Purpose: To study the imaging patterns of Posterior cortical atrophy (PCA) and Dementia with Lewy bodies (DLB) on fluoro-deoxyglucose positron emission tomography computed tomography ([F]FDG PET/CT), identify areas of overlap and differences and to develop a prediction model to assist in diagnosis using univariate and multivariate analysis.
Methods: A retrospective analysis of 72 patients clinically suspected of having posterior dementia was done. All patients underwent [FF]FDG PET/CT of the brain and dopamine transporter imaging with [[Tc]TRODAT-1 SPECT scan on separate days. The patients were divided into PCA with normal TRODAT uptake (n=34) and DLB with abnormal TRODAT uptake (n=38). The FDG PET/CT uptake patterns were recorded and areas of significant hypometabolism by z score analysis were considered as abnormal. Receiver operator characteristics (ROC) curve analysis was used to determine cutoff z scores and binary logistic regression analysis was used to determine the Odds ratio of being in the predicted groups.
Results: Significantly hypometabolism was found in parieto-temporo-occipital association cortices and cingulate cortices in PCA patients. DLB patients showed significantly reduced uptake in the visual cortex. No significant difference was found between z score of occipital association cortex which showed hypometabolism in both groups. The cut-off z-score values derived from the ROC curve analysis were as follows- parietal association (cut-off-3, sensitivity-65.6%, specificity - 68.7%), temporal association (cut-off-2, sensitivity-78%, specificity-75%) and posterior cingulate (cut-off-0.5, sensitivity-93.7%, specificity-40.6%), their respective Odds ratio (with 95% confidence interval) for being in the PCA group as derived from univariate logistic regression were 3.66 (1.30-10.32), 10.71 (3.36-34.13) and 7.85 (1.57-39.17). The cut-off z score of primary visual cortex as derived from ROC curve was zero with sensitivity of 87.5%, specificity of 71.9%, and the Odds ratio for being the in the DLB group was 24.7 with 95% confidence interval of 5.99-101.85.
Conclusion: [F]FDG PET may be useful as a non-invasive diagnostic modality in differentiating the two posterior cortical dementias, despite significant overlap. Primary visual cortical hypometabolism can serve as an independent diagnostic marker for DLB, even in the absence of TRODAT imaging.