Predictive Accuracy of Amyloid Imaging for Progression From Mild Cognitive Impairment to Alzheimer Disease With Different Lengths of Follow-up: A Meta-analysis

In the past decade, amyloid deposition has been shown to begin many years before the clinical symptoms of dementia in mild cognitive impairment (MCI) due to Alzheimer disease (AD). Longitudinal studies with different follow-up durations have suggested that (11)C-Pittsburgh compound B positron emission tomography ((11)C-PIB-PET) may play a role in stratifying patients with MCI into risk levels for developing AD. However, the predictive accuracy of amyloid imaging for the progression from MCI to AD with different follow-up durations has not yet been systematically evaluated. A formal systematic evaluation of the sensitivity, specificity, and other properties of (11)C-PIB-PET was performed. This study aimed to systematically review and meta-analyze published data on the diagnostic performance of (11)C-PIB-PET for predicting conversion to AD in patients with MCI and to determine whether long-term follow-up has a positive effect on predictive accuracy. Relevant studies were systematically identified through electronic searches, which were performed in MEDLINE (OvidSP), EMBASE (OvidSP), BIOSIS Previews (ISI Web of Knowledge), Science Citation Index (ISI Web of Knowledge), PsycINFO (Ovid SP), and LILACS (Bireme). The methodological quality of each study was assessed by QUADAS-2. Sensitivities and specificities of (11)C-PIB-PET in individual studies were calculated, and the studies underwent meta-analysis with a random-effects model. A summary receiver-operating characteristic curve (SROC) was constructed with the Moses-Shapiro-Littenberg method. Pooled estimates of sensitivity, specificity, positive likelihood ratio (LR+), negative likelihood ratio (LR−), diagnostic odds ratio (DOR), and the SROC curve of each subgroup were determined. Heterogeneity was tested, and potential sources for heterogeneity were explored by assessing whether certain covariates significantly influenced the relative DOR. Eleven eligible studies consisting of a total of 352 patients with MCI at baseline were included. Overall, the studies were of moderate-to-high methodological quality. The sensitivity and specificity of (11)C-PIB-PET for predicting conversion to AD ranged from 83.3% to 100% and 41.1% to 100%, respectively, with pooled estimates of 94.7% (95% confidence interval [CI]: 89.8%–97.7%) and 57.2% (95% CI: 50.1%–64.2%), respectively. Moderate heterogeneity was observed for specificity between the included studies (I(2) = 42.1%). The pooled estimates for the long-term follow-up subgroup were 95.5% sensitivity (95% CI: 84.5%–99.4%) and 72.4% specificity (95% CI: 59.1%–83.3.8%), whereas the pooled estimates for the short-term follow-up subgroup were 94.4% sensitivity (95% CI: 88.2%–97.9%) and 51.0% specificity (95% CI: 42.6%–59.5%). Homogeneity in each subgroup was significantly higher than that of the included studies, and most diagnostic indicators in the long-term follow-up subgroup were far superior to those in the short-term follow-up subgroup or the entire group. Not all of the methodological quality scores of studies included in this systematic review were high. Current evidence suggests that prolongation of the follow-up duration tended to yield greater accuracy of (11)C-PIB-PET for predicting the progression from MCI to AD. In particular, the specificity, which reflects the exploratory nature of the use of amyloid imaging to identify the process of MCI to AD, was improved with a longer follow-up period.