Skip to content
MRD Watch

Wednesday, November 18, 2026

Subscribe
Sponsor this briefing. One site and newsletter slot, $250/week.

For clinicians

Cross-assay MRD/ctDNA limit-of-detection table

Last updated November 17, 2026

Prepared 2026-11-17 (CT). Every figure was checked live on 2026-11-17 at the URL given.

1 ppm = 0.0001% (VAF or tumor fraction). ppm = % x 10,000. Molecules/mL, copies/uL and cell counts cannot be converted without input-DNA data.

All explainers ยท Tissue-free vs tumor-informed

Vendor and study figures are not head-to-head. Definitions differ: LOD95 (lowest level detected in 95% of replicates) vs detection threshold vs 'observed' detection vs LOQ or reporting floor; ppm vs VAF vs methylation-based tumor fraction vs molecules/mL; LOD depends on input DNA (ng of cfDNA) and number of variants tracked; most figures come from contrived (cell-line) dilutions and many are vendor-reported. Lower LOD does not by itself mean better clinical sensitivity or outcomes. Rows are grouped by approach; order within a group carries no meaning and is not a ranking.

Tumor-informed

NeXT Personal

Personalis; U.S. commercialization with Tempus

tumor-informed; Tumor + matched-normal WGS, personalized panel sequenced in plasma

  • Detection threshold (mean): 1.67 PPM

    ppm: 1.67

    VAF or TF %: 0.000167

    1.67 ppm x 0.0001 = 0.000167%

  • LOD95: 3.45 PPM

    ppm: 3.45

    VAF or TF %: 0.000345

    3.45 ppm x 0.0001 = 0.000345%

LOQ / reporting floor
Linearity 0.8 to 300,000 PPM (not stated as an LOQ)
Variants / input
Up to ~1,800 somatic variants
Evidence
peer-reviewed analytical validation (all authors Personalis employees)
Medicare (A58454, R20)
Named in A58454: neoadjuvant breast; recurrence/relapse with personal history of cancer for NSCLC and advanced breast; ICI-response monitoring in stage IV solid tumors. A58454
Source(s), checked

Signatera (exome-based, bespoke multiplex PCR)

Natera

tumor-informed; Tumor + matched-normal WES, 16 clonal variants by multiplex PCR, ultra-deep sequencing

  • Limit of detection: not independently published

    ppm: not independently published

    VAF or TF %: not independently published

    not independently published

LOQ / reporting floor
not independently published
Variants / input
not independently published
Evidence
not independently published
Medicare (A58454, R20)
Named in A58454: advanced colorectal; muscle-invasive bladder and ovarian; neoadjuvant breast; NSCLC and advanced breast recurrence; ICI-response monitoring (advanced solid tumors). A58454
Source(s), checked

    Signatera Genome

    Natera

    tumor-informed; WGS-based design

    • Limit of detection: not independently published

      ppm: not independently published

      VAF or TF %: not independently published

      not independently published

    LOQ / reporting floor
    Not published
    Variants / input
    not independently published
    Evidence
    not independently published
    Medicare (A58454, R20)
    Not named separately in A58454 (Signatera listings do not specify Genome). A58454
    Source(s), checked

      RaDaR ST

      NeoGenomics

      tumor-informed; Tumor sequencing to select variants; patient-specific amplicon panel

      • Limit of detection: not independently published

        ppm: not independently published

        VAF or TF %: not independently published

        not independently published

      LOQ / reporting floor
      not independently published
      Variants / input
      Up to 48 tumor-specific variants (Gale 2022, original RaDaR)
      Evidence
      variant count is peer-reviewed; the limit of detection is not independently published
      Medicare (A58454, R20)
      Named in A58454: HPV-negative HNSCC; advanced breast recurrence; ICI-response monitoring in stage IV solid tumors (added in R20, effective 4/9/2026). A58454

      Haystack MRD

      Quest Diagnostics (Haystack Oncology)

      tumor-informed; Tracks patient-specific somatic mutations with duplex error correction

      • LOD95 (peer-reviewed): 0.063 mean ctDNA molecules/mL

        ppm: not convertible

        VAF or TF %: not convertible

        Not convertible: molecules per mL of plasma depend on total cfDNA per mL

      LOQ / reporting floor
      Analytical measurement range 0.25 to 1000 mean ctDNA molecules/mL
      Variants / input
      Not stated in abstract
      Evidence
      peer-reviewed analytical + clinical validation (Quest/Haystack authors)
      Medicare (A58454, R20)
      Not named in A58454. A58454
      Source(s), checked

      Oncodetect (next-generation, WGS + MAESTRO)

      Exact Sciences

      tumor-informed; WGS-based design with MAESTRO enrichment

      • Limit of detection: not independently published

        ppm: not independently published

        VAF or TF %: not independently published

        not independently published

      LOQ / reporting floor
      Not published
      Variants / input
      not independently published
      Evidence
      not independently published
      Medicare (A58454, R20)
      Named in A58454 as Oncodetect for advanced colorectal cancer. Whether that listing covers a later version is not stated in the Medicare article. A58454
      Source(s), checked

        Pathlight

        SAGA Diagnostics

        tumor-informed; Tumor WGS to find structural variants; personalized multiplex digital PCR

        • LoD95 at 70 ng cfDNA input: 0.00052% (paper rounds to 5 ppm)

          ppm: 5.2

          VAF or TF %: 0.00052

          0.00052% / 0.0001 = 5.2 ppm

        • LoD95 at 300 ng input: 0.00011% (1.1 ppm)

          ppm: 1.1

          VAF or TF %: 0.00011

          0.00011% / 0.0001 = 1.1 ppm

        LOQ / reporting floor
        Not published
        Variants / input
        Up to 16 SVs per tumor (median 14, range 4-16 in the study); LoD95 at 70 ng and 300 ng input
        Evidence
        peer-reviewed (cell-line dilution, probit)
        Medicare (A58454, R20)
        Named in A58454: advanced breast recurrence/relapse with personal history of cancer. A58454

        Precise MRD

        Myriad Genetics

        tumor-informed; WGS-based personalized panel

        • Limit of detection: not independently published

          ppm: not independently published

          VAF or TF %: not independently published

          not independently published

        LOQ / reporting floor
        Not published
        Variants / input
        not independently published
        Evidence
        not independently published
        Medicare (A58454, R20)
        Not named in A58454. A58454
        Source(s), checked

          Invitae Personalized Cancer Monitoring (PCM)

          Invitae (select assets acquired by Labcorp, 2024)

          tumor-informed; Tumor + germline WES; patient-specific panel

          • LOD, 60 ng input, 18-50 variants, baseline threshold: 0.008% AF

            ppm: 80

            VAF or TF %: 0.008

            0.008% / 0.0001 = 80 ppm

          • LOD, 10 ng input, 18 variants, monitoring threshold: 0.05% AF

            ppm: 500

            VAF or TF %: 0.05

            0.05% / 0.0001 = 500 ppm

          LOQ / reporting floor
          Not published
          Variants / input
          18-50 variants; 10-60 ng cfDNA
          Evidence
          peer-reviewed analytical validation (Invitae authors)
          Medicare (A58454, R20)
          Not named in A58454. A58454
          Source(s), checked

          FoundationOne Tracker

          Foundation Medicine

          tumor-informed (from tissue CGP results); Personalized plasma assay from FoundationOne tissue CGP alterations

          • Analytical sensitivity (not an LOD95 in ppm): >97.3% at >=5 mean tumor molecules/mL (2 monitorable alterations)

            ppm: not convertible

            VAF or TF %: not convertible

            Not convertible (molecules/mL)

          LOQ / reporting floor
          Not published
          Variants / input
          Depends on tissue CGP report
          Evidence
          peer-reviewed analytical validation
          Medicare (A58454, R20)
          Not named in A58454. A58454
          Source(s), checked

          Tumor-naive / tissue-free

          Guardant Reveal

          Guardant Health

          tumor-naive (tissue-free); Plasma-only, methylation-based tumor fraction

          • LOD95 at 3-4 ng cfDNA (as cited in a peer-reviewed paper): 0.02%

            ppm: 200

            VAF or TF %: 0.02

            0.02% / 0.0001 = 200 ppm (tumor fraction)

          • LOD95 at 15 ng cfDNA (as cited in a peer-reviewed paper): 0.005%

            ppm: 50

            VAF or TF %: 0.005

            0.005% / 0.0001 = 50 ppm (tumor fraction)

          LOQ / reporting floor
          Minimum reported tumor fraction 0.05% (500 ppm) in the lung assay, per the MGB methods section: July 2023 range 0.05%-8.7% with LOD 0.05%; December 2024 range 0.05%-10% with LOD lowered to 0.01%, minimum reported value still 0.05%.
          Variants / input
          Fixed methylation panel; 3-4 ng and 15 ng cfDNA inputs in LOD studies
          Evidence
          peer-reviewed citation of an analytical abstract, plus a reporting floor from an independent cohort
          Medicare (A58454, R20)
          Named in A58454: advanced colorectal cancer (progression, recurrence, relapse). A58454

          Northstar Response

          BillionToOne

          tumor-naive (tissue-free); treatment-response monitoring; Methylation panel

          • Limit of detection: not independently published

            ppm: not independently published

            VAF or TF %: not independently published

            not independently published

          LOQ / reporting floor
          Not published
          Variants / input
          not independently published
          Evidence
          not independently published
          Medicare (A58454, R20)
          Not named in A58454. A58454
          Source(s), checked

            NavDx

            Naveris

            virus-informed, tissue-free (HPV-driven cancers only); Tumor tissue-modified viral HPV DNA (types 16, 18, 31, 33, 35) by digital PCR

            • LOD range across HPV genotypes: 0-1.10 copies/uL

              ppm: not convertible

              VAF or TF %: not convertible

              Not convertible (copies per microliter)

            LOQ / reporting floor
            LOQ <1.20-4.11 copies/uL; LOB 0-0.32 copies/uL
            Variants / input
            5 HPV genotypes
            Evidence
            peer-reviewed (Naveris authors)
            Medicare (A58454, R20)
            Named in A58454: HPV-driven oropharyngeal and anal squamous cell cancers (recurrence/relapse). A58454
            Source(s), checked

            Blood-cancer assays (reference only, different biology)

            clonoSEQ (blood cancers; reference only)

            Adaptive Biotechnologies

            tumor-informed immune-receptor sequencing (not solid-tumor ctDNA); Patient-specific Ig/TCR rearrangements

            • LoD: 1.903 malignant cells

              ppm: not convertible

              VAF or TF %: not convertible

              Not convertible without stated DNA input (cells, not a fraction)

            LOQ / reporting floor
            LoQ 2.390 malignant cells; LoB 0
            Variants / input
            Depends on DNA input
            Evidence
            peer-reviewed (Adaptive author)
            Medicare (A58454, R20)
            Not in A58454 (solid-tumor article). A58454
            Source(s), checked

            Foresight CLARITY (PhasED-Seq; lymphoma; reference only)

            Foresight Diagnostics

            tumor-informed phased variants; PhasED-Seq

            • LOD95 at 120 ng input: 0.7 parts per million

              ppm: 0.7

              VAF or TF %: 7e-05

              0.7 ppm x 0.0001 = 0.00007%

            LOQ / reporting floor
            Not published
            Variants / input
            120 ng input DNA
            Evidence
            peer-reviewed (Foresight authors), B-cell malignancies
            Medicare (A58454, R20)
            Not in A58454 (solid-tumor article). A58454
            Source(s), checked

            Row notes

            • NeXT Personal: Also called xM (NeXT Personal Dx). 1.67 PPM is the paper's mean detection threshold; LOD95 is 3.45 PPM.
            • Signatera (exome-based, bespoke multiplex PCR): No peer-reviewed analytical-validation paper was found for this limit of detection.
            • Signatera Genome: Cell-line dilution figures were vendor data on file. Not independently published.
            • RaDaR ST: The 11 ppm LOD95 was vendor data on file and is not shown.
            • Haystack MRD: The paper reports molecules per mL. A percent figure on the company site is not independently published and is not shown.
            • Oncodetect (next-generation, WGS + MAESTRO): The stated limit of detection came from a company page and is not shown.
            • Precise MRD: The stated limit of detection came from a company page and is not shown.
            • Invitae Personalized Cancer Monitoring (PCM): Current commercial name/availability under Labcorp not re-verified today.
            • Guardant Reveal: Tumor fraction, not variant allele fraction. A company-page limit of detection is not shown. The 0.05% floor is documented for the lung assay in the Mass General Brigham paper.
            • Northstar Response: The stated limit of detection came from a company page and is not shown.
            • clonoSEQ (blood cancers; reference only): Regulatory status beyond the peer-reviewed methods paper is not restated from a company site.

            Mass General Brigham transient-positive study

            Fluctuating Methylation-based ctDNA Detection in NSCLC Surveillance: Implications for Serial Monitoring and Specificity, Cho JM, et al. (Mass General Brigham / Dana-Farber), JTO Clinical and Research Reports, 2026 (Collection Dec 2026), PMID 43036776 (checked 2026-11-17). DOI.

            Assay: Guardant Reveal (named in Methods; abstract says 'a commercially available methylation-based ctDNA test').

            Design: Retrospective, single center; 85 patients after lung resection or SBRT for NSCLC; surveillance testing 2024-2025.

            • 17 of 85 (20%) had at least one positive test
            • 10 of 17 (59%) later tested negative with no recurrence at the time
            • 11 of 17 (65%) were positive only at or near 0.05% tumor fraction, the minimum reported value, without radiographic recurrence; one patient had 3 serial low-level positives without recurrence
            • 3 patients with tumor fractions above that level (>10%, 0.25%, 0.55%) all had recurrence; 1 more patient at 0.2% later tested negative and had no recurrence on imaging
            • No recurrences in the ctDNA-negative group during follow-up

            Authors' conclusion: Isolated low-level positivity may not warrant clinical action; results should be read as trends.

            Limits: Retrospective, small, single center, short follow-up, no systematic comparison with mutation-based testing, only 3 SBRT patients.

            Funding: DF/HCC Lung SPORE Career Enhancement award; no Guardant funding or disclosure found.

            Context: Transient early positives are not unique to tissue-free tests: a tumor-informed NSCLC study (Gale 2022, RaDaR) found ctDNA 1-3 days after surgery in 25% of patients that was not associated with recurrence. Gale 2022.

            Left out (could not verify a number today)

            • Tempus xM (tumor-naive, CRC): No analytical LOD published (Nakamura CCR 2025 reports clinical performance only: landmark sensitivity 61.1%, specificity 87.9%; longitudinal 83.3%/89.5%). Do not apply NeXT Personal figures. https://pubmed.ncbi.nlm.nih.gov/39513962/ (checked 2026-11-17)
            • Northstar Clear (tissue-free MRD): No limit of detection is independently published. (checked 2026-11-17)
            • Guardant Reveal Ultra: 'LOD at one part per million or below' appears only in a Q3 2026 earnings-call transcript in the dossier; no live spec sheet or paper found. /workspace/dossiers/guardant-reveal/2026-11-06_q3_transcript_ultra.txt (checked 2026-11-17)
            • Veracyte TrueMRD: No commercial LOD found in a live primary source (research MRDetect paper only). Medicare-listed for MIBC. https://www.cms.gov/medicare-coverage-database/view/article.aspx?articleid=58454 (checked 2026-11-17)
            • Labcorp Plasma Detect: Only a conference abstract (2024) was found earlier; not re-verified. (checked 2026-11-17)
            • Precise MRD AACR 2026 abstract 2598 (<5 ppm genome equivalents): AACR page timed out / returned 403 today; vendor 1.45 ppm kept instead, labeled vendor. https://doi.org/10.1158/1538-7445.AM2026-2598 (checked 2026-11-17)

            Medicare source

            CMS MolDX billing article A58454, revision R20, revision effective 10/01/2026, revision ending N/A (checked 2026-11-17). Coverage is indication-specific; a listing means the test met MolDX criteria for the named uses only.

            Fact-only. Tempus-marketed assays appear like any other vendor's. No ranking.