For clinicians
Tissue-free vs tumor-informed MRD: by shedding level and clinical setting
Last updated November 17, 2026
Prepared 2026-11-17 (CT). All sources checked live on 2026-11-17.
Not a single best test. Fit depends on how much DNA the tumor sheds, the clinical setting, whether tissue is available, and the evidence for that cancer type.
Figures come from different studies, assays, cancer types, stages, sampling schedules and follow-up lengths, and are not head-to-head unless the row says so. LOD definitions differ (see lod_table.md).
The two approaches
Tumor-informed
Sequences the patient's tumor (and usually a matched normal blood sample) to build a personal panel, then looks for those variants in plasma.
Tissue needed. Yes. Example: one tumor-informed test needs an FFPE block or slides plus a blood tube for the first test.
Turnaround. Example: one tumor-informed test reports a first result 3-4 weeks after tumor tissue and blood are received, and later draws in 7-10 days. Turnaround for other tests was not checked on this date.
- Knows which variants to look for and can filter CHIP/germline with a matched normal.
- Published LOD95s range from about 1 ppm to about 100 ppm depending on assay, input DNA and number of variants (see LOD table).
- Depends on tissue quality and amount; panel design needs time.
- In a CRC meta-analysis, tumor-informed tests had higher sensitivity with serial sampling (0.88 vs 0.59); landmark (single-draw) differences were not significant; false-positive rates did not differ significantly.
- MRD Watch LOD table draft (this folder) (checked 2026-11-17)
- Camblor et al., Cancer Treat Rev 2026, PMID 41401645 (checked 2026-11-17)
Tissue-free (tumor-naive)
Uses blood only, usually a fixed methylation and/or genomic panel, with no tumor sequencing.
Tissue needed. No tumor tissue.
Turnaround. Example: one tissue-free test reports a 7-day median turnaround from sample receipt to the first result. Turnaround for other tests was not checked on this date.
- Available when tissue is missing, too small, or old.
- Published limits of detection are generally higher (tens to hundreds of ppm) than recent tumor-informed whole-genome assays. Figures are not head-to-head.
- Low-level results near the reporting floor can be transient. In a lung cohort, 10 of 17 detected results later tested undetectable.
- Sensitivity varies with where the cancer recurs. In one colorectal study of a tissue-free test, detection was 100% for liver, 53% for lung, and 40% for the peritoneum.
- MRD Watch LOD table draft (this folder) (checked 2026-11-17)
- Cho et al. (MGB), JTO Clin Res Rep 2026, PMID 43036776 (checked 2026-11-17)
- Parikh et al., Clin Cancer Res 2024, PMID 39110016 (checked 2026-11-17)
By tumor shedding level
Higher-shedding (examples: colorectal, bladder, ovarian, pancreatic, gastroesophageal, head and neck, melanoma, liver; in advanced disease)
Evidence. ctDNA detected in >75% of patients with these advanced cancers; in localized disease detection was 73% (CRC), 57% (gastroesophageal), 48% (pancreatic), 50% (breast).
What it means. More tumor DNA in blood narrows the practical gap between approaches. Most published tissue-free MRD data so far are in colorectal cancer.
- Bettegowda et al., Sci Transl Med 2014, PMID 24553385 (checked 2026-11-17)
Lower-shedding (examples: primary brain, kidney, prostate, thyroid; early-stage lung adenocarcinoma)
Evidence. ctDNA was detected in fewer than half of patients with primary brain, kidney, prostate, or thyroid cancers in an advanced-disease series. In early-stage lung adenocarcinoma, a whole-genome tumor-informed assay detected preoperative ctDNA in 81%, including 53% of stage I; patients with levels below 80 ppm still had worse survival than patients with undetectable ctDNA. In one non-small cell lung cancer study, pretreatment detection was 24% at stage I.
What it means. When signal is very low, assay LOD, input DNA and variant number matter more. An undetectable result in a low shedder is less reassuring regardless of approach.
- Bettegowda et al., Sci Transl Med 2014, PMID 24553385 (checked 2026-11-17)
- Black et al., Nat Med 2025, PMID 39806071 (checked 2026-11-17)
- Gale et al., Ann Oncol 2022, PMID 35306155 (checked 2026-11-17)
Breast (intermediate; varies by subtype)
Evidence. In one breast study, 15 of 34 ctDNA-positive patients (44.1%) first turned positive below 100 ppm. In triple-negative breast cancer, a tissue-free assay and a multivariant tumor-informed assay had similar lead times (7.6 vs 7.1 months; P = .15); among 41 patients detected by both, 12 were detected earlier by the tumor-informed assay and 1 earlier by the tissue-free assay.
What it means. Many detections sit in ultra-low ranges; both approaches have prognostic data in TNBC.
- ALIENOR, ESMO Open 2026, PMID 42242004 (checked 2026-11-17)
- c-TRAK TN tissue-free vs tumor-informed, JAMA Oncol 2026, PMID 42593771 (checked 2026-11-17)
By clinical setting
Post-surgical MRD (landmark, about 2-12 weeks after surgery)
Tumor-informed
Largest randomized and cohort evidence so far: DYNAMIC (ctDNA-guided stage II colon cancer: chemotherapy in 15% vs 28%, non-inferior 2-year recurrence-free survival 93.5% vs 92.4%); a postsurgical cohort in stage II-III colorectal cancer reported a hazard ratio of 10.82 for ctDNA positivity. The first result waits on tissue, so it comes weeks after surgery.
Tissue-free
One tumor-naive colorectal assay at the landmark time point: sensitivity 61.1%, specificity 87.9%. A meta-analysis found no significant difference between tumor-informed and tumor-agnostic tests at that single time point in colorectal cancer.
Caveat. Very early draws can mislead. In one non-small cell lung cancer study, ctDNA in the first 1 to 3 days after surgery (25% of patients) was not associated with recurrence.
- Tie et al., NEJM 2022 (DYNAMIC), PMID 35657320 (checked 2026-11-17)
- Kotani et al., Nat Med 2023 (GALAXY), PMID 36646802 (checked 2026-11-17)
- Nakamura et al., Clin Cancer Res 2025, PMID 39513962 (checked 2026-11-17)
- Camblor et al., Cancer Treat Rev 2026, PMID 41401645 (checked 2026-11-17)
- Gale et al., Ann Oncol 2022, PMID 35306155 (checked 2026-11-17)
Surveillance (serial testing after treatment)
Tumor-informed
Pooled colorectal sensitivity was 0.88 with serial sampling. One tumor-informed colorectal assay reported 100% sensitivity and specificity at 3-year follow-up, and 83.3% sensitivity with 100% specificity at 5 years, in a small cohort.
Tissue-free
Pooled colorectal sensitivity was 0.59 with serial sampling. One tissue-free colorectal assay reported longitudinal sensitivity of 81% in stage II or higher colon cancer, with specificity 98.2%. Another tumor-naive assay reported longitudinal sensitivity 83.3% and specificity 89.5%. A lung cohort found frequent transient low-level detected results.
Caveat. Read results as trends; single low-level positives need confirmation.
- Camblor et al., Cancer Treat Rev 2026, PMID 41401645 (checked 2026-11-17)
- Clark et al., J Mol Diagn 2026, PMID 42624249 (checked 2026-11-17)
- Parikh et al., Clin Cancer Res 2024, PMID 39110016 (checked 2026-11-17)
- Nakamura et al., Clin Cancer Res 2025, PMID 39513962 (checked 2026-11-17)
- Cho et al. (MGB), JTO Clin Res Rep 2026, PMID 43036776 (checked 2026-11-17)
Response monitoring (advanced disease on therapy)
Tumor-informed
Medicare article A58454 lists specific tumor-informed assays for monitoring response to immune-checkpoint inhibitors.
Tissue-free
The same Medicare article lists a tissue-free assay for that monitoring. Company-only performance figures for other tissue-free monitoring tests are not repeated here.
Caveat. Higher tumor burden in advanced disease usually means more ctDNA, so LOD matters less than in MRD.
- CMS MolDX A58454, R20, revision effective 10/01/2026 (checked 2026-11-17)
No tissue available (or too little, too old, or failed sequencing)
Tumor-informed
Not possible without tumor sequencing.
Tissue-free
The practical option. c-TRAK TN authors say their data support tissue-free MRD testing in a trial setting, 'permitting ctDNA testing when tissue is not available.' ASCO 2026 guideline: ctDNA testing may be offered when a specific, evidence-based action can be taken.
Caveat. Coverage is indication-specific.
- c-TRAK TN tissue-free vs tumor-informed, JAMA Oncol 2026, PMID 42593771 (checked 2026-11-17)
- Lockwood et al., ASCO ctDNA guideline, JCO Oncol Pract 2026, PMID 42314080 (checked 2026-11-17)
- CMS MolDX A58454, R20, revision effective 10/01/2026 (checked 2026-11-17)
Fact-only. No ranking and no product names on this page.