Smailagic · The Cochrane database of systematic reviews 2015 · systematic review and meta-analysis · n=14 studies (421 participants)

¹⁸F-FDG PET for the early diagnosis of Alzheimer's disease dementia and other dementias in people with mild cognitive impairment (MCI).

Cited 205 times in the scientific literature.

Level 1 - systematic review of randomized trials

Systematic review and meta-analysis of diagnostic test accuracy studies

PubMed 25629415 · doi:10.1002/14651858.CD010632.pub2 · record verified 2026-08-28

What was done

This Cochrane systematic review searched multiple databases up to January 2013 for delayed-verification studies evaluating the diagnostic accuracy of brain ¹⁸F-FDG PET to predict clinical conversion from mild cognitive impairment (MCI) to Alzheimer's disease dementia or other dementias. Two reviewers independently extracted data and assessed study quality. Hierarchical summary ROC (HSROC) modeling was used to estimate pooled sensitivity and likelihood ratios at fixed specificity values.

What was found

Fourteen studies involving 421 participants met inclusion criteria. For conversion from MCI to Alzheimer's disease dementia, individual study sensitivities ranged from 25% to 100% and specificities ranged from 15% to 100%. At the median study specificity of 82%, the HSROC model estimated sensitivity was 76% (95% CI: 53.8 to 89.7), with a positive likelihood ratio of 4.03 (95% CI: 2.97 to 5.47) and a negative likelihood ratio of 0.34 (95% CI: 0.15 to 0.75). For all-cause dementia (5 studies), sensitivity ranged from 46% to 95% and specificity from 29% to 100%, but no meta-analysis was performed. More than 50% of studies had poor methodological quality for the index test, and most had unclear risk of bias for participant selection and reference standards.

Why it matters

Due to significant heterogeneity, lack of standardized cutoff thresholds, and prevalent methodological weaknesses, current evidence does not support routine clinical use of ¹⁸F-FDG PET scans to predict dementia progression in people with MCI.

Limits

The total sample size was small (421 participants across 14 studies). Included studies had poor reporting, lacked defined thresholds for test positivity, had poor methodological quality in over half of the index test evaluations, and exhibited unclear risk of bias for participant selection and reference standards.

Cited by