Integrity Report: 'Network pharmacology and animal experimental study on Fangji Huangqi Decoction activating autophagy via AMPK/mTOR pathway in knee osteoarthritis'

Academic fraud report · Geng Detector

Summary

This report assesses a 2026 online-first paper in China Journal of Chinese Materia Medica (DOI: 10.19540/j.cnki.cjcmm.20260523.502) by Wang Houyuan et al. The verdict is severe: multiple core findings indicate likely data fabrication and methodological inconsistency. Independent t-test calculations on Table 6 LC3B data yield P≈0.069 (model vs low-dose) and P≈0.04 (low-dose vs high-dose), not the reported P<0.001. The methods section lists the formula's constituents as 'Baizhu (Atractylodes macrocephala)' in the abstract but uses 'Cangzhu (Atractylodes lancea)' as the TCMSP search term—two botanically and chemically distinct species. Table 4's header lists IL-1β, IL-18, TNF-α, while its column headers and the cited kit catalog (Beyotime PI328 = IL-6) indicate IL-6 was actually measured. The corresponding author's email is a student account. Image-level analysis was not possible. Confidence is high for findings 1–3; finding 4 is circumstantial; figure integrity remains unverified.

Verdict

🔴 Confirmed / Severe concerns. Multiple independent inconsistencies in statistics, methodology, and reporting strongly suggest data fabrication or patchwork assembly.

Key findings

Fabricated P-values (Table 6): Recalculated t-tests on reported LC3B means do not support P<0.001.
Herbal identity mismatch (Atractylodes species): 'Baizhu' vs 'Cangzhu' conflict between formula description and TCMSP retrieval query.
Inflammatory-factor mismatch (Table 4): Discrepancy between narrative, table title, column headers, and catalog numbers (PI328 = IL-6).
Corresponding-author email irregularity: A 'stu.gzucm.edu.cn' address with 2024 student ID prefix.
Figure integrity: Not assessed (text-only extraction). Western blot and IHC images require pixel-level inspection.

Evidence highlights

Table 6, LC3B (mean ± SD, n=3): Model 10.25 ± 0.94; Low-dose 20.26 ± 6.95; High-dose 37.40 ± 7.07.
Model vs Low-dose: mean diff = 10.01; pooled SE ≈ 4.05; t ≈ 2.47, df=4 → two-tailed P≈0.069 (not significant). Paper claims 3) P<0.001.
Low-dose vs High-dose: mean diff = 17.14; pooled SE ≈ 5.72; t ≈ 3.00 → two-tailed P≈0.04. Paper claims 5) P<0.001.
Section 2.1.1: TCMSP search keywords include '苍术' (Cangzhu, Atractylodes lancea) while abstract lists '白术' (Baizhu, Atractylodes macrocephala).
Table 4 columns: TNF-α, IL-1β, IL-6; kit catalog PI328 (Beyotime) corresponds to IL-6.
Corresponding author email: 20242120156@stu.gzucm.edu.cn.

Notes

DOI preserved: 10.19540/j.cnki.cjcmm.20260523.502.
All numeric statistics above were recomputed from the paper's own reported values; no external data introduced.
Figure 4–6 (Western blot, immunohistochemistry) were not image-analyzed in this report; follow-up ImageJ examination is recommended.
Findings 1–3 are internally verifiable from the PDF; finding 4 is circumstantial; finding 5 is an open verification gap.
Final determination of misconduct requires institutional investigation.