Editors Retract Breast Cancer Study After Overlapping Migration and Invasion Images Surface

The editors of Cancer Cell International have retracted a 2015 study that claimed the cell-cycle protein cyclin D3 could predict disease-free survival in patients with breast cancer, bringing an eleven-year publication saga to a close and adding another entry to the growing catalogue of cancer papers withdrawn over compromised figure integrity. The retraction notice, published on 7 October 2026 as volume 26, article number 333 of the journal, states that the Editors-in-Chief no longer have confidence in the results and conclusions presented in the original article, which had appeared in the journal on 26 September 2015 under the title Cyclin D3 predicts disease-free survival in breast cancer.

The concerns that triggered the retraction are strikingly specific and, to anyone familiar with laboratory practice, deeply troubling. According to the notice, the Migration image in the Control group partially overlaps with the Invasion image in the same group in Figure 3 of the paper. A second, parallel problem affects the experimental arm of the same figure: the Migration image in the siCCND3 group partially overlaps with the Invasion image in that group as well. In plain terms, two pairs of images that were supposed to document two distinct biological measurements in two distinct experimental conditions appear, at least in part, to be the same pictures. The editors concluded that these duplications undermined the reliability of the figure and, by extension, the paper’s central claims.

To understand why such duplication matters, it helps to understand what migration and invasion assays actually measure and why they are treated as independent lines of evidence. In cancer biology, a migration assay typically assesses the ability of cells to move across a surface or through a gap, often using a scratch wound or a porous membrane without any coating. An invasion assay asks a harder question: can the cells not only move but also digest their way through a matrix that mimics the extracellular barriers they would encounter in tissue, usually a layer of Matrigel or a similar basement-membrane substitute coating the membrane? The two assays are conceptually related but biologically distinct, and a paper arguing that silencing a gene alters cancer cell behavior is expected to show coherent, distinguishable changes in both. When the control-group migration image overlaps with the control-group invasion image, the two assays collapse into one measurement. When the same overlap appears in the knockdown group, the entire comparative framework of the experiment is called into question, because the reader can no longer tell whether the reported differences between conditions reflect biology or image handling.

The gene at the center of the retracted work, CCND3, encodes cyclin D3, one of three D-type cyclins that serve as regulatory subunits for cyclin-dependent kinases 4 and 6. These kinase complexes phosphorylate the retinoblastoma protein, releasing transcription factors that drive cells from the G1 phase of the cell cycle into DNA synthesis. Because D-type cyclins sit at the gateway to proliferation, their dysregulation has long been implicated in cancer, and the related cyclin D1 has become a well-established diagnostic and prognostic marker in several tumor types, most famously in mantle cell lymphoma and in certain breast cancers. Cyclin D3 has a more nuanced profile: it is essential for the development of certain immune cell lineages, it is amplified or overexpressed in subsets of lymphomas and other malignancies, and its role in breast cancer specifically has been debated in the literature. The retracted 2015 paper had positioned cyclin D3 as a predictor of disease-free survival in breast cancer patients, a claim that, if validated, could have influenced how clinicians stratify patients for follow-up intensity or adjuvant therapy decisions.

The retraction notice makes clear that the concerns were raised by the authors themselves following publication, a detail that distinguishes this case from many retractions initiated by external sleuths or journal readers. The notice states that concerns were raised by the authors, including but not limited to the two image-overlap issues in Figure 3, indicating that the listed problems may not be exhaustive. Equally notable is what the notice records about the authors’ subsequent behavior: the authors have not responded to any correspondence from the Editor about this retraction. That silence leaves the record without an author explanation for the duplicated images, without a corrected version, and without any authorial dissent from the retraction itself. Under the retraction guidelines that most major publishers follow, a retraction is not a finding of misconduct by the journal; it is a statement that the published record can no longer be trusted, and it stands regardless of whether authors cooperate.

The original study emerged from the Department of Breast Surgery at the Breast Cancer Institute of Fudan University Shanghai Cancer Center, one of China’s leading oncology institutions, with contributing authors also affiliated with the center’s Department of Pathology and with the Department of Oncology at Shanghai Medical College of Fudan University. The author list included Yayun Chi, Sheng Huang, Mengying Liu, Liang Guo, Xuxia Shen and Jiong Wu, with Chi, Huang and Liu credited as equal contributors and Wu listed as the corresponding author. The paper was published open access in 2015 as volume 15, article 89 of Cancer Cell International, under the DOI 10.1186/s12935-015-0245-6, and the retraction notice itself carries the DOI 10.1186/s12935-026-04480-3. Open-access articles are licensed under Creative Commons terms that permit wide sharing and reuse, which means retracted copies can circulate freely on repositories, course packs and preprint-style aggregators long after the retraction notice appears, a structural challenge that the scholarly publishing industry continues to grapple with through initiatives such as CrossMark, which flags the version-of-record status of a document to readers who encounter it elsewhere.

Retractions of this kind rarely arrive in isolation, and the eleven-year gap between publication and withdrawal invites questions about how such overlaps survived peer review in the first place. Peer reviewers evaluate the scientific logic, the novelty and the statistical presentation of a manuscript, but they are volunteers, typically unpaid, and they are rarely equipped with the software tools now used to detect image manipulation. Over the past decade, the field has developed increasingly sensitive approaches to figure forensics, including tools that scan manuscripts for duplicated regions within and across images, even when the duplicates have been rotated, flipped, resized or adjusted for brightness. Journals have also begun screening at submission rather than after publication. A partial overlap between a migration image and an invasion image is exactly the class of error such screening is designed to catch, and its detection a decade after publication illustrates both how far the technology has come and how much of the older literature remains unexamined.

For clinicians and researchers who cited the original paper, the retraction carries practical consequences. Systematic reviews and meta-analyses that aggregated prognostic markers for breast cancer may have incorporated the cyclin D3 disease-free survival findings, and those analyses now require revision to exclude the retracted data. Databases that track gene-disease associations and prognostic signatures will typically annotate the paper as withdrawn, but the speed and completeness of such updates vary. For patients, the direct clinical impact is likely limited, since a single retrospective biomarker study from 2015 would not, on its own, have established cyclin D3 as a standard clinical test; prognostic markers require independent validation in large, multi-institution cohorts before entering practice. The retraction nonetheless removes one data point from the evidence base and, more importantly, removes a claim that may have been cited as supporting evidence in downstream papers, a phenomenon known in the literature-watching community as citation contamination, where a retracted finding continues to accrue citations from authors unaware of its status.

The case also speaks to the specific vulnerabilities of functional assays in cancer research. Migration and invasion images, along with flow cytometry plots, western blots and colony-formation photographs, are among the most frequently implicated elements in retractions, because they are generated in large numbers, look superficially similar across conditions and are easy to mislabel or reuse when assembling multi-panel figures. The pattern in this retraction, where the duplication occurs within each experimental group rather than between groups, suggests the kind of assembly error that can arise when many images from the same experiment are sorted and placed into panels, though the notice does not speculate on cause, and no author explanation is on record. What the notice does establish is the editorial standard applied: when two supposedly distinct measurements within the same figure cannot be distinguished, the editors’ confidence in the results and conclusions is lost, and the appropriate remedy is withdrawal of the paper rather than correction, because the integrity of the underlying experiment cannot be reconstructed from the published images alone.

The retraction notice closes the chapter on a study that once contributed to the conversation about cyclin D3 and breast cancer prognosis, but it does not settle the underlying biology. Whether cyclin D3 expression genuinely stratifies disease-free survival in breast cancer remains an open empirical question that would need to be addressed by studies with verified, independently generated data. What the case does settle, for the record, is that the 2015 paper cannot serve as evidence in that conversation. Readers who encounter the original article, its PDF or its citations in the wild are directed to the retraction notice at Cancer Cell International, and the scholarly consensus it represents is unambiguous: the figure evidence no longer supports the conclusions, the authors have not contested the withdrawal, and the article now stands as a cautionary example of why image integrity, quiet as it may seem beside headlines about genes and survival, is the load-bearing wall of experimental science.

Subject of Research: Retraction of a 2015 breast cancer biomarker study on cyclin D3 and disease-free survival due to duplicated images

Article Title: Retraction Note: Cyclin D3 predicts disease-free survival in breast cancer

Article References: Chi, Y., Huang, S., Liu, M., Guo, L., Shen, X., & Wu, J. (2026). Retraction Note: Cyclin D3 predicts disease-free survival in breast cancer. Cancer Cell International, 26(1), Article 333. https://doi.org/10.1186/s12935-026-04480-3

Image Credits: AI Generated

DOI: 10.1186/s12935-026-04480-3

Keywords: retraction, breast cancer, cyclin D3, CCND3, disease-free survival, image duplication, migration assay, invasion assay, Cancer Cell International, research integrity, prognostic markers, cell cycle