Amnestic Syndrome in Memory Clinics: Similar Morphological Brain Patterns in Older Adults with and without Alzheimer’s Disease

INTRODUCTION

Amnestic syndrome of the hippocampal type (ASHT), a frequent clinical presentation among patients evaluated in Memory Clinics, is characterized by a deficit in both free and cued verbal episodic memory recalls. In this context, the Free and Cued Selective Recall Reminding Test (FCSRT) was established as reference tool [1]. ASHT is a clinical criterion for typical Alzheimer’s disease (AD) and can be used to distinguish prodromal AD from other neurodegenerative conditions [2, 3]. However, recent findings suggest that ASHT is not specific to AD. A neuropathological study on patients consulting to tertiary care Memory Clinics has reported that one-third of patients with AD did not meet FCSRT criteria for ASHT, whereas half of individuals with pathologies other than AD presented with an amnestic syndrome [4]. Non-amyloid pathologies such as behavioral-variant frontotemporal lobar degeneration (FTLD), primary age-related tauopathy (PART), and limbic-predominant age-related TDP-43 encephalopathy (LATE) should be considered within the differential diagnosis of AD when evaluating patients with ASHT [4, 5].

Brain morphometry represents a promising tool to support the diagnosis of AD. While PET and cerebrospinal fluid (CSF) are the current recommended biomarkers for the diagnosis of AD pathology, limitations such as availability, cost, invasiveness, and adverse effects promote the investigation of alternate methods, such as brain morphometry derived from magnetic resonance imaging (MRI) [6, 7]. Manual segmentation of brain volumes is currently the gold standard, but automated methods are comparable to manual segmentation and help to address limitations such as time constraints and reliance on subjective judgement [8]. Yet, further investigations are required for the comparison of brain atrophy patterns, especially in the hippocampus, between AD and non-AD patients with ASHT [5].

The aims of this study were to 1) identify whether brain morphometry patterns including hippocampal volumes differ between amnestic older adults with and without AD as defined by AD CSF biomarkers, and 2) investigate the relationship between memory performance, hippocampal volumes, and AD CSF biomarkers. We hypothesized that 1) brain morphometry would be comparable between AD and non-AD amnestic patients, and 2) memory impairment would be explained by hippocampal atrophy. In the era of disease-modifying therapy specific to amyloid pathology, exploring non-invasive and non-expensive biomarkers for the diagnosis of AD will improve our management strategy.

MATERIALS AND METHODS

Ninety-two consecutive patients with ASHT (mean age 72.5±6.8 years old, 36 females) assessed at the Leenaards Memory Center of the Lausanne University Hospital between August 2013 and February 2023 were retrospectively selected. Inclusion criteria were 1) being referred to our Memory Clinic; 2) > 55 years of age; 3) presence of ASHT; 4) availability of comprehensive neuropsychological evaluation, CSF AD biomarkers, and structural magnetic resonance imaging (sMRI); 5) A + T+, A-T+, or A-T- AT classification based on CSF biomarkers (see below). A + T- patients were excluded because AD would represent an unlikely cause of their clinical presentation. Patients received a clinical diagnosis at a consensus case conference involving behavioral neurologists and neuropsychologists, based on international consensus guidelines [9]. ASHT was defined following the criteria of Sarazin, i.e., a score < 40/48 on the total recall of the FCSRT [3]. Research was conformed with the Declaration of Helsinki and approved by the local Institutional Review Board. All patients provided their informed consent to the reuse of data and had a consultation at the Leenaards Memory Center.

RESULTS

Thirty-two patients (35%) met the criteria for amnestic syndrome with Alzheimer’s pathology (“AD”), and 60 patients (65%) for amnestic syndrome in the absence of Alzheimer’s pathology (“non-AD”) (Table 1).

AD and non-AD were comparable in demographics, except for an increased number of females in the AD group (p = 0.045) and a trend for a lower education level in the non-AD group (p = 0.070). CSF biomarkers differed between groups, with higher t-Tau in AD. Patients in the non-AD group had heterogeneous diagnoses, including 18% cerebrovascular disease (n = 11), 15% alcoholic encephalopathy (n = 9), 10% FTLD (n = 6), 7% Lewy body dementia (n = 4), 3% normal pressure hydrocephalus (n = 2), 3% sleep apnea (n = 2), 3% chronic traumatic encephalopathy (n = 2), 3% toxic encephalopathy (n = 2), and 2% infectious encephalopathy (n = 1). Twenty non-AD patients (33%) were classified as suspected non-amyloid pathology (SNAP) [32]. According to clinical diagnoses, a subgroup of AD patients had possible mixed pathology with cerebrovascular disease (22%, n = 7), alcoholic encephalopathy (9%, n = 3), cerebral amyloid angiopathy (6%, n = 2), toxic encephalopathy (3%, n = 1), FTLD (3%, n = 1), and Lewy body dementia (3%, n = 1).

Both clinical severity and the MoCA were comparable between the two groups (Table 1). Episodic memory was equally impaired in AD and non-AD patients as the FCSRT total recall, free recall, delayed free recall, and delayed total recall scores did not differ between groups. Short term and working memory were more severely affected in non-AD compared to AD patients when focusing on the Digit Span Forwards score (p = 0.015, d = 0.56), although the Digit Span Backwards score was comparable between the groups. There was a trend for worst impairment in executive functions in non-AD patients as measured by the Literal Verbal Fluency score (p = 0.089), although the Trail Making Test B and Victoria Stroop Test did not differ between patients with and without AD. Semantic memory and attention were comparable between groups.

Concerning brain morphometry, total intracranial absolute volumes were comparable between AD and non-AD (Table 2). The frontal, parietal, occipital, temporal, and hippocampal GM volumes’ z-scores were negative on average (Table 2), indicating significant hippocampal and cortical atrophy in both patient groups. However, the range of z-scores was wide in both groups, with greater variability in non-AD. Whole-brain and regional relative brain volumes were similar between AD and non-AD when adjusting for age and sex (Table 2). Results were consistent when adjusting for t-Tau. Finally, we conducted a sensitivity analysis by excluding the 22 patients with A-T+ profile. Brain volumes remained similar between groups (A + T+ versus A-T-).

We used linear (FCSRT total recall score ∼ Hippocampal volume) and multilinear (FCSRT total recall score ∼ Hippocampal volume, Aβ₄₂, p-Tau, t-Tau, Group) regression models to assess whether hippocampal relative volumes explained episodic memory performances across the whole group of patients. Total (β= 0.44, 95% CI [0.19,0.67], p = 0.001), left (β= 0.48, 95% CI [0.22,0.73], p = 0.001), and right hippocampus volume (ß = 0.35, 95% CI [0.12,0.59], p = 0.004) strongly correlated with FCSRT total recall score. When we included AD CSF biomarkers and AD/non-AD group as additional predictors to the model, hippocampal volumes explained the larger fraction of FCSRT total recall inter-subject variability (Table 3). Aβ₄₂ was a significant predictor of FCSRT total recall with smaller effect size than the hippocampus (Table 3) while group, p-Tau and t-Tau were not.

DISCUSSION

We found that AD and non-AD amnestic patients present similar brain morphological patterns, including hippocampal volumes. Memory performance was positively correlated to hippocampal volume, also after adjusting for CSF biomarkers. These results suggest that brain morphometry may not be used as neuroimaging evidence of AD in amnestic older adults and reaffirm the importance of pathophysiological biomarkers for the diagnosis of AD.

AD and non-AD amnestic patients had similar brain morphological patterns. The presence of a stereotypical pattern of brain atrophy is well established in AD, and accurately differentiates AD patients from healthy older adults [33]. Neuroimaging biomarkers, such as hippocampal volume, are therefore significant contributors to the diagnostic confidence when evaluating patients with cognitive impairment [34]. While brain morphometry may be useful in distinguishing patients with AD from healthy older adults, this is not the case for identifying amnestic patients with amyloid and non-amyloid pathologies. Indeed, a recent study reported similar findings using brain morphometry for differentiating amnestic patients with AD versus SNAP as assessed by PET [5]. The authors reported similar medial temporal lobe atrophy between groups. Similarly, PART may present with AD-like memory loss, often misleading to AD pathology [35]. LATE and hippocampal sclerosis (HS) have emerged as key mimics of AD in postmortem analyses, and both have high prevalence among the oldest old, with HS detected in individuals over age 85 in up to 20% of autopsies [36, 37]. Interestingly, some of these pathologies share similar brain atrophy pattern to AD, including in the hippocampus. In FTLD, patients who manifest amnesia as initial clinical presentation exhibit medio-temporal atrophy comparable to patients with AD [38], and hippocampal atrophy has also been observed in PART, LATE, HS, cerebrovascular disease, and depression [36 , 39–41]. Amnestic patients with these AD mimics may therefore be indistinguishable from those with AD based on brain morphometry alone. Consistently, in our cohort 67% of non-AD patients had a clinical diagnosis of cerebrovascular disease, FTLD, Lewy body dementia, normal pressure hydrocephalus, sleep apnea, alcoholic, chronic traumatic, toxic or infectious encephalopathy. We note that PART and LATE are biological constructs that are not directly translated into current clinical diagnosis’ protocols and more likely represented in the SNAP group.

In this study, amnestic patients with and without AD presented with similar cognitive profiles and comparable impairment in episodic memory. Amnestic patients without AD may have a performance at the FCSRT total recall test that is equivalent to amnestic patients with AD due to episodic memory impairment in FTLD, PART, LATE, HS, cerebrovascular disease, and depression. In these diseases, although memory deficit is not the hallmark of the pathology, episodic memory is reported to be impacted. The bvFTLD has been reported to present with performance on the FCSRT that is worse than healthy older adults, although it may be better than patients with AD in immediate and delayed free and total recall [4, 38]. Moreover, recent studies have shown that the cognitive impairment observed in definite PART is characterized by episodic memory deficits, although there may be sparing in comparison to AD [42]. While less frequent in populations younger than 80 years old, LATE is typically associated with slowly progressive impairments in episodic memory, which may be accompanied by deficits in working and semantic memory [43, 44]. LATE and AD present with nuanced differences in cognition, with similar profiles but worse impairment in AD than in isolated LATE, and more severe deficits when both pathologies are combined. Often misdiagnosed as AD, HS presents with a similar type of amnesia [37]. Due to the variable nature of cerebrovascular lesions, vascular dementia may also manifest with episodic memory impairment, although often to a lesser extent than AD [45].

Hippocampal volumes explain episodic memory impairment independently of Alzheimer’s pathology. In our study, total, left and right hippocampal volumes positively correlated with FCSRT total recall in the whole groups of patients. This corroborates a previous study on 35 ASHT patients, which reported free and total recall to be significantly correlated with measures of medio-temporal atrophy [46]. As AD and non-AD groups differed significantly in levels of t-Tau, p-Tau, and Aβ₄₂, our analyses were adjusted for CSF AD biomarkers and group. The difference in t-Tau could indicate more advanced neurodegeneration in patients with AD, and tau pathology has been suggested to affect memory function through different mechanisms in patients with and without AD [47]. However, there was no correlation between either of t-Tau, p-Tau or group, and FCSRT total recall. This suggests that hippocampal volume explains episodic memory performance irrespectively of AD pathology, and reinforces the importance of molecular markers in the diagnosis of AD.

Although our study includes a high number of amnestic patients with CSF biomarkers and brain morphometric measurements, it is not without limitations. Automated morphometry algorithms may present with intrinsic limitations in accuracy, although they have the advantage of availability and are less time-consuming than manual methods. Moreover, this study focused on volumetry of relatively large brain areas. While this limits multiple comparisons, other morphological features such as cortical thickness and shape assessed at higher spatial resolution may highlight morphometry patterns more specific to individual pathologies. At a fined-grain level of investigation, voxel-based morphometry may provide additional sensitivity for spatially constrained atrophy patterns and group-differences. Yet, volume-based morphometry has demonstrated comparable-to-superior accuracy or complementary value to voxel-based morphometry in distinguishing MCI from demented patients [29, 48]. The use of a Memory Clinic retrospective dataset reflects an actual clinical scenario but is dependent on patient research consent and may lead to bias with inclusion of more educated patients. Furthermore, the Leenaards Memory Center proposes a lumbar puncture for CSF AD biomarkers’ assessment mainly to patients younger than 80 years old, which leads to a recruitment bias with over-representation of younger patients. Absence of older amnestic patients from our sample because of lack of CSF biomarkers may lead to under-representation of certain pathologies affecting the oldest, such as LATE and HS, in favor of diseases with higher prevalence in patients under 80 years old such as bvFTLD and DLB [35–37 , 50]. Finally, the Leenaards Memory Center is also a referral center for complicated diagnosis related to cognitive impairment for the Western Switzerland [9]. The group of non-AD patients is therefore heterogeneous in terms of clinical diagnoses. The relatively high prevalence of non-AD profiles in our amnestic cohort may be partly explained by the presence of some amyloid false negatives, considering that the Aβ₄₂ concentration and not the Aβ₄₂/Aβ₄₀ ratio was available for the majority (80%) of AT patients’ classification (unavailable Aβ₄₀ data). This aspect supports the usefulness of the CSF Aβ₄₂/Aβ₄₀ ratio in the interpretation of CSF biomarker profiles in Memory Clinic patients [13] and motivated our sensitivity analysis on A + T+ versus A-T- patients only. In addition, there could also be patients’ selection biases due to our clinical practice: patients identified as “typical AD” early in the clinical investigation may not have undergone all the tests required by the inclusion criteria of this retrospective study (i.e., MRI, lumbar puncture, and neuropsychological assessment). Moreover, A + T- amnestic patients were excluded from the study because their clinical profiles is likely explained by non-AD pathologies [15]. Absence of longitudinal information may limit interpretation of the relationship between brain morphometry and degree of memory impairment. In pathologies such as LATE, patients at an earlier stage of the disease initially present with similar levels of atrophy as of amnesia, while progression of the condition is characterized by greater increase of atrophy than of memory impairment [51, 52]. Finally, our results reflect the clinical population of our local geographic area in Western Switzerland and may not be generalized world-wide.

With the increasing understanding of AD-mimics such as HS and bvFTLD, and the description of new entities such as LATE and PART, this study demonstrates similar brain atrophy patterns between AD and non-AD pathologies and highlights the importance of biological markers for the diagnosis of AD in patients with amnestic syndrome of the hippocampal type, in particular when assessing eligible patients for disease-modifying therapies.

AUTHOR CONTRIBUTIONS

Hadrien M. Lalive (Conceptualization; Formal analysis; Writing – original draft); Alessandra Griffa (Conceptualization; Supervision; Writing – review & editing); Sabrina Carlier (Conceptualization; Writing – review & editing); Mirco Nasuti (Data curation; Writing – review & editing); Tommaso Di Noto (Methodology; Writing – review & editing); Bénédicte Maréchal (Methodology; Writing – review & editing); Olivier Rouaud (Conceptualization; Writing – review & editing); Gilles Allali (Conceptualization; Funding acquisition; Supervision; Writing – review & editing).