Machine learning tool helps match treatments to cancer patients

Can immunotherapy treatment help this cancer patient? And, if possible, what specific treatment to apply? Oncologists regularly ask themselves these questions. Insurers are also asking these questions because immunotherapy is expensive. Patients wonder if this new treatment could save their lives. Now, a new study by Professor Keren Yizhak, of the Ruth and Bruce Rappaport School of Medicine at the Technion – Israel Institute of Technology, uses artificial intelligence to create a simple and inexpensive way to answer this question for every patient, writes the research center in a press release.

Immunotherapy is a recent development in the world of cancer treatments. It provided complete remission to patients who could not be helped by other means, and it reduced many side effects of chemotherapy. There are multiple immunotherapeutic treatments. The principle by which they all work is stimulation of the patient’s immune system to attack cancer cells.

Simplification of the measurement of the burden of tumor mutations

How does the immune system distinguish between cancerous cells it needs to attack and healthy cells in the body? The more mutations the tumor has accumulated, the more it differs from “normal” cells, and thus immunotherapy may be more effective. This feature is called tumor mutation load (TMB). A higher BMR means more new mutations. Professor Yizhak’s method greatly simplifies the measurement of BMR.

Currently, to measure TMB, cells are taken from the tumor and their DNA is compared to the DNA of the patient’s healthy cells. Professor Yizhak and his group propose two important modifications to this process.

The first modification, already explored in a paper previously published by the group, is to compare RNA molecules rather than DNA molecules. This makes a difference because DNA molecules contain the entire human genome, while RNA molecules are small parts of the genetic code, copied to be used as instructions within the cell. In their previous study, the group showed that RNA molecules can also be used to identify cancer-specific mutations.

Dual innovation

The innovation in the group’s most recent paper is twofold: first, to eliminate the need to compare cancerous tumor RNA to healthy cell DNA. As a result, less genetic material needs to be sequenced, so patients can undergo fewer procedures. Instead of comparing tumor genetic material to healthy patient genetic material, Professor Yizhak’s team developed a machine learning algorithm that was trained to recognize healthy genome aberrations and distinguish them from the natural variation that occurs. exists between people. .

Second, using these predictions, they were able to calculate an RNA-based TMB metric. In fact, this method was found to be more effective than the standard method in estimating the predicted efficacy of immunotherapy for a given patient. This is thought to be the case because RNA contains the parts of the genome that are constantly in use and therefore can initiate an immune response. Mutations in parts of the genome that are not used are less likely to affect cell function.

The development of the algorithm was made possible by using a large existing database of sequenced RNA from cancer patients, on which the algorithm could be trained. In fact, Professor Yizhak’s lab is a “dry” computer lab. Computer labs use the vast amounts of clinical data collected by the scientific community around the world, using it to make new discoveries and develop new tools to help patients. The study was led by Dr. Rotem Katzir and B.Sc. student Noam Rudberg, both of the Henry and Marilyn Taub School of Computer Science.

Sherry J. Basler