Clinical Clues to the Diagnostic Yield of Genetic Testing in Adults With Late-Onset Behavioral Change

The diagnosis of behavioral variant frontotemporal dementia is often difficult because behavioral change has a broad differential diagnosis. Genetic testing may aid in the diagnostic process. We investigated the prevalence of pathogenic genetic variants (PGVs) in individuals referred to our memory clinic with late-onset behavioral change and identified clinical "red flags" for PGV carriership, specifically in diagnostically ambiguous cases.

Individuals presenting with late-onset behavioral change were included from the Late Onset Frontal Lobe Syndrome study (n = 88), Social Brain Project (n = 265), and Amsterdam Dementia Cohort (n = 349). PGV prevalence was calculated. Among diagnostically ambiguous individuals at baseline, univariate logistic regression models were fitted to identify clinical cues for PGV carriership. Based on these results, we fitted multivariate logistic regression models. We also assessed the association of cortical thickness and subcortical volumes with PGV carriership.

Among 702 individuals, 228 received a diagnosis in the frontotemporal lobar degeneration (FTLD) spectrum at baseline and 474 were diagnostically ambiguous. A total of 106 individuals (15%) carried a PGV (20% in FTLD; 13% in ambiguous cases). The most common PGV in both groups was the C9orf72 repeat expansion (56% and 57%), followed by microtubule-associated protein tau (13% and 11%) and GRN (11% and 10%). A Huntingtin repeat expansion was found in 5 ambiguous cases. In multivariate analyses, PGV carriership was associated with a family history of dementia (OR_FH [95% CI] 3.1 [1.7-5.5], p < 0.001), younger age (OR_age,10yr [95% CI] 2.0 [1.4-2.9], p < 0.001), female sex (OR_female [95% CI] 2.0 [1.1-3.6], p = 0.02), a Frontal Assessment Battery score <13 (OR_FAB [95% CI] 2.1 [1.1-4.1], p < 0.05), and medial temporal and posterior atrophy (OR_MTA [95% CI] 3.2 [1.0-10], p < 0.05; OR_PCA [95% CI] 13 [2.0-81], p < 0.01). In additional MRI analyses, atrophy in the thalamus (standardized β ± standard error = -1.26 ± 0.28), putamen (-1.15 ± 0.24), and superior parietal cortex (-1.07 ± 0.22) was most strongly associated with PGV carriership.

Genetic testing for dementia-associated genes should be considered in all late-onset behavioral change cases. While we propose several clinical cues as "red flags" for PGV carriership, their absence should not preclude genetic counseling. The higher PGV prevalence among diagnostically ambiguous women suggests that FTLD may be underrecognized in women compared with men.