Acoustically Altered Speech Perception in Frontotemporal Dementia Syndromes

Hearing impairment is a major clinical issue in dementia. Acoustically altered speech (AAS) is a typical real-world listening challenge, but markers of AAS perception, especially in frontotemporal dementias (FTDs), remain undefined. We investigated whether AAS perception differentiates FTD syndromes and predicts real-world hearing symptoms.

In this cross-sectional case-control study, participants meeting diagnostic criteria for FTD and primary progressive aphasia syndromes were recruited at a UK tertiary center. Participants performed pure-tone audiometry (PTA) and 3 AAS perception tests: speech-in-babble (SIB), spectrally filtered speech (SFS), and time-compressed speech (TCS). Real-world hearing abilities were assessed through the modified Amsterdam Inventory for Auditory Disability and Handicap (mAIAD). Group characteristics were compared using analysis of variance and Fisher exact tests. Regression models adjusted for PTA and Wechsler Abbreviated Scale of Intelligence Matrices scores compared group performance on AAS tests. The Spearman test examined correlations with mAIAD scores.

Among 44 participants, the mean (SD) age was 66 (6.23) years and 13 (29.5%) were female. 29 patients with FTD (6 with nonfluent/agrammatic variant PPA [nfvPPA], 8 with semantic variant PPA [svPPA], and 15 with behavioral variant FTD [bvFTD]) were compared with 15 age-matched controls. Patients with nfvPPA had significantly higher mean PTA thresholds than controls (mean difference 12 decibels [95% CI 0.7-23.4], p = 0.034). Those with nfvPPA performed significantly worse on the SFS test than all other groups (vs controls: difference 6.3 [9.4-3.1], p < 0.001; vs svPPA group: difference 2.4 [0.5-7.3], p = 0.0018; vs bvFTD group: difference 2.0 [1.45-7.0], p = 0.0012) and significantly worse on the TCS test than controls and patients with bvFTD (vs controls: difference 88.2 [29.7-146.7], p < 0.01; vs bvFTD group: difference 76.1 [26.7-125.5], p < 0.01). The svPPA group performed significantly worse than controls on TCS (difference 57.4 [16.38-98.4], p = 0.01). Patients with nfvPPA (difference 27.2 [9.3-45.2], p < 0.001) and svPPA (difference 25.7 [11.8-39.5], p = 0.0013) had significantly worse mAIAD scores than controls. Over the combined patient cohort, mAIAD scores correlated more strongly with AAS than PTA thresholds (TCS, ρ = 0.64, p < 0.001; SFS, ρ = 0.53, p = 0.003; SIB, ρ = 0.5, p = 0.004; PTA, ρ = 0.37, p = 0.028).

AAS perception stratifies FTD syndromes and constitutes a "real-world audiogram" in these diseases.