An intratumorally discovered extracellular vesicle-derived biomarker enables ultrasensitive digital quantification in plasma for early detection of non-small cell lung cancer

Heng Huang, Taketo Kato, Yuichi Abe, Akira Yokoi, Masami Kitagawa, Eri Asano‐Inami, Yoshito Imamura, Yuji Nomata, Hiroki Watanabe, Hirofumi Takenaka, Taiki Ryo, Yuta Kawasumi, Keita Nakanishi, Yuka Kadomatsu, Harushi Ueno, Shota Nakamura, Tetsuya Mizuno, Ayumu Taguchi, Toyofumi F. Chen‐YoshikawamedRxiv 2025 · DOI

TL;DR

Researchers have identified GLUT1—a protein involved in glucose transport—as a highly specific biomarker on the surface of tumor-derived extracellular vesicles (EVs). By combining this discovery with an ultrasensitive digital detection technique, they achieved a diagnostic accuracy (AUC) of 0.944 for detecting Stage I lung cancer, significantly outperforming current clinical blood tests.

What Problem Does This Paper Address?

Non-small cell lung cancer (NSCLC) is the leading cause of cancer-related deaths worldwide, largely because it is often diagnosed too late. While tissue biopsies are the gold standard, they are invasive and difficult to perform repeatedly. "Liquid biopsies"—tests that look for cancer signals in the blood—offer a minimally invasive alternative.

Extracellular vesicles (EVs) are tiny "bubbles" released by cells into the bloodstream, carrying proteins and genetic material from their parent tumor. However, using them for early detection is incredibly difficult: tumor-derived EVs are "diluted" by a massive background of healthy EVs, and the specific cancer markers they carry are often present in extremely low concentrations. This paper tackles the dual challenge of biomarker specificity (finding the right target) and detection sensitivity (seeing that target at low levels).

Methods

The researchers employed a "tissue-to-plasma" strategy:

  1. Discovery: They first isolated EVs directly from lung cancer tissues and paired healthy tissues using density gradient flotation. They then used mass spectrometry proteomics to identify proteins that were uniquely overexpressed in the cancer-derived EVs.
  2. Integration: They cross-referenced their findings with public transcriptomic data (TCGA) and the Human Protein Atlas to ensure the candidate markers were both cancer-specific and localized on the cell membrane (making them accessible on the EV surface).
  3. Validation: They used immunogold labeling and transmission electron microscopy (TEM) to visually confirm that their top candidate, GLUT1, was actually present on the EV membrane.
  4. Digital Quantification: To detect these rare signals in patient blood, they used a modified Digital EV Screening Technique (DEST). This method captures EVs on magnetic beads and uses Tyramide Signal Amplification (TSA)—a technique that can boost protein signals up to 10,000-fold—to enable quantification via flow cytometry.

Datasets

  • Nagoya University Hospital Cohort: A primary clinical dataset of 47 patients (38 with NSCLC and 9 non-cancer controls) used for the DEST assay validation.
  • The Cancer Genome Atlas (TCGA): Public transcriptomic data from 973 cancer and 107 non-cancer samples used to filter and rank biomarker candidates.
  • Human Protein Atlas (HPA): Used as a reference set for 2,074 genes encoding plasma membrane-localized proteins.

Key Insights and Ideas

  • The "Tissue-First" Advantage: By looking at EVs isolated directly from tumors first, the team bypassed the "noise" of the bloodstream to find the most potent markers.
  • GLUT1 as a Metabolic Beacon: While GLUT1 is well-known for its role in the "Warburg effect" (cancer's sugar-hungry metabolism), this study is the first to demonstrate its clinical utility as a surface marker on circulating EVs for NSCLC.
  • Subpopulation Specificity: Interestingly, the GLUT1 signal was significantly elevated only in EVs carrying the CD63 marker (typically associated with exosomes), but not in those carrying CD9 or CD81. This suggests that cancer-specific GLUT1 is packaged into very specific types of vesicles.
  • Synergy with Standard Care: The researchers found that their EV-GLUT1 marker didn't just replace existing tests; it complemented them. Combining it with traditional markers like CEA and CYFRA 21-1 created a "three-marker panel" with superior predictive power.

Key Results

  • Superior Accuracy: In the whole cohort, the CD63+ EV-derived GLUT1 marker achieved an Area Under the Curve (AUC) of 0.817.
  • Early Detection Power: For Stage I patients, the accuracy jumped to an AUC of 0.898, outperforming conventional serum biomarkers.
  • The Power of Three: When combined into a panel (GLUT1 + CEA + CYFRA 21-1), the diagnostic performance reached an impressive AUC of 0.944 for Stage I NSCLC.
  • Survival Correlation: High expression of the gene encoding GLUT1 (SLC2A1) was significantly correlated with decreased overall survival in lung cancer patients.

Limitations and Future Work

While promising, the authors acknowledge several hurdles:

  • Sample Size: The study was conducted on a relatively small, single-center cohort (47 patients). Large-scale, multi-center prospective trials are needed to confirm these results.
  • Technical Complexity: The modified DEST assay with TSA is highly sensitive but requires specialized equipment and high-quality antibodies, which may limit its immediate adoption in standard clinical labs.
  • Mechanistic Gaps: The study focuses on the diagnostic utility of GLUT1 on EVs but does not fully explore the biological mechanism of how or why GLUT1 is selectively packaged into these vesicles.

Future research will likely focus on automating the DEST platform and investigating whether this GLUT1-EV signature can also be used to monitor treatment response or detect cancer recurrence.

Summary updated

Abstract

Abstract Background Extracellular vesicle (EV)-based liquid biopsy remains challenging due to the limited specificity and low abundance of EV cargoes. This study aimed to identify cancer-specific molecules from tissue-derived EVs and evaluate their potential as circulating biomarkers for non-small cell lung cancer (NSCLC). Methods Tissue-derived EVs were isolated by density gradient flotation and subjected to proteomic profiling to identify differentially expressed proteins. Public transcriptomic data, functional enrichment, and variable importance scores were integrated to select candidate biomarkers. Immunogold labeling transmission electron microscopy (TEM) was used to visualize protein expression on the EV membrane using, and an ultrasensitive modified digital EV screening technique (DEST) was employed to quantify protein signals in circulating EVs. The diagnostic performance was evaluated by the area under the curves (AUC). Results Tissue-derived EVs were isolated from cancer tissues and paired non-cancer lung tissues from patients with NSCLC. Proteomic profiling identified 1,568 tissue-derived EV proteins, of which 904 were differentially expressed. Integrated transcriptomic and proteomic analyses yielded eight cancer-specific, membrane-localized candidate biomarkers, with GLUT1 showing the highest variable importance score. Immunogold labeling TEM confirmed the localization of GLUT1 on the EV membrane. Modified DEST analysis of 47 patients revealed significantly elevated levels of CD63 + EV-derived GLUT1 in the cancer group than in the non-cancer group ( P < 0.01). Diagnostic performance reached an AUC of 0.817 in the whole cohort and 0.898 in the stage-I cohort, exceeding that of conventional serum biomarkers. A three-marker panel (CD63 + EV-derived GLUT1, CYFRA 21-1, and CEA) further improved discrimination, with AUCs of 0.853 in the whole cohort and 0.944 in the stage-I cohort. Conclusion Circulating EV-derived GLUT1 was preliminarily identified as a novel biomarker for NSCLC, and its combination with conventional serum biomarkers enhanced diagnostic performance, particularly in early-stage cases.
BibTeX
@misc{huang2025intratumorally,
  author = {Heng Huang and Taketo Kato and Yuichi Abe and Akira Yokoi and Masami Kitagawa and Eri Asano‐Inami and Yoshito Imamura and Yuji Nomata and Hiroki Watanabe and Hirofumi Takenaka and Taiki Ryo and Yuta Kawasumi and Keita Nakanishi and Yuka Kadomatsu and Harushi Ueno and Shota Nakamura and Tetsuya Mizuno and Ayumu Taguchi and Toyofumi F. Chen‐Yoshikawa},
  title = {{An intratumorally discovered extracellular vesicle-derived biomarker enables ultrasensitive digital quantification in plasma for early detection of non-small cell lung cancer}},
  year = {2025},
  doi = {10.1101/2025.06.09.25329294},
  howpublished = {medRxiv},
}
Figure