Logo: University of Southern California

How Does Cancer Move?

Stealing a page from Google’s playbook: AME professor predicts where cancer will go next.
By: Adam Smith
October 23, 2012 —


METASTASIS IN PROGRESS: The human circulatory system in many ways resembles an elaborate highway system. In the artist's rendering above, we get a sense of the roughly 100,000 mile length of an adult's system of blood vessels. In this maze, Professor Paul Newton is forecasting the many detours of a ciculating tumor cell (CTC). Illustration by Yang Liu.


The woods are on fire, and a valiant group of firefighters are racing against time to put the fires out.
In the midst of this chaos, appears a “combustion scientist,” someone with no experience putting out fires. In the midst of the smoke and the flame, he decides to convince the firefighters that understanding the subtleties of combustion science will be of great help in the firestorm all around them. Predictably, the firefighters ignore him. They’ve got a job to do.

This is the scenario Paul Newton is walking into, and he knows it. The USC Viterbi professor of aerospace and mechanical engineering, a career applied mathematician, is deep in the woods of oncologists and hematologists — people at the front lines of fighting cancer — and he’s essentially offering them numbers, equations, and computer models.

But Newton’s models satisfy one fundamental question: how does cancer move in the body? Where did it originate? And where is it going next?

And Newton’s collaborators at the front lines have begun to realize what a powerful weapon these models are in the war on cancer.

Said Peter Kuhn, a molecular biologist with the Scripps Research Institute: “This is one of the first times that a mathematical model can potentially directly result in a change of treatment approaches. This is the very essence of bringing mathematics and physics to the challenges of cancer care.”

But first, a question: why does it matter how cancer moves in the body? The truth is, a primary cancer tumor typically isn’t fatal. Matters turn deadly, however, when that cancer metastasizes, when bits of that primary tumor flake off like murderous colonists, seeking other parts of the body. The veins and arteries of the circulatory system are like an elaborate highway system, the cancer riding the bloodstream.

The standard view of cancer is that it’s a unidirectional process. Let’s say the primary tumor site is in the lungs. It then proceeds to the regional lymph nodes or liver or brain. But Newton’s mathematical models proved what some oncologists had long suspected: that cancer doesn’t just move in a straight line, sometimes it doubles back and re-infects the primary tumor where it all started. Newton likens this to drawing compound interest at a bank, except, in this case, it’s not money that’s compounding in size, it’s the tumor.

The models showed that a certain property of the first metastatic site the cancer spreads to is surprisingly important and can dictate its future path: some parts of the human body were clearly “sponges” and others were “spreaders.” For example, the lymph nodes are sponges — if a rogue tumor cell takes up residence there, it’s not liable to re-circulate to other areas. However, the adrenal glands are among the body’s most notorious “spreaders” — they’re like the endocrine system’s version of the Connectors from Malcolm Gladwell’s “The Tipping Point.” `What makes it complicated’, says Newton, "is that the adrenal gland may act as a spreader for one kind of cancer, like lung cancer, but not for another, like breast cancer. So it’s actually a combination of where the cancer originated and where it goes to first that matters – what I would call the one-two punch."

To figure out how cancer moves, Newton turned to an unlikely source of inspiration: Google’s famous PageRank algorithm. The same logic that Google applies to ranking web pages, Newton is now applying to probable cancer sites within the human body.

Said Newton: “If you’re just randomly surfing the web, Google can track the probability of you going from one site to another just by looking at thousands and thousands of people who surf the web. They’ll say, ‘This guy is at the REI web-site; his probability of next jumping to the Lands End web-site is such and such because we have a collection of ten billion people who were on the REI web-site, and we know where they’re going to go next. They know how the REI web-site is connected to the Lands End web-site is connected to the Costco web-site and so forth.”

In the same way, Newton can tell an oncologist, hey, if there’s a primary tumor in the lungs, here’s the probability it will move to the liver. If it’s in the liver, here’s the probability of where it will go next. For the oncologist, these are invaluable forecasts.

Instead of a rich data set of user’s Google search histories, Newton turned to 3,827 cadavers from the years 1914-1943. They say that dead men tell no tales, but these all provided illuminating biographies on the life and times of cancer. All were specifically chosen because the time period — before radiation, before chemotherapy — provided a wealth of information on how lung cancer, left completely untreated, would make its way through the complex plumbing of the human circulatory system.

From this, Newton could assemble a timeline, a computer model of how cancer might flow from a primary tumor. It is his hope that such a baseline model could be tailored to an individual patient, and he and his collaborators at The Scripps Research Institute are working on just that. Imagine a cancer model based on millions of human beings. Treatment, whether through drugs, radiation or resection, could be more focused and personalized to each patient’s unique characteristics.

One collaborator, Dr. Jorge Nieva, chair of hematology and oncology at the Montana-based Billings Clinic, noted: “In essence, what Paul is creating is the beginning of a new model. You know, we doctors have plenty of our own models, all created by other physicians. But it’s the difference of trying to predict the weather on an 18th century sailing ship versus using a ship with modern technology. I can take someone’s PSA, size of tumor, grade of tumor and turn to charts that tell me about the type of cancer progression they might have. But that’s like trying to predict tomorrow’s weather using today’s calendar date, temperature and barometer reading. Paul is the equivalent of a global weatherman – he can give me the forecast for cancer using thousands of variables instead of three.”

And it’s more than that. Newton’s models can even help save lives where current medical imaging may fall short.

“If you have a field,” observed Dr. Nieva, “and it’s full of weeds, and you have one weed so big you can see it from your farmhouse — what do you do? Do your take your tractor and your backhoe and remove the one big weed? Most farmers will tell you no. You have to treat the whole field.”

Modern imaging, PET and CAT scans, allows the doctor to see every cancer tumor in high resolution, in three dimensions, just so long as that tumor is dime-sized or larger. “What we can’t see,” said Dr. Nieva, “are tumors the size of, say, gravel or dust. And those could contain thousands or hundreds of thousands of cancer cells.”

And so, just as with a large weed, Dr. Nieva argues, it’s better not to subject the patient to removing the dime-sized tumor they can see, if it will only be replaced by another in six months.

Another example is where exactly to focus the imaging. “The best example is the brain,” said Dr. Nieva, “that’s something we don’t normally look for. Imagine a patient with breast cancer walks into my office, and they’ve got a tumor in organ X. Based upon Paul’s models, we know if there’s a tumor in organ X, it’s likely that cancer has spread to the brain. We know which patients need to get an MRI for that.”

Newton’s work also sits nicely at the crossroads of what USC Viterbi Dean Yannis C. Yortsos refers to as engineering+, using engineering as the great enabler to solve problems from the arts to health care. His collaboration with Paul Macklin, assistant professor of the Keck School of Medicine of USC, was awarded a 2012 Zumberge Interdisciplinary Award last April. In addition, Newton is serving as a link between two of the largest cross-disciplinary cancer centers in the United States, one at USC and the other at the Scripps Research Institute in La Jolla. There are only 12 Physical Science-Oncology Centers in the United States, as funded by the National Cancer Institute, and Newton has found willing collaborators in both.

For Newton, he wanted to have real clinical impact. “If you sit in your office and read scientific papers,” he said, “you might develop models that are very interesting, but the risk is that they may bear very little resemblance to the real world.” Like his famous namesake, Newton couldn’t resist the challenge.