Personalizing Cancer Care with Precision Medicine

When Matt Ellefson received a diagnosis of advanced lung cancer on the last day of 2009, he quickly joined a clinical trial combining standard chemotherapy and radiation treatments with an investigational drug that managed to send his cancer into remission within six months.

By the time the South Dakota engineer experienced a recurrence 12 months later, an entirely new treatment option had become available. Based on an analysis of Ellefson’s new tumor, removed from a lymph node in his neck, doctors discovered a mutation in a gene called anaplastic lymphoma kinase, or ALK, which occurs in about 5 percent of lung cancers. In 2011, the Food and Drug Administration (FDA) approved a targeted drug, Xalkori (crizotinib), to tackle ALK-fueled lung cancers by blocking a protein produced by the gene. Ellefson now takes two Xalkori pills daily and, at age 51, is cancer-free.

"This provides so much hope,” says Ellefson, a lifelong nonsmoker. “It was amazing how well the drug worked. I started taking it at the end of October, and by mid-November I was training for a half-marathon.”

He developed a small brain metastasis in late 2012—not unexpected since Xalkori can’t cross the blood-brain barrier—that was soon eradicated with stereotactic radiation therapy. “I could noticeably tell I was getting my strength back,” he says of Xalkori’s effectiveness. “I had more energy, and I was more positive.”

Ellefson's experience demonstrates how biomarkers are transforming cancer care, enabling doctors to personalize their approach to diagnosing and managing the disease. Cancer biomarkers include identifiable gene mutations, proteins, growth factors and other biomolecules that can help diagnose malignancies at earlier stages, classify tumors to determine ideal therapies or monitor patients' treatment responses. Essentially, identifying the right biomarker could affect the entire continuum of cancer care for patients.

Promising Advances, Yet Obstacles Persist

While hundreds of potential biomarkers have been studied over the past decade—coinciding with quicker, cheaper next-generation gene sequencing abilities—fewer than two dozen cancer biomarkers are currently FDA-approved, and less than half of those are protein biomarkers detectable in blood. Researchers widely consider minimal or noninvasive biomarker tests, such as simple blood tests, to be the holy grail of the field.

“Ultimately, the goal we’ve all been working toward is trying to understand the way each tumor is physically built and how they develop in each person,” explains Kenna Mills Shaw, executive director of the Institute for Personalized Cancer Therapy at the MD Anderson Cancer Center in Houston. “We want to deconstruct the actual problem. That’s where biomarkers offer a specific target—and downstream, a specific treatment. If we can understand all the instructions tumors use, we can go after their specific vulnerabilities, their Achilles’ heels.”

But biomarker research is complex. Understanding the molecular alterations that characterize cancer cells requires a combination of genomics (analyzing DNA), proteomics (studying proteins) and transcriptomics (assessing RNA molecules, which translate the DNA code into proteins). And the research process involves three essential steps: discovery, verification and validation.

Discovering potentially important biomarkers isn't a problem. Groups such as The Cancer Genome Atlas (TCGA), a consortium launched in 2006 to use DNA sequencing to identify the most common and significant mutations in cancers, recently learned that 127 genes were significantly mutated among 12 different types of cancers, including breast, colon, lung and ovarian malignancies as well as acute myeloid leukemia. Endeavors such as this have produced biomarker breakthroughs that already affect treatment for millions of people with cancer, including the discovery of BRAF mutations in metastatic melanoma. HER2 mutations in some breast and gastric cancers, and several serum marker proteins that can signal recurrence and help determine prognosis in ovarian cancer.

[Biomarkers Could Help Lower Treatment Costs and Side Effects]

Too Much Information

In the hands of groups such as TCGA, next-generation gene sequencing is fueling much of the biomarker revolution, says Daniel Hayes, a professor of medicine and breast cancer research at the University of Michigan in Ann Arbor. Between 300 and 350 important abnormalities are expressed in humans’ 15,000 or so genes, he notes.

“The wave of the future is doing complete transomics of tumors—next-generation sequencing that either sequences the entire genome or the part that encodes for expressed genes,” Hayes says. “The brave new world right now is that we really don’t know what to do with these kinds of massive analyses . . . because we’re not talking about a single test. That gets complicated, too.”

And expensive. At the center of this information glut is the burgeoning field of bioinformatics, which develops methods for storing, retrieving, organizing and analyzing biological data to support the identification of trends and patterns that prompt new therapies. But while the costs of sequencing personal genomes has plummeted to a few thousand dollars, the downstream expenses associated with tracking and mining this compiled data are seldom considered, says Mark Gerstein, co-director of the Yale Computational Biology and Bioinformatics Program in New Haven, Conn.

Gerstein co-authored a 2011 paper published in the journal Genome Biology explaining how the real costs of sequencing are higher than thought, creating a gate-keeping effect for scientists hoping to explore potential biomarkers.

“Sequencing the genome of a person who has cancer—even multiple times—is quite reasonable because the cost has come down so much,” Gerstein says. “But the bottleneck now is the interpretation of all the sequencing and coming up with a way to understand all this data in simple terms. We don’t really have that yet.”

Money, Time, Technology Constraints

With so much focus on research, why are so few biomarkers considered reliable? Scientists say a serious breakdown in the research process—fueled by a lack of money, time and technology—hits the pause button on crucial clinical trials that might offer cancer patients wider and better drug options.

Moreover, current technologies can’t consistently verify the presence of candidate biomarkers in patient samples, says Ross Molinaro, an assistant professor of pathology and laboratory medicine at Emory University School of Medicine in Atlanta. “Verifying and validating biomarkers may take years to achieve, especially in cases where disease prevalence is low and the ability to collect samples from these patients is limited.”

Sometimes, stored tissue from prior clinical trials can be used in biomarker research, offsetting cost and availability obstacles, but these samples aren’t always ideal because the cells could have changed due to freezing or from the effects of earlier cancer drugs, researchers say.

Hayes contends that many more biomarkers would be considered clinically useful if pharmaceutical companies and government agencies invested as much in biomarker research as in drug development.

“In general, tumor biomarkers have sort of been the funny cousin in the corner, compared to drugs and therapeutics,” says Hayes, who helped establish guidelines for the development and validation of biomarkers. “It’s hard to make evidence-based guidelines because we don’t have the evidence.”

Cell Variations Also a Culprit

Cancer itself is also to blame for the dearth of reliable biomarkers, since even within single tumors, gene mutations may differ from cell to cell. That means a targeted treatment aimed at one biomarker-identified mutation may miss the mark with another, allowing the malignancy to proliferate.

Shaw explains that tumors often evade some treatments because, as cancer cells spread, they send out error messages to new cells, essentially ensuring that not all tumor cells will be identical. “When we ‘kill off’ part of the tumors that might have a given biomarker profile, that doesn’t mean the residual tumor cells or recurrences are necessarily the same genomically,” she says.

On the flip side, however, researchers from TCGA and other projects have found that some tumor types that weren’t considered similar share an underlying genetic likeness, unexpectedly opening the possibility of treating them with the same drugs. For example, nearly 70 percent of head and neck squamous cell carcinomas appear to have the same mutant p53 gene found in 95 percent of ovarian cancers.

Because tests looking for a single biomarker may not be sensitive or specific enough to both distinguish normal from diseased samples and tease out cancers from other conditions—important for preventing falsepositive and false-negative results—clusters of cancer biomarkers are beginning to be evaluated in single tests. One such test already available, OVA1, measures five biomarkers to determine the likelihood that an ovarian mass is cancerous before surgery.

But finding the "right" patients for testing new biomarkers in clinical trials can also be elusive, Shaw notes, especially as tests for multiple biomarkers begin to proliferate.

“We can intellectually design those experiments, but finding a big enough sample size to test is where it becomes difficult,” she says. “Add five biomarkers to one test for breast cancer, for example, and you’d have a much smaller population with that specific disease. We’d potentially have a very sensitive biomarker signal that’s testable, but a very small set of patients to actually work with.”

Biomarker Workhorses Spawn New Treatments

By tapping into an existing or readily identifiable patient pool, scientists are using some well-established biomarkers to produce more effective treatments for patients whose first-line therapies have failed. Consider patients whose breast tumors overexpress the protein HER2, which is found in about 20 percent of invasive breast cancer cases. This easily identifiable biomarker offers a fixed target for a steadily expanding array of therapies.

Jennifer Schreck’s life was transformed by the latest of four approved HER2-targeted drugs. After receiving a diagnosis of advanced breast cancer in 2006 at age 33, her condition worsened while on standard taxane-based chemotherapies combined with Herceptin (trastuzumab), the first anti-HER2 drug, which was approved in 1998.

Schreck, now 42, had to use a wheelchair by 2010—with the cancer in her liver and lungs as well as most of her bones—when she entered a clinical trial for Kadcyla (ado-trastuzumab emtansine, also known as T-DM1). The therapy links Herceptin with a potent tumor-killing drug to destroy only HER2-positive cancer cells, minimizing side effects to healthy cells. Schreck received Kadcyla every three weeks as a 30-minute infusion.

Kadcyla was Schreck’s ticket back to camping, hiking, biking and the active lifestyle her family enjoys, which metastatic disease had initially forced her to abandon.

“After the second treatment, there was dramatic improvement,” says the Aurora, Ohio, resident. “I felt like a windup doll—I was walking and not in pain. My life is so much better.”

While Schreck’s cancer has diminished and even disappeared in spots, she’s under no illusion that the biomarker-driven treatment will be a cure. Still, Kadcyla—shown to extend survival by up to five months on average compared with alternatives—gave her time with her husband and two young sons she might not have had, and she experienced only minor side effects, including constipation and low platelet counts.

“For whatever reason, I’m still responding and I’m very thankful,” she says. “I’m not just existing, I’m living.”

Aiming for similar successes, scientists remain hopeful about the potential of increasingly translating biomarker science into patient care. But they are also realistic about its inherent problems.

“Any time there’s a goal to combine multiple sets of inherently complex information, challenges may arise,” Molinaro says. “I compare this to doing your taxes. In some cases, a tax accountant can be used to simplify the complex process. In the biomarker pipeline, a universally accepted ‘biomarker accountant’ has yet to be identified.”

Talking Points:

> Patients should ask if a biomarker could help determine the ideal therapy.

> Some biomarkers can also reveal treatments that won't be effective.