Detection and Localization of Solid Tumors Utilizing the Cancer-Type-Specific Mutational Signatures

Area of Science:

• Oncology
• Genomics
• Biomarker Discovery

Background:

• Accurate tumor detection and origin identification are critical for effective cancer diagnosis and personalized therapy.
• Current methods for diagnosing lesions with unclear histology rely heavily on experience, leading to low accuracy and efficiency.
• Developing objective and reliable biomarkers for tumor origin tracing is a significant unmet need.

Purpose of the Study:

• To develop and validate a logistic regression model utilizing mutational signatures (MS) for accurate cancer-type prediction and tumor origin tracing.
• To identify cancer-type-specific mutational signature patterns (CTS-MS) and evaluate their diagnostic performance across diverse solid tumors.
• To assess the utility of CTS-MS in identifying the tissue-of-origin for metastatic cancers and improving cancer detection from cell-free DNA (cfDNA).

Main Methods:

• Development of a logistic regression model based on mutational signatures (MS) for each cancer type.
• Training and validation using extensive datasets from ten tumor types (7,581 samples total).
• Analysis of somatic mutations and CTS-MS in cell-free DNA (cfDNA) for improved cancer-type prediction.

Main Results:

• Mutational signatures (MS) effectively distinguished cancer from inflammation and healthy samples.
• Cancer-type-specific MS patterns (CTS-MS) demonstrated robust performance in distinguishing primary and metastatic solid tumors (AUC: 0.76–0.93).
• The model achieved high accuracy (90%) in distinguishing breast and prostate cancers using cfDNA and CTS-MS, improving prediction accuracy.

Conclusions:

• Mutational signatures (MS) represent a novel and reliable biomarker for solid tumor diagnosis and tissue-of-origin prediction.
• Cancer-type-specific MS patterns (CTS-MS) offer a powerful tool for identifying primary and metastatic cancers.
• Combining cfDNA analysis with CTS-MS enhances the accuracy of cancer-type prediction, paving the way for non-invasive diagnostics.