All cancers have a genome that has become somewhat corrupted, whether through random mutations, viral infections, or both. This corruption fuels tumor growth and produces abnormal protein fragments known as neoantigens, which are unique to each patient. Through personalized treatments like cell therapies and cancer vaccines, doctors can train the immune system to target neoantigens on patients’ tumors.
Identifying which neoantigens will make the best immunotherapy targets, however, has been a difficult challenge for the field to overcome. Now, an international collaboration of scientists from academia, industry, and nonprofit organizations has created a more reliable way to predict ideal neoantigen targets, according to new findings published online today in the journal Cell.
Co-conceived by the Cancer Research Institute (CRI), the Parker Institute for Cancer Immunotherapy (PICI), and Sage Bionetworks, the Tumor Neoantigen Selection Alliance (TESLA), was launched in 2016 and brought together 36 teams of scientists, led by PICI principal data scientist Daniel Wells, Ph.D., and Nadine Defranoux, Ph.D. Each team was asked to apply its own neoantigen prediction algorithms to identical tumor samples and share their data with a centralized analysis platform. The analysis revealed five qualities, or keys, that could reliably improve our ability to determine a neoantigen’s usefulness in the context of immunotherapy.
Other senior authors who made crucial contributions to the work include Robert D. Schreiber, Ph.D., of the Washington University School of Medicine, and Ton N. Schumacher, Ph.D., of the Netherlands Cancer Institute, who serve as an associate director and a member, respectively, of the CRI Scientific Advisory Council.
Schreiber, in particular, has long been at the leading edge of the neoantigen field, and was the first to prove that checkpoint immunotherapy unleashes T cells against these abnormal targets.
“Until now, neoantigen prediction has been a black box,” Schreiber said in a press release announcing the study results. But TESLA’s new model enabled the scientists to accurately predict 75 percent of “good” neoantigen targets, while filtering out 99 percent of “bad” ones when tested in a separate cancer sample, by taking into account the neoantigen’s abundance, degree of abnormality, and other factors that ultimately dictate its ability to stimulate a successful immune response.
Consequently, this model could provide an important foundation moving forward when designing personalized immunotherapy approaches for cancer patients.
“Our aim is that data produced from TESLA becomes the reference standard when developing a new neoantigen-based treatment,” added Wells. By using this dataset, all of which is available freely to researchers, as a benchmark, he believes that, “the whole field would be able to collaborate and iterate on new methods much more quickly.”
For more on the potential of these exciting personalized immunotherapies, check out our story on Dr. Schreiber and his quest to cure cancer.