If you have ever wondered how drugs are selected for clinical trials and assumed, as often described, that they must be selected after lengthy preclinical studies in animal models, here is an article that will disappoint you.
50 scientists in Scotland conducted a two-stage systematic review to select the first two drugs for evaluation in Motor Neuron Disease-Systematic Multi-arm Adaptive Randomised Trial (MND-SMART: NCT04302870).
Obviously this could only lead to drugs that were already tested either in preclinical studies or in drugs trials.
Indeed the repurposing of drugs reduces costs and barriers to clinical development because they have been assumed by someone else, this is for example the strategy used by Amylyx for AMX0035.
First, the authors reviewed clinical studies in Motor neuron disease, Alzheimer's disease, Huntington's disease, Parkinson's disease and multiple sclerosis, identifying drugs described in at least one Motor neuron disease publication or publications in two or more other diseases.
The authors scored and ranked drugs using a metric evaluating safety, efficacy, study size and study quality. In stage two, the authors reviewed efficacy of drugs in Motor neuron disease animal models, multicellular eukaryotic models and human induced pluripotent stem cell studies.
An expert panel reviewed candidate drugs over two shortlisting rounds and a final selection round, considering the systematic review findings, late breaking evidence, mechanistic plausibility, safety, tolerability and feasibility of evaluation in Motor neuron disease-SMART.
Curiously those experts eliminated Sodium phenylbutyrate which is one component of AMX0035.
We learn also they rightly eliminated drugs that were already trialed four times or more. As the saying tells: "Errare humanum est, perseverare autem diabolicum."
The seven candidate drugs remaining were memantine, acetyl-l-carnitine, simvastatin, ciclosporin, melatonin, fluoxetine and N-acetyl cysteine.
For future drug selection, the authors will incorporate automation tools, text-mining and machine learning techniques to the systematic reviews and consider data generated from other domains, including high-throughput phenotypic screening of human iPSCs.
This statement is in itself a bit bizarre, there are many AI tools for discovering drugs, some are open source, some are commercial, but in most cases they are of good to excellent quality. Drug selection with a literature review sounds like a process from the dark ages.