health · digital · family: it worked in the lab
every cancer,a different screening
No test can screen for multiple cancers at stage 1 from a home sample
Problem statement
Cancer screening today is organ-specific (mammography for breast, colonoscopy for colon, and PSA for prostate, which the U.S. Preventive Services Task Force leaves to an individual decision for men aged 55 to 69), invasive, facility-dependent, and covers only a handful of cancer types. Of the more than 100 types of cancer, recommended screenings cover only four (breast, cervical, colon, and lung), and most require clinical visits, specialist equipment, or blood draws. Multi-cancer early detection (MCED) tests can screen for several cancers at once, but the commercially available ones often fail to catch Stage 1 solid tumors: in the validation study of Grail's Galleri test, sensitivity was 16.8% for Stage I cancers and 40.4% for Stage II, against 99.5% specificity. No technology exists that can reliably detect 30+ cancer types at Stage 1 from a self-administered home sample (urine or breath) without clinical infrastructure.
Why this matters
Cancer is the second leading cause of death in the U.S., and the WHO counts nearly 10 million cancer deaths worldwide in 2024. Chances of survival are high when cancer is detected at Stage 1, and late-stage diagnoses drive higher rates of mortality. ARPA-H's program manager says late-stage diagnoses are projected to account for 44% of all new cancer diagnoses in the U.S. over the next 30 years, and nearly 60% of diagnosed cancers have no recommended screening test. If a simple, at-home test could screen for dozens of cancers at Stage 1, it would fundamentally shift cancer from a late-diagnosed disease to an early-detected one. ARPA-H puts the potential at millions of lives saved; its draft solicitation estimates that if half of U.S. late-stage diagnoses could be prevented by early detection, over 8 million lives would be saved over the next 30 years.
What’s been tried and why it hasn’t worked
Commercially available MCED tests often fail to catch Stage 1 solid tumors. Grail's Galleri blood test, in its clinical validation study, detected 16.8% of Stage I cancers, which means it missed about 5 out of 6 at that stage. Blood tests of this kind look for tumor DNA in the blood, and early tumors often shed too little of it: in a digital-PCR study of 640 patients, circulating tumor DNA was detected in only 48% to 73% of patients with localized colorectal, gastroesophageal, pancreatic, or breast cancer. Most cancer screenings require a hospital visit for blood testing or another laboratory procedure, which limits access. Routine screening is typically recommended only for older or at-risk people. Even with testing, most cancers are not detected through routine screens. They are caught at later stages, after symptoms have developed.
What would unlock progress
Synthetic biology offers a fundamentally different approach: rather than relying on the trace endogenous signals that tumors naturally shed, engineered biological sensors could be administered (orally or via inhalation) that actively seek out and amplify cancer-specific signals into easily detectable reporters in urine or breath. This requires: (1) programmable biosensors that can detect multiple cancer-specific molecular markers (not just one); (2) tunable reporter systems that produce distinct, quantifiable signals for different cancer types; (3) a self-administered delivery format and a home-compatible detection device (smartphone-assisted or handheld reader) that is accurate enough for clinical decision-making without laboratory infrastructure.
Entry points for student teams
A student team could design and test a synthetic gene circuit in cell culture that produces a detectable fluorescent or enzymatic reporter in response to a cancer-associated microRNA panel. A more engineering-focused team could prototype a handheld reader device for detecting synthetic reporter molecules in urine samples, benchmarking sensitivity and specificity against known concentrations. Relevant disciplines: synthetic biology, bioengineering, diagnostics, electrical engineering, human-computer interaction (for the home test UX).
Genes
Structural cousins: stuck the same way, in other fields
Sources
ARPA-H, "Platform Optimizing SynBio for Early Intervention and Detection in ONcology (POSEIDON)," ARPA-H press release, "ARPA-H launches program to develop at-home multi-cancer screening test," 2024; accessed 2026-02-23; Klein EA, Richards D, Cohn A, Tummala M, Lapham R, Cosgrove D, Chung G, Clement J, Gao J, Hunkapiller N, Jamshidi A, Kurtzman KN, Seiden MV, Swanton C, Liu MC, "Clinical validation of a targeted methylation-based multi-cancer early detection test using an independent validation set," Ann Oncol, 32(9):1167-1177, 2021, (abstract read at Europe PMC, PMID 34176681), accessed 2026-09-29; GRAIL, "Galleri test," (cited only for which study validates the product), accessed 2026-09-29; World Health Organization, "Cancer," fact sheet, 3 July 2026, accessed 2026-09-29; National Cancer Institute, "What Is Cancer?," accessed 2026-09-29; U.S. Preventive Services Task Force, "Prostate Cancer: Screening," Final Recommendation Statement, May 08, 2018, accessed 2026-09-29; Bettegowda C, Sausen M, Leary RJ, et al., "Detection of circulating tumor DNA in early- and late-stage human malignancies," Science Translational Medicine, 6(224):224ra24, 2014, (abstract read at Europe PMC, PMC4017867), accessed 2026-09-29; ARPA-H, "ARPA-H-SOL-24-109, POSEIDON Innovative Solutions Opening (ISO)," draft dated August 9, 2024, copy hosted at accessed 2026-09-29 go to source 1 ↗ go to source 2 ↗ go to source 3 ↗ go to source 4 ↗ go to source 5 ↗ go to source 6 ↗ go to source 7 ↗ go to source 8 ↗
Working notes
Notes made while writing and checking this brief, shown as written. They use working shorthand.
Related briefs: `health-cervical-cancer-screening-access-equity` (screening access as an equity problem — POSEIDON's at-home test would address this for many cancer types); `health-multiplexed-biosensor-field-translation` (multiplexed sensing challenges relevant to multi-analyte cancer detection). The `failure:not-attempted` tag reflects that ARPA-H describes the at-home synthetic MCED tests it wants as first-in-class. The `temporal:newly-tractable` tag reflects the judgment that programmable gene circuits, engineered probiotic sensors, and synthetic biology tools for mammalian systems have matured in recent years. `stakeholders:systemic` because deployment at population scale would require coordination across FDA approval, healthcare systems, insurance reimbursement, and public health infrastructure. Source-bias note: ARPA-H's framing is highly optimistic; the technical barriers to achieving the program's targets (the draft solicitation asks for at least 90% sensitivity, 99.9% specificity, and 95% tissue-of-origin accuracy) for 30+ cancer types at Stage 1 from a home test are enormous.
Reconciliation 2026-09-29: Fact repair after a site review read this write-up against its two cited sources. Both were fetched in full on 2026-09-29: the ARPA-H POSEIDON program page (https://arpa-h.gov/explore-funding/programs/poseidon) and the ARPA-H press release of August 1, 2024 (https://arpa-h.gov/news-and-events/arpa-h-launches-program-develop-home-multi-cancer-screening-test). A second press release linked from the program page, dated September 30, 2025, was also read in full (https://arpa-h.gov/news-and-events/arpa-hs-poseidon-program-kicks-develop-cancer-screening-kits-home-use). Many figures and names in the write-up were not in any of these pages, and no other source was cited for them. They were removed or replaced with what the pages say, and were not re-sourced. (1) The text said "Of the 200+ cancer types, fewer than 10 have recommended screening tests" and gave "PSA for prostate" as an example. The program page says recommended screenings account for four cancer types: breast, cervical, colon and lung. The sentence now says that, and the prostate example is gone. (2) The text said blood tests "like Grail's Galleri" reach "only 17–44% sensitivity for Stage 1 cancers", and later that "Galleri's 17% sensitivity for Stage 1 means it misses 5 out of 6 early-stage cancers." ARPA-H names no test and gives no sensitivity figure. It says commercially available MCED tests often fail to catch Stage 1 solid tumors. The text now says that. (3) Three figures in "Why This Matters" were not in the sources: "nearly 10 million" cancer deaths a year worldwide, five-year survival that "exceeds 90%" at Stage 1 and "drops below 30%" at Stage 4, and "over 40% of cancers are diagnosed at Stage 3 or 4." The section now uses figures from the 2024 press release: cancer is the second leading cause of death in the U.S., late-stage diagnoses are projected to be 44% of new U.S. cancer diagnoses over the next 30 years, and nearly 60% of diagnosed cancers have no recommended screening test. The 44% figure is a projection quoted from the program manager, and it is for the United States. (4) "The potential lives saved are in the millions annually" is now "millions of lives saved." The program page asks whether a simple test could save millions of lives and does not say per year. (5) The section on what has been tried named Grail Galleri, Exact Sciences, Freenome and CancerSEEK, explained why blood tests and protein panels miss early tumors, and said no urine or breath test has achieved clinical validation. None of this is in the ARPA-H pages. The section now reports what the program page says about current screening. (6) In the notes above, "no prior serious attempt exists" now says that ARPA-H calls the tests it wants first-in-class. The words "is verified" and "since 2018" were removed from the sentence about the newly-tractable tag, because the sources give no date. ">90% sensitivity" is now "high sensitivity and specificity", because the pages give no numeric target. Confirmed against the sources and left as written: the program name and the press release title and year in the Source line; 30+ cancers at Stage 1; breath or urine samples; a self-administered home kit; sensors taken by mouth or inhaled, tunable reporters, and a smartphone-assisted or handheld reader (these are in the 2025 press release). NOT checked: the suggestions for student teams, which are advice and are not attributed to ARPA-H; and the POSEIDON funding solicitation, which the press release mentions but the write-up does not cite. Not in the ARPA-H pages and left as written: the examples of mammography and colonoscopy, and the statement that tumors shed only trace signals of their own. Genes untouched.
Reconciliation 2026-09-29 (re-sourcing): The editor asked for the claims removed in round 5 to be re-sourced, so that anything put back rests on a source that was fetched and read. Six things came back, each at the scope of its source. (1) The Galleri figures. The validation study of Grail's multi-cancer blood test reports sensitivity of 16.8% for Stage I cancers, 40.4% for Stage II, and 99.5% specificity (Klein et al., Annals of Oncology 32(9):1167-1177, 2021; abstract read at Europe PMC, PMID 34176681, 2026-09-29). That the test in this study is Galleri comes from Grail's own product page (https://grail.com/galleri-test/, read 2026-09-29), which is cited only for that. The original text said "17-44%"; the sourced figures are 16.8% and 40.4%, and "misses 5 out of 6" is now stated as about 5 out of 6, which is what 16.8% works out to. (2) The prostate example. PSA screening is back as an example, with the U.S. Preventive Services Task Force's scope: it leaves the decision to the individual for men aged 55 to 69 (Final Recommendation Statement, May 08, 2018, https://www.uspreventiveservicestaskforce.org/uspstf/recommendation/prostate-cancer-screening, read 2026-09-29). (3) The count of cancer types. The original said "200+"; the National Cancer Institute says there are more than 100 types of cancer (https://www.cancer.gov/about-cancer/understanding/what-is-cancer, read 2026-09-29), so the text now says that. The "fewer than 10 have recommended screening tests" phrasing stays out; the sentence keeps ARPA-H's count of four. (4) Global deaths. The WHO cancer fact sheet dated 3 July 2026 says cancer accounted for nearly 10 million deaths in 2024 (https://www.who.int/news-room/fact-sheets/detail/cancer, read 2026-09-29). The text now gives that figure with its year. (5) Why blood tests miss early tumors. A digital-PCR study of 640 patients found circulating tumor DNA in only 48% to 73% of patients with localized colorectal, gastroesophageal, pancreatic, or breast cancer (Bettegowda et al., Science Translational Medicine 6(224):224ra24, 2014; abstract read at Europe PMC, PMC4017867, 2026-09-29). The text now says that, at that scope, in place of the original "often below detection limits." (6) The program's numeric target and its lives-saved estimate. The POSEIDON solicitation (ARPA-H-SOL-24-109, draft dated August 9, 2024, read 2026-09-29 from a copy hosted by the University of Colorado Anschutz at https://research.cuanschutz.edu/docs/librariesprovider148/comirb_documents/draft-solicitation.pdf?sfvrsn=cc9366b4_1) sets performance metrics of at least 90% sensitivity, 99.9% specificity, and 95% tissue-of-origin accuracy, and estimates that if half of U.S. late-stage diagnoses could be prevented by early detection, over 8 million lives would be saved over the next 30 years. Both are now in the text, marked as coming from the draft solicitation; the final ISO PDF on SAM.gov has been deleted and no archived copy was found. Searched for and left out: the survival figures (over 90% at Stage 1, under 30% at Stage 4, "for most types"), because the SEER all-sites page gives a single 70.5% five-year relative survival with no split by stage and no primary source was found for a figure across most cancer types; "over 40% of cancers are diagnosed at Stage 3 or 4" as a present-day count, because the only figure found is the solicitation's 44% projection over the next 30 years, which the text already carries; "millions of lives annually," because no source says per year (the solicitation's estimate is over 30 years); the statement that CancerSEEK and protein panels "face high false-positive rates," because the CancerSEEK paper says the opposite, reporting specificity greater than 99% with 7 of 812 healthy controls testing positive (Cohen et al., Science, 2018, abstract read at Europe PMC, PMC6080308); the names Exact Sciences and Freenome and the claim that no urine or breath test has reached clinical validation, for which no source was fetched; and the "since 2018" date on the newly-tractable tag, for which no source gives a date. Genes untouched.