MCED Blood Tests for Multi-Cancer Detection: How Close Are We?

THE PROTOHUMAN PERSPECTIVE#

A single blood draw that flags cancer before symptoms appear — this isn't speculative futurism anymore. It's a clinical reality being deployed at scale, with over 100,000 tests already administered in real-world settings. The implications for human healthspan optimization are hard to overstate.

Right now, only four cancers have recommended screening programs in the US. That leaves roughly 83% of cancer deaths coming from types we don't routinely screen for at all. MCED tests represent a fundamental shift: instead of waiting for organ-specific symptoms, you interrogate the blood for pan-cancer signals. For anyone tracking their biological age, optimizing autophagy pathways, or investing in longevity protocols, cancer remains the single largest wildcard threat. No amount of NAD+ supplementation or HRV optimization matters if an undetected malignancy is silently metastasizing. These tests don't replace those strategies — they fill the gap that no biomarker panel or wearable currently covers. The technology is imperfect, but the trajectory is unmistakable: blood-based cancer surveillance is becoming a routine part of proactive health management.

THE SCIENCE#

What MCED Tests Actually Measure#

Multi-cancer early detection tests interrogate several classes of tumor-derived biomarkers circulating in peripheral blood. The most validated approach, used by the Galleri test (GRAIL, Inc.), relies on targeted methylation sequencing of cell-free DNA (cfDNA) — fragments of tumor DNA shed into the bloodstream that carry cancer-specific epigenetic signatures ^[3]. Machine learning algorithms then classify these methylation patterns to both detect a cancer signal and predict the anatomical cancer signal origin (CSO).

Other platforms take different molecular routes. OncoSeek integrates a panel of seven protein tumor markers (PTMs) with AI algorithms, achieving its detection at a reagent cost under $25 per test ^[1]. The Exact Sciences platform combines methylated DNA markers with protein biomarkers and somatic mutation analysis in a multi-target "reflex" strategy ^[2]. And CANSCAN, published in Nature Medicine, uses whole-genome sequencing to analyze multidimensional cfDNA fragmentomics — the size, distribution, and fragmentation patterns of circulating DNA rather than just its sequence ^[5].

Each approach exploits a different biological signal. But they share one core insight: tumors leak molecular information into the blood long before they become clinically apparent.

The Numbers That Matter#

Let me lay out the performance data across the major platforms, because this is where things get both promising and complicated.

OncoSeek, validated across 15,122 participants from seven centers in three countries, demonstrated an AUC of 0.829, with 58.4% sensitivity and 92.0% specificity. It detected 14 cancer types accounting for 72% of global cancer deaths, with per-cancer sensitivities ranging from 38.9% to 83.3%. In symptomatic patients specifically, sensitivity jumped to 73.1% at 90.6% specificity ^[1].

Galleri, evaluated across 111,080 real-world tests, showed a cancer signal detection rate of 0.91%. Among cases with reported outcomes, 258 had confirmed invasive cancers spanning 32 types. The test correctly predicted the cancer signal origin in 87% of cases, with a median 39.5 days from result to diagnosis. Specificity was 99.5% ^[3].

Exact Sciences' MP-reflex classifier achieved 55.2% overall sensitivity at 98.5% specificity in a prospectively collected cohort, with particular strength in cancers with poor 5-year survival ^[2].

The Sensitivity Problem — Which Is Annoying, Actually#

Here's the thing most headlines skip: overall sensitivity of 50–58% means these tests miss roughly half of all cancers. That's the uncomfortable truth sitting underneath the excitement. A specificity of 92–99.5% is genuinely impressive — it means false positives are rare — but sensitivity is the metric that determines whether the test catches your cancer.

The sensitivity varies dramatically by cancer type and stage. Late-stage cancers shed more cfDNA and more protein markers, so they're easier to detect. Early-stage cancers — the ones you most want to catch — are harder. This is the central tension of MCED: the clinical value proposition depends on early detection, but early-stage sensitivity is precisely where performance drops.

I'm less convinced by the case-control cohort designs used in some of these validations. Case-control studies, where you assemble known cancer patients and known healthy controls, tend to inflate performance metrics compared to prospective screening in asymptomatic populations. The Galleri real-world data is more informative here because it reflects actual clinical deployment, but even that dataset has selection bias — these are people who chose to get tested and whose doctors chose to order the test ^[3].

The Janus Particle Wildcard#

A genuinely novel approach published in Nature Biomedical Engineering in March 2026 sidesteps DNA entirely. Kumar et al. developed an assay using Janus particles — di-hemispherical microspheres with one fluorescent side — that detect cancer-derived small extracellular vesicles (sEVs) directly in plasma ^[4].

The mechanism is elegant: when vesicles bind to the particles, their size alters the Brownian rotation-induced blinking frequency. Smaller proteins don't change the frequency, ensuring selectivity without isolation steps. Using less than 10 μl of sample, the assay detects approximately 200 vesicles per microlitre in under one hour.

In a blind study of 87 subjects with colorectal cancer, pancreatic ductal adenocarcinoma, glioblastoma, Alzheimer's disease, and healthy controls, it achieved AUCs of 0.90–0.99 for disease identification ^[4]. That's a small study, and I'd want to see this replicated in hundreds or thousands of patients before getting too excited. But the sensitivity and speed — 100x better than ultracentrifugation plus surface plasmon resonance — suggest a fundamentally different detection paradigm.

Proteomics as Triage#

Separately, a Swedish study profiled 1,463 plasma proteins in patients presenting with non-specific symptoms and developed a cancer prediction model with an AUC of 0.80, validated in an independent replication cohort at 0.82 ^[6]. The team identified 29 proteins associated with new cancer diagnoses. What makes this approach distinctive is its intended use case: not population-level screening, but triage of symptomatic patients whose diffuse complaints don't point to any specific organ. The model distinguished cancer from autoimmune, inflammatory, and infectious diseases — which is clinically useful given how much diagnostic overlap exists in that patient population.

COMPARISON TABLE#

THE PROTOCOL#

For individuals interested in integrating MCED testing into their health surveillance strategy, here's a practical framework based on currently available evidence.

Step 1: Assess your baseline risk. MCED tests are most clinically useful for individuals aged 50+ where annual cancer incidence approaches 1% ^[2]. If you're younger with no family history, the pre-test probability is low enough that even a highly specific test will generate a meaningful proportion of false positives relative to true positives. Know your baseline before testing.

Step 2: Continue all guideline-recommended single-cancer screenings. MCED tests are designed to complement — not replace — mammography, colonoscopy, cervical screening, and low-dose CT for lung cancer in high-risk individuals. The evidence is unambiguous on this point ^[3]. Skipping a colonoscopy because you got a negative MCED result is a bad trade.

Step 3: Choose an available MCED test based on your geography and goals. In the US, Galleri is currently the only commercially available MCED test, ordered through a healthcare provider. Its 99.5% specificity and cancer signal origin prediction make it the most validated option for asymptomatic screening. OncoSeek is designed for lower-resource settings and carries CE-IVD marking. Discuss options with a physician who understands the test's limitations.

Step 4: Establish a testing cadence. Based on current evidence, annual or biennial testing appears reasonable for individuals at elevated risk, though optimal screening intervals are not yet established in prospective mortality-reduction trials. The honest answer is we don't have randomized data yet showing that MCED testing reduces cancer deaths — that evidence is still being generated in trials like NHS-Galleri.

Step 5: Interpret results with appropriate context. A negative result does not rule out cancer — sensitivity ranges from 38.9% to 83.3% depending on cancer type ^[1]. A positive result requires diagnostic workup guided by the predicted tissue of origin. In the Galleri real-world data, median time from positive result to cancer diagnosis was 39.5 days ^[3]. Do not panic; do act promptly.

Step 6: Track your results longitudinally. Serial testing may improve detection over time by catching cancers that were below the detection threshold at earlier timepoints. This is speculative but biologically plausible — as a tumor grows, it sheds more biomarkers. Maintain a personal health record that logs MCED results alongside standard screening outcomes and relevant blood biomarkers (CRP, LDH, CA-125 if applicable).

What is a multi-cancer early detection (MCED) blood test?#

An MCED test is a blood-based assay that screens for signals from multiple cancer types simultaneously, typically by analyzing cell-free DNA methylation patterns, protein biomarkers, or extracellular vesicles in plasma. Unlike organ-specific screening (mammography for breast, colonoscopy for colon), these tests aim to detect a broad range of cancers — up to 32 types — from a single blood draw. They use machine learning algorithms to both identify cancer signals and predict the tissue of origin.

How accurate are current MCED tests at detecting early-stage cancer?#

Overall sensitivity across available MCED tests ranges from approximately 50% to 58%, meaning they catch roughly half of cancers present at the time of testing ^[1][2]. Sensitivity is generally lower for early-stage (stage I–II) cancers and higher for late-stage disease. Specificity is consistently high, ranging from 92% to 99.5%, which means false positives are uncommon. The Janus particle approach showed AUCs of 0.90–0.99 in a small pilot study, but this requires much larger validation ^[4].

Who should consider getting an MCED test?#

Current evidence best supports testing in individuals aged 50 and older, where cancer prevalence is highest and the positive predictive value of the test is most favorable. People with elevated cancer risk — family history, prior cancer, known genetic predispositions — may benefit most. However, no regulatory body has yet issued formal screening recommendations for MCED tests. They should be used alongside, not instead of, existing recommended cancer screenings ^[3].

When will MCED tests be able to detect all cancers?#

Honestly, we don't know yet. Current tests cover 14–32 cancer types, but sensitivity for individual cancers varies widely. Some cancers shed minimal biomarkers into blood, making detection inherently difficult. The convergence of cfDNA analysis, proteomics, and extracellular vesicle detection may eventually cover more cancer types, but universal detection at early stages remains an engineering and biological challenge. Prospective mortality-reduction trials are still ongoing.

How much do MCED blood tests cost?#

The Galleri test costs approximately $949 in the US and is generally not covered by insurance. OncoSeek was specifically designed for low- and middle-income countries with a reagent cost under $25 per test ^[1], though the full clinical cost including analysis and interpretation would be higher. As multiple platforms compete and evidence accumulates, costs are expected to decrease — but widespread insurance coverage likely depends on mortality-reduction data from ongoing trials.

VERDICT#

7.5/10. The science is real, the trajectory is clear, and the clinical deployment of MCED tests at scale (100,000+ Galleri tests) marks a genuine inflection point in cancer screening. But I can't give this higher marks when overall sensitivity hovers around 50–58% and we still lack prospective data proving that MCED testing actually reduces cancer mortality. The Janus particle technology is the most exciting development in this space — if it validates at scale, it could fundamentally change detection thresholds. OncoSeek's sub-$25 price point is a legitimate breakthrough for global health equity. The technology works. The question is whether it works well enough, often enough, to change outcomes at the population level. That answer is probably yes, eventually — but we're being honest about what we know and what we don't.