Phenotypic subtypes of progressive dysexecutive syndrome due to Alzheimer’s disease: a series of clinical cases

Abstract

Diagnostic criteria for a progressive dysexecutive syndrome due to Alzheimer’s disease (dAD) were proposed. Clinical observations suggest substantial variability in the clinico-radiological profiles within this syndrome. We report a case series of 6 patients with dAD highlighting this heterogeneity. Average age at diagnosis was 57.3 years, and patients were followed annually with clinical, cognitive, and multimodal imaging assessments for an average of 3.7 years. Cases were divided based into three subtypes based on their pattern of FDG–PET hypometabolism: predominantly left parieto-frontal (ldAD), predominantly right parieto-frontal (rdAD), or predominantly biparietal (bpdAD) (n = 2 for each). Prominent executive dysfunction was evidenced in all patients. ldAD cases showed greater impairment on measures of verbal working memory and verbal fluency compared to other subtypes. rdAD cases showed more severe alterations in measures of visual abilities compared to language-related domains and committed more perseverative errors on a measure of cognitive flexibility. bpdAD cases presented with predominant cognitive flexibility and inhibition impairment with relative sparing of working memory and a slower rate of clinical progression. rdAD and bpdAD patients developed neuropsychiatric symptoms, whereas none of the ldAD patients did. For each subtype, patterns of tau deposition relatively corresponded to the spatial pattern of FDG hypometabolism. dAD cases could be differentiated from two clinical cases of atypical AD variants (language and visual) in terms of clinical, cognitive and neuroimaging profiles, suggesting that dAD subtypes represent clinical entities separable from other variants of the disease. The recognition of distinct dAD phenotypes has clinical relevance for diagnosis, prognosis, and symptom management.

Introduction

Diagnostic criteria for a dysexecutive syndrome due to Alzheimer’s disease (dysexecutive AD or dAD) have been recently proposed [81]. These include the presence of a progressive and predominant degradation of executive functioning (i.e., cognitive flexibility, working memory and/or inhibition) in the setting of AD pathology, with the absence of predominant behavioral features. One critical aim of this set of criteria was to distinguish dAD from “frontal” dementia syndromes such as the behavioral variant of frontotemporal dementia (bvFTD) [66] and other AD phenotypic presentations such as the amnestic, visual (i.e., posterior cortical atrophy; PCA) [17] and language (logopenic primary progressive aphasia; lvPPA) [24] variants [25]. This latter distinction was notably supported by the finding of selective degeneration of the parieto-frontal network in dAD, which is closely related to executive functioning [10, 43, 56, 71] as evidenced with ¹⁸F-fluorodeoxyglucose–positron emission tomography (FDG–PET) [81]. This parieto-frontal working memory network has also been observed to accumulate tau pathology to a greater degree in younger-onset cases of AD [42], and the extent of tau accumulation within this network correlates with the degree of executive functioning impairment across AD phenotypes [7].

Clinico-radiological heterogeneity is commonly observed in neurodegenerative diseases and can even be found within well-established dementia syndromes sharing identical pathological substrates. For instance, PCA can be divided into different phenotypic subtypes (i.e., dorsal, ventral, caudal) with distinct predominant cognitive deficits [17], and right-sided lvPPA patients show less severe aphasia but faster rate of disease progression compared to those with left-sided lvPPA, even though AD is the most frequent pathology in both phenotypes [15]. Similarly, although all dAD patients present with prominent executive dysfunction, observations stemming from our clinical practice in behavioral neurology and neuropsychology suggest inter-individual variability in clinical presentation and progression, imaging findings and neuropsychological profiles within this phenotype. This heterogeneity would seemingly be underpinned by differences in patterns of degeneration of the parieto-frontal network across patients. More precisely, we observed three distinct phenotypic dAD subtypes based on their pattern of FDG–PET hypometabolism: left-predominant (ldAD), right-predominant (rdAD) and bi-parietal predominant (bpdAD).

This study aims to provide a thorough description of a clinical case series of phenotypic subtypes of dAD in terms of clinical presentation and progression, multimodal imaging findings (magnetic resonance imaging; MRI; FDG–PET; amyloid-PET; tau-PET) and neuropsychological profiles. All dAD patients were examined in clinical and research settings in a tertiary behavioral neurology clinic at Mayo Clinic Rochester. Subtype classification was based on patterns of FDG–PET hypometabolism given the clinical utility of this imaging technique for differential diagnostic in dementia syndromes [32] and its relevance to assess network integrity underlying inter-individual cognitive impairment across the spectrum of neurodegenerative disorders [40]. Finally, we compared clinical, imaging, and neuropsychological profiles of dAD patients to those of two cases of atypical AD, one with PCA and one with lvPPA, to demonstrate that dAD, albeit its intra-syndrome heterogeneity, represents a clinical entity discernible from other atypical AD phenotypes.

Methods

Patient consent

This retrospective study was approved by the Mayo Clinic Institutional Review Board. All participants or their designee provided written informed consent at their first clinical visit to allow their clinical data to be used for research. All patients were subsequently enrolled in the Alzheimer’s disease research center (ADRC) protocol, which allowed for annual follow-up with neurological examination, multimodal imaging, and neuropsychological assessment.

Recruitment and case selection

Patients selected for this study were referred to our behavioral neurology clinical practice at Mayo Clinic Rochester between March 2010 and May 2019, and data collection was performed up to April 2021. Cases were selected by experienced clinical neurologists subspecialized in behavioral neurology (DTJ, HB) according to clinical diagnosis (dAD; see criteria below), AD biomarkers positivity, and FDG–PET image availability. Three dAD cases (one ldAD, one rdAD and one bpdAD) out of six were part of another study from our group [81]. Two other cases of biomarker-confirmed atypical AD, one diagnosed with lvPPA and another with PCA, were also included to allow for comparison of clinical features, neuroimaging findings and neuropsychological profiles with dAD cases. A retrospective review of clinical Electronic Medical Record and research amyloid-PET and tau-PET images was additionally performed for all patients.

Diagnosis and subtype classification

All dAD patients met criteria for progressive dysexecutive syndrome due to AD [81], in that they presented 1) a persistent, predominant and progressive dysexecutive syndrome for at least 6 months in the absence of predominant behavioral features (i.e., not meeting criteria for bvFTD), 2) positive biomarkers for AD pathology as evidenced by CSF biomarkers (decreased Aβ42: increased P-tau; low Amyloid-Tau Index or ATI) or PET-imaging (i.e., elevated tracer retention on amyloid-PET and tau-PET), and 3) an absence of an alternative medical condition severe enough to account for the clinical presentation. Thresholds for AD biomarkers positivity are described in the following sections. Given that dAD diagnostic criteria were not published at the time of data collection, all patients initially received a diagnosis of early-onset AD [20]. They were then retrospectively attributed a provisional diagnosis of dAD in the context of another publication (see [81]. The lvPPA and PCA cases met well-established consensus diagnostic criteria, including imaging-supported features [17, 24].

Clinical diagnoses were assigned following a clinical visit with a behavioral neurologist (DTJ, DSK, RCP, JGR). Clinical interview was performed with the patient and an informant. Neurological examinations were based on clinical judgement of the treating neurologist. All patients were administered the Kokmen Short Test of Mental Status (STMS) [46] at all clinical visits, which is a bedside cognitive test. Neurological examination assessed a combination of the following domains: ideomotor praxis, simple motor programming (Luria’s motor series) [52], language (standardized auditory comprehension, reading comprehension, naming, repetition, narrative picture description and writing), visuospatial and visuoconstruction abilities (cube and clock drawing) and simultagnosia (Navon figures, [58], Ishihara plates [13], overlying figures with five items [68]).

dAD subtype classification was based on visual inspection of FDG–PET hypometabolism pattern and was further confirmed via consensus opinion among subspeciality trained behavioral neurologists (DTJ, HB, JGR). dAD cases were divided into 3 subtypes: left-predominant (n = 2), right-predominant (n = 2), or biparietal-predominant (n = 2). This classification was done blinded from clinical information profile, aside from the fact that patients were assigned a diagnosis of dAD.

CSF biomarkers

Six out of 8 patients (5 dAD and the PCA) underwent a lumbar puncture within approximatively 1–2 weeks of the initial clinical visit. Two milliliters (ml) of CSF were collected and conserved in polypropylene transfer tubes, frozen at −85 °C. CSF was transported to Athena Diagnostics in Worcester, MA, where CSF analysis was performed using an enzyme-linked immunosorbent assay. Diagnostic thresholds provided by Athena Diagnostics are based on P-tau levels and the ATI and are the following: not AD (P-tau < 54 pg/ml; Aβ42-Tau Index (ATI) > 1.2); borderline AD (P-tau 54–58 pg/ml; ATI 0.8–1.2); AD (P-tau > 58 pg/ml; ATI < 0.8). The ATI measure used by Athena has been proven to have significant clinical relevance for differentiating clinically diagnosed AD from non-AD disorders, such as vascular dementia and bvFTD with respective sensitivity and specificity of 85–94% and 83–89% [1, 34]. Of note, borderline CSF ATI is commonly observed in dAD [81], because 24% of cases have low amyloid but normal P-tau levels and thus patients within this range are defined as possible dAD based on CSF biomarkers. dAD individuals included in this study with low amyloid and normal P-tau levels also underwent tau-PET imaging which demonstrated tau-PET positivity despite the normal CSF P-tau results.

Neuroimaging

FDG–PET images were acquired using a PET/CT scanner (GE Healthcare) and were performed in both clinical and research contexts following identical procedure. Patients were injected with FDG in a dimly lit room and waited for a 30-min uptake period. FDG scans lasted 8 min, separated into four 2-min dynamic frames following a low-dose CT transmission scan. Scans were processed through CortexID software (GE Healthcare). Standardized uptake value ratios (SUVRs) were calculated by normalizing regional glucose uptake to the pons and were compared with an age-segmented normative database. Whole-brain hypometabolism patterns were projected onto a three-dimensional stereotactic brain rendering.

An “asymmetry index” was derived for FDG–PET scans at clinical presentation for all patients by averaging SUVRs from several brain regions (lateral and medial prefrontal, sensorimotor, anterior and posterior cingulate, precuneus, superior and inferior parietal, occipital lateral, primary visual, lateral and medial temporal) for each hemisphere, and then subtracting the right hemisphere mean from the left hemisphere mean. Therefore, a positive score indicated hypometabolism asymmetry toward the right hemisphere, whereas a negative score indicated hypometabolism asymmetry toward the left hemisphere. This asymmetry index was interpreted qualitatively given the lack of standard to determine the significance of asymmetry.

Amyloid-PET and tau-PET scans were performed as part of the ADRC protocol. These were, respectively, done with Pittsburgh compound B (PiB) and 18Flortaucipir (AV1451) radiotracers which were synthetized using an on-site cyclotron. Image processing was done using procedures described in separate publications from our group [37, 38]. Both amyloid-PET and tau-PET images were scaled to the cerebellar crus region, which yielded regionwise standard uptake value ratios (SUVR). We used a validated meta-region of interest (ROI) to derive global amyloid-PET and tau-PET SUVR scores for each patient. Abnormal thresholds were set at > 1.42 and > 1.23 for amyloid-PET and tau-PET, respectively [38].

All PET images were co-registered to a corresponding MRI T1-weighted image acquired within a week from PET images acquisition. MRI images were acquired using a GE scanner at either 1.5 T or 3 T following a standardized procedure.

Neuropsychological assessment

Neuropsychological assessment was performed on all patients at initial visit. Four of the six dAD and the PCA patients completed annual assessments. Selection of tests was not standardized and varied depending upon whether the assessment was performed in clinical or research settings and/or degree of clinical impairment. Of note, two dAD (one ldAD and one rdAD) and the PCA patients underwent neuropsychological assessment prior to their referral at Mayo Clinic. When available, standardized and raw scores were collected from outside neuropsychological reports and were integrated into the present documentation.

Raw scores were converted into age-adjusted scaled scores (SS). Age-adjusted norms for the WAIS, WMS, D-KEFS, WCST, CVLT, HVLT, WRAM-L and BVMT were computed using their respective standard manuals. Mayo Older Americans Normative Studies (MOANS) were used for the remaining tests [36, 51],M M [53, 65, 75]. As MOANS norms are only available for ages down to 56, the youngest age bracket (56–60) was used to standardize scores for patients younger than 56. Severity ranges were above average (SS 12–14); average (SS 9–11); below average (SS 7–8); borderline (SS 5–6); moderately impaired (SS 3–4); severely impaired (SS 1–2).

Neuropsychological batteries included a combination of the tests listed in Table 1 and covered a wide variety of cognitive domains including cognitive flexibility, inhibition, verbal and visuospatial working memory, verbal and visual episodic memory, abstract verbal reasoning, word and cultural knowledge, verbal fluency, confrontation naming, visuoconstruction abilities, and visuospatial reasoning. Since these cognitive domains were measured using different combinations of tests across patients, composite scores were computed for each cognitive domain. This was done by averaging SS of tests used to assess each domain. Verbal and visual episodic memory composite scores were divided according to immediate recalls, delayed recalls, and recognition scores. In-text description of neuropsychological results refers to cognitive domains rather than specific tests, unless specified otherwise.

Table 1.

Neuropsychological tests used across patients

Cognitive domain	Test	References
Cognitive flexibility	Trail making test (TMT)	[74]
	D-KEFS: design fluency	[18]
	Wisconsin card sorting test (WCST; perseverative errors)	[27]
	Stroop test: switching condition	[18]
Inhibition	Stroop test: inhibition condition	[76]
Verbal working memory	WAIS-III/WAIS-IV: digit span, arithmetic, letter-number sequencing	[84, 85]
	WMS-III: mental control	[86]
Visuospatial working memory	WMS-III: spatial span	[86]
Verbal episodic memory	Rey auditory verbal learning test (RAVLT)	[67]
	California verbal learning test short form (CVLT)	[88]
	WMS-III logical memory I and II	[86]
	Hopkins verbal learning test (HVLT)	[8]
	Wide range assessment of memory and learning (WRAML)	[72]
Visual episodic memory	WMS-III visual reproduction I and II WMS-III faces I and II	[86]
	Brief visuospatial memory test (BVMT)	[9]
Abstract verbal reasoning	WAIS-III/WAIS-IV: similarities	[84, 85]
Word and cultural knowledge	WAIS-III/WAIS-IV: vocabulary, information	[84, 85]
Verbal fluency	Animal and phonemic fluency	[78]
Confrontation naming	Boston naming test (BNT)	[44]
Visuoconstruction abilities	WAIS-III/WAIS-IV: block design	[84, 85]
	Rey-Osterrieth complex figure copy (ROCF)	[63]
Visuospatial reasoning	WAIS-IIIWAIS-IV: picture completion, matrix reasoning, visual puzzles	[85]

Results

Summary results for CSF AD biomarkers can be found in Table 2 and neuropsychological results at initial visit are displayed in Table 3. Table 4 describes clinical and imaging features supporting the differential diagnosis of each dAD case with the lvPPA and PCA cases as well as other common dementia syndromes (AD, bvFTD) at initial presentation. Figure 1 highlights multimodal imaging (FDG–PET, amyloid-PET, tau-PET), along with asymmetry indexes and SUVRs. Although most patients had serial PET images, only one set of each modality is displayed per patient to avoid redundancy. FDG–PET images that were acquired the closest to the initial clinical visit are displayed, alongside with the closest set of year-matching amyloid-PET and tau-PET images. Figure 2 displays longitudinal neuropsychological scores for dAD, PCA and lvPPA patients.

Table 2.

CSF AD biomarkers

Patient	Days between first clinical visit and lumbar puncture	Aβ42	P-Tau	ATI	Interpretation
ldAD1	6	415.25	39.3	0.83	Borderline*
ldAD2	6	629.75	168	0.32	AD
rdAD1	16	262.8	62.85	0.44	AD
rdAD2	2	325.2	59	0.45	AD
bpdAD1	6	504.05	58.5	0.89	Borderline*
bpdAD2	N/A		N/A	N/A	N/A
lvPPA	6	416.85	118.25	0.4	AD
PCA	N/A		N/A	N/A	N/A

^*Tau positivity in these patients was demonstrated through tau-PET imaging (see Fig. 1). CSF results are expressed in pg/ml

CSF Cerebrospinal fluid, AD Alzheimer’s disease, ldAD left-predominant dysexecutive Alzheimer’s disease, rdAD right-predominant dysexecutive Alzheimer’s disease, bpdAD biparietal-predominant dysexecutive Alzheimer’s disease, lvPPA logopenic variant of primary progressive aphasia, PCA posterior cortical atrophy, ATI Aβ₄₂–Tau index

Table 3.

Summary of neuropsychological results by cognitive domain at initial visit

Patient	Cog Flex	Inh	Verbal WM	VP WM	Verbal imm recall	Verbal delayed recall	VerbalRecog	Visual imm recall	Visual delayed recall	VisualRecog	Abstract verbal reasoning	Word and cultural knowledge	Verbal fluency	Naming	VC	VSP reasoning
IdAD1	4.5	2	4	–	5	2.5	6	2	1	5	6	9.5	5	5	3.5	6
ldAD2	1*	1	4*/2	–	5*/2.5	2	7.5	7*	–	–	7*	8*	2.5	7	13*/7	6*
rdADl	1	3	6	5	9.3	5.5	12	10	7	9	7	8	9		4	4.5
rdAD2	1.8	1	6	–	3	2.5	7	1	2	4	7	–	7.5	5	5	–
bpdADl	6.5	7	9	–	4.7	1.5	8	5	5	–	9	9.5	8.5	7	7.5	10
bpdAD2	5	8	10	–	3.7	2.5	3	7	2	–	–	10	6.5	12	8	–
lvPPA	6	2	6	–	7	6	6	–	–	–	–	–	4	–	8	–
PCA	2	2	8	–	8	12	10	–	–	–	–	–	10	–	2.5	–

^*Scores obtained from an outside neuropsychological assessment performed 2 years prior to referral at Mayo. Scores are expressed in scaled scores (SS)

ldAD = left-predominant dysexecutive Alzheimer’s disease; rdAD = right-predominant dysexecutive Alzheimer’s disease; bpdAD = biparietal-predominant dysexecutive Alzheimer’s disease; lvPPA = logopenic variant of primary progressive aphasia; PCA = posterior cortical atrophy; Cog Flex = Cognitive flexibility; Inh = Inhibition; WM = Working memory; VSP = Visuospatial; Imm = Immediate; Recog = Recognition; VC = Visuoconstruction

Table 4.

Clinical and imaging features supporting differential diagnosis between dAD cases and the lvPPA and PCA cases as well as common dementia syndromes (AD, bvFTD) at initial presentation

	AD		lvPPA		PCA		bvFTD
	Clinical	Imaging	Clinical	Imaging	Clinical	Imaging	Clinical	Imaging
ldAD1	Early EF > EM impairment	FDG-PET frontal involvement;	Early EF > language impairment; faster rate of clinical progression	FDG-PET parieto-frontal involvement; lesser degree of lateralization	Early EF > visuospatial & primary visual impairment; faster rate of clinical progression	FDG-PET parieto-frontal involvement; sparing of primary visual cortex	Absence of prominent behavioral features	FDG-PET parietal involvement, including the posterior cingulate & precuneus
ldAD2	Early EF > EM impairment; young onset	FDG-PET frontal involvement; preserved hippocampus on MRI	Early EF > language impairment; faster rate of clinical progression	FDG-PET parieto-frontal involvement; lesser degree of lateralization	Early EF > visuospatial & primary visual impairment; faster rate of clinical progression	FDG-PET parieto-frontal involvement; sparing of primary visual cortex	Absence of prominent behavioral features	FDG-PET parietal involvement, including the posterior cingulate & precuneus
rdAD1	Early EF > EM impairment; young onset	FDG-PET frontal involvement;	Early EF > language impairment; faster rate of clinical progression	FDG-PET parieto-frontal involvement; right hemisphere involvement; lesser degree of lateralization	Early EF > visuospatial & primary visual impairment; faster rate of clinical progression	FDG-PET parieto-frontal involvement; sparing of primary visual cortex	Absence of prominent behavioral features in the early disease stages	FDG-PET parietal involvement, including the posterior cingulate & precuneus
rdAD2	Early EF > EM impairment; young onset	FDG-PET frontal involvement;	Early EF > language impairment; faster rate of clinical progression	FDG-PET parieto-frontal involvement; right hemisphere involvement; lesser degree of lateralization	Early EF > visuospatial & primary visual impairment; faster rate of clinical progression	FDG-PET parieto-frontal involvement; sparing of primary visual cortex	Absence of prominent behavioral features in the early disease stages	FDG-PET parietal involvement, including the posterior cingulate & precuneus
bpdAD1	Early EF > EM impairment; young onset	FDG-PET parietal > temporal involvement; preserved hippocampus on MRI	Early EF > language impairment	FDG-PET parietal > temporal involvement; equally bilateral involvement	Early EF > visuospatial & primary visual impairment	FDG-PET parietal involvement; sparing of primary visual cortex	Absence of prominent behavioral features	FDG-PET parietal involvement, including the posterior cingulate & precuneus
bpdAD2	Early EF > EM impairment; young onset	FDG-PET parietal > temporal involvement; preserved hippocampus on MRI	Early EF > language impairment	FDG-PET parietal > temporal involvement; equally bilateral involvement	Early EF > visuospatial & primary visual impairment	FDG-PET parietal involvement; sparing of primary visual cortex	Absence of prominent behavioral features in the early disease stages	FDG-PET parietal involvement, including the posterior cingulate & precuneus

Fig. 1.

Multimodal imaging for dAD, lvPPA and PCA patients. Patients are separated by different rows, with lAD patients on the upper part of the figure, rdAD patients in the upper middle, bpdAD patients in the lower middle, and the lvPPA and PCA patients in the lower part. Within each column, FDG–PET (left), PiB-PET (middle) and AV1415-PET (right) findings are displayed for each patient, with regional SUVR Z-scores scales at the top of each column. Global composite SUVRs are displayed for PiB-PET and AV1451-PET images, except for rdAD2 for whom PiB-PET imaging was not available. Asymmetry indexes are displayed below each FDG–PET image, with a positive index indicating left predominance in hypometabolism pattern and a negative index indicating right predominance. ldAD left-predominant dysexecutive Alzheimer’s disease, rdAD right-predominant dysexecutive Alzheimer’s disease, bpdAD biparietal-predominant dysexecutive Alzheimer’s disease, lvPPA logopenic variant of primary progressive aphasia, PCA posterior cortical atrophy, PiB Pittsburgh compound B

Fig. 2.

Summary of longitudinal clinical and neuropsychological scores that best discriminated between dAD subtypes and between dAD and other atypical variants of AD. Visit 1 on the X axis represents either initial visit at Mayo Clinic or at an outside neuropsychology clinic, aside from the WCST for which scores have only been collected at initial visit at Mayo Clinic. Scores are expressed in scaled scores (SS), aside from the STMS for which raw scores can range from 0 to 38. ldAD left-predominant dysexecutive Alzheimer’s disease, rdAD right-predominant dysexecutive Alzheimer’s disease, bpdAD biparietal-predominant dysexecutive Alzheimer’s disease, lvPPA logopenic variant of primary progressive aphasia, PCA posterior cortical atrophy, STMS short test of mental status, WCST Wisconsin card sorting test

Case 1—ldAD1

The first ldAD case was a right-handed, 66-year-old female with 18 years of education. She was referred for a 1-year history of progressive cognitive difficulties. Upon interview, she struggled to provide specific examples of her difficulties, but reported being more forgetful and inattentive, which caused her to retire early and quit driving. She also mentioned that her husband assists her with performance of many tasks. This is more suggestive of executive dysfunction as the cause for her functional impairment rather than episodic memory. Neurological examination was unremarkable aside from points lost on the STMS (see below). She had no family history of dementia. She was followed-up for 1 year and 6 months. Over the course of the follow-up, she noticed mild disinhibition, word-finding problems, and increased difficulties with performing tasks involving executive control and tracking conversations. Neurological examination remained stable. Medication included memantine (prescribed at her local institution), donepezil, vitamin E and levothyroid.

Her MRI performed at the initial visit showed ventriculomegaly and revealed bilateral hippocampal atrophy, and a follow-up MRI conducted a year later showed no significant change. Of note, ventriculomegaly is radiological feature associated with normal pressure hydrocephalus [26]. It is, however, doubtful whether this condition contributed to the clinical profile this patient did not present with clinical features associated with this condition, such as gait imbalance or incontinence. In addition, ventriculomegaly in the setting of a cortically based neurodegenerative clinical syndrome with abnormal FDG–PET is more suggestive of ex-vacuo ventriculomegaly [79]. Consistent with this interpretation, an FDG–PET performed at initial visit revealed hypometabolism in frontal, parietal, temporal and lateral occipital lobes, as well as within the posterior cingulate and precuneus. The asymmetry index for this scan was − 0.97, indicating a predominance of hypometabolism towards the left hemisphere for cortical areas. Hypometabolism was also observed in the cerebellar areas bilaterally, although to a greater extent in the right hemisphere. A follow-up FDG scan acquired 14 months later highlighted a similar although more severe pattern of hypometabolism. One set of amyloid-PET and tau-PET images were acquired 5 months following her initial visit. Amyloid-PET scan showed widespread PiB uptake in heteromodal association cortices (posterior cingulate, medial prefrontal cortex, angular gyrus) with sparing of sensorimotor areas. Tau-PET scan showed AV1451 uptake mostly in posterior parietal and occipital areas bilaterally, and to a lesser extent in the frontal lobes, with a slight predominance towards the left hemisphere.

Her score on the STMS at initial visit was 18/38. A neuropsychological evaluation performed at initial visit revealed varying degree of cognitive flexibility impairment. More precisely, the Trail Making Test B was severely impaired and performance on the Wisconsin Card Sorting Test was moderately impaired for correct response and category completed, but with perseverative errors in the average range. These discrepancies might be explained by poor working memory capacities, given that performance on tasks measuring verbal working memory was moderately to severely impaired. Inhibition was severely impaired. Scores on immediate and delayed recalls were lower for visual episodic memory (severe impairment) than for verbal memory (borderline and moderate impairment, respectively), while recognition was borderline in both cases (with four false positives for verbal episodic memory). Animal fluency was severely impaired and confrontation naming was borderline. Verbal abstract reasoning was also borderline. Measures of visuospatial reasoning and visuoconstruction abilities were in the borderline range, aside from the ROCF’s copy which was severely impaired. The remainder of the assessments was in the below average or average range.

Case 2—ldAD2

The second ldAD case was a right-handed, 55-year-old female with 12 years of education. She was referred with a 5-year history of progressive cognitive difficulties. She initially received a diagnosis of multi-domain mild cognitive impairment (MCI) in outside settings 2 years prior to her referral at Mayo. The first major symptom she reported was being unable to learn a new software program at work which was highly unusual for her and not attributable to episodic memory impairment. Inability to perform and learn novel tasks under executive control was considered the likely etiology for this significant functional impairment. She also experienced word-finding issues and reported being more forgetful in day-to-day life (e.g., forgetting where she put things at home, forgetting conversations). Her husband also noted increased mild rigidity in her personality and lack of empathy, but these features were not at the core of the clinical profile. Neurological examination revealed difficulty with the Luria motor series. She had no family history of dementia. Her medication included acetaminophen, escitalopram, levothyroid, naproxen, omeprazole and oxybutynin (which was changed to trospium following initial visit). Donepezil was initiated shortly after her initial visit once AD pathophysiology was confirmed. Follow-up was performed annually for 1 year and 4 months. Over this period, her cognitive difficulties increased significantly to the point, where she had to retire from work. Follow-up examination revealed increased impairment on the Luria series, difficulties in spelling and calculation (i.e., acalculia) and some left/right confusion without finger agnosia.

This patient had two sets of MRI and PET scans. MRI at initial visit highlighted mild white matter changes associated with chronic small vessel disease, and the follow-up MRI showed mild biparietal atrophy. Both FDG scans revealed hypometabolism in temporal and parietal lobes, including the posterior cingulate and precuneus. Milder, yet significant hypometabolism was also found in frontal areas bilaterally. Asymmetry index for the first FDG scan was −0.44, indicating a predominance of hypometabolism towards the left hemisphere for cortical areas. Mild hypometabolism was also observed in the cerebellar areas, mostly in the left hemisphere. Amyloid-PET at initial presentation and follow-up showed widespread PiB uptake across the neocortex, mostly concentrated in areas forming the heteromodal association cortex with sparing of sensorimotor areas. Tau-PET revealed widespread AV1451 uptake across the neocortex, with slightly higher uptake in the left frontal lobe.

On formal cognitive testing, she scored 17/38 on the STMS at initial visit, and performance declined to 16/38 and 10/38 on the two following annual examinations, respectively. A neuropsychological assessment performed 2 years prior to her referral revealed severe cognitive flexibility impairment and moderate verbal working memory impairment. Immediate and delayed recall on tasks of episodic memory ranged from borderline to below average. Performance on tasks assessing visuospatial reasoning ranged from borderline to below average. A neuropsychological assessment performed at her initial visit at Mayo highlighted severely impaired performance on tasks of inhibition and verbal working memory. Immediate and delayed recalls on a verbal memory task were severely impaired, with below average recognition, although thirteen false positives were observed, indicating a high bias towards positive answers. Performance on fluency tasks were moderately to severely impaired, while performance on a task of confrontation naming was below average. The remainder of the assessments was in the below average or average range.

Case 3—rdAD1

The first rdAD case was a right-handed, 59-year-old male with 14 years of education. He was referred for a 4-year history of progressive difficulty with task performance, short-term memory, navigation, and time perception. The first symptom he noticed was difficulty with wiring a light switch at home, which he typically would be able to do without a problem; however, he had difficulty visualizing the connection and was unable to complete this task. He would also forget the place he just left (e.g., store), get lost while driving in unfamiliar places, and get confused with the time at which he was supposed to start his work shifts. These symptoms were suspected to be related to visuospatial executive dysfunction and supporting formal testing of cognitive function was completed to further evaluate this. Neurological examination revealed disproportionate difficulties with visuospatial tasks (i.e., clock and cube drawing), as well as difficulties with complex calculations. He had a strong family history of dementia, where both of his parents and several aunts and uncles were diagnosed with late-onset dementia. Clinical follow-up was performed annually for 2 years and 9 months. Cognitive symptomatology progressed to the point, where he could no longer perform complex tasks under executive control (e.g., temporal sequencing) in addition to increased visuospatial and languages impairment and was unable to work or complete cognitive testing. The neurological examination conducted a year following his initial visit highlighted increased difficulties with the Luria motor series, as well as overlying and Navon figures. A year later, he was no longer capable of performing most tests, had impaired semantic knowledge (i.e., famous faces and buildings), and manifested slight hypomimia. He also developed neuropsychiatric features, such as delusions, paranoia, hallucinations, and depressive symptoms. His wife reported increased irritability and verbal aggressivity. She also reported decreased hours of sleep and episodes of nocturnal confusion in the absence of dream enactment behavior. His medications included fish oil, acetaminophen. Memantine was initiated but eventually stopped due to the intolerance of side effects.

This patient underwent four annual MRI scans. Overall, MRI findings revealed mild cerebral volume loss with bilateral decreased hippocampal size and mild to moderate leukoaraiosis in the subcortical and periventricular white matter, with no significant interval change. Three serial FDG–PET scans were acquired over follow-up. The initial FDG scan showed hypometabolism bilaterally in the parieto-frontal areas. Asymmetry index was 0.74, indicating a predominance of hypometabolism towards the right hemisphere. Subsequent FDG scans revealed greater involvement of the posterior cingulate and precuneus bilaterally, as well as cerebellar areas. Three sets of amyloid-PET and tau-PET images were acquired over the same period. PiB uptake was distributed throughout the neocortex with more prominent signal in heteromodal association cortex with sparing of sensorimotor areas, and this pattern did not significantly change between scans. All three tau-PET scans showed AV1451 uptake in temporal lobes bilaterally and parieto-frontal areas bilaterally with higher uptake in the right hemisphere.

He scored 27/38 on the STMS, and his performance was similar (28/38) the following year, before declining to 24/38 the year after that. Neuropsychological results showed prominent severe cognitive flexibility impairment and moderate inhibition impairment. Performance on a measure of visuospatial working memory was lower (borderline) than for a measure of verbal working memory (below average). A similar pattern was observed for episodic memory, where performance for delayed recalls was lower with visual material (borderline) than for verbal material (below average). Performance on measures of verbal fluency was average and performance on a measure of confrontation naming was borderline. Performance on measures of visuoconstruction abilities varied between borderline and moderate impairment, except for ROCF copy which was severely impaired. Of note, this patient also underwent a brief neuropsychological assessment 1 year following his initial visit in outside settings. This assessment highlighted impairment on visual episodic memory measures but average performance on measures of verbal episodic memory. The remainder of the assessments was in the below average or average range.

Case 4—rdAD2

The second rdAD case was a right-handed, 48-year-old female with 16 years of education. She received a diagnosis of pseudo-dementia by a local neurologist a month prior to her referral. She was referred to Mayo for a dysexecutive syndrome that progressed in the recent years which was accompanied by depressive symptomatology. She observed increased difficulties while performing tasks under executive control. For instance, she experienced difficulties managing finance and having problems with complex thinking and task performance in general (e.g., temporal sequencing of steps while performing a given task). Neurological examination revealed word finding issues and difficulties with repetition, and she could not complete the Luria motor series. She had a sister who was diagnosed with early onset Parkinson. Her medications were mostly natural products including vitamin D3, C, B4 and B12, fish oil, iron and omega 3. Memantine and vitamin E were initiated shortly following her examination after AD pathophysiology was confirmed. She was followed for 1 year and 4 months. Over this period, cognitive symptomatology progressed significantly (i.e., increased difficulties with task performance, forgetfulness and trouble managing medication). A follow-up examination revealed the emergence of ideomotor and ideational apraxia. She also developed episodes of delusion and visual hallucinations.

An MRI scan revealed diffuse cerebral atrophy, including bilateral hippocampi atrophy. Nonspecific white matter changes compatible with mild chronic small vessel ischemic changes were also observed. An FDG–PET scan revealed considerable hypometabolism in parietal, temporal, and frontal areas. Significant hypometabolism was also noted in the posterior cingulate gyrus. Asymmetry index was 0.65, indicating a predominance of hypometabolism towards the right hemisphere for cortical areas. Mild hypometabolism was also observed in the cerebellar areas bilaterally. A tau-PET scan revealed elevated AV1451 uptake across the neocortex.

She scored 19/38 on the STMS, and 16/38 at follow-up 1 year later. Neuropsychological results showed severe impairment on tasks of cognitive flexibility and inhibition, while verbal working memory performance was in the borderline range. Immediate and delayed recalls on a verbal episodic memory task were moderately impaired. Performance was lower for visual episodic memory, where immediate and delayed recalls were severely impaired, and recognition was moderately impaired. Performance on a measure of animal fluency was in the moderate impairment range. Performance on a measure of visuoconstruction abilities was borderline (WAIS-IV block design). The remainder of the assessment was in the below average or average range.

Case 5—bpdAD1

The first bpdAD case was a right-handed, 58-year-old woman with 16 years of education. She was referred by a local neurologist, who initially diagnosed her with an adjustment disorder with depressive and anxious mood. She reported difficulties using her passwords or follow directions while driving under executive control. She also mentioned increased forgetfulness (e.g., forgets conversations, needs to write things down), but these difficulties appeared to be related to executive dysfunction rather than pure episodic memory impairment as per neuropsychological results. She mentioned her cognitive symptomatology has been present for the past years but was unable to specify a clear time-line of symptoms. She, however, noted symptoms worsened in the past year due to a concussion without loss of consciousness. She had no family history of dementia. Her medications included fish oil, a multivitamin and vitamin D. Donepezil was initiated shortly after her first visit following confirmation of AD pathophysiology. She was followed-up for 3 years and 3 months. Over this period, her cognitive symptomatology progressed significantly to the point, where she could no longer work, drive, or manage her medications. On her last visit, she failed the Luria motor series. She also developed visual and somatic hallucinations (i.e., sensations of sharp teeth and strings in her mouth) and episodes of delusions.

This patient had four serial MRIs. Her first MRI performed at an outside facility showed minimal nonspecific white matter changes possibly compatible with chronic small vessel disease. Subsequent MRIs revealed mild generalized cerebral volume loss, possibly greater in parietal areas, with no interval change between scans. Three FDG–PET scans were acquired over follow-up. The initial FDG showed moderate hypometabolism in bilateral temporoparietal areas and mild patchy bilateral frontal hypometabolism. There was also marked hypometabolism in the bilateral posterior cingulate regions. Asymmetry index was 0.15, indicating a very mild predominance towards the right hemisphere. Subsequent FDG–PET scans highlighted slightly more hypometabolism in these areas. Two sets of amyloid-PET and tau-PET were acquired over the last 2 years of follow-up. Amyloid-PET scans showed PiB uptake distributed throughout the neocortex with more prominent accumulation in heteromodal association cortex with sparing of sensorimotor areas. Tau-PET scans acquired over the same period revealed AV1415 uptake mostly concentrated in posterior parietal areas.

She scored 31/38 on the STMS, and performance increased to 34/38 the following year, before declining to 28/38 on her last visit. Neuropsychological assessment showed borderline impairment across tasks on measures of cognitive flexibility and inhibition, while performance on a verbal working memory task was average. Scores on immediate and delayed recalls on a verbal episodic memory task were in the borderline and severe impairment ranges, respectively. Both immediate and delayed recalls on a visual episodic memory task were in the borderline range. Two additional neuropsychological assessments were conducted annually over the follow-up, and mostly showed a decline in executive functioning, where cognitive flexibility was severely impaired and inhibition was moderately impaired. The remainder of the assessments was in the below average or average range.

Case 6—bpdAD2

The second bpdAD case was a right-handed, 58-year-old male with 13 years of education. He manifested a 2–3-year history of memory, calculation, and planning problems. As a carpenter he experienced difficulties measuring and planning tasks using executive control and performing mathematical operations. He often found himself repeating operations and had to double-check his own work, despite being able to do these types of tasks without difficulty in the past. He also needed to get increased bookkeeping support for his business. He was also noted to forget recent conversations, but this was not the most prominent or limiting aspect of his cognitive dysfunction. Neurological examination was unremarkable aside from points lost on the STMS (see below). His mother died from the complication of AD which was proven upon autopsy. His medications included aspirin, multivitamin, prilosec and vitamin E, and donepezil and memantine were eventually initiated. He was followed for 7 years and 11 months. Over this period, his cognitive symptomatology progressively worsened, and he developed behavioral disturbances, such as increased irritability, hoarding, disinhibition, and agitation, along with depressive symptoms. Neurological examinations performed over follow-up showed moderate difficulties with verbal comprehension, difficulties with the Luria motor series, ideomotor apraxia and mild parkinsonism. Dream enactment behavior suggesting probable REM-sleep behavior disorder (RBD) was observed 4 years after initial visit.

This patient had eight serial MRIs. The initial MRI showed white matter changes related to small vessel ischemic disease. Subsequent MRI examinations additionally revealed mild generalized cerebral and cerebellar atrophy, increased focal white matter signal in the left anterior temporal lobe, moderate to advanced hyperintensities in cerebral white matter and pons and hippocampal atrophy bilaterally. Six FDG–PET scans were acquired. They all showed hypometabolism in the temporoparietal areas bilaterally, including the posterior cingulate gyrus, which progressed over time. Asymmetry index was −0.10 at initial scan, indicating a very mild predominance towards the left hemisphere. Seven amyloid-PET scans were acquired, which initially showed PIB uptake in areas forming the default mode network with sparing of sensorimotor areas, before spreading to other cortical areas with disease progression. One tau-PET scan was performed 6 years following initial visit, which highlighted AV1451 uptake mostly concentrated in posterior parietal areas.

His score on the STMS was 31/38, and performance remained relatively stable or slowly declined for the four following years with scores of 32/38, 30/38, 28/38, 27/38, before declining in the last years of follow-up with scores of 20/38, 20/38 and 14/38. Neuropsychological results at initial visit showed performance in the below average and borderline range for tasks assessing inhibition and cognitive flexibility, respectively, while verbal working memory was average. Performance on measures of verbal and visual episodic memory (immediate and delayed recalls and recognition) ranged from moderate to severe impairment, except for visual immediate recall which was below average. Performance on a task of animal fluency was borderline. Seven additional neuropsychological assessments were conducted yearly following the initial visit. Over this period, performance on tasks measuring cognitive flexibility and inhibition declined to the severe impairment range. Verbal working memory performance remained relatively stable before declining to the borderline range at the last visit. Surprisingly, recognition on a verbal episodic memory task increased to the borderline range on the last visit when it was assessed, with five false positives. Performance on fluency tasks eventually declined to the severe impairment range. One measure of visuoconstruction ability (WAIS-IV Block Design) declined from below average to moderate impairment, whereas the ROCF copy declined to severe impairment.

This patient died 10 years following his initial visit. Upon post-mortem examination, he was found to have high AD neuropathological changes with A3, B3 and C3 scores [35]. He also had transitional limbic LBD, which could have explained the presence of parkinsonism and RBD over the course of the disease. He was also found to have moderate to severe amyloid angiopathy, cerebrovascular disease (moderate arteriolosclerosis, left inferior temporal gyrus subacute infarct, rarefied perivascular white matter in the basal ganglia, intraparenchymal hemosiderin), Alzheimer type II gliosis, aging-related tau astrogliopathy (ARTAG) and minimal TAR DNA-binding protein (TDP-43) amygdaloid lesions.

Case 7—lvPPA

The lvPPA case was a right-handed 64-year-old female with 14 years of education. She was referred for a 1-year history of progressive difficulties with word-finding and expressing her thoughts. Her neurological examination showed difficulties with repetition, naming, comprehension, and two-step commands. She was followed for 2 years and 4 months. Over the course of this follow-up, symptoms remained relatively stable. Her medication included alendronate, cyanocobalamin, hydrochlorothiazide, multivitamins, vitamin D3. Memantine was initiated, but eventually stopped due to undesired side effects.

Three MRI scans were acquired over follow-up. The initial MRI scan showed nonspecific volume loss in the left anterior temporal area and nonspecific multifocal white matter hyperintensities. Subsequent MRI scans did not reveal any significant change. One FDG–PET scan acquired at initial visit showed moderate hypometabolism in left temporo-parietal areas and mild hypometabolism in the left occipital lobe and right temporal lobe. Asymmetry index was −0.88, indicating a predominance towards the left hemisphere. One amyloid-PET scan acquired at initial visit showed a widespread neocortical PiB uptake concentrated in heteromodal association cortex with sparing of sensorimotor areas. The tau-PET scan revealed AV1451 uptake in temporo-parietal junctions as well as within the occipital lobes, with a clear predominance towards the left hemisphere.

Her score on the STMS was 28/38 and remained stable over her two following annual visits (i.e., 27/38 and 28/38). Only one neuropsychological assessment was conducted at initial visit. Performance on tasks measuring executive functions varied between borderline (cognitive flexibility) to severe impairment (inhibition) ranges. Immediate recall, delayed recall and recognition on a verbal episodic memory task ranged from below average to borderline range, with four false positives for recognition. Performance on a measure of phonemic fluency was moderately impaired. Results on tasks measuring visuoconstruction abilities varied between average (WAIS-IV block design) and borderline range (ROCF copy).

Case 8—PCA

The PCA case was a left-handed, 57-year-old female with 14 years of education. She was referred by a local neurologist for a 2-year history of difficulties with writing, reading, getting dressed, and navigating. She had retired and stopped driving 1 month before her initial visit due to her cognitive problems. Neurological examination revealed simultagnosia (i.e., impaired color discrimination, unable to name any of the Ishihara plates), but intact color naming and matching. She exhibited mild ideomotor apraxia bilaterally. She had no family history of dementia. Her medications at initial visit included donepezil, aspirin, calcium, fish oil, levothyroid, and multivitamins. She was followed for 4 years and 10 months. Over the years, symptoms remained relatively stable. Results from neurological examinations were mainly unchanged over follow-up.

This patient had one MRI performed in outside settings and five MRIs acquired annually at Mayo. The first MRI showed a mild degree of atrophy, most prominent in the occipital and parietal lobes bilaterally. Additional MRIs revealed a similar pattern of atrophy that slightly progressed over the years, in addition to mild leukoaraiosis. Two FDG–PET scans were acquired, one at initial visit and one at her last visit, which both revealed hypometabolism in the occipital lobes bilaterally, which extended to parietal and temporal lobes over follow-up. Asymmetry index for the first scan was −0.51, indicating a predominance towards the right hemisphere. Four amyloid-PET scans were acquired. Her initial amyloid-PET scan revealed PiB uptake mostly concentrated in posterior parietal areas (i.e., posterior cingulate, precuneus), whereas following scans highlighted uptake in heteromodal association cortex with sparing of sensorimotor areas. Five tau-PET scans were acquired. The first tau-PET scan revealed a pattern that was mostly concentrated in occipital and posterior parietal areas bilaterally, whereas subsequent scans showed accumulation progressively extending to frontal areas.

Her score on the STMS was 28/38 and remained relatively stable over the two following years at 30/38 and 29/38, respectively, before declining to 22/38, 22/38 and 19/38 on following annual examinations. Of note, a neuropsychological assessment performed prior to her referral revealed severe visuospatial deficits, which impacted any task using visual material, whereas other cognitive domains (language, episodic memory) were preserved. All three neuropsychological assessments performed at Mayo showed moderate to severe impairment across tasks using visually presented material (cognitive flexibility, inhibition, visuoconstruction abilities), whereas performance on verbal episodic memory and language were in the average to above average ranges.

Discussion

We described a series of 6 clinical cases highlighting the clinico-radiological heterogeneity in a progressive dysexecutive syndrome due to AD pathology. Although all dAD patients presented with predominant executive functioning impairment, differences were observed between dAD subtypes in terms of clinical features, neuroimaging findings, and neuropsychological profiles. dAD subtypes were nonetheless discernible from other forms of atypical AD phenotypes. These findings have clinical relevance for diagnosis, prognosis, and symptom management in individuals who develop different subtypes of dAD.

Subtypes of dAD patients were identified according to their pattern of FDG–PET hypometabolism, which resulted in three categories: ldAD, rdAD and bpdAD. ldAD and rdAD patients were defined by predominant hypometabolism in the left and right parieto-frontal network, respectively, while bpdAD patients were defined by hypometabolism mostly concentrated in posterior parietal areas with relative sparing of frontal regions. Molecular imaging revealed a relative tau deposition as captured by AV1451 relatively overlapped with patterns of FDG–PET hypometabolism in each subtype, whereas patterns of amyloid deposition did not differ across dAD subtypes. This is consistent with previous studies demonstrating that tau accumulates along large-scale cognitive networks underlying phenotypic variability and overlaps with patterns of neurodegeneration across AD phenotypes [19], 22, 23, 39, 42, 61, 62, 70, 83. It is worth noting that cerebellar hypometabolism was observed in ldAD2 and both rdAD cases, which was generally observed in the contralateral hemisphere to the cortical pattern of hypometabolism. Considering the decussation of axonal tracts in the superior cerebellar peduncles, this pattern likely reflects diaschisis caused by the dysfunction of functional circuits linking posterior cerebellar areas and prefrontal regions (i.e., the dorsolateral prefrontal cortex), which support executive functions [41, 47, 69].

The heterogeneity observed in radiological findings across dAD subtypes is reflected in their respective clinical and neuropsychological profiles. One main cognitive feature of ldAD patients is the degree to which verbal working memory and verbal fluency were impaired in these patients, which was similar to what was observed in the lvPPA patient. This is consistent with the pattern degeneration of the parieto-frontal network in the left hemisphere, which is thought to underlie the “phonological loop” component of working memory [21, 59, 60]. According to Baddeley’s model of working memory, the phonological loop is a repository allowing one to temporarily retain and manipulate speech-like elements in mind relevant to an ongoing task [3, 4]. Interestingly, ldAD patients generally showed greater impairment on verbal working memory and verbal fluency tasks compared to confrontation naming. Thus, verbal fluency impairment observed in ldAD might be better explained by dysfunction of the phonological loop rather than a pure primary language impairment per se, given that confrontation naming does not involve working memory. In contrast, lvPPA patients, in addition to diminished verbal working memory, also manifest considerably impaired performance on confrontation naming tasks [49, 77]. Thus, confrontation naming tasks might be useful in differentiating ldAD from lvPPA.

rdAD patients generally exhibited lower performance on tasks using visual material compared to verbal material. For instance, spatial span was lower than digit span for the rdAD patient for which both measures were available, and visual episodic memory recognition was lower than verbal recognition in both rdAD patients. Moreover, the discrepancy between performance on measures of visuospatial and visuoconstruction abilities and those tapping into language was generally greater in rdAD patients than for other dAD subtypes. This is consistent with the pattern of selective degeneration of the right hemispheric parieto-frontal network specifically seen in rdAD patients. This portion of the parieto-frontal network is thought to support visuospatial working memory, also referred to as the “visuospatial sketchpad” from Baddeley’s working memory model, which temporarily handles and manipulates visual and spatial information while performing a task [2, 3]. Impairment of the visuospatial component of working memory in rdAD probably had an impact on the impairment involving the mental manipulation and integration of visual material seen across cognitive domains. Moreover, both rdAD patients committed a lot of perseverative errors on the WCST (severe impairment). Evidence suggests that integrity of the parieto-frontal network would be essential to perform this task [31, 57]. Interestingly, findings from a study using functional MRI showed that the right ventromedial and dorsolateral prefrontal areas were involved in visuospatial working memory operations, whereas the temporo-parietal junction bilaterally was essential for error detection [50]. This spatial pattern of activation is highly reminiscent of the pattern of parieto-frontal network degeneration specifically observed in rdAD patients, which might explain the high rate of perseverative errors on the WCST in these patients.

bpdAD patients generally presented with a mild pattern of cognitive dysfunction. Most affected cognitive functions were cognitive flexibility and inhibition, which might have negatively impacted performance on immediate and delayed episodic memory recalls. Other cognitive domains, including working memory, were relatively preserved. This is consistent with a body of literature highlighting the putative and independent role of parietal areas, notably the inferior parietal lobule, a core aspect of executive functioning, such as task-switching, set-shifting and inhibition [82, 89], whereas working memory processes rely on both frontal and parietal areas [21, 56, 64]. Moreover, STMS scores remained relatively stable or slowly declined over the first years of follow-up for both bpdAD patients. Hence, this dAD subtype might be characterized by a selective degradation of cognitive flexibility and inhibition and a slow clinical progression.

It is also noteworthy that all rdAD and bdpAD patients developed neuropsychiatric symptoms over disease course (i.e., depressive symptoms, hallucinations, delusions, paranoia), whereas none of the ldAD patients did. Some studies and systematic reviews suggest a relationship between right hemisphere dysfunction and atrophy, delusions [12, 14] and hallucinations [11]. These symptoms were documented in both rdAD patients. In bpdAD, the underpinnings of such symptoms remain unclear, as the literature on the involvement of parietal areas in neuropsychiatric symptoms in dementia is scarce. However, one bpdAD patient suffered from tactile hallucinations, which could be related to abnormal metabolism extending to sensory areas, but this remains speculative. An alternative although not mutually exclusive possibility for the occurrence of hallucinations would be the presence of co-pathologies, such as DLB, which is often observed in AD [33, 73]. Indeed, visual hallucinations are seen in up to 80% of patients with DLB pathology [29] compared to 9–16% in patients with a clinical diagnosis of AD [6, 91]. However, this again remains speculative given that only one bpdAD patient underwent post-mortem examination which revealed transitional DLB restricted to limbic areas. This relatively low amount of DLB pathology might thus reflect end-stage AD rather than being a driver of clinical symptoms. It is also worth emphasizing that none of the dAD patients exhibited any of the core features of DLB at clinical presentation [55]. Overall, rdAD and bpdAD might be at a higher likelihood of developing neuropsychiatric symptoms, but additional investigations in larger groups will be required to yield prevalence estimates and uncover the neural correlates of such symptoms.

The existence of different dAD subtypes suggested by this clinical case series has relevant clinical implications. It is essential to identify clinical, cognitive, and imaging features of different dAD subtypes to accurately classify patients, predict and manage symptoms, and estimate disease progression. When available, FDG–PET should be considered in clinical settings to differentiate dAD subtypes, given the power of this tool to assess network integrity across the neurodegenerative diseases spectrum and categorize AD phenotypes [40], 48. Referral to neuropsychology and speech pathology can also be useful to characterize the nature and extent of cognitive impairment, and thus provide critical information for diagnostic purposes. Accurate subtype classification is also crucial for symptom treatment and/or management. For instance, ldAD cases presented with severe impairment on tasks tapping into verbal fluency, which can be managed through speech and language therapy [16]. Moreover, neuropsychiatric symptoms such as delusions, hallucinations, paranoia, and depressive symptoms were present in rdAD and bpdAD patients. This can prompt for adjustments in pharmacological treatment as well as counselling family and caregivers about the manifestation and occurrence of such symptoms. Patients with bpdAD are seemingly characterized by a slow progression, which could help guide patients and families in planning for the future. It has been proposed that expected clinical outcomes in clinical trials should be adapted to symptomatology specific to different AD phenotypes [25]. The same logic should apply to dAD subtypes given the considerable inter-individual variability in clinical features and cognitive domain impairment. Variability in patterns of parieto-frontal network dysfunction should also be considered in the design of potential nonpharmacologic treatments, such as transcranial magnetic stimulation (e.g., see [45], that target large-scale networks.

However, it remains important to keep in mind the similarities among dAD subtypes. Despite the within-syndrome clinico-radiological heterogeneity described above, all dAD patients presented with cognitive complaints pertaining to short-term memory and complex thinking, showed predominant executive functioning compared to other cognitive domains, and exhibited selective neurodegeneration of the parieto-frontal network. Thus, although differences highlighted in this study are clinically meaningful and essential to disentangle the AD syndromic variability, different dAD subtypes should nonetheless be grouped under the label of dAD, and not be considered as independent phenotypes. They should also not be considered as subtypes of other atypical AD variants, which is supported by the comparison between dAD patients and lvPPA and PCA patients. While the lvPPA patient showed a similar degree of impairment on verbal working memory and verbal fluency tasks compared to ldAD patients, executive dysfunction was not at the core of this patient’s clinical profile. In addition, the pattern of FDG–PET hypometabolism in the lvPPA patient, although asymmetrically left-sided, specifically targeted the language network [87] rather than the left parieto-frontal network as observed in the ldAD patients. Similarly, the PCA patient presented with core visual and visuospatial impairment, whereas executive functioning was relatively spared when measured through verbal tasks. Moreover, although this patient showed prominent FDG–PET hypometabolism in posterior areas, a much greater occipital involvement was observed compared to bpdAD patients. It is also worth discussing particular clinical features of the rdAD1 case such as self-reported visuospatial problems as well as difficulties with Navon figures and cube and clock drawing documented through neurological examination, which can raise questions about a PCA syndrome. However, the cognitive symptomatology of this patient was primarily dysexecutive in nature rather than pertaining to primary visual capacities. This is also supported by neuropsychological findings, where tasks relying on processing of visual stimuli (e.g., naming, visuoconstruction) were impaired to a lesser extent than those tapping into executive functions (borderline versus severe impairment for cognitive flexibility and inhibition). Moreover, inspection of FDG–PET hypometabolism revealed a relative sparing of primary visual cortices compared to the PCA patient, in addition to a much more prominent parieto-frontal involvement.”

It is important to mention that prior to the characterization of dAD, early onset cases mimicking an amnestic syndrome (e.g., subjective sentiment of increased forgetfulness, poor performance on immediate and delayed recalls) and with positive biomarkers for AD pathology were almost universally thought to suffer from the canonical, typical amnestic variant of AD but with a precipitated symptom onset [20]. However, in addition to subjective reports of impaired task performance under executive control, all dAD patients in this report generally showed much lower performance on measures of executive functioning compared to measures of episodic memory, and generally worse performance on episodic memory free recalls compared to recognition. This contrasts what is seen in amnestic AD, where encoding/learning deficits dominate the cognitive profile from the early phase of the disease [30],Mary M [54], while executive dysfunction occurs in later stages [90]. In addition, FDG–PET images highlighted selective hypometabolism in the parieto-frontal network, which is specific to dAD comparatively to other AD variants, including the amnestic phenotype (Graff-radford et al., n.d.; [81]. Thus, although both dAD and amnestic AD share identical pathological substrates, this prompts clinicians to consider syndrome and etiology separately when it comes to formulate a diagnostic impression. It is also important to point out that dAD patients did not present with predominant behavioral and/or personality changes, which are clinical hallmarks of bvFTD. Even though both dAD and bvFTD are generally characterized by early onset of symptoms and executive dysfunction [5, 66], predominant deficits in core executive functions in the absence of behavioral/personality changes should suggest dAD as an etiological diagnosis rather than bvFTD.

Our results must obviously be considered in the light of limitations. Since this study is a clinical case series by design, only a handful of dAD patients that fit within our coarse clinical observations was presented. While clinically relevant differences are reported between dAD subtypes, our findings are pending replication by leveraging much larger data sets and data-driven statistical analyses suited to disentangle clinico-radiological variability in AD (e.g., see [28, 40, 80, 83]. Moreover, clinical examinations were not standardized and hence the presence of certain symptoms might have gone unexamined in some patients. Lack of standardization was also evident in neuropsychological assessments across patients. This might have introduced bias in the computation of demographically adjusted SS, since tasks assessing similar cognitive constructs can vary in terms of difficulty and normative methods. Although bpdAD patients seemingly show a slower rate of decline compared to ldAD and rdAD patients based on the present clinical case series, dAD patients have been followed-up for varying periods of time. It thus remains challenging to accurately compare rates of progression between dAD subtypes. However, a potential explanation for this observed difference is survival bias, where patients with slower cognitive decline might be assessed for longer periods of time. Related to this later point, estimating actual disease progression from a pathobiological standpoint was impossible given that not all patient presented at the same stage of the disease. This inevitably introduces variability in neuropsychological performance that is unrelated to dAD subtype categorization.

Conclusions

dAD is a recently defined syndrome characterized by a generally early onset, progressive and prominent decline of core executive functions that is tied to a selective degradation of the parieto-frontal network due to AD pathology [81]. This clinical case series suggests the existence of meaningful clinico-radiological heterogeneity within this syndrome. Different dAD subtypes present with distinct patterns of parieto-frontal network degeneration, which most likely underlies inter-individual variability in clinical features and cognitive impairment at onset and over the disease course. Additional investigations in larger samples will be essential to better characterize clinical and cognitive profiles associated with dAD subtypes and uncover biological mechanisms underlying this variability.