Before a surgeon's first incision, a patient's blood already holds the story of what may come next. Researchers from KAIST and two major South Korean medical centers have found that colorectal cancer rewires the body's entire amino acid metabolism in ways that are readable in a single preoperative blood sample — not by measuring one marker in isolation, but by mapping how eighteen amino acids speak to one another. Published in Advanced Science in June 2026, the work suggests that prognosis, like most things worth understanding, is less about any single fact than about the relationships between
Blood amino acid network patterns predict colorectal cancer recurrence better than current tests
The amino acid network is a map of that rewiring
Mark
Why does it matter that amino acids interact with each other? Couldn't you just measure them individually and get the same information?
Mimi
You could measure them individually, and doctors have been doing that for years. But a network tells you something different—it tells you how the body is allocating resources. When glycine goes up even though cancer cells are consuming it, that's not just a number. That's evidence the whole system is being reprogrammed.
Mark
So the cancer is changing the body's metabolism, not just growing in one place.
Mimi
Exactly. The tumor is like a factory demanding raw materials. The body responds by shifting production. Those shifts show up in the blood as a pattern. A single amino acid level is like looking at one worker. The network is like seeing the whole assembly line reorganize.
Mark
And the machine-learning model can spot that reorganization better than the old CEA test?
Mimi
Yes. CEA has been the standard for decades, but it's blunt. The amino acid network is more granular. It captures the metabolic signature of advancing disease. When you combine both—CEA plus the network—you get the best prediction.
Mark
What happens next? Does this go straight into hospitals?
Mimi
Not immediately. It needs validation in larger patient populations, and the test itself needs to be standardized for clinical use. But the proof of concept is there. If it works in practice the way it works in the research, it could change how doctors decide who needs aggressive follow-up after surgery.
Der Puls
- Colorectal cancer silently reshapes the body's metabolic landscape long before recurrence becomes visible, leaving patients and doctors with imprecise tools to anticipate what comes after surgery.
- The standard CEA blood test, relied upon for decades, measures only one voice in a complex biochemical conversation — a limitation this research now exposes with striking clarity.
- A team of graduate students and professors turned to network science, using fluorine-19 NMR spectroscopy to map not just amino acid levels but their interactions, catching a systemic metabolic remodeling that single-marker tests cannot see.
- Branched-chain amino acids that fuel healthy muscle decline as disease advances, while glycine — consumed voraciously by tumor cells yet paradoxically rising in the blood — emerges as a key signal of systemic disruption.
- Machine-learning models trained on these interaction networks outperformed CEA alone, and combining both approaches pushed predictive accuracy higher still, pointing toward a clinically actionable tool.
- The path forward is translation: moving this network-based blood test from the research lab into surgical wards, where it could give patients a more honest picture of their risk and open the door to earlier, more personalized intervention.
Before a surgeon's first incision, a patient's blood already holds the story of what may come next. Researchers from KAIST and two major South Korean medical centers have found that colorectal cancer rewires the body's entire amino acid metabolism in ways that are readable in a single preoperative blood sample — not by measuring one marker in isolation, but by mapping how eighteen amino acids speak to one another. Published in Advanced Science in June 2026, the work suggests that prognosis, like most things worth understanding, is less about any single fact than about the relationships between facts.
Before surgery for colorectal cancer, a patient gives blood. In a laboratory, researchers measure eighteen circulating amino acids — not merely their quantities, but the way they interact with one another, the network of relationships that emerges from their collective behavior. That network, it turns out, tells a more complete story than any single number can.
A team from KAIST, Gangnam Severance Hospital, and Asan Medical Center has found that as colorectal cancer advances, it systematically reshapes the body's amino acid metabolism in ways that are visible in the bloodstream. Tumor cells are metabolically voracious, consuming amino acids for growth and energy, and the body reorganizes itself in response. Branched-chain amino acids like valine and leucine — which support muscle and energy metabolism — decline. Glycine and serine, which cancer cells need to proliferate and synthesize DNA, rise. Glycine's increase is paradoxical: even as tumors consume it rapidly, its blood concentration climbs, and a new interaction pattern forms around it. This contradiction is itself evidence of deep systemic remodeling.
Using fluorine-19 nuclear magnetic resonance spectroscopy, the researchers fed these interaction patterns into machine-learning models designed to identify patients at risk for recurrence or metastasis. The amino acid network model outperformed the long-established CEA test, and combining both approaches improved predictive accuracy further still. The interaction data captured something that individual measurements, taken in isolation, had always missed.
The project began as an interdisciplinary proposal from graduate students who saw cancer progression through a network lens rather than a reductionist one. Professor Ji Min Lee frames the result as a step toward precision medicine: a single preoperative blood draw that could tell a patient — and their surgeon — more accurately what lies ahead, and make targeted intervention possible. The research, published in Advanced Science in June 2026, points toward a future where prognosis is not a single number but a map of how the body's chemistry has been rewired.
Before surgery for colorectal cancer, a patient gives blood. The sample sits in a lab. Researchers run it through a machine that measures eighteen different amino acids circulating in the bloodstream—not just how much of each one is present, but how they interact with one another, the way they talk to each other in the body's chemistry. This network map, it turns out, tells a story that a single number cannot.
A team from KAIST, Gangnam Severance Hospital, and Asan Medical Center has discovered that as colorectal cancer advances, the amino acid network undergoes a systematic reshaping. The finding matters because it offers a way to predict which patients will face recurrence or metastasis—and it does so more reliably than the tests doctors currently rely on. The research, published in June 2026 in the journal Advanced Science, suggests that the future of cancer prognosis may lie not in measuring individual markers in isolation, but in understanding how they move and change together.
Cancer is metabolically voracious. Tumor cells need amino acids—the building blocks of protein—for growth and energy production, and they need them in vast quantities. Colorectal cancer in particular leaves a distinctive metabolic fingerprint. The body's amino acid metabolism shifts in response to the tumor's demands. These changes don't stay confined to the tumor itself; they leak into the bloodstream, making circulating amino acids a window into what is happening systemically.
Until now, researchers had focused mainly on measuring individual amino acids one at a time. The new work asks a different question: What if you looked at the relationships among them? Using fluorine-19 nuclear magnetic resonance spectroscopy, the team measured eighteen amino acids in a small serum sample and then analyzed not just their abundance but their interconnections. As colorectal cancer progressed, a clear pattern emerged. Branched-chain amino acids—valine and leucine, which fuel muscle and energy metabolism—declined in proportion. Meanwhile, glycine and serine, which cancer cells need to synthesize DNA and proliferate, increased. The shift reflects a fundamental reprogramming of how the body allocates its amino acids as disease advances. Glycine was particularly striking: even though rapidly dividing cancer cells consume it voraciously, its concentration in the blood actually rose, and a new interaction pattern centered on glycine appeared. This paradox itself became evidence of systemic metabolic remodeling.
The researchers then fed these amino acid interaction patterns into machine-learning models designed to identify patients at risk for recurrence or metastasis. The results were clear. A model based solely on the amino acid interaction network outperformed the standard CEA test—a marker that has been used in colorectal cancer care for decades. When the team combined the interaction data with CEA levels and individual amino acid measurements, the predictive power climbed higher still. The interaction network, in other words, captured something that individual measurements missed.
The work began as an interdisciplinary proposal from graduate students in medical science and chemistry who saw an opportunity to study cancer progression through a network lens rather than a reductionist one. Professor Ji Min Lee, who led the effort, frames the finding as a step toward precision medicine—the ability to predict recurrence risk from a single blood draw and tailor treatment accordingly. For a patient facing surgery, the implications are tangible: a more accurate sense of what lies ahead, and the possibility of more targeted intervention.
What the research reveals is that cancer doesn't just change the tumor. It rewires the body's metabolism in ways that are legible in the blood. The amino acid network is a map of that rewiring. Reading it carefully—understanding not just what is there but how it all connects—offers a clearer picture of disease state and risk than any single marker can provide. The next step is moving this from the research setting into clinical practice, where it could change how doctors assess and treat colorectal cancer patients.
Bemerkenswerte Zitate
We hope this will lead to new precision medicine technologies that can predict recurrence risk more accurately using a blood sample alone and help establish personalized treatment strategies.— Professor Ji Min Lee, KAIST
Möchten Sie die ganze Geschichte?
Das Original lesen bei News-Medical ↗
Health & Medicine