Academic Integrity Review: Extrachromosomal Circular DNA: A New Target in Cancer (Frontiers in Oncology, 2022)

Academic fraud report · Geng Detector

Summary

Verdict: Cleared (no fraud indicators). This Frontiers in Oncology review article (DOI: 10.3389/fonc.2022.814504) by Wu et al. is a narrative review containing no original experimental data; all four figures are author-illustrated conceptual schematics (BFB model, detection methods, chromatin regulation, tumor evolution). Because most fraud-detection dimensions—image reuse, splicing, statistical end-digit checks, p-hacking—depend on primary experimental data, they are structurally inapplicable here. Forensic tool signals flagged (S1 column-difference uniformity, I4/I5/I10/I17/I19/I21 copy-move matches in Figures 1–3, I24 PRNU similarity PCE=61.5 between img-004 and img-006) were assessed and explained benignly: schematics contain repeated rendered motifs (chromosome bars, circular eccDNA icons, arrows, panel borders) that legitimately trigger copy-move detection, and all figures share the same vector export pipeline. No text, citation, funding, or authorship red flags were found. Confidence: high for the inapplicability of data-based checks; medium for the visual forensic signals due to inherent limitations of forensic tooling on vector schematics.

Verdict

Cleared. The article is a literature review with no original data, so the principal fraud-detection surfaces (data manipulation, image reuse from other papers, statistical anomalies) are not applicable. No actionable misconduct evidence was identified.

Key findings

Paper type: review article (REVIEW), no Methods/Results sections, zero data tables (Table count: 0), four author-illustrated conceptual figures (Figures 1–4).
Forensic S1 statistical check (column-difference uniformity) has no valid target in a review: no experimental measurement tables exist.
Forensic I4/I5/I10/I17/I19/I21 copy-move matches in Figures 1–3 are attributable to legitimately repeated schematic motifs (chromosome bars, circular eccDNA icons, arrows, axis borders, panel label boxes), not to data reuse.
Noise-variance CV values (e.g., 1.716 in img-002; 1.26 in img-005) are consistent with vector-to-raster export and do not imply manipulation.
Forensic I24 PRNU similarity (PCE=61.5 between img-004 and img-006) is inapplicable: figures are software-rendered vectors with no camera sensor fingerprint, and the article does not claim different imaging devices for different panels.
No text-level red flags: 168 references spanning 1965 (Cox et al., Lancet) to 2021 (eccDNA apoptosis work), DOIs well-formed, funding consistent with the Central South University Cancer Research Institute group (NSFC 81672683; Hunan Provincial Natural Science Foundation 2018SK21211; 111 Project 111-2-12).
No productivity anomalies: corresponding author Wei Xiong's publication pattern in eccDNA/nasopharyngeal carcinoma is consistent and discipline-coherent.

Evidence highlights

DOI: 10.3389/fonc.2022.814504; received 2021-11-13, accepted 2022-03-17, published 2022-04-14.
Article-type label on first page: REVIEW; submission to "Molecular and Cellular Oncology" section.
Cited values are single aggregate numbers from primary literature (e.g., "1 to 3 trillion base pairs", "480 kb circular domain", "20–50 enhancers", "3212 patients")—not multi-column experimental measurements.
Classic eccDNA references cited appropriately: Cox D, Yuncken C, Spriggs AI. Lancet (1965) 1(7402):55–8; Wu et al. Nature 2019 (575:699–703).
Forensic tool annotation explicitly cautions that "regularly repeated elements (multi-panel axes/borders/lane stripes/ticks) also form same-offset matches and must be interpreted in image context."

Notes

Limitations: visual forensic algorithms are optimized for photographic/blot data images and produce many incidental matches on schematic illustrations; reviewers should not over-interpret copy-move counts or PRNU NCC/PCE values for vector art. The clearing is conditional on the review's data-free nature—if any portion of the figures were later found to contain uncredited reused material from other publications, this assessment would need to be reopened. No follow-up actions (no author contact, no PubPeer post, no journal or institutional report) are recommended.