Could A Genome-Savvy Computer Help Change The Way We Treat Cancer?

For years, the war on cancer has been about attacking specific body parts—developing treatment and funding research around lung cancer, for instance, or breast or brain cancer, to name others.

But it turns out the disease is more complex: Scientists now know that with genetic analysis, they can attack specific mutations of cancer cells rather than just the organs in which they appear.  And so the fight against cancer is entering a new phase, one that focuses on developing tactics and medications to fight cancer cell by cell.

One of the most daunting tasks in the medical community, for instance, has been not only developing drugs that keep pace with the many ways cancer takes hold in a patient's body, but also finding medication that will respond to specific abnormalities; in some cases, doctors must cycle through drug after drug with a patient to find an effective treatment. 

This summer, researchers in England could take a big step forward in shaping a more sophisticated approach to attacking cancer cells by refining how clinical trials test new cancer treatments. While that may not sound transformative, it could help patients access effective drugs more quickly.

In most clinical trials, researchers test just one drug, given at random to patients whose results are measured against those who receive a control medication or placebo.  While this serves the researchers' purposes, it often doesn't do much for patients who receive the placebo; likewise, patients for whom the new drug is ineffective must wait for another trial to try a different medication.

This upcoming trial, however, will test as many as 14 different drugs, provided by the pharamaceutical companies AstraZeneca and Pfizer, on patients with advanced lung cancer. Scientists will test samples of the patients' tumors when the trial begins, and assign a drug based on the abnormalities found in a given tumor's cells. 

Fifteen to 20 patients will receive each of the medications. Drugs that are successful could be fast-tracked to market; those that aren't can quickly be dropped. More drugs could be added to the trial as time goes on. 

The approach not only does away with randomization, it also gives doctors better insight into which medications work for certain abnormalities. If successful, the trial could mean oncologists will focus far less on the part of the body where cancer has developed and instead will zero in on the mutations in cells that caused the cancer. The hope: such customized treatment will turn cancer from a killer to a chronic illness that can be managed with medication.

A job for Watson

One of the biggest challenges of treatment guided by genetic analysis is the sheer volume of information doctors need to absorb in order to develop more personalized plans.

Every day, scientists are making new discoveries about the DNA sequencing of cancer cells and which mutations are harmful, but there’s no one place doctors can find that research. Instead, they need to comb through medical journals and clinical records, a time-consuming and daunting process.

It sounds like the kind of job best handled by a super computer. And that's what the scientists at the New York Genome Center are preparing to do. Last month, the biomedical research nonprofit announced a partnership with IBM to use Watson, the enormously powerful computing system best known for humiliating human Jeopardy champions on national TV, for genetic research.

Since its shining moment as a game show contestant, Watson has consumed more than 23 million medical abstracts. And because of its ability to understand language and evaluate hypotheses, Watson can go beyond storing data to also find meaning in it. 

For the New York Genome Center’s pilot study, Watson will focus on glioblastoma, the most aggressive of malignant brain tumors.  Scientists will sequence the genomes of the brain tumors in 20 patients, then turn that data over to Watson.

The computer’s job is to first identify the mutations in the various tumor cells. Based on the vast amount of medical knowledge it has accumulated, Watson will then provide a hypothesis of how those mutations cause cancer (some cancer mutations, for instance cause cells to ignore signals to stop growing while others stimulate the growth of nearby blood vessels).  Finally, Watson will create a list of drugs that could be effective in treating different kinds of cell mutations—some of which may have never been used in cancer treatment before. 

Ultimately, it will be up to hospital boards to weigh Watson’s treatment recommendations; any given hospital could decide to use one medication or a handful of them. But either way, having Watson do the heavy lifting allows doctors to give patients effective treatment much more quickly than they could have with human analysis alone.

Getting personal

Here are some other recent advances in cancer research:

· Aspirin power:  A new study published in Science Translational Medicine has found that taking daily aspirin may reduce the risk of colon cancer among people who have a high level of a certain type of gene in their colons.

· Early detection:  Researchers at Stanford University say they have devised a method through which lung cancer can be detected with a simple blood test.  Using this technique, they identified about 50 percent of the study participants with stage-1 lung cancer and almost all of the patients with more advanced lung cancer. They believe that their approach eventually could make it possible for blood tests to detect many types of cancer.

· In the blood:  Meanwhile, at Johns Hopkins University Kimmel Cancer Center, scientists say they’ve developed a blood test that can detect the recurrence of breast cancer with 95 percent accuracy.  The test is able to monitor 10 breast cancer-specific genes in patients' blood.

· When cancer drugs fail:  A team of researchers at Hebrew University say they’ve identified the process through which tumors become resistant to certain drugs. And they believe that it could lead to a reliable way to predict which patients will be helped by chemotherapy and recover and which the drugs will not help them.

Randy Rieland | | READ MORE

Randy Rieland is a digital media strategist and contributing writer in innovations for Smithsonian.com.