3: Atypical Alzheimer’s disease

Sharon J. Sha^1,2 and Gil D. Rabinovici¹

¹ University of California, San Francisco, San Francisco, CA, USA

² Stanford Center for Memory Disorders, Stanford, CA, USA

Introduction

Alzheimer’s disease (AD) is the most common pathologic cause of dementia [1]. Clinically, AD typically presents with early episodic memory loss and visuospatial dysfunction. Less prominent deficits in executive function, attention, and language are common as well. Behavioral disturbances such as psychosis do not typically occur until late disease stages [2]. It is increasingly recognized, however, that AD pathology can be found in patients with nonamnestic clinical presentations [3–5]. AD is the most common cause of posterior cortical atrophy (PCA) [6, 7] and is found to be the causative pathology in 20–50% of patients with corticobasal syndrome (CBS) [3, 8] and in 20–40% of patients with primary progressive aphasia (PPA) [9, 10], focal cortical syndromes that were initially postulated to be pathologically distinct from AD [11–13]. Identifying patients with atypical clinical syndromes who have underlying AD is important clinically as symptomatic therapies are available for AD, but not yet for other degenerative dementias, and disease-specific therapies for AD are on the horizon [14]. Whereas previous criteria for AD included obligatory decline in memory [15], the new criteria propose to include nonamnestic presentations as well [16, 17].

Epidemiology

AD affects 5.2 million people in the United States and 17 million people worldwide [18, 19]. The prevalence of AD is about 1% at age 60–65 and doubles every 5 years, approaching 40% in 85–90-year-olds. The prevalence of atypical presentations of AD is difficult to estimate. Nonamnestic presentations might account for up to 15% of patients seen in dementia referral centers [5]. The average age of onset in patients with atypical syndromes is typically in the 60s [6, 7, 20, 21], and it has been suggested that early age-of-onset AD (EOAD) patients (defined in most studies as under age 65 at symptom onset) are more likely to show nonamnestic presentations [22, 23]. It is not known whether patients with atypical presentations differ from typical patients in disease progression or survival, although rapidly progressive forms of AD recently have been recognized [24].

Diagnosis

In 1984, the National Institute of Neurological Disorders and Stroke–Alzheimer’s Disease and Related Disorders Association (NINCDS-ADRDA) working group formulated diagnostic criteria that remained in practice through 2011 (Table 3.1) [15]. The NINCDS-ADRDA criteria had several limitations. First, they were only about 70–80% sensitive and 70% specific compared to pathology [25, 26]. Furthermore, patients with AD who presented with atypical patterns of cognitive impairment often did not meet NINCDS-ADRDA criteria, which require memory impairment as one of the core features. [5] In addition, biomarkers such as molecular, functional, and structural imaging modalities, cerebrospinal fluid (CSF) evaluation, as well as genetic information were not available when these criteria were developed and therefore were not included in the original criteria.

Table 3.1 1984 NINCDS-ADRDA criteria for probable Alzheimer’s disease [15].

Insidious onset after age 40 with gradual progression
Dementia
Deficit in at least 2 areas of cognition (one must include memory) for at least 12 months
Progressive worsening of cognition
No disturbance of consciousness
Other systemic disease or brain disorder does not account for the disease

*Possible AD: atypical onset, presentation, or clinical course of dementia without a systemic disease or brain disorder that could account for the disease
*Definite AD: pathological confirmation by biopsy or autopsy + criteria listed for probable AD

2011 updated clinical criteria for AD [16]

Insidious onset with gradual progression
Clear-cut history of worsening of cognition by report or observation
The initial and most prominent cognitive deficits are evident on history and examination in one of the following categories.
- Amnestic presentation
- Nonamnestic presentations:
  - Language presentation
  - Visuospatial presentation
  - Executive dysfunction
The diagnosis of probable AD dementia should not be applied when there is evidence of
- Substantial concomitant cerebrovascular disease
- Core features of dementia with Lewy bodies
- Prominent features of behavioral variant frontotemporal dementia
- Prominent features of semantic variant primary progressive aphasia or nonfluent/ agrammatic variant primary progressive aphasia
- Other systemic disease or brain disorder accounting for symptoms

Integrating biomarkers and genetics into diagnostic criteria has been an ongoing process [27]. There are currently two partly overlapping sets of criteria set forth by expert workgroups (Tables 3.1 and 3.2). Both sets of criteria recognize nonamnestic presentations of AD, and both allow the integration of imaging and fluid biomarkers to supplement clinical criteria, but in different ways. The criteria proposed by the US National Institutes of Health National Institute on Aging (NIA) and the Alzheimer’s Association (AAS) (NIA-AAS) workgroup allow the diagnosis of probable AD to be made on clinical grounds alone (Table 3.1). If available, biomarkers can be used to supplement the clinical evaluation. Biomarkers are divided into two categories: markers of amyloid beta (Aβ, including CSF Aβ₄₂ levels or amyloid positron emission tomography (PET)) and markers of neuronal injury (CSF measures of total or phosphorylated tau, atrophy on MRI or hypometabolism/hypoperfusion on fluorodeoxyglucose PET (FDG-PET) or single-photon emission computed tomography (SPECT)). Various combinations of these markers modify the likelihood of underlying AD pathophysiology, for example, from low (if biomarkers from both categories are negative) to high (if there are positive markers in both categories) (Table 3.2) [16].

Table 3.2 Various criteria for AD incorporating clinical presentation and biomarkers.

Source: McKhann et al. [16]. Reproduced with permission of Elsevier.

Updated AD dementia criteria incorporating biomarkers [16].
Diagnostic Category	Biomarker probability of AD etiology	Aβ (PET or CSF)	Neuronal injury (CSF tau, FDG-PET, structural MRI)
Probable AD dementia
Based on clinical criteria	Uninformative	Unavailable, conflicting, or indeterminate	Unavailable, conflicting, or indeterminate
With 3 levels of evidence of AD pathophysiological process	Intermediate	Unavailable or indeterminate	Positive
	Intermediate	Positive	Unavailable or indeterminate
	High	Positive	Positive
Possible AD dementia (atypical clinical presentation)
Based on clinical criteria	Uninformative	Unavailable, conflicting, or indeterminate	Unavailable, conflicting, or indeterminate
With evidence of AD pathophysiological process	High but does not rule out second etiology	Positive	Positive
Dementia unlikely due to AD	Lowest	Negative	Negative
Proposed criteria for typical AD (must have A and B) [17]
Presence of early and significant episodic memory impairment that includes: Gradual and progressive change in memory function > 6 months Objective evidence of an amnestic syndrome of the hippocampal type In vivo evidence of Alzheimer’s pathology (one of the following): Decreased Aβ_1–42 together with increased t-tau or p-tau in CSF Increased tracer retention on amyloid PET AD autosomal dominant mutation present (in PSEN1, PSEN2, or APP)
Exclusion criteria:
Sudden onset Early occurrence of gait disturbances, seizures, and major and prevalent behavioral changes Focal neurological signs Early extrapyramidal signs Early hallucinations Cognitive fluctuations Non-AD dementia Major depression Cerebrovascular disease Toxic, inflammatory, and metabolic disorders MRI FLAIR or T2 signal changes in the medial temporal lobe that are consistent with infectious or vascular insults
Proposed criteria for atypical AD (must have A and B) [17]
Specific phenotype (one of the following): Posterior variant including an occipitotemporal variant defined by the presence of early, predominant, and progressive impairment of visuoperceptive functions or of visual identification of objects, symbols, words, or faces or a biparietal variant defined by the presence of early, predominant, and progressive difficulty with visuospatial functions, features of Gerstmann’s syndrome or Balint’s syndrome, limb apraxia, or neglect Logopenic variant defined by the presence of early, predominant, and progressive impairment of single-word retrieval and in repetition of sentences in the context of spared semantic, syntactic, and motor speech abilities Frontal variant defined by the presence of early, predominant, and progressive behavioral changes including association of primary apathy or behavioral disinhibition or predominant executive dysfunction on cognitive testing Down’s syndrome variant defined by occurrence of a dementia characterized by early behavioral changes and executive dysfunction in people with Down’s syndrome In vivo evidence of Alzheimer’s pathology (as noted above) Exclusion criteria Sudden onset Early and prevalent episodic memory disorder Major depression Cerebrovascular disease Toxic, inflammatory, or other metabolic disorders

An International Working Group (IWG) has proposed an alternative set of criteria, which require both a suggestive clinical syndrome (“typical” amnestic or “atypical” nonamnestic) and a biomarker evidence of AD pathophysiology by either amyloid PET or CSF Aβ₄₂ and tau measures [17]. MRI and FDG-PET/SPECT are conceptualized as “topographical markers” of disease progression, but are not included in the criteria because they are not specific to AD pathophysiology. The IWG criteria are intended to maximize accuracy in research studies but may have limited utility in the clinical setting where access to CSF biomarkers and amyloid PET is limited. Importantly, both sets of criteria also recognize mixed or atypical presentations (which can be distinguished from nonamnestic presentations that are still characteristic of underlying AD, such as PCA or the logopenic variant of PPA).

Case 1

Mr. M is a 72-year-old right-handed gentleman who has had memory problems for 4 years. He asks repetitive questions, cannot remember plans for business trips, and forgets conversations. Recently, he has had problems with navigation. Executing complex tasks such as cooking has become more difficult. Functionally, he is no longer able to work or drive. His wife has taken over bill paying and assisted him with cooking. He needs reminders to shower but is able to perform all basic activities of daily living (ADLs). Neurological examination is normal. He scores 22/30 on the Mini-Mental Status Examination (MMSE). Neuropsychological testing reveals significant deficits in episodic memory and visuospatial and executive functions. Laboratory tests for thyroid function and B12 are normal. An MRI demonstrates cortical atrophy, primarily in temporoparietal regions, and prominent hippocampal atrophy (Figure 3.1). He passed away at age 77. Autopsy diagnosis is high-likelihood AD (NIA-Reagan; see neuropathology in the following text).

Figure 3.1 Coronal T1-weighted MRI of case 1 showing bilateral hippocampal and less severe frontal and temporal cortical atrophy.

This case represents a “typical” AD presentation in a patient who developed problems with recent memory in late life. Visuospatial ability, language, and executive function are affected more variably.

Neuropathology

The core neuropathological features of AD are neuritic plaques (NPs) and neurofibrillary tangles (NFTs). NPs are extracellular, florid (flowerlike) appearing structures composed largely of the 42-amino-acid amyloid-beta polypeptide (Aβ_1–42), a cleavage product of the amyloid precursor protein (APP). Mature NPs have a dense core surrounded by dystrophic neurites; NPs are more specific for AD than the less fibrillar, diffuse plaques often seen in normal aging [28]. Plaques form seemingly simultaneously throughout the association isocortex, including parietal, prefrontal, and lateral temporal regions [29, 30]. Primary sensorimotor, visual, and auditory cortices, medial temporal cortex, and hippocampus are relatively spared of plaques in AD. NFTs are flame-shaped, intracellular inclusions composed of hyperphosphorylated species of the microtubule-associated protein tau (MAPT). NFTs first appear in the entorhinal cortex and then spread to limbic and paralimbic regions and to the temporal and parietal neocortex, with later involvement of prefrontal regions. Primary visual and sensorimotor regions are the last to develop pathology [29]. Pathologic criteria for AD include rating the distribution and burden of NPs (using Consortium to Establish a Registry for AD (CERAD) criteria [31]) and NFTs (using Braak staging [29]). Combined CERAD and Braak staging is used to establish NIA-Reagan criteria, which use these pathology ratings to state whether an individual suffered from AD with low, intermediate, or high probability (NIA-Reagan criteria). In 2012, a new set of neuropathological criteria were proposed by the NIA-AAS that integrate Thal Aβ plaque score [30] with traditional CERAD and Braak staging [32].

Genetics

Approximately 1–6% of AD patients present under the age of 65, and 60% of these cases have a positive family history with 13% showing an autosomal dominant pattern [33]. Autosomal dominant AD has been associated with mutations in three genes: ‘presenilin 1 (PS1, chromosome 14), the most common gene associated with familial AD; presenilin 2 (PS2, chromosome 1), and APP (chromosome 21). Both PS1 and PS2 are components of the gamma-secretase complex that cleaves APP into the toxic species Aβ_1–42. Mutations in PS1 have been reported to cause early behavioral changes similar to frontotemporal dementia (FTD) [34] and, in some cases, have been associated with Pick bodies, a pathologic feature of FTLD, in addition to AD pathology. Patients with trisomy 21 (Down’s syndrome) develop AD pathology in the fourth and fifth decade, likely related to the presence of 3 copies of wild-type APP [35].

APOE, the polymorphic genetic locus for apolipoprotein E on chromosome 19, is the strongest genetic determinant in sporadic AD. There are three allelic variants of APOE: ε3 is the most common, ε2 might decrease the risk of AD, whereas carriers of the ε4 allele are at higher risk for developing the disease [36, 37]. Although the ε3 allele is the most common in the general population, 50–65% of AD patients have at least one ε4 allele [38, 39]. Furthermore, there is a strong gene dose effect, such that ε4 heterozygotes are at approximately three-fold greater risk than ε4 noncarriers for developing AD, whereas homozygotes have a 15-fold greater risk [40]. Each ε4 allele is associated with an approximately 10-year younger age of onset [40]. The relationship between APOE genotype and AD phenotype (aside from early age of onset) is not clear. One study found that homozygosity for the ε4 allele was present in 17 of 71 patients presenting with an amnestic phenotype compared to only one patient of 29 patients presenting with nonamnestic phenotype [41]. Another study found a paucity of ε4 carriers (2 out of 10) in patients presenting with PCA [42], but this finding has not been replicated by other groups [6, 43]. A study from our center found no difference in the frequency of APOE ε4 between EOAD patients presenting with typical AD, PCA, or logopenic variant PPA (lvPPA), though the frequency of ε4 carriers was higher in all patient groups compared to controls [20]. Curiously, in European cases of rapidly progressive AD, ε4 carriers were underrepresented [24]. Additional risk factors for sporadic late-onset AD are being uncovered via genome-wide association studies and next-generation sequencing [44].

Structural and functional neuroimaging: MRI, FDG-PET, and SPECT

MRI in typical AD demonstrates atrophy in the areas affected by NFTs, including the hippocampus, medial temporal cortex, lateral temporoparietal cortex, and posterior cingulate/precuneus, with relatively less involvement of dorsolateral prefrontal cortex until advanced disease stages [45, 46]. A similar topographic pattern is seen with FDG-PET (reflecting hypometabolism) and SPECT (reflecting hypoperfusion) [47]. The degree of atrophy correlates with neurofibrillary pathology [48] and with clinical severity and can be used to track clinical progression [49]. It is increasingly recognized that there is a “hippocampal-sparing” endophenotype of AD which deviates from traditional Braak staging in that the medial temporal lobes are spared on imaging and at autopsy. Hippocampal-sparing AD correlates with younger age of onset and a nonamnestic clinical presentation [50, 51].

CSF/amyloid imaging

An exciting recent development in the field has been the emergence and validation of biomarkers for molecular pathology. Patients with AD show decreased levels of Aβ_1–42 and increased levels of total and phosphorylated tau in the CSF, and a ratio of tau/Aβ_1–42 can distinguish AD patients from controls with high sensitivity and specificity [52–54]. NPs can be imaged using a variety of PET tracers, including ¹¹C-PIB [55], ¹⁸F-florbetapir [56], ¹⁸F-flutemetamol [57], ¹⁸F-florbetaben [58], and ¹⁸F-NAV4694 [59]. Both CSF biomarkers and amyloid PET have been validated against autopsy-confirmed cases [53, 60, 61]. These biomarkers might be helpful for ruling in AD in patients with atypical clinical presentations [45, 62, 63] as will be demonstrated in the vignettes below. More recently, PET tracers specific to NFTs have been developed and used in pilot human studies [64–66]. These tracers will allow us for the first time to see in vivo how amyloid and tau interact with each other and with brain structure and function in aging and AD.

Case 2

Mr. D is a 51-year-old right-handed man with a history of dyslexia who developed word-finding difficulties over the past 4–5 years. When speaking, he loses phonemes (the speech sound in language of which words are represented) at the end of words and occasionally stutters. He has significant problems producing and comprehending long sentences as well as retaining long strings of numbers. Grammar is intact, but spelling is difficult. Memory has started to decline in the past year. He and his family deny any problems with visuospatial or executive function. He has been anxious and irritable. Functionally, he was forced to stop working but remains independent with all basic ADLs. On exam, his speech is hesitant and deliberate with normal syntax. Repetition is significantly impaired—for example, when asked to repeat “the ship crashed into the shore,” he states, “the boat crashed into the sand” (video 1). He has word-finding difficulties for which he compensates with circumlocutions (video 2). The remainder of his neurological exam is normal. The MMSE score is 26/30, missing points for repetition and orientation to place. Neuropsychological testing shows significant deficits in naming, repetition, echoic memory (i.e., short-term auditory storage; e.g., digit span forwards), and calculations with only minor deficits in episodic memory and executive functioning. Laboratory studies of vitamin B12 and thyroid function are normal. An MRI demonstrates global atrophy with more prominent atrophy in the left parietal and posterior temporal lobe with normal hippocampal size bilaterally (Figure 3.2). A Pittsburgh compound B PET scan is positive for amyloid. He is started on a trial of donepezil and referred to speech therapy.

Figure 3.2 (a) Axial T1-weighted MRI of case 2 showing left parietal atrophy. (b) Coronal T1-weighted MRI showing normal hippocampal size. (c) PIB-PET showing cortical PIB binding (yellow to red indicate increasing spectrum of PIB binding). Orientation of MRIs is radiologic (left is right). Orientation of PET scan is neurological (right is right).

The patient in case 2 meets Mesulam criteria for PPA as he evolved progressive difficulties with language with relative sparing of other cognitive functions for at least 2 years from symptom onset [67] (Table 3.3). As is often the case, neurodegeneration appears focal and asymmetric. PPA typically is divided into three distinct clinical variants based on the pattern of aphasia: a nonfluent variant (nfvPPA), also referred to as progressive nonfluent aphasia (PNFA), characterized by motor speech deficits and agrammatism; a semantic variant (svPPA), previously referred to as semantic dementia, characterized by fluent speech with loss of meaning for single words; and a logopenic variant (lvPPA), defined by anomia and impaired repetition (especially for long sentences with unpredictable content) with intact grammar, motor speech, and single-word comprehension [69]. Each variant is associated with a selective atrophy pattern with left inferior frontal and perisylvian involvement in nfvPPA, anterior temporal in svPPA, and left temporoparietal junction in lvPPA [21, 62, 68, 69].

Table 3.3 Criteria for primary progressive aphasia.

Source: Adapted from Mesulam [67]. © Wiley.

Gradual progression of word finding, naming, or comprehension problems
No other cognitive domains affected until 2 years after onset with language as the primary deficit (apraxia and acalculia may be present)
ADLs limited by language only in the first 2 years
No other systemic illness or brain disorder (stroke) accountable for the disease

Gorno-Tempini and colleagues characterized the language and cognitive deficits in lvPPA in depth and have proposed diagnostic criteria [69] (Table 3.4). Mr. D in case 2 meets core criteria and four supportive features for lvPPA. Furthermore, his MRI and PET qualify for an imaging-supported diagnosis of lvPPA.

Table 3.4 Criteria for logopenic variant PPA.

Clinical diagnosis of lvPPA (both core features must be present)

Impaired single-word retrieval in spontaneous speech and confrontational naming
Impaired repetition of sentences and phrases

At least three of the following other features must be present:

Speech (phonological) errors in spontaneous speech and naming
Spared single-word comprehension and object knowledge
Spared motor speech
Absence of frank agrammatism

Imaging-supported diagnosis of lvPPA (both criteria must be present)

Clinical diagnosis of lvPPA
Imaging must show at least one of the following results:
- Predominant left posterior perisylvian or parietal atrophy on MRI
- Predominant left posterior perisylvian or parietal hypoperfusion or hypometabolism on SPECT or PET

lvPPA with definite pathology (clinical diagnosis of LV and 1 or 2 must be present)

Histopathological evidence of a specific neurodegenerative pathology (e.g., AD, FTLD-tau, FTLD-TDP, and others)
Presence of a known pathogenic mutation

Determining the PPA variant can help predict underlying pathology, as AD is frequently the cause of lvPPA, whereas nfvPPA and svPPA are usually associated with FTLD pathology [3, 10]. Several studies have shown that the majority of, but not all, lvPPA patients have AD pathology using PIB, autopsy, or CSF biomarkers [62, 70, 71]. Therefore, whereas lvPPA appears to be a marker for AD pathology, the syndrome is pathologically heterogeneous, and both CSF biomarkers and amyloid imaging can be helpful in determining whether AD is the causative pathology. For Mr. D, a positive PIB scan (Figure 3.2) greatly increases the likelihood of AD as the pathologic substrate.

In most studies, PPA patients with pathological AD have been found to have increased NFTs in the left hemisphere [10] compared [72] to the right, though this was not observed in all patients [5]. Results regarding plaque distribution have been more variable, with some studies reporting a left-sided predominance in PPA [5] and others finding a symmetric distribution of plaques indistinguishable from a typical AD pattern [10, 62].