Amyloid-Beta Imaging in Prodromal Alzheimer's
Amyloid-Beta Imaging in Prodromal Alzheimer's
This prospective study showing the robust independent predictive utility of Aβ imaging with FBB PET provides further evidence that prodromal AD or MCI due to AD is accurately identified by the use of a biomarker specific to Alzheimer's pathology in patients with a history of cognitive decline and who have objective impairment. This allows an earlier diagnosis with a high level of confidence. Economic models have suggested that shortening the time to diagnosis of AD by confirming the presence of AD pathology with FBB PET may be cost-effective. There was an excellent inter-rater reliability for the visual reads as well as an almost perfect concordance with the classification of a scan as positive or negative for Aβ by semiquantitative assessment with the neocortical SUVR, supporting its use in routine clinical settings. Our finding that FBB imaging alone had 83% (95% CI 61% to 94%) accuracy over 2 years for prediction of AD is supported by pooled data from all published longitudinal Aβ imaging studies of MCI to date. The pooled data are results from 297 MCI subjects followed for an average of 34 months. Progression to AD was observed in 121/174 (70%) when Aβ imaging was positive compared with 10/123 (8%) if Aβ imaging was negative, yielding a predictive accuracy of 81% (95% CI 74% to 86%).
In this study, we did not find that predictive accuracy was further improved by the combination of FBB imaging with aMCI status. This is at odds with our previous findings in the AIBL study and likely reflects differences in subject selection and sample size.
Of those MCI who had dementia at 4 years, 21/26 (81%) were FBB+ at baseline. This figure is consistent with postmortem studies of MCI. For example, one study reported that 68%–88% of those with MCI who progressed to dementia had AD neuropathology at postmortem depending on the neuropathological criteria used. The rate of progression to dementia in FBB− MCI was 24% at 4 years. This figure is also close to the AIBL study which showed that 29% of MCI participants with a negative C-PiB Aβ baseline scan progressed to dementia at 3 years follow-up.
HV showed limited prognostic value. Consistent with these findings, a review pooling results across AD imaging biomarkers reported the highest diagnostic and prognostic accuracy for amyloid imaging and the lowest for MRI. HA is observed with ageing in the absence of neurodegenerative disease and also occurs in other dementias including DLB and FTLD. There is no universally accepted method to measure HV, define HA, or determine the threshold for detecting or predicting progression to AD. Consequently, we used a commercially available software package that is operator and site independent, and applied it to a 3D MP-RAGE MRI sequence to measure HV. We used the double ROC approach to determine the optimal threshold for distinguishing AD from healthy elderly in an unrelated study population and applied this threshold to determine the predictive accuracy of HV in this MCI cohort. It could be argued that the selected threshold might not be optimal, but we have subsequently applied different data-derived thresholds and were not able to improve the predictive accuracy of HV.
The strengthening of the correlation between HV and EM and the concomitant weakening of the correlation between Aβ and EM over 2 years of serial study is consistent with the hypothetical sequence of biomarker change in AD, where progressive HA continues as Aβ accumulation plateaus after dementia develops. This weakening of the strength of the correlation between Aβ and EM with disease progression may account for some of the variability reported on the correlation of Aβ and EM between MCI studies. Furthermore, despite a non-significant mild strengthening of the correlation between Aβ and HV after 2 years, the correlation between Aβ and EM disappeared when accounting for HV. These results, along with our previous baseline report suggest that while at very early stages, Aβ and HV are significantly and independently related to EM impairment, and as disease progresses the relation between Aβ and EM is mediated by HV. This is in agreement with previous reports and supports the concept that disease modifying therapy, when available, should be given very early to achieve the greatest benefit.
Longitudinal studies of Aβ with PET indicate that accumulation takes approximately 15–20 years to progress from the Aβ+ threshold to the levels seen in most patients with mild AD. During this period, the slow increase in Aβ accumulation remains constant, but as disease progresses the rates of accumulation slow down, tending towards a plateau. In this study, we observed an increase in FBB retention at a rate of 0.037 SUVR/year in FBB+ participants, similar to the rates of 0.043 SUVR/year reported in longitudinal studies with C-PiB PET. This also suggests that FBB PET is sensitive enough to monitor the therapeutic response to anti-Aβ therapy.
Among the limitations of the present study are the single-centre setting, the fact that repeat scans and neuropsychology tests were only obtained for 2 years and that the clinical evaluation was not blinded to FBB results beyond 2 years. Nonetheless, the 4-year follow-up reinforces the 2-year findings and provides valuable insight into the eventual outcome for both Aβ+ and Aβ− scan individuals with MCI. Other limitations include possible confounding effects from acetylcholinesterase inhibitor medication on cognition that was commenced in seven participants who completed FBB and neuropsychological studies to 2 years. Given that six out of these seven declined in EM scores at 24 months compared with baseline, impact on the overall results is likely to be minor.
Discussion
This prospective study showing the robust independent predictive utility of Aβ imaging with FBB PET provides further evidence that prodromal AD or MCI due to AD is accurately identified by the use of a biomarker specific to Alzheimer's pathology in patients with a history of cognitive decline and who have objective impairment. This allows an earlier diagnosis with a high level of confidence. Economic models have suggested that shortening the time to diagnosis of AD by confirming the presence of AD pathology with FBB PET may be cost-effective. There was an excellent inter-rater reliability for the visual reads as well as an almost perfect concordance with the classification of a scan as positive or negative for Aβ by semiquantitative assessment with the neocortical SUVR, supporting its use in routine clinical settings. Our finding that FBB imaging alone had 83% (95% CI 61% to 94%) accuracy over 2 years for prediction of AD is supported by pooled data from all published longitudinal Aβ imaging studies of MCI to date. The pooled data are results from 297 MCI subjects followed for an average of 34 months. Progression to AD was observed in 121/174 (70%) when Aβ imaging was positive compared with 10/123 (8%) if Aβ imaging was negative, yielding a predictive accuracy of 81% (95% CI 74% to 86%).
In this study, we did not find that predictive accuracy was further improved by the combination of FBB imaging with aMCI status. This is at odds with our previous findings in the AIBL study and likely reflects differences in subject selection and sample size.
Of those MCI who had dementia at 4 years, 21/26 (81%) were FBB+ at baseline. This figure is consistent with postmortem studies of MCI. For example, one study reported that 68%–88% of those with MCI who progressed to dementia had AD neuropathology at postmortem depending on the neuropathological criteria used. The rate of progression to dementia in FBB− MCI was 24% at 4 years. This figure is also close to the AIBL study which showed that 29% of MCI participants with a negative C-PiB Aβ baseline scan progressed to dementia at 3 years follow-up.
HV showed limited prognostic value. Consistent with these findings, a review pooling results across AD imaging biomarkers reported the highest diagnostic and prognostic accuracy for amyloid imaging and the lowest for MRI. HA is observed with ageing in the absence of neurodegenerative disease and also occurs in other dementias including DLB and FTLD. There is no universally accepted method to measure HV, define HA, or determine the threshold for detecting or predicting progression to AD. Consequently, we used a commercially available software package that is operator and site independent, and applied it to a 3D MP-RAGE MRI sequence to measure HV. We used the double ROC approach to determine the optimal threshold for distinguishing AD from healthy elderly in an unrelated study population and applied this threshold to determine the predictive accuracy of HV in this MCI cohort. It could be argued that the selected threshold might not be optimal, but we have subsequently applied different data-derived thresholds and were not able to improve the predictive accuracy of HV.
Aβ, HA and EM
The strengthening of the correlation between HV and EM and the concomitant weakening of the correlation between Aβ and EM over 2 years of serial study is consistent with the hypothetical sequence of biomarker change in AD, where progressive HA continues as Aβ accumulation plateaus after dementia develops. This weakening of the strength of the correlation between Aβ and EM with disease progression may account for some of the variability reported on the correlation of Aβ and EM between MCI studies. Furthermore, despite a non-significant mild strengthening of the correlation between Aβ and HV after 2 years, the correlation between Aβ and EM disappeared when accounting for HV. These results, along with our previous baseline report suggest that while at very early stages, Aβ and HV are significantly and independently related to EM impairment, and as disease progresses the relation between Aβ and EM is mediated by HV. This is in agreement with previous reports and supports the concept that disease modifying therapy, when available, should be given very early to achieve the greatest benefit.
Aβ Accumulation Over Time
Longitudinal studies of Aβ with PET indicate that accumulation takes approximately 15–20 years to progress from the Aβ+ threshold to the levels seen in most patients with mild AD. During this period, the slow increase in Aβ accumulation remains constant, but as disease progresses the rates of accumulation slow down, tending towards a plateau. In this study, we observed an increase in FBB retention at a rate of 0.037 SUVR/year in FBB+ participants, similar to the rates of 0.043 SUVR/year reported in longitudinal studies with C-PiB PET. This also suggests that FBB PET is sensitive enough to monitor the therapeutic response to anti-Aβ therapy.
Limitations
Among the limitations of the present study are the single-centre setting, the fact that repeat scans and neuropsychology tests were only obtained for 2 years and that the clinical evaluation was not blinded to FBB results beyond 2 years. Nonetheless, the 4-year follow-up reinforces the 2-year findings and provides valuable insight into the eventual outcome for both Aβ+ and Aβ− scan individuals with MCI. Other limitations include possible confounding effects from acetylcholinesterase inhibitor medication on cognition that was commenced in seven participants who completed FBB and neuropsychological studies to 2 years. Given that six out of these seven declined in EM scores at 24 months compared with baseline, impact on the overall results is likely to be minor.
Source...