# Toward the simultaneous detection of multiple diseases with a highly cost-effective cell-free DNA methylome test

- **DOI:** [10.1073/pnas.2518347123](https://doi.org/10.1073/pnas.2518347123)
- **Rating:** Sound (3.7 out of 5)
- **Verdict:** MethylScan advances low-cost cfDNA methylome diagnostics across multiple tasks, but internal validation limits clinical confidence.
- **Framework version:** v5.5
- **Pipeline version:** v0.4
- **Evaluation date:** 2026-07-21
- **Evaluation page:** https://www.nabu.science/paper/10.1073%2Fpnas.2518347123

## Key claims

1. MethylScan achieved an area under the receiver operating characteristic curve (AUROC) of 0.938 (95% CI: 0.920 to 0.954), with a sensitivity of 63.3%
   - Support: Supported
   - Type: Primary finding
   - Basis: internally cross-validated diagnostic performance is consistently reported, without independent external validation
   - Location: Abstract; Results, Fig. 6C
2. For early-stage cancers, the AUROC was 0.916 (95% CI: 0.890 to 0.940), with 55.3% sensitivity
   - Support: Supported
   - Type: Secondary
   - Basis: stage-specific cross-validation metrics are reported, though sensitivity remains moderate
   - Location: Abstract; Results, Fig. 6D
3. In liver cancer surveillance, MethylScan achieved an AUROC of 0.927 (95% CI: 0.889 to 0.959), with 79.6% sensitivity
   - Support: Supported
   - Type: Secondary
   - Basis: high-risk surveillance analysis reports appropriate metrics, but remains internally validated
   - Location: Abstract; Results, Liver Cancer Detection, p. 7
4. we developed MethylScan, a highly cost-effective approach for cfDNA methylome sequencing
   - Support: Tentative
   - Type: Secondary
   - Basis: assay enrichment and depth reductions support direction, but direct cost benchmarking is limited
   - Location: Abstract; Introduction pp. 1–2; Results, Fig. 2–3
5. The assay also demonstrated strong performance in additional diagnostic tasks
   - Support: Tentative
   - Type: Exploratory
   - Basis: additional tasks are heterogeneous, with partial liver-disease classification and uneven class performance
   - Location: Abstract; Results pp. 7–8

## Strengths

- **Coherent lower-cost assay strategy** The assay targets regions where blood-cell cfDNA background is consistently hypomethylated and uses MSRE digestion plus targeted capture to enrich disease-informative hypermethylated fragments. This directly addresses the sequencing-depth cost barrier identified in the introduction.
  - Dimension: contribution
  - Location: Introduction pp. 1–2; Results, Fig. 1 and Fig. 2
- **Analytical spike-in validation included** The paper evaluates tumor signal detection using a liver tumor gDNA dilution series spanning 0% to 5% tumor fraction, with replicates and comparison to the assay without digestion. This provides an analytical bridge between assay chemistry and downstream diagnostic modeling.
  - Dimension: methodological rigour
  - Location: Results, Fig. 3; Methods, Panel Validation by the Spike-In Dilution Series Experiment
- **Main cancer metrics transparently reported** For multicancer detection and liver cancer surveillance, the article reports AUROC, sensitivity, specificity, confidence intervals, and repeated cross-validation details. Stage-specific breakdowns are also provided for cancer detection, including early-stage performance.
  - Dimension: reporting
  - Location: Results, Fig. 6C–D; Results, Liver Cancer Detection, p. 7

## Limitations

- **No independent external validation cohort** The headline diagnostic results are based on stratified fivefold cross-validation repeated 50 times within the study cohort. That is useful for development-stage evidence but does not establish performance in independent intended-use populations.
  - Dimension: methodological rigour
  - Location: Results, Fig. 6C; Methods, Marker Selection and Predictive Model for Cancer Detection
- **Broad simultaneous-detection framing overreaches** The title, abstract, and significance statement describe simultaneous detection and comprehensive health monitoring, but the reported evidence consists of separate models for distinct applications. This creates a reliability-relevant framing concern rather than a contradiction in the numerical results.
  - Dimension: reporting
  - Location: Title; Abstract; Significance statement; Discussion pp. 8–9
- **Uneven non-cancer task maturity** The liver disease classifier assigns reliable labels to only 62.4% of patients, and ALD and MASLD performance is weaker than HBV and HCV. These caveats matter because the abstract summarizes additional tasks as strong platform evidence.
  - Dimension: methodological rigour
  - Location: Results, “Using MethylScan For Noninvasively Classifying Multiple Liver Diseases,” p. 7

## Reliability checks

Caveats (4 of 4 checks).

- **Internal coherence:** Noted. The headline numerical performance metrics are internally consistent, but the article’s broad simultaneous-detection framing runs ahead of the separately evaluated retrospective models. Liver disease classification is disclosed with a substantial non-classification rate, but that caveat is less prominent than the overall multi-application framing.
- **Research conduct:** Noted. Human genomic data availability and code availability are specifically stated, and commercial conflicts are explicitly disclosed. The disclosed concentration of author employment, stock, consulting, executive roles, and patent interests in a company tied to the assay is a conduct watch-out for interpretation, especially given internal-only validation.
- **Reference integrity:** Clean. 35 references verified via PubMed, Crossref, or OpenAlex.
  - References not verified automatically are resolved manually within 24 hours.

## Post-publication record

- **Status:** Clean
- **Summary:** No retraction or expression of concern found in Crossref (includes Retraction Watch).
- **Last checked:** 2026-10-17
