Academic Fraud Investigation Report: 'Prognostic value of Musashi 2 (MSI2) in cancer patients: A systematic review and meta-analysis' (DOI: 10.3389/fonc.2022.969632)

Academic fraud report · Geng Detector

Summary

This report flags the 2022 Frontiers in Oncology systematic review/meta-analysis on MSI2 as highly suspicious. The central finding is systematic arithmetic fabrication in the N (sample size) column of Table 3. Recomputing totals from the constituent studies listed in Table 1 yields substantial discrepancies in nearly every clinicopathological subgroup, including Gender (reported 1650 vs. recomputed 1603, +47), Liver metastasis (338 vs. 306, +32), Age (1414 vs. 1404, +10), Tumor size (1012 vs. 1002, +10), Degree of invasion, Lymph node metastasis, and Degree of differentiation. Strikingly, Table 2 subgroup sums were arithmetically exact (e.g., CRC 105+180+85+164=534), indicating the authors can add correctly. The selective pattern in Table 3 strongly suggests reuse or copying of statistical output from a different biomarker meta-analysis, with N values left unedited. A secondary issue involves an I² of 86.2% in the Degree of differentiation analysis, which is not discussed or subjected to sensitivity analysis. Limitations: the underlying extracted values themselves cannot be independently verified without the authors' raw extraction files and Stata logs.

Verdict

🟠 Highly suspicious. The investigation identifies systematic arithmetic inconsistencies in Table 3 that are inconsistent with the authors' demonstrated arithmetic capability in Table 2, raising substantial concern that statistical results may have been copied or fabricated.

Key findings

Systematic N discrepancies in Table 3. Sums of study-level sample sizes from Table 1 do not match the totals reported in Table 3 for nearly every clinicopathological subgroup.
Magnitude of discrepancies ranges from +10 to +47 patients, with Gender showing the largest absolute mismatch (reported 1650; recomputed 1603; Δ = +47).
Selective inconsistency pattern. Table 2 subgroup arithmetic is exact (e.g., CRC: 105 + 180 + 85 + 164 = 534), demonstrating the authors are capable of correct summation; the failure is therefore not innocent error.
Unexplained high heterogeneity. I² = 86.2% in the Degree of differentiation analysis is reported without discussion, sensitivity analysis, or subgroup exploration, undermining the validity of the pooled estimate.
Likely reuse of statistical output. The pattern (precise OR/CI values alongside fabricated N totals) is consistent with the authors having copy-pasted meta-analytic output from a different biomarker review and failing to update the sample-size column.

Evidence highlights

DOI: 10.3389/fonc.2022.969632
Location of primary anomaly: Table 3, Page 07 (Meta-analysis of MSI2 and clinicopathological features in cancer patients)
Recomputed vs. reported totals (from Table 1 study-level N):
Age (young vs. old), 12 studies (16, 19, 21, 22, 24, 25, 27, 28, 30, 32, 33, 34): 105 + 75 + 180 + 126 + 82 + 162 + 62 + 164 + 67 + 181 + 149 + 51 = 1404; reported 1414 (Δ = +10).
Gender (male vs. female), 11 studies: summed = 1603; reported 1650 (Δ = +47).
Liver metastasis, 3 studies (23, 28, 34): 91 + 164 + 51 = 306; reported 338 (Δ = +32).
Tumor size, 9 studies: summed = 1002; reported 1012 (Δ = +10).
Degree of invasion, Lymph node metastasis, Degree of differentiation: also failed independent verification.
Counter-evidence (Table 2, Page 05): Subgroup sums reconcile precisely, e.g., CRC 105 + 180 + 85 + 164 = 534.
Secondary anomaly: I² = 86.2% in Degree of differentiation (Table 3) is reported without methodological discussion or sensitivity analysis; pooled result presented as a definitive null (P=0.418).

Notes

All recomputations are derived solely from the published Table 1 study-level N values; the underlying extracted data themselves cannot be verified without the authors' raw extraction sheets and Stata/R logs.
Discrepancies could theoretically reflect inclusion of patients from overlapping cohorts within a single subgroup; however, this would not explain the precise arithmetic in Table 2 nor the selective failure restricted to Table 3.
Recommended follow-up: request original extraction tables and complete statistical output logs from the corresponding author; consider formal notification to the journal editorial office and the authors' institutional research integrity committee (affiliation reported: Nanjing Medical University Affiliated Taizhou People's Hospital).