U.K. researchers have repurposed algorithms that astrophysicists employ to discover black holes to create a digital twin model that can accurately predict treatment responses for different chemotherapy drug classes in people with cancer.
The technology, called FarrSight®-Twin, could “help health professionals to match individuals to the right chemotherapy drug,” said Uzma Asghar, PhD, co-founder and chief scientific officer at Concr and a consultant medical oncologist, currently working at The Royal Marsden NHS Foundation Trust, London. “This increases the probability of response to the drug and, as we have shown, improve survival in a retrospective analysis.”
Speaking at the 36th EORTC-NCI-AACR Symposium on Molecular Targets and Cancer Therapeutics in Barcelona, Spain, Asghar explained that a digital twin is a virtual replicate of an individual and their cancer. It is created from biological data from thousands of patients with cancer who have been treated in different ways and considers information about the patient, their demographic profile, the type of cancer that they have, including molecular data on their tumor, and any potential drug they may be exposed to.
“The great thing about using this technology is that the more information you feed into the digital twin about that individual, the better the predictions become, but it also allows you to follow the patient’s cancer journey,” she said.
Asghar told Inside Precision Medicine that the patients she saw in her clinic were the driving force behind her decision co-found Concr and develop the FarrSight-Twin technology.
“I see far too many people not benefiting from chemotherapy in clinic and I wanted a way that we could get better insight into who will and will not respond to chemotherapy and which drug had a higher chance of working,” she said. “At the moment there are not very strong biomarkers that we can use to predict response to chemotherapy.”
So far, the digital twin has been trained on over 10,000 cancer patients using more than 20 different solid cancer types. The data has come from publicly available datasets such as The Cancer Genome Atlas as well as clinical trial data from research partners. In addition, Asghar and team are now collecting real-world data to “convince people that this technology does what it says it does.”
In the current study, Asghar and colleagues used FarrSight-Twin to recreate published phase II or III clinical trials of patients with either breast, pancreatic, or ovarian cancer with a digital twin representing each real patient who took part in the trial. Overall, the digital trials accurately predicted the outcome of the actual clinical trials in all simulated clinical studies.
Further testing showed that when patients received the treatment predicted by FarrSight-Twin to be best, they had a 75.0% response rate. This fell to 53.5% when patients received a suboptimal treatment.
All the trials compared two different drug therapies, including anthracyclines, taxanes, platinum-based drugs, capecitabine, and hormone treatments.
Asghar said: “We are excited to apply this type of technology by simulating clinical trials across different tumor types to predict patients’ response to different chemotherapies and the results are encouraging.”
“This technology means that researchers can simulate patient trials at a much earlier stage in drug development and they can re-run the simulation multiple times to test out different scenarios and maximize the likelihood of success. It is already being used to simulate patients to act as controls for comparing the effect of a new treatment with the existing standard of care.”
“We are currently developing this technology so that it can predict treatment response for individual patients in the clinic and help doctors understand which chemotherapy will or will not be helpful, and this work is ongoing.”
Although the technology is still “a few years from validation and approval for clinical use,” Asghar hopes that in future, “it will be routinely used to delivering precision medicine in the clinic or at the bedside to by predicting and individual’s response to cancer treatment.”
Indeed, once the patient information has been uploaded, FarrSight-Twin can create predictions within minutes, making it feasible for point-of-care use.
Asghar and colleagues also anticipate that pharma and biotech companies will use FarrSight-Twin to support development of their novel compounds from the laboratory to phase I trials, and at the concept point of developing clinical studies.
They are currently testing the technology to see if it could help predict which available treatments will work best for patients with triple-negative breast cancer in the VISION study, an observational collaborative trial between Concr, The Institute of Cancer Research, Durham University, and the Royal Marsden Hospital.
