Academic Integrity Report: Knockdown of the EEF1A2 gene reduces lung cancer brain metastasis by downregulating the BCL10/NFκB pathway (DOI: 10.20517/2394-4722.2024.77)

Academic fraud report · Geng Detector

Summary

This report flags serious concerns of potential data fabrication and methodological inconsistency in the above paper, published in Journal of Cancer Metastasis and Treatment (2025). The verdict is HIGHLY SUSPICIOUS. The most damning finding concerns Table 1 (page 6, clinical information of LC and LCBM patients): subgroup counts fail basic arithmetic (e.g., LC gender: 8+12=20 instead of the stated n=21; LC age groups: 7+11=18), and several P-value cells contain nonsensical integers such as '12' and '11', strongly suggesting copy-paste of footnote markers or typographic fabrication. A second contradiction appears between Methods (21 LC and 21 LCBM patients enrolled) and the Figure 6 caption (only five LC and five LCBM samples analyzed by Western blot), without any explanation. Additional concerns include an implausibly compressed timeline (sample collection ending August 2023, manuscript received July 2024) covering transcriptomics, proteomics (TMT), stable cell lines, full in vitro panels, and a nude-mouse intracardiac/intracarotid brain metastasis model, plus unreported animal mortality. Western blot panels (Figures 3D, 4D–G, 6B) could not be pixel-level analyzed but warrant raw-image scrutiny. Confidence is moderate-to-high; final determination requires institutional investigation.

Verdict

Highly suspicious (🟠). Primary clinical data table contains arithmetic contradictions and implausible P-values suggestive of fabrication or careless data transcription. Secondary concerns involve sample-size inconsistency and unrealistic experimental timelines.

Key findings

Table 1 arithmetic failure: Stated LC group n=21, but Male 8 + Female 12 = 20 (one patient unaccounted); age distribution ≥65 (7) + <65 (11) = 18 (three patients missing).
Implausible P-values in Table 1: Tumor-marker columns (CA125, CA153, CA242, CA724, CA199) list P-values as "12", "11", "12" — impossible output from any statistical software; likely mis-pasted footnote markers.
Unrealistic P-value precision: Remaining P-values retain four decimal places (0.6262, 0.3722) despite very small subgroup denominators (e.g., only ~11 cases with tumor-marker data).
Sample-size contradiction: Methods state 21 LC and 21 LCBM patients enrolled; Figure 6 caption states Western blots used five LC and five LCBM samples with no justification for the reduction.
Timeline compression: Sample collection ended August 2023; manuscript received July 2024 — under one year to complete microarray analysis, ceRNA network construction, stable cell-line generation, full in vitro assays, TMT proteomics, and a 50-day intracardiac/intracarotid nude-mouse brain-metastasis model.
Missing animal mortality data: Intracarotid injection models typically incur substantial mortality; none is reported despite "clean" endpoint statistics.
Western blot panels (Figures 3D, 4D–G, 6B) flagged for raw-image review: No pixel-level duplication analysis possible from the supplied PDF; multi-molecular-weight composites in Figure 4 are particularly prone to hidden splicing.

Evidence highlights

Direct numeric evidence from Table 1:
LC sex split: Male 8 / Female 12; sum = 20 ≠ 21.
LC age split: ≥65 = 7, <65 = 11; sum = 18.
P-value cells: CA125 = "12"; CA153 = "11"; CA242, CA724, CA199 = "12".
Reported P-values with four-decimal precision: 0.6262, 0.3722.
Methods text: "From September 2020 to August 2023… included 21 patients with LC and 21 with LCBM."
Figure 6 caption: "Western blot analysis of EEF1A2… in tumor samples from five LC patients and five LCBM patients."
DOI: 10.20517/2394-4722.2024.77.

Notes

Confidence is moderate-to-high for the Table 1 arithmetic issue; lower for Western-blot splicing, which requires raw uncropped images.
Authors should be asked to (1) reconcile Table 1 counts, (2) explain the integer P-values, (3) justify the n=21 → n=5 reduction for Figure 6, and (4) provide original uncropped blot images and animal mortality records.
Affiliations cited in the report include Kunming Medical University Third Affiliated Hospital and Yunnan University of Chinese Medicine; institutional ethics committees may need to review sample provenance and animal-experiment records.
This automated report is an aid to discussion, not a final determination of misconduct.