Objective: Identify common and disease specific molecular drivers of rate of change in Neurofilament light (NfL) in Huntington’s disease (HD), Parkinson disease (PD) and Alzheimer disease (AD) using reverse engineering forward simulation, REFS^TM platform.

Background: NfL is recognized as a promising biomarker for diagnosis, prognosis and monitoring for both clinical and research purposes in neurodegenerative conditions. Multi-modal data in HD, PD and AD was analyzed using Aitia’s patented causal AI and simulation technology, REFS^TM. Causal connectivity of biomarkers or genetic traits to NfL was reverse-engineered and simulated annealing. The causal effect was appropriately estimated in silico by Do-calculus.

Method: The platform developed two complementary ensembles of Bayesian causal networks to discover causal relationships among clinical, single nucleotide polymorphisms, protein transcriptomic, biomarker and imaging variables in PD, AD and HD to investigate drivers of rate of change of NfL. First model included participants from TRACK-HD/TRACK-ON (n=73) with HD-ISS stage 2 and 3 and participants from ADNI (n=275) diagnosed with MCI and dementia, while a second model included cohort from TRACK-HD/TRACK-ON studies (n=74) with PD participants from the PPMI study (n=311). Features identified to have the strongest causal effects across the network were further evaluated in disease specific simulations.

Results: In the first model, a total of 218 causal drivers of blood NfL change rate were identified, which were then grouped as 18 HD-AD-common, 18 HD-specific, 6 AD-specific, and 22 shared drivers with distinct disease-specific effects. Majority of disease-specific drivers showed opposite directionality of causal effects (i.e., risk vs. protective) in HD and AD. Similarly, the second model selected 59 HD-PD-common drivers as well as 55 disease-specific drivers. While the blood NFL level measured at baseline was a strong driver of its future change rate in all three diseases, the nature of that relationships was markedly different between diseases.

Conclusion: Digital twins showed that higher baseline NfL in blood would likely lead to accelerated accumulation in both the AD and PD, but to slower accumulation in HD. Small samples size in HD may make the HD-specific effect estimation less powered than the other two disease specific effect or common effect.

« Back to 2023 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/gemini-digital-twins-identified-neuro-common-and-disease-specific-drivers-of-rate-of-change-in-nfl-in-huntingtons-disease-parkinson-disease-and-alzheimer-disease/