New Method Facilitates Early Detection and Localization of Tumors

To help with early tumor diagnosis and tumor localization, researchers have devised a technology that concurrently detects many cancer-specific DNA circulating hallmarks.

The study, published as a Reviewed Preprint in eLife, offers a thorough evaluation of a recently created predictive model for identifying cancer and identifying the tissue of origin.

Using an extensive set of patient samples to design the test and a different collection of patient samples to validate it, the authors offer strong support and shed light on the early diagnosis of cancer with non-invasive approaches.

Since it contains epigenetic and genetic modifications from the original tumor DNA, circulating tumor DNA (ctDNA) in the blood has emerged as a viable biomarker for diagnosing early-stage cancers.

It is feasible to identify a corresponding unknown primary cancer and its location by detecting distinct patterns of these modifications that are unique to tumor types and originating tissue.

Existing diagnostics, however, lack accuracy in detecting the source of tumor DNA and/or require in-depth gene sequencing, which is too expensive for wider population screening.

Despite their great potential, there remain several challenges that ctDNA tests must solve to deliver accessible and reliable clinical adoption for the general population, including the low amount of ctDNA in the blood of early-stage cancer patients, and the diversity in ctDNA from different cancer types, subtypes and stages.

Van Thien Chi Nguyen, Study Co-Lead Author and Research Associate, Medical Genetics Institute

“Recent studies have focused on a multi-analyte approach—combining genomic and protein features—but these are also time-consuming and costly. We set out to develop a multi-modal approach that uses several properties of ctDNA to detect and locate cancer from a single screening test,” added study co-leader author Trong Hieu Nguyen, who is a senior data scientists and PhD at Gene Solutions and the Medical Genetics Institute.

The SPOT-MAS method (Screening for the Presence of Tumor by DNA Methylation and Size) combines shallow whole-genome sequencing with targeted high-depth gene sequencing to uncover a variety of ctDNA characteristics.

By comparing the DNA of tumors with healthy DNA, it can identify the methylation patterns, DNA copy number, size of ctDNA fragments, and sequence motifs at the ends of the fragments.

This method was utilized by the researchers to analyze samples from 738 patients with breast, colorectal, liver, lung, and stomach cancer as well as 1,550 healthy individuals. The data were then entered into an algorithm to find the ctDNA signatures for each of the cancer types.

The method was then tested on a new set of 239 individuals, each of whom had one of the five cancer types, to evaluate how well it did at identifying the presence and type of cancer.

They discovered that SPOT-MAS accurately identified 73% of cancers while only employing shallow-depth sequencing. Furthermore, the test had a 70% accuracy rate in predicting the tissue of origin.

When they examined models with only one feature, such as concentrating solely on DNA copy numbers, they discovered that models with many characteristics outperformed those with just one feature.

The test was least effective for breast cancer, where it only caught about half of instances, and most successful for liver cancer, where it recognized over 90% of cases. This is consistent with earlier research that indicates liver tumors shed more ctDNA than breast cancers, which shed less.

More study is needed to compare this novel methodology to existing ctDNA methods to demonstrate that it performs better and is less expensive than testing utilizing single features alone. It will also be required to examine the test in larger numbers of people who have cancer at an earlier stage than those in the current study.

Our work demonstrates that SPOT-MAS, with its unique combination of ctDNA feature and innovative machine learning algorithms, can successfully detect and localize multiple types of cancer through low-cost sequencing techniques. A large, multi-center prospective study with several years’ follow-up is currently underway to fully validate the performance of SPOT-MAS as an early cancer screening test before it’s considered for use in a wider clinical setting.

Le Son Tran, Study Senior Author and Principal Investigator, Gene Solutions