Research report
A research report is the sourced material a domain
dossier is synthesized from — generated on a plan and a cadence, one topic per file.
A report carries no confidence tags. Its bracketed markers say who might have
an incentive to shade a line; none of them says anyone checked it. To reach a score, a line has
to be drafted onto a candidate as unverified and pass the Verifier or the
Corroborator, like everything else.
Section technology · Version 2026-09-01 · Cadence annual · Evidence class 2 published · Sources 8 · Supersedes none
Domain: ctdna-mrd-assay-validation · Scope: Public primary sources reachable by this repo's connectors, WebSearch and WebFetch (US jurisdiction, plus peer-reviewed publications regardless of the authoring institution's country). Excludes CPT descriptors, veterinary datasets, fda.gov guidance documents and case law, anything behind an unset PATENTSVIEW_API_KEY, and non-US/non-EU regulatory frameworks, per the default boundary in knowledge-base/research/_template.md.
Sourcing: eight peer-reviewed papers (2018-2026) tracing digital-PCR reference-measurement-procedure development, the recurring low-VAF precision ceiling, and the newest reference-material and sequencing technologies; no manufacturer, market-research or news source appears in this report.
Digital PCR (dPCR) moved from a research technique to the metrological backbone for characterizing low-variant-allele-fraction (VAF) ctDNA reference materials over the decade captured here, validated as an SI-traceable primary reference measurement procedure and then used both to calibrate contrived reference materials and to cross-calibrate next-generation sequencing (NGS) assays against it [1][2][4]. Across independent interlaboratory and single-laboratory studies published between 2018 and 2023, quantification precision consistently degrades and platform-specific systematic error appears as VAF falls below roughly 0.1%-0.5%, a limit reported for both droplet digital PCR (ddPCR) and NGS platforms [2][3]. Reference-material design itself has diversified since 2025 away from a single contrived VAF dilution series toward materials that each target one specific pre-analytical or analytical failure mode — native ctDNA fragmentation, fragment-length-dependent isolation recovery, and methylation status — rather than extending the same synthetic-oligo dilution-series design [5][6][7].
_Baseline (v1). No prior version; this establishes the starting point for future diffs._
By the end of this decade window, three independent methods papers had each separately established digital PCR as the primary quantification method for characterizing ctDNA reference materials at low VAF, rather than treating it as one option among several [1][2][4]. A 2018 Clinical Chemistry paper validated dPCR as an SI-traceable primary reference measurement procedure for a KRAS c.35G>A (G12D) copy-number target: concentration values varied by less than 1.2-fold across five different detection chemistries and less than 1.3-fold across four commercial dPCR platforms [single-source][1]. Verbatim: "The candidate dPCR reference measurement procedure showed linear quantification over a wide range of copies per reaction and high repeatability and interlaboratory reproducibility (CV, 2%-8% and 5%-10%, respectively)." — PMID 29903874 [1]. A 2020 interlaboratory study built a gravimetrically-mixed, SI-traceable BRAF V600E "ctDNA" reference material and used it to expose a systematic error specific to one commercial ddPCR platform (Naica Crystal), caused by uncorrected droplet volume, detectable only because the reference material's assigned value was independently known; that study set the limit of blank/detection/quantification for its ddPCR assay at 0.01%/0.02%/0.1% VAF [single-source][2]. Verbatim: "a systematic error caused by uncorrected droplet volume in Naica Crystal ddPCR platform was found by using the ctDNA RM." — PMID 31594564 [2].
A 2020 PLOS ONE paper reported that early and random PCR errors remain a principal source of NGS noise even after duplex molecular-barcode error correction, clonal-hematopoiesis filtering, and patterned-error suppression are all applied, and that sample duplicates — not barcoding alone — are necessary to remove the residual stochastic noise [single-source][3]. Verbatim: "early and random PCR errors are a principal source of NGS noise that persist despite duplex molecular barcoding... sample duplicates are necessary to eliminate the stochastic noise associated with NGS." — PMID 32084206 [3].
A 2023 American Journal of Clinical Pathology study addressed a related but distinct problem — that NGS-based cfDNA output is reported "in misleading units that are confounded by non-disease-related factors" — by calibrating NGS assays with spiked synthetic normalizer DNA and cross-referencing the result to ddPCR, using Epstein-Barr virus genome copies as the model target in n=12 patient-plasma and n=12 mock-plasma specimens [single-source][4]. Verbatim: "Next-generation sequencing was equally sensitive to ddPCR, with improved linearity when NGS values were normalized for spiked DNA read counts (R2 = 0.95 for normalized vs 0.91 for raw read concentrations)." — PMID 37244060 [4]. The authors describe the strategy as a step toward "a universal reference material to overcome biological and preanalytical variables" confounding NGS-based disease-burden quantification [single-source][4].
Three papers published within the final 18 months of this search window each target one specific failure mode rather than extending the synthetic-oligo dilution-series design the domain dossier's incumbent vendors (Seraseq, Horizon/Revvity, Twist) already use. A 2025 Genes paper prepared ctDNA reference material by digesting nucleosomal DNA from cultured cancer cell lines with micrococcal nuclease rather than synthesizing oligonucleotides, explicitly to better approximate native ctDNA, and paired it with two new digital PCR assays for TP53 R175H and R248W achieving limits of detection of 0.143% and 0.092% VAF respectively, with repeatability RSD of 0.16%-7.65% and linearity (R²) of 1.0000 and 0.9981 across a 50%-0.1% VAF range [single-source][5]. Verbatim: "this method can closely mimic the properties of clinical ctDNA." — PMID 40428398 [5]. A 2026 Journal of Molecular Diagnostics paper from a national metrology institute (TÜBİTAK UME, Turkey) targets a pre-analytical failure mode the three incumbent vendors' own pages do not address: fragment-length-dependent recovery bias during cfDNA isolation. It produced two certified reference materials — one with 80 bp and 240 bp double-stranded DNA fragments, one with 80/120/160/240 bp fragments — gravimetrically prepared and characterized by duplex and multiplex droplet digital PCR, with certified copy-number concentrations and stated measurement uncertainty assigned per fragment length, and demonstrated applicability spiked into a plasma matrix [single-source][6]. A 2026 Clinical Epigenetics paper reports that NIST and LGC Clinical Diagnostics each independently developed candidate methylated cfDNA reference materials — NIST's with five components at 0%, 5%, 25%, 50% and 100% methylation, LGC's with two components — and that six participating laboratories using qPCR, dPCR and bisulfite sequencing found a strong correlation between designed and observed methylation levels alongside assay-specific variation [paywalled][7]. This finding was read from the NIST-hosted publication summary page; the underlying Springer-hosted article itself sat behind an access gate and was not retrieved [paywalled][7].
A 2025 Nature Methods paper reports duplex error-corrected whole-genome sequencing (WGS) of cell-free DNA at roughly 120x depth, run on a stated lower-cost sequencing platform (Ultima Genomics), reaching an error rate of 7.7×10⁻⁷ and using that error suppression to assess disease burden in melanoma and urothelial-cancer patients without a matched tumor sequence at all — a tumor-naive route distinct from every tumor-informed assay the domain dossier's own analytical-validation table describes [single-source][8]. Verbatim: "we applied deep (~120×) lower-cost WGS (Ultima Genomics) for tumor-informed circulating tumor DNA detection within the part-per-million range. We further leveraged lower-cost sequencing by developing duplex error-corrected WGS of ccfDNA, achieving 7.7 × 10-7 error rates, allowing us to assess disease burden in individuals with melanoma and urothelial cancer without matched tumor sequencing." — PMID 40217113 [8].
[1] Assessment of Digital PCR as a Primary Reference Measurement Procedure to Support Advances in Precision Medicine — Clinical Chemistry (published 2018-09, day not given in the PubMed record; accessed 2026-09-01). PMID 29903874 — https://pubmed.ncbi.nlm.nih.gov/29903874/ [peer-reviewed] [2] Interlaboratory assessment of droplet digital PCR for quantification of BRAF V600E mutation using a novel DNA reference material — Talanta (published 2020-01-15; accessed 2026-09-01). PMID 31594564 — https://pubmed.ncbi.nlm.nih.gov/31594564/ [peer-reviewed] [3] The stochastic nature of errors in next-generation sequencing of circulating cell-free DNA — PLoS ONE (published 2020, month not given in the PubMed record; accessed 2026-09-01). PMID 32084206 — https://pubmed.ncbi.nlm.nih.gov/32084206/ [peer-reviewed] [4] Calibration of cell-free DNA measurements by next-generation sequencing — American Journal of Clinical Pathology (published 2023-09-01; accessed 2026-09-01). PMID 37244060 — https://pubmed.ncbi.nlm.nih.gov/37244060/ [peer-reviewed] [5] Establishment of Reference Measurement Procedure for TP53 R175H/R248W Detection and a Novel Preparation Method for ctDNA Reference Material — Genes (published 2025-05-14; accessed 2026-09-01). PMID 40428398 — https://pubmed.ncbi.nlm.nih.gov/40428398/ [peer-reviewed] [6] Certified Reference Materials for Standardization of Cell-Free DNA Isolation Recovery in Liquid Biopsy — The Journal of Molecular Diagnostics (published 2026-06, day not given in the PubMed record; accessed 2026-09-01). PMID 41936819 — https://pubmed.ncbi.nlm.nih.gov/41936819/ [peer-reviewed] [7] Development, characterization, and inter-laboratory validation of methylated human cell free DNA candidate reference materials — Clinical Epigenetics (published 2026-05-18; accessed 2026-09-01, via the NIST-hosted publication summary page, full Springer-hosted text not retrieved). DOI 10.1186/s13148-026-02156-3 — https://www.nist.gov/publications/development-characterization-and-inter-laboratory-validation-methylated-human-cell-free [peer-reviewed] [8] Error-corrected flow-based sequencing at whole-genome scale and its application to circulating cell-free DNA profiling — Nature Methods (published 2025-05, day not given in the PubMed record; accessed 2026-09-01). PMID 40217113 — https://pubmed.ncbi.nlm.nih.gov/40217113/ [peer-reviewed]
Well established: Digital PCR's role as the reference-measurement backbone for low-VAF ctDNA reference materials, and the recurring ~0.1%-0.5% VAF precision ceiling across platforms, both rest on multiple independent, non-affiliated peer-reviewed papers spanning 2018-2023 [1][2][3][4] — though every individual quantitative figure within them (each specific LOD, RSD, CV, or CI) is itself supported by only the one paper that reported it.
Thin: every specific LOD, precision figure, and reference-material-composition detail in section 3 rests on exactly one paper [single-source][1][2][3][4][5][6][8]. The methylated-cfDNA reference-material finding [7] rests on one paper, and this report read it only through a NIST-hosted summary page rather than the underlying peer-reviewed text, which sat behind an access gate [paywalled][7].
Rescoped from class 3: none in this report — every finding here is a published measured quantity (evidence class 2) or a described technical method, not a question about what a person would do.
Out of scope: CPT descriptors, veterinary datasets, fda.gov guidance documents and case law, and any PatentsView-backed patent search (unset PATENTSVIEW_API_KEY, the same structural gap the domain dossier already records for idea-003). Non-US/non-EU regulatory frameworks were not queried, though a peer-reviewed publication from a non-US metrology institute [6] is used here as a published finding, not as a claim about that country's regulatory framework. The 42 CFR 493.1256 daily-QC requirement the domain dossier already establishes in its care-pathway section is not re-derived here, to avoid duplicating a fact this report's technology-trajectory scope does not add to.
Not searched vs. not found: A search for "digital PCR uncertainty measurement ultra-low allele fraction circulating tumor DNA" was run for the domain dossier's own technology section and returned zero PubMed records — confirmed not found, not merely not searched, and not re-run here. A targeted search combining "fragmentomics" and "reference material," which the domain dossier's open-questions table flags as not yet run, was not searched in this pass either. Whether the TÜBİTAK UME [6] or NIST/LGC [7] materials described above are commercially available for purchase, as opposed to published as candidate reference materials, was not searched. A broader literature sweep on whole-genome sequencing cost trends returned mostly off-topic records (antimicrobial resistance, NICU genomics, pediatric survey data) and no on-point cost-per-genome time series; that specific quantity is recorded as not found in the window searched.
[inference] The clustering of the three most recent reference-material papers ([5], [6], [7], 2025-2026) around one specific failure mode each — native fragmentation, fragment-length-dependent isolation recovery, and methylation status — read against the earlier, broader dPCR-validation and cross-calibration papers ([1], [2], [4]), looks like a shift from single, one-size dilution-series materials toward matched-purpose components addressing one pre-analytical or analytical variable at a time. This is this report's own synthesis of a pattern across independently-published papers, not a conclusion stated in any one of them.
| Proposition | Evidence class | Resolvable identifier | Dossier section |
|---|---|---|---|
| A 2018 study validated digital PCR as an SI-traceable primary reference measurement procedure for a KRAS G12D copy-number target, with concentration values varying by <1.2-fold across 5 detection chemistries and <1.3-fold across 4 commercial platforms | 2 published | PMID 29903874 | 7 |
| A 2020 interlaboratory ddPCR study using a gravimetrically-prepared BRAF V600E ctDNA reference material found a systematic error specific to one commercial ddPCR platform, caused by uncorrected droplet volume | 2 published | PMID 31594564 | 7 |
| Stochastic PCR errors persist as a principal source of NGS noise in cfDNA sequencing even after duplex molecular barcoding, clonal-hematopoiesis filtering, and patterned-error suppression, per a 2020 study | 2 published | PMID 32084206 | 7 |
| A 2023 study calibrated NGS-based cfDNA quantification against droplet digital PCR using spiked synthetic normalizer DNA, improving linearity from R²=0.91 (raw) to R²=0.95 (normalized) in n=12 patient and n=12 mock plasma specimens | 2 published | PMID 37244060 | 7 |
| A 2025 study prepared ctDNA reference material by micrococcal-nuclease digestion of nucleosomal DNA from cancer cell lines and reported digital-PCR limits of detection of 0.143% VAF (TP53 R175H) and 0.092% VAF (TP53 R248W) | 2 published | PMID 40428398 | 7 |
| A 2026 study from a national metrology institute produced two fragment-length-resolved certified reference materials (80/240 bp and 80/120/160/240 bp fragments) for cfDNA isolation-recovery standardization, characterized by droplet digital PCR | 2 published | PMID 41936819 | 7 |
| A 2026 study reports NIST and LGC Clinical Diagnostics each developed candidate methylated cfDNA reference materials, validated across six laboratories using qPCR, dPCR and bisulfite sequencing | 2 published | DOI 10.1186/s13148-026-02156-3 | 7 |
| A 2025 study achieved duplex error-corrected whole-genome sequencing of cell-free DNA at an error rate of 7.7×10⁻⁷ using ~120x deep, stated lower-cost sequencing (Ultima Genomics), enabling tumor-naive ctDNA detection | 2 published | PMID 40217113 | 7 |