Utilizing PiB PET, this study has demonstrated that Aβ
burden does increase at a slow rate over time. The rate is greater in nondemented persons with amyloid already present at baseline (high PiB) than in those without, and Aβ
deposition clearly continues in mild to moderate DAT patients. These increases in Aβ
deposition in AD are in agreement with 2 recent reports,33,34
and challenge the widespread belief that Aβ
burden reaches a plateau at a much earlier stage of the disease.23
Ninety-one percent of DAT patients showed an increase in PiB SUVR, ranging from 1 to 18% over 2 years (average increase of 0.08 SUVR or 3.4% per year). The fact that SUVR increases were positively correlated with baseline MMSE in the DAT group (see Supporting Information Fig) suggests that Aβ
deposition might slow down with disease progression in the later stages of DAT. Our preliminary results from 3-year follow-up point to the same conclusion. In nondemented participants, PiB retention increased on average by 1 to 2% per year, but there was marked individual variation that in many cases exceeded the range of test-retest variability for PiB PET (see ). We obtained an average of 3.5% for test-retest variability, whereas others have reported somewhat higher values.22,35
The individual variability observed here and in previous reports might explain the inconsistent results from prior brain biopsy studies, all of which had <6 subjects.6–8
Our findings are somewhat at odds with a 2-year follow-up study in 16 DAT subjects in which participants showed stable or even decreased PiB retention.22
Further review of that data showed that 60% of the subjects actually had a small increase in PiB retention.36
In other recently reported studies, significant increases in PiB retention were found in individuals but not on group analysis.21,23,24
The larger sample size of our study is the most likely reason that we have found significant group increases in PiB. Not surprisingly, the significant increases were in high PiB groups and DAT, with lesser, nonsignificant increases in the low PiB groups.
The longitudinal aspects of this study help elucidate the relationship between Aβ burden and cognitive trajectory in healthy individuals and MCI subjects. That is, those who had high PiB retention were much more likely to progress from normal to impaired cognition or from MCI to AD. Aβ deposition appears to be an age-dependent process, particularly at the early stages of the disease course (normal cognition or MCI). Therefore, the cross-sectional aspect of the study allows comparison of age-matched HC with MCI subjects and with AD patients who present with either early onset or late onset of the disease.
Our data also show, in agreement with previous studies, that Aβ
deposition is strongly associated with ApoE status, particularly in nondemented individuals.18–20
There is a dose-dependent association between increases in Aβ
deposition and number of ε
4 alleles when the groups are examined together but not when the groups are examined separately, and although both ε
4 homozygotes and heterozygotes had higher increases in Aβ
deposition than ε
4 noncarriers in DAT, the difference was not significant. These findings suggest that the number of ε
4 alleles does influence Aβ
deposition in the prodromal DAT period, but not once DAT is established. This is to some extent at odds with a recent study of 24 DAT patients, where a significant ApoE dose-dependent increase in Aβ
burden was reported.34
In our study, there were no significant differences in baseline MMSE, CDR, or PiB SUVR in the DAT group between ε
4 homozygotes, heterozygotes, and noncarriers, but the baseline PiB SUVR values were much higher than those reported by Grimmer and colleagues (2.3 vs 1.7, respectively).34
Although only a trend, a higher increase in Aβ
deposition was observed in those DAT patients with a lower baseline PiB SUVR, which added to the fact that there were no significant differences in PiB SUVRs between ε
4 carriers and noncarriers in DAT subjects, might explain the different results.
This study found a higher rate of conversion (29%/year) from MCI to DAT than the usually reported 8 to 16%/year.37–39
The likely explanation is that MCI participants recruited from a tertiary referral memory disorders clinic are more likely to progress than subjects detected through community screening. In addition, our MCI cohort was also older (mean age, 73.4 years) than in many studies, and had a higher percentage (90%) of ApoE ε
4 carriers. Thus, they likely represent what has been termed late MCI.40
Other features of our MCI participants indicate that most fell toward the severe end of the MCI spectrum, with episodic memory impairment and baseline PiB retention equivalent to the DAT group. Indeed, the only features—besides informant report of functional capabilities—that separated our MCI from DAT were the better performance on MMSE, CDR, and nonmemory cognitive tests (see ). Aggregating the results from 3 recently reported smaller PiB follow-up studies reveal figures very similar to those reported here, with 26 of 41 MCI subjects with high PiB progressing to DAT.24,41,42
The fact that 67% of MCI subjects with high PiB retention progressed to DAT over 2 years, compared to 5% of those MCI subjects with low PiB retention, suggests that MCI subjects with high Aβ
deposition in the brain are 13× more likely to progress to DAT than those with low Aβ
As previously reported,14
there was a correlation between PiB retention and memory impairment in the MCI group, but this has not been consistently reported by other PiB studies assessing amnestic MCI. This discrepancy may be due to the fact that 12 of 65 MCI subjects had nonamnestic MCI. Nonamnestic MCI subjects are more likely to have low PiB retention and therefore drive a correlation between memory scores and PiB retention within the MCI group. Some measures of progressive cognitive decline such as MMSE and nonmemory scores correlated weakly with the increase in Aβ
burden as measured by PiB. Technical factors could mask a stronger correlation such as the introduction of medication for symptomatic treatment of DAT. However, only 3% of HC, 11% of MCI, and 3% of DAT subjects commenced acetyl cholinesterase inhibitors or memantine between the baseline and follow-up cognitive assessments. Given their relatively weak effects, particularly on episodic memory tasks, it is unlikely that this has influenced our findings. Worsening partial volume effect from progressive brain atrophy could also mask an increase in PiB retention over time, but we found that correction of PiB images for partial volume effect with a 3 compartment MRI model did not significantly alter the percentage difference in PiB measures over a 20-month period. For longer follow-up, this may be more important. We also found no evidence of Aβ
deposition in the cerebellar cortex that would either mask an increase or show a reduction in SUVR at follow-up, with no differences in cerebellar cortex PiB SUVs between baseline and follow-up within our study groups. Finally, the fact that the annual rate of increase in Aβ
deposition is similar to the test-retest variability of PiB measurements along with the potentially high test-retest variability of cognitive assessments also could pose difficulties for establishing a correlation between increase in amyloid and reduction in cognition. However, in numerous cases, definite progression of cognitive decline occurred with no increase or even a decrease in PiB retention (see ). Therefore, the more likely reason for finding only a weak correlation between change in PiB retention and change in cognition is that by the time cognitive decline manifests, Aβ
has already triggered downstream mechanisms that play a more direct role in driving cognitive decline.9
Such mechanisms may include formation of neurofibrillary tangles, neuroinflammatory processes, and oxidative stress driven by Aβ
oligomers, leading to progressive synaptic failure and neuronal death.1
Our data have several implications for therapeutic trials. The slow rate of Aβ deposition indicates that the time window for altering amyloid accumulation prior to DAT may be very wide. It also supports the growing consensus that antiamyloid therapy may need to be given early in the course of the disease, perhaps even before symptoms appear, to be effective, and that downstream mechanisms may also need to be addressed to successfully prevent the development of DAT.
This study also has shown that Aβ deposition precedes the development of objective cognitive impairment in at least some healthy elderly individuals, and that imaging Aβ may be useful to stratify the risk of objective cognitive decline over the following 2 years. Based on our results, there is a >98% (95% confidence interval, 90–100%) chance that an asymptomatic elderly individual with low PiB retention will remain cognitively stable over 20 months, in contrast to the 16% risk of developing MCI or DAT in those with high PiB retention and a 25% risk by 3 years. It is likely that our healthy control population is biased toward the early development of DAT. Respondents to advertising may be concerned about subtle subjective cognitive decline or have a family history of DAT. Therefore, although these risks may not equate to those of the general population, they are relevant for those persons most likely to present to a medical practitioner requesting investigation for very early DAT.
In conclusion, Aβ deposition increases slowly and continuously from cognitive normality to moderately severe DAT. A significant PiB SUVR increase at 20-month follow-up was observed in DAT and MCI subjects, and significant PiB SUVR increases were also observed in the subgroup of HC and MCI subjects who presented with high PiB at baseline. Aβ deposition was associated with ApoE genotype, with homozygote ε4 carriers presenting with higher PIB SUVR and higher PiB SUVR increases at follow-up. Extensive Aβ deposition precedes cognitive impairment and is associated with a higher risk of cognitive decline and progression from MCI to DAT over 1 to 2 years, denoting the nonbenign nature of Aβ deposition. However, cognitive decline is not closely related to change in Aβ burden, suggesting that downstream factors have a more direct effect on symptom progression.