Testing causality in the association between amyloid‐β and tau in genetically identical twins.

Background: The amyloid‐cascade hypothesis of Alzheimer's disease posits that there is a causal relationship between amyloid‐β (Aβ) and tau pathology. However, the mechanism through which Aβ would cause tau remains unclear. We cannot exclude the possibility of an additional underlying factor influencing Aβ at one time‐point and tau at a later time‐point. One such potential underlying factor can be genetic variation among individuals. This study aimed to examine the relationship between Aβ and tau, taking potential genetic confounding into account using a genetically identical twin design. Methods: We included 70 identical twins (35 pairs) from the EMIF‐AD PreclinAD study with normal baseline cognition. We extracted baseline global [18F]flutemetamol (Aβ)‐PET binding potential (BPND) and 4‐year follow‐up [18F]flortaucipir (tau)‐PET BPND in a temporal meta‐region‐of‐interest. If Aβ is causally related to tau, the twin with more Aβ should also have more tau compared to the co‐twin. We tested this using regional and voxel‐wise within‐pair difference models, in which the within‐pair difference in Aβ is regressed on the within‐pair difference in tau. The resulting effect represents an association that is free of confounding due to factors that are shared between twins of the same pair, which includes genetic factors (twins are genetically identical), but also shared demographic (e.g. sex, age) and shared environmental (e.g. raised in the same family) factors. Results: Participant characteristics are shown in Table 1. Total group analysis showed that global Aβ‐PET BPND was associated with temporal tau‐PET BPND (β=0.49, p<0.001) (Fig‐1A). The twin‐difference model replicated this. We observed a highly comparable effect size for the association between within‐pair differences in global Aβ‐PET and within‐pair differences in temporal tau‐PET (β=0.56, p<0.001) (Fig‐1B). Compared to the total group voxel‐wise analysis, we observed an overlapping (and more widespread) pattern when regressing within‐pair differences in global Aβ‐PET on within‐pair differences in voxel‐wise tau‐PET, except for effects on tau in the medial temporal lobe (Fig‐2). Conclusion: Our within‐pair difference results show that the association between Aβ and tau is minimally affected by confounding by genetic variation. Our voxel‐wise results support a causal association between global Aβ and tau outside of the medial temporal lobe. [ABSTRACT FROM AUTHOR]