Skip to Content

Cancer Genomes Let Drugmakers Get Personal

One startup, H3 Biomedicines, uses genome data to design drugs aimed at small groups of patients.

A growing effort to personalize cancer medicine tries to link existing drugs to patients based on the specific genetic and molecular anomalies of a patient’s cancer (see “Foundation Medicine: Personalizing Cancer Drugs”). But H3 Biomedicines, a startup in Cambridge, Massachusetts, wants to personalize cancer drugs from the beginning by designing drugs to target specific patient populations.

The company is taking advantage of cancer genome data—the DNA sequences of tumors from thousands of patients—made available by the National Institutes of Health as The Cancer Genome Atlas (TCGA), and by the International Cancer Genome Consortium. When the company analyzed the first 3,000 cancer genomes in the TCGA web portal, it looked for cancer mutations that were shared by at least 1 percent of different cancer types. “Everything that occurs in greater than 10 percent was a known oncogene or known tumor suppressor [the two main classes of cancer-associated genes],” says Markus Warmuth, CEO of the company. But that was not the case for the less common mutations. For mutations shared between 1 and 10 percent of cancers, especially those shared by 5 percent or less of cancers, “most everything was novel,” says Warmuth.

The lesson may be that cancer drugs will not play into the pharmaceutical industry’s previous search for blockbuster drugs. “There’s not some big novel cancer gene that no one had known before, one that induces 50 percent of breast cancer and 65 percent of lung cancer. That’s not what the data tells us,” says Warmuth. 

So the company plans to develop drugs based on promising targets that may benefit only a small percentage of patients with a given cancer type. But another emerging theme in cancer discovery may increase the number of potential patients for any of H3’s potential drugs: tumors in very different parts of the body can share some of the same genetic and molecular abnormalities. H3’s lead target mutation can be found in some skin, breast, and blood cell cancers, with frequencies ranging from 3 percent to 10 percent for each.

They may need that time to address another challenge—how to find the right patients to test its drugs. Many drug development trials in the past are thought to have failed because too many patients in the trial could not respond to the treatment. “Those patients would never respond to the drug because they don’t actually have the genes that would respond,” says Kevin Dalby, a medicinal chemist at the University of Texas at Austin. “So by knowing what you are looking for, you can choose your population much more astutely, so you would have a much better chance of having a successful clinical trial.”

Warmuth thinks that the dropping cost of DNA sequencing will provide an opportunity to solve the patient problem. Rather than enrolling patients in a clinical trial and then determining whether they carry the particular mutation targeted by a candidate drug, patients could have their tumor’s genome sequenced and then share the information in a database available to pharmaceutical companies. “Then you could more proactively approach patients in the future, and ask them if they are willing to become part of a clinical trial as drugs are being developed for particular genetic [changes].”

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.