For the better part of a century, brain tumors have been judged by their appearance. Where a tumor was located, how much it spread, and what it looked like under the microscope all determined whether a patient was given a good or bad prognosis, and how they were treated. But these rough measures can only tell so much about the aggressiveness of a particular tumor, its potential response to therapy, or longer term implications for the patient.
Over the last 25 years, researchers have started to see that there is far more to cancer than meets the eye. Longtime collaborators Robert Jenkins, M.D., Ph.D., at Mayo Clinic and Margaret Wrensch, M.P.H, Ph.D., at the University of California, San Francisco have applied the latest molecular technologies to probe the genetic wiring deep within gliomas, the most common kind of brain tumors. Based on their findings, they have developed a new test that will be available later this year to guide prognosis and treatment, and, in some cases, even direct targeted therapies to improve the outlook for patients with this terrifying disease.
Tracing the origins of cancer
Cancer comes in a variety of makes and models. Like an automobile manufactured by Ford or Honda, tumors are usually labeled according to their origin. Breast cancers, skin cancers, and brain cancers are thus named for tumors found in the breast, skin, and brain. Gliomas take their namesake from the glial cells that provide support for the brain’s neurons. Within gliomas, subtypes called astrocytomas, oligodendrogliomas, and mixed oligoastrocytomas each arise from astrocytes, oligodendrocytes, or a combination thereof.
Even though such designations can help to separate the most aggressive tumors from the slower growing ones, they fail to predict the best course of treatment for a significant number of cases. It is like walking onto a used car lot and trying to figure out how fast a car will drive or how reliable its brakes are by looks alone. One sports car might move faster than another, and a perfectly nice-looking sedan might have perfectly shoddy brakes. The only way to know for sure is to take a look under the hood.
“Often there are critical differences between tumors that don’t show up on histology,” says Dr. Jenkins, the Ting Tsung and Wei Fong Chao Professor of Individualized Medicine Research and a professor of laboratory medicine and pathology at Mayo Clinic. “The underlying genetics can be different and what’s driving the tumor development can be different. I think we’re learning that molecular features trump histologic features.”
All cancers begin with mutations in genes that control how cells function, particularly the way they grow and divide. Some mutations have been there all along, like a part that has been defective since it left the factory floor, whereas others are acquired over time, similar to the wear and tear on a vehicle.
These cancer-causing genes can take on three different forms. Oncogenes act like the accelerator of the car: mutations in these genes put the pedal to the metal, sending cells growing wildly out of control. Tumor suppressor genes are the brakes: when they are faulty cells continue to divide even when they should stop. Caretaker genes play the part of mechanic: mutations mean that defects don’t get repaired like they should and the cells accumulate mutations, setting them on a path to destruction.
Hundreds of different genetic mutations have been identified in gliomas, many of them by the Jenkins and Wrensch laboratories. The two researchers recognized that these mutations provide a map of the complex genetic wiring underlying every tumor, showing how it developed, how quickly it will progress, and if it will respond to specific therapies. Several years ago, they began to investigate whether this wiring could form the basis of a new and improved classification system for gliomas.
Their study focused on a powerful trio of genetic alterations, the 1p/19q codeletion, the IDH mutation, and the TERT mutation, chosen because they occur early during cancer development, are more common to gliomas, and are sometimes associated with desirable outcomes. First, the researchers scanned genetic data on 317 gliomas from the Mayo Clinic Case-Control Study and marked each of the tumors as either positive or negative for the three alterations.
Most gliomas fell into five of the nine possible molecular categories — triple-positive, TERT- and IDH-mutated, IDH-mutated-only, TERT-mutated-only, and triple-negative. The researchers compared patient characteristics among each of these different molecular categories. They found that the patients shared a similar age of onset and overall survival with other members of their same category. Patients with triple-negative gliomas, for example, had poorer overall survival than TERT– and IDH-mutated and triple-positive gliomas. When the researchers repeated the analysis in 351 gliomas from the UCSF Adult Glioma Study and 419 gliomas from the Cancer Genome Atlas (TCGA) Study, they got the same results both times.
The results, published in the New England Journal of Medicine, clearly indicate that molecular methods can do a better job than current methods of predicting which specific treatment course is necessary for each individual patient. For instance, Drs. Jenkins and Wrensch found that the molecular classification can identify patients with histologically defined lower-grade tumors who have less favorable outcomes and deserve more aggressive therapy. Likewise, it can identify patients in other categories who have a better prognosis and could be spared harsh treatments and their side effects.
Taking it to the clinic
Dr. Jenkins and his colleagues at Mayo are now working to translate this approach into clinical practice. They have expanded their starter set of three cancer alterations to more than 50 to provide even more actionable information to clinicians and patients.
“We already knew that other defects went along with the three defining mutations. Why not generate a test that could find those things too?” says Dr. Jenkins. “It’s as cheap to test 50 genes as it is to test three. It’s as cheap to evaluate the whole genome for chromosomal alterations as it is to just look for the 1p19q codeletion. By widening our net, it would strengthen these categories and, from the patients’ perspective, it may generate more personalized data about their disease.”
The team is using microarrays or “gene chip” technology to map where segments of DNA are duplicated or missing in patient tumors, like the 1p19q codeletion. They are also sequencing the genomes of the tumors in search of mutations in 50 genes that might predispose to cancer or that might be acquired as tumors grow and progress. The researchers have shown that these other genetic changes follow specific patterns within the molecular categories, making it even easier to group patients.
As they’ve added more genes to each of the categories, their new method of naming gliomas has quickly become outdated. Tumors once thought of as IDH only were found to have defects in two other genes, p53 and ATRX. But they couldn’t be called triple-positive, because that would put them in the same category as gliomas that are positive for the 1p19q codeletion, the IDH mutation, and the TERT mutation. That distinction is not trivial.
Both kinds of triple-positive glioma are considered low-grade. But their outcomes are vastly different. Patients who have the 1p19q/IDH/TERT variety can expect to live for 15 to 20 years, whereas those with IDH/p53/ATRX mutations are only given seven years. Knowing what is ahead can help patients decide whether to subject themselves to the damaging effects of radiation, which can leave them with the equivalent of early Alzheimer’s disease, or endure multiple punishing rounds of chemotherapy, which can backfire and allow cancer to come back with a vengeance.
Unraveling a tumor’s wiring could not only guide treatment with traditional methods, but it may also point to new therapies that precisely target cancer-causing defects. Such targeted therapies are generally more effective at treating cancer and come with less severe side effects. Some of the most prominent members of this novel class of drugs go after the BRAF oncogene, which is mutated in the majority of melanomas and a small percentage of gliomas. Jenkins and his team included BRAF in their 50-gene panel so they can pick out the patients most likely to benefit from this targeted therapy. They are also on the lookout for other genes linked to targeted therapies or associated with glioma risk that might make the test even more powerful.
“We’re already working on version 2 of the test, which will probably include 70 genes,” says Dr. Jenkins. “That is one of the big challenges of clinical cancer genetics. People like me are continually doing science and finding new things. It’s going to be like software packages. We’re going to have a version 2 and a version 2.1, a version 3, a panther, a lion…”
Building gliomas from scratch
Glioma patients seen at Mayo Clinic can already receive the microarray test – the one that assesses chromosomal aberrations like the 1p19q codeletion — as part of their treatment. Jenkins hopes that the other new test, the 50-gene panel, will be available by September, with other versions still ahead. In the meantime, he is interested in understanding how the combination of genetic mutations lurking within the different molecular groups could ultimately give rise to cancer. To find out, Dr. Jenkins plans to build gliomas in a dish, like a vintage car aficionado trying to resurrect a model T.
Dr. Jenkins’ laboratory will use the revolutionary genome-editing tool CRISPR to introduce these mutations into cells, and then will wait to see how the disease unfolds. They’ll model the IDH/p53/ATRX tumors first before tackling other combinations of cancer-causing mutations, including both the underlying variants (those factory-manufactured defects) and the acquired ones (the wear-and-tear damage). The trick, Jenkins says, will be hitting the cells at the exact time in their development to mimic what happens when these cancers form in the human body.
“It’s possible that these tumors have been around since the person was born,” he says. “It may very well have started in utero, and it just takes 30 years for them to come to clinical attention. Or the vulnerable period may be much later, but frankly we’ll never know until we do the experiments.”
Dr. Jenkins was awarded a $300,000 grant from the National Brain Tumor Society* to work on the problem. He hopes that building models of gliomas can lead to new therapies that target the entire ensemble of mutations underlying specific types of tumors.
Every year, approximately 17,000 Americans are diagnosed with a primary brain tumor, and more than 13,000 succumb to the disease. Dr. Jenkins believes that to save more patients, clinicians need to stop focusing on individual aspects of tumors, such as appearances or single genetic defects, and instead look at the bigger picture. The genetic wiring that drive tumors may ultimately provide the key for stopping deadly gliomas in their tracks.