A logarithm derived from Mokken scaling analysis of the Addenbrooke’s Cognitive Examination–Revised discriminates with high accuracy Alzheimer’s disease from behavioral variant frontotemporal dementia
Viviane Amaral‐Carvalho, Thais Bento Lima‐Silva, Luciano Inácio Mariano, Leonardo Cruz de Souza, Henrique Cerqueira Guimarães, Valeria S Bahia, Ricardo Nitrini, Maira Tonidandel Barbosa, Monica Sanches Yassuda, Paulo Caramelli- Psychiatry and Mental health
- Cellular and Molecular Neuroscience
- Geriatrics and Gerontology
- Neurology (clinical)
- Developmental Neuroscience
- Health Policy
- Epidemiology
Abstract
Background
The ACE‐R is an accurate and brief cognitive battery for the detection of mild dementia, especially for the discrimination between Alzheimer’s disease (AD) and behavioral variant frontotemporal dementia (bvFTD). The aim of this study was to develop a logarithm based on discriminative items of the ACE‐R for AD combined with relevant demographic characteristics that may help to differentiate AD from bvFTD.
Method
The ACE‐R was administered to 102 patients with mild dementia due to probable AD and 37 patients with mild bvFTD from two Brazilian research centers. All individuals were submitted to the Mattis Dementia Rating Scale (DRS) and the ACE‐R. The performance of the patients was compared and analyzed. Mokken scaling analysis was applied to identify the latent trait on the AD Group, and later, multivariate logistic regression and ROC curve analysis were carried out to assess the ACE‐R’s ability to discriminate AD and bvFTD.
Result
Mean ± SD total scores in the ACE‐R were 70.2 ± 10.8 in AD and 72.2 in bvFTD. AD Mokken ACE‐R (MokACE‐R‐AD) comprises 12 items measuring the same latent concept. In our sample, logistic regression with cross‐validation pointed that MokACE‐R‐AD, Age, Male sex, and items Orientation to time and location and the item of Memory (name and address recall) share importance as independent variables (p<0.05) to differentiate AD from bvFTD. The proposed logarithm reached an area under the ROC curve of 0.922, with 88% sensitivity, 88% specificity, 71% PPV and 96% NPV.
Conclusion
The new logarithm using the ACE‐R items achieved high diagnostic accuracy in identifying mild AD versus bvFTD. Furthermore, the final ROC curves showed the superiority of the model proposed in relation to the analysis of the subscales individually or the VLOM ratio, as the literature previously reported. Further analysis in larger samples, with biomarkers or pathological confirmation, are necessary to confirm these findings.