This Test Could Give Cancer Patients a More Personalized Prognosis

Adam Hayden was wide awake as the neurosurgeon pried a coin-sized piece of bone from his skull, exposing both his brain in its clear sac and a tumor the size of a plum. The neurosurgeon removed the tumor in pieces, sealing them in a freezer colder than the surface of Mars until a pathologist could confirm their identity: a deadly brain cancer called glioblastoma. Eventually the surgeon reached a point where he could go no deeper without cutting healthy tissue. Stopping meant some cancer would remain. But if he kept cutting, Hayden would likely be paralyzed on the left side of his body. No one could make the choice but Hayden. He was glad to be awake for it.

The prognosis for glioblastoma is grim. Two years after diagnosis, just 30% of patients are still alive. After five years, it’s 3%. But patients still have to decide how they’ll be treated, and all of them currently use crude survival estimates like the ones above. That’s not good enough for Jill Barnholtz-Sloan and her colleagues, who recently published a glioblastoma nomogram—a statistical tool that synthesizes test results into an individual prediction.

A typical untreated brain with glioblastoma

“The easiest statistic to quote is the median survival,” said Barnholtz-Sloan, a cancer epidemiologist at Case Western Reserve. “But we all know two things. One, not everyone is the median, and two, every patient wants to think that they’re going to be the patient with exceptional survival.” Nomograms tackle both problems.

On the surface, a nomogram is a series of questions about you and your tumor. As you answer questions, you accumulate points, and the point total is converted into your likelihood of having a particular outcome. The backbone of the nomogram is a set of equations which determine how many points to assign each answer.

Though nomograms offer patients better information about their potential outcomes, getting clinicians to adopt them has been a challenge, said Michael Kattan, the Chairman of Quantitative Health Sciences at the Cleveland Clinic. Kattan’s prostate cancer nomograms are widely used today, but he doesn’t think that would be the case if not for his collaborator Peter Scardino, a surgeon who, in Kattan’s words, “spread the Gospel” about nomograms to other doctors.

Kattan met Scardino while both men were working at Baylor College of Medicine. “I’d hear him and the others talk about how they really wanted to counsel patients better,” Kattan said. “And I could see the way they were putting the pieces together and just be thinking, ‘that’s not really the best way to do it.’” So Kattan roughed out a prostate cancer nomogram for Scardino.

The nomogram revealed staggering problems. “We’re telling these patients they’re high-risk. and they’re actually low. And we’re telling these patients they’re low-risk, and they’re actually high,” Kattan remembers telling Scardino. “We’re just making mistakes right and left by doing it that way.”

Kattan didn’t plan to go into medicine. He was working on a PhD in management information systems when he learned he had advanced Hodgkin lymphoma, a cancer of the immune system. After two months of chemotherapy, Kattan faced a decision: whether to undergo radiation. Radiation would destroy any lingering cancer cells in his bone marrow, but it might damage his heart permanently. The risk estimates his doctor gave him were too broad—they were for populations, not an individual. He chose the radiation.

Hayden points out that this problem comes up all the time, most famously when women who are genetically predisposed to breast cancer decide whether or not to get mastectomies. “That is a life-altering decision based on statistical information,” Hayden said. “I think what we need is more sophisticated statistics.” He could just as easily be talking about Kattan, or himself.

More sophisticated statistics, like Barnholtz-Sloan’s new nomogram, might easily have helped Hayden as he weighed the relative disadvantages of paralysis and death. But his surgeon had another method of reducing uncertainty. He told Hayden to just ignore the future and decide based on his “quality of life today.”

“That sort of set off the lightbulb,” Hayden said. Today, he could chase his toddlers around the yard. Today he could play hide-and-seek. He told the surgeon to stop. “I said…Let’s preserve the function that I have. And I would make that decision exactly the same, over and over and over and over.”