Mammograms remain the backbone of breast cancer screening, but they struggle when breast tissue is dense and they say nothing about whether a treated tumor will come back. Scientists in Israel now report an experimental blood test that tackles both problems, with results published in Nature this week.
Biologist Efrat Shema of Rehovot’s Weizmann Institute led the project together with physicist Guy Ron of Hebrew University in Jerusalem, and doctors from several Israeli hospitals joined in. Rather than hunting for mutations, the team reads epigenetic marks, the chemical tags that tell a cell which genes to switch on. Whenever a cell dies, tumor cells included, it releases nucleosomes, tiny spools of genetic material wound on a protein core, into the blood. Using single-molecule fluorescence microscopy, the researchers scanned millions of these spools for dozens of tags at once and let machine-learning models search the data for a tumor’s fingerprint.
For the main trial the researchers compared blood drawn from 77 women with breast cancer, each monitored for half a decade or more by doctors in Jacksonville, with blood from 30 cancer-free volunteers. Tumors that had not yet spread beyond the breast and measured two centimeters or less were identified with 95 percent accuracy. Performance held up regardless of breast density and across tumor types, including in an expanded set of 188 patient samples.
The study also uncovered a small, highly aggressive subpopulation of cells hiding inside tumors that otherwise looked fairly tame. The more of its signature showed up in a patient’s blood, the likelier a relapse. Combined with other markers, that allowed the team to separate women whose cancer returned from those who stayed well with 81 percent accuracy. This is important: even after apparently successful treatment, somewhere between a fifth and two fifths of patients see the cancer return.
The test is not yet approved and needs larger prospective trials. Shema’s next aim is therapies that rob tumors of the ability to reshape their epigenetic profile.
