Course Navigation

What is Fragile X?

Fragile X Syndrome (FXS) is the most common inherited cause of learning disability, affecting around 1 in 4,000 males and 1 in 8,000 females. It can cause a wide range of learning difficulties as well as social, language, emotional and behavioral problems. 

 FXS is caused by a defect in a gene that makes a brain protein needed for development. The mutated gene is located on the X chromosome and is called the FMR1 gene.

 For more information you can visit the Fragile X Society website 

In this lesson we will cover: 

  • Early Fragile X research: the role of FRAXA 
  • Repurposing for FXS: the role of Healx
  • The benefit of AI technology in repurposing 

Early Fragile X research: the role of FRAXA 

FRAXA is a charity which was set up in 1994 by parents whose children were affected by Fragile X. Its primary goal is to address the lack of funding in biomedical research related to the condition.

FRAXA started by investing in basic research into Fragile X. They raised funds from the FRAXA community and gave grants on a competitive basis to researchers. Once the gene and protein responsible for Fragile X were identified, FRAXA recruited neuroscientists to complement the work of the geneticists. 

 The missing protein was shown to play roles in learning, memory and brain plasticity and quickly became a hot topic in neuroscience (and still is to this day). FRAXA hoped that by knowing the gene and understanding the protein, an obvious drug target would be found. However, clinical trials kept failing and efforts to find a drug were confounded when it became apparent the role of the gene is far more complicated than once thought.

 After 22 years of research, FRAXA had taken novel drug discovery routes as far as they could go. The disease was characterised at the molecular level, targets were validated and large scale international clinical trials were run with big pharma companies. None of this resulted in a drug that would make a meaningful difference to patients’ lives. FRAXA started to change its strategy. Instead of relying on academic labs they started using contract research organizations (CROs). But they were also open minded and prepared to try something new. In 2016, a friend of the CEO was in Cambridge and introduced him to Healx.

Repurposing for FXS: the role of Healx 

Ultimately, FRAXA went as far as a patient group could go towards developing a treatment for Fragile X. Despite 22 years of good progress, an effective drug had still not been found. FRAXA approached Healx to collaborate in late 2016. Its CEO, a clinician from America, was attracted by Healx’s fresh approach to drug discovery and was willing to try something new.

Healx’s ‘big data’ approach has proved to be very quick and cost effective when compared to traditional drug development. It takes less than half the time and 3% of the cost of traditional drug development to reach FDA approval. Different drug combinations can be tested together by a computer before testing in animal models. New pathways and targets, previously unknown to researchers, can also be identified.

Healx was initially attracted to FRAXA’s cause because it was a well-developed patient group that possessed a good model system in which drugs could be tested. The members were also open minded and willing to help Healx design clinical trials if a drug was found and offer insight into why past clinical trials were unsuccessful (as they had worked with the pharmaceutical companies who ran them). FRAXA also knew its community well and what really mattered to them.

Healx at first needed to know what symptoms the drugs were trying to treat. FRAXA was instrumental in this role. Fragile X can be a complicated disease with multiple symptoms. Healx wanted to make the algorithms prioritise finding drugs that treated the symptoms that mattered the most to the patients and families.  Research showed these to be:

  • Intellectual disabliilty 
  • ADHD, hyperactivy 
  • Anxiety, sensory issues 
  • Autistic behaviours 
  • Seizures

 Once the aims were identified Healx gathered as much biological data as it could for its algorithms to use. They searched through public data bases and found 19 potential sources. But only the data in 3 was deemed to be of high enough quality for use.

 Once the biological data was gathered…. 

  1. The algorithms looked at the data and assessed which drug’s expression profiles would counterbalance the aberrant gene’s expressions profiles for this specific diseas
  2. The data from all these different sources was collected into a knowledge graph (a representation of all the knowledge gathered and how it links together in a complex web). This acted as a map on which drug-gene-symptom-disease interactions could be traced on. Every possible drug was guided along it to see if it had the potential to help relieve one or more symptoms of the disease.

Once the computer had analyzed the data, a list of compounds that may relieve the symptoms of Fragile X was produced. Then a team within Healx, made up of pharmacologists with previous drug discovery experience, combed through the list. They eliminated compounds that would not work for biological reasons the algorithms had not been programmed to consider.

 From this, 8 drugs were shown to be promising. The computer showed 3 of them treated all symptoms FRAXA had define as very important to patients and their families. One of the drugs was a well-established anti-inflammatory drug used to treat arthritis. No one had previously thought about applying it to treat Fragile X.

 The 3 most promising compounds were tested on the mice model and the most effective drug is now in Phase 2a clinical trials.

The benefit of AI technology in repurposing 

These developments were a major success, largely because of a computers ability to compare combinations of drugs and see what their additive effect might be on the outcome of the disease.

 This is very useful for a complex disease such as Fragile X. For example, if there are 500 drugs for which gene expression profiles could be collected and the software could combine them in groups of 2, it would produce 250,000 different combinations. This is what Healx did with 4,280 different monotherapies, resulting in 9,157,069 potential therapies. The computers sorted through them and found 18 combinations which could be effective, 8 of them altered 4 major symptoms and all were tested in mouse models for safety and efficacy. 

The 3 best options are now being taken by Healx into clinical trials with close consultation with FRAXA.