Analysis of clinical failure of anti-tau and anti-synuclein antibodies in neurodegeneration using a quantitative systems pharmacology model

Abstract

Misfolded proteins in Alzheimer's disease and Parkinson's disease follow a well-defined connectomics-based spatial progression. Several anti-tau and anti-alpha synuclein (aSyn) antibodies have failed to provide clinical benefit in clinical trials despite substantial target engagement in the experimentally accessible cerebrospinal fluid (CSF). The proposed mechanism of action is reducing neuronal uptake of oligomeric protein from the synaptic cleft. We built a quantitative systems pharmacology (QSP) model to quantitatively simulate intrasynaptic secretion, diffusion and antibody capture in the synaptic cleft, postsynaptic membrane binding and internalization of monomeric and oligomeric tau and aSyn proteins. Integration with a physiologically based pharmacokinetic (PBPK) model allowed us to simulate clinical trials of anti-tau antibodies gosuranemab, tilavonemab, semorinemab, and anti-aSyn antibodies cinpanemab and prasineuzumab. Maximal target engagement for monomeric tau was simulated as 45% (semorinemab) to 99% (gosuranemab) in CSF, 30% to 99% in ISF but only 1% to 3% in the synaptic cleft, leading to a reduction of less than 1% in uptake of oligomeric tau. Simulations for prasineuzumab and cinpanemab suggest target engagement of free monomeric aSyn of only 6-8% in CSF, 4-6% and 1-2% in the ISF and synaptic cleft, while maximal target engagement of aggregated aSyn was predicted to reach 99% and 80% in the synaptic cleft with similar effects on neuronal uptake. The study generates optimal values of selectivity, sensitivity and PK profiles for antibodies. The study identifies a gradient of decreasing target engagement from CSF to the synaptic cleft as a key driver of efficacy, quantitatively identifies various improvements for drug design and emphasizes the need for QSP modelling to support the development of tau and aSyn antibodies.

Figure 1

Schematic representation of the combined PBPK-QSP model. The model combines a minimal PBPK model (top left) with various brain compartments using appropriate volumes (bottom left) for derivation of both antibody concentrations in the various compartments (plasma, CSF, ISF, synaptic cleft) and levels of tau/aSyn in CSF. Availability of the antibodies is modulated by tortuosity and off-target capture by microglia in the ISF. The QSP part simulates key processes in the synaptic cleft, including secretion of free monomeric (blue) and oligomeric Tau/aSyn (yellow) from the presynaptic synaptic compartment, diffusion, binding to antibodies, binding to postsynaptic “acceptors (HSPG and LRP1)” and internalization of 2 forms of Tau/aSyn (monomeric and two oligomeric forms) into the postsynaptic neuronal compartment. This is assumed to be a proxy for misfolded protein propagation.

Figure 2

Plasma PK and pharmacodynamic effect. (a) Plasma PK profiles of the five antibodies at their highest dose from the fitted profiles in the Phase 1 study. (b) Simulated and experimentally observed pharmacodynamic effect of different gosuranemab doses on free CSF monomeric tau, normalized to baseline, as reported in their Phase 1 study (dots are experimentally reported data).

Figure 3

Effect of gosuranemab on tau in different compartments in a 52-week PSP study. Effect of gosuranemab on free tau (a) in different compartments at 4000 mg Q4 W. The pharmacology leads to an almost full depletion of free CSF and ISF monomeric and oligomeric tau and a 10–15% decrease in monomeric synaptic cleft tau but no change in synaptic cleft or neuronal uptake of oligomeric tau. (b) The reduction in neuronal oligomeric tau uptake by gosuranemab is very small relative to placebo (0.008%).

Figure 4

Effect of tilavonemab on free monomeric and oligomeric tau in different compartments in PSP. (a) Simulated profiles of free monomeric tau in CSF, ISF and synaptic cleft oligomeric tau for a 52-week tilavonemab study at 4000 mg Q4 W. The level of target engagement decreases substantially from the CSF to the synaptic cleft. (b) Dynamics of neuronal oligomeric tau in a 52-week study at 2000 and 4000 mg. Despite good target engagement on free monomeric CSF tau, there was almost no reduction in the uptake of synaptic oligomeric tau (0.01%).

Figure 5

Effect of semorinemab on free monomeric and oligomeric tau in different compartments in AD. Simulated outcomes of free tau dynamics in different compartments in the 8100 mg semorinemab dose Q4 W study for 18 months (a) and intraneuronal oligomeric tau dynamics over 73 weeks (b). The target engagement of the free monomeric tau is substantial in CSF and ISF and less so in the synaptic cleft; however, the drug can only reduce neuronal uptake of oligomeric tau by 0.3%.

Figure 6

Effect of cinpanemab on free monomeric and oligomeric aSyn in different compartments. Dynamics of synaptic free monomeric and PFF aSyn in different compartments during treatment for 52 weeks (Q4 W) in PD with the highest dose of 3500 mg cinpanemab (a). Due to the high selectivity of the antibody for PFF aSyn, the reduction is much more pronounced than for monomeric aSyn. (b) Dynamics of neuronal PFF a-Syn uptake with and without cinpanemab at different doses (in mg/kg) during the 52-week study, showing a substantial dose-dependent reduction in neuronal uptake.

Figure 7

Effect of prazineuzumab on free monomeric and oligomeric aSyn in different compartments. (a) Dynamics of monomeric and oligomeric aSyn in different compartments during a 52-week study with 4500 mg Q4 W prazineuzumab in PD. (b) Neuronal uptake dynamics for a 52-week study with and without different doses (1500 and 4500 mg) Q4 W of prazineuzumab, showing considerable reduction of PFF aSyn uptake.