The cingulate island sign in a mixed memory clinical cohort: Prevalence and diagnostic accuracy

Introduction: Visual rating of the cingulate island sign (CIS) on [18F]fluorodeoxyglucose ([18F]FDG) positron emission tomography (PET) has a high specificity for dementia with Lewy bodies (DLB) in selected cohorts such as DLB versus Alzheimer's disease (AD). In a mixed memory clinical population this study aimed to uncover the prevalence of CIS, the diagnostic accuracy for DLB, and the relationship between CIS and disease severity. Methods: CIS on [18F]FDG-PET was retrospectively assessed with the visual CIS rating scale (CISRs) in 1000 patients with a syndrome diagnosis of mild cognitive impairment (MCI) or dementia with no restrictions in etiological diagnosis. Results: In this cohort 24.3 % had a CISRs score >= 1 and 3.5 % had a CISRs score = 4. The prevalence of a CISRs score >= 1 was highest in DLB (74.0 %, n = 57). A CISRs score >= 1 was present in at least 9 % in other diagnostic groups. The prevalence of CIS across disease severities showed no statistically significant difference (p = 0.23). To differentiate DLB from non-DLB the optimal cut-off was a CISRs score >= 1 (balanced accuracy = 77.1 %) in MCI/mild dementia and a CISRs score >= 2 (balanced accuracy = 80.6 %) in moderate/severe dementia. The positive predictive value of a CISRs score = 4 for DLB was 57.7 % in MCI/mild dementia and 33.3 % in moderate/severe dementia. Conclusion: The CISRs is useful in differentiating DLB from other etiologies in a mixed memory clinical population. Balanced accuracy and positive predictive value may vary across disease severities in the population studied.