Geng Report: Impact of HMGA1 on tumorigenesis, prognosis and immune microenvironment in HNSCC — a multi-omics study (Ziang Xu et al., npj Precision Oncology, 2025)

Academic fraud report · Geng Detector

Summary

This academic-integrity report assesses Ziang Xu and colleagues' multi-omics study on HMGA1 in head and neck squamous cell carcinoma, published in npj Precision Oncology in 2025 (DOI: 10.1038/s41698-025-01068-4). The overall verdict is 'confirmed fraud' (实锤) based on three main findings. First, Figure 6A shows unmistakable vertical demarcation lines between HOK/HSC3/HN4 lanes and HN6/CAL27 lanes and across GAPDH bands, with abrupt changes in background, contrast and noise patterns inconsistent with single-membrane exposure, indicating image splicing. Second, CCK-8 proliferation curves (Figure 6C) and xenograft tumor-volume curves (Figure 6H) are unnaturally smooth with uniform error bars across all time points, lacking expected biological variability. Third, the Methods section states cells were cultured in DMEM with 1% FBS, a concentration insufficient for routine maintenance of HNSCC lines, suggesting either a critical writing error or fabrication. The reviewer recommends reporting to the journal, demanding raw blots and primary numerical data, and potential retraction if unaddressed. Confidence is high for the splicing issue; medium for the data-quality concerns, which require original records for final confirmation.

Verdict

🔴 Confirmed fraud (实锤) — The paper exhibits multiple serious concerns including apparent image splicing, biologically implausible quantitative data patterns, and a critical methodological error that undermine the validity of the wet-lab experiments.

Key findings

Figure 6A image splicing: Vertical demarcation lines between HOK/HSC3/HN4 and HN6/CAL27 lane groups, and around GAPDH bands, show abrupt changes in background, contrast, and noise — incompatible with a single membrane/exposure.
Unnaturally smooth quantitative curves (Figure 6C and 6H): CCK-8 proliferation and in-vivo tumor-volume curves display near-perfect smoothness and uniform error bars across 6h, 12h, 24h, 36h, 48h and longitudinal time points, inconsistent with expected biological variability.
Cell-culture methodology error: Methods state 1% FBS in DMEM for routine culture; standard HNSCC lines (HSC3, HN4, HN6, CAL27) require 10% FBS, while 1% is normally used only for starvation or as a chemoattractant in Transwell assays.
Recurring pattern across multiple panels: Visual anomalies identified in both HMGA1 and GAPDH regions, plus consistently unnatural error-bar uniformity, strengthen the case for fabricated or improperly processed raw data.

Evidence highlights

DOI: 10.1038/s41698-025-01068-4
Figure 6A caption (as quoted in report): *"A PCR and WB analysis demonstrated significantly elevated HMGA1 in HNSCC cells compared to HOK cells."*
Figure 6C caption: *"The effect of siHMGA1 on the proliferation of HN6 and CAL27 cells in-vitro was detected by CCK-8 assay."*
Figure 6H caption: *"Statistical graph of tumor volumes."*
Methods text (page 12): *"The cell lines were cultured in dulbecco's modified eagle medium (DMEM) supplemented with 1% fetal bovine serum and 1% antibiotic..."*
Visual-analysis summary independently flagged both the WB demarcation and the unnaturally smooth curve patterns.

Notes

Severity: Finding 1 (image splicing) is rated 🔴 high; Findings 2 (data implausibility) and 3 (1% FBS error) are rated 🟠 medium–high.
Limitations: Without access to the original uncropped WB scans and raw CCK-8/tumor-volume measurements, the smoothness concern cannot be fully adjudicated; however, the visible evidence already meets a strong prima facie case, especially for the splicing issue.
Recommended actions: (1) report to *npj Precision Oncology* editorial office; (2) request full unedited WB scans and raw numerical data from corresponding and first authors; (3) the journal should initiate an independent investigation and consider retraction if the authors cannot provide satisfactory originals.
Disclaimer: This assessment is based solely on the supplied text, visual-analysis summary, and standard peer-review norms. Final determination rests with the journal and relevant academic-integrity bodies.