for this ‘large-scale network’ approach comes from MRI studies suggesting partially distinct patterns of cortical atrophy in AD and FTD. Although direct comparisons of the anatomical distribution of atrophy in AD and FTD have been rare, distributions of atrophy in the temporal lobe appear to be somewhat different in these conditions (Baron et al., 2001; Chan et al., 2001; Galton et al., 2001; Good et al., 2002; Boxer et al., 2003; Busatto et al., 2003; Karas et al., 2003). Each FTD subgroup also appears to have a partially distinct anatomical distribution of disease, according to in vivo measures such as single photon emission computed tomography (Jagust et al., 1989; Miller et al., 1997), PET (Grossman et al., 1996b; Turner et al., 1996) and quantitative analyses of atrophy in high resolution MRI studies. Patients with SD thus seem to have prominent left anterior temporal disease (Mummery et al., 2000; Chan et al., 2001; Galton et al., 2001), while inferior frontal regions of the left hemisphere appear to be most compromised in patients with PNFA (Grossman et al., 1996b; Rosen et al., 2002b; Nestor et al., 2003) and disease in NON-APH patients is said to be focused in frontal regions of the right hemisphere (Miller et al., 1993; Rosen et al., 2002a). Occasional non-quantitative neuroimaging studies of CBD suggest a distribution of disease primarily involving the parietal cortex, although the temporal and frontal cortical regions also may be atrophic (Caselli et al., 1992; Grisoli et al., 1995; Savoiardo et al., 2000).
Patients with AD, FTD and CBD have not been directly compared with respect to the neural basis for their naming deficit. In this study, we correlated VBM analyses of volumetric structural MRI with naming to assess the hypothesis that partially shared and partially unique patterns of cortical atrophy across patient groups account for the profiles of impaired naming seen in these patients.
|
Methods |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
Subjects
We studied 50 patients diagnosed with a neurodegenerative condition at the Department of Neurology at the University of Pennsylvania. Initial clinical diagnosis was established by an experienced neurologist (M.G.) using published criteria. Subsequently, at least two trained reviewers of a consensus committee confirmed the presence of specific diagnostic criteria based on an independent review of the semi-structured history, mental status examination and neurological examination. If the reviewers disagreed in their diagnosis (which occurred in 11% of 70 cases, including patients with other diagnoses not participating in this study), consensus was established through open discussion by the entire committee. These patients and their legal representatives participated in an informed consent procedure approved by the Institutional Review Board at the University of Pennsylvania.
Among the participants, 12 patients were given the diagnosis of AD based on National Institute of Neurological and Communicative Disorders and Stroke Alzheimer’s Disease and Related Disorders Association (NINCDS–ADRDA) criteria (McKhann et al., 1984). This included a progressive syndrome involving prominent episodic memory difficulty, associated with circumlocutory speech, a visual constructional impairment and/or an executive limitation.
Another 29 patients were given the diagnosis of FTD, according to published criteria (Lund and Manchester Groups, 1994; McKhann et al., 2001). We also used the consensus mechanism described above to establish subgroup diagnosis. The subgroups were based on published criteria (Neary et al., 1998), which have been modified to improve reliability (Davis et al., 2001; Price et al., 2001). We studied 15 FTD patients with a progressive form of aphasia and 14 FTD patients with a non-aphasic pattern of cognitive and social impairment. One aphasic subgroup consisted of patients presenting with a fluent form of progressive aphasia, also known as SD (n = 8). This was characterized by fluent and circumlocutory spontaneous speech, which may be empty in content, and is associated over time with difficulty understanding single words. Since these patients were relatively early in the course of their disease, they had a prominent naming deficit without widespread difficulty understanding single words. Another progressive aphasic subgroup of FTD patients consisted of those presenting with PNFA (n = 7). These patients had effortful speech, which may be associated with dysarthria and impaired grammatical comprehension, but relatively good single word comprehension. The NON-APH subgroup of FTD patients (n = 14) presented with social and behavioural difficulties, and a limitation of executive functioning.
Finally, nine patients were given the clinical diagnosis of CBD based on clinical-pathological studies reported in the literature and our own autopsy series (Rinne et al., 1994; Grimes et al., 1999; Riley and Lang 2000; Forman et al., 2002). These patients had progressive cortical sensory loss, apraxia and/or a lateralized extrapyramidal disorder (unilateral limb rigidity, myoclonus, dystonia, alien hand syndrome, gait difficulty and/or axial rigidity, but minimal resting tremor).
The initial clinical diagnosis of a neurodegenerative disease was consistent with the results of serum studies, structural imaging studies such as MRI or CT, studies of CSF (when available) and clinical functional neuroimaging studies such as single photon emissions computerized tomography (SPECT) or PET (these studies were not available to the consensus committee). Exclusion criteria included the presence of other neurological conditions such as stroke or hydrocephalus, primary psychiatric disorders such as depression or psychosis, or a systemic illness that can interfere with cognitive functioning. Most of these patients were taking a fixed dosage of an acetylcholinesterase inhibitor (e.g. donepezil, rivastigmine or galantamine). Some of these patients may have been medicated with a low dosage of a non-sedating anti-depressant (e.g. serotonin-specific re-uptake inhibitors such as sertraline) or an atypical neuroleptic agent (e.g quetiapine), as indicated clinically, but none of the patients demonstrated any evidence of sedation suggesting over-medication. Table 1 summarizes the demographical and clinical features of these patients. AD, FTD and CBD patient groups were selected to match in terms of age [F(2,47) = 1.49; not significant (ns)], education [F(2,47) = 2.37; ns), duration of disease [F(2,47) = 0.61; ns] and Mini-Mental State Examination (MMSE) score (Folstein et al., 1975) [F(2,47) = 0.37; ns]. Similarly, FTD patient subgroups were matched in terms of their age [F(2,26) = 0.52; ns], education [F(2,26) = 0.10; ns], duration of disease [F(2,26) = 0.03; ns] and MMSE score [F(2,26) = 2.40; ns].
|
Confrontation naming in these patients was compared with cohorts of 25 healthy older control subjects who were closely matched in age and education to each group of patients. The performance of each patient was converted to a Z-score based on each group’s matched control subjects; we compare these normalized Z-scores statistically across the patient groups in Results below. It may be noted that the subgroups of control subjects did not differ statistically in their naming performance. The patients were participating in a longitudinal protocol. For the purpose of the present study, we selected the naming performance dataset closest in time to the MRI (naming data were obtained on the same day as the MRI in the overwhelming majority of patients) and these data were typically obtained early in the course of disease. Imaging data in these patients was compared with 12 right-handed, healthy, control subjects who were matched for age [mean (±SD) = 68.5 (±9.4) years]) and education [mean (±S.D.) = 15.4 (±1.8) years].
Cognitive materials and procedure
We administered several measures to assess confrontation naming and the major processes thought to contribute to naming. We selected a validated measure corresponding most closely to each component of naming that we hypothesized. The patients were offered rest breaks between tasks as necessary during the performance of these measures.
Confrontation naming
To assess confrontation naming (Morris et al., 1989), we used an abbreviated version of the Boston naming test (Kaplan et al., 1983), which was thought to be representative of the full protocol. Each subject was asked to name orally each test stimulus (n = 15). All visual stimuli were black-and-white line drawings, and target names were divided equally between high frequency, mid-frequency and low frequency items. Patients were given as much time as they needed to respond.
Lexical retrieval
To assess semantically-guided lexical search, retrieval and associated downstream processes such as phonological assembly and articulation, patients were asked to name orally as many different words as possible belonging to a target semantic category, i.e. ANIMALS (Mickanin et al., 1994). They were given 60 s to complete this task. We report the number of unique words meeting the category criterion in this time span. Two FTD patients (one SD patient and one NON-APH patient) did not complete this task.
Semantic category membership judgement
To assess semantic memory with a simple task that required little expression and minimized executive resource demands, patients were asked to judge the semantic category membership of 48 individually presented stimuli in response to a simple probe (‘Is it an X?’) (Grossman et al., 1996a). One target category was natural (VEGETABLES) and one manufactured (TOOLS); half of the stimuli in each category were targets and half foils, and half of each category of stimuli were printed words and half colour photos (matched for frequency, familiarity and visual complexity across categories). Stimuli were presented in a manner blocked by category and material. Patients were given as much time as they needed to complete the task. Performance on this task has been validated by behavioural and SPECT correlation studies in AD patients with impaired semantic memory (Grossman et al., 1997). Two FTD patients (both NON-APH patients) and one CBD patient did not complete this task.
Visual–spatial functioning
To assess visual perceptual–spatial functioning (Morris et al., 1989) in a manner involving a constructional element, patients were asked to copy four geometric designs graded according to their perceptual–spatial complexity. Performance was evaluated on an 11-point scale. One patient in each group (including one NON-APH FTD patient) did not complete this task.
Imaging procedure
All images were acquired by a GE Horizon Echospeed 1.5 T MRI scanner (GE Medical Systems, Milwaukee, WI, USA). Each study began with a rapid sagittal T1-weighted image to determine patient position. High resolution T1-weighted 3D spoiled gradient echo images were then acquired with TR (repetition time) = 35 ms, TE (echo time) = 6 ms, slice thickness = 1.3 mm, flip angle = 30°, matrix size = 128 x 256, and a rectangular FOV (field of view) giving an in-plane resolution of 0.9 x 0.9 mm. The brain volumes were normalized by registration to the T1 template (Evans et al., 1993) of 305 averaged brain volumes in SPM99 (Frackowiak et al., 1997) using 12-parameter affine registration, non-linear registration using 12 non-linear iterations and 7 x 8 x 7 basis functions. The brains were normalized to Talairach and Tournoux brain coordinates (Talairach and Tournoux, 1988). We examined the results of both unmodulated and modulated routines, but a comparison found no regional differences between these techniques. We report below the results from the unmodulated analysis.
We used VBM to analyse brain volumes (Ashburner and Friston, 2000). SPM99 was used to segment the brain volumes into four tissue types (grey matter, white matter, CSF and other). Based on a priori MRI information, the segmentation algorithm in SPM99 calculates a Bayesian probability for each voxel of each tissue group in the volume. We inspected each slice of each segmented volume to ensure that no voxels from the dural sinuses or other adjacent non-brain structures were misclassified as grey matter (Good et al., 2001; Karas et al., 2003). Lastly, using SPM99, the grey matter volume was smoothed with a 12 mm FWHM (full width at half maximum) Gaussian filter to minimize individual gyral variations.
SPM99 was used for all statistical analyses. This included a two sample t-test routine to compare the grey matter volume of each patient group to the control group of 12 healthy seniors, and to assess relative grey matter atrophy across patient groups by comparing pairs of contrasts involving each patient group and the healthy seniors. A proportional analysis threshold was used to include only voxels with 
40% of the grand mean value. Implicit masking was used to ignore zeros and global calculation was based on the mean voxel value. We set our statistical threshold for the atrophy studies of each patient group relative to control subjects at a value of P < 0.0001. We did not correct for multiple comparisons in this and other analyses due to the hypothesis-driven nature of the statistical tests. Moreover, the small size of the voxels would have made a Bonferroni-like statistical correction too conservative. Instead, for all analyses, we accepted only clusters comprised of 100 adjacent voxels because this would also demonstrate a consistent effect in a particular neuroanatomical distribution (Forman et al., 1995). The between group comparisons of grey matter atrophy derived from pairs of patient–control contrasts were set at a statistical threshold of P < 0.001. The correlation analyses involved a regression of the confrontation naming Z-score on grey matter atrophy derived from the contrast of each patient group with the elderly control subjects. We set a statistical threshold for these analyses at P < 0.001. We focused on the significant correlations of naming and cortical volume that correspond to areas of significant grey matter atrophy. These correlations occur in demonstrably abnormal brain regions, and it is these correlated voxels that are likely to reflect interruptions of the large-scale neural network underlying naming. In comparison, naming–cortical correlations involving brain regions without significant grey matter atrophy are difficult to interpret unequivocally as these correlations are in regions where the large-scale neural network for naming is less likely to be compromised.
|
Results |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
Behavioural analyses of naming performance
Confrontation naming accuracy is summarized in Table 1. As can be seen, the average naming deficit in AD, FTD and CBD patient groups differed significantly from older control subjects’ performance, using a criterion of Z < –1.96 (significant at the P < 0.05 level). However, an analysis of variance (ANOVA) did not reveal a difference in naming accuracy across patient groups [F(2,47) = 0.38; ns]. All three FTD subgroups were significantly impaired in their confrontation naming relative to the performance of healthy seniors as well (at least at the P < 0.05 level, according to Z-score distributions). Although there was no statistical difference between FTD patient subgroups in their confrontation naming [F(2,26) = 0.48; ns], Table 1 shows that SD patients were the most impaired FTD subgroup. SD patients were also more impaired than AD patients and CBD patients. An evaluation of individual patient performance profiles revealed that 32 (64%) of the patients differed significantly from older control subjects in their naming accuracy at the P < 0.05 level (Z < –1.96). This included seven (58%) of the AD patients, four (44%) of the CBD patients and 21 (72%) of the FTD patients [including seven (88%) of the SD patients, four (56%) of the PNFA patients and 10 (71%) of the NON-APH patients]. These findings confirm previous observations that confrontation naming is widely compromised in patients with these neurodegenerative diseases and that confrontation naming is particularly impaired in the SD subgroup of patients with FTD.
Table 1 also summarizes the performance of the patient groups with regard to measures of each of the major cognitive components thought to contribute to confrontation naming. All patient groups differed from control subjects in their lexical retrieval performance at the P < 0.01 level (according to Z-score distributions), including each of the FTD patient subgroups. However, lexical retrieval performance did not differ between the patient groups [F(2,47) = 0.20; ns] nor between subgroups of patients with FTD [F(2,26) = 0.65; ns]. Inspection of individual patient performance profiles revealed that 42 (88%) of the patients differed significantly from older control subjects in their lexical retrieval performance at the P < 0.05 level (Z < –1.96). This included all 12 (100%) of the AD patients, seven (78%) of the CBD patients and 23 (85%) of the 27 FTD patients who performed this task (including six (86%) of the seven SD patients, seven (100%) of the PNFA patients and 10 (77%) of the 13 NON-APH patients). None of the patient groups differed significantly from the older control subjects in their performance on the simple measure of semantic memory we administered (including each of the FTD patient subgroups). Nevertheless, an ANOVA revealed a difference between groups [F(2,46) = 5.44; P < 0.01]. AD patients (33% of whom were impaired according to individual Z-score analyses) differed significantly from FTD patients and CBD patients in their semantic memory performance at the P < 0.05 level according to a Newman–Keuls procedure. CBD patients differed significantly from older control subjects in their visual–spatial performance at the P < 0.01 level (according to the Z-score distribution), although AD patients and FTD patients (and each of the FTD patient subgroups) did not differ from control subjects. We also found a difference between groups on the visual–spatial measure [F(2,46) = 24.69; P < 0.001]. CBD patients (seven of whom were impaired according to individual Z-score analyses) differed significantly from AD patients and FTD patients in their visual–spatial performance at least at the P < 0.05 level according to a Newman–Keuls procedure.
A correlation analysis is summarized in Table 2. As can be seen, confrontation naming accuracy correlated with lexical retrieval performance in AD, FTD and CBD. We also saw a correlation between naming accuracy and lexical retrieval in each of the FTD subgroups. We also found that CBD patients show a correlation between confrontation naming and visual–spatial functioning. In FTD, confrontation naming accuracy also correlated with semantic memory performance.
|
These findings confirmed significant naming difficulty in mildly-to-moderately demented patients with AD, CBD and FTD. Moreover, lexical retrieval was significantly impaired across all patient groups and correlated with impaired confrontation naming in each patient group. These observations emphasize the importance of lexical retrieval in these patients and suggest that impaired naming may be due to the disruption of a single component of the naming process across groups of patients with different neurodegenerative diseases. However, semantic memory appeared to play a role in naming difficulty in FTD and visual–spatial functioning contributed to the naming deficits of CBD patients. These findings are more consistent with the hypothesis that various components of the naming process may be disrupted differentially in neurodegenerative diseases; this may result in partially distinct patterns of impaired naming across these groups of patients. Indeed, the lexical retrieval component of naming and associated downstream processes can be compromised in many ways.
VBM analyses of cortical atrophy
The anatomical loci of peak grey matter atrophy in patients with AD, FTD and CBD, and the extent of the associated clusters are summarized in Table 3. These findings showed partially overlapping distributions of cortical atrophy across the patient groups. AD patients demonstrated significant grey matter atrophy in the bilateral temporal, left frontal and right parietal regions. FTD patients also showed cortical atrophy in the bilateral temporal and frontal brain regions. CBD patients showed grey matter atrophy in the bilateral temporal and frontal regions as well as the bilateral parietal regions. As illustrated in Fig. 1A, patients with AD, FTD and CBD showed overlapping distributions of significant cortical atrophy in the lateral temporal cortex of the left hemisphere. There was also a very small area of overlapping significant atrophy in the right inferolateral temporal cortex (11 voxels). An area of cortical atrophy common to AD, FTD and CBD suggests the possibility that the basis for impaired naming may be shared in part across these patient groups.
|
|
We also examined cortical atrophy in AD, FTD and CBD patient groups compared with each of the other patient groups. These group-by-group contrasts, summarized in Table 3 and illustrated in Fig. 1, emphasized differences in the distribution of grey matter atrophy across these patient groups. Consider first the pattern of atrophy in AD compared with FTD and CBD. Relative to FTD, AD patients have significantly greater atrophy in the bilateral posterolateral temporal-parietal and lateral occipital regions (Fig. 1B). We also observed significantly greater atrophy in the left hippocampus in AD relative to FTD (not shown). Relative to CBD, AD patients have greater atrophy in a left anterior and ventral temporal distribution (Fig. 1C). AD patients also have greater atrophy than CBD patients in the left anterior cingulate cortex (not shown). Atrophy in FTD also differed from that seen in AD and CBD. FTD patients have relatively greater atrophy in the right prefrontal (Fig. 1D) and left medial frontal (not shown) regions compared with AD patients. Relative to CBD, FTD patients have greater atrophy in the left anterior temporal (Fig. 1E) and the left anterior cingulate (not shown) regions. CBD patients showed greater atrophy in bilateral parietal and right frontal and temporal regions compared with AD patients (Fig. 1F). Compared with FTD patients, CBD patients have greater atrophy in bilateral temporal-parietal regions (Fig. 1G), including the medial parietal cortex (not shown). These partially distinct patterns of cortical atrophy are consistent with the hypothesis that the neural basis for impaired naming may differ in part across these patient groups.
We performed similar analyses of grey matter atrophy in subgroups of patients with FTD. These findings are summarized in Table 4. As can be seen, significant grey matter atrophy was evident in several areas of the left temporal cortex in SD. Patients with PNFA also showed significant cortical atrophy in portions of the left temporal lobe, as well as significant grey matter atrophy in several left frontal regions. NON-APH patients with FTD also demonstrated some left temporal grey matter atrophy, as well as cortical atrophy in a right frontal distribution. Figure 2A illustrates the distribution of grey matter atrophy in these FTD subgroups. An area of significant cortical atrophy shared across all FTD patients was seen in the left anterior temporal cortex; this is consistent with the possibility that disruption of a single component of the naming process may explain naming difficulty in all subgroups of patients with FTD.
|
|
Areas of significant cortical atrophy that differed across subgroups of FTD patients are summarized in Table 4. For example, Fig. 2B shows that SD patients have significant grey matter atrophy in the left posterolateral temporal cortex relative to PNFA patients. Figure 2D illustrates the distribution of significant grey matter atrophy in PNFA relative to SD in the frontal cortex of the right hemisphere. Other pairwise FTD subgroup comparisons are shown in Fig. 2. These differences suggest that confrontation naming difficulty in subgroups of FTD patients may be due to partially distinct sources of impairment.
Correlations between confrontation naming accuracy and cortical volume
Table 5 summarizes the coordinates of the peak foci of significant correlations between confrontation naming performance and grey matter volume in patients with AD, FTD and CBD. In AD, significant correlations between cortical volume and confrontation naming were found in the left lateral temporal cortex and the left anterior cingulate cortex. Fig. 3 shows the distribution of areas of significant correlation between naming and cortical volume, which overlap with areas of significant cortical atrophy. The correlations between naming performance and grey matter volume corresponded to areas of significant cortical atrophy in an anterior-lateral temporal distribution of the left hemisphere, as well as a small inferolateral temporal area in the right hemisphere (Fig 3A).
|
|
Table 5 shows that patients with FTD have significant correlations between confrontation naming accuracy and grey matter volume in the bilateral anterior temporal and bilateral frontal regions. The correlation between naming and cortical volume in FTD corresponds to the area of significant cortical atrophy in the left anterior-lateral temporal cortex, as well as small areas in the right temporal cortex and left frontal cortex (Fig. 3B).
Table 5 also shows that naming performance correlated with grey matter volume in CBD in the bilateral frontal and temporal cortices. These naming-cortical correlations correspond to areas of significant cortical atrophy in CBD in small areas of the frontal and temporal cortex bilaterally (Fig. 3C). These observations are consistent with the hypothesis that the neural basis for impaired naming partially overlaps in lateral aspects of the left temporal cortex across patients with AD, FTD and CBD. Unique correlations in each patient group also appear to provide evidence consistent with the claim that partially distinct components of the large-scale neural network for naming are interrupted in patients with neurodegenerative diseases.
We also examined patterns of correlation between confrontation naming performance and cortical volume in subgroups of patients with FTD. Table 5 summarizes the coordinates of the peak foci of significant correlation between confrontation naming and grey matter volume in these patients. In SD, significant correlations were seen in bilateral temporal cortex. Fig 4A shows the area of correlation between naming and cortical volume in SD that overlaps with significant cortical atrophy. This was in the left lateral temporal region. Table 5 indicates that the anatomical distribution of significant naming-volume correlation in PNFA was in the frontal and temporal areas bilaterally. The naming-volume correlations overlapping with areas of significant grey matter atrophy in PNFA were in the left anterior temporal cortex, as well as the inferior, orbital, dorsolateral prefrontal and premotor portions of the left frontal lobe (Fig. 4B). In NON-APH patients, significant naming-volume correlations were seen in bilateral frontal cortex extending to the anterior cingulate cortex and in the left temporal and parietal regions (Table 5). Significant naming-volume correlations overlapping with regions of significant cortical atrophy in NON-APH (Fig. 4C) were in the left anterior temporal cortex and the right dorsolateral prefrontal cortex. These distinct naming-cortical correlations are consistent with the hypothesis that a large-scale neural network for naming is interrupted in several unique ways in subgroups of patients with FTD.
|
|
Discussion |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
Naming is a complex process involving components such as interpreting a visual stimulus, identifying the corresponding concept in semantic memory, and selecting and expressing the name that best labels this concept. Since naming is so complex, a neurobiologically valid model of naming is likely to require converging evidence from multiple sources. This may include: neuroanatomical observations based on studies of stroke patients (Hillis et al., 2001
); functional neuro imaging studies of healthy subjects using PET (Howard et al., 1992; Whatmough et al., 2002) or functional MRI (Smith et al., 2001; Simon et al., 2002); fine-grained temporal observations of the naming process with electrocortical studies of naming during the presurgical evaluation of epilepsy patients (Ojemann and Schoenfield-McNeill, 1999; Crone et al., 2001); and event-related scalp potential studies of healthy subjects (Indefrey and Levelt, 2000). The present study examines the nature of naming difficulty in patients with AD, FTD and CBD, and the relationship between these naming deficits and the patterns of cortical atrophy seen in these patients. We argue below that naming is supported by a large-scale neural network involving multiple cortical regions. Our observations are consistent with the hypothesis that this network is interrupted in a manner which is partially shared across these patient groups and in a manner that is partially unique to each patient group. This yields a pattern of naming difficulty that includes a component common to all groups, namely lexical retrieval, as well as components distinct to each patient group involving specific aspects of retrieval as well as impairments of semantic memory and visual perceptual–spatial functioning.
Behavioural observations of impaired naming: partially shared and partially distinct patterns of naming difficulty
We found that naming difficulty is significantly compromised in AD, FTD and CBD. This is consistent with many previous observations of naming performance in these patient groups. Moreover, the level of naming difficulty was quantitatively equivalent across groups. One factor appeared to contribute to naming difficulty across all three patient groups: consistent with previous observations, naming accuracy correlated with lexical retrieval in AD (Hodges et al., 1991; Cronin-Golomb et al., 1992), in FTD (Thompson et al., 1997; Lambon Ralph et al., 1998) and in CBD (P. Moore, K. Dennis and M. Grossman, unpublished data). This suggests that a single component, i.e. lexical retrieval, may play a role in naming difficulty shared across patient groups.
It is also apparent in our data and in previously published work that these patients demonstrate some qualitatively distinct features in their impaired naming. Consider first patients with AD. Only lexical retrieval showed a statistically significant correlation with naming difficulty in the AD patients participating in this study. This differed from the pattern seen in FTD and CBD, where additional factors appeared to contribute to their naming deficit. Although some work has related semantic memory to impaired naming in AD (Huff et al., 1986; Hodges et al., 1996; Lambon Ralph et al., 1997), we may not have observed a naming-semantic correlation in these AD patients because of their degree of dementia and heterogeneity in semantic functioning within the AD group (Grossman et al., 1996a, 1997). We adopted the strategy of studying AD patients with mild-to-moderate dementia—even though this limits the ability to generalize our findings across the entire spectrum of AD severity—to allow matching for severity with groups of FTD and CBD patients. With this measure of semantic memory, we also minimize the potential confounds associated with task-related resource demands. In FTD, by comparison, naming accuracy appeared to correlate with lexical retrieval as well as with the semantic component of naming. This has been seen in other work (Hodges et al., 1995; Lambon Ralph et al., 2001; P. Moore, K. Dennis and M. Grossman, unpublished data). Naming in CBD correlated with lexical retrieval and performance on a measure of visual perceptual–spatial functioning (P. Moore, K. Dennis and M. Grossman, unpublished data). These findings suggest that impaired naming may have qualitatively distinct features in AD, FTD and CBD. While the measures of lexical retrieval, semantic and visual perceptual–spatial functioning were selected to reflect each of these components of naming, our observations may be limited by the fact that each was assessed by a single measure and thus may not capture its full scope in the naming process. Moreover, the numbers of patients we examined in each group was relatively small. This limits the statistical treatments of the data, suggests caution in generalizing our findings and emphasizes the importance of additional work with larger and more varied groups of these patients. With these caveats in mind, our observations are consistent with the hypothesis that a deficit in lexical retrieval and associated downstream processes is common to all patients suffering from AD, FTD or CBD, and that this component of naming may play a role in the naming difficulty that is shared by these patients. Further, the behavioural findings suggest some unique aspects of impaired naming in each of these neurodegenerative diseases, consistent with the hypothesis that a large-scale neural network for naming is interrupted in several distinct ways in these patients.
Our observations also confirm a statistically significant naming deficit in SD, PNFA and NON-APH subgroups of patients with FTD. While clinical observations and empirical studies emphasize the profound naming impairment in patients with SD (Snowden et al., 1989; Hodges et al., 1992a; Neary et al., 1998), naming difficulty is also evident in PNFA and NON-APH patients (Weintraub et al., 1990; Hodges and Patterson, 1996; Thompson et al., 1997; Croot et al., 1998; P. Moore, K. Dennis and M. Grossman, unpublished data). SD patients showed the greatest naming impairment and almost all the individual SD patients had a significant naming deficit. However, we were not able to confirm a statistically greater deficit in SD than other patients with FTD due to the relatively small number of patients participating in this study.
We also examined whether the naming deficit in FTD subgroups is due to an impairment of a single critical component of the naming process or to the interruption of different components of a large-scale neural network underlying naming. SD, PNFA and NON-APH patients all showed a correlation with lexical retrieval, suggesting that a single component of naming is compromised across all three FTD subgroups. Previous work has also emphasized the contribution of other cognitive components to impaired naming in FTD subgroups. For example, patients with SD have semantic memory impairments that appear to play a role in their naming deficit (Hodges et al., 1995; Lambon Ralph et al., 1998, 2001; P. Moore, K. Dennis and M. Grossman, unpublished data). We may not have observed this because the SD patients participating in the present study were early in the course of their disease and had a deficit in understanding occasional words that was not sufficiently extensive to be detected by a standard, group-based, semantic protocol (Lambon Ralph et al., 2001). Semantic difficulty may also be present in PNFA and NON-APH patients—as manifested in their respective deficits in verb knowledge and social knowledge (Bak et al., 2001; Grossman et al., 2003; Wood and Grafman, 2003; P. Moore, K. Dennis and M. Grossman, unpublished data). In this context, we may have seen a correlation of naming with semantic memory across all FTD patients, but not in each subgroup either due to the relatively small number of participants in each subgroup or because of the different kinds of semantic memory that may be compromised in these subgroups. The deficit in NON-APH patients may depend to some extent on their pattern of non-aphasic cognitive impairment, such as the presence of distractibility and other non-specific test-taking issues unrelated to the naming process per se (Rahman et al., 1999; Perry and Hodges, 2000; Rosen et al., 2002a). PNFA patients also appear to have some executive deficits that may limit their naming (Rhee et al., 2001). Finally, it is important to emphasize that lexical retrieval itself is a complex process involving multiple components. Despite the overwhelmingly common occurrence of a specific kind of naming deficit, i.e. failure to retrieve a word, which superficially appears to be identical in all groups of patients, the neurodegenerative process may interrupt different aspects of retrieval and expression in a large-scale neural network for naming in subgroups of FTD patients. This possibility can be examined by looking for different cognitive-cortical correlative patterns in patients with superficially similar profiles of impaired naming. We discuss below cortical atrophy in neurodegenerative diseases and the relationship of this atrophy to their naming deficit.
Correlations of naming difficulty with cortical atrophy: a single component of naming compromised across all patient groups
The present study sought to take advantage of partially shared and partially distinct patterns of naming difficulty across neurodegenerative diseases to learn about the neural basis for confrontation naming. In particular, we tested the hypothesis that qualitative similarities in language and cognition related to naming are due in part to patterns of cortical atrophy shared across AD and FTD, while qualitative differences in the language and cognitive components of naming are also associated with the partially distinct patterns of cortical atrophy seen in AD and FTD. We add to these observations the first quantitative study of grey matter atrophy in CBD. While an important advantage of this approach is the prominent anomia in these patients, one shortcoming is the absence of histopathologically confirmed diagnosis. Clinical diagnosis is a marker for regional cortical abnormality that interferes with the naming process, and additional work will be needed in the future to relate naming deficits more directly to histopathological abnormalities.
We used VBM analyses of structural MRI to identify the neuroanatomical distribution of grey matter atrophy in neurodegenerative diseases. Although other techniques are available, imaging studies of regional cortical atrophy have generally adopted one of two approaches: (i) a region-of-interest (ROI) analysis; or (ii) VBM. Few direct comparisons of VBM and ROI approaches have been published (Good et al., 2002), and discrepancies may be due to several differences between techniques that emphasize their relative advantages and disadvantages. The VBM technique does not allow the careful preservation of gyral and sulcal patterns of individual patients which can be achieved with an ROI approach, even though there is significant variability in the sulcal patterns across individuals that may be difficult to interpret with the ROI approach (Steinmetz and Seitz, 1991). Careful histological analyses also underline the poor correspondence between gross sulcal anatomy and microscopically-defined architectonic boundaries (Amunts et al., 1999). These individual anatomical features are blurred in VBM analyses that normalize each brain to a template and smooth the images in preparation for statistical analysis of group data. Although labour-intensive and time-consuming, the ROI technique is less reliable than a computer-based, fully-automated VBM approach. There are potential problems associated with the automatic identification of tissue types during extraction of the brain from the skull prior to segmentation (Good et al., 2001; Karas et al., 2003). The brain to which experimental subjects are normalized can be vexing: since the most widely used average brain uses young adults (Evans et al., 1993), greater deformation is required when studying the brains of elderly subjects. Potential solutions include using a ‘local’ template composed of the participants in a study, normalizing with a high dimensional algorithm, or implementing a deformation-based approach to normalization (Gee and Haynor, 1999; Good et al., 2001; Karas et al., 2003)—although these are not problem-free,. Other shortcomings of VBM have been detailed elsewhere (Ashburner and Friston, 2000; Baron et al., 2001; Good et al., 2001; Karas et al., 2003). We attempted to minimize some of these issues by: (i) using age-matched seniors as a reference group; (ii) by inspecting each slice of each segmented image of each patient for inaccurately labelled voxels; and (iii) by performing direct comparisons of cortical atrophy across age-matched patient groups that take into account a common reference group. A final issue is that cortical atrophy may not always reflect the full extent of functional cortical disease identified by other techniques such as PET (Chételat et al., 2003). Additional work is needed to determine whether discrepancies between techniques reflect equally necessary components of a large-scale neural network supporting a cognitive function.
With these caveats in mind, the results of the present study are consistent with the claim that naming difficulty is due in part to the interruption of a single neural component across all groups of patients. For example, we found an area of cortical atrophy that is common to AD, FTD and CBD, i.e. the lateral temporal cortex in the left hemisphere. A very small area of the right inferolateral temporal cortex also shared atrophy across groups. Dorsolateral prefrontal regions of the left hemisphere of these patient groups with cortical atrophy were adjacent to each other, although there was no overlap across all three groups.
The observation of significant atrophy on its own permits only very limited inferences about brain–behaviour relationships in neurodegenerative diseases. Demented patients have many different kinds of cognitive deficits, and any of these may be related to an area of cortical atrophy. We therefore performed direct correlations between naming performance and cortical volume. We found that confrontation naming correlates with cortical volume in the left lateral temporal lobe of patients with AD, FTD and CBD. This is somewhat consistent with previous work showing a correlation between confrontation naming and the left anterior temporal cortex in a combined group of AD patients and SD patients (Galton et al., 2001). To confirm that a specific anatomical distribution of correlation between naming and cortical volume contributes to the naming impairment of these patients, we constrained our inferential reasoning further by requiring that the correlation correspond to a region of significant cortical atrophy. With this constraint, we observed naming-cortical volume correlations in a left lateral temporal distribution in each patient group which was also significantly atrophic. Based on the observations that (i) naming difficulty correlated with impaired lexical retrieval in groups of patients with AD, FTD and CBD, (ii) there was a correlation between naming and cortical volume in left lateral temporal cortex in all three groups of patients and (iii) left temporal cortex was atrophic in the area of significant correlation in all of these patients, it seems reasonable to infer that this left lateral temporal region contributes to naming in AD, FTD and CBD. Since lexical retrieval appears to be the only component associated with naming in all three patient groups, it is likely that left lateral temporal cortex plays a role in the lexical retrieval component of naming.
Much work has associated naming with the left temporal lobe. Naming difficulty has been observed following focal ischaemic insult to the left temporal lobe (Benson, 1979; Kohn and Goodglass, 1985; Goodglass, 1993). Functional neuroimaging studies in healthy adults have shown activation of the left temporal cortex during naming (Howard et al., 1992; Mummery et al., 1998; van Turennout et al., 2000; Whatmough et al., 2002; Burgund et al., 2003). One PET study demonstrated activation in healthy subjects during naming of the left lateral temporal cortex, which was in the same anatomical distribution as patients with structural insult causing a lexical retrieval deficit during naming (Damasio et al., 1996). Based on the observation that semantic memory was preserved in their patients, these investigators hypothesized that the left lateral temporal cortex serves as an interface between the neural representation of a concept and the abstract representation of its name. The left lateral temporal area in this study is very similar to the anatomical distribution of the correlation we observed. It is also possible to speculate that the inferolateral temporal cortex of the right hemisphere contributed to the lexical retrieval component of naming difficulty across these patients, although the very small volume of tissue implicated in this region should be borne in mind. The right temporal cortex has been associated with sparse representations of word meaning (Beeman et al., 1994) and may contribute to retrieval by helping to limit the scope of the search for a target concept. Regardless of the contribution to naming played by lateral temporal cortex, these findings suggest that the impairment of one component of naming is likely to be shared across patients with AD, FTD and CBD.
Correlations between naming and cortical atrophy: multiple interruptions of a large-scale neural network result in partially distinct patterns of impaired naming in AD, FTD and CBD
While the observations described above lend support to the ‘single component’ hypothesis of impaired naming, several findings suggest that this may not fully explain naming difficulty in these patients. Consider in more detail AD patients, where impaired naming correlated only with lexical retrieval. We found a significant correlation between naming and cortical volume in the left lateral temporal cortex in AD as well as in FTD, and a direct comparison showed that AD patients have significantly greater atrophy in the left lateral temporal cortex than FTD patients. Other imaging and autopsy work has shown significant atrophy in this anatomical distribution in AD as well (Grady et al., 1988; Arnold et al., 1991; Haxby et al., 1990; Johnson et al., 1993; Karbe et al., 1994; Alsop et al., 2000; Baron et al., 2001; Busatto et al., 2003). The ‘single component’ hypothesis would predict that naming should be more impaired in AD than FTD since left lateral temporal atrophy is greater in AD than in FTD. In fact, this was not found: naming difficulty was statistically equivalent in AD and FTD. Similarly, the lexical retrieval deficit was equivalent statistically across these patient groups. While the left lateral temporal cortex may contribute to the lexical retrieval component of naming, the severity of disease in this area does not appear to predict the relative severity of the impairment in naming or lexical retrieval.
Additional evidence for a multifactorial approach to naming comes from the observation of different behavioural correlation profiles in FTD and CBD, and the distinct patterns of significant cortical atrophy across these patient groups. Consider in this context patients with FTD. The pattern of correlations between naming and cortical atrophy appears to reflect a combination of the profiles seen across FTD subgroups. SD, PNFA and NON-APH patients all showed a significant correlation between confrontation naming accuracy and lexical retrieval, although correlations between naming and cortical volume showed a distinct profile in each subgroup. This observation suggests that a single neural locus for all aspects of lexical retrieval and its associated phonological assembly and articulatory processes is unlikely. Instead, different parts of a large-scale neural network appear to be implicated in different aspects of lexical retrieval during naming. The lexical retrieval component of naming is known to be quite complex. Retrieval may involve selecting the correct name from among many possible choices in semantic memory that are equally accurate because they label objects with overlapping features. This appears to be a component of retrieval associated with the left lateral temporal cortex that is compromised across AD, FTD and CBD groups (as noted above). Other components of retrieval include, but are not limited to, inhibiting names that share many phonological features but do not name the target object, translating a material-neutral representation of the name into a material-specific form that can be expressed in a specific modality, and assembling phonological or graphemic comp