Artificial Intelligence (AI) is already having a transformative impact in healthcare, including on breast cancer detection and diagnosis, risk assessment, and prognosis. And there is much excitement building around AI’s potential role in drug discovery and clinical trials.
In the final installment of BCRF’s series on AI in breast cancer, we delve into how it’s reshaping drug discovery, patient matching to clinical trials, and data gathering and analysis (big data mining).
Read part one on AI’s potential in breast cancer research, part two on breast cancer screening and diagnosis, and part three on precision medicine and predictive analytics.
Traditional drug development is a slow process. Case in point: It took 14 years from the time the HER2 protein was discovered to the moment the FDA approved the lifesaving targeted therapy Herceptin for patients. That’s because there are several steps involved: identification of a target protein that’s specific for the disease, screening for drugs that bind the protein to identify the best candidates, testing and optimization of lead candidate(s) in laboratory cells and models, and clinical trial testing in humans. All of this takes tremendous amounts of time, manpower, and funding.
Central to drug discovery and development is the fact that researchers need to determine the exact 3D structure of the protein target so that drugs can be designed to fit into a specific protein pocket. But predicting a 3D structure of a protein is not an easy task, and in fact, researchers have dubbed this the “protein-folding problem.” How does a string of amino acids that make up the building blocks of a protein also form the complex 3D structure of a functional protein? It is analogous to the letters of the alphabet creating words, sentences, and a coherent novel—all by themselves. The complexity of the task makes traditional experimental methods complicated and time-consuming, taking as much as several years to solve a single protein’s structure.
Enter AlphaFold, an AI deep learning system developed by DeepMind, a subsidiary of Google. AlphaFold was trained on mountains of data from prior experiments that successfully defined protein structures and uses this information to help determine the 3D structures of yet-unsolved proteins.
In 2022, AlphaFold deciphered the 3D structures of 200 million known proteins in just one year. Imagine how this technology would have accelerated the development of HER2-targeted drugs and, moving forward, the avenues it will open for future breast cancer drugs.
DeepMind made this data open source, so it’s freely available to the research community and pharmaceutical companies at large. In 2023, DeepMind’s was awarded the American Nobel Prize, the Lasker Award, for their solution to the protein-folding problem.
“Thanks to AI, new medical breakthroughs may one day move from years to mere months,” Demis Hassabis, AlphaFold’s founder, said in a recent TED Talk.
The first application of AlphaFold in oncology sought to unravel the structure of CDK20, a newly identified protein target that is important in liver cancer. Remarkably, in only 30 days, scientists from AlphaFold and Insilico Medicine, an AI drug discovery company, collaborated to decipher the CDK20’s protein structure. Using a fully automated AI machine learning program, they generated approximately 9,000 molecules that could target CDK20. From these, seven were selected for synthesis and biological testing with one lead CDK20 small molecule inhibitor found to be active in liver cancer laboratory models.
This is one example of how scientists are using AI to fuel drug discovery. The ultimate test will be if the combination of AlphaFold and the pharmaceutical industry’s AI-intensive strategies lead to successful FDA approvals in oncology. If this type of work continues, we can shave years off the early steps in the process and speed up drug development in the future.
Only time will tell if AI advances lead to improvements in patient outcomes, but the future of AI-based drug development is looking bright. Knowing the potential importance of AI to accelerate drug development, BCRF is committed to funding novel AI research in breast cancer and, moving forward, plans to fund more innovative AI research.
Patient screening is a key component of clinical research that determines if people are eligible to participate in specific trials. That process is labor intensive, time consuming, and subject to human errors. A clinical research team must scour patient medical records and laboratory results to determine who fits a given trial’s long inclusion and exclusion criteria. AI solutions that can sift through large amounts of information like this and determine who is eligible for specific clinical trials would be invaluable.
In fact, these practices have already begun to take shape in oncology. A team of researchers compared AI to the standard method of clinical trial screening by reviewing data from three oncology clinical trials. They found that AI-assisted screening led to a 24 to 50 percent increase in the number of patients that were correctly identified as potential participants compared to standard practice. Further, no patients correctly identified by the standard practice were missed by AI. Most strikingly, compared to standard practices that can take many days to screen patients, AI took only minutes. The ability to shorten research timelines will ultimately benefit more patients.
AI is also leveraged in big data mining, the process of sifting through mounds of information to find meaningful nuggets to achieve a given goal. In oncology, big data mining allows researchers to efficiently assess changes in the genome that could be associated with disease. Building on human genome sequencing, genomic profiling blossomed as an important area in cancer biology. But the amount of information generated can be daunting to analyze, so bioinformatic scientists have developed ways to use AI-mediated big data mining to analyze mountains of data for key patterns or trends.
For example, diagnostic companies like Caris and Tempus are using state-of-the-art AI algorithms to examine patient molecular data for biomarkers that could predict a tumor’s response or chance of developing resistance. The results could inform clinical trials to validate biomarkers or inform how cancer care teams make clinical decisions and tailor personalized treatments.
As we continue to harness AI in drug discovery and clinical trials, it’s paramount to consider what this means for patients. Testing, transparent protocols, and ongoing oversight to prevent any unintended consequences are needed, particularly for underserved communities.
Nevertheless, the integration of AI into oncology will accelerate progress and improve outcomes for patients with breast cancer. Indeed, AI is poised to revolutionize the field as it may be leveraged to speed up the drug development process, increase the availability of innovative and more effective treatments, and enable greater participation in clinical trials to test them—thereby personalizing care for patients worldwide. BCRF will remain at the forefront of AI advances by funding innovative projects that can unlock its potential and help move the needle in the fight against breast cancer.
Maadi, H., Soheilifar, M. H., Choi, W., Moshtaghian, A., & Wang, Z. (2021). Trastuzumab mechanism of action; 20 years of research to unravel a dilemma. Cancers, 13(14), 3540. https://doi.org/10.3390/cancers13143540
Ren, F., Ding, X., Zheng, M., Korzinkin, M., Cai, X., Zhu, W., Mantsyzov, A., Aliper, A., Aladinskiy, V., Cao, Z., Kong, S., Long, X., Liu, B. H. M., Liu, Y., Naumov, V., Shneyderman, A., Ozerov, I. V., Wang, J., Pun, F. W., . . . Zhavoronkov, A. (2023). AlphaFold accelerates artificial intelligence powered drug discovery: efficient discovery of a novel CDK20 small molecule inhibitor. Chemical Science, 14(6), 1443–1452. https://doi.org/10.1039/d2sc05709c
TED. (2024, April 29). How AI is unlocking the secrets of nature and the universe | Demis Hassabis | TED [Video]. YouTube. https://www.youtube.com/watch?v=0_M_syPuFos
AI and Cancer. (2024, May 30). Cancer.gov. https://www.cancer.gov/research/infrastructure/artificial-intelligence
Calaprice-Whitty, D., Galil, K., Salloum, W., Zariv, A., & Jimenez, B. (2020). Improving Clinical Trial Participant Prescreening With Artificial Intelligence (AI): A Comparison of the Results of AI-Assisted vs Standard Methods in 3 Oncology Trials. Therapeutic Innovation & Regulatory Science, 54(1), 69–74. https://doi.org/10.1007/s43441-019-00030-4
Chua, I. S., Gaziel-Yablowitz, M., Korach, Z. T., Kehl, K. L., Levitan, N. A., Arriaga, Y. E., Jackson, G. P., Bates, D. W., & Hassett, M. (2021). Artificial intelligence in oncology: Path to implementation. Cancer Medicine, 10(12), 4138–4149. https://doi.org/10.1002/cam4.3935
Please remember BCRF in your will planning. Learn More
