Alzheimer's pathology in primary progressive aphasia

1Primary progressive aphasia (PPA) refers to a group of neurodegenerative disorders with language impairment as the initial symptom. These disorders are of high neurobiological and clinical importance because they illustrate the potentially focal nature of neurodegenerative disease and the potential heterogeneity of clinical presentations even where the underlying pathological process is uniform. The best characterized subtypes of PPA are progressive nonfluent aphasia (PNFA) and semantic dementia (SD). Patients with PNFA have nonfluent speech characterized by agrammatism and/or a motor speech impairment (usually an apraxia of speech, i.e., hesitancy and effortfulness attributable to impaired planning of articulation). SD presents with fluent aphasia, anomia, and single word comprehension deficits secondary to verbal semantic impairment. “Fluency” in this context refers to the flow of speech. However, dysfluency may arise from a variety of underlying deficits, including agrammatism, impaired articulation (motor deficits such as apraxia of speech), decreased phrase length or slower speech rate (e.g., due to word-finding pauses); patients referred to as having a “nonfluent aphasia” may have various more or less distinct primary language or speech impairments. This theme is well illustrated by the recently recognized entity of logopenic/phonological aphasia (LPA), which constitutes a third major syndrome within the PPA spectrum. Patients with LPA have word-finding pauses and anomia as well as impaired speech repetition, particularly sentences.

2Most cases of PPA have a non-Alzheimer pathological substrate within the frontotemporal lobar degeneration spectrum, and are usually associated predominantly with either tau- or TAR (trans-activation-response) DNA binding protein 43 (TDP-43)-positive cellular inclusions (known as FTLD-tau or FTLD-TDP pathology), respectively. However, it has long been recognized that PPA syndromes may also be associated with Alzheimer's disease (AD) pathology and in recent years more detailed series have been reported. In particular, recent evidence has suggested that LPA is underpinned by AD pathology in a high proportion of cases and may be the most common aphasia phenotype of AD. However both PNFA and SD have also been reported with AD pathology, as have syndromes that do not fit clearly into a single category, so-called “mixed” aphasia. As AD is the most common neurodegenerative disease of later life, the range of phenotypic variation in AD and the mechanisms that drive this variation are key issues in the field of neurodegenerative disease.

3Here we review the clinical, neuropsychological and cross-sectional neuroimaging features of a retrospective series of patients with a clinical diagnosis of PPA and AD pathology either demonstrated directly or presumed on the basis of cerebrospinal fluid (CSF) biomarker profiles. We consider these cases in relation to previously published series of PPA patients with either pathologically confirmed AD or a positive Pittsburgh compound B (PIB)-positron emission tomography (PET) scan suggestive of AD.

4From the Dementia Research Centre patient database comprising a consecutive series of patients seen between 1992 and 2008, we extracted all cases meeting criteria for PPA and who had either AD pathology at postmortem/cerebral biopsy or CSF biomarker data consistent with Alzheimer pathology (raised CSF total tau level with reduced amyloid Aβ42 fraction; ). In total, 14 patients were included in the series: 9 had pathologically confirmed Alzheimer's disease (7 who came to postmortem and 2 on cerebral biopsy) and 5 had CSF biomarkers consistent with AD (these 5 patients were previously reported in ). Clinical notes and neuropsychological data were reviewed, and the clinical diagnosis at the time the patient was initially assessed and a revised clinical diagnosis based on current descriptive criteria for PPA were recorded in each case. Neuropsychological data were also recorded where available. Ethical approval for the study was obtained from the National Hospital for Neurology and Neurosurgery Local Research Ethics Committee. Written research consent was obtained from all patients participating in the study.

5All subjects had been scanned on a 1.5 T GE Signa unit scanner (General Electric, Milwaukee, WI) with T1-weighted volumetric images obtained with a 24-cm field of view and 256 × 256 matrix to provide 124 contiguous 1.5-mm-thick slices in the coronal plane. Mean (standard deviation) age at scan was 60.2 (6.2) years. A control group of 23 age- and gender-matched cognitively normal subjects (mean age 63.5 [7.3] years at time of scan) was used for comparison. No subject had significant cerebrovascular disease or other secondary pathology on neuroimaging. Image analysis was performed using the MIDAS software package. A rapid, semiautomated technique of brain segmentation which involves interactive selection of thresholds, followed by a series of erosions and dilations was performed for each scan. This yielded a brain region which was separated from surrounding CSF, skull, and dura giving a baseline brain volume. Ventricles were also segmented within MIDAS. Scans and associated brain regions were initially transformed into standard space by registration to the Montreal Neurological Institute (MNI) Template. Left and right hemispheric regions were defined using the MNI average brain which was split by dividing the whole volume along a line coincident with the interhemispheric fissure. An intersection of each individual's brain region and the hemispheric regions defined on the MNI template was generated to provide a measure of brain volume in left and right hemispheres and left/right volume ratios were also calculated. The 2 disease groups and the healthy control group were compared statistically based on contrasts between the group means using a linear regression model in STATA10 (Stata Corporation, College Station, TX).

6We investigated changes in imaging patterns with severity using cortical reconstruction and thickness estimation methods with the Freesurfer image analysis suite (surfer.nmr.mgh.harvard.edu/) as previously described. We used performance on the Graded Naming Test (, total number of items equals 30) (i.e., degree of anomia) as a measure of disease severity, splitting the group according to their score: group 1 (less severe: 9 patients) scored > 0 (mean 7.7, standard deviation 9.2) and group 2 (more severe: 4 patients) were unable to score. One case (AD-PPA6) with greater right than left hemisphere atrophy was not included in this analysis; this atrophy profile might reflect either a different disease phenotype or reversed hemisphere language dominance, however inclusion of this case could potentially bias any group-level correlations between cortical thickness and disease severity.

7Effect size maps were generated based on the difference in mean thickness in each of these severity subgroups and in the whole group, comparing each to the controls and expressing the disease-control difference as a percentage of the mean control group thickness.

8Demographic and clinical data for patients are presented in Table 1; neuropsychological data (where available) are presented in Table 2. All patients had language impairment as their primary presenting feature. This was usually difficulty finding words although 1 patient complained of a return of a childhood stutter shortly before the onset of word-finding difficulties. Spontaneous speech was relatively nonfluent and occasional phonemic errors were made by all patients, with occasional emergence of neologistic jargon errors. None of the patients was described as having had apraxia of speech. All of the patients who came to postmortem or had a cerebral biopsy had initially received a diagnosis of PPA, PNFA, or language variant frontotemporal lobar degeneration although prior to death the diagnosis in 2 of these cases was changed to atypical language variant of AD. The 5 patients with CSF biomarkers consistent with AD were ascertained more recently and had been diagnosed with LPA before CSF analysis. On review of the clinical notes of the 7 patients who came to postmortem and the 2 patients with cerebral biopsy-proven AD, all would also have met criteria for LPA based on their initial symptoms and neurocognitive assessment. A family history of dementia was present in only 2 cases: these patients each had a single parent with a diagnosis of Alzheimer's disease in the eighth decade. Myoclonus was noted in 2 patients and 2 patients developed generalized seizures. One patient exhibited axial rigidity late in the course of the disease; no other features of parkinsonism or motor neuron disease were present in this series. Behavioral impairment was unusual early in the illness but aggression, anxiety, and irritability were noted in some patients later in the course.

9Although all patients had had an initial neurocognitive assessment, for many patients formal neuropsychological testing was only performed later in the illness (e.g., when AD-PPA4 was tested, Mini Mental State Examination [MMSE] score was 4/30 and he performed poorly across multiple domains). Consistent with a diagnosis of LPA, neuropsychological assessment showed severely impaired digit span in all but 3 patients, who scored in the low (but not defective) range. Naming was in the impaired range at initial assessment in over half of the patients and became impaired in all cases as the disease progressed, also consistent with LPA. Single-word comprehension was intact in 9 of 14 patients as has been described in LPA but impaired in the more severely affected patients (intact in those with MMSE 18 or above, impaired in those with an MMSE below 17). None of the patients complained of episodic memory impairment at presentation, however verbal memory was impaired in 8 of 11 patients tested while visual memory was affected less frequently (5 of 14 patients). Reading was affected in most patients and some were noted to have phonological dyslexia. Limb apraxia and dyscalculia were noted in most patients, however visuospatial skills were intact in all but 1 severely affected patient. Executive dysfunction was also seen in most patients.

10Six of the 7 patients who came to postmortem had severe Alzheimer pathology with Braak stage VI, and Consortium to Establish a Registry for Alzheimer's Disease (CERAD) frequent plaques (Table 1). For the seventh case, no staging information was available but this case had been reported as showing severe Alzheimer pathology with frequent plaques and tangles. Four cases were also noted to have cerebral amyloid angiopathy. The 2 patients who had cerebral biopsies were noted to have frequent amyloid plaques and neurofibrillary tangles.

11Volumetric magnetic resonance imaging (MRI) data for patients and controls are presented in Table 3. Whole brain and hemisphere volumes were smaller in patients and there was evidence of left/right hemispheric asymmetry at group level and in all but 1 of the individual patients; 1 (right-handed) patient showed reverse asymmetry. Asymmetry became more marked with increasing disease duration (Fig. 1, R = 0.55, p = 0.04).

12In the cortical thickness analysis versus healthy controls (Fig. 2), group 1 (with less severe disease) showed areas of cortical thinning predominantly in the left hemisphere, most marked in the inferior parietal and posterior superior temporal lobes. Other areas involved in the left hemisphere were posterior cingulate, precuneus, medial temporal lobe, and prefrontal cortex. In the right hemisphere, only the posterior cingulate and precuneus and a small area in the medial temporal lobe were affected. In group 2 with more severe anomia, cortical thinning remained asymmetrical but was more extensive within both hemispheres. In the left hemisphere there was additional involvement of anterior superior and middle temporal lobe, posterior medial temporal lobe, and inferior frontal lobe areas. In the right hemisphere there was involvement of areas similar to those initially involved in the left hemisphere, i.e., lateral parietal, posterior superior temporal, posterior cingulate, precuneus, medial temporal, and prefrontal cortices.

13Here we have described a series of 14 patients with PPA in association with proven or probable AD pathology. The key clinical features of the cases in this series were initial presentation with word-finding difficulty, and relatively nonfluent spontaneous speech with occasional phonemic errors but without motor speech impairment. “Word-finding difficulty”, like fluency, refers to a cluster of related deficits: though often related to anomia, patients with conversational pauses but with relatively intact naming may also present with a word-finding problem. Reviewing the diagnoses in this series revealed that all cases fulfilled (or would likely have fulfilled) descriptive criteria for LPA. The neuropsychological findings of impaired digit span, dyscalculia, limb apraxia, and phonological dyslexia were consistent with LPA. However, verbal memory, although not a presenting feature in any of the patients, was also affected in most cases. Although this feature has not been emphasized in some previous studies of LPA, in 1 previous series 5 of 6 patients were impaired on verbal memory tasks. In contrast, visuospatial processing (a right hemisphere function) was generally well preserved. Cross-sectional brain imaging revealed asymmetrical left-sided atrophy predominantly affecting the posterior superior temporal lobe and inferior parietal lobe but also the posterior cingulate, precuneus, and medial temporal lobe. These features corroborate previous neuroanatomical findings in LPA. In more severe disease there was evidence of atrophy spread to the left frontal lobe, more anterior left temporal lobe areas, as well as posterior superior temporal lobe, inferior parietal lobe, and posterior cingulate areas within the right hemisphere.

14The nosology of patients with language impairment and AD pathology remains controversial. Such patients have been classified either as having the symptomatic description of PPA (with an LPA phenotype in most cases) or having the predictive clinicopathological description of an atypical “language variant” within the AD spectrum. While there should not be conflict between these 2 descriptions as they are essentially at 2 different levels of classification, predicting which patients with a PPA syndrome will have AD pathology (particularly in the absence of a PIB-PET scan or CSF markers) is nevertheless often challenging during life. The extent of involvement of other cognitive domains may be helpful, however the present evidence suggests that the presence and severity of extralinguistic impairments depends on disease stage. Furthermore, the clinical salience of these additional impairments is variable: in this series, a number of patients that performed poorly on episodic memory tasks did not complain of amnestic symptoms, whereas 2 patients who came to postmortem exhibited widespread cognitive impairment prompting a reformulation of the clinical diagnosis as an atypical language variant of AD. We would argue that the presenting syndrome at an early disease stage is likely to provide the more rational basis for classifying language dysfunction associated with AD, particularly as language impairments are very common as “typical” AD advances. This distinction is clinically important, as recognition of PPA features that predict AD pathology could help direct the use of investigations such as CSF and PIB-PET, and ultimately, the selection of patients for clinical trials and disease-modifying therapies.

15Previous series from 5 research groups have reported PPA patients with either pathologically confirmed AD or a positive PIB-PET scan showing amyloid deposition (Table 4; ). Prior to the first detailed description of LPA, patients with both PNFA and SD were reported with AD pathology but since that time LPA has been the clinical syndrome most closely associated with AD pathology. In, all patients with LPA versus 1 of 6 patients with PNFA and 1 of 5 patients with SD had positive PIB-PET scans; in, 7 of 11 logopenic patients had AD pathology, compared with none of the 6 agrammatic patients, 3 of 5 of the “mixed” patients, and the single semantic patient. It is important to recognize that classification of PPA phenotypes generally depends on syndromic characterization, and overlap between syndromes is frequent, particularly with disease evolution (e.g., LPA overlaps both with PNFA and SD). It is unclear whether older series of PPA cases included patients that would now be described as having LPA, e.g., in many of the patients with PNFA were diagnosed before the initial description of LPA. In that study, 5 of 7 patients with a mixed aphasia (including LPA) patients had AD pathology, compared with 2 of 20 with SD and 12 of 26 with PNFA. Improved understanding of the specific disease phenotypes has refined clinicopathological correlations in PPA, e.g., patients with motor speech deficits (e.g., apraxia of speech) appear to show an association with FTLD-tau rather than AD pathology. For the clinical syndrome of SD there is an association chiefly with FTLD-TDP rather than AD pathology. The SD syndrome underpinned by AD may be associated with asymmetrical temporal lobe atrophy focused on the left hippocampus and superior temporal lobe, rather than the temporal pole and anteroinferior temporal lobe as in classical SD caused by FTLD-TDP pathology. More marked superior temporal lobe atrophy has been associated with LPA in other studies.

16An outstanding neurobiological question concerns the overlap of LPA/atypical language-presentation AD with typical amnestic AD (and with other atypical variants of AD such as posterior cortical atrophy). Neuropsychologically, there are few data to compare amnestic-onset AD with atypical language variants but studies of language impairment in typical AD have shown that patients can be logopenic with an early anomia, and that phonological and semantic impairments also occur. Motor speech impairment (apraxia of speech) has been reported only rarely in association with AD. From an anatomical perspective, LPA is associated with asymmetrical atrophy compared with the relatively symmetrical atrophy of amnestic AD. However, certain key areas of atrophy or cortical thinning are implicated in both LPA-AD and typical AD, i.e., the temporo-parietal junction, the precuneus, posterior cingulate, and the medial temporal lobe. One recent study has shown an overlap of patterns of atrophy in these areas in early onset amnestic AD, posterior cortical atrophy, and LPA. The present study has certain limitations, including relatively small patient numbers, retrospective ascertainment, and most importantly, lack of uniform histopathological confirmation. Taking these caveats into account, the present evidence in conjunction with previous work suggests that the LPA syndrome might be regarded, very broadly, as a “uni-hemispheric” presentation of AD. Further detailed longitudinal prospective studies comparing amnestic and language presentations of AD are needed to elucidate the pathophysiological mechanisms that instigate and sustain neuropsychological and anatomical asymmetry.

17Dr. Rohrer has received research support from Brain (Exit Scholarship). Dr. Rossor serves on a scientific advisory board for Elan Corporation and Wyeth; serves as Editor-in-Chief of the Journal of Neurology, Neurosurgery and Psychiatry, and on the editorial boards of Practical Neurology, Dementia and Geriatric Cognitive Disorders, Neurodegenerative Diseases, and the British Medical Journal; receives royalties from publishing Brain's Diseases of the Nervous System (11th Ed.), Oxford University Press (2001), and Brain's Diseases of the Nervous System (12th Ed.), Oxford University Press (2009); and receives research support from the Department of Health and the Alzheimer's Research Trust. Dr. Warren has received research support from the Wellcome Trust (Intermediate Clinical Fellowship).

18Ethics approval was obtained from the local ethics committee at the National Hospital for Neurology and Neurosurgery, London, UK. Written research consent was obtained from all patients participating in the study.