Early-onset Alzheimer’s Disease: Nonamnestic Subtypes and Type 2 AD

Abstract

Patients with Alzheimer’s disease (AD), the most prevalent neurodegenerative dementia, are usually elderly; however, ~4–5% develop early-onset AD (EOAD) with onset before age 65. Most EOAD is sporadic, but about 5% of patients with EOAD have an autosomal dominant mutation such as Presenilin 1, Presenilin 2, or alterations in the Amyloid Precursor Protein gene. Although most Alzheimer’s research has concentrated on older, late-onset AD (LOAD), there is much recent interest and research in EOAD. These recent studies indicate that EOAD is a heterogeneous disorder with significant differences from LOAD. From 22–64% of EOAD patients have a predominant nonamnestic syndrome presenting with deficits in language, visuospatial abilities, praxis, or other non-memory cognition. These nonamnestic patients may differ in several ways from the usual memory or amnestic patients. Patients with nonamnestic EOAD compared to typical amnestic AD have a more aggressive course, lack the apolipoprotein E ε4 (APOE ε4) susceptibility gene for AD, and have a focus and early involvement of non-hippocampal areas of brain, particularly parietal neocortex. These differences in the EOAD subtypes indicate differences in the underlying amyloid cascade, the prevailing pathophysiological theory for the development of AD. Together the results of recent studies suggest that nonamnestic subtypes of EOAD constitute a Type 2 AD distinct from the usual, typical disorder. In sum, the study of EOAD can reveal much about the clinical heterogeneity, predisposing factors, and neurobiology of this disease.

Introduction

In 1907, Alois Alzheimer described a 51-year-old patient who developed dementia with predominant language and behavioral changes (1). This patient, Auguste Deter, proved to have the amyloid plaques and neurofibrillary tangles (NFTs) that have come to define the neuropathology of AD. For the next 70 years, early-onset dementia was the embodiment of AD (EOAD), with the much more common late-onset disorder (LOAD) with onset after age 65, considered senile dementia from normal aging. In 1968, however, Blessed and colleagues showed that the brains of patients with senile dementia had plaques and tangles that were qualitatively the same as early-onset forms (2). The consolidation of these two age-related forms of the disease was completed when a 1976 editorial by Robert Katzman emphasized that AD and senile dementia were a single disease (3). By the 1984 NINCDS-ADRDA criteria for AD, the definition of AD was transformed to the usual progressive memory deficit presentation seen in LOAD, with the range of AD spanning onsets from 40–90 years of age (4). From this point on, many clinicians and investigators forgot the example of Auguste Deter and the lessons to be learned from the nonamnestic subtypes of EOAD. Recently, however, many investigators have rediscovered EOAD and have focused on understanding what AD that begins at or before age 65 years can reveal about the clinical manifestations and pathophysiology of this disease.

EOAD is an important clinical problem. Alzheimer’s disease (AD), which is the most common neurodegenerative dementia (5), is particularly devastating when it occurs at a young age. Beyond the psychological and medical toll, EOAD disproportionately impacts individuals during their most productive years, and the cost of treating patients with EOAD is significant. The few epidemiological studies on EOAD indicate an incidence rate of about 6.3/100,000 (6) and a prevalence rate of about 24.2/100,000 in the 45- to 64-year-old age group (7) or between 220,000 and 640,000 people in the United States (8). This compares to a 10–20 times greater incidence and prevalence rates for LOAD. In some specific populations such as U.S. veterans, the frequency of EOAD may be even higher (9). Finally, although the cut-off age of 65 years for EOAD is arbitrary and harkens back to the original AD vs. senile dementia distinction, it remains useful for distinguishing different AD syndromes. Many EOAD patients have nonamnestic syndromes infrequently present among those with LOAD.

Genetic and Predisposing Factors

Not only is there an increased family history of dementia among persons with EOAD, compared to those with LOAD (10), but there are genes that increase the risk of developing EOAD. The most important are the apolipoprotein E (APOE) ε4 alleles. APOEε4 is a susceptibility gene that decreases the age of onset for typical AD (11, 12), and it correlates with the presence of memory loss and hippocampal atrophy in AD (13, 14). The APOEε4 allele may be responsible for those EOAD patients with a typical amnestic presentation (11, 15–17). In contrast, EOAD patients with atypical nonamnestic presentations tend to be APOEε4 negative (11, 15, 16).

Autosomal dominant familial forms of AD account for only ~1% of all patients with AD, but they usually occur at a young age and have EOAD. The most common cause for familial EOAD is a mutation in the Presenilin 1 (PSEN1) gene, followed by Presenilin 2 (PSEN2) and the Amyloid Precursor Protein (APP) gene (18, 19). Patients with familial EOAD are less likely to present with nonamnestic deficits than those with sporadic EOAD (See Table 1) (20). In addition to a much younger age of onset (41.8 ± 5.2 years for PSEN1 vs. 55.9 ± 4.8 for sporadic EOAD), clinicians may be able to distinguish familial EOAD patients with PSEN1 mutations from those with sporadic EOAD by the presence of neurological abnormalities including a history of headaches and pseudobulbar affect, as well as myoclonus and gait abnormality on examination (20). Detailed family history is indicated as part of the evaluation of EOAD because patients with AD genes usually have a positive family history of autosomal dominant transmission (21). EOAD may be present if there are at least three affected family members in two or more generations, some of whom are the patient’s first-degree relatives (21). Some patients with familial EOAD, however, may lack a known family history or may have incomplete penetrance (22). For these reasons, it is important to consider evaluating EOAD patients with a very early age of onset or abnormal neurological findings for genetic factors. Actual genetic testing, however, should only occur with the participation of someone who has expertise in the area of genetic counseling (21).

Table 1.

Presenilin-1 (PSEN1) familial EOAD vs. sporadic (non-familial) EOAD: presenting symptoms on initial clinic visit

Symptom	PSEN1 familial EOAD (n = 32)	Non-familial EOAD (n = 81)	Significance, p
Memory	27 (84.37%)	47 (58.02%)	χ2 = 5.91; p <0.05
Visuospatial	1 (3.12%)	14 (17.28%)	χ2 = 2.86; ns
Language	3 (9.38%)	14 (17.28%)	χ2 = 1.48; ns
Other	1 (3.12%)a	6 (7.4%)b	χ2 = 0.18; ns

^aApathy in PSEN-1 familial EOAD patient.

^bPersonality change (3); writing impairment (2); paranoia (1).

ns, nonsignficant.

There may be additional predisposing factors for EOAD. Susceptibility genes other than APOEε4, such as the single nucleotide polymorphisms CR1, CLU, BIN, and PICALM, or tau gene H1/H1 haplotypes, or even prion gene polymorphisms, could increase the risk of EOAD, but only by a modest degree (23). One study, which used the analysis of parent-child concordance rates to calculate heritability among those with EOAD, proposes an autosomal recessive pattern of inheritance for EOAD, possibly from unknown genes (24–26). Prior psychological trauma may also predispose to EOAD, and there are reports linking an increased incidence of dementia with post-traumatic stress disorder (27, 28). In addition, there is an increased risk of dementia among individuals who have had a traumatic brain injury (29), and dementia pugilistica in boxers or exposure to blast or other closed head injury may lower the age of onset of dementia (30).

EOAD Clinical Subtypes

About 22–64% of cases of EOAD may have an atypical or nonamnestic presentation (31–34). Experts have developed the new National Institute on Aging and the Alzheimer’s Association (NIA-AA) criteria for AD, in part, to take into account the nonamnestic presentations of EOAD (35). These nonamnestic EOAD subtypes differ from typical amnestic AD not only in their non-memory presentations (11, 15, 16, 32, 34), but in having a more aggressive rate of progression with a shorter duration of disease (16, 36, 37),

The EOAD nonamnestic subtypes currently constitute a confusing list of syndromes and classifications (Table 2) (33, 34, 38). Similar to Alois Alzheimer’ original patient, the most common may be a language-impaired or aphasic subtype (32), with recent indications that progressive logopenic aphasia (PLA) may be the same as this nonamnestic EOAD (39, 40). Others suggest that a biparietal subtype, with visuospatial deficits and ideomotor apraxia, is the most nonamnestic EOAD (33). Many investigators describe a posterior cortical atrophy (PCA) subtype of EOAD that has prominent visuospatial or visuoperceptual deficits (41, 42). Some include the biparietal subtype as PCA and divide PCA into temporo-occipital forms and a basic visual variant (32, 43). Still others describe a nonamnestic EOAD subtype with predominant executive deficits (34). Patients with the corticobasal syndrome, characterized by progressive limb apraxia as well as motor changes, have AD in about half of their autopsies (44), and may constitute an additional nonamnestic EOAD subtype. Different combinations of aphasia, apraxia, and agnosia occur as well, and there is no consensus on how to group these nonamnestic subtypes.

Table 2.

Studies with nonamnestic EOAD

Potential localization	Koedam et al. 2010 n = 87	Alladi et al. 2007 n = 34	Stopford et al. 2007 n = 17
Left parietal	Apraxia/visuospatial (12%)	Corticobasal syndrome (17.5%)	Praxis 23.5%
Left parietal	Language (9%)	Language (56%):	Language 23.5%
Left temporal-occipital	Aphasia-apraxia-agnosia (8%)	(nonfluent 35%, semantic 6%, mixed 15%)
Dorsolateral frontal	Dysexecutive (2%)	Beh. variant fronto- temporal dem. (6%)	Dysexecutive (41.2%)
Right parietal	Posterior cortical atrophy (1%)	Posterior cortical atrophy (20.5%)	Perceptuospatial (11.8%)
Right temporal-occipital

In a 10-year retrospective study, our group divided 125 patients with EOAD into four main subgroups: typical amnestic EOAD, a language group consistent with PLA, a visuospatial group consistent with PCA, and a smaller limb apraxia group not meeting motor criteria for corticobasal syndrome, and termed progressive ideomotor apraxia (PIA) (Table 3) (31). On comparison with 56 patients with LOAD, 80 (64%) of our EOAD patients had a nonamnestic presentation, compared to only 7 (12.5%) of the LOAD patients. On magnetic resonance imaging (MRI), similar to typical AD, the amnestic EOAD patients showed more hippocampal atrophy, but, different from typical AD, the nonamnestic EOAD patients with language presentations had more left parietal changes and the nonamnestic EOAD patients with visuospatial presentations had more right parietal-occipital changes. In sum, the cohort of EOAD contains nonamnestic subtypes whose disease focuses not in the usual memory-hippocampal system, but in posterior neocortex.

Table 3.

EOAD subgroups: demographic features and presenting symptoms

	Amnestic- aEOAD, n = 45	Language n = 33	Visuospatial n = 35	Limb praxis n = 12	Significance 4 groups
Age pres.	55.6 (5.34)	59.73 (5.02)	60.91 (4.6)	58.92 (8.7)	F = 7.07, p <0.001
Age onset	52.52 (5.76)	55.91 (5.02)	56.82 (4.89)	55.83 (5.62)	F = 4.99, p <0.01
Sex-M/F	21/24	17/16	19/16	5/7	χ2 = 0.81, ns
1° FHx	13 (28.9%)	18 (54.5%)	12 (34.3%)	4 (33.3%)	χ2 = 7.18, ns
MMSE	22.18 (6.02)	18.36 (7.33)	21.22 (6.07)	22.1 (3.41)	F = 2.64, ns
Areas of difficulty on presentation	45 memory	16 fluency- effortful, halting, 13 word-finding 8 reading or spelling, 2 auditory comprehension	16 localization 9 visual perception, 8 visual reading, 6 env. orientation, 4 dressing ability, 4 object recognition, 2 driving ability	6 mechanics of writing, 2 manual dexterity, 4 complex motor (2 acalculia; 9 aphasia)

From Mendez et al. 2012.

Investigators have proposed specific clinical criteria for PLA and PCA, two of these nonamnestic EOAD subtypes. PLA criteria include an insidious onset and progression of both of the following (40): 1) impaired single-word retrieval in spontaneous speech and naming (anomia) and 2) impaired repetition of sentences and phrases. In addition, there are at least three of the following other features: Speech (phonologic) errors, spared single-word comprehension and object knowledge, spared motor speech, and absence of frank agrammatism. In addition, PLA results in reading difficulty due to phonological alexia. PCA criteria include an insidious onset and progression of all of the following (42): 1) visual complaints with intact primary visual functions, except for possible visual field deficits; 2) evidence of predominant complex visual disorder, such as oculomotor apraxia, optic ataxia, environmental disorientation (header, topographic, landmark), dressing apraxia, abnormal anti-saccades, neglect, constructional difficulty, simultanagnosia, visual agnosia, prosopagnosia; and 3) proportionally less impaired deficits in memory and verbal fluency. In addition, PCA most commonly affects the dorsal visual stream involving spatial localization rather than the ventral visual stream involving object identification (41).

Neuropsychological studies have also shown a higher prevalence of non-memory symptoms as the initial presentation of EOAD (33). Among EOAD patients compared to LOAD patients, most recent studies find worse visuospatial functions, including spatial localization and hand-eye coordination, with relative sparing of memory (45–48). After age correction, EOAD patients also have worse motor praxis when compared to LOAD patients (45). In one neuropsychological comparison of 81 EOAD and 91 LOAD, the EOAD patients had relative sparing of memory and worse visuospatial functions, attention, and executive functions, whereas the LOAD patients tended to have worse memory (38). In contrast, different investigators found worse frontal-executive performance among LOAD patients rather than among EOAD patients (49). This discrepancy may be due to the presence of decreased working memory in EOAD, not from frontal-executive dysfunction but from an impaired capacity-dependent, episodic buffer located in the parietal lobe (50).

Neuroimaging Results

Patients with EOAD, in general, have a different structural neuroimaging pattern than patients with LOAD, probably due to the presence of the nonamnestic subtypes (51–53). For nonamnestic EOAD, investigators have reported differences in grade and distribution of gray matter atrophy in posterior parts of the brains compared to more medial temporal regions in LOAD (48, 54–57). In EOAD (with embedded variants), voxel based morphometry (VBM) studies have shown more widespread changes and less atrophy in anterior hippocampus and amygdala compared to LOAD (58). In another VBM study of EOAD, cortical atrophy maps showed changes affecting all lobes, whereas they are more confined in LOAD (48). Moreover, we used tensor based morphometry to create 3D Jacobian maps of local differences in tissue volume, and showed that EOAD subjects had significantly smaller volume in superior temporal and posterior parietal regions, compared to LOAD subjects (Figure 1).

Figure 1.

Regions where EOAD had significantly smaller volume compared to LOAD.

Magnetic resonance imaging (MRI) shows temporoparietal atrophy, even in the presence of hippocampal sparing, in nonamnestic EOAD subtypes (59). Analysis of cortical thickness patterns in EOAD subtypes suggests that different clinical presentations of AD represent points in a phenotypic spectrum of neuroanatomic variation (60). These investigators found overlapping temporoparietal areas on cortical thickness measures among 15 LPA, 25 PCA, and 14 amnestic EOAD patients compared to 30 normal controls. Areas of involvement in PLA and PCA are regionally distributed in posterior neocortex with involvement of corresponding white matter tracts (39, 61, 62), and there may be greater parietal precuneus involvement in EOAD vs. LOAD (63). Others have found that there are bilateral overlapping areas of parietal and posterior temporal involvement in PLA and PCA (61). However, in PLA, there is thinning of left posterior temporal, inferior parietal, medial temporal, and posterior cingulate (64), and in PCA there is greater involvement of the right parietal and bilateral occipital lobes (31, 61) . In sum, despite overlap, in parietal neocortex, there are clear differences among subtypes with more involvement in superior temporal lobe in LPA, bilateral occipital-parietal areas in PCA, and bilateral medial temporal and posterior parietal areas in typical amnestic EOAD.

Patients with EOAD not only show more severe cortical atrophy, but also have more glucose hypometabolism than patients with LOAD (48, 55, 57, 63, 65–69). On ¹⁸fluoro-deoxyglucose positron emission tomography (FDG-PET) studies, comparably demented patients with EOAD (with embedded variants) showed greater regional hypometabolism, particularly in parietal areas, than normal controls and those with LOAD (67, 68, 70, 71–74). In our study, EOAD patients, compared to LOAD patients, also had greater focal hypometabolism in the parietal regions (left worse than right), whereas LOAD patients, compared to the EOAD patients, had greater hypometabolism in the bilateral inferior temporal regions and inferior frontal regions (right worse than left) (Figure 2) (Kaiser, in press). In a recent study, younger patients with EOAD (<62 years) had decreased FDG in the parietal cortex associated with more severe impairment in visuospatial fuctions, attention, and executive function vs. decreased posterior cingulate uptake and impairment in memory in LOAD (75).

Figure 2.

Base comparison of EOAD relative to LOAD. Two-sample t-test comparison shows relative hypometabolism in the EOAD group (blue) and relative hypometabolism in the LOAD group (red) (uncorrected p = 0.01, T = 2.43, k_c = 10 voxels). Picture is in neurological convention; brain’s left is your left. From Kaiser et al., in press.

In addition to FDG-PET, nuclear imaging probes can detect amyloid-β deposits in the brain with [¹¹C]6-OH-BTA-1 Pittsburgh Compound B (PIB) and other forms of amyloid PET imaging (76). [¹¹C]PiB is a radioactive diagnostic agent tagged with a radioisotope that binds to fibrillar amyloid plaques and has been sensitive to the detection of typical AD (77). Patients with EOAD, like those with LOAD, have increased global [¹¹C]PIB retention without specific differences in distribution (78), and some, but not all, studies have found greater amyloid burden on [¹¹C]PiB imaging in EOAD compared to LOAD (78, 79). With the exception of increased PIB-PET signal in the neostriatum in familial EOAD, distinct distributions of amyloid load have been absent in patients with PLA (79, 80) or PCA (81, 82), and there have not been regional differences in amyloid burden when EOAD was compared to LOAD (57). One recent study, however, has succeeded in showing that patients with EOAD have increased [¹¹C]PIB binding in parietal areas associated with decreased visuospatial functions when compared to LOAD (75). Finally, there are greater regional changes on [¹⁸F]FDG-PET than on [¹¹C]PIB-PET in nonamnestic EOAD syndromes (82), suggesting that not amyloid plaque deposition but metabolic impairment drives the clinical presentation of the nonamnestic EOAD subtypes.

Neurobiological Differences from Load

Nonamnestic EOAD is different from typical amnestic AD in having greater initial pathology in posterior parietal and adjacent areas rather than in the more typical entorhinal-hippocampus region (58, 61, 83). Postmortem, there are higher burdens of amyloid plaques and NFTs in EOAD than in LOAD (84), and these appear concentrated in posterior neocortical areas. There are relatively greater focal neocortical NFTs and less hippocampal atrophy in PCA, PLA, and PIA when compared to typical amnestic AD (43, 85–87). In comparison, typical AD is associated with early volume loss in the hippocampus and medial temporal regions, as well as the temporoparietal cortex, posterior cingulate, and precuneus (88, 89). In further support of this, neuropathological findings report a hippocampal sparing subgroup of AD that is younger at death and has higher than expected NFTs in inferior parietal, superior temporal, and middle frontal cortices (83).

Although there are no consistent differences in cerebrospinal fluid (CSF) biomarkers across the EOAD subtypes, they do point to a specific sequence of pathological changes. Amyloid-β_1–42 and hyperphosphorylated tau levels in CSF do not differ according to age at onset (90). Compared to amnestic EOAD, however, patients with nonamnestic EOAD tend to have higher tau levels in the CSF but do not differ in levels of amyloid-β (91). These findings suggest a faster course with greater NFT density in nonamnestic vs. amnestic EOAD. This is further examplifyed in those at risk for autosommal dominant AD due to PSEN1, PSEN2, and APOE (92). Because the age at clinical onset is similar among generations with an autosomal dominant form of AD, this allows for predicting the age of onset of AD and for demontrating the sequence of pathological changes (92). Preceding dementia, there is decreased CSF amyloid-β 25 years before; positive [¹¹C]PIB-PET, increased CSF tau, and increased brain atrophy 15 years before;, and decreased FDG metabolism and memory 10 years before (92, 93). Higher CSF tau levles in nonamnestic AD is, therefore, consistent with a faster rate of disease.

Type 2 AD

The clinical and neuropathological differences of nonamnestic EOAD from typical AD suggests that nonamnestic EOAD is a Type 2 AD with overlapping nonamnestic presentations and a relatively aggressive disease course not affected by the age-lowering effects of APOEε4 (Figure 3). Parietal lobe dysfunction is an early characteristic of nonamnestic EOAD. Specifically, Type 2 AD may be made up of clinically overlapping visuospatial, language, or praxis syndromes which differ, as a group, from typical amnestic AD of early or late onset. Defining Type 2 AD challenges the dominant clinical view of AD as necessarily an insidious memory decline in the elderly.

Figure 3.

Nonamnestic (Type 2 AD) vs. typical, amnestic EOAD. This graph also contains postulated differences in APOEε4. Adapted and modified from Van der Flier et al. 2011.

Type 2 AD also challenges the prevailing sequence of initial amyloid-β_1–42 deposition followed by NFT formation in the entorhinal-hippocampal cortex and suggests an alternate cascade in the expression of Alzheimer pathophysiology. The prevailing Alzheimer cascade theory indicates a specific sequence of neuropathological changes. First, there is accumulation of amyloid-β in the brain which drives the neurodegeneration. Amyloid-β levels in CSF decrease early and reach plateau levels long before dementia develops whereas tau levels, which reflect neurodegeneration, continue to increase as cognition declines. Investigators postulate that typical AD begins with amyloid-β deposition followed by NFTs in the entorhinal-hippocampal region (94, 95). In nonamnestic EOAD, there is selective vulnerability of posterior neocortical areas and relative hippocampal sparing (83, 95). Whereas typical AD, which includes amnestic EOAD as the younger end of the age distribution, follows the proposed Alzheimer cascade (94), nonamnestic EOAD represents a continuum of a common early deposition of NFTs in posterior parietal-precuneus/posterior cingulate with involvement of adjacent neocortex and white matter tracts. The alternate cascade in Type 2 AD still begins with amyloid-β deposition but proceeds to earlier and more prominent NFTs in posterior neocortex rather than in the more typical hippocampal-posterior cingulate system. Type 2 AD is composed of overlapping syndromes with a similar or proximate hub of onset in posterior parietal-precuneus/posterior cingulate and adjacent gray and white matter (34, 61, 96). In nonamnestic EOAD, the origin of an apparent predilection for early amyloid-β deposits and neurodegeneration in parietal and related areas is unclear, but there are several possible considerations. Typical amnestic AD impairs the default mode neural network disrupting hippocampal-cortical memory systems, and nonamnestic EOAD may impair other neural networks early in the course such as the frontoparietal control network (97) leading to early parietal amyloid-β deposition. Type 2 AD may also affect temporoparietal connectivity as compensation to counteract a parietal-posterior cingulate gyrus disconnection (98). There may be different amyloid-β oligomer profiles in EOAD vs. LOAD (99). These oligomers, rather than the amyloid-β insoluble fibrils, may the the molecular pathogens that trigger synaptic dysfunction in EOAD and negatively affect the number of nicotinic acetylcholinesterase receptors (99). Moreover, levels of anti-amyloid-β IgG (IgG1 and IgG3), which may help control the development of amyloid plaques, are lower in PCA vs. typical AD (100). These serum anti-amyloid-β antibodies may differentially react to amyloid-β oligomers in nonamnestic EOAD. This work is clearly preliminary, but it indicates how researchers are beginning to explore the underlying reasons for Type 2 AD. In conclusion, EOAD, in comparison to LOAD, reveals insights regarding the underlying pathophysiology of AD. EOAD has many non-memory presentations (language, visuospatial, executive) which tend to be APOEε4 negative, lack a determinative Alzheimer mutation, and have distinct focal gray and white matter changes and regional hypometabolism. We propose a Type 2 AD, which results from early, regional vulnerability to amyloid deposition in posterior neocortex. At onset, nonamnestic EOAD subtypes involve parietal and associated neocortex whereas typical amnestic AD involves the hippocampi and related structures. Although amyloid β_1–42 deposition may initiate the cascade in both forms of AD, the effects result in an initial impact of neurodegeneration and NFTs in neocortex in nonamnestic EOAD as opposed to the usual Braak and Braak staging with onset in transentorhinal cortex in typical amnestic AD. Researchers are just beginning to explore the underlying pathophysiological factors for this selective vulnerability in Type 2 AD. Finally, clinicians need to recognize patients with nonamnestic EOAD because of implications for clinical care, clinical trials, developing AD diagnostic criteria, genetic testing, and quality-of-life factors. Dementia devastates at any age, but when it affects someone pre-retirement and with financial and family responsibilities, prompt diagnosis and mobilization of resources is absolutely vital (101).