The existing data for available MCED tests, with sources cited.
This page is for informational purposes only and is not medical advice. Every value below is reproduced from a company page or a published study, linked next to it, without interpretation, and none of it is a recommendation. If you are considering one of these tests, please discuss it with your own clinician.
Three multi-cancer early detection tests can be ordered in the United States today: Galleri, Cancerguard, and Caris Detect. The table below puts what each company and each published study has reported side by side, with a link to the source next to every value. "Not reported" means I could not find the value on any company or study page.
Last checked September 12, 2026.
* Not yet available in the US.
| What it looks for How many cancers the company says the test can pick up, who it is meant for, and which cancers it is weak on. | ||||
| Galleri | Cancerguard | Caris Detect | Shield MCD* | |
|---|---|---|---|---|
| Cancer types claimed by the company | Signal shared by more than 50 types of cancer [1] | Signals associated with 50+ cancer types and subtypes [2] | Signals from 55+ types of cancer [3] | 10 cancers: bladder, breast, colorectal, esophageal, gastric, liver, lung, ovarian, pancreatic, prostate [4] |
| Intended population, as stated by the company | Recommended for adults with an elevated risk for cancer, such as those aged 50 or older [5] | Adults ages 50-84 with no known cancer diagnosis in the preceding 3 years [6] | Adults 21 and older [7] | Adults 45 and older at average risk [8] |
| Cancers with low sensitivity, as stated by the company | Low sensitivity for cancers that shed little DNA, for example brain, skin, and early breast and prostate cancers [1] | Not reported | Breast had the lowest sensitivity (53%) in the Achieve 1 study, per a news report [9] | Per-cancer sensitivity ranged from 96% (esophageal and stomach) down to 21% (prostate), and was 3% for stage I-II prostate cancer [10] |
| Technology, as described by the company | Next-generation sequencing of methylation patterns in cell-free DNA, with a predicted cancer signal origin [11] | Methylated cell-free DNA by real-time PCR plus protein biomarkers by immunoassay [6] | Whole genome and whole transcriptome sequencing with an artificial intelligence model [12] | Methylation patterns and cell-free DNA fragment positions (fragmentomics), with genomic analysis [13] |
| Performance in screening studies These numbers come from studies that gave the test to healthy people and then followed them. This is the closest thing to what would happen if you took the test. | ||||
| Galleri | Cancerguard | Caris Detect | Shield MCD* | |
| Sensitivity, all cancers | 39.3% (95% CI 34.9-44.0), 12-month episode sensitivity; PATHFINDER 2, 32,007 participants [14] | No prospective screening data for the commercial test [15] | No prospective screening data | No prospective screening data; enrolled in the NCI Vanguard study (up to 24,000 people ages 45-75) since July 2025 [16] |
| Sensitivity, 12 pre-specified deadly cancers | 69.8% (95% CI 62.8-76.0); PATHFINDER 2 [14] | Not reported | Not reported | Not reported |
| Specificity | 99.6% (95% CI 99.6-99.7); PATHFINDER 2 [14] | Not reported | Not reported | Not reported |
| Positive predictive value | 60.3% (95% CI 54.5-65.8); 173 cancers among 287 positive results; PATHFINDER 2 [14] | Not reported | Not reported | Not reported |
| Cancer signal origin accuracy | 91.3% (95% CI 86.2-94.7); PATHFINDER 2 [14] | Not applicable; the test does not report a cancer signal origin [6] | Not reported | Not reported |
| Share of detected cancers at stage I-II | 53.0% (80 of 151); PATHFINDER 2 [14] | Not reported | Not reported | Not reported |
| Randomized trial: primary endpoint | Not met; stage III and IV cancers combined across 12 pre-specified cancers, 706 versus 688, incidence rate ratio 1.03; NHS-Galleri [17] | Not reported | Not reported | Not reported |
| Randomized trial: stage IV cancers | Reduced 14% (342 versus 397); by round, 9% at year 0, 22% at year 1, 26% at year 2; NHS-Galleri [17] | Not reported | Not reported | Not reported |
| Mortality data | None reported [17] | Not reported | Not reported | Not reported |
| Performance in case-control studies These studies mix known cancer patients with healthy volunteers. They make a test look better than it will in real screening. | ||||
| Galleri | Cancerguard | Caris Detect | Shield MCD* | |
| Sensitivity, overall | 51.5% (95% CI 49.6-53.3); CCGA validation set, 4,077 participants [18] | 55.6% (95% CI 49.0-62.0), excluding breast and prostate; validation study, 324 cancers of 19 types and 800 controls. The company notes real-world sensitivity may be lower than in case-control data [15] | Not reported | 59.7% (224 of 375) for the 10 cancers in the signal-origin model; 55.7% (263 of 472) across all 14 tumor types; blinded case-control study, 962 evaluable participants (472 with cancer, 442 without) [10] |
| Sensitivity by stage | Stage I 16.8%, II 40.4%, III 77.0%, IV 90.1%; CCGA [18] | Stage I 26.8%, II 42.9%, III 63.6%, IV 89.3%; validation study, excluding breast and prostate [15] | Stage I 56.8%, II 67.7%, III 79.0%, IV 98.6%; Achieve 1 study, 3,014 subjects enrolled on the basis of high-risk screening, symptoms, or a mass on imaging; reported in a company press release, not published or peer reviewed [19] | Stage I-II 35%, III-IV 84% for the 10 cancers in the signal-origin model; 31% and 81% across all 14 tumor types [10] |
| Specificity | 99.5% (95% CI 99.0-99.8); CCGA [18] | 97.4% (95% CI 96.0-98.3); validation study [15] | 99.2% in 141 asymptomatic subjects; 96.0% in 2,051 subjects with benign conditions or high risk; Achieve 1; reported in a company press release, not published or peer reviewed [19] | 98.6% (436 of 442) [10] |
| Cancer signal origin accuracy | 88.7% (95% CI 87.0-90.2); CCGA [18] | Not applicable; the test does not report a cancer signal origin [6] | Not reported | 92%; primary or secondary prediction, for the 10 cancers in the signal-origin model [10] |
| Evidence and practical details A randomized trial is the strongest evidence. A company brochure is the weakest. | ||||
| Galleri | Cancerguard | Caris Detect | Shield MCD* | |
| Evidence, as reported | Randomized trial (NHS-Galleri) and prospective screening study (PATHFINDER 2), both presented at ASCO 2026 as abstracts; the earlier PATHFINDER study was published in The Lancet in 2023 [14] | Case-control studies reported in a conference abstract (AACR 2026, abstract 1108) [15] | Company press release only; not peer reviewed; not a screening population [19] | Case-control conference data only (ASCO 2025, abstract 10550) [10] |
| Time to results, as stated by the company | About 2 weeks after the sample arrives at the lab; up to 4 weeks in some cases [1] | About 2 weeks after the lab receives the sample [20] | About 2 to 3 weeks after the blood draw [3] | Not reported |
* Not yet available in the US.
| Study | Test | Design | Participants | Where reported |
|---|---|---|---|---|
| PATHFINDER 2 [14] | Galleri | Prospective screening study, single arm, results returned to participants | 35,878 enrolled; 32,007 in the performance analysis | Nat Med, September 22, 2026 (doi:10.1038/s41591-026-04618-w); first presented at ASCO 2026, abstract LBA10509 |
| NHS-Galleri [17] | Galleri | Randomized controlled trial, up to three annual screening rounds (years 0, 1 and 2) | 142,250 randomized | N Engl J Med, September 22, 2026 (doi:10.1056/NEJMoa2505723); companion test-performance paper in Nat Med (doi:10.1038/s41591-026-04652-8); first presented at ASCO 2026, abstract LBA100; NCT05611632 |
| PATHFINDER [23] | Galleri | Prospective screening study, results returned to participants | 6,662 enrolled; 6,621 with analyzable results | Lancet 2023;402:1251-60 (summarized in The ASCO Post, October 19, 2023) |
| CCGA validation substudy [18] | Galleri | Case-control, independent validation set | 4,077 (2,823 with cancer, 1,254 without) | Annals of Oncology 2021;32(9):1167-77 |
| Cancerguard development and validation studies [15] | Cancerguard | Case-control; headline figures exclude breast and prostate | Development: 729 cancers, 2,434 controls. Validation: 324 cancers, 800 controls | AACR 2026, abstract 1108 (Cancer Res 2026;86(7_Suppl)) |
| Achieve 1 [19] | Caris Detect | Enrolled on the basis of high-risk screening, symptoms, or a mass on imaging; not a screening population | 3,014 evaluable | Caris press release March 31, 2026; not peer reviewed |
| Shield MCD case-control study [10] | Shield MCD | Blinded case-control | 962 evaluable (472 with cancer across 14 tumor types, 442 without) | ASCO 2025, abstract 10550 (J Clin Oncol 2025;43(16_suppl)); NCT05334069 |
| NCI Vanguard Study [16] | Shield MCD | Feasibility study for future randomized trials of multi-cancer tests | Up to 24,000, ages 45-75 | Ongoing; enrolling since July 2025 |
Each bracketed number on this page opens one of these. Values were confirmed against the linked page on September 12, 2026.