Comparing parametric methods for longitudinal measurement of β‐amyloid pathology with PET in elderly individuals
Fiona Heeman, Janine Hendriks, Catarina Tristão‐Pereira, Lyduine E. Collij, Peter Young, Bart N.M. van Berckel, Pieter Jelle Visser, Bernard J Hanseeuw, Rik Vandenberghe, Valentina Garibotto, Giovanni B Frisoni, Daniele Altomare, Mahnaz Shekari, Christopher Buckley, Gill Farrar, Mark E Schmidt, Rossella Gismondi, Andrew W. Stephens, Craig W Ritchie, Catriona Wimberley, Pablo Martinez‐Lage, Richard Manber, Robin Wolz, Juan Domingo Gispert, Michael Schöll, Isadora Lopes Alves, Frederik Barkhof, David Vállez García, Adriaan A. Lammertsma, Maqsood Yaqub- Psychiatry and Mental health
- Cellular and Molecular Neuroscience
- Geriatrics and Gerontology
- Neurology (clinical)
- Developmental Neuroscience
- Health Policy
- Epidemiology
Abstract
Background
Amyloid‐ß (Aß) PET is commonly used for studying the earliest phases of Alzheimer’s disease (AD) in cognitively unimpaired (CU) individuals. In this group, the expected changes in Aß pathology are small, which emphasizes the importance of selecting a method with the highest possible precision for measuring these changes. This study compared several methods for quantifying Aß pathology longitudinally in mostly CU individuals from the AMYPAD‐PNHS cohort.
Method
Participants were scanned with either [18F]flutemetamol (baseline N = 360, follow‐up N = 243) or [18F]florbetaben (N = 66 baseline, N = 49 follow‐up), according to a dual‐time window protocol(Table 1). A subset of participants had = 2 timepoints available (N = 206, N = 43, respectively). SUVRs were calculated, and parametric modelling was performed using RPM, SRTM2, RLogan, MRTM0, MRTM and MRTM2 using PPET software to generate parametric BPND or DVR images for a global cortical region, all with cerebellar cortex as reference tissue. Annual percentage change (APC) was calculated and compared between methods using an ANOVA. To check for changes in Aß pathology over time, a linear mixed effects model was used. Bland‐Altman analyses were used to explore agreement between SUVR, and the outcome parameter from the parametric methods. All analyses were run per tracer, separately for visual Aß‐positive and Aß‐negative individuals, with a p‐value threshold of p<0.05.
Result
For both tracers, there were no differences in APC between methods. However, method‐dependent differences in interquartile range of the APC were observed(Figure 1,2). For [18F]flutemetamol, only the Aß‐negative group showed a significant change in Aß pathology for SUVR and SRTM2 (both ‐0.004/year). For [18F]florbetaben, only the Aß‐positive group showed a significant change in Aß pathology for all methods except MRTM2 (range: +0.020‐0.032/year). No significant bias was observed between APC in SUVR and the other methods for either tracer or Aß‐group.
Conclusion
There were no significant differences in APC between methods, despite method‐dependent variability. In the [18F]flutemetamol‐cohort, two methods showed a minimal decrease in Aß pathology in the Aß‐negative‐group, possibly related to measurement variability, which requires further investigation. In the [18F]florbetaben‐cohort, which included more participants with a very mild cognitive impairment, a significant increase in Aß pathology was observed in the Aß‐positive‐group, except for MRTM2.