Investigation Report: 'Identification of TMEM59L as a potential diagnosis, prognosis and immunotherapy biomarker for colon adenocarcinoma' (DOI: 10.1038/s41598-026-36478-2)

Academic fraud report · Geng Detector

Summary

This academic-integrity review concludes that the article by Wang, Jia, Du, Zhao and Shi, published in Scientific Reports, contains strong evidence of fabricated or corrupted clinical baseline data in Table 1. The high-expression group (n = 228) shows impossible arithmetic: age strata sum to 227, and three unrelated clinical variables (histological type, CEA level, colon polyp history) all share the identical '39 vs 39' patient counts with suspiciously similar percentages (17.11% / 17.12%), a pattern incompatible with authentic clinical records. A second major finding is a logical contradiction between results and conclusions: multivariate Cox analysis reportedly retained only pathologic M stage and pathologic stage as independent prognostic factors, yet the authors nevertheless constructed a Nomogram incorporating TMEM59L and promoted it as an independent biomarker. A minor methodological concern is the non-standard citation of a '2020 version' of the TIDE algorithm, whose reference work is Jiang et al. (Nature Medicine, 2018). Image-level and raw-data forgery checks were not feasible from the submitted text. Overall verdict: confirmed ('实锤') data fabrication in Table 1 and unsupported prognostic claims; medium-to-low confidence on the TIDE methodology issue pending verification.

Verdict

🔴 Confirmed (实锤). The paper presents multiple, mutually reinforcing anomalies that are highly consistent with fabricated Table 1 data and logically unsupported conclusions. Confidence is high for the Table 1 arithmetic anomalies and for the Cox/Nomogram inconsistency; confidence is moderate for the TIDE methodology remark; image-based fraud detection was not possible with the materials supplied.

Key findings

Fabricated or duplicated clinical baseline data in Table 1 (high-expression group, n = 228). Three independent clinical variables share an identical '39 vs 39' split, and the age strata sum to 227 rather than 228, with missing patients unaccounted for.
Logical contradiction in prognostic claim. Multivariate Cox analysis reportedly retained only pathologic M stage and pathologic stage (both p < 0.05) as independent prognostic factors, but the authors still built a TMEM59L-inclusive Nomogram and described TMEM59L as a promising independent prognostic biomarker.
Non-standard methodology citation. The paper refers to a '2020 version' of the TIDE algorithm while citing the original 2018 Jiang et al. Nature Medicine paper (Ref 21); the versioning convention is atypical for the field.
Limited verification scope. Western blot, IHC, and flow cytometry panels (e.g., Figs. 8B, 8E) could not be examined for splicing, duplication, or PS artifacts because no high-resolution images or raw data were available.

Evidence highlights

Table 1 (Page 6), Age row (high-expression group): <=65 = 90, >65 = 137; sum = 227, one patient unaccounted for versus the stated total of 228.
Table 1 (Page 6), Histological type (high-expression group): Adenocarcinoma = 39 (reported 17.7%; computed 39/228 = 17.1%); Mucinous adenocarcinoma = 39 (17.8%). Sum = 78, leaving 150 patients with no histological classification.
Table 1 (Page 6), CEA level (high-expression group): <=5 = 39; >5 = 39; sum = 78, 150 patients missing.
Table 1 (Page 6), History of colon polyps (high-expression group): NO = 39 (17.11%); YES = 39 (17.12%); identical 39/39 split with near-identical percentages across an unrelated variable.
Results (Page 7), Fig. 2F caption and surrounding text: Multivariate Cox reportedly identified only pathologic M stage and pathologic stage as independent prognostic factors (both p < 0.05).
Fig. 3A and Discussion (Page 11): A Nomogram incorporating TMEM59L is presented and TMEM59L is promoted as an independent prognostic biomarker, contradicting the Cox result.
Materials and methods (Page 3): Reference to the '2020 version' of the TIDE algorithm, with the underlying citation being Jiang et al., Nature Medicine, 2018 (Ref 21).

Notes

The Table 1 '39 vs 39' duplication across three clinically distinct variables, combined with the broken age total, is the strongest indicator of data fabrication or uncontrolled copy-paste between records; probability of coincidental occurrence in real electronic medical records is negligible.
The Cox-vs-Nomogram contradiction means the prognostic Nomogram is statistically unjustified as presented; even if the bioinformatics pipeline were sound, the stated conclusion does not follow from the authors' own analysis.
The TIDE '2020 version' wording is suggestive but not conclusive of misconduct; it may reflect copy-pasted methods text from another paper or an idiosyncratic informal citation style. Verification against current TIDE documentation is recommended.
Western blot, IHC, and flow cytometry images were not assessable; given the confirmed issues elsewhere, these figures should be examined against original raw data if obtainable.
DOI of the investigated paper: 10.1038/s41598-026-36478-2. Any final determination of misconduct must be made by the publisher (Scientific Reports editorial office) and/or the authors' institutions (Dalian Medical University First Affiliated Hospital; First Affiliated Hospital of Wannan Medical College).