Abstract
Our findings suggest that subthreshold amyloid levels may contribute to conversion to dementia in patients with amyloid-negative amnestic MCI.
Introduction
Amnestic Mild Cognitive Impairment (aMCI) is frequently a precursor to Alzheimer’s Disease (AD), yet its progression remains difficult to predict. In clinical practice, Amyloid PET scans are the gold standard for identifying $\beta$-amyloid (A$\beta$) plaques. A "negative" scan often leads to the assumption of lower risk for dementia conversion. However, nearly half of aMCI patients are amyloid-negative, and a significant portion of this group still progresses to dementia. This study investigates whether amyloid levels falling below traditional clinical thresholds—"subthreshold" levels—impact cognitive decline and dementia conversion.
Methods
This retrospective cohort study utilized two distinct datasets: the AMC Cohort (103 patients from Asan Medical Center, Seoul) and the ADNI Cohort (48 patients for external validation). All participants were diagnosed with aMCI and classified as amyloid-negative via visual inspection of $[^{18}F]$-florbetaben (FBB) PET scans.
Researchers employed quantitative Standardized Uptake Value Ratio (SUVR) analysis to measure tracer uptake in specific Volumes of Interest (VOIs), including the frontal, temporal, parietal, and occipital cortices. Receiver Operating Characteristic (ROC) curves determined the predictive power (Area Under the Curve, AUC) of these regional SUVR values for dementia conversion over a 36-month follow-up period.
Key Insights: The Continuum Hypothesis
The study challenges the binary "positive/negative" view of amyloid pathology, supporting a "Continuum Hypothesis." Amyloid accumulation is a slow, non-linear process; thus, "amyloid-negative" is not synonymous with "amyloid-free."
Specific "canary" regions, particularly the parietal cortex and precuneus, emerged as highly sensitive predictors. Even marginal increases in amyloid within these areas—undetectable by standard visual rating—were associated with a higher risk of conversion. This suggests that low-level amyloid might lower the brain's threshold for other pathologies, such as tau protein or vascular damage, to trigger clinical symptoms.
Results and Predictive Accuracy
In the AMC cohort, approximately 36% (39 out of 107) of patients initially labeled amyloid-negative converted to dementia within three years. Quantitative analysis revealed that the SUVR of the left parietal cortex was the strongest individual predictor of conversion (AUC = 0.762, $P < 0.001$). Combined models incorporating the precuneus, parietal cortex, and global composite scores achieved an AUC of 0.763.
Validation using the ADNI dataset confirmed these trends, with the left precuneus SUVR showing the highest predictive value (AUC = 0.596, $P = 0.006$). Furthermore, higher amyloid levels in the frontal cortices of "converters" correlated with poorer performance on visual memory tasks, specifically the RCFT recognition test.
Limitations and Future Directions
While these findings are significant, the study has limitations. It did not account for other contributing biomarkers like tau, TDP-43, or $\alpha$-synuclein. Additionally, the conversion rate observed is higher than some previous reports, suggesting a possible selection bias toward higher-risk patients in memory clinic settings. Future longitudinal studies tracking amyloid levels over time, rather than relying on a single baseline scan, are necessary to confirm these findings and explore the cellular-level implications of subthreshold signals.
Summary updated