Clinical outcomes up to 9 years after [18F]flutemetamol amyloid-PET in a symptomatic memory clinic population

Abstract

Background: Previous studies demonstrated increases in diagnostic confidence and change in patient management after amyloid-PET. However, studies investigating longitudinal outcomes over an extended period of time are limited. Therefore, we aimed to investigate clinical outcomes up to 9 years after amyloid-PET to support the clinical validity of the imaging technique.

Methods: We analyzed longitudinal data from 200 patients (M_age = 61.8, 45.5% female, M_MMSE = 23.3) suspected of early-onset dementia that underwent [¹⁸F]flutemetamol-PET. Baseline amyloid status was determined through visual read (VR). Information on mortality was available with a mean follow-up of 6.7 years (range = 1.1-9.3). In a subset of 108 patients, longitudinal cognitive scores and clinical etiological diagnosis (eDx) at least 1 year after amyloid-PET acquisition were available (M = 3.06 years, range = 1.00-7.02). VR - and VR + patients were compared on mortality rates with Cox Hazard's model, prevalence of stable eDx using chi-square test, and longitudinal cognition with linear mixed models. Neuropathological data was available for 4 patients (mean delay = 3.59 ± 1.82 years, range = 1.2-6.3).

Results: At baseline, 184 (92.0%) patients were considered to have dementia. The majority of VR + patients had a primary etiological diagnosis of AD (122/128, 95.3%), while the VR - group consisted mostly of non-AD etiologies, most commonly frontotemporal lobar degeneration (30/72, 40.2%). Overall mortality rate was 48.5% and did not differ between VR - and VR + patients. eDx at follow-up was consistent with baseline diagnosis for 92/108 (85.2%) patients, with most changes observed in VR - cases (VR - = 14/35, 40% vs VR + = 2/73, 2.7%, χ² = 26.03, p < 0.001), who at no time received an AD diagnosis. VR + patients declined faster than VR - patients based on MMSE (β = - 1.17, p = 0.004), episodic memory (β = - 0.78, p = 0.003), fluency (β = - 1.44, p < 0.001), and attention scores (β = 16.76, p = 0.03). Amyloid-PET assessment was in line with post-mortem confirmation in all cases; two cases were VR + and showed widespread AD pathology, while the other two cases were VR - and showed limited amyloid pathology.

Conclusion: In a symptomatic population, we observed that amyloid-status did not impact mortality rates, but is predictive of cognitive functioning over time across several domains. Also, we show particular validity for a negative amyloid-PET assessment, as these patients did not receive an AD diagnosis at follow-up.

Keywords: Diagnosis; Early-onset dementia; Neuropathology; Survival; [18F]flutemetamol amyloid-PET.

Fig. 1

Flow diagram of study design and patient selection In the panel on the left (gray), the study design of the original clinical study is described [5]. In the panels on the right, the selection (green) and exclusion (orange) process of the current longitudinal study is shown

Fig. 2

Kaplan–Meier survival curve Survival rate stratified by visual read (VR) status, showing no difference between visually negative and positive patients. The number of patients at each time point is also provided. Shaded represent the 95% confidence interval

Fig. 3

Flow diagram of changes in the etiological diagnosis of visually amyloid-negative cases AD, Alzheimer’s disease; FTLD, frontotemporal lobar degeneration; DLB, Lewy body disease; ND, neurodegeneration

Fig. 4

Longitudinal cognitive decline depends on visual read status Spaghetti plot illustrating the results of the linear mixed models. A Visually amyloid-positive subjects (red) show a steeper decline in global cognitive functioning as measured with the MMSE and B in episodic memory (visual association test) compared to visually amyloid-negative (blue) patients

Fig. 5

Post-mortem cases CAA, cerebral amyloid angiopathy; PSEN, presenilin; AD, Alzheimer’s disease; FTLD, frontotemporal lobar degeneration; CBD, corticobasal degeneration