3 Questions: Building predictive models to predict tumor growth | MIT News

Just as Darwin’s finches evolved in response to natural selection for endurance, the cells that make up a cancerous tumor similarly resist selective pressures to survive, mutate and spread. In fact, tumors are complex collections of cells with their own unique structure and capacity to change.

Today, Artificial Intelligence and machine learning tools offer an unparalleled opportunity to illuminate the generalizable rules that govern tumor progression at the genetic, epigenetic, physiological, and microenvironmental levels.

Matthew G. Jonesassistant professor at MIT Department of Biologyi Koch Institute for Integrative Cancer Researchonce Institute of Medical Engineering and Sciencehopes to use computational methods to create predictive models – playing a chess game with cancer, making sense of a tumor’s ability to mutate and resist treatment with the ultimate goal of improving patient outcomes. In this interview, he describes his current work.

Question: What part of the tumor progression are you working to assess and characterize?

A: A very common issue with cancer is that patients will respond to treatment at first, and then eventually that treatment stops working. The reason this happens so much is that tumors have an amazing, and very challenging, ability to evolve: the ability to change their genetic makeup, protein expression structure, and cellular dynamics. The tumor as a system also changes at the structural level. Most of the time, the reason why a patient succumbs to the tumor is because the tumor may have changed into a condition that is out of control, or it appears in an unexpected way.

In many ways, Cancers can be thought of as, on the one hand, incredibly chaotic and disorganized, and on the other hand, as having their own inner, ever-changing mind. The central thesis of my lab is that tumors follow regular patterns in space and time, and we hope to use statistical and experimental techniques to determine the molecular mechanisms underlying these changes.

We focus on one particular way tumors arise through a type of DNA amplification called extrachromosomal DNA. Excluded from the chromosome, these ecDNAs circulate and exist as their own separate set of DNA particles in the nucleus.

Originally discovered in the 1960s, ecDNA was thought to be a rare phenomenon in cancer. However, as researchers began using next-generation sequencing on large groups of patients in the 2010s, it became apparent that these ecDNA enhancers gave tumors the ability to adapt to stressors, and to heal, quickly, but that they were more prevalent than originally thought.

We now know that these ecDNA amplifications are seen in about 25 percent of cancers, in the most aggressive cancers: brain, lung, and ovarian cancer. We discovered that, for a variety of reasons, ecDNA amplifications are able to change the rule book in which tumors evolve in ways that allow them to accelerate the most invasive disease in surprising ways.

Question: How do you use machine learning and artificial intelligence to learn ecDNA amplification and tumor evolution?

A: There is translational work I do in the lab to improve the lives of patients. I want to start with patient data to find out how different evolutionary pressures are driving the diseases and changes we see.

One of the tools we use to study tumor evolution is single cell lineage tracing technology. Broadly, they allow us to study individual cell lines. If we sample a particular cell, we not only know what that cell looks like, but we can (ideally) pinpoint when the dominant mutation appeared in the history of the tumor. That evolutionary history gives us a way to study these dynamic processes that we wouldn’t be able to see in real time, and helps us make sense of how we can stop that evolution.

I hope to get better at classifying patients who will respond to certain drugs, to anticipate and overcome drug resistance, and to identify new therapeutic targets.

Question: What excited you about joining the MIT community?

A: One of the things I was most attracted to was the combination of excellence in both engineering and biology. At the Koch Institute, everything is designed to develop this link between engineers and basic scientists, and beyond the campus, we can connect with all biomedical research businesses in the greater Boston area.

Another thing that drew me to MIT is its strong emphasis on education, training, and investing in student success. I am a personal believer that what separates academic research from industrial research is that academic research is a service profession, because we are training the next generation of scientists.

It has always been my mission to bring excellence to both the technical disciplines of integration and testing. The types of interns I hope to take on are those who are willing to collaborate and solve big problems that require both disciplines. KI [Koch Institute] it is specially prepared for this kind of hybrid lab: my dry lab is next to my wet lab, and it is a source of collaboration and communication, and that reflects the general idea of ​​KI.