Grail’s Galleri test can look for signals from more than 50 types of cancer, but its accuracy varies by cancer type. Galleri’s sensitivity ranges from 93.5% for liver and bile-duct cancer to 11.2% for prostate cancer and 0% for thyroid cancer. This is from published cancer-type results.
To understand what a negative result means, it helps to separate three questions: which cancer, which stage, and which study?
Which cancers does Galleri detect most often?
Sensitivity is the fraction of people with cancer whose test detects a cancer signal. Specificity is the proportion of people without cancer who test negative.
The chart below shows the 25 cancer classes in the Grail’s CCGA performance table, including a pooled “other” category. The highest detection rates were for liver/bile-duct, head-and-neck, esophageal, pancreatic, and ovarian cancers. Breast, kidney, prostate, and thyroid cancers were much less likely to produce a positive result.
Dots show CCGA sensitivity and lines show 95% Wilson intervals calculated from published counts; asterisks mark cancers with routine screening for eligible people. Source: Klein et al., Annals of Oncology, 2021.
Note that the thyroid result came from just 14 cases, none detected (it doesn’t establish that Galleri can never detect thyroid cancer). Subgroup size and the mix of stages affect these estimates, so small differences between adjacent estimates shouldn’t be treated as a reliable ranking. CCGA study poster.
Cancer stage changes detection accuracy
Galleri’s overall sensitivity was 51.5%, but accuracy rose sharply with stage. Sensitivity at detecting Stage III cancer was 77% and stage IV was 90.1%:
| Stage | Sensitivity |
|---|---|
| I | 16.8% |
| II | 40.4% |
| III | 77.0% |
| IV | 90.1% |
Source: Klein et al., Annals of Oncology, 2021.
A cancer-type percentage combines the stages included in that group. “83.7% for pancreatic cancer,” for example, isn’t a claim that the test detects 83.7% of stage I pancreatic cancers. Early detection is the intended benefit, but early-stage disease is also where detection is hardest.
Does that mean Galleri finds half of cancers in screening?
CCGA was a case-control validation study, with known cancer cases and non-cancer controls. It wasn’t a trial of repeatedly screening an otherwise healthy population.
In the later prospective PATHFINDER 2 study, the MCED-V2 assay detected 39.3% of cancers diagnosed within 12 months. That’s an episode-sensitivity estimate in a screening cohort. It shouldn’t be substituted into every CCGA cancer-type bar, or treated as the result for every version of the assay. PATHFINDER 2.
The study designs answer different questions. A validation study can show whether a test distinguishes samples with and without cancer. A screening study helps establish how often signals appear, what follow-up finds, and which cancers are diagnosed after negative results.
What does a negative Galleri result mean?
A negative result means the test didn’t detect a cancer signal. It doesn’t rule out cancer, especially a type or stage with low sensitivity.
Galleri should supplement recommended screening. It doesn’t replace mammography, colorectal screening, cervical screening, or low-dose lung CT when you’re eligible. Symptoms still need evaluation even after a negative test. NCI’s guide to multi-cancer detection.
For the tradeoff between detection, false alarms, and follow-up, see The Galleri cancer test produced fewer false positives than I expected. Our original Galleri analysis provides additional background. The cancer-type results explain why “screens for more than 50 cancers” is a description of scope, not a promise of equal detection.
Get your free 30-day heart health guide
Evidence-based steps to optimize your heart health.
