Cross-Platform Methylation-Based Site of Origin Classification for Squamous Cell Carcinomas

Area of Science:

• Oncology
• Genomics
• Molecular Pathology

Background:

• Squamous cell carcinomas (SCCs) are common and can metastasize widely.
• Distinguishing primary SCCs from metastases is clinically challenging due to overlapping features.
• Accurate site-of-origin determination is critical for appropriate patient management.

Purpose of the Study:

• To develop and validate a methylation-based classifier (SquaMOS) for predicting the site-of-origin of squamous-appearing carcinomas.
• To assess the accuracy of SquaMOS on primary and metastatic tumors across diverse data sources.
• To evaluate the applicability of SquaMOS to low-depth Nanopore sequencing data.

Main Methods:

• Developed SquaMOS using array-based methylation data from 1062 primary SCCs and urothelial carcinomas.
• Tested SquaMOS on internal (n=458) and external (n=78) primary tumor datasets.
• Validated SquaMOS performance on metastatic tumors (n=51) and shallow Nanopore sequencing data (n=36).

Main Results:

• SquaMOS achieved high accuracy in predicting the site-of-origin for primary tumors (96.1% internal, 97.4% external).
• The classifier demonstrated 96.1% accuracy on metastatic SCCs.
• SquaMOS showed 91.7% accuracy on low-depth Nanopore sequencing data, with 100% accuracy for high-confidence predictions.

Conclusions:

• SquaMOS is a highly accurate methylation-based classifier for determining the site-of-origin of squamous-appearing carcinomas.
• The classifier performs robustly on primary and metastatic tumors and is transferable to rapid sequencing technologies.
• SquaMOS offers a promising tool for clinical settings to guide treatment decisions for SCC patients.