Fraud-detection report: "Carcinosarcoma is an aggressive subtype of bladder cancer: A population-based study" (Cancer Medicine, DOI 10.1002/cam4.4611)

Academic fraud report · Geng Detector

Summary

Verdict: Highly suspicious (orange). This Chinese-language fraud-detection report, associated with id geng_geng_6a1e7a00c0d465.46377602, reviews Lin Liu, Jinglan Zhu, and Yun Tian's 2022 Cancer Medicine article (DOI 10.1002/cam4.4611) on bladder carcinosarcoma using SEER data. The report flags four major concerns. First, a methodological anomaly: the authors state they used SPSS 11.0 (released circa 2002) alongside R 3.6.3, an implausible combination for a 2021 manuscript, suggesting template reuse or paper-mill provenance. Second, a statistical inconsistency: although the authors claim "most factors were matched" after 1:3 propensity-score matching (PSM), Table 2 shows post-PSM p-values of 0.030 for Race, <0.001 for Year of diagnosis, and <0.001 for Grade, all below 0.05, indicating matching failure. Third, an unusually fast review timeline: revised 22 Dec 2021 and accepted 23 Dec 2021 — a one-day turnaround from revision to acceptance. Fourth, a data-coverage discrepancy: a referenced earlier study (1973–2004, SEER 17) identified 221 bladder carcinosarcoma cases, whereas the present study (2000–2018, SEER 18) identified only 152, without justification. Image forensics were not performed. Overall, evidence supports serious doubt regarding methodological integrity and PSM validity.

Verdict

🟠 Highly suspicious. The paper shows strong internal inconsistencies between claimed PSM success and reported post-matching p-values, uses an anachronistic software stack (SPSS 11.0 alongside R 3.6.3), and displays an unusually compressed revision-to-acceptance window. These signals are consistent with template reuse, careless methods reporting, or paper-mill involvement. Image-level forensics were not performed.

Key findings

Anachronistic software declaration (Method Section 2.5, p. 2218): SPSS 11.0 (released circa 2002) cited alongside R 3.6.3 for a manuscript submitted November 2021. Highly atypical for contemporary SEER/PSM research and suggestive of copy-pasted boilerplate.
Failed propensity-score matching contradicted in narrative (Section 3.3 & Table 2, pp. 2219–2220): Authors state "most factors were matched" after 1:3 PSM, yet post-PSM p-values are 0.030 (Race), <0.001 (Year of diagnosis), and <0.001 (Grade) — all significant at α = 0.05. The matching is therefore demonstrably unsuccessful, undermining downstream Cox regression.
Implausibly fast editorial turnaround (title page): Received 21 Nov 2021; revised 22 Dec 2021; accepted 23 Dec 2021. Only 1 day between revision and acceptance for a large SEER analysis with PSM and subgroup analyses.
Imprecise p-value reporting (Section 3.2, p. 2219): Body text reports "p < 0.05" for advanced-stage prevalence comparison (66.4% vs. 68.8%), while Table 2's pre-PSM p-value for SEER stage is 0.002 — vague reporting inconsistent with the underlying data.
Unjustified case-count reduction (Section 4 Discussion, p. 2221): A cited Wang et al. study (SEER 17, 1973–2004) found 221 bladder carcinosarcoma (BCS) cases with median OS 9 months; the current study (SEER 18, 2000–2018) finds only 152 BCS cases, despite broader database coverage and a more recent period. No methodological explanation is provided.

Evidence highlights

DOI: 10.1002/cam4.4611
Title: *Carcinosarcoma is an aggressive subtype of bladder cancer: A population-based study*
Authors: Lin Liu, Jinglan Zhu, Yun Tian — *Cancer Medicine*, 2022
Timeline: Received 21 Nov 2021; Revised 22 Dec 2021; Accepted 23 Dec 2021
Software declaration: SPSS 11.0 + R 3.6.3 (Section 2.5)
Table 2 post-PSM p-values (CS vs. TCC): Race 0.030; Year of diagnosis <0.001; Grade <0.001
Table 2 pre-PSM SEER stage p-value (CS vs. SCC): 0.002, vs. "p < 0.05" in text
BCS case counts: prior SEER 17 (1973–2004) = 221; current SEER 18 (2000–2018) = 152

Notes

Image forensic analyses (Geng Forms 1 and 3) were not conducted because original high-resolution figures were unavailable; only the PDF text was reviewed.
The flagged issues concern methodology reporting and statistical logic rather than confirmed data fabrication; formal institutional investigation would be required to establish misconduct.
Reviewers and editors may wish to verify the peer-review file and request the authors' raw R/SPSS code, PSM logits, and inclusion/exclusion counts to reconcile the 221→152 case-count discrepancy.
Report generated with AI assistance; final determination of academic misconduct rests with qualified institutional bodies.