Data integrity investigation of Geng et al. (2021) "Butyrate Suppresses Glucose Metabolism of Colorectal Cancer Cells via GPR109a-AKT Signaling Pathway and Enhances Chemotherapy"

Academic fraud report · Geng Detector

Summary

This report evaluates the Frontiers in Molecular Biosciences article by Hong-Wei Geng et al. (DOI: 10.3389/fmolb.2021.634874) and concludes the paper is highly suspicious of data fabrication. Two numerical contradictions are documented as definitive (red-level) findings. First, in HCT116 cells treated with 2 mM butyrate for 24 hours, BrdU incorporation OD values are reported as 0.65 in Figure 3E but as 1.37 in Figure 4C for the same condition. Second, the proportion of 2-NBDG-positive LoVo cells after butyrate treatment is stated as 24% in Figure 1A but as 39% in Figure 2C. Additional concerns include suspiciously round percentages for GLUT1 reduction (15%, 24%, 60%, 40%) and a Figure 2C caption that describes butyrate/PBS treatment while the main text discusses SC79 rescue. Pixel-level image analysis (Western blot duplication, splicing) could not be performed because only extracted text was available. The verdict rests on internal logical contradictions in the reported numbers and text-figure mismatches; final determination of misconduct requires institutional investigation and access to raw data.

Verdict

Highly suspicious. Internal numerical contradictions across figures describing identical experimental conditions strongly indicate fabricated or carelessly assembled data. The original PDF and raw data (flow cytometry FCS files, BrdU OD tables, full Western blot images) must be examined before a final determination of misconduct.

Key findings

Inconsistent BrdU baseline values for HCT116 under identical butyrate treatment: 0.65 OD in Figure 3E vs 1.37 OD in Figure 4C — a roughly two-fold discrepancy for the same condition.
Inconsistent 2-NBDG uptake percentages for LoVo under identical butyrate treatment: 24% in Figure 1A vs 39% in Figure 2C.
Round, suspiciously convenient GLUT1 reduction percentages: 15% and 24% (total) and 60% and 40% (membrane fraction), with no trailing decimals typical of real biological measurements.
Text–figure mismatch in Figure 2C: main text describes SC79 co-incubation rescue, while the legend refers only to 2 mM butyrate or PBS vehicle for 24 h.
Western blot reuse/splicing risk: Coomassie-stained loading controls across multiple panels (AKT, p-AKT, GLUT1, G6PD) could not be verified pixel-by-pixel; flagged for original-image review.

Evidence highlights

Figure 3E description: HCT116 BrdU OD "from 1.5 to 0.65" after butyrate.
Figure 4C description: HCT116 BrdU OD "from 1.37 to 1.64" for NaB + Con (butyrate control).
Figure 1A description: LoVo 2-NBDG-positive cells "from 36 to 24%" after butyrate.
Figure 2C description: LoVo 2-NBDG-positive cells "from 39 to 80%" with SC79 rescue, butyrate-only baseline 39%.
Figure 1C description: GLUT1 total reduced by "15 and 24%"; membrane fraction reduced by "60 and 40%".
Figure 2C legend: "2-NBDG… after incubating with 2 mM butyrate or PBS vehicle for 24 h" (no mention of SC79).

Notes

Limitations: only extracted text was available; pixel-level image duplication and splicing checks (Style 1 and Style 3 of forensic image analysis) were not possible.
The BrdU and 2-NBDG contradictions alone are sufficient to warrant contacting the corresponding author for raw data and to alert the journal editorial office of *Frontiers in Molecular Biosciences*.
Recommended follow-up: obtain original PDF, inspect β-actin and Coomassie loading controls across figures, request FCS files for all flow cytometry panels, and cross-check Figure 2C's actual content against its legend.