PET Imaging in Alzheimer Disease: Diagnostic and Prognostic Applications

INTRODUCTION

FDG PET in Alzheimer Disease

18F-FDG PET measures cerebral glucose metabolism and provides an indirect marker of neuronal and synaptic dysfunction. AD commonly produces hypometabolism involving the posterior cingulate cortex, precuneus and lateral temporoparietal association cortices. The spatial pattern can help distinguish AD from other neurodegenerative disorders. The Alzheimer’s Association specialty-care guideline recommends considering FDG PET when the etiological diagnosis remains equivocal after structural imaging, particularly in early MCI or atypical presentations [1]. The guideline reports sensitivity of approximately 80–99% and specificity of 63–98% for differentiating AD from frontotemporal lobar degeneration, while reported sensitivity and specificity for AD versus Lewy body disease are approximately 70–92% and 74–100%, respectively. FDG PET may also contribute to staging and prognosis because the magnitude and distribution of metabolic abnormalities correlate with neurodegenerative burden. However, interpretation becomes less useful in severe dementia, when diffuse hypometabolism can obscure disease-specific patterns [1].

Amyloid PET Imaging

Amyloid PET directly identifies fibrillar amyloid-β plaques. Available fluorine-18 tracers include florbetapir, florbetaben and flutemetamol. A positive scan indicates substantial amyloid pathology, whereas a negative scan substantially lowers the probability that significant amyloid pathology is responsible for a patient’s cognitive syndrome. Amyloid PET is especially useful when clinical diagnosis is uncertain, including selected patients with MCI, atypical presentations, or early-onset cognitive impairment. However, amyloid positivity is not synonymous with dementia because amyloid pathology may occur in cognitively unimpaired older adults or alongside other diseases. Accordingly, PET findings must be interpreted within the clinical context [2]. The clinical relevance of amyloid PET has expanded with anti-amyloid therapies. Recent literature describes amyloid PET as increasingly important for confirming treatment eligibility and for quantifying biological treatment response using standardized approaches such as Centiloid scaling [3].

Tau PET Imaging

Tau PET provides a spatial representation of aggregated tau pathology and is particularly valuable because tau burden is more closely related to disease severity and cognition than amyloid burden alone. Tau pathology typically begins in medial temporal regions and spreads to association cortices as disease progresses. 18F-flortaucipir was the first FDA-approved tau PET tracer. Nevertheless, interpretation requires awareness of off-target binding and incomplete detection of all tau species. Next-generation tracers are being developed to improve specificity [4]. Current appropriate-use guidance recognizes selected applications of tau PET for disease staging, atypical presentations, prognosis and treatment-related decision-making [2]. Tau PET therefore has strong potential to move PET from binary diagnosis toward individualized biological staging.

PET and Prediction of Disease Progression

Prediction of progression from MCI to dementia is an important application of PET. Amyloid PET can identify individuals with underlying AD pathology, but amyloid positivity alone does not determine when dementia will occur. Prognosis is influenced by tau burden, cognitive impairment, neurodegeneration, age, vascular disease and other pathologies. FDG PET can provide prognostic information through the extent and distribution of hypometabolism, while tau PET may be particularly informative because tau accumulation more closely parallels clinical severity. Recent work also suggests that quantitative and multimodal approaches may improve prognostic accuracy compared with single biomarkers. Therefore, PET-based prognosis should be communicated probabilistically rather than as a precise individual prediction (Table 1).

Table 1: Major PET Approaches in Alzheimer Disease

.

Differential Diagnosis

PET can assist differentiation of AD from frontotemporal degeneration, dementia with Lewy bodies and other neurodegenerative disorders.  FDG PET is particularly useful because diseases demonstrate different metabolic patterns. AD typically produces posterior temporoparietal and posterior cingulate hypometabolism, whereas frontotemporal disorders more often show frontal and anterior temporal abnormalities. Lewy body disease may show occipital hypometabolism. Amyloid PET can help determine whether AD amyloid pathology is present, while tau PET can demonstrate whether tau distribution is consistent with AD. Nevertheless, mixed pathology is common in older patients and PET findings should be integrated with clinical history, examination, MRI and other biomarkers [1] (Table 2).

Table 2: Diagnostic and Prognostic Roles of PET

.

PET and Treatment Decisions

The emergence of disease-modifying anti-amyloid therapies has changed the clinical role of PET. Updated appropriate-use criteria emphasize that amyloid and tau PET should be performed when results are expected to influence care, including clarification of diagnosis and assessment of eligibility for disease-modifying treatment [2]. Amyloid PET is particularly relevant because anti-amyloid therapy targets Aβ pathology. Tau PET may provide complementary information about disease stage and expected clinical benefit. Recent reviews emphasize that PET is shifting from a confirmatory research biomarker toward an active component of therapeutic decision-making and treatment monitoring [3] (Table 3).

Table 3: Typical PET Patterns in Major Neurodegenerative Disorders

.

Quantitative PET and Artificial Intelligence

Visual PET interpretation remains important, but quantitative methods can improve reproducibility and detect subtle abnormalities. Common approaches include standardized uptake value ratios, cortical composite measures, voxel-based analysis and Centiloid scaling. Quantification is increasingly relevant for longitudinal treatment monitoring. Artificial intelligence and machine learning may further enhance PET by integrating imaging with MRI, cognitive testing, APOE status and blood biomarkers. However, models require large and diverse datasets and external validation before widespread clinical adoption. The emerging multimodal approach is likely to combine molecular imaging with accessible blood biomarkers rather than rely on any single test.

Limitations

PET has several limitations. First, amyloid positivity is not equivalent to clinical AD dementia. Second, tau tracers differ in their binding properties and may demonstrate off-target signal. Third, PET is expensive and not universally available. Fourth, radiation exposure limits unnecessary repeated imaging. Fifth, interpretation varies with tracer, scanner, acquisition protocol, reference region and quantitative threshold. Mixed pathology is another major challenge. A patient may have AD pathology together with vascular disease, Lewy body pathology, TDP-43 proteinopathy, or other conditions. PET should therefore be interpreted as one component of a multimodal diagnostic process. The appropriate-use criteria explicitly emphasize clinical correlation, technical quality and patient-centered decision-making [2].

Future Directions

Future PET research is likely to focus on next-generation tau tracers, improved imaging quantification, multimodal artificial intelligence and integration with blood biomarkers. The development of blood-based biomarkers may allow PET to be reserved for cases in which molecular localization or therapeutic decision-making requires additional certainty. Recent reviews emphasize that tau PET retains a unique advantage because it provides topographic information that fluid biomarkers cannot provide [5]. Quantitative amyloid and tau PET may also become increasingly relevant for monitoring biological response to therapy. These developments could establish a more personalized imaging framework in which the choice of tracer is determined by the clinical question rather than by a one-size-fits-all diagnostic pathway.