David Eidelberg, MD Seminar #1

Abstract: Brain imaging has been used extensively to identify and validate disease-specific functional networks as biomarkers in neurodegenerative disorders [Perovnik et al. Nat Rev Neurol 2022]. Although disease networks are highly reproducible across patient populations, it is not known whether these topographies reflect pathological connectivity patterns that worsen with advancing disease, or beneficial adaptations that may be promoted by treatment. To distinguish between these possibilities, we used graph theory to study connectivity patterns in an extensively validated metabolic network termed the Parkinson’s disease-related metabolic pattern (PDRP) [Schindlbeck and Eidelberg Lancet Neurol 2018; Ko et al. Cereb Cortex 2018; Perovnik et al. Nat Rev Neurol 2022; Barbero et al. Neurotherapeutics 2023]. In particular, we focused on assortativity, a metric that captures the tendency of connections to form between nodes with similar properties. In graph theory, assortativity has been linked to network stability. High values are associated with ine cient information flow through the network, and with increased susceptibility to fragmentation. Low assortativity, by contrast, is defined by greater diversity of connections, resulting in improved e ciency of information flow and greater network robustness [Noldus and Van Mieghem J Complex Networks 2015; Barabasi Network Science 2016].

Accordingly, PDRP assortativity increased with disease progression in multiple independent patient populations [Vo et al. Cereb Cortex 2023]. Moreover, this metric was elevated in clinically aggressive PD genotypes (GBA-1 variants) compared to sporadic disease, and was reduced in patients with slow progression mutations (LRRK2-G2019S). A similar dichotomy was seen in the response to PD treatment with levodopa infusion compared to subthalamic nucleus (STN) deep brain stimulation (DBS). Despite titration to the same degree of motor improvement, the two interventions had opposite e ects on PDRP assortativity, with increasing values (i.e., network destabilization) during levodopa infusion, but with reductions to near normal levels (i.e., network stabilization) during STN stimulation. Notably, an analogous reduction in assortativity was observed in the PDRP space 12 months after STN AAV2-GAD gene therapy, an experimental surgical procedure that slowly remodels basal ganglia-cortical motor networks [Niethammer et al. Sci Transl Med 2018; Vo et al. Cereb Cortex 2023]. In aggregate, the findings suggest that stereotyped changes in network architecture underlie disease progression and the response to treatment in PD patients.