Skip to Content

Imaging the Genetic Profile of a Tumor

MRI scans could be used to determine which drug will work best against a brain tumor.

When patients are diagnosed with glioblastoma, the most common form of brain cancer, they often have only months to live. Even though researchers’ knowledge about this tumor’s biology and genomics has expanded in recent years, no significant treatment strategies have been developed during the past 25 years. Now there is preliminary but strong evidence that the appearance of these tumors in magnetic resonance images (MRIs) can be used to predict their genomic profiles. Researchers hope that MRI will soon be used for dividing glioblastomas into genomic subtypes, in turn allowing doctors to put patients on the best drug before a biopsy is even taken.

Visual cues: Researchers at the University of California, San Diego, have made connections between features in magnetic resonance images of the most prevalent form of brain cancer (above) and gene-expression patterns. This information could lead doctors to the best treatments for individual patients.

At the genetic level, two patients with glioblastoma (or any particular cancer type) may have very different tumors. For example, one patient might respond well to a therapy targeting tumor blood-vessel growth, while the other patient’s tumor might have activated a genetic program that allows it to resist such a therapy. Right now, it’s difficult for doctors to tell these patients apart and to predict responses to many other targeted therapies, so all glioblastoma patients are given the same treatments. Pathologists can perform gene-expression studies on biopsies, but these tests are expensive and not in wide use; MRI scans are standard.

“By tying imaging features to specific biology, we hope to give insights into treatment targets and patient prognosis,” says Michael Kuo, a radiologist at the University of California, San Diego, who led the study connecting MRI scans with glioblastoma genetics. In work described in today’s issue of the Proceedings of the National Academy of Sciences, Kuo’s group identified five visually discernable kinds of tumors strongly associated with particular gene-expression profiles that are tied to targeted therapies. The paper also describes for the first time a particularly aggressive subtype of the disease.

According to the paper, Kuo and his collaborators defined a set of traits present in MRI scans of 22 glioblastoma tumors. “Normally, when a radiologist looks at a tumor, he’s focused on diagnosis: is this a primary tumor, a metastasis, or an infection?” explains Kuo. The San Diego researchers defined a longer list of characteristics, including morphology and the interaction of the tumors with the surrounding tissues. “The premise here is, there is a lot more information in the images than is currently accounted for,” says Kuo.

Then the researchers looked for connections between the ten types of tumors shown in the MRIs and the activity of seven genetic programs by studying the patients’ biopsies using microarrays. These genetic programs included groups of genes associated with blood-vessel growth, cell proliferation, and other characteristic aspects of cancer biology, all of which are targeted by existing drugs. The team also looked for associations between tumor appearance and overexpression of one gene in particular, coding for epidermal growth factor receptor, a cell receptor that’s overactive in many glioblastomas.

Radiologists have seen some of these connections anecdotally but haven’t had the data to back them up, says Patrick Wen, a neuro-oncologist at the Dana-Farber Cancer Institute, in Boston. For example, Wen says, oncologists have suspected the connection between tumor appearance and response to therapies targeting tumor blood vessels demonstrated in Kuo’s study. If further data back up this result, it is one of many that would prove therapeutically useful. Wen says that the results are a very important first step “towards using imaging to tailor treatment without having to take tissue.”

“Gene-expression patterns result in anatomical changes you can see in these images,” says Webster Cavanee, director of the Ludwig Institute for Cancer Research at the University of California, San Diego. Cavanee, who was not involved in Kuo’s research, believes that the connection between imaging and genetics will hold in other cancers, and perhaps other diseases. Last year, Kuo published a paper connecting characteristics of liver tumors in CT scans with gene-expression patterns. This suggests that the connection between anatomical changes and gene expression will hold up across imaging and tumor types alike.

Cavanee and Wen agree that if Kuo’s work holds up in larger studies, it could have a major clinical impact. “An imaging protocol could be powerful,” says Cavanee, because medical imaging is already part of standard cancer care. “The images are already there. It’s a matter of layering information on something that already exists without adding cost–you’re just adding precision.”

Another advantage of using images for molecular profiling rather than biopsies and microarrays is the global view that this affords. “In general, a biopsy does represent the whole tumor,” says Kuo. But some tumors, particularly glioblastomas, may be heterogeneous–part of a tumor might be more vulnerable or resistant to particular drugs than others–and a biopsy only gives information about one region. Microarrays do give much more specific information, but this level of detail might not be needed: MRI scans might be good enough to rapidly get patients on a drug that’s likely to work. And to get a biopsy, “you need to go in physically and get tissue,” which carries risks, says Kuo.

Kuo’s group also identified a particularly malignant version of glioblastoma and connected its MRI scans and genetic profile with poor patient outcomes. “Instead of saying all patients with [glioblastoma] tumors are the same, we can sort them into two subtypes based on outcome,” says Kuo. His work suggests that these patients should be identified and treated more aggressively, while patients with less malignant cancers can be spared the side effects of aggressive treatment.

Wen describes Kuo’s work as an important first step, but a small one. Kuo’s team is currently working to validate its results in a larger group of glioblastoma patients. “Our data suggest we’re going in the right direction,” he says.

Keep Reading

Most Popular

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

OpenAI teases an amazing new generative video model called Sora

The firm is sharing Sora with a small group of safety testers but the rest of us will have to wait to learn more.

Google’s Gemini is now in everything. Here’s how you can try it out.

Gmail, Docs, and more will now come with Gemini baked in. But Europeans will have to wait before they can download the app.

This baby with a head camera helped teach an AI how kids learn language

A neural network trained on the experiences of a single young child managed to learn one of the core components of language: how to match words to the objects they represent.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.