Integrity review of "Regulatory pathways and drugs associated with ferroptosis in tumors" (Cell Death and Disease, 2022)

Academic fraud report · Geng Detector

Summary

Verdict: No evidence of misconduct; the article is cleared. The paper is a purely narrative review (DOI: 10.1038/s41419-022-04927-1) in which all four figures (Fig. 1–4) are author-drawn schematic diagrams of ferroptosis-related signaling pathways (defense pathways, core machinery, NRF2 network, p53 roles), and all five tables (Table 1–5) are textual summaries of miRNAs, lncRNAs, circRNAs, preclinical drugs, and nanomaterials. No Western blots, microscopy images, flow cytometry plots, or quantitative bar/scatter charts are present, so the primary targets of image-forensic checks do not exist in this manuscript. Forensic signals reported by the automated tool (noise-variance anomalies, copy-move matches, weak cross-image PRNU similarity, NCC ≈ 0) have benign explanations: schematic diagrams contain inherently repetitive geometric elements (bilayer ribbons, circular organelles, rectangular protein icons, arrows, text boxes) that naturally produce high copy-move hit rates and uneven noise variance, while near-zero cross-figure PRNU actively excludes any shared-sensor fingerprint. The Data Availability statement confirms no original experimental data were generated. Limits: forensic interpretation is necessarily qualitative for vector schematics, and the review itself could not be audited for citation accuracy or AI-assisted text characteristics beyond noting that Editage-style polishing is acknowledged.

Verdict

Cleared (✅). No actionable integrity concerns identified. The manuscript is a narrative review containing no experimental figures or quantitative datasets subject to image or statistical manipulation checks.

Key findings

The paper is a literature review; Fig. 1–4 are author-illustrated mechanistic schematics (ferroptosis defense pathways; core molecular machinery; NRF2 network; p53 roles), and Table 1–5 are textual compilations of miRNAs, lncRNAs, circRNAs, drugs, and nanomaterials.
No Western blots, gel images, microscopy photos, flow cytometry plots, or quantitative bar/scatter charts exist in the article, so image-manipulation forensics has no primary target object to evaluate.
Automated image-forensic signals (noise variance on img-000.png, img-003.jpg, img-005.jpg; copy-move matches on img-002.jpg, img-004.jpg; weak cross-image PRNU NCC ≈ 0) are attributable to benign features of vector schematics: repeated bilayer ribbons, circular organelles, rectangular protein icons, arrows, and text boxes produce natural copy-move hits, and large flat-color fills next to thin lines produce uneven noise variance.
Near-zero cross-figure PRNU (NCC ≈ 0) actively excludes any shared sensor fingerprint between figures, ruling out experimental-image reuse.
The Data Availability statement ("All data generated or analyzed during this study are included in this published article") is consistent with a review that generates no raw experimental data.
No quantitative elements (sample sizes, SD/SEM, p-values, dose–response curves, F/t statistics, ethics approval numbers, reagent dates) are present in the text, and there is no evidence of short-term high-output production patterns from this author group.

Evidence highlights

Fig. 1 caption: "Four ferroptosis defense pathways" (schematic).
Fig. 2 caption: "Core molecular machinery and regulatory pathways of ferroptosis" (schematic).
Fig. 3 caption: "NRF2 and the ferroptosis network" (signaling network schematic).
Fig. 4 caption: "The various roles of p53 in ferroptosis" (signaling network schematic).
Table 1–5 contents: miRNA / lncRNA / circRNA / preclinical drug / nanomaterial textual summaries.
Data Availability statement confirms inclusion of all generated/analyzed data within the article, consistent with no original experimental data.
DOI: 10.1038/s41419-022-04927-1.

Notes

Forensic flags on schematics should be interpreted with caution: rule-based geometric repetition and uniform-fill regions can mimic copy-move and noise-variance artifacts, and PRNU analysis is not designed for vector illustrations.
Acknowledgement of Editage-style language polishing does not, on its own, indicate AI-generated content beyond a reasonable threshold.
A review cannot be audited for experimental-image manipulation or p-hacking in the usual sense; only citation accuracy and conceptual originality remain as theoretical concerns, which were not flagged here.
Recommended actions: no need to contact authors for raw data, post on PubPeer, notify the journal editor, or report to institutional research integrity committees.