Cancer often seems frustratingly capricious — striking some light smokers while leaving their three-pack-aday counterparts unaffected, for example, or yielding readily to chemo in some while resisting the therapy in others afflicted by the same tumor type.

Some of the mystery surrounding the way cancer works has been explained. Scientists discovered years ago that certain types of breast cancer are promoted by specific hormones. Yet while science’s ability to make such distinctions is getting better, the nature of many tumors and the reasons why they progress as they do remain challenging topics.

Peter Sorger’s among those trying to shed light on such issues. A professor of biology, Sorger has a familial pedigree that might seem to have made his professional path nearly inevitable. His mother has a doctorate in microbiology. His dad was a pathologist and chief of staff at Cambridge’s Mt. Auburn Hospital.

Given all that, why did Sorger — self-described as not all that great at math — in effect go back to school, as an MIT faculty member, to study subjects like linear algebra? It’s because he decided that to accomplish his research goals, he had to become a modeler.

Some scientists say that computer modeling techniques could lead to cancer treatment’s Holy Grail: individualized therapies. Others are skeptical that modeling has much to offer.

As a modeler as well as a biologist, Sorger says that modeling something as immensely complicated as cancer does pose tough hurdles. One example: the simple phenomenon of a signaling protein ordering a specific cell to grow and divide — a process that can go awry in cancer — may involve dozens of individual biochemical steps.

Getting all the measurement data you need to create a truly complete model of an individual’s tumor from a biopsy plus blood tests, in short, is almost certainly impossible. But that doesn’t rule out an important role for modeling.

“Modeling techniques are designed to deal with situations where you don’t have all the information you’d like,” notes Sorger. In fact, he says, if you can measure and use as the basis for your model, say, 50-to-100 proteins, the model can indeed be medically useful.

The first clinical benefits from the MIT modeling work will likely flow from helping doctors make better decisions about who will benefit from a particular drug regimen — and that day may not be far off. “We’re working with a couple of pharmaceutical companies,” notes Douglas Lauffenburger, head of biological engineering and a collaborator of Sorger’s, “and they’re testing some of our models to see if they can categorize cancer patients based on measurements of proteins in those patients.”


These and other advances are among the reasons cancer center head Tyler Jacks says this is an exhilarating time to be doing cancer work.

Jacks’ MIT roots are deep. His dad is a former professor of management at the MIT Sloan School, so as a child the future scientist spent long hours on campus. Maybe partly because of that history, Jacks has a perspective on the MIT way of tackling tough challenges.

Institute people he says, haven’t minded being out of step with their peers elsewhere — in fact, it’s often a point of pride to be so. Numerous other institutions, he notes, are doing cancer work similar in many respects to what’s going on at MIT. He is delighted by this, noting that the more smart people there are working on cancer, the better the chances for progress.

Nonetheless, Jacks believes there’s a difference between MIT and other institutions, in cancer research as elsewhere. “We don’t like to be where everybody else is,” he says. “We like to be out there on the edge. And that’s how we’re approaching the task of understanding and curing cancer.”