Blood Test for Multiple Cancers ― Too Many False Positives?

Roxanne Nelson and Zosia Chustecka

June 10, 2021

New data on the Galleri blood test (from Grail) show that it detected a cancer signal for a broad range of cancer types and that more than half of the cancers that were detected were of early stage, when cancer generally is more treatable.

The new results come from the PATHFINDER Trial and were reported at the annual meeting of the American Society of Clinical Oncology.

However, an expert not involved in the trial pointed out that "there were more false positives than true positives, which could be a problem if rolling this out to a whole population setting."

Last month, Grail announced that the test was about to make its clinical debut in the United States, and Providence Health System is now offering the test on prescription.

New Results

The PATHFINDER Trial enrolled 6662 participants older than 50 years who had varying risk factors but for whom there was no suspicion of active cancer. Of this group, 6629 were evaluable.

At the meeting, an interim analysis of the results was presented by Tomasz M. Beer, MD, deputy director at the Knight Cancer Institute, Oregon Health and Sciences University, Portland, Oregon.

A cancer signal was detected in 92 participants (1.4%); of these, diagnostic resolution had been achieved for 65 as of March 2021, he reported.

Among these 65 participants, true positive results were found for 29 patients, and a cancer diagnosis was confirmed. For 36 patients, false positive results occurred.

"Thirty-six participants are classified as false positives based on the initial diagnostic workup, and their final status will be defined after the full 12 month follow-up," said Beer.

"Of the 63 patients with diagnostic workup triggered by a multi cancer early detection test, 90% had at least one imaging test," he said.

The median number of imaging tests was 1.26, and 88% of patients underwent minimally invasive procedures, he noted.

It was these results that were concerning to the discussant of the study, Max Diehn MD, PhD, associate professor of radiation oncology at Stanford University, Stanford, California.

He highlighted the fact that there were more false positives than real positives and noted that "there were a significant number of invasive procedures in false positives which could cause harm to these patients who don't have cancer."

There were more false positives than true positives.  Dr Max Diehn

In addition, Diehn pointed out that the majority of true positives were for lymphoid malignancies, not solid tumors, and it is not known whether the early detection of lymphoid malignancy has clinical utility. "The majority of tumors were lymphoid, and since we usually think of those as systemic diseases at diagnosis, it's not clear if discovering these earlier will have a survival benefit," he said.

Detection Rate

In his presentation, Beer said that at this interim analysis, the overall cancer detection rate was 1.4%, and the overall positive predictive value (PPV) for cancer detection was 44.6%.

For those participants who had an additional risk factor for cancer (eg, smoking history, genetic predisposition, or history of cancer more than 3 years in the past), the PPV increased to 57.1%, but for participants without these risk factors, the PPV fell to 30%. However, Beer pointed out that the study was not designed for a statistical comparison between these two cohorts.

"Accuracy of the cancer signal of our original prediction among true positives was high, at 85% for the first cancer signal of origin and 96.3% for the first or second," he said.

"There were 13 different cancer types diagnosed amongst 28 participants, and of those, 82% of those cancers were de novo, and 18% represent cancer recurrences," he said.

More than half the new cancers that were detected were at early stages, when cancer generally is more readily treatable. Of the participants who received a new cancer diagnosis, 57% were found to have cancer of stage I–III, and 39% had cancer of stage I–II.

Beer concluded by emphasizing that the participants will continue to be followed. "We will identify the incidence of cancer diagnosis for all participants within 12 months of their initial blood draw, at which time the specificity and negative predictive value of the multicancer early detection test will be evaluated," he said.

"The interim results of PATHFINDER demonstrate that a routine blood test is capable of detecting many different cancers even before symptoms arise, an approach that has significant potential advantages," Beer commented in a press release from Grail.

"Most importantly, it can detect cancers that have no recommended screening tests today, and more than two-thirds of cancers go unscreened for this reason. These results are a pivotal step toward extending early detection to many more types of cancer," he added.

Strengths and Limitations

In his discussion of the paper, Diehn noted that the study had both strengths and limitations and that it left some questions unanswered.

"The strengths of this study are that it is large and prospective and shows that real-time application of a multicancer early detection test is feasible," he said. "There was a relatively short time to diagnostic resolution and a significant subset of detected solid tumors that were stage I and II."

Limitations include the fact that follow-up was incomplete and sensitivity and specificity could not be assessed.

Several key questions remain, he said. "Could cases without diagnostic resolution — which accounted for almost a third of cases — be enriched for false positives, and if so, would it decrease the positive predictive value? We await the final results to assess that," he noted.

It is also unknown whether many cancers will be diagnosed among the patients whose test results were negative, he said. "This is important for assessing the sensitivity of the test, particularly important for the early stage solid tumors."

The study was funded by Grail. Beer disclosed the following: stock and other ownership interests – Arvinas, Salarius; consulting or advisory role – Arvinas, Astellas, AstraZeneca, Bayer, Bristol-Myers, Squibb, Clovis, Constellation, GRAIL, Janssen, Myovant, Novartis, Pfizer, Sanofi, Tolero; research funding (inst) – Alliance Foundation Trials, Astellas, Bayer, Boehringer Ingelheim, Corcept, Endocyte, Freenome, GRAIL, Harpoon, Janssen, Medivation, Sotio, Theraclone, Zenith Epigenetics. Diehn disclosed the following: stock and other ownership interests – CiberMed, Foresight Diagnostics; consulting or advisory role – AstraZeneca, BioNTech, Genentech, Gritstone Oncology, Illumina, Novartis, Reflexion, Roche; Patents, Royalties, other intellectual property – patent filings on ctDNA detection assigned to Stanford University (inst); patent filings on tumor treatment resistance mechanisms assigned to Stanford University (inst); travel, accommodations, expenses – AstraZeneca, Roche, Varian Medical Systems.

American Society of Clinical Oncology (ASCO) 2021.

For more from Medscape Oncology, join us on Twitter and Facebook.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.