On Wednesday 24th of September Dr Nat Hastings presented at the Charles River Labs Science Day at Chesterford Research Park on a topic that is relevant to Cellestial’s drug development activities and which stirred many discussions – there is no such things as a model of Parkinson’s disease.
What does that mean, when many so-called models are described in the publications and are freely commercially available?
First, we need to examine what a disease model should be. It is a system that should have predictive value in terms of the human condition and be translatable to clinical findings, while being cheaper, quicker, and more humane than testing straight in patients.
To create a system like this that has value we should have confidence that the model replicates the correct type of pathology in the correct type of cells. Historically, Parkinson’s has been viewed through the lens of dopaminergic denervation which leads to characteristic motor symptoms. However, recent research calls the completeness of this dogma into question as multiple brain regions and many cell types have now been shown to be affected in this condition.
A few commonly used types of in vitro and in vivo models of Parkinson’s include:
- Dopaminergic toxin models – original way of mimicking Parkinson’s symptoms
- + Cheap and quick, replicate motor features in vivo
- – May not recapitulate the actual causes of Parkinson’s; may not have the relevant effect on other cell types like astrocytes
- When to use: if testing dopamine-replacement strategies only, including dopaminergic replacement drug formulations and dopamine cell transplantation
- Alpha-synuclein-based models
- + Can mimic alpha-synuclein spread that is reminiscent of Braak staging of Parkinson’s
- – Lengthy; expensive; predictive value of clinical translatability in idiopathic Parkinson’s remains to be fully demonstrated; may be most relevant to early stages of Parkinson’s
- When to use: if testing disease-modifying treatments where effects on alpha-synuclein aggregation would indicate the engagement of the correct biology
- Genetic models (SNCA, DJ-1, PINK1, LRRK2, GBA1)
- + Have high relevance to people with Parkinson’s who have specific mutations; easier stratification in further clinical trials
- – Relevance to idiopathic Parkinson’s (largest unmet clinical need, >85% of people with Parkinson’s) remains to be demonstrated
- When to use: if targeting specific sub-populations with Parkinson’s-associated mutations
- Inflammatory models
- + Links between neuroinflammation and Parkinson’s are increasingly acknowledged; effects on multiple brain regions can be studied
- – Quite non-specific; may be most relevant to early stages of Parkinson’s
- When to use: if testing disease-modifying treatments with anti-inflammatory properties
None of these alone has proven to have a predictive value in a complex condition like Parkinson’s. So, what do we do?
- Experimentally ensure that the chosen model has relevance to the pathology and cell type that the treatment targets (e.g. MPP+, a toxic product of MPTP, is taken up by dopaminergic neurones but extruded by astrocytes – is it a relevant system to study astrocytic pathology? Is the promoter for overexpressing a gene favouring the cell type of interest? Does the biology of interest exist in the model organism, e.g. connexin 43 in astrocytes in Drosophila flies?)
- Use more than one relevant model, if feasible
- Have patience. By definition, models are reductionist: Parkinson’s may take 60+ years to develop in humans. Replicate relevant results in slower-developing alpha-synuclein aggregation models
The day closed with a panel discussion chaired by Justin Bryans (CRL) with Damian Crowther (Trimtech Therapeutics), Marion Cubitt (Isogenica), Chris Hurley (CRL), and Nat Hastings (Cellestial Health) discussing the challenges and opportunities of different treatment modalities.
Also, a bet was made: that the first drug product to reach market without any form of in vivo testing would be a topical biologic. What do you think?