Exact Sciences Falls Behind GRAIL. Kind Of.

Bottom-Up Insights
  • This initiates coverage of Exact Sciences, which is the 11th company in our coverage ecosystem.
  • A case control study of the multi-cancer early detection tool in development showed far lower sensitivity than Galleri from GRAIL, especially in early-stage cancers.
  • Liquid biopsy tools are an important source of revenue growth potential in the second half of this decade, assuming the opportunity pans out for the field broadly and Exact Sciences specifically. The business remains on solid footing thanks to the unstoppable force known as Cologuard.
  • A second article will dig into my forecasts and explain the price targets.
MVP Article Disclosure: Please note this article was from our MVP platform and was written prior to September 2023. We've made numerous refinements, which means article structure, image and data visualization formats, and terms may have changed.

In mid-September 2022, Exact Sciences announced data from a case-control study evaluating its multi-cancer early detection (MCED) diagnostic. This is perhaps the most important liquid biopsy tool in development by the company, which could lay the groundwork for generating billions of dollars in additional annual revenue later this decade.

Unfortunately, the data suggest the tool isn't as good at detecting early-stage cancers as the MCED diagnostic from GRAIL, which is sold under the brand name Galleri. That's important because "early detection" is, well, it's literally in the name. Exact Sciences delivered Stage II and Stage III detection rates of 46% and 68%, respectively. Galleri from GRAIL delivered 77% and 84%, respectively, at a similar point of development.

There's an avalanche of nuance in how these data are collected, how sensitivity (detection rates) and specificity (false positive rates) are calculated, and how results or reported metrics are interpreted. A closer look at the nuances suggest Exact Sciences isn't as far behind GRAIL as the headline numbers might indicate, but it still doesn't seem nearly as good as Galleri from a technical standpoint.

Competitive, but Not the Best

Let's unpack a few nerdy terms first.

  • "Liquid biopsy" means the diagnostics uses blood samples. A tissue biopsy isn't an ideal first step for cancer screening because it's an invasive procedure. It can also be difficult to access many tissues inside the body, complicating procedures. A liquid biopsy tool can be developed to screen for a single cancer, or for many types of cancers at once.
  • "Case control" means the study used biobanked patient samples. The data weren't collected from individuals walking into their doctor assuming they were healthy. Rather, the data were collected from samples with known outcomes – did the patient who provided the sample in the past have cancer or not, what type of cancer, and at what stage of progression? Healthy individuals of similar ages with non-cancer diseases were included to tease out false positive potential, too. These studies are meant to inform the design of larger, real-world studies that can be used to earn regulatory approval.
  • Sensitivity is the ability to correctly identify individuals with cancer. It can also be called the detection rate.
  • Specificity is the ability to correctly identify individuals without cancer. It can also be called the inverse false positive rate. If an MCED diagnostic has a specificity of 99%, then 1 in 100 individuals will be incorrectly assumed to have cancer.
  • Cancers are often diagnosed using an ascending scale of stages. A Stage I cancer hasn't spread beyond the origin tissue, whereas a Stage IV cancer has likely spread to nearby lymph nodes and other organs in the body. It's basically a quick way of answering, "how much cancer is present and where is it located?"
  • The idea for a liquid biopsy tool is to detect as many signals as possible to increase sensitivity and specificity. "Multi-omics" means multiple molecules were collected and analyzed to make the diagnosis. These molecules – fragments of DNA, RNA, proteins, fats, and so on – are often referred to as biomarkers. A single molecule can also be analyzed with multiple techniques. For example, a fragment of DNA can be analyzed using genomics (sequencing the DNA), epigenomics (sequencing the chemical groups attached to the DNA), fragmentomics (understanding what tissue the DNA originated from), and so on.

Exact Sciences evaluated its MCED diagnostic using nearly 600 cancer samples and analyzed four types of biomarkers. The tool combines technology developed in-house, acquired from Thrive Earlier Detection (2020) and Base Genomics (2020), and developed in collaboration with researchers at Johns Hopkins and Mayo Clinic.

The case-control study achieved an overall sensitivity of 61% and an overall specificity of 98.2%. That means the tool correctly identified 61% of all cancers in the dataset regardless of severity. Additionally, the results suggest for every 1,000 individuals who receive the test roughly 18 will be incorrectly diagnosed with cancer. (Further testing would reveal they didn't have cancer, but it's generally not great to scare the shit out of people with inaccuracy.)

The headline numbers only tell us so much. That's because the sensitivity and specificity will be different for each stage of cancer and for each type of cancer. An MCED diagnostic might detect only 9% of Stage II bladder cancers, but sniff out 80% of Stage II colon cancers. Both numbers influence the overall sensitivity and specificity. A tool that excluded bladder cancer could therefore have higher sensitivity than a competing diagnostic that included it, but that doesn't necessarily mean the first product is better. Tricky tricky.

This highlights the nuances of MCED diagnostics, especially when attempting to make apples-to-apples comparisons between different tools.

  • An MCED diagnostic's overall sensitivity and specificity will be influenced by the types of cancers included in screening. There are two reasons. First, we haven't identified liquid biomarkers for all types of cancers at all stages. It remains difficult to detect early-stage bladder cancer using a liquid biopsy. Second, cancers have different incidence rates. Colon cancer is twice as common as bladder cancer. That amplifies the impact of overall sensitivity for a tool that detects 9% of early-stage bladder cancers and 80% of early-stage colon cancers.
  • An MCED diagnostic that screens for more types of cancers will always have a lower overall sensitivity than one that screens for fewer types of cancers. A tool that attempts to screen for 50 cancers could be weighed down by including 10 rare cancers with relatively low sensitivity rates.
  • An MCED diagnostic with a lower sensitivity could detect more total cancers in a population than one with a higher sensitivity. A tool that includes the three "easiest-to-detect" cancers might have an overall sensitivity of 80%. A tool that screens for 50 cancers and has an overall sensitivity of 40% will detect more cancers overall. Which one is more valuable from a public health standpoint?

All that nuance is important for understanding that there aren't any apples-to-apples comparisons in MCED diagnostics. So let's make some comparisons!

  • Exact Sciences presented data at the European Society of Medical Oncology (ESMO) Congress 2022. The study was called DETECT-A.
  • The closest comparison would be GRAIL's Galleri data presented at the 2019 American Society of Clinical Oncology (ASCO) Annual Meeting. The study was called the Circulating Cell-Free Genome Atlas (CCGA) 2019.

Both provide a snapshot at roughly equivalent stages of development.


Exact Sciences (DETECT-A 2022)

(CCGA 2019)

Overall sensitivity



Overall specificity



Cancers screened






Stage I sensitivity



Stage II sensitivity



Stage III sensitivity



Stage IV sensitivity



Data Source: Press releases.

The headline numbers suggest Exact Sciences is falling behind GRAIL in developing MCED diagnostics. The nuances suggest it's not as far behind as the headline numbers imply.

For example, DETECT-A screened for more cancers overall, screened more difficult-to-detect cancers such as bladder cancer, and included fewer overall samples. All weigh on the overall sensitivity.

It's worth pointing out that Galleri is being evaluated to screen over 50 different types of cancer. The data in the comparison above excluded certain difficult-to-detect cancers, such as bladder cancer. Follow-up data from 2021 showed the test demonstrated a sensitivity of 9.1% in Stage II bladder cancer, which lowered the overall sensitivity of the diagnostic. If those samples were excluded from the recent DETECT-A case-control study, then it would have reported a higher overall sensitivity.

Get ready for more comparisons in the coming years – many of them likely to ignore the nuances described above. Exact Sciences is on track to begin enrolling individuals in the Study Of All comeRs (SOAR) trial in 2023. The registrational study will collect data that can be submitted to the U.S. Food and Drug Administration (FDA) for formal approval, although results won't be available for many years.

Keep Expectations In Check

Investors should remember that MCED diagnostics won't be the endgame in cancer screening, but rather one of the arrows in the quiver. They'll be used side-by-side other screening and diagnostic tools. Positive tests will need follow up from more precise diagnostics and imaging.

To be sure, MCED diagnostics will detect cancers that would otherwise go undetected. They could more than double the number of cancers detected by screening, which will save lives and improve outcomes. Detecting early-stage cancers in asymptomatic individuals could revolutionize cancer screening.

But they won't be perfect. MCED diagnostics won't detect every cancer from every individual who signs up for screening. Importantly, if sensitivity rates from larger population studies are too low, then the economic opportunity could be lower than investors currently expect.

  • The FDA could balk at approval, request certain cancers be removed from screening, and/or require unique labeling.
  • These diagnostics may not be included in screening guidelines set by medical bodies.
  • The ability to earn widespread insurance coverage may depend on regulatory approvals and inclusion in screening guidelines. MCED diagnostics may have much lower selling prices and patients may need to pay out of pocket, which would significantly reduce revenue potential.

MCED diagnostics could also be a nightmare for science communication. If you thought explaining the WiFi to grandma was painful, then try telling her these tools may miss 40% of cancers and still provide tremendous clinical value.

In a similar vein, there's a small but growing misinformation risk associated with living technologies. That could ensnare MCED diagnostics, too. I can already envision conspiracy theories on blood collection and genome tracking spreading across the internet, which could limit adoption and use. However, many individuals make statements to signal their group identity that contradict their individual actions. An anti-vaxxer may share misinformation online and then get vaccinated anyway. The good ol' human condition.

Exact Sciences Remains Well Positioned in MCED

Despite all the nuances and potential for reduced market opportunities, Exact Sciences is well positioned to become one of the primary providers of MCED diagnostics.

The early data don't lead the competitive landscape and the development timeline remains years behind Galleri. However, the company has the most robust commercial infrastructure of any peer aside from Roche. Commercial infrastructure is defined as the labs processing samples, relationships with regional cancer centers and oncologists, and sales and marketing teams.

GRAIL appears to have the best technical capabilities. But it has virtually no commercial infrastructure, nor does it have much revenue to lean on. The business generated only $22 million in first-half 2022 revenue. That will impede the company's ability to scale efficiently and effectively. It may need $2 billion (with a "b") to build the commercial infrastructure required to adequately compete with Exact Sciences, Roche, NeoGenomics, and Guardant Health.

Achieving success in MCED diagnostics will require a Goldilocks combination of technical capabilities, clinical trial data, commercial infrastructure, and financial resources. Objectively speaking, no company is better positioned than Exact Sciences to seize the opportunity for shareholders.

Price Targets & Allocation

(No change from October 9.)

Note: I broke up the analysis of the MCED diagnostic and my modeling into two articles. These price targets are based on forecasts through 2025 that are explained in more detail in the next article.

Exact Sciences is considered an Anchor position. Current entry range price targets for the company are as follows:

  • Current Price (market close Oct. 18): $35.29 per share
  • Prioritize Below:                $61.52 per share
  • Midpoint:                            $79.95 per share
  • Caution Above:                  $98.38 per share
  • Allocation Range:              Up to 15%

Exact Sciences reported 176.96 million shares outstanding as of August 1, 2022. The price targets above assume 185 million shares outstanding, which prices in 4.5% dilution expected in near-term payouts for recent acquisitions.

Further Reading

  • May 2019 press release announcing CCGA data from Galleri
  • June 2021 study discussing more mature CCGA data from Galleri and the nuances of MCED diagnostics
  • September 2022 press release announcing DETECT-A case-control study data from Exact Sciences