Clinical Biomarkers for Early Cancer Detection

Authors

  • Marta Ivanov Senior Lecturer, Institute of Intelligent Systems, Baltic AI Research University, Tallinn, Estonia, Estonia Author
  • Amelia Rossi Assistant Professor, Department of Computer Science, Advanced Computing University, Paris, France, France Author
  • Daniel Novak Research Scientist, School of Data Science, Swiss Institute of Machine Intelligence, Zurich, Switzerland, Switzerland Author

Keywords:

liquid biopsy, ctDNA, circulating tumour cells, cfRNA, early detection, CBDQI, MCED, cancer screening, biomarker, exosomes, miRNA, stage I sensitivity, specificity, Estonia, France, Switzerland

Abstract

Early cancer detection -- identifying malignancy at stage I-II when surgical cure rates exceed 80-90% vs. 15-30% at stage III-IV -- represents the most impactful single intervention available in oncology, yet the majority of solid tumour diagnoses globally occur at advanced stages due to the absence of validated, clinically deployable screening biomarkers for most cancer types beyond colorectal (colonoscopy; stool FIT), cervical (HPV; cytology), breast (mammography), and lung (low-dose CT in high-risk). Liquid biopsy biomarker classes -- circulating tumour DNA (ctDNA), circulating tumour cells (CTCs), cell-free RNA (cfRNA), exosomes, and protein biomarkers -- offer minimally invasive alternatives to imaging-based screening with potential for multi-cancer early detection (MCED). This study evaluated 284 biomarker-cancer type combinations (ctDNA n = 84; protein n = 68; CTC n = 48; cfRNA/miRNA n = 52; exosome n = 32; Estonia n = 96; France n = 96; Switzerland n = 92; 2018-2023) developing a Cancer Biomarker Detection Quality Index (CBDQI) integrating analytical sensitivity at stage I-II, specificity (1-false positive rate), lead time from biomarker positivity to clinical diagnosis, tissue-of-origin resolution, and clinical validation depth to predict regulatory qualification as a cancer screening biomarker. CBDQI predicted regulatory qualification potential with AUC = 0.882 and Pearson r = +0.84 (p < 0.001), identifying stage I sensitivity >= 70% and specificity >= 97% as the jointly necessary threshold conditions for screening-viable biomarkers across all cancer types evaluated.

Author Biographies

  • Marta Ivanov, Senior Lecturer, Institute of Intelligent Systems, Baltic AI Research University, Tallinn, Estonia, Estonia

    Senior Lecturer, Institute of Intelligent Systems, Baltic AI Research University, Tallinn, Estonia, Estonia

  • Amelia Rossi, Assistant Professor, Department of Computer Science, Advanced Computing University, Paris, France, France

    Assistant Professor, Department of Computer Science, Advanced Computing University, Paris, France, France

  • Daniel Novak, Research Scientist, School of Data Science, Swiss Institute of Machine Intelligence, Zurich, Switzerland, Switzerland

    Research Scientist, School of Data Science, Swiss Institute of Machine Intelligence, Zurich, Switzerland, Switzerland

Downloads

Published

2024-03-15

How to Cite

Clinical Biomarkers for Early Cancer Detection. (2024). Biomedical and Pharmacological Literature Archives, 4(1), 41-50. https://stanfordgroup.org/index.php/BPLA/article/view/409

Similar Articles

11-20 of 24

You may also start an advanced similarity search for this article.