Tailoring Tumor Treatments

In May 2012, an eight-year-old boy named Kelvin was diagnosed with an unusual cancer called a sarcoma, a tumor not typically found in the brain. His doctors took an aggressive approach: They removed the tumor and gave him six cycles of high-dose chemotherapy. The tumor subsided.

But one year later, the fast-dividing tumor cells had spread from his brain into his lungs. Kelvin lost his appetite and 20 pounds, and he was too tired to go to school. A fresh round of chemotherapy brought no results. The situation was grim. “Most people with relapsed sarcoma of that kind would live not more than three months,” says Rod Rassekh, Kelvin’s oncologist at the British Columbia Children’s Hospital in Vancouver.

At this point, deciding that more chemotherapy would be futile, Rassekh decided to try a new approach called “precision medicine”, in which patients get personalized treatments tailored to their genetic make-up and history. He contacted Olena Morozova, a bioinformatician at the the Treehouse Childhood Cancer Initiative, a project at UC Santa Cruz, a group that uses genomic data to identify therapy. Treehouse is one of five nationwide precision medicine consortiums aimed at pediatric cancer, and it’s the only one on the U.S. West Coast. Rassekh hoped that Morozova could supplement his own analysis to help find a treatment that would work for Kelvin’s unique condition.

Bioinformatician Olena Morozova of the Treehouse Childhood Cancer Initiative at UC Santa Cruz Genomics Institute. Photo: Devika G. Bansal

Each day, an average of 42 children are diagnosed with cancer in the U.S. Many of them follow the same path as Kelvin: Standard chemotherapy either doesn’t work for them or is too toxic for their developing bodies. Newer drugs on the market target only the abnormal cancer-causing proteins. That works a lot better than the sledgehammer approach of chemotherapy, which targets all growing and dividing cells. But even though targeted drugs have shown promise for treating cancer in adults, the same isn’t true for kids. Most of these drugs haven’t been tested in children, and among the ones that have, very few have worked.

Finding effective cures for children is a struggle because childhood cancers carry a wide variety of genetic changes, far more than adult cancers, and drugs don’t exist to target the many root causes found in kids. To make matters worse, pediatric cancers make up a scant 1% of all cancers diagnosed each year, deterring pharmaceutical companies from developing new drugs specifically to target pediatric cancer. The market just isn’t large enough to justify the high costs.

To get around that problem, Morozova and the the Treehouse Childhood Cancer Initiative are looking for adult and pediatric cancer drugs that might work in individual children, such as Kelvin, by matching cancer therapies to the specific characteristics of their tumors. “We can’t change which drugs are available,” says Morozova. “But we’re realizing there are a lot of opportunities [for pediatric cancer]—a lot more than we previously thought.”

Kids aren’t small adults

The incidence of childhood cancers has slowly increased over the past four decades. One in 285 children in the U.S. are diagnosed with cancer, according to a 2017 American Cancer Society report; that number has grown by 0.6 percent per year since 1975. But treatments have improved too. Overall, more than 80 percent of children with common types of cancer survive at least five years from diagnosis, compared to the death sentence that pediatric cancer once was in 1950s.

Cancer types that occur in children are different from those in adults. Leukemia accounts for close to 30 percent of all pediatric cancers, followed closely by brain tumors. In contrast, breast, skin and lung cancers are the most common types for adults. The causes of most childhood cancers are unknown and are not strongly linked to lifestyle or environmental risk factors, unlike many adult cancers.

Pediatric cancer is not just adult cancer in small people.

“Pediatric cancer is not just adult cancer in small people,” says oncologist Mark Kieran of the Dana Farber Cancer and Blood Disorders Center in Boston. “It’s fundamentally different.”

Rapidly dividing cells are the hallmark of cancer, but they also are part of normal childhood growth and development. The most common kinds of kid cancers occur in blood and brain tissues, where cells grow quickly. “Maybe the way pediatric cancers trick the system is to convince the body they’re rapidly dividing because that’s how you grow,” Kieran says.

Many targeted drugs have been developed in adults to block specific proteins that cause cancer. On the contrary, only three such drugs were developed specifically for childhood cancers in the last 20 years, Morozova says. Precision medicine aims to match targeted drugs to kids’ tumors based on their specific molecular signatures.

Tumor cells hold clues

As a graduate student in Vancouver, Olena Morozova studied pediatric cancers. When she moved to Santa Cruz for her postdoctoral work, her colleague’s six-month-old daughter was going through chemotherapy. Moved by that experience, Morozova wanted to foster awareness of the disease by raising money for Alex’s Lemonade Stand Foundation, a patient community-driven non-profit to fund pediatric cancer research.

“We thought about putting a desk in the biomedical sciences building with a jar of lemonade. We would make $100 and it would be just a gesture,” Morozova remembers with a smile. But the community at the UC Santa Cruz Genomics Institute picked up the idea, and they ended up raising $12,000 for the foundation. Impressed by the momentum this had created, institute director David Haussler told Morozova: “Don’t stop now. Look at how many people are moved by this cause.”

“Basically, we all wanted to do something more than just basic research,” Haussler recalls. “But it was really Olena’s determination that pulled all of this together.”

Around the time Kelvin’s cancer relapsed, Morozova began mining the Cancer Genome Hub at UCSC, then world’s largest public collection of tumor signatures, for clues to pediatric cancer. The database contains genetic information of tumors from thousands of people, a useful tool when trying to understand what makes cancer cells go awry.

Tumor cells are different from other cells in the body. Normal cells divide under very tight control, but tumor cells go haywire and divide rapidly. Cells know when and how much to grow and divide through genetic pathways. In cancer cells, one or more proteins in these pathways can go rogue, commanding the cells to grow nonstop. The loss of control can happen due to mutations at many points along the pathway.

Illustration: Taylor Maggiacomo

Scientists can identify the nature of these mutations by taking tumor cells and reading the strings of DNA bases, labeled A, G, T, and C, that make up their genetic code using a technology called genome sequencing. The sequencing reveals which genes in tumor cells are different from those in normal tissues, and therefore could be causing the disease. A detailed map of these molecular signatures in tumor cells is useful for targeted therapies; having a list of suspect genes means researchers can aim specifically at them, Morozova says.

In the last few years, scientists have begun to understand the genetic alterations that occur in different tumor types. Often, cancers that originate in different parts of the body can arise from errors in the same genetic pathways—and thus can be treated with the same drugs that target these genetic pathways. It’s a bit like using the same drug, ibuprofen, to reduce pain in all parts of the body, as long as inflammation is the root cause. “People are now starting to apply that knowledge clinically,” says William Parsons, pediatric oncologist at the Baylor School of Medicine. “But we’re at the edge of doing that.”

Precision medicine in pediatric cancer is a relatively new approach, and only a handful of groups have tried it. Published studies have looked for tumor mutations for which FDA-approved drugs exist in the market, Morozova says. By focusing only on DNA mutations, however, researchers can miss a lot of information, she adds. A gene could be perfectly normal in its DNA, but still affect a pathway adversely if tumor cells produce it in abnormally high amounts—just as a lot of puny street bullies can together do as much damage as a couple of burly toughs.

Morozova and her colleagues use DNA and RNA sequences from patient tumor cells to look for genes that may be producing too much protein. They do this by comparing the patients’ RNA sequences to thousands of other tumor sequences. Through their comparative genomic analysis, the Treehouse team gathers information on all the genes that are present in unexpected amounts, suggesting that they may be important for the tumor to grow.

This method creates a list of possible suspects that behave unusually and might be targeted by a drug. For example, BRAF is a tumor-causing protein in skin cancer. Its mutation turns it on nonstop, so it tells cells to keep growing and dividing. To add to the DNA mutation analysis already performed by others, the Treehouse analysis checks whether a cell with normal BRAF is nonetheless making too much of the protein, which could also cause tumors to grow.

“There’s a lot of interest in personalized medicine for pediatric cancer, but often people think of that as looking just at the level of mutations and DNA,” says pediatric oncologist Alejandro Sweet-Cordero at UC San Francisco, who enrolled 59 patients into the Treehouse study while at Stanford University. But pediatric patients often don’t have these DNA mutations, which are primarily the sort targeted by existing drugs.

“The purpose of the collaboration with Santa Cruz is to dig deeper into the genetic information to try and find targetable pathways that may be relevant to pediatric patients,” says Sweet-Cordero, who hopes to open a similar study at UCSF in 2017.

A jacked-up JAK pathway

The molecular detectives in Vancouver started looking for a good treatment match for Kelvin in the spring of 2014. To their surprise, they found that Kelvin’s sarcoma was churning out a protein called JAK, in large amounts, which is atypical for this type of cancer. The next step was to try out ruxolitinib, a drug that blocks the JAK pathway.

“You would never think to use that drug for a sarcoma without doing this type of work,” says Rassekh, Kelvin’s oncologist. “But biologically, it made sense to try it.”

Rassekh shared the news with Kelvin’s mother. “This is experimental; it may or may not work,” he told her. By that time, Kelvin’s condition had worsened, so his mom decided to try the ruxolitinib treatment.

Hunched over her computer one July morning in 2014, Morozova was peering into the alphabet soup she had just received from one of the patients studied by Treehouse when an email popped into her inbox. It was from Rassekh.

“I’m including this patient on the JAK drug consistent with your analysis,” he wrote, alerting her that Kelvin was about to become the first patient who would be treated with a drug based on abnormal amounts of a protein rather than a DNA mutation.

Kelvin started taking ruxolitinib that July. Within one week he felt completely different, Rassekh remembers. Over the next three weeks, Kelvin regained 20 pounds and returned to school full-time. “He went from being a sick-looking oncology patient to someone who you’d walk by in the hallway and not realize he had a disease,” Rassekh says. “His clinical response was dramatic.”

Precision for all kids

Doctors hope many other patients could see the same benefits Kelvin experienced, but a few obstacles are holding them back.

One of the challenges is to get every child’s tumor mapped out in a standard way across all hospitals. Having more tumor sequences in the same form would ultimately allow data miners like Morozova to find a match in the large dataset for any new case, yielding clues on how to treat it better.

Another issue is that tumor sequencing is now largely reserved for patients with relapsed or high-risk tumors. Newly diagnosed patients, who might have curable diagnoses, often don’t get tailored treatments and are instead stuck with standard chemotherapy, says William Parsons of Baylor. This could change as oncologists become more comfortable with using genomics in their treatment paradigm.

Clinical decisions have traditionally been very passive, says David Haussler of UCSC. “You basically looked it up in a flowchart: ‘If the result of this test is this, then give that drug,’” he says. “It’s much more complicated than that.” The precision medicine approach makes clinical decisions very active by allowing clinicians to go above and beyond the standard protocols, if the case needs it.

Graphic: Taylor Maggiacomo

Moving forward, it will also be important to incorporate precision treatments with regular chemotherapy, because most patients’ tumors eventually mutate to evade targeted therapies. “I would much rather try to cure patients with lower doses of chemo with targeted agents alongside,” Rassekh says. “That’s where I think targeted agents are going to be really useful.”

Finally, most pharmaceutical companies are not making drugs for kids; they’re making them for adults. The majority of adult drugs don’t come in formulations that children can use. Most come as tablets or capsules that are difficult for a three-year-old to swallow, for example. In addition, oncologists don’t always have information on the amount of drug that is safe to prescribe to a child.

Experts believe the field will overcome these challenges as more cancer centers adopt the approach of tailored therapy.

* * * * *

Kelvin’s benefits were short-lived. Rassekh found that while the drug put the brakes on Kelvin’s tumor, it didn’t really shrink it—a common outcome for his type of cancer. Although the treatment could not spare Kelvin’s life, it did help him live 23 months post relapse, long enough to see his 12th birthday.

“He was normal until the very last month because of targeted agents and this approach,” Rassekh says.

Rassekh remembers Kelvin as a very bright, quiet boy, who was also wise. Near the end of his life, Kelvin told Rassekh that he was worried people would forget about him after he died.

“You’re the type of kid we remember our whole lives when we look after you,” Rassekh told him. “Part of what people are going to remember about you is how you were so brave to enroll in this clinical study.”

“Is my name a part of it?” Kelvin asked Rassekh.

No, Rassekh said, he wouldn’t release Kelvin’s name to protect his privacy.

But Kelvin didn’t like that answer. “When you give a talk, I want you to use my name,” he told Rassekh.

Honoring his wish, Treehouse stopped referring to Kelvin as ‘Patient 1,’ and revealed his name.

“The legacy Kelvin wanted was to never be forgotten,” Rassekh says. “His name will always be remembered.”

© 2017 Devika G. Bansal / UC Santa Cruz Science Communication Program