Beware Liquid Biopsies to Guide PARP Blockade

In August, the FDA approved the first liquid biopsy test to guide the use of PARP inhibitors in men with advanced prostate cancer. Now, a report has raised concerns about the potential for false positives due to interference from clonal hematopoiesis—which could lead to unsuitable treatment decisions (JAMA Oncol 2020 Nov 5 [Epub ahead of print]).

Clonal hematopoiesis of indeterminate potential (CHIP) occurs when blood cells share an acquired genetic variant that gives them a competitive advantage. Although these cells are not malignant, the mutations they carry can occur in cancer-associated genes, including those linked to DNA repair or tumor suppression.

Because mutations from both clonally expanded blood cells and tumor cells are found in cell-free DNA (cfDNA), CHIP-related variants can confound liquid biopsy tests performed on blood plasma samples—an approach commonly used by blood-based genomic profiling tests such as FoundationOne Liquid CDx (Foundation Medicine) and Guardant360 CDx (Guardant Health).

On November 9, the FDA approved Foundation's liquid biopsy test as a companion diagnostic to determine whether men with prostate cancer should receive the PARP inhibitor olaparib (Lynparza; AstraZeneca); the test was approved in August to guide the use of another PARP blocker, rucaparib (Rubraca; Clovis Oncology). Guardant's assay is approved only for general tumor profiling, although prostate cancer specialists routinely use the test to select candidates for PARP blockade.

Yet recent findings suggest that such tests should be interpreted with caution. Colin Pritchard, MD, PhD, of the Fred Hutchinson Cancer Center in Seattle, WA, led an analysis of blood plasma cfDNA from 69 men with advanced prostate cancer. In 20 patients, he and his colleagues identified a total of 23 pathogenic variants in DNA-repair genes—including in BRCA1/2, ATM, and CHEK2. These variants would suggest the 20 men are good candidates for PARP inhibition.

However, when the researchers ran the same DNA-sequencing assay on whole blood samples—a strategy that can distinguish between CHIP-related variants and those shed by tumors—they found that eight of the variants arose via clonal hematopoiesis rather than from a tumor. Thus, only 65% of the DNA repair–gene variants detected were true positives, and only 14 of the 20 patients had tumor mutations that would warrant treatment with a PARP inhibitor.

“This is a sort of warning,” says Alexander Wyatt, DPhil, of the Vancouver Prostate Centre in Canada, who was not involved in the study. “We should be addressing these issues and saying, ‘If we want to use this testing modality, we need to find a way of doing it properly.’”

One possible solution is analyzing plasma cfDNA alongside paired whole-blood or peripheral blood mononuclear cell DNA. Studying matched samples introduces extra cost and some logistic complexity but “really does add so much value,” says Pritchard, adding that “it's the right thing to do.”

“We can't be having substantial numbers of false positives for this type of testing because there's no recourse,” he says. “Liquid biopsy tests are often used as the sole biomarker to make a pretty important treatment decision, so we've got to get this right.”

The problem of CHIP-related false positives is not restricted to prostate cancer. A recent study of plasma and corresponding white blood cell samples from 332 patients with late-stage non–small cell lung cancer revealed that 72% had at least one CHIP mutation (JCO Precis Oncol 2020;4:1271–9). “I think there are going to be other reports describing CHIP interference,” says Joshi Alumkal, MD, of the University of Michigan Rogel Cancer Center in Ann Arbor. “This may be the tip of the iceberg in terms of genes that are used for therapy selection that could be CHIP-clone driven.” –Elie Dolgin

For more news on cancer research, visit Cancer Discovery online at http://cancerdiscovery.aacrjournals.org/CDNews.