By Sonja Blum, MD, PhD
Synopsis: In this multicenter, memory clinic-based cohort targeting earliest at-risk states for dementia of Alzheimer’s type, limbic-predominant and hippocampus-sparing atrophy subtypes were identified. Limbic-predominant atrophy was linked to worse cognitive outcomes over time, including in individuals who were asymptomatic or only had subjective memory symptoms at the time of baseline imaging.
Source: Baumeister H, Vogel JW, Insel PS, et al. A generalizable data-driven model of atrophy heterogeneity and progression in memory clinic settings. Brain. 2024;147(7):2400-2413.
The authors of this study aimed to characterize atrophy patterns in the rising population of patients who still are cognitively normal or have only subjective cognitive impairment at the time of presentation to memory clinics. Prevalence and clinical implications of atrophy heterogeneity in this patient population with absence of significant cognitive impairment is needed.
We know from patients with symptomatic Alzheimer’s disease (AD) that there are distinct categories of atrophy trajectories, and that AD does not always follow the commonly seen progression of pathology and atrophy starting from the medial temporal lobe (MDL) and then spreading throughout the cortex. Posterior cortical atrophy (PCA), behavioral variant AD, and logopenic primary progressive aphasia are examples of disease variants where atrophy patterns deviate from the classic amnestic AD.
The authors studied the DELCODE cohort, a multicenter, memory clinic-based cohort targeting the earliest at-risk states for dementia of Alzheimer’s type. A total of 813 participants (52% women) had available baseline structural magnetic resonance imaging (MRI) segmentation. Of these, 285 were cognitively unimpaired, 342 had subjective cognitive disorder, 118 had mild cognitive impairment, and 68 had AD. Fluid biomarkers characterizing this cohort included apolipoprotein E (APOE) genotype (33.91% APOE e4 carrier), plasma neurofilament light, amyloid beta (Abeta)42, and Abeta40. Cerebrospinal fluid (CSF) samples were available in a subset of 385 patients (47.36%). CSF biomarkers measured were p-tau 181, total tau, Abeta42, and Abeta40.
Standard FreeSurfer and automatic segmentation of hippocampal subfields (ASHS) were used for MRI segmentation. The authors then used SuStain, a machine learning algorithm that is able to analyze large datasets and to identify distinct subtypes of a disease and stage their progression simultaneously.1 They identified 10 atrophy events and stages. Participants who did not exhibit significant atrophy were not assigned to subtype but were referred to as a separate atrophy-negative group.
There were 459 participants (56.46%) who were atrophy-negative. In the remaining 354 participants, two distinct atrophy progression sequences were identified: limbic-predominant (188 participants, 23.12%) and hippocampus-sparing (166 participants, 20.42%).
Participants with limbic-predominant atrophy were older and had more pathological levels across biomarkers (more likely to be Abeta-positive and had higher p-tau181 and total tau levels), except neurofilament light. Cognitive scores were lower in the limbic-predominant subtype in both learning and memory, as well as in executive function and processing speed, and they declined more with increasing SuStain stage. The hippocampus-sparing atrophy subtype had a more pronounced deficit in executive function than in memory scores, overall having a greater impairment in executive function with increasing SuStain stage.
Those who were atrophy-negative at baseline but had lower memory scores had increased odds of progression to limbic-predominant atrophy. In the atrophy-negative group, those with higher baseline plasma neurofilament light level had increased odds of progression to hippocampal-sparing atrophy.
Results were replicated using same protocol in the BioFINDER-2 sample, finding atrophy progression sequences of the limbic-predominant and hippocampus-sparing subtype, with highly similar associations with cognitive performance.
Commentary
The number of patients seeking care in memory clinics is rising rapidly. In particular, there is an influx of patients with very subtle cognitive concerns who are coming for early diagnosis of AD. They are responding to the recent evidence that early diagnosis and treatment with amyloid-clearing monoclonal antibodies, as well as other emerging treatments on the horizon, offer hope for stabilization of the disease.
The authors focused on a memory clinic cohort, which was heavily skewed toward the asymptomatic or only subjectively symptomatic individuals (627 of the 813 participants, or 77.12%). Using modeling of cross-sectional MRI-based atrophy markers, the authors uncovered two atrophy subtypes, limbic-predominant and hippocampus-sparing. These atrophy subtypes resembled known biological subtypes of AD.
The limbic-predominant subtype showed more abnormal AD fluid biomarker levels, and appears to represent a more homogeneous group of patients, presumably with AD as their primary cause of neurodegeneration. The hippocampus-sparing atrophy subtype appears to be a more heterogeneous group, of which only 29.86% had biomarkers suggestive of AD. Hippocampus-sparing atrophy might occur as the result of various neuropathological processes that include but are not limited to hippocampus-sparing AD.
The model was trained on purely cross-sectional data, but validity of the model was strengthened by follow-up MRI analysis, which demonstrated that atrophy deviating from the proposed progression sequence was rarely observed. In addition, the model was shown to be generalizable to a second cohort, at group and subject level.
This line of work elucidates a path toward developing approaches to predict future atrophy subtype from a set of cross-sectional clinical markers, such as structural imaging, fluid biomarkers, and cognitive scores.
Sonja Blum, MD, PhD, is Associate Professor of Clinical Neurology at Weill Cornell Medical College.
Reference
1. Young AL, Marinescu RV, Oxtoby NP, et al. Uncovering the heterogeneity and temporal complexity of neurodegenerative diseases with Subtype and Stage Inference. Nat Commun. 2018;9(1):4273.
In this multicenter, memory clinic-based cohort targeting earliest at-risk states for dementia of Alzheimer’s type, limbic-predominant and hippocampus-sparing atrophy subtypes were identified. Limbic-predominant atrophy was linked to worse cognitive outcomes over time, including in individuals who were asymptomatic or only had subjective memory symptoms at the time of baseline imaging.
You have reached your article limit for the month. Subscribe now to access this article plus other member-only content. TESTING
- Award-winning Medical Content
- Latest Advances & Development in Medicine
- Unbiased Content