Journal Club: Shield MCD (Guardant)

A note on sourcing: This journal club is based on a conference presentation, company press releases, FDA regulatory filings, and the National Cancer Institute’s public disclosures about the Vanguard Study. No peer-reviewed manuscripts evaluating Shield MCD have been published at the time of this book.

Study Design

In April 2025, Guardant presented clinical validation data for its MCED test (Shield MCD) at AACR. The presentation evaluated samples from 778 participants in a blinded case-control design (similar to CCGA and DETECT-A). Of these 778 participants, 375 had a known cancer diagnosis, while 403 did not have cancer. Participants ranged in age from 40 to 78 (median 62), 55% were female, and 79% were white. The performance of the test was studied in a prespecified group of cancer types, which were bladder, breast, colorectal, esophageal, gastric (combined with esophageal in reporting), liver, lung, ovarian, pancreatic, and prostate.

Results

At 98.5% specificity, overall sensitivity was 59.7% across all ten cancer types, and sensitivity for the six “most aggressive” cancers (as stated by the investigators), including esophageal/stomach, liver, lung, ovarian, pancreas, and prostate, was 74%. For stage I-II cancers, sensitivity was 34.6% while it was 84.2% for stage III-IV cancers. Sensitivity for the cancers without guideline-recommended screening was 73%. Per-cancer sensitivity ranged from 96% for esophageal/gastric cancer to 21% for prostate cancer. The cancer signal origin accuracy for the top two most likely organs of cancer origin was 89%.

What I Think

The good news:

The Shield MCD data are encouraging as a foundation for future validation. The cancer-type sensitivity pattern aligns with known cfDNA shedding biology (cancer types of the gastrointestinal system, like esophageal/gastric, were at the top).

Although direct comparison between Shield MCD and other tests is tempting, it cannot be done reliably, given that the tests were evaluated in different types of studies (case-control versus prospective interventional in PATHFINDER and PATHFINDER 2). That said, I believe some high-level observations are fair. Shield MCD's overall sensitivity of ~60% and its sensitivity of ~75% for the "most aggressive" cancers are in a range that is broadly comparable to what GRAIL reported in CCGA (their case-control study) for Galleri and what Exact Sciences reported in their presentations and clinician brochure for Cancerguard, as is the 89% cancer signal origin accuracy.

Several limitations exist.

As previously mentioned, Shield MCD was studied in a case-control study (similar to GRAIL’s CCGA study and Exact Science’s DETECT-A study) and not a prospective interventional trial. This study was not meant to determine performance metrics in a screening population. Typically, MCED tests that have moved from case-control to prospective screening have seen performance metrics change.

These data are from conference presentations and company press releases, and, therefore, the study has not undergone peer review or been published in a journal. The full dataset, including subgroup analyses and detailed methodology, is not yet available for independent evaluation.

As with other tests described in this book, Shield MCD had higher sensitivity for advanced-stage cancers (84.2%) than early-stage cancers (34.7%), which goes against the premise that the tests can be used reliably to identify early-stage cancers so they can be treated with curative intent.

What Came After

FDA Breakthrough Device Designation

In June 2025, the FDA granted Breakthrough Device Designation to Shield MCD for screening of bladder, colorectal, esophageal, gastric, liver, lung, ovarian, and pancreatic cancer (the cancer types evaluated in the 2025 AACR presentation) in individuals aged 45 or older. Breakthrough Device Designation means the FDA has agreed that the device has the potential to provide more effective diagnosis of life-threatening diseases than current options, and that the agency will provide expedited development, assessment, and review.

Real-World Data Initiative

In Q3 2025, Guardant announced that clinicians who order a Shield colorectal cancer screening test can opt in to receive a multi-cancer report. In Guardant's Q3 2025 earnings call, they reported real-world data from 9,251 participants showing specificity of 99% and a PPV of 41% for Shield MCD (although these data have not been presented at a conference or published in a medical journal as of the time of this book).

National Cancer Institute Vanguard Study

In January 2025, the National Cancer Institute announced that Shield MCD had been selected for the Vanguard Study, a four-year pilot study designed to enroll up to 24,000 participants and evaluate the feasibility of using MCED tests in future, large, randomized controlled trials. The Vanguard Study is run by the Cancer Screening Research Network, a new National Cancer Institute-sponsored clinical trial network. The study is randomized and has three arms, all of which are encouraged to receive standard-of-care cancer screenings. The first arm is a control arm that receives blood draws, the second arm will receive annual Shield MCD testing for two years, and the third arm will receive annual testing with ClearNote Health's Avantect MCED test for two years. Participants are aged 45 to 75 with no cancer diagnosis in the past five years. Enrollment opened on June 16, 2025. Practically, Vanguard is a feasibility study aimed at assessing willingness to undergo randomization, adherence to testing and diagnostic follow-up, and how diagnostic workflows vary by location, and the data will inform the design of future trials. The National Cancer Institute evaluated tests from multiple companies using blinded reference sample sets, and one could argue that by choosing Shield MCD, it was an independent endorsement of the platform's performance.