Every tumor has a unique molecular address, like a biological zip code that distinguishes it from healthy tissue and other types of cancer. The problem is that until now, no one had a complete map of these addresses. No one knew precisely where to look in the human body. Natural biomarkers (DNA, proteins) are too few, difficult to find, and often give false alarms because they are also produced by normal cellular activity. Now, however, a team of Georgia Institute of Technology He built the first version of a true tumor atlas, cataloging the specific enzymatic activity of 14 different cancer types. With this map, bioengineered sensors can circulate throughout the body, recognize the tumor's code, and release synthetic markers detectable with standard tests. A single test to detect breast, lung, prostate and eleven other types of cancer.
The project that maps every cancer
When Georgia Tech received a $49,5 million contract from theAdvanced Research Projects Agency for Health (ARPA-H), the goal was clear but risky: to build a tumor atlas capable of guiding multi-cancer testing before tumors become visible on traditional scans. As reported in the official press release, it wasn't guaranteed money. The team led by the bioengineer Gabe Kwong He had to prove that the idea worked, otherwise the funding would stop.
Two years later, they passed the critical threshold. The team built the first instrument capable of measuring enzyme activity around tumors and healthy cells, then used it to map the unique molecular signatures of 14 different cancer types. This tumor atlas is the basis for designing sensors that circulate in the body, recognize the specific “barcode” of a tumor, and release easily detectable markers into the blood.
As Kwong explains:
"If I want to send a sensor to a specific region of the body, there's no way to target it today. We administer it systemically, and it spreads throughout the tissues. The powerful thing is that we're now defining tissue sites with a specific molecular barcode. When a sensor is administered systemically, it should only activate when the barcode matches the local tissue."
Why a tumor atlas was needed
About 20% of people in the world will develop cancer during their lifetime (in America 40%, according to theAmerican Cancer Society). Most are discovered at an advanced stage, when treatment is more difficult, expensive, and less effective. Current screening methods (colonoscopy, mammography, PSA testing) work, but each detects only one type of cancer at a time. And often, they do so only when the tumor is already large enough to be visible.
The hunt for natural biomarkers (circulating tumor DNA, specific proteins) has run into a problem: these substances are present in minute quantities and are also produced by normal cellular activity, generating false positives. Traditional sensors don't know where to look. They activate everywhere, creating background noise instead of clear signals.
The project CODA (Cancer and Organ Degradome Atlas) They changed their approach. Instead of searching for rare molecules in the blood, the team mapped enzyme activity around tumors. Each type of cancer has a unique enzyme profile, like a molecular fingerprint. It's like having a specific address instead of wandering around hoping to bump into someone by chance.
How smart sensors work
In the second phase of the project, the team is finalizing the tumor atlas and testing three different types of sensors. All use "molecular logic" to recognize and respond to tumor cells. It's like a multi-factor authentication system: a single enzyme isn't enough; multiple enzymatic signals at the same location are needed to activate the sensor.
When the sensor recognizes the complete code (the specific combination of enzymes for that tumor), it releases a synthetic marker into the blood. This marker is designed to be easily detectable with standard laboratory tests, without the background noise of natural biomarkers.
Ross Uhrich, program manager at ARPA-H who oversees the CODA project, emphasizes that "preliminary studies in preclinical models are very promising. The team's sensor prototypes already outperform comparable biomarkers on the market in detecting small tumors."
As has already happened with other innovative diagnostic technologies, the goal is to achieve reliable, affordable and accessible tests on a large scale.
An ever-evolving cancer atlas
Kwong collaborates with John Blazeck of School of Chemical and Biomolecular Engineering e Peng Qiu of the Coulter Department of Biomedical Engineering at Georgia Tech. Key partners include bioengineering Danino Valley of Columbia University e Min Xue of the Mount Sinai Health System.
The first version of the tumor atlas includes several models for each type of cancer, demonstrating that the approach works. But the atlas is designed as a living document: as new data arrives, the map expands and becomes more refined. In the future, it will also be available to other researchers who want to develop new cancer screening tools.
“The fundamental premise is that addressing breast cancer is different from addressing lung cancer, which is different from addressing healthy lung tissue,” explains Kwong.
Each fabric has its own signature, like a neighborhood with unique architectural features.
From science fiction to the clinic
Does this all sound incredible? It is. The ARPA-H model is designed to accelerate high-risk healthcare ideas that require massive investment. Kwong estimates that with the traditional research approach, achieving this result would have taken 20-30 years. The CODA project aims to do this in 3-5 years from now: 2028-2030.
“It's a matter of scale and scope,” Kwong says. “With a typical research approach, I'm not sure I'd ever get there. This compresses everything and does the work in three to five years.”
Once the tumor atlas is complete and the sensors validated, the goal is to commercialize multi-cancer tests that can be implemented on a large scale. Tests that can replace mammograms, colonoscopies, and PSA. with a single blood test capable of detecting 14 different types of cancer when they are still small and curable.
The next time a sensor circulates in the body and recognizes the molecular code of an invisible tumor, it won't be luck. It'll be because it had a map.
And he knew exactly where to look.
