What specific cognitive tests and imaging studies distinguish frontotemporal dementia from Alzheimer’s disease?
Executive summary
Two complementary pillars help separate frontotemporal dementia (FTD) from Alzheimer’s disease (AD): cognitive/neuropsychological testing that probes frontal‑executive, language, and social cognition versus episodic memory, and brain imaging that maps frontal–temporal versus temporoparietal/hippocampal degeneration or metabolic changes on MRI and PET; neither approach is perfectly specific, and combined multimodal assessment (imaging + CSF or advanced MRI analytics) yields the best individual diagnostic probabilities [1] [2] [3].
1. Cognitive profile: which tests point toward FTD versus AD
Standard neuropsychological batteries detect a divergent pattern: AD typically shows early, prominent episodic memory impairment on tests like list‑learning and recall (e.g., Free and Cued Selective Reminding paradigms), whereas FTD—especially behavioral variant FTD (bvFTD) and nonfluent/semantic primary progressive aphasias—shows early deficits in executive functions, social cognition, behavior inventories, and specific language domains [1] [4] [5]. Tests cited in the literature that aid this separation include the Frontal Assessment Battery (FAB) and Addenbrooke’s Cognitive Examination (ACE‑III) for frontal/executive and language deficits, the Frontal Behavioral Inventory for behavioral change sensitivity to FTD, and theory‑of‑mind or social cognition tasks to expose bvFTD deficits that are less marked in typical AD [1] [5] [4]. Importantly, atypical AD variants (behavioral/dysexecutive AD or focal cortical presentations) can mimic FTD on these tests, so cognitive testing is necessary but not sufficient [6] [7].
2. Structural MRI: atrophy patterns that separate the diseases
Structural MRI is the workhorse imaging study: FTD characteristically shows frontal and anterior temporal cortical atrophy (including asymmetric temporal pole and insular changes in semantic variants), whereas AD more commonly produces hippocampal and temporoparietal cortical volume loss and medial temporal atrophy [2] [8]. Quantitative approaches—hippocampal volumetry, deformation‑based morphometry, and machine‑learning classifiers trained on structural MRI—improve differentiation beyond visual reads and can track longitudinal atrophy patterns distinct to each disease [2] [9] [10]. Yet early disease and overlapping atrophy (or focal AD variants) limit single MRI specificity, prompting the use of combined imaging and fluid biomarkers [5] [6].
3. Functional imaging and molecular PET: metabolism and amyloid/tau markers
FDG‑PET reveals hypometabolism patterns complementary to MRI—frontal and anterior temporal hypometabolism favors FTD, whereas posterior temporoparietal and posterior cingulate hypometabolism favors AD—helping when structural scans are equivocal [8] [4]. Molecular PET adds molecular specificity: amyloid PET reliably identifies AD pathology when positive and serves as a strong negative discriminator for FTD when absent; the recent approval of amyloid tracers has made amyloid PET a practical tool to rule in/out AD pathology in ambiguous cases [11] [8]. Tau PET and tracers for all tau isoforms remain more limited for universal FTD detection, and FTD pathologies (TDP‑43, non‑AD tau isoforms) can evade current tau ligands, limiting sensitivity for FTD molecular imaging [12] [11].
4. CSF, combined biomarker strategies, and machine learning
Cerebrospinal fluid (CSF) biomarkers typical of AD (reduced Aβ42, elevated phosphorylated tau) provide high diagnostic value when combined with MRI or PET, and adding CSF neurofilament light (NfL) and other markers can raise confidence for FTD versus AD; recent work shows probabilistic algorithms that fuse MRI and CSF greatly improve individual diagnostic precision [3] [10]. Machine‑learning models applied to multimodal MRI (structural, longitudinal, or multiparametric) are promising and outperform single measures in research cohorts, but they remain dependent on training sets and may overfit or struggle with atypical clinical presentations in routine practice [13] [9].
5. Practical interpretation, caveats and competing incentives
Best practice integrates history, targeted neuropsychological tests (e.g., FAB, ACE‑III, Frontal Behavioral Inventory), structural MRI, and functional/molecular imaging or CSF biomarkers when available; even then, diagnostic uncertainty persists early or in atypical cases, and overreliance on any single test can mislead [4] [5] [1]. Readers should note potential biases: commercial and research interests accelerate PET tracer adoption and machine‑learning claims [11] [13], and many studies use specialized cohorts that may not reflect community clinics—hence multimodal, contextual assessment remains the recommended, evidence‑based approach [3] [10].