ObjectiveTo explore the value of the amyloid-tau-neurodegeneration (ATN) framework in neuroimaging based on PET for diagnosing mild cognitive impairment (MCI) and Alzheimer′s disease (AD), and analyze its relationship with clinical cognition.

MethodsFrom May 2022 to March 2024, a total of 98 cases (23 males and 75 females, age (67.8±8.6) years) with a diagnosis of AD, MCI, or non-AD (control patients, CP) who underwent ¹⁸F-FDG, ¹⁸F-AV45, and ¹⁸F-AV1451 PET/CT imaging in the Second Affiliated Hospital of Guangzhou Medical University were included retrospectively. The clinical data, Mini-Mental State Examination (MMSE), and Montreal Cognitive Assessment (MoCA) scores were recorded. Cases were divided into MCI group, mild AD group, moderate AD group, moderate-severe AD group, and CP group. PET images were visually and semi-quantitatively evaluated. SUV _mean and SUV ratio (SUVR) were obtained from independent brain regions of ¹⁸F-FDG ( n=8), ¹⁸F-AV45 ( n=14) and ¹⁸F-AV1451 ( n=14). ROC curve analysis was performed with clinical diagnosis as a criterion. The consistency between visual assessment and the clinical diagnosis was analyzed by Cohen′s Kappa coefficient. Semi-quantitative comparisons between groups were performed using the independent-sample t test, one-way analysis of variance, Mann-Whitney U test, or Kruskal-Wallis rank sum test. Age was used as a covariate to calculate the partial correlation coefficient between SUVR and cognitive scores.

ResultsThe sensitivity and specificity of comprehensive visual assessment in diagnosing AD+ MCI were 87.65%(71/81) and 14/17 respectively, showing a moderate consistency with clinical diagnosis ( Kappa=0.60, P<0.001). Semi-quantitative analysis showed that ¹⁸F-FDG uptakes in all independent brain regions of MCI patients were higher than those of AD patients, whereas the uptakes of ¹⁸F-AV45 and ¹⁸F-AV1451 were lower ( t values: 2.66-3.95, z values: 4.98-15.04, all P<0.05). The difference in ¹⁸F-AV45 uptake among the three subgroups of AD was relatively small ( H values: 0.46-4.06, F values: 0.03-0.08, all P>0.05). Except for the medial temporal and occipital lobes, the ¹⁸F-AV1451 uptake in the moderate-severe AD group tended to be higher than that in the moderate and mild AD groups, though not statistically significant ( H values: 0.20-5.17, all P>0.05). ¹⁸F-FDG PET semi-quantitatively distinguished MCI from CP with a high sensitivity (13/14), ¹⁸F-AV45 demonstrated a high sensitivity for diagnosing AD+ MCI (92.59%, 75/81), and ¹⁸F-AV1451 had a high specificity for distinguishing AD from MCI (14/14) (AUCs: 0.87, 0.90 and 0.92). The uptakes of ¹⁸F-FDG in gray matter of AD and MCI patients were positively correlated with MMSE and MoCA scores ( r values: 0.30-0.43, 0.29-0.45, all P<0.05), while the uptakes of ¹⁸F-AV45 and ¹⁸F-AV1451 were negatively correlated with MMSE and MoCA scores ( ¹⁸F-AV45, r values: from -0.39 to -0.30, from -0.38 to -0.30, all P<0.05; ¹⁸F-AV1451, r values: from -0.50 to -0.28, from -0.53 to -0.28, except for medial temporal lobe P>0.05, all others P<0.05).

ConclusionThe PET-based neuroimaging ATN framework is helpful for early diagnosis of MCI and AD, as well as for AD staging, and may reflect the disease progression and clinical cognitive status of AD to a certain extent.