CORTEX1427_proof ■ 13 April 2015 ■ 1/12 c o r t e x x x x ( 2 0 1 5 ) 1 e1 2

Available online at www.sciencedirect.com 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64

ScienceDirect Journal homepage: www.elsevier.com/locate/cortex

Research report

Q4

Q3

Is the logopenic-variant of primary progressive aphasia a unitary disorder? Cristian E. Leyton a,b,c,*, John R. Hodges b,c,d, Olivier Piguet b,c,d, Catriona A. McLean e, Jillian J. Kril f and Kirrie J. Ballard a,b a

Faculty of Health Sciences, The University of Sydney, Lidcome, NSW, Australia Neuroscience Research Australia, Randwick, NSW, Australia c ARC Centre of Excellence in Cognition and its Disorders, Sydney, NSW, Australia d School of Medical Sciences, The University of New South Wales, Sydney, NSW, Australia e Department of Anatomical Pathology, Alfred Hospital, Melbourne, VIC, Australia f Department of Pathology, Sydney Medical School, The University of Sydney, Sydney, NSW, Australia b

article info

abstract

Article history:

Logopenic progressive aphasia is one of the clinical presentations of primary progressive

Received 11 November 2014

aphasia and formally defined by the co-occurrence of impaired naming and sentence

Reviewed 19 January 2015

repetition. Impaired naming is attributed to failure of lexical retrieval, which is a multi-

Revised 30 January 2015

staged process subserved by anatomically segregated brain regions. By dissecting the

Accepted 17 March 2015

neurocognitive processes involved in impaired naming, we aimed to disentangle the clinical

Action editor Peter Garrard

and neuroanatomical heterogeneity of this syndrome. Twenty-one individuals (66.7% fe-

Published online xxx

males, age range 53e83 years) who fulfilled diagnostic criteria for logopenic variant and had at least two clinical and language assessments, 1 year apart, were recruited and matched for

Keywords:

age, sex distribution and level of education with a healthy control sample (n ¼ 18). All

Primary progressive aphasia

participants underwent a structural brain scan at the first visit and surface-wise statistical

Logopenic variant of primary pro-

analysis using Freesurfer. Seventeen participants with logopenic variant underwent amy-

gressive aphasia

loid imaging, with 14 demonstrating high amyloid retention. Based on their performance on

Alzheimer's disease

single-word comprehension, repetition and confrontation naming, three subgroups of

Anomia

logopenic cases with distinctive linguistic profiles and distribution of atrophy were identified. The first subgroup (n ¼ 10) demonstrated pure anomia and left-sided atrophy in the posterior inferior parietal lobule and lateral temporal cortex. The second subgroup (n ¼ 6), presented additional mild deficits in single-word comprehension, and also exhibited bilateral thinning of the fusiform gyri. The third subgroup (n ¼ 5) showed additional impaired single-word repetition, and cortical thinning focused on the left superior temporal gyrus. The subgroups differed in the proportion of cases with high amyloid retention and in the rate of decline of naming performance over time, suggesting that neurodegeneration spreads differentially throughout regions subserving word processing. In line with previous reports, these results confirm the extensive damage to the language network and, in part, explain the clinical heterogeneity observed across logopenic cases. © 2015 Published by Elsevier Ltd.

* Corresponding author. Faculty of Health Sciences, The University of Sydney, 75 East St. Libcombe, NSW, 2141, Australia. E-mail addresses: [email protected] (C.E. Leyton), [email protected] (J.R. Hodges), [email protected] (O. Piguet), [email protected] (C.A. McLean), [email protected] (J.J. Kril), [email protected] (K.J. Ballard). http://dx.doi.org/10.1016/j.cortex.2015.03.011 0010-9452/© 2015 Published by Elsevier Ltd.

Please cite this article in press as: Leyton, C. E., et al., Is the logopenic-variant of primary progressive aphasia a unitary disorder?, Cortex (2015), http://dx.doi.org/10.1016/j.cortex.2015.03.011

65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129

CORTEX1427_proof ■ 13 April 2015 ■ 2/12

2 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

1.

c o r t e x x x x ( 2 0 1 5 ) 1 e1 2

Introduction

The linguistic profile of the logopenic variant of primary progressive aphasia (lv-PPA) is defined by the co-occurrence of anomia, word-finding difficulties and impaired sentence repetition (Gorno-Tempini et al., 2011). These deficits result from the breakdown of several cognitive processes, including verbal short-term memory (Gorno-Tempini et al., 2008), lexical retrieval (Leyton, Piguet, Savage, Burrell, & Hodges, 2012) and phonological processing (Bonner & Grossman, 2012; Brambati, Ogar, Neuhaus, Miller, & Gorno-Tempini, 2009; Goll et al., 2011; Leyton & Hodges, 2013). Reflecting the involvement of these cognitive processes, the distribution of cortical atrophy in lv-PPA comprises an extensive swathe of the left hemisphere which underpins most of the components of the language network (Leyton et al., 2012; Mesulam, Wieneke, Thompson, Rogalski, & Weintraub, 2012; Rohrer et al., 2010). Although atrophy of the left parietal-temporal junction represents the anatomical signature of lv-PPA, the extent of brain atrophy varies considerably from case to case, pointing to the presence of lv-PPA endophenotypes with slightly different clinical profiles, disease severity and decline over time (Leyton, Ballard, Piguet, & Hodges, 2014; Machulda et al., 2013). Accordingly, although the vast majority of cases with lv-PPA have Alzheimer's disease (Chare et al., 2014; Harris et al., 2013; Mesulam et al., 2014; Rohrer, Rossor, & Warren, 2012), this pathology demonstrates variable neuroanatomical extension which results in heterogeneous clinical presentations (Warren, Fletcher, & Golden, 2012). This clinical and anatomical heterogeneity opens the possibility that specific cognitive processes are predominantly damaged in some cases, but not in others. In this sense, the investigation of confrontation naming provides a suitable paradigm to explore the diversity of cognitive deficits in lvPPA, as this task relies on the integration of separate, albeit interactive, steps requiring multiple cognitive processes anatomically segregated (Damasio, Tranel, Grabowski, Adolphs, & Damasio, 2004; DeLeon et al., 2007). Models of lexical production specify a number of discrete stages (Dell & O'Seaghdha, 1992; Levelt, 2001), each of which can be separately damaged and result in impaired naming. In the first stage, the item to be named should be recognised, for which the integrity and access to semantic representations are crucial. Failure at this stage, as in the semantic variant of PPA, not only results in profound anomia, but also in failure on object recognition and word-comprehension tasks. In the intermediate stage, referred to as lexical retrieval, the semantic representation is linked to its arbitrary phonological word form. In other words, although the item can be recognised, the specific target word is not yet retrieved. At the final, or postlexical, stage the phonological information is temporarily stored in the phonological buffer in order to execute the motor plan of the intended utterance. Consequently, failure at this level can result in marked difficulties with repetition, particularly for multisyllabic words and long sentences (Leyton, Savage, et al., 2014). Given that diagnostic criteria for lv-PPA explicitly exclude significant impairments in single-word comprehension, object knowledge or motor aspects of speech, it can be presumed

that the main mechanism underlying anomia in lv-PPA is impaired retrieval of the phonological form. Nevertheless, given the broad extension of pathological changes in lv-PPA (Leyton et al., 2012; Rohrer et al., 2010; Teichmann et al., 2013) over most of the left-sided language regions involved in semantic and lexical processing (Indefrey & Levelt, 2004; Price, Devlin, Moore, Morton, & Laird, 2005), it is possible that other stages of naming processing are also compromised. On these grounds, the concurrent analysis of performance across a range of single-word tasks and structural neuroimaging measures can not only contribute to decipher the clinical and neuroanatomical heterogeneity of the logopenic syndrome, but also reveal sub-groups with distinctive neurobiological features and prognosis. Furthermore, information related to the preservation and involvement of various linguistic components can provide the rational basis for planning and implementing behavioural interventions (Best et al., 2013). The aim of this study was to ascertain the neurocognitive processes involved in impaired naming in lv-PPA by examining patterns of performance on single-word processing tasks and naming errors. Accordingly, we hypothesised that this set of tasks would allow the identification of coherent clinical subgroups with distinctive patterns of brain atrophy and neurobiological behaviour. As such, a second aim was to observe the clinical progression of lv-PPA over consecutive assessments and infer the proportion of cases with Alzheimer pathology in each group.

2.

Material and methods

2.1.

Participants

Twenty-one lv-PPA participants having at least two separate assessment sessions were recruited between 2007 and 2013 through the FRONTIER frontotemporal dementia clinical research group in Sydney Australia. Participants with limited English proficiency (high proficiency was assumed for those who had English as a second language but had lived and worked in an English speaking country for over 10 years) or with concomitant motor neuron disease, significant extrapyramidal features, past history of stroke, epilepsy, alcoholism, or significant traumatic brain injury were excluded from the study. All participants underwent a complete neurological evaluation, a routine neuropsychological assessment, and structural brain MRI. The clinical diagnosis of lv-PPA was retrospectively conducted at baseline assessment using a clinical protocol previously described (Leyton & Hodges, 2014; Leyton et al., 2011) and based on the current International Consensus recommendations (Gorno-Tempini et al., 2011). As such, PPA cases with anomia, word finding difficulties and impaired sentence repetition in absence of apraxia of speech, frank agrammatism or dissolution of semantic knowledge were included in the study. Cases with impaired naming and mild single-word comprehension in absence of other evidence of semantic involvement, but impaired sentence repetition were also classified as lv-PPA. This profile contrasted with that

Please cite this article in press as: Leyton, C. E., et al., Is the logopenic-variant of primary progressive aphasia a unitary disorder?, Cortex (2015), http://dx.doi.org/10.1016/j.cortex.2015.03.011

66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130

CORTEX1427_proof ■ 13 April 2015 ■ 3/12 c o r t e x x x x ( 2 0 1 5 ) 1 e1 2

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

of individuals diagnosed with the semantic variant of PPA. These cases not only experienced severe naming deficits, but also displayed marked single-word comprehension deficits accompanied by other evidence of semantic involvement, such as impaired object knowledge and surface dyslexia. Similarly, PPA cases with evidence of apraxia of speech, frank agrammatism or both deficits were diagnosed as non-fluent/ agrammatic variant of PPA (Gorno-Tempini et al., 2011). Seventeen of the 21 lv-PPA participants underwent a PiB-PET scan (Klunk et al., 2004) and those cases with a standardised uptake value ratio for neocortical PiB equal or higher than 1.5 were regarded as having high burden of amyloid brain denoting the presence of Alzheimer pathology (Rowe et al., 2010). Healthy control subjects with no history or clinical evidence of major neurological or psychiatric disorder (n ¼ 13) were recruited from the research group's volunteer panel and matched according to the number of years of formal education and age at the moment of assessment. The study received approval from the South Eastern Sydney and Illawarra Area Health Service and the University of New South Wales human ethics committees.

2.2.

Q1

Cognitive status and functional ability assessment

All participants were administered the Addenbrooke's Cognitive Examination revised (ACE-R) (Mioshi, Dawson, Mitchell, Arnold, & Hodges, 2006) to estimate their current global cognitive ability. This screening cognitive measure comprises items that evaluate attention and orientation, memory, verbal fluency, language, and visuospatial abilities. Additional neuropsychological testing included the auditory-verbal digit span task (forward and backward) from the WAIS-III (Wechsler, 1997), ReyeOsterrieth Complex Figure (Osterrieth, 1945; Rey, 1941), the Trail Making Test A and B (Reitan, 1955). A sentence repetition task selected from the Multilingual Aphasia Examination (Benton & Hamsher, 1989), in which participants have to repeat one-by-one 14 sentences of increasing length from 3 to 18 words, was also administrated. In addition, functional ability was determined using a carerbased questionnaire, the Frontotemporal Dementia Rating Scale (Mioshi, Hsieh, Savage, Hornberger, & Hodges, 2010).

2.3.

Single-word tasks

The Sydney Language Battery (Savage et al., 2013) consists of four single-word tasks using the same set of stimuli: visual confrontation naming, repetition of multisyllabic words, word comprehension, and semantic association. Visual confrontation naming, which is always administrated first, requires the participants to name 30 different colour photographs presented one at a time. These include 12 living and 18 nonliving semantic-category items, matched by word frequency and number of phonemes per word. The word repetition task requires the participant to listen and repeat the same 30 multisyllabic words one after the other. Any phonological error, pausing or re-starts were considered abnormal singleword repetition and coded as phonological or nonphonological errors, respectively. Single-word comprehension (word-picture matching), which is always administered

3

before the semantic association task (see below), is assessed by asking the participant to point to the picture that best matches the word spoken by the examiner in an array or photographs containing the target item and six semantically related or visually similar foils. Finally, the semantic association task (pictureepicture matching) requires the participant to select the picture most closely associated with the target picture from a set of four options. The four options are semantically related to each other, but only one option is semantically related to the target (e.g., strawberry: cream, butter, cheese, milk). While the same pictures of the target items are used in the naming and in semantic association subtests, alternative versions are used in the word comprehension task, so that participants cannot identify the item based on visual memory.

2.4.

Analysis of naming errors

For the naming task, any marked hesitancy, phonological or semantic substitutions as well as omissions or ‘don't know’ responses were scored as errors. The classification of errors was conducted by a rater (C.E.L.) and based on the procedures described in Hodges, Graham, and Patterson (1995). In case of uncertain classification, errors were reviewed and reached consensus by second raters (K.J.B. and J.R.H.) Phonological errors and marked fragmentation of the utterance were coded as “sub-lexical” naming errors. Co-ordinate substitutions (e.g., rhinoceros for hippopotamus) and superordinate substitutions (e.g., animal for giraffe) were considered “semantic” errors. Approximate responses, circumlocutions or coherent explanations related to the item to be named, but not part of the aforementioned semantic errors, were considered as “lexical retrieval” errors. Examples of these errors include expressions like ‘pyramids and all that’ referring to hieroglyphic; ‘prehistoric animal’ or ‘they don't exist any more’ to dinosaur; and ‘something for cooking, you can eat’ referring to potatoes. Abandoned testing, empty or unrelated responses (e.g., ‘I've seen it before’, ‘I know it’); and incomplete, incomprehensible, or ‘I don't know’ responses were considered as ‘non-responses’.

2.5.

Statistical analysis

Statistical analyses were undertaken using SPSS 20.0 (IBM Corporation).

2.5.1.

Baseline performances

Given the non-normal distribution of scores on several tasks in the healthy control group due to ceiling effects, nonparametric ManneWhitney U tests were used to compare performances of continuous variables between lv-PPA and healthy controls. Chi square test was used to estimate differences in the distribution of categorical variables.

2.5.2.

Cluster analysis

A two-step cluster analysis was conducted to identify lv-PPA sub-groups based on their performance on single-word tasks. To select the single-word tasks to be entered to the cluster analysis, a preliminary analysis of covariance was conducted using naming performance as the dependent

Please cite this article in press as: Leyton, C. E., et al., Is the logopenic-variant of primary progressive aphasia a unitary disorder?, Cortex (2015), http://dx.doi.org/10.1016/j.cortex.2015.03.011

66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130

CORTEX1427_proof ■ 13 April 2015 ■ 4/12

4 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

c o r t e x x x x ( 2 0 1 5 ) 1 e1 2

variable and all the other three single-word tasks as independent variables. This analysis revealed that single-word comprehension (F(1,17) ¼ 7.5, p ¼ .014) and single-word repetition (F(1,17) ¼ 4.7, p ¼ .045), but not the semantic association task (F(1,17) ¼ 2.0, p ¼ .17), were significantly associated with naming performance. As a result, naming, single-word repetition and single-word comprehension performances were entered to an exploratory hierarchical cluster method to determine a priori the number of partitions in the sample. In this analysis, squared Euclidean distances were used as a proximity measure, and the Ward's method used as a clustering algorithm. The number of clusters to be used was determined by visual inspection of the dendrogram and then used as input for the k-means analysis. Given the iterative nature of this clustering method, this outcome was matched with the initial clustering solution and used as confirmatory method (Lange, Iverson, Senior, & Chelune, 2002). One-way analysis of variance (ANOVA) tests were carried out to estimate differences across resultant clusters for the relevant continuous variables. In addition, three separate repeatedmeasures ANOVAs were conducted to analyse naming performances on specific semantic-category, to compare frequency of type of naming errors, and to compare type of errors in single-word repetition across clusters. A non-significant test of Mauchly was assumed as sphericity; in case of violation of this assumption, degrees of freedom were adjusted using the Greenhouse-Geisser method. Overall significant differences in the models were further explored using post hoc pair-wise comparisons with Bonferroni corrections. The level of statistical significant was set at .05 unless stated otherwise.

2.5.3.

Longitudinal changes

To examine the rate of decline over time in each resultant cluster, performances on total ACE-R and each of the four single-word tasks were selected as outcome variables. Linear mixed-effect models (Laird & Ware, 1982) were used to evaluate performances on the selected outcome variables over time. Fixed effects included the clustering group, followup time, and the interaction of both effects. Patient's individual variability at baseline was the only random effect included (where we used a random intercept model). The variability of any estimated parameters was determined by the fixed and random components in the model. A significant effect of follow-up time would indicate that the outcome variable changes linearly with time, while a significant interaction between clustering group and follow-up time would indicate that the rate of change of the outcome variable differed across clustering cohorts. Residual errors of the models were assumed to be normally distributed, as were the random intercepts for the subjects' baseline response. All participants were assumed to be independent.

2.6. Imaging acquisition and cortical thickness calculation Whole-brain T1-weighted images were acquired for all participants at the baseline assessment using a 3T Philips MRI scanner with standard quadrature head coil (eight channels). The 3D T1-weighted images were acquired as follows: coronal

orientation, matrix 256  256, 200 slices, 1 mm in-plane resolution, slice thickness 1 mm, echo time/repetition time ¼ 2.6/ 5.8 msec, flip angle a ¼ 19 . To estimate cortical thickness, T1 images were processed using Freesurfer version 5.1 (http://surfer.nmr.mgh.harvard. edu/) (Fischl & Dale, 2000), following the pipeline described elsewhere (Leyton, Ballard, et al., 2014). The identification of cortical areas was conducted using automatic parcellation of the cerebral cortex (Destrieux, Fischl, Dale, & Halgren, 2010). The calculated cortical thickness was smoothed with a 20 mm full-width at half height Gaussian kernel. This level of blurring kernel was chosen to reduce the impact of imperfect alignment between cortices and thereby improving the signal-tonoise ratio (Lerch & Evans, 2005). Imaging statistical analysis was performed vertex-byvertex using general linear models to examine differences between each resultant lv-PPA cluster and the healthy control sample. A false discovery rate (Genovese, Lazar, & Nichols, 2002) was set at .001 to adjust p values for multiple comparisons. p values for group comparison analyses were mapped onto the inflated cortical surface representation of an average brain. Effect sizes for each cluster were also calculated according to the formula by Cohen, in line with previous studies (Hilti et al., 2013).

3.

Results

3.1. Demographic, cognitive and single-word tasks at baseline The average age of the lv-PPA cohort at baseline assessment was 66.9 ± 7.6 years, with estimated symptom duration of 3.5 ± 2.2 years, and 13.2 ± 3.6 years of formal education. The control sample was matched in age (67.7 ± 4.4; t(37) ¼ .4, p ¼ .69), education (12.7 ± 2.4; t(37) ¼ .5, p ¼ .61) and sex distribution (both samples had 66.7% of females). The lv-PPA sample performed lower than the healthy control sample on all cognitive tests and on total scores for each of the four single-word tasks (Table 1). An analysis of naming performance for living versus nonliving items revealed that lv-PPA named living items (53.6%) better than nonliving items (44.2%) (two-tailed, paired t(20) ¼ 3.1, p ¼ .006). Naming error analysis revealed that almost half (48.2%) of all incorrect responses on picture naming were omissions or abandoned test, followed by semantic errors (25.9%), lexical retrieval breakdown (14.0%) and post-lexical impairments (11.9%).

3.2.

Cluster analysis

The resultant dendrogram demonstrated three clusters made up of 10, 6 and 5 cases (Fig. 1). The K-means clustering analysis demonstrated complete congruence with the initial 3-cluster solution. The analyses of variance revealed no differences in demographic features or length of symptoms across cluster (Table 2). Except for ACE-R performance, which was higher in Cluster 1, none of the other cognitive tasks were significantly different across clusters. All clusters had low performance on sentence repetition, but post hoc non-

Please cite this article in press as: Leyton, C. E., et al., Is the logopenic-variant of primary progressive aphasia a unitary disorder?, Cortex (2015), http://dx.doi.org/10.1016/j.cortex.2015.03.011

66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130

CORTEX1427_proof ■ 13 April 2015 ■ 5/12

5

c o r t e x x x x ( 2 0 1 5 ) 1 e1 2

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

Table 1 e Neuropsychological performance in healthy controls and lv-PPA patients. lv-PPA (n ¼ 21)

Healthy controls (n ¼ 18)

ACE-R (/100) Digit span Forwards Digit span Backwards RCF Copy (/36) RCF Recall (/36) TMT A (sec) TMT B (sec) Single-word tasks Naming (/30) Comprehension (/30) Repetition (/30) Semantic association (/30)

ManneWhitney U

Mean

SD

Range

Mean

SD

Range

z

p

96.7 11.9 8.3 32.7 18.7 28.8 73.0

±2.5 ±2.0 ±2.4 ±3.1 ±6.7 ±9.0 ±23.2

[91e100] [8e14] [6e13] [25e36] [7e35] [13e50] [36e112]

65.6 6.3 3.2 24.4 6.6 73.0 260.9

±11.1 ±2.2 ±1.4 ±9.1 ±5.4 ±64.9 ±141.4

[44e91] [2e10] [0e6] [5.5e35] [0e18.5] [24e300] [77e450]

5.32 5.05 5.29 3.64 4.41 3.68 4.73

Is the logopenic-variant of primary progressive aphasia a unitary disorder?

Logopenic progressive aphasia is one of the clinical presentations of primary progressive aphasia and formally defined by the co-occurrence of impaire...
1MB Sizes 1 Downloads 8 Views