Neurofilaments light chain (NfLs) are currently recognized as a marker of axonal injury and degeneration. Their measurement in biological fluids has a promising role in the diagnosis, prognosis, and monitoring of the therapeutic response in neurological diseases, including neurodegenerative dementias. In recent years, their relationship with clinical phenotypes and measures of disease severity has been extensively studied. Here, we reviewed studies investigating the association between NfLs and imaging measures of grey matter (GM) and white matter (WM) damage in neurodegenerative dementias. We identified a large number of studies investigating this association in Alzheimer’s disease (AD) and disorders of the frontotemporal dementia (FTD) spectrum. Results were heterogeneous, possibly due to different methodological approaches—both in NfL measurements and imaging analyses—and inclusion criteria. However, a positive association between NfL levels and GM atrophy, WM microstructural disruption, glucose hypometabolism, and protein accumulation emerged invariably, confirming the role of NfLs as a reliable biomarker for neurodegenerative dementias, albeit not specific.

Objective: Neurofilament light chain proteins (NfLs) are considered a promising biomarker of neuroaxonal damage in several neurological diseases. Their measurement in the serum and cerebrospinal fluid (CSF) of patients with dementia may be especially useful. Our aim was to compare the NfL measurement performance of two advanced technologies, specifically the Ella™ microfluidic platform and the Lumipulse™ fully automated system, in patients with cognitive disorders. Methods: Thirty subjects with neurodegenerative cognitive disorders (10 with Alzheimer’s Disease, 10 with Frontotemporal Dementia, and 10 with non-progressive Mild Cognitive Impairment) seen at the Cognitive Neurology Clinic of Modena University Hospital (Italy) underwent CSF and serum NfL measurement with both the Ella™ microfluidic platform (Bio-Techne, Minneapolis, MN, USA)) and the Lumipulse™ fully automated system for the CLEIA (Fujirebio Inc., Ghent, Belgium). Correlation and regression analyses were applied to assess the association between NfL concentrations obtained with the two assays in CSF and serum. The Passing–Bablok regression method was employed to evaluate the agreement between the assays. Results: There were high correlations between the two assays (r = 0.976, 95% CI. 0.950–0.989 for CSF vs. r = 0.923, 95% CI 0.842–0.964 for serum). A Passing–Bablok regression model was estimated to explain the relationship between the two assays, allowing us to switch from one to the other when only one assay was available. Conclusions: We found a good degree of correlation between the two methods in patients with neurocognitive disorders. We also established a method that will allow comparisons between results obtained with either technique, allowing for meta-analyses and larger sample sizes.

Frontotemporal dementia (FTD) and amyotrophic lateral sclerosis (ALS) are recognized as part of a disease continuum (FTD-ALS spectrum), in which the most common genetic cause is chromosome 9 open reading frame 72 (C9ORF72) gene hexanucleotide repeat expansion. The clinical phenotype of patients carrying this expansion varies widely and includes diseases beyond the FTD-ALS spectrum. Although a few cases of patients with C9ORF72 expansion and a clinical or biomarker-supported diagnosis of Alzheimer’s disease (AD) have been described, they have been considered too sparse to establish a definite association between the C9ORF72 expansion and AD pathology. Here, we describe a C9ORF72 family with pleomorphic phenotypical expressions: a 54-year-old woman showing cognitive impairment and behavioral disturbances with both neuroimaging and cerebrospinal fluid (CSF) biomarkers consistent with AD pathology, her 49-year-old brother with typical FTD-ALS, and their 63-year-old mother with the behavioral variant of FTD and CSF biomarkers suggestive of AD pathology. The young onset of disease in all three family members and their different phenotypes and biomarker profiles make the simple co-occurrence of different diseases an extremely unlikely explanation. Our report adds to previous findings and may contribute to further expanding the spectrum of diseases associated with C9ORF72 expansion.

Cognitive reserve (CR) reflects brain's resilience to pathology, enabling to maintain function despite structural damage. This study investigates its role in young-onset cognitive impairment (<65 years) beyond brain integrity and neurodegeneration. Participants underwent neuropsychological assessment - including the Cognitive Reserve Index questionnaire (CRIq) -, magnetic resonance imaging (MRI), and blood neurofilaments light-chain (NfLs) measurement. Scores of global cognition and domain-specific cognition were derived from Principal Component Analyses of neuropsychological results. Linear regression models estimated CR's contribution to global and domain-specific cognition, alongside age, sex, MRI measures, and NfLs as predictors. Among the 115 participants, global cognition was significantly explained by CR [effect size (ES) = 0.229], grey matter volume (ES = 0.348), and NfLs (ES = -0.302). The effect of CR was prominent on language and attentional-executive functions: while the CRIq subscore related to education predicted performance in both these domains, the subscore related to leisure activities was positively associated with the language domain only. These findings highlight CR's protective role in young-onset cognitive impairment, particularly for non-amnestic cognitive domains. Since a high CR can mask or compensate for neurological cognitive disorders delaying its diagnosis, our results suggest that measures of CR, including time spent on leisure activities, should be considered when interpreting neuropsychological tests.

Despite the increasing availability of biomarkers in clinical settings, diagnosing individuals with subtle cognitive/behavioral symptoms and normal structural brain imaging remains challenging. In real-world settings, it is not feasible to subject all such patients to costly and invasive second-level assessments, and there are no clear guidelines for identifying those who should undergo these procedures beyond clinical follow-up and structural imaging. The present study explores the potential of repeated magnetic resonance imaging (MRI) in this context. We investigated whether detecting pathological brain volume (BV) changes exceeding normal aging could aid in deciding whether advanced biomarker testing is reasonable. Individuals with subtle cognitive/behavioral complaints underwent baseline and 18-month follow-up assessments (neuropsychological evaluation, MRI, and biomarker tests). Annualized percentage of BV change (PBVC/y), adjusted for age, sex, and scanner, was calculated with SIENA, and classified whether exceeding the 80th/95th percentiles of normative data. At follow-up, participants were classified (I) Converters/Non-Converters based on clinical information, (II) Non-converters+/Non-Converters- based on biomarkers status, and (III) Underlying disease based on both clinical follow-up and biomarkers. Logistic regressions assessed PBVC/y, baseline BV, and Mini-Mental State Examination change as predictors of participants' classification. Among 110 participants, PBVC/y exceeding both the 80th/95th percentiles increased the probability of being classified Converters or Alzheimer's disease/Frontotemporal dementia. Exceeding the 80th percentile increased the odds of being classified Non-Converters+. Our results suggest that the application of SIENA normative data to repeated MRIs may help in deciding whether or not performing more invasive and high-cost second-level procedures in individuals with early-stage and subtle cognitive/behavioural changes.

Importance Accurate clinical and pathological diagnoses are essential in neurodegenerative diseases, especially given the emergence of pathology-specific disease-modifying therapies. However, diagnostic accuracy remains challenging due to heterogeneous clinical presentations, complexity of integrating multimodal data, and limited access to multidisciplinary expertise. Primary Progressive Aphasia (PPA) exemplifies these challenges, requiring specialized clinical, neuropsychological, and imaging evaluations. Generative artificial intelligence (AI), powered by large language models, may offer scalable diagnostic support in this context. Objective To evaluate the diagnostic performance of an agentic generative AI system in classifying prototypical PPA cases by clinical syndrome and underlying pathology. Design Retrospective diagnostic validation study using a multi-agent generative AI architecture simulating expert-level reasoning. Setting Single tertiary academic referral center (University of California San Francisco, Memory and Aging Center). Participants Fifty-four individuals with a definite diagnosis of PPA and post-mortem confirmation (18 semantic [svPPA], 17 logopenic [lvPPA], 19 nonfluent [nfvPPA]), selected as prototypical cases with congruent clinical, imaging, and pathological profiles. Exposure Multimodal input data, including clinical notes, neuropsychological and language assessments, and MRI brain images, were processed through a multi-agent architecture. The system generated diagnostic predictions under two conditions: (1) open-ended diagnosis from a set of 15 neurodegenerative clinical syndromes; (2) constrained classification of PPA variant and underlying neuropathology. Main Outcomes and Measures Generative AI system diagnostic accuracy for clinical syndrome and pathology, based on expert clinical diagnoses and post-mortem confirmations as gold standard. Results In the open-ended setting, the system correctly identified PPA in 49 of 54 cases (90.7%, chance level=6.7%). When constrained to PPA, it achieved 100% accuracy for svPPA and nfvPPA, and 94.1% for lvPPA as primary prediction. Neuropathological predictions were most accurate for FTLD-TDP type C (100%) and FTLD-4R tau (100%), and high for Alzheimer’s disease (94.4%). The full diagnostic pipeline of all 54 cases was completed in under 10 minutes. Conclusions and Relevance The AI system demonstrated expert-level performance in classifying prototypical PPA cases, integrating multimodal data and mirroring specialist reasoning. Its speed and accuracy support its potential role in extending access to specialized diagnostic expertise, particularly in non-tertiary settings. Further validation in larger and more heterogeneous populations is warranted.