Although the treatment was not a cure, Rosie’s mobility and quality of life improved dramatically. Conyngham and the scientists involved are working on booster shots to treat her resistant tumors.
Image credits:Jake Willis, NSW
LLosing a beloved pet is hard for anyone to accept, but one Australian technology entrepreneur didn’t give up when her five-year-old rescue dog Rosie was diagnosed with terminal cancer. Turning to ChatGPT and AlphaFold, Paul Coningham worked with scientists to create personalized mRNA vaccines. “We took Rosie’s tumor, sequenced her DNA, converted it from tissue to data, and used that to search for problems in her DNA and develop treatments based on that,” Rosie’s owner Paul Conyngham said in an interview on The Today Show Australia. “ChatGPT helped us through the entire process.”
“What’s so fascinating about this is that [Conyngham] “He used ChatGPT to identify the sequence and generate the RNA vaccine, and then he was able to ultimately produce it and inject it very quickly,” said Steven Hsesheng Lin, a physician-scientist and radiation oncologist at MD Anderson Cancer Center who was not involved in the project. “So going from concept to translation into actual patients, in this case animal patients, is amazing.”
Machine learning steps after standard treatments fail
Rosie had a tennis ball-sized mast cell tumor (MCT) in her foot, which reduced her mobility. MCTs, the most common type of skin cancer in dogs, secrete histamine and can cause a variety of health problems.1 Rosie had already undergone surgery and chemotherapy, but to no avail, and veterinarians gave her only months to live.
Conyngham, a data analyst, has experience in machine learning but no biology background. We asked ChatGPT to help us find a cure. The first step was to sequence the DNA of Rosie’s tumor, for which Mr Conyngham paid thousands of Australian dollars out of his own pocket. He collaborated with computational biologist Martin Smith from the University of New South Wales (UNSW) Ramaciotti Genomics Center, who was initially skeptical of the unusual request due to the computational burden of tackling genome sequencing. Conyngham assured that there were no problems in analyzing the data, using ChatGPT to identify neoantigens present on tumors and Google DeepMind’s AlphaFold to predict protein structures.
After selecting which parts of the protein could be used to produce a personalized mRNA vaccine, he consulted Paul Thorderson, an expert in biomimetic chemistry at the UNSW RNA Institute, and asked him to produce Rosie’s neoantigen mRNA from the DNA template. “That’s a really special part of this story, that part was done by. [Paul]”Somebody didn’t have a background in biology, medicine or chemistry,” Thorderson said. [from] Martin worked with us on genome sequencing work, from which we used AI to identify neoantigens. ”
Thorderson’s team had produced hundreds of different mRNA molecules for therapeutic purposes, but had never worked on cancer vaccines, so they jumped at the opportunity to collaborate. Although he thought it was a great idea, he said he was skeptical and thought by the time a vaccine was produced it would be too late for Rosie. However, with Conyngham’s determination and leadership, they were able to quickly turn things around. “There was a good level of communication between all of us, Martin, Paul and myself, for several months as he finalized the design,” Thorderson explained. “And as soon as we had the design, we plugged it into our workflow… amplify it, create mRNA from it, and convert that mRNA into [lipid nanoparticles] And we are ready to mix. ”
Within two months, Mr Thorderson and his colleagues had produced the vaccine and Mr Conyngham worked on a lengthy ethics application to have Rosie’s veterinarian at the University of Queensland administer the vaccine. Conyngham said the change has been rapid and surprising. In less than a month, Rosie went from being immobile to jumping over fences and chasing rabbits.
Combination therapy shrinks tumors but is not a cure
The personalized mRNA vaccines were given in combination with immune checkpoint inhibitors, making it difficult to determine which treatment had the greatest effect, Thorderson said. Although the vaccine significantly shrunk Rosie’s largest tumor, it was not a cure. “I think Rosie’s lifespan and healthspan have been significantly extended,” Conyngham said. He and his team are currently working to develop a booster shot that will address the resistance mutations in Rosie’s remaining tumor. Because the goal was to find a personalized treatment for Rosie, rather than a marketable product, they did not take into account whether the neoantigens they were targeting were unique or common to dogs with MCT, Thorderson said.
Mr Lin said the key question was whether the degree of personalization was necessary in Rosie’s case. In recent research Published in Nature magazineLin and his collaborators demonstrated that a SARS-CoV-2 mRNA vaccine can sensitize tumors to immune checkpoint inhibitor therapy.2 “In our preclinical model, just by actually delivering these non-specific RNAs; [tumor responses]’ explained Lin.[But] I think it would be very interesting to do the following, since a personalized approach might actually get a better response. [bigger] study. ”
For those wondering whether this approach can also be applied to humans, Professor Lin pointed out that phase III clinical trials are currently investigating the use of personalized mRNA vaccines as neoadjuvant therapy after treatment with immune checkpoint inhibitors in human lung cancer patients.3 “What’s holding us back is the speed at which we can actually produce these vaccines,” he added.
Adam Gripin, a physician-scientist at MD Anderson Cancer Center and the University of Florida and co-author of the study with Lin, is working to develop personalized mRNA vaccines for human brain tumor patients. Gripin said Rosie’s case was a great example of the power of mRNA vaccines and artificial intelligence in treating both animals and humans, and he was enthusiastic about applying this approach to human patients. “The most attractive use of machine learning is how to significantly reduce the time it takes to predict these antigens,” Gripin commented. “…This is a very exciting time to be in this field, and we are very optimistic about the opportunity to develop highly effective personalized treatments for patients.”
Mr Thorderson, director of the UNSW mRNA Institute, said he strongly believed that mRNA technology should be accessible in an equitable way and that the decentralized workflow used in Rosie’s case was an example of the power of doing things differently. “Our technology has the means to make products cheaper at the manufacturing stage,” he said. “…Ultimately, if we want to make personalized medicine more accessible, we need to rethink both our reimbursement models and our regulatory environment.”
