More than 580 scientific papers on central nervous system tumors have been retracted, and the vast majority were pulled for research misconduct rather than honest error, according to a systematic bibliometric analysis published in the Chinese Neurosurgical Journal. Their findings paint a sobering picture of a research field in which flawed studies accumulate citations for years after being formally withdrawn from the scientific record.
The scale of the problem has grown sharply. Of 68,154 retraction records retrieved from the database on December 31, 2025, screening and exclusion narrowed the field to 582 CNS tumor articles. The earliest retracted paper had been published in 1975, and the first retraction occurred in 1977, but the annual numbers climbed steeply in recent years. Retractions peaked at 130 in 2023, while the number of originally published papers that were later retracted exceeded 30 in 2015 and peaked at 82 in 2022. Statistical trend tests confirmed the pattern: both retracted publications and retractions showed strong upward trends over time, with Kendall’s tau coefficients of 0.69 and 0.72 respectively, both statistically significant.
Tumor type mattered. Among the twelve categories of central nervous system tumors defined by the 2021 WHO Classification of Tumors of the Central Nervous System, retracted papers spanned ten categories, but gliomas, glioneuronal tumors and neuronal tumors dominated, accounting for 74.57 percent of retractions, or 434 of the 582 papers.
The reasons behind the retractions were strikingly skewed toward misconduct. In total, 469 articles, or 80.58 percent, were retracted due to misconduct. The three most common specific misconduct reasons were paper mill involvement, accounting for 30.58 percent of all retractions; duplication of or within images, at 28.35 percent; and compromised peer review, at 13.57 percent. The single most frequent reason overall was investigation by journal or publisher, cited in 47.42 percent of cases, a detail the authors interpret as evidence that editorial scrutiny after publication has become a crucial line of defense when pre-publication peer review fails.
Geographically, the analysis found that China led in absolute numbers of first-author retractions, contributing 69.02 percent of the 581 papers with country information, followed by the United States at 11.70 percent and India at 8.09 percent. The authors are careful to stress that these are counts, not rates: without denominator data on total publication output per country, a country with a large research volume may naturally show more retractions. The composition of the subgroups also differed markedly. The median publication year for US first-author papers was 2011, compared with 2019 for China and 2022 for India, and 69.1 percent of US papers predated 2015, versus roughly 11 percent for both China and India.
Most of the retracted work was preclinical. Basic research made up 75.26 percent of the dataset, while clinical research accounted for 17.01 percent, with reviews, meta-analyses and commentaries comprising the remaining 45 papers. The authors note that this clinical proportion is substantially lower than those reported for retracted neurosurgery and spine literature, suggesting that research integrity problems in this field concentrate heavily in laboratory-based studies rather than in studies directly involving patients.
Perhaps the most troubling finding concerns what happens after retraction. Of the 569 retracted papers indexed in Web of Science, 515 had been cited at least once, accumulating 12,151 citations in total, and 89.71 percent of those cited papers, 462 in all, continued to receive citations after their retraction dates. Even more striking, 35.34 percent received more citations after retraction than before. The median counts were five citations before retraction, three after, and ten in total. Because experimental findings in neuro-oncology may inform glioma grading, treatment selection, imaging interpretation and prognostic assessment, the authors warn that this persistent citation of withdrawn work can contaminate how later evidence is framed and interpreted, creating what they describe as a chain of flawed evidence.
Correction also came slowly. The median time from publication to retraction was 805.5 days, roughly 2.2 years, with the longest lag stretching to 9,973 days. Retraction lag varied significantly by reason category, with misconduct cases showing the longest median lag of 849 days, likely reflecting the lengthy investigations required by journals, publishers and institutions. Basic science articles took the longest to retract, with a median of 1,025 days, while commentaries were withdrawn fastest at 130 days. Lag also increased with author count, from 560 days for papers with one to three authors to 1,211 days for those with ten or more, a pattern consistent with the idea that larger author teams complicate investigation logistics. The notably long lag for US first-author papers, a median of 3,416.5 days, appears to reflect the older publication years in that subgroup rather than national differences in retraction processes, since publication year and retraction lag were strongly negatively correlated.
Citation behavior also differed across journal quality. Articles published in first-quartile journals and in journals with impact factors of five or higher attracted the highest annual citation rates both before and after retraction, meaning that flawed work published in prestigious venues circulated most widely. The journal landscape itself raised concerns: the 582 papers appeared in 217 journals, and 30 of those journals lacked a 2024 impact factor, including 19 that had been dropped by Clarivate. Those 19 dropped journals alone accounted for 16.15 percent of the retracted papers, a pattern the authors associate with compromised editorial standards and systematic manipulation of the publication process, though they caution that a cross-sectional, count-based design cannot establish whether such journals actively facilitated flawed research.
The study has limitations the authors acknowledge openly. The Retraction Watch Database may not capture every retraction, the title-level keyword search may have missed relevant articles, citation data came only from Web of Science with limited coverage of non-English journals, and classifications of reasons and article types involve subjective judgment. Country of origin was assigned solely by first-author affiliation, and because only retracted articles were analyzed, the findings describe absolute distributions rather than relative retraction risk. Even so, the core message stands: retractions in CNS tumor research have risen sharply, misconduct dominates as the cause, the median correction time exceeds two years, and nearly nine in ten retracted articles that were ever cited continued to be cited afterward. The authors advise clinicians and researchers to check retraction status through databases such as Retraction Watch, PubMed or Web of Science before relying on any published finding, and they argue that current mechanisms for identifying and communicating retractions remain insufficient to stop flawed research from spreading through the neuro-oncology evidence base.
Subject of Research: Bibliometric analysis of retracted publications in central nervous system tumor research
Article Title: Retractions of publications on central nervous system tumors based on the Retraction Watch Database
Article References: Gao, X., Mao, R., Lin, F., Shang, J., Wu, J., Chen, L., & Wang, H. (2026). Retractions of publications on central nervous system tumors based on the Retraction Watch Database. Chinese Neurosurgical Journal, 12(1), Article 24. https://doi.org/10.1186/s41016-026-00444-8
Image Credits: AI Generated
DOI: 10.1186/s41016-026-00444-8
Keywords: retractions, CNS tumors, research misconduct, Retraction Watch Database, gliomas, paper mills, peer review, bibliometrics, neuro-oncology, citation analysis, research integrity, retraction lag
News Source: Nathaniel Bowman. (October 29, 2026). Misconduct Drives Most Retractions in Central Nervous System Tumor Research. Scienmag.
