AI Tool Predicts Cancer Therapy Outcomes
Cancer immunotherapy drugs can produce dramatic recoveries, but they only work for some patients. To identify who will benefit, associate professor of biomedical informatics Marinka Zitnik and colleagues have developed an artificial intelligence system that analyzes thousands of genes in a patient’s tumor. Trained on data representing more than 10,000 tumors and 33 types of cancer from prior patients, the model outperformed the best prediction methods across multiple cancers and immunotherapy drugs, and made successful predictions in cancers and treatments it had not previously encountered. The AI model also identified the immune processes inside each tumor that may be helping or blocking treatment, giving researchers clues for overcoming resistance. The tool could someday help doctors match patients to the right immunotherapy, deploy more effective clinical trial enrollment, and identify new drug targets.
Treating Disease with Captive Bacteria
Bacterial infections can cause major problems in healthcare settings. Pinkas family professor of bioengineering David Mooney has found a way to attack those infections at their source with another type of bacteria, engineered to deliver therapeutic drugs. Mooney and his colleagues have developed Implantable Living Materials, or ILMs: helpful bacteria encapsulated in a tough, stiff hydrogel that could be surgically implanted in a patient.
In their demonstration—a simulation of an infected stainless steel prosthetic joint in a mouse— researchers programmed the genetically-engineered bacteria to detect Pseudomonas aeruginosa, a major cause of hospital-acquired infections, and to respond by releasing antimicrobials. The ILM they implanted in the mouse contained the helpful bacteria for up to six months, resisted the physical stresses that break down other types of hydrogels, and significantly reduced the severity of infection. Microbial therapies using this technique have the potential to treat cancer, metabolic disease, and other illnesses that respond to localized drug delivery.
Understanding Why Mono Leads to MS
Five years ago, Kjetil Bjornevik and his colleagues at the Harvard Chan School of Public Health found that infection with Epstein-Barr Virus (EBV)—the virus that causes mononucleosis—greatly increased a patient’s risk of developing multiple sclerosis (MS). Now Bjornevik, an assistant professor of epidemiology and nutrition, and Natalia Drosu, a research fellow in neurology at Harvard Medical School, have uncovered a clearer picture of how that process works. People with untreated MS generate unusually high numbers of an immune cell that uniquely responds to EBV virus particles. The researchers speculate that this abnormal immune reaction to EBV infection may help drive MS. Although more research is needed, the work “provides a framework for developing EBV-targeted therapies, including vaccines and antivirals,” according to the study authors.
