How to follow up and cluster subjects by longitudinal changes of fibrillary amyloid imaging and CSF biomarkers? a 24-month follow up


      Presence of amyloid is associated with cognitive decline. Fibrillar amyloid deposition levels in brain can be detected by increase uptake of amyloid in [18F] Florbetapir (AV45) PET and decrease of csfAβ1-42. Combination of biomarkers performed better for diagnosing AD. Tau was strongest predictor of conversion but Jack et al. propose that CSFAβ1-42 was abnormal more often than t-tau. Here, we studied relationship between regional distributions of brain fibrillar amyloid deposition, neurodegenerative biomarkers in brain(FDG) and CSF(tau), brain structural change and cognitive function at 24-month follow-up to find the best method to follow up the subjects.


      We analyzed 182 participants from ADNI and try to cluster the subjects by MCLUSTforR. We divided the subjects into 3 groups by CSF biomarkers and AV45 (progressive, regressive, discordant) and put threshold (2SD, 1SD, Z-score) in each group. We used the baseline csf biomarkers to divide subjects into 2 subgroups (positive or negative at baseline). We compare the results from every methods to find the best way. We also analyzed by baseline diagnosis (CN, EMCI, LMCI, AD). FDG voxel-based group comparisons were calculated with PET-SUVR resampled and blurred with an 8mm Gaussian filter. Voxel-based group comparisons were calculated using RMINC. Student t tests or ANOVA was used. Statistical tests were carried out at p<0.05.


      We had only2cluster with MCULST. We divided subjects by csf Aβ1-42and AV45 into 3 groups but when put threshold, we had only 7 subjects left (3 progressive, 2 regressive, 2 nonconcordant). Finally we used cutoff from zero. We found that subjects in regressive amyloid accumulation and CSFAβ1-42 positive at baseline have lowest cognitive function at baseline. We divided subjects by another 3 methods (relation between csf Aβ1-42 and AVV45 with baseline tau; relation between csf tau and AV45 with baseline tau; relation between csf Aβ1-42, and tau with baseline tau). For structural brain change, it was significant change in first method (relation between csf Aβ1-42 and AV45 with baseline csfAβ1-42) in FDG between progressive with CSFAβ1-42 positive and negative at baseline. Cognitive function was significant change between progressive group with baseline csf tau positive and negative in LDEL (only the combination use of biomarker(fig1)). When analyzed in group base on baseline diagnosis, we found significant between progressive group in FDG in cognitive normal subjects and LDEL in EMCI subjects.


      Relation between amyloid change in csf and imaging are useful to follow up neurodegeneration. Combination of amyloid and tau biomarkers in the csf and imaging are better to follow up the cognitive function. We suggest more than 24-month follow up to see more significant change.
      Figure thumbnail fx1
      Figuret-statistical contrast of FDG between nonprogresser and progress group (representing brain region showing higher rates of metabolism declines)
      Figure thumbnail fx2
      Figure 1Shows FDG change in 24 months in all subjects by the four methods of cluster
      Figure thumbnail fx3
      Figure 2Shows LDEL change in 24 months in all subjects by the four methods of cluster
      Table 1Show baseline characteristic and FDG change of the subjects in each group (cluster by [I8F]Florbetapir (AV45) PET and csf Aβ1-42 with baseline csf Aβ1-42)
      progressive csf Aβ1-42 (+)progressive csf Aβ1-42 (-)regressive csf Aβ1-42 (+)regressive csf Aβ1-42 (-)nonconcordant csf Aβ1-42 (+)nonconcordant csf Aβ1-42 (-)
      Gender (m/f)22/1711/195/313/826/2413/21
      FDG change-0.076850.01070-0.035400.01228-0.01134-0.03661