Research from the Korea Advanced Institute of Science and Technology (KAIST), Gangnam Severance Hospital, and Asan Medical Center suggests that a single pre-surgery blood sample could help identify colorectal cancer patients at higher risk of recurrence or metastasis.
Colorectal cancer alters the body’s metabolism, detectable in blood samples. Instead of analyzing individual amino acid levels, scientists investigated the interactions between them, forming a metabolic “network.” This approach may offer better insights into cancer progression than current methods focusing on individual amino acid measurements.
Cancer cells consume many nutrients for growth and division, relying heavily on amino acids for protein building, energy production, and DNA synthesis. While previous studies focused on amino acid concentrations, recent research examined the relationships among 18 circulating amino acids in small serum samples.
Using fluorine-19 nuclear magnetic resonance spectroscopy, the researchers discovered that amino acid networks became reorganized as colorectal cancer advanced. Metabolic changes included decreased prominence of branched-chain amino acids like valine and leucine, critical for muscle and energy metabolism, while glycine and serine, vital for DNA synthesis and rapid growth, became more prominent.
Notably, glycine, actively consumed by cancer cells, showed increased abundance in the bloodstream as the disease progressed. A glycine-centered interaction pattern in the amino acid network suggested widespread metabolic remodeling.
The team used these interaction features in machine-learning models to predict patient recurrence or metastasis. Models based on amino acid interactions outperformed those using only the carcinoembryonic antigen (CEA), a standard blood marker in colorectal cancer. Combining CEA, individual amino acid levels, and interaction-derived features resulted in the best performance, surpassing models based solely on individual amino acid measurements.
Professor Ji Min Lee, leading the study, expressed hope the methodology could enhance treatment precision. “New precision medicine technologies may emerge, predicting recurrence risk accurately via blood samples and tailoring treatment strategies,” Lee said.
External experts find the results intriguing, although further validation is necessary before influencing patient treatment. Dr. Michael F. Driscoll from Norton Healthcare noted the need for larger studies, given the small sample size, and highlighted challenges in adopting blood-based biomarker tests due to insurance coverage classifications.
Currently, many gastrointestinal oncologists rely on circulating tumor DNA (CT-DNA) after surgery to assess recurrence risk, as it remains the most sensitive predictor, validated in extensive studies.
These findings, while not translating to a ready-to-use clinical test, suggest that circulating amino acid interactions might serve as promising biomarkers for tracking cancer progression and identifying high-risk patients.
Study co-authors include Ji-Yeon Lee and Dr. Jumi Kim, with Professors Ji Min Lee, Hyunwoo Kim, and Eun Jung Park as co-corresponding authors. Findings were published in Advanced Science on June 9, 2026.