Please can you outline the basic principles of game theory?
Game theory has been used over the last several decades to explain how people or things interact.
The classic description of game theory was described by the prisoner's dilemma, which is a situation in which two players have two options where the outcome depends on the simultaneous choice made by the other.
It's often thought about in terms of two prisoners separately deciding whether to confess to a crime and what happens to them if one confesses or both of them confess.
Basically this idea of how people interact and make choices, where they're either altruistic or not, suggest rules by which we can think about all kinds of systems.
Game theory has been applied especially to economics, social science situations, political science, and now more and more to biology.
Why is game theory increasingly being applied to biology?
In classic teaching physics involves learning about atoms and forces. Chemistry is how these atoms and forces build up to form molecules. Then biology is how those molecules come together to form life and life systems.
What's missing more and more as we understand those basic ideas is how you describe how cells interact and how organisms interact. This opens up fields like ecology and evolution; however, these really don’t describe how they interact in a truly dynamic fashion.
So game theory is being applied more and more to try and understand the interactions, the networks, and how organisms or cells interact with each other.
Essentially we find explanations lacking when we simply look at physics, chemistry, and biology in their current forms.
What sparked your interest in game theory and the potential application to understand more about cancer cells?
It actually started back many years ago when I was working with a guy by the name of Bob Axelrod at the University of Michigan. He is one of the true founders of game theory, especially a form of game theory, called cooperation theory.
Bob’s a political scientist, but in the late 70s he described cooperation theory as something that should be looked at in biology.
Cooperation theory is really game theory and describes how two organisms can cooperate with each other to help each other grow and prosper.
We came up with a theory back about seven years ago trying to understand how cancer cells potentially cooperate with each other to actually facilitate tumorigenesis.
The idea is that every cancer cell in a mass does not need to have all the characteristics of cancer to be successful. If you have some cancer cells that have the characteristics to attract new blood vessels for example, they'll attract new blood vessels for even cancer cells that don't have that ability.
Then if another sort of cancer cell doesn't have that ability, but has the ability to spread, then those can spread even though they may not have the ability to grow new blood vessels.
That sort of flies in the face of biology dogma, so that got me interested in studying how cancer cells can potentially cooperate or interfere with each other.
We started developing game theory models based around that, the most recent of which is this idea of really looking at game theory in terms of tumor metabolism.
Please can you tell us about the cooperative-like and competitive-like states of different cells in tumors?
What we find is that if you have two cancer cells sitting next to each other, they can interact in basically one of four ways.
They can each be autonomous and not contribute anything to each other in any way, or one cancer cell can basically secrete a hormone or a cytokine that's going to help the other cancer cell grow without getting anything in return. In that case, that cell is facilitating the growth of the other. They have a facilitating relationship.
The third is that each of them gives each other something that helps each grow better, and that's called mutualism. That is true cooperation.
The fourth aspect is one can basically steal a resource from the other so it's stronger and it parasitizes that resource from the other, causing the other one to not grow, while it grows. That's parasitism.
So you have those four basic states that can exist. If you think about it, that makes sense when you have millions of tumor cells all trying to grow, many with different characteristics, that some of them are going to facilitate one, some of them are going to be mutually helping, some are going to be parasitizing, and some are just going to be doing their own thing.
What we've been really interested in is what happens as a tumor grows and you create a malignant niche, basically in our terms, what we like to call the cancer swamp. This environment of hypoxic that's low oxygen, low pH, low nutrient, how do cells survive in that swamp and actually flourish there?
Knowing that that happens in almost every growing cancer, what does that malignant niche really do for the cancer? Does it increase biological diversity? That's where we really started to try and understand how hypoxic and oxygenated tumor cells, as well as those that have glucose nearby and those that don't, how they start to interact in the setting of a cancer swamp.
Can you please outline how you used mathematical and computer tools to try to gain an understanding of the biological interactions between tumor cells?
It's really not so much raw computing power in this case. It's not like we're cranking big data sets or big data. This is not really a bioinformatics approach. This is a mathematical question. Can we use game theory type math?
Can we use cooperation type math that's been developed in other systems like economics, like cooperation theory in social sciences and evolution, to really try and understand and apply that to cancer?
If it's been done somewhere else, if we can draw a parallel potentially to a cancer system, we don't need to rediscover the wheel. We should just try and apply those systems to cancer or biology in general.
I try to practice what I call collision science, which is I try to bring together diverse disciplines around a problem so that we don't keep having to rediscover the wheel all the time. If somebody has done it somewhere else, I should just try and apply that to cancer rather than rediscover everything.
So really what we're doing here is simply using the math of other systems, in this case game theory, to see if we can model how cancer cells potentially get their energy and oxygen in a hypoxic low energy, low nutrient environment.
What were your main findings?
Ardeshir Kianercy, my post-doctoral fellow who did his thesis work in game theory, looked at how tumor states might develop over time and found that as a tumor changes its environment, as it develops from a growing mass, before it outstrips its blood supply, all the cancer cells are existing in a state where they're getting oxygen and getting nutrients.
But as the cancer cells grow and outstrip their blood supply you develop this low oxygen, low nutrient state, where some of the cells closer to the blood vessel remain oxygenated and remain using glucose as their main source of energy. But as the cells get further away from that they actually switch to a state, and we model this state, where they're starting to use more anaerobic glycolysis and lactate as their energy source.
We demonstrated that the oxygenated cells could secrete lactate that the non-oxygenated cells could use for their energy source. What we hypothesized based on the math is that there's sort of a facilitation of the oxygenated cells to help the unoxygenated ones continue to grow and be robust, essentially a lactate shuttle.
As I was describing earlier, if you imagine two cancer cells sitting next to each other, one giving something to the other, facilitating the growth, that's what we found. That could help explain how cancer cells can continue to survive as a mass even when they don't have access to a good nutrient supply. There's a shuttle effect essentially: one cancer cell is acting as a facilitative species to the other.
What is the benefit to the oxygen-rich cell of doing this?
In this case it's not an example of mutualism where both are benefiting. This is really an example of facilitation, which exists in game theory, in many systems. There are many examples of facilitation, just not in cancer.
Did your experiments enable you to identify the ideal time to disrupt metastatic cancer cell cooperation?
What it did was give us a theoretical framework by which we can set up testing in the laboratory to test whether our theory was correct. So it helped us design our laboratory experiments.
What impact do you hope this research will have and what are your further research plans?
There are a couple things. I'm a guy who takes care of patients, so ideally everything we do ends up helping a patient somehow. Many folks would say this is so theoretical, how can it possibly help a patient?
My vision is, here we have a theoretical framework that's helping us explain how energy is transferred and nutrients are transferred around a tumor mass to facilitate its ability to grow. Our next step is to prove that that's true in experimental systems.
If it's true in experimental systems, then I can develop therapeutic strategies to block that from happening. Potentially, I can use a drug at different parts of the pathway to block this, and nobody's thought about using drugs in that way before. So, if that's the case, then I can take that all the way up to patients.
It's not a process that's fast. It's going to take me some years to do that.
Game theory is a new way of thinking about how cancers grow. Not only will it inform us for these metabolism experiments, it's informing us around how other cells interact with the cancer cells.
For example, what we have also done is said if you play the cooperation game in game theory and look at the other cells that are in the cancer system, you find that the mass of tumor cells, the thing we call a tumor, actually consists of not only cancer cells, but 30 other normal host cells: white blood cells; red blood cells; immune cells; fibroblasts; fat cells etc.
One of the most common cells that we found is something called a tumor-associated macrophage. We modelled and used game theory to describe how tumor-associated macrophages interact with cancer cells and found that they actually facilitate the growth of the cancer cells.
There is some mutualism that occurs, in that the macrophages give the cancer cells several cytokines and hormones that help them grow, and the cancer cells give other cytokines and hormones to the macrophages. So we've developed a whole therapeutic strategy of actually blocking the facilitative species, the tumor-associated macrophage, as a cancer therapy.
We're trying to develop, using the paper we just published in theory, as a way to look for new targets along the metabolism pathway. We've already used it to describe how to find other interactions between host cells and cancer cells and interrupt those. We're hoping that other people start to think like this and develop even more therapeutic strategies.
What’s the most important point to take away from this research?
I think that we really believe that we have to think about cancer in terms of it being a dynamic ecosystem. We have to get the world away from thinking that cancer cells are just a bunch of cells growing in isolation, and all we think about is what genes get turned on and off to make these cells grow.
They actually exist in a complex environment interacting with the host at the local and systemic levels. We will do much better by thinking about how that local community of cancer with its host, that local ecosystem, is interacting within the entire host patient.
Just like we, as humans, exist on the earth as the biosphere of all ecosystems, for cancer, the patient is that biosphere, and nothing happens in isolation. So what a cancer cell is doing in a prostate, a breast, or liver, is affecting the entire patient on that level. I think the more people understand that you have to think about cancer that way, the more we're going to make progress against it.
About Professor Pienta
Kenneth J. Pienta is the Donald S. Coffey Professor of Urology and serves as the Director of Research at the Brady Urological Institute of the Johns Hopkins University School of Medicine. He is also a Professor in the Departments of Oncology and Pharmacology and Molecular Science.
He serves as a faculty member in the Cellular and Molecular Medicine Program at the Johns Hopkins School of Medicine and is a Professor of Chemical and Biomolecular Engineering in the Whiting School of Engineering at Johns Hopkins University.
Dr. Pienta is a two-time American Cancer Society Clinical Research Professor Award recipient. He is the author of more than 375 peer-reviewed articles, and been the principal investigator on numerous local and national clinical trials. Throughout his career, Dr. Pienta has effectively mentored more than 45 students, residents, and fellows to successful careers in medicine.
Study shows AI can predict prognosis in triple-negative breast cancer