New Blood Test Combines DNA Sequencing and ML for Early Cancer Detection

Researchers at the Johns Hopkins Kimmel Cancer Center have developed a novel blood testing technology called GEMINI (Genome-wide Mutational Incidence for Non-Invasive detection of cancer), which combines genome-wide sequencing of single molecules of DNA with machine learning to enable earlier detection of lung and other cancers.

GEMINI, which was described in a Nature Genetics paper,  involves collecting a blood sample from individuals at risk of developing cancer and extracting cell-free DNA (cfDNA) shed by tumors from the plasma. The cfDNA is then sequenced using cost-efficient whole genome sequencing, and single DNA molecules are analyzed for sequence alterations, providing mutation profiles across the genome. A machine learning model is employed to distinguish individuals with cancer from those without by identifying changes in cancer and non-cancer mutation frequencies in different regions of the genome. The classifier generates a score ranging from 0 to 1, with a higher score indicating a higher probability of having cancer.

In laboratory tests, GEMINI combined with computerized tomography imaging detected over 90% of lung cancers, even in patients with stage I and II disease. The research demonstrates the potential of GEMINI for early cancer detection and monitoring patients during therapy.

The study mainly focused on detecting lung cancer in high-risk populations, but altered mutational profiles were also detected in cfDNA from patients with other cancers, such as liver cancer, melanoma, or lymphoma, suggesting broader applicability.