When 57-year-old Margaret Chen was diagnosed with liver cancer, her doctors faced a difficult question: which treatment would give her the best chance? Now, researchers at Johns Hopkins University in Baltimore may have found a way to answer that question before treatment even begins — using computer simulations of a patient's tumor.
Scientists at the Johns Hopkins Kimmel Cancer Center have developed virtual tumors that can predict which liver cancer patients will respond to combination immunotherapy. The work, published in the Proceedings of the National Academy of Sciences on July 14, could eventually help doctors choose the right treatment faster.
The tool combines two types of computer models. One uses math equations to simulate how a patient's whole body might respond to treatment. The other tracks how individual cells behave and where they sit within a tumor. Together, they create a digital twin of the tumor that researchers can test different drug combinations on, without risking any harm to real patients.
"Many cancers have a very fast progression time, and doctors may not necessarily have time to try surgery or different treatments," said Dr. Atul Deshpande, senior study author and assistant professor of oncology at Johns Hopkins. "Our idea was to create a computational model where we could simulate trying different doses or combinations of cancer therapies, and it could help guide physicians toward the best options for patients."
The researchers tested their virtual tumors using a combination of two drugs: nivolumab, an immunotherapy that helps the immune system attack cancer, and cabozantinib, a targeted therapy that blocks signals tumors need to grow. The predicted response rates matched closely with real clinical trial results — a sign that the virtual patients behave like real ones.
But the team discovered something unexpected. In patients who did not respond to treatment, their simulations revealed a physical wall of cells called fibroblasts — tissue cells that had somehow reorganized themselves to block immune cells from reaching the tumor.
"Even if immune cells were located near the tumor, the fibroblast would block the immune cells from reaching the tumor," Deshpande explained. The remarkable finding? This protective barrier is visible in scans before treatment even starts.
Another major advantage of the computational model is speed. From just 15 patients in an early-phase study, the researchers generated a virtual population large enough to predict outcomes for a massive Phase III trial — something that would normally take years and thousands of patients.
The tool still needs further testing before doctors could use it in clinics. But Deshpande hopes that one day, when multiple treatments exist for a cancer, this technology could help determine which ones are worth trying and which to avoid for each individual patient.