AdComm votes in favor of new colorectal cancer screening test, but questions about serial testing and labeling remain

Quick background: The Guardant SHIELD test

• The Guardant Shield Blood Collection Kit is being considered under a Pre-Market Approval (PMA) application. The Shield blood test is an in vitro diagnostic (IVD) product intended to screen for colorectal cancer (CRC) markers in blood. Specifically, the sponsor is proposing an indication for the detection of “colorectal cancer derived alterations in cell-free DNA from blood collected in the Guardant Blood Collection Kit.” Per the proposed indication, “Patients with an “Abnormal Signal Detected” may have colorectal cancer or advanced adenomas and should be referred for colonoscopy evaluation. Shield is not a replacement for diagnostic colonoscopy or for surveillance colonoscopy in high-risk individuals.” Following collection, which is completed in a provider’s office, the test itself is performed at Guardant Health.
• On May 23, the FDA’s Medical Device Advisory Committee panel on Molecular and Clinical Genetics Devices (MCGP) convened to discuss Guardant’s PMA for the Shield blood test.
• Let’s get the conclusion out of the way up front: The panel voted 7-2 in favor of the device’s approval. We’ll spend the rest of this piece explaining how the panel came to this conclusion.
• Shield is a screening test, meaning that an abnormal result indicates that the patient should be referred for a colonoscopy evaluation. As a screening test, Shield does not necessarily diagnose cancer, but rather identifies blood markers of CRC that warrant follow up. It is intended for use in individuals at average risk for CRC who are 45 or older.
• Guardant recently published results from its ECLIPSE study on the Shield blood test. ECLIPSE was intended to evaluate the performance of the Shield blood test compared to a screening colonoscopy. The study included participants 45-84 years of age who were of average risk for CRC; the program enrolled almost 25,000 subjects from 265 sites. The final clinical evaluation dataset included 7,861 participants (i.e., those with valid Shield tests results and a valid colonoscopy for comparison. Overall, it found the test to have an 83% sensitivity in detecting individuals with CRC; however, the sensitivity was found to vary by stage. While the Shield blood test demonstrated 100% sensitivity for CRC Stages II, III and IV, the sensitivity was 65% for pathology-confirmed Stage I and 55% for clinical Stage I. Further, the test had 13.2% sensitivity detecting advanced adenomas (AA).
• In effect: the test had high sensitivity for later stages of CRC, but faced challenges with identifying early-stage cancer or pre-cancerous lesions. As for specificity, the false positive rate of 10% identified during the study yielded a specificity of 90%.

Shield is already available as a laboratory developed test (LDT), which allowed the sponsor to utilize a unique data set: Adherence

• The PMA for the Shield blood test was submitted to FDA in May 2023. However, sponsor Guardant Health has made the test available as an LDT since May 2022; since that time, “it has been used by more than 20,000 people and more than 90% of patients who were prescribed the test in the real-world clinical setting completed it,” according to company press releases.
• That last point about patient uptake is a key factor in Guardant’s materials on the Shield blood test. “CRC is the fourth most diagnosed cancer and second leading cause of cancer-related death in the US,” as outlined in materials for the panel from Guardant, and “an estimated 76% of CRC-related deaths occur individuals who are not up to date with screening.” Notably, CRC screening methods are currently available and have generally high performance (i.e., sensitivity and specificity) – including colonoscopy procedures and an FDA-approved stool-based test (Cologuard) – but comparatively low uptake, given the prevalence of CRC in the population. As sponsor Guardant put it, there is “suboptimal adherence with current screening options.”
• Shield would be the first blood-based method; “Guardant Health’s market research found that 64% of patients prefer a blood-based screening test over all other screening methods, including colonoscopy or stool-based tests,” the company has stated, and blood tests do not require preparation or sedation.
• The sponsor was able to leverage some of the test’s experience as an LDT to make the case for unmet need. The sponsor made the case that adherence should be incorporated into assessments of the public health outcomes – in effect, incorporating information about rates of tests ordered versus tests completed in participation. As the sponsor presented, “A retrospective review of completion rates from the first 10,000 clinical orders for Shield (operated as a laboratory-developed test, LDT) demonstrated 96% adherence in screening age-eligible individuals. While this may be an overestimate skewed by early adopter bias, these data support that blood-based screening will yield higher adherence relative to current CRC screening modalities.” The early adopter (i.e., the first people to adopt something are more likely to want to use it) bias did appear to impact the numbers, but “Even when assessing Shield at an adherence of 80%, which is significantly below the 96% observed with the LDT implementation, CRC detection remains at or above that of other CRC screening testing modalities (66%).”
• According to Guardant, higher-than-average adherence should help bolster sensitivity. As the sponsor’s representatives put it during the meeting, “the tests must be completed to be effective.” At the panel meeting, Guardant’s representatives made the case that the relatively low burden (i.e., single blood draw, no sedation) of the Shield for patients would result in higher uptake, and, importantly, repeated uptake. This information could even help address concerns about the limited sensitivity for AA (as noted above, 13.2%), Guardant argued in its meeting materials: “Shield’s limited advanced adenoma detection could result in harm if the screening test is used only once in a lifetime,” because “expected higher adherence” and “extended dwell time” of AA progressing to CRC may mitigate the risk of accrued harms.
• Notably, ECLIPSE itself didn’t include repeat testing – so all of the information about testing adherence in the real world did come from the LDT experience. Longitudinal testing is underway to evaluate repeat testing, but this wasn’t included in the original analysis of ECLIPSE. Panelist CHARITY MORGAN (University of Alabama at Birmingham) asked about “the decision not to do repeat testing in the two-year follow-up period” from the study, citing the sponsor’s assertions about the value of repeat testing. Guardant Vice President VICTORIA RAYMOND responded that the study was intended to demonstrate CRC sensitivity and specificity in the average risk screening population – as expected in an FDA package – and “was designed in that way, and so repeat testing was not part of the study design.”
• Adherence and access were a key point for the panel. During the meeting, panelists grappled with the idea of where Shield would fit in the screening and diagnostic landscape, especially as the gold-standard options are currently available. Acting Director of the CDRH Office of Health Technology 7 (OHT7, the diagnostics office)’s Division of Molecular Genetics and Pathology DONNA ROSCOE did acknowledge that assessing the availability of options like colonoscopy were technically outside of the FDA’s purview when considering the PMA for a screening test, but did encourage discussion from the panel on the way that these considerations impact the benefit/risk profile of the screening test. She went on to (later in the meeting) acknowledge that a blood-based test would be expected to increase compliance with testing regimens – but, again, this information was not provided under the ECLIPSE study.
• Several panelists focused on this point, noting that repeated testing would be critical to fully understanding the performance of the device. As Guardant Chief Medical Officer CRIAG EAGLE acknowledged, the data on update is anecdotal, and is not from ECLIPSE (more on that below): “We do have a single point adherence in the range of 88-90%” compared to 30-60% for other CRC testing methods. However, given the novelty of the test itself, he also noted that this was part of the bigger expectation for post-market follow up, and that Guardant would commit to continue to assess adherence going forward. “But certainly, this is a very valid point,” he acknowledged to the panel.

The panel voted 7-2 in favor of the device; this followed panel deliberation on three discussion questions

• The panel meeting included both discussion questions and voting questions. The discussion questions focused on key issues related to the device and supporting data: Its context in the existing clinical workflow, AA, and longer-term follow up. The voting questions included the typical three part vote: If the device is safe, it is effective, and if the benefits outweigh the risk (i.e., whether it should be approved). The panel voted 7-2 on the last question, supporting approval of the PMA.
• Question one: Appropriate patient population and clinical scenario (including primary screening or more specific, targeted groups). The FDA raised the question of whether Shield should be considered a “first line” test or a “second line” test. First line tests are those that are indicated as a primary screening method for average risk adults (e.g., Cologuard). Second line tests are those that are indicated for individuals at average risk who decline first line tests. Guardant’s proposed indication is “most similar to “first line” claim,” per FDA, and Guardant Vice President VICTORIA RAYMOND confirmed that Shield is seeking a primary test indication, as “the performance of Shield is within range of [already available] primary screening options.” From CDRH’s perspective, “this test is falling in the middle,” acknowledged Roscoe (CDRH) when compared to stool-based tests, “but there are advantages to this test,” she said.
• The panel roundly agreed that the Shield test offered some significant value, but grappled with where the test would fall in the clinical context. The panel had a wide ranging discussion about what exactly it looks like in a clinical workflow to be a primary versus secondary option – in effect, the process or expectation for turning down colonoscopy. “I think it makes more sense to have this as a second line,” said panelist Morgan, for patients who decline or do not have access to colonoscopy; for this group “I think this is a good option for them,” she concluded. However, “I understand that people who don’t have access to [colonoscopy], this is better than nothing… but I don’t want to downplay the issue that this test is going to miss a lot of cancers, and advanced adenomas,” said Morgan. ALEXANDER BOROWSKY (UC Davis School of Medicine) questioned whether the test could be considered “not just as an alternative secondary screen, but as a primary screen which is recognized to be an imperfect primary screen.” Panelist PADMA RAJAGOPAL (National Cancer Institute) further highlighted the potential (and anecdotal from clinical experience) confusion already seen as to whether the test was primary or secondary, and that labeling and materials will need to be extremely clear. Roscoe (CDRH) asked for some input from the panel about what that would look like, and panelists pointed to the varying efficacy of methods to communicate relative risk in other cancer screening methods.
• Question two: AA. As noted above in the proposed indication statement, Shield’s output will show “Abnormal Signal Detected” in patients who “may have colorectal cancer or advanced adenomas.” The ECLIPSE study demonstrated a 13.2% sensitivity for the detection of AA, compared to 83% for CRC. The false negatives for AA and earlier stages “is the reason for the panel,” Roscoe (CDRH) confirmed during the meeting.
• In general, the panel agreed that “the sensitivity for AA of Shield is not at the level of these other approved noninvasive tests,” as Morgan put it. “For this test, I think that a possible mitigation is very clear labeling for the product,” she further concluded. “The committee still has concerns about the lack of sensitivity for detection” of AA, Panel Chair ANDREA FERREIRA-GONZALEZ (Virginia Commonwealth University) summed, noting that education about the test’s limitation will be critical.
• Question three: Post-Approval Studies (PAS). Given the ongoing questions about sensitivity in certain contexts, the potential value of repeat testing (but without ECLIPSE data) and the real-world implications of the test in clinical practice, the FDA requested information about what data should be collected in ongoing studies.
• The panel focused on information that was already flagged by the agency, including “more information about those false positives we’re seeing, and more long term data for those subjects,” according to Morgan. Panelist SEAN SPENCER concurred, and added that the added context of understanding how CRC screenings are “performed more so in a primary care setting… that is ultimately where we envision this being used” would be helpful going forward.
• As noted above, the panel voted 7-2 in favor of the device’s approval. The panel also voted 8-1 in favor of the device’s safety, 6-3 in favor of the device’s efficacy. While panelists generally agreed that there is still a need for ongoing studies and ensuring appropriate education on the role of the Shield (including its limitations) for providers and patients. For panelists that voted against efficacy, the general consensus was a concern about the efficacy levels in AA, which they generally noted could be mitigated through labeling – “I think in the labeling, it should clearly indicate that this is not to detect adenomas and is not designed as a preventative strategy,” said Spencer. Panelist KARLA BALLMAN (Mayo Clinic College of Medicine and Science) concurred, saying “I do feel like it is a good colon cancer screening test but for later stage colon cancer, [and] not as good for stage one, and definitely not good for advanced adenomas.”

Analysis & What’s Next

• Up front, the advisory committee panel meeting touched on several hot topics in diagnostics regulation. This included moving more tests to primary care settings, the role of screening in the diagnostic landscape, the way that LDT experience can be used in a marketing application, and the concept of repeat or serial testing.
• Repeat testing as a way to bolster the benefit/risk profile: As described above, the sponsor prioritized the benefits from the way that Shield will be incorporated into the clinical care landscape, and particularly noted that repeated testing should buffer against any accrued risk from lower sensitivity. This should be a familiar position for diagnostics developers over the last few years; in particular, it follows the FDA’s longstanding precedent that comparatively lower sensitivity antigen tests for Covid-19 could meet the thresholds for cumulative sensitivity through repeat testing strategies (i.e., serial or orthogonal testing), although Covid-19 antigen tests are labeled to include the expectation of repetition in a set amount of time. While the Shield test wouldn’t include a similar labeling strategy (e.g., the test could not be indicated for annual repeats, that would come through clinical guidelines), this was a key point of discussion for the panel, FDA regulators, and the sponsor. The experience that Guardant had before the panel will be informative for development programs of less invasive (and potentially less sensitive) screening tests going forward – in effect, if the increased access and expectation of compliance and adherence will work like a serial testing strategy to limit the accrued risk of lower sensitivity.
• The data on repeat testing and adherence came from the LDT experience. This makes two interesting points; first, while the sponsor did not rely on real-world data (RWD) and evidence (RWE) from offering the test as an LDT (which has been done for other LDTs seeking full market access authorizations), they were able to provide practical information on the clinical experience of Shield from its tenure as an LDT. This included the information on adherence, uptake and compliance with the screening test, especially in comparison to the other available options. With the FDA currently working to implement a rule that would require most LDTs to submit a marketing application to the FDA over the coming years, the way that the information gathered while the test was offered as an LDT can be used to support such a marketing application is likely to be top of mind for many developers.
• This was particularly useful as the ECLIPSE study did not include information on repeat testing, in part because repeat testing would not necessarily align with the types of test data is expected in an FDA package. As Panelist SEAN SPENCER (Stanford School of Medicine) noted during the discussion (particularly during question three, about a post-approval study) concluded that “the ECLIPSE study had to be designed the way that it was,” generally to support the overall validation of the test (i.e., point in time comparison between the gold-standard method of colonoscopy and the performance of the Shield investigational test), as expected by the FDA in a submission. This does raise some interesting questions about how a sponsor would reasonably go about designing a study that could account for a longer-term repeat testing strategy, of which there is currently limited information from the FDA outside of the pandemic experience for Covid-19 tests.
• These issues especially highlight one of CDRH’s overarching goals for the future of diagnostics: moving them into lower-acuity settings. Diagnostic tests have historically been primarily used in clinical laboratories, but public health officials and regulators have been indicating interest in bringing access to new tests to other settings – including non-laboratory tests (point of care, primary care) and at-home or direct-to-consumer options. The Shield test’s blood-based sample method would allow for CRC screening to take place in the course of regular primary care, potentially expanding access and uptake. Several of the questions raised at the Shield advisory panel meeting will likely apply to these technologies going forward, including how increased access or adherence is considered in the benefit/risk framework and what labeling and educational materials for primary care providers offering different lines of screening should look like.

To contact the author of this item, please email Laura DiAngelo ( ldiangelo@agencyiq.com).
To contact the editor of this item, please email Alexander Gaffney ( agaffney@agencyiq.com)

Key Documents and Dates

• May 23, 2024: Molecular and Clinical Genetics Panel of the Medical Devices Advisory Committee Meeting Announcement
• MCGP Meeting Recording (YouTube)