“How accurate is it?” is the first question almost everyone asks about the Cancerguard test, and it is a completely reasonable one. It is also, as it turns out, a question that cannot be answered with a single number — and the reason why is genuinely worth understanding, because it applies to every screening test you will ever be offered, including ours.
Accuracy in medicine is not one measurement. It is at least three, they answer different questions, and the one that matters most to you personally is usually the one nobody quotes in the marketing. Once you can read these numbers properly, you can evaluate any test that comes along — this year’s blood tests, next year’s, and the imaging options too.
This article walks through what each number means, what Cancerguard’s published figures actually are, and how to interpret a result you have already received.
Why “accurate” is the wrong single word
Imagine a test that simply says “no cancer detected” to everybody, every time. Because most people screened do not have cancer, that test would be right about 99% of the time. Technically, 99% accurate. Practically, useless — it never finds anything.
That is the trap in the word “accurate.” A test’s usefulness depends on how it performs in two very different situations: when cancer is present, and when it is not. Those two performances are measured separately, and they are called sensitivity and specificity. A third number, positive predictive value, tells you what a result actually means for the person holding it.
Three numbers, three different questions. Here they are in plain English.
Sensitivity: how often it catches cancer that is there
Sensitivity answers a simple question — of the people who genuinely have cancer, what percentage does the test correctly flag?
If a test has 68% sensitivity, it means that out of 100 people with the cancers being measured, roughly 68 would receive a positive result and about 32 would be told nothing was detected, even though something was there. Those 32 are called false negatives.
Sensitivity is the number that determines how much reassurance a negative result is worth. A test with modest sensitivity can still be worth taking, but it cannot honestly be described as a clean bill of health. This is why sensitivity is the first figure to look for when a test is marketed on peace of mind.
Specificity: how often it correctly clears healthy people
Specificity is the mirror image — of the people who do not have cancer, what percentage does the test correctly tell so?
A specificity of 97.4% means that among 1,000 healthy people, roughly 974 are correctly reassured and about 26 receive a positive result despite having no cancer. Those 26 are false positives.
At first glance 97.4% sounds close to flawless, and this is exactly where most people’s understanding stops. It is also where the most important idea in screening begins.
Positive predictive value: the number that actually affects you
Sensitivity and specificity describe the test. Positive predictive value, or PPV, describes your result. It answers the question you would actually be asking if a positive result arrived in your inbox: given that this test came back positive, what are the odds I really have cancer?
The answer depends not only on the test but on how common cancer is in the group being screened. And because cancer is rare in any given year among people without symptoms, the arithmetic works out in a way that surprises nearly everyone.
Here is a worked illustration. Picture 10,000 people being screened, and suppose 100 of them — 1% — have a cancer the test is designed to find. With 64% sensitivity — the overall figure excluding breast and prostate, which is the right one to use here because it covers the full range of cancers in scope rather than only the six deadliest — the test would correctly flag about 64 of those 100. Meanwhile, 9,900 people do not have cancer, and with 97.4% specificity, about 2.6% of them will still get a positive result. That is roughly 257 people.
So the total number of positive results is about 321 — of which only 64 are real. That works out to a positive predictive value of around 20%. Four out of five people receiving a positive result would not have cancer.
This is not a flaw in the test. It is what happens when you apply a very good test to a rare condition, and it is unavoidable. Published multi-cancer early detection trials show exactly this pattern:
| Study | Population | Positive predictive value | Share of positives that were false alarms |
|---|---|---|---|
| DETECT-A, blood test alone (Science, 2020) | 10,006 women aged 65–75 | 19.4% | About 81% |
| DETECT-A, blood test plus PET-CT | Same group | 28.3% | About 72% |
| PATHFINDER (The Lancet, 2023) | 6,621 adults aged 50+ | 38% | 62% |
| PATHFINDER 2 (ASCO, 2026) | 35,878 adults aged 50+ | 60.3% | About 40% |
Two things are worth noting. First, the trend is genuinely encouraging — predictive value has improved substantially as these tests have developed. Second, and importantly: these figures come from studies of Galleri and CancerSEEK, not Cancerguard. Cancerguard has not yet published an equivalent large prospective study of its own, so no one can quote you a real-world PPV for it. That gap is worth knowing about before you interpret any result.
It is also fair to say that imaging faces the same mathematics. Any test looking for something uncommon will produce findings that turn out to be nothing, and whole-body MRI is no exception — incidental findings are common, and we tell patients that before they book, not after.
What Cancerguard’s published numbers actually are
In its test-development studies, Cancerguard reported 68% sensitivity across six of the deadliest cancers — pancreatic, lung, liver, esophageal, stomach and ovarian — and 64% overall sensitivity when breast and prostate cancers are excluded. Specificity was 97.4%. It also detected more than one in three of the stage I and II cancers.
Two pieces of context belong alongside those figures. Cancerguard is a laboratory-developed test rather than an FDA-approved product, it is not generally covered by insurance, and it is indicated for adults aged 50 to 84 with no cancer diagnosis in the past three years. It is also not indicated for breast, prostate, or pre-cancerous lesion screening — and breast and prostate together account for roughly 31% of new cancer diagnoses in the United States.
The other point is about the source of the numbers themselves. These are development-study figures, generated during the process of building and validating the test, not results from a large prospective screening trial of the kind Galleri has now published twice. Development studies tend to enroll people already known to have cancer, which is scientifically appropriate but is not the same as screening a general population and seeing what happens. Performance in the real world is usually more modest. That distinction is not a criticism of Cancerguard — it is simply where the evidence currently stands, and it is the honest way to read the numbers. We compare the two tests more directly in our guide to Cancerguard versus Galleri.
Why stage matters so much
There is one more layer, and it matters more than any of the headline percentages.
These tests work by detecting circulating tumor DNA — fragments of genetic material shed into the bloodstream by tumor cells. How much of that material is circulating depends largely on how much tumor is present. A large, advanced tumor sheds a great deal. A small, early one may shed almost nothing.
That is why sensitivity is consistently lower for stage I and II disease than for stage III and IV, and why sensitivity at stage I has been reported in the low tens of percent across MCED validation studies. The biology sets a floor, and the floor sits precisely where early detection would be most valuable.
So when a test reports 68% sensitivity, understand that the figure blends together easier-to-detect advanced disease with harder-to-detect early disease. It is not evenly distributed.
What a negative result does and does not mean
A negative Cancerguard result means no cancer signal was detected in that blood sample, at that moment, above the test’s detection threshold.
It does not mean you do not have cancer. No multi-cancer early detection test is sensitive enough or broad enough for that conclusion, and the manufacturers say so themselves. The concern researchers raise most often is that a reassuring blood result may lead someone to postpone a mammogram or skip a colonoscopy — the very screenings with decades of evidence behind them. That would be a poor exchange.
Keep your guideline-based screening exactly as it is. A negative MCED result is a small piece of additional information, not a replacement for anything.
Where imaging fits
If a blood test raises a question, something has to answer it. In the PATHFINDER study, 92% of participants with a positive signal went on to have at least one imaging test — 61% had PET-CT, 39% CT, and 21% MRI. The median time to a definitive answer was 79 days: 57 days for those whose cancer was real, and 162 days for those whose result was a false alarm.
Imaging asks a fundamentally different question. A blood test asks whether a signal is present; imaging asks what is physically there, and reports location, size and shape. Our radiologists read whole-body scans without ionizing radiation and without contrast dye, and patients from age 18 are eligible.
We are equally clear about what imaging cannot do. Whole-body MRI is poor at detecting small lung nodules — published sensitivity for indeterminate pulmonary nodules is around 23%. It does not evaluate the lining of the colon, so it cannot replace a colonoscopy. A whole-body protocol is not the same as a dedicated breast or prostate MRI. And like every MCED test, whole-body MRI screening has no published evidence that it extends life. Anyone who tells you otherwise is ahead of the data.
We think of these tools as sequential rather than competing, which is the theme of our fuller comparison of cancer blood tests and whole-body MRI.
Reading the next accuracy claim you see
If you take one thing from this article, let it be a habit: whenever you see a screening test advertised with a single impressive percentage, ask three questions. Sensitivity for which cancers, and at which stages? Specificity — and how many false positives does that produce in a group this size? And is there a published predictive value from a real screening population, or only from a development study?
Those three questions will tell you more than any brochure. They are the same questions we apply to our own scan, and Dr. Edward Tavel and our radiologists would rather you arrive with them than without them. If you have already taken a test and are unsure what your result means, that is a reasonable thing to talk through rather than carry alone.
Taking a proactive approach to your health starts with understanding what is actually happening inside your body — and with knowing what each number on a report can honestly tell you.
Frequently asked questions
How accurate is the Cancerguard test?
In its development studies, Cancerguard reported 68% sensitivity across six of the deadliest cancers, 64% overall sensitivity excluding breast and prostate, and 97.4% specificity. These are development-study figures rather than results from a large prospective screening trial, so real-world performance may differ.
What does Cancerguard's 97.4% specificity mean in practice?
It means about 26 in every 1,000 people without cancer would still receive a positive result. Because cancer is uncommon in any screening year, those false positives can outnumber true findings — published MCED trials have reported positive predictive values ranging from roughly 19% to 60%.
Is the Cancerguard test reliable enough to rule out cancer?
No. A negative result means no cancer signal was detected above the test's threshold, not that cancer is absent. Sensitivity is substantially lower for early-stage disease, and guideline-based screening such as mammography and colonoscopy should continue unchanged.
Is Cancerguard FDA-approved?
No. Cancerguard is a laboratory-developed test rather than an FDA-approved product, it is not generally covered by insurance, and it is indicated for adults aged 50 to 84 with no cancer diagnosis in the previous three years.
What happens after a positive Cancerguard result?
A positive result is the start of an investigation, not a diagnosis. In the PATHFINDER study, 92% of people with a detected signal had at least one imaging test, most commonly PET-CT, and the median time to a definitive answer was 79 days — 162 days for those whose result proved to be a false alarm.
Sources
- Cancer statistics, 2026
- Cancerguard™ Clinician Brochure: Test Information for Healthcare Providers
- Feasibility of blood testing combined with PET-CT to screen for cancer and guide intervention
- Blood-based tests for multicancer early detection (PATHFINDER): a prospective cohort study
- GRAIL Presents PATHFINDER 2 Results of More Than 35,000 Participants Showing the Galleri® Test Substantially Increased Cancer Detection With Robust Performance and Favorable Safety at 2026 ASCO Annual Meeting
- Multi-Cancer Early Detection Tests: State of the Art and Implications for Radiologists
- Questions and Answers about Multi-Cancer Detection Tests
- Whole-body MRI for opportunistic cancer detection in asymptomatic individuals: a systematic review and meta-analysis
